UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Towards an understanding of self-efficacy with using a manual wheelchair Sakakibara, Brodie Masaru 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_fall_sakakibara_brodie.pdf [ 4.24MB ]
Metadata
JSON: 24-1.0074105.json
JSON-LD: 24-1.0074105-ld.json
RDF/XML (Pretty): 24-1.0074105-rdf.xml
RDF/JSON: 24-1.0074105-rdf.json
Turtle: 24-1.0074105-turtle.txt
N-Triples: 24-1.0074105-rdf-ntriples.txt
Original Record: 24-1.0074105-source.json
Full Text
24-1.0074105-fulltext.txt
Citation
24-1.0074105.ris

Full Text

    TOWARDS AN UNDERSTANDING OF  SELF-EFFICACY WITH USING A MANUAL WHEELCHAIR   by   BRODIE MASARU SAKAKIBARA   B.Sc., The University of British Columbia, 2007   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY   in   THE FACULTY OF GRADUATE STUDIES  (Rehabilitation Sciences)  THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)    August 2013    ? Brodie Masaru Sakakibara, 2013        ii Abstract  Self-efficacy with using a manual wheelchair is the belief individuals have in their ability to use their wheelchair in challenging situations. It is a new construct that may have important implications on the health and well-being of wheelchair users, but has received minimal investigation. There is a need to develop an understanding of this construct in community-dwelling wheelchair users.   Purpose To investigate: the associations between self-efficacy, participation frequency (Chapter 2), and life-space mobility (Chapter 3); the health, personal, and environmental factors that predict self-efficacy (Chapter 4); and the measurement properties of the 65-item Wheelchair Use Confidence Scale (WheelCon) (Chapter 5).   Methods Multiple regression analyses were used to: examine the self-efficacy effects on participation frequency, measured using the Late-Life Disability Instrument, and life-space mobility, measured using the Life-Space Assessment; and develop a predictive model of self-efficacy, measured with the WheelCon, in a sample (n=124) of wheelchair users, ?50 years old.   Principal components analyses were used to evaluate the dimensionality of the WheelCon. Rasch analyses were used to examine the WheelCon?s item reliability in a sample (n=220) of wheelchair users, ?19 years old.  Results Self-efficacy was a statistically significant predictor of participation frequency and life-space mobility, after controlling for important confounders. The association between self-efficacy and participation frequency was mediated by life-space mobility and perceived participation limitations. The association with life-space mobility was mediated by wheelchair skills. The models accounted for 55.0% and 39.0% of the participation frequency, and life-space mobility variance, respectively.   iii Age, sex, need for a seating intervention, hours of daily wheelchair use, and formal training and assistance with wheelchair use were statistically significant predictors of self-efficacy. The model accounted for 44.0% of the self-efficacy variance.  The WheelCon was found to be comprised of two dimensions. Several items were eliminated due to their non-compliance with the Rasch model. The 13-item mobility efficacy, 8-item self-management efficacy subscales, and the combined 21-item short form have good reliability, and provide accurate and precise measurements.   Conclusion Self-efficacy has important implications on the participation frequency and life-space mobility in community-dwelling wheelchair users, ?50 years old. The construct may be assessed efficiently and precisely.                    iv Preface  This dissertation is comprised of research that was coordinated out of the GF Strong Rehabilitation Research Lab, Vancouver, British Columbia. All projects and associated methods were developed by the student (Brodie M. Sakakibara), in consultation with a supervisory committee (William C. Miller, supervisor, Catherine L. Backman, and Janice J. Eng, committee members), and approved by the University of British Columbia?s Behavioural Research Ethics Board (certificate #s: H10-02172 and H07-02570), Vancouver Coastal Health Research Institute (certificate #: V10-02172), Providence Health Care (certificate #: H10-02172), Fraser Health Research Ethics Board (certificate #: FHREB 2010-090), Interior Health Research Ethics Board (certificate #: 2011-12-023-E), and the Institut de r?adaptation en d?ficience physique de Quebec (certificate #: 2010-216).   Versions of Chapters 2, 3, 4 and 5 will be submitted for publication. BMS, and WCM conceptualized each of the studies and developed the research design. BMS supervised data collection in British Columbia, analyzed the data, and drafted the chapters/ manuscripts. WCM supervised the projects, analyzed the data, provided feedback on and edited the chapters/manuscripts. CLB and JJE were involved in the early stages of study and concept formation, and provided feedback on and edited the manuscripts. Fran?ois Routhier supervised data collection in Quebec, provided feedback on and edited the manuscripts.              v Table of Contents  Abstract ........................................................................................................................................... ii Preface ............................................................................................................................................ iv Table of Contents ............................................................................................................................ v List of Tables ................................................................................................................................. ix List of Figures ................................................................................................................................. x Glossary ......................................................................................................................................... xi Acknowledgements ..................................................................................................................... xvii Dedication .................................................................................................................................. xviii  CHAPTER 1: Introduction ..................................................................................................... 1 1.1 The wheelchair ...................................................................................................................... 1 1.2 The wheelchair user .............................................................................................................. 2 1.3 Wheelchair use is increasing ................................................................................................. 4 1.4 Population aging and wheelchair use in Canada ................................................................... 5 1.5 Subjective quality of life of manual wheelchair users .......................................................... 6 1.6 Participation and mobility of older, community-dwelling manual wheelchair users ........... 7 1.6.1   The International Classification of Functioning, Disability, and Health ......................... 8 1.6.2   Correlates of participation in manual wheelchair users ................................................... 9 1.6.3   Correlates of mobility in manual wheelchair users ........................................................ 21 1.6.4   Predictive models of participation and mobility in manual wheelchair users ............... 25 1.7 Self-efficacy ........................................................................................................................ 25 1.7.1   Self-efficacy with using a manual wheelchair ............................................................... 26 1.8 Aging and self-efficacy ....................................................................................................... 27 1.8.1   Aging and self-efficacy with using a manual wheelchair .............................................. 27 1.9 Predictive ability of self-efficacy ........................................................................................ 28 1.9.1   Predictive ability of self-efficacy with using a manual wheelchair ............................... 29 1.10 Modifiable nature of self-efficacy .................................................................................... 30 1.10.1   Modifiable nature of self-efficacy with using a manual wheelchair ............................ 30 1.11 Research purpose .............................................................................................................. 31  	 ?	 ?	 ?	 ? vi CHAPTER 2:  Direct and mediated self-efficacy effects on participation frequency ......... 34 2.1 Introduction ......................................................................................................................... 34 2.2 Methods ............................................................................................................................... 36 2.2.1   Participants ..................................................................................................................... 36 2.2.2   Recruitment .................................................................................................................... 36 2.2.3   Outcome measures ......................................................................................................... 37 2.2.4   Study protocol ................................................................................................................ 41 2.2.5   Data analyses ................................................................................................................. 42 2.2.5.1   The direct effect of self-efficacy on participation frequency .................................. 42 2.2.5.2   The mediated effect of self-efficacy on participation frequency ............................ 45 2.3 Results ................................................................................................................................. 46 2.3.1   Sample characteristics .................................................................................................... 46 2.3.2   The direct effect of self-efficacy on participation frequency ......................................... 47 2.3.3   The mediated effect of self-efficacy on participation frequency ................................... 50 2.4 Discussion ........................................................................................................................... 52 2.5 Conclusion ........................................................................................................................... 55  CHAPTER 3:  Direct and mediated self-efficacy effects on life-space mobility .................. 56 3.1 Introduction ......................................................................................................................... 56 3.2 Methods ............................................................................................................................... 58 3.2.1   Participants and recruitment .......................................................................................... 58 3.2.2   Outcome measures ......................................................................................................... 58 3.2.3   Study protocol ................................................................................................................ 60 3.2.4   Data analyses ................................................................................................................. 60 3.2.4.1   The direct effect of self-efficacy on life-space mobility ......................................... 61 3.2.4.2   The mediated effect of self-efficacy on life-space mobility ................................... 63 3.3 Results ................................................................................................................................. 65 3.3.1   The direct effect of self-efficacy on life-space mobility ................................................ 65 3.3.2   The mediated effect of self-efficacy on life-space mobility .......................................... 67 3.4 Discussion ........................................................................................................................... 69 3.5 Conclusion ........................................................................................................................... 72  CHAPTER 4: Health, personal, and environmental predictors of self-efficacy with using a manual wheelchair ....................................................................................... 73 4.1 Introduction ......................................................................................................................... 73 4.2 Methods ............................................................................................................................... 76 4.2.1   Participants and recruitment .......................................................................................... 76 4.2.2   Outcome measures ......................................................................................................... 76 4.2.3   Study protocol ................................................................................................................ 78 4.2.4   Data analyses ................................................................................................................. 78  vii 4.2.4.1   Maximum model ..................................................................................................... 78 4.2.4.2   Regression modeling ............................................................................................... 79 4.3 Results ................................................................................................................................. 80 4.3.1   Maximum model ............................................................................................................ 80 4.3.2   Regression modeling ...................................................................................................... 83 4.4 Discussion ........................................................................................................................... 85 4.5 Conclusion ........................................................................................................................... 90  CHAPTER 5:  Rasch analyses of the Wheelchair Use Confidence Scale ............................. 91 5.1 Introduction ......................................................................................................................... 91 5.1.2   The Wheelchair Use Confidence Scale .......................................................................... 91 5.1.3   Important differences between CTT and IRT ................................................................ 92 5.1.3.1   CTT sample dependency versus IRT (Rasch) invariance ....................................... 93 5.1.3.2   CTT test versus IRT item focus .............................................................................. 93 5.1.3.3   CTT ordinal versus IRT interval measurement ...................................................... 94 5.2 Methods ............................................................................................................................... 95 5.2.1   Participants ..................................................................................................................... 95 5.2.2   Data analyses ................................................................................................................. 95 5.2.2.1   Rasch Rating Scale Model ...................................................................................... 98 5.2.2.2   Response format category collapsing ..................................................................... 98 5.2.2.3   Response format functioning .................................................................................. 99 5.2.2.4   Evaluating Rasch assumptions .............................................................................. 100 5.2.2.5   Assessing item fit and item elimination ................................................................ 102 5.2.2.6   Reliability and validity .......................................................................................... 103 5.3 Results ............................................................................................................................... 104 5.3.1   Participants ................................................................................................................... 104 5.3.2   Response format functioning ....................................................................................... 104 5.3.3   Evaluating Rasch assumptions ..................................................................................... 106 5.3.4   Assessing item fit, reliability, and validity .................................................................. 107 5.4 Discussion ......................................................................................................................... 117 5.5 Conclusion ......................................................................................................................... 121  CHAPTER 6:  Discussion and future directions .................................................................. 122 6.1 Self-efficacy, participation, and mobility with using a manual wheelchair ...................... 122 6.1.2   Self-efficacy as a body function .................................................................................. 122 6.1.3   The direct effects of self-efficacy on participation frequency and life-space        mobility ...................................................................................................................... 123 6.1.4   The mediated effects of self-efficacy on participation frequency and life-space   mobility ...................................................................................................................... 124 6.1.5   The need for self-efficacy enhancing interventions ..................................................... 125  viii 6.2 Health, personal, and environmental predictors of self-efficacy ...................................... 126 6.3 Rasch analyses of the Wheelchair Use Confidence Scale ................................................. 127 6.4 Strengths and limitations ................................................................................................... 130 6.5 Future directions ................................................................................................................ 133  REFERENCES .......................................................................................................................... 137  APPENDICES ........................................................................................................................... 153 Appendix A: Variables/measures organized by the International Classification of Functioning, Disability and Health ................................................................. 154 Appendix B:    Barthel Index .................................................................................................. 158 Appendix C:    Demographic Information Form ..................................................................... 161 Appendix D:    Functional Comorbidity Index ....................................................................... 163 Appendix E:    Home and Community Environment Instrument ........................................... 164 Appendix F:    Hospital Anxiety and Depression Scale ......................................................... 166 Appendix G:    Interpersonal Support and Evaluation List ? 6 Item ....................................... 169 Appendix H:    Late Life Disability Instrument ...................................................................... 170 Appendix I:    Life-Space Assessment ................................................................................... 173 Appendix J:    Mini Mental State Examination ...................................................................... 174 Appendix K:    Seating Identification Tool ............................................................................. 177 Appendix L:    Wheelchair Skills Test ? Questionnaire 4.1 ................................................... 178 Appendix M:   Wheelchair Use Confidence Scale for Manual Wheelchair Users v3.0 ......... 180 Appendix N:    Wheelchair Use Confidence Scale for Manual Wheelchair Users v2.4 ......... 186 Appendix O:    Wheelchair User Shoulder Pain Index ............................................................ 194 Appendix P:    Late Life Disability Instrument correlation matrix ......................................... 197 Appendix Q:    Life-Space Assessment correlation matrix ..................................................... 198 Appendix R:    WheelCon correlation matrix ......................................................................... 199 Appendix S:    Fit statistics of the items in the Mobility efficacy dimension ........................ 200 Appendix T: Fit statistics of the Mobility efficacy dimension after eliminating items with   misfitting outfit statistics ................................................................................ 202 Appendix U:    13-item Mobility efficacy subscale score conversion, SEM and reliability ... 203 Appendix V:    Fit statistics of the items in the Self-management efficacy dimension .......... 204 Appendix W: Fit statistics of the Self-management efficacy dimension after eliminating  items with misfitting outfit statistics .............................................................. 205 Appendix X: 8-item Self-management efficacy subscale score conversion, SEM and reliability ......................................................................................................... 206 Appendix Y:    21-item WheelCon short form score conversion, SEM and reliability .......... 207 Appendix Z:     Regression syntax ........................................................................................... 209	 ?      ix List of Tables  Table 1.1: Correlates of participation in wheelchair users ..................................................... 14 Table 1.2: Correlates of mobility in wheelchair users ........................................................... 23 Table 2.1: Descriptive statistics and correlations with/mean differences in participation frequency ............................................................................................................... 49 Table 2.2: The direct effect of self-efficacy on participation frequency ................................ 51 Table 2.3: Multiple mediator effects of self-efficacy on participation frequency ................. 51 Table 3.1: Descriptive statistics and correlations with/mean differences in life-space  mobility ................................................................................................................. 66 Table 3.2: The direct effect of self-efficacy on life-space mobility ....................................... 68 Table 3.3: The mediated effect of self-efficacy on life-space mobility ................................. 68 Table 4.1: Descriptive statistics and correlations with/mean differences in self-efficacy with using a manual wheelchair .................................................................................... 82 Table 4.2: Regression modeling to identify predictors of self-efficacy with using a manual wheelchair ............................................................................................................. 84 Table 5.1: Sample characteristics ......................................................................................... 105 Table 5.2: 11-point response format outfit statistics and average measure values .............. 105 Table 5.3: Mobility efficacy subscale items in order by difficulty and fit statistics ............ 109 Table 5.4: Self-management efficacy subscale items in order by difficulty and fit       statistics ............................................................................................................... 112 Table 5.5: WheelCon short form items in order by difficulty and fit statistics ................... 115          x List of Figures  Figure 1.1:  Study variables organized by the International Classification of Functioning, Disability and Health .............................................................................................. 9 Figure 2.1: The direct and mediated paths of self-efficacy on participation frequency .......... 46 Figure 3.1: The direct and mediated paths of self-efficacy on life-space mobility ................. 64 Figure 5.1: Analytical process used to perform Rasch analyses on the WheelCon ................ 97 Figure 5.2: Mobility efficacy items along the self-efficacy continuum ................................ 110 Figure 5.3: Self-management efficacy items along the self-efficacy continuum .................. 113 Figure 5.4: WheelCon short form items along the self-efficacy continuum ......................... 116                          	 ?	 ?	 ?	 ? xi Glossary  Activity: The execution of an act or task by an individual. Qualified as capacity, which describes what an individual can do at a given point in time (World Health Organization [WHO], 2001).  Adjusted R2: A measure of the loss of predictive power or shrinkage in regression. An indictor of how much variance in the outcome would be accounted for if the model had been derived from the population, instead of the sample (Field, 2009).  Average measure: Used to evaluate the ordering of a response format. It is calculated as the average ability of the respondents in a particular response option, averaged across all items. Average measure values advancing monotonically with higher response option is indication of appropriate response option ordering (Linacre, 2002).  Body functions: Physiological functions of body systems, including psychological functions (WHO, 2001).  Bootstrapping: A technique from which a sampling distribution of a statistic is estimated by taking repeated samples with replacement from the data set. Through the use of the distribution derived from the bootstrap samples a confidence interval, or a standard error can be determined (Fields, 2009).  Chunkwise regression: A set of predictor variables that are logically related and equally important, and entered together as a ?chunk/block? into a regression model (Kleinbaum, Kupper, Nizam & Muller, 2008).     xii Confounding effect: The condition in which a relationship of interest has a meaningful different interpretation when variables are ignored or included in the data analyses (Kleinbaum et al., 2008).  Confounding variable: A variable that has a causal effect on both the independent and dependent variables (Kleinbaum et al., 2008).  Environmental factors: The physical, and social environments in which people live and conduct their lives. Also includes assistive technologies for mobility (WHO, 2001).  Factor complexity: When an item loads strongly on two or more components/dimensions in factor analysis (e.g. Principal Components Analysis) (Norman & Streiner, 2008).  Fit statistic:  A measure of how well the data matches the Rasch model?s expectations. Can be either an outfit statistic or an infit statistic. Fit statistics follow a chi-square distribution with a range from 0 to infinity, with an expectation of 1. Values substantially greater than 1 reflect more variability than expected, and values less than 1 reflect less variability than expected (Linacre, 2009).  Hierarchical regression: A regression modeling strategy in which the researcher determines the order that variables are entered into the regression equation. The order of variable entry is typically based on theory and empirical evidence (Fields, 2009).  Hierarchically well-formulated models: The condition in which a regression model contains all of the lower order components of any interaction term in the model. If a model is not hierarchically well-formulated then tests about the highest order terms in the model are dependent upon the coding of variables in the model.  xiii Tests should be independent of the coding, and they are if the model is hierarchically well-formulated (Kleinbaum & Klein, 2010).  Infit mean square statistic:  Inlier-pattern-sensitive fit statistic is the information-weighted average of the squared residuals. Infit statistics provide information on unexpected responses to items with difficulty levels ?near? person ability levels (Linacre, 2009).  Interaction effect: The condition in which a relationship of interest is different at different levels/values of another variable (Kleinbaum et al., 2008).   International Classification of Functioning Disability and Health (ICF): A biopsychosocial model of health that provides a standard language and theoretical framework for the description of health and health-related states (WHO, 2001).  Item separation:   An indication of how well respondents are able separate or distinguish items in term of their ?difficulty? (Linacre, 2009).  Joint Maximum Likelihood Estimation (JMLE):  A procedure used to simultaneously estimate item difficulty and person ability parameters in Rasch models. The procedure begins by using temporary estimates of person ability, and then estimates item difficulty. After estimating the item difficulty, the person ability is reestimated. This process continues until there is no difference in parameter estimates between successive iterations (de Ayala, 2009).    Local independence: A person?s response to an item in a measurement tool is only due to his/her ability that is being measured (Streiner & Norman, 2008).    xiv Log-of-odds (logit): Rasch models determine the odds of a person responding to an item in a certain way given their ability. The definition of odds is the ratio between the probability of a response occurring versus the probability of it not occurring. Because odds range from 0 to infinity with an odds ratio of 1 meaning no difference, the scale is asymmetrical. To remedy the asymmetry, the natural log of the odds ratio is taken to make the ?no difference? equal to 0. This log-of-odds interval-level continuum represents the construct of interest, and is used to locate the item difficulty and person ability parameters (Streiner & Norman, 2008).   Mediation effect: The condition in which a variable exert influences on others through a mediating variable (Preacher & Hayes, 2008).  Mediating variable: A variable that is caused by an independent variable that in turn has a causal effect on the dependent variable. Explains the association between an independent and dependent variable (Frazier, Tix & Barron, 2004).  Life-space mobility: Movement extending from within one?s home to movement beyond one?s town or geographic region (Baker, Bodner & Allman, 2003).  Monotonicity: The proportion of people selecting a high response option in a response format increases for those with higher ability levels (Linacre, 2002).  Multiple mediation: The condition in which some variable exert influences on others through multiple mediating variables (Preacher & Hayes, 2008).    xv Orthogonal rotation: The process in which the x- and y-axes remain perpendicular during rotation procedures used with factor analysis (e.g. principal components analysis) (Norman & Streiner, 2008).  Outfit mean square statistic:  Outlier-sensitive fit statistic is the average of the squared standardized residuals. Outfit statistics provide information on unexpected responses to items with difficulty levels ?far? from person ability levels (Linacre, 2009).  Participation: The involvement in life situations. Qualified as performance, which describes what an individual does in his/her current environment (WHO, 2001).  Personal factors: The background of an individual?s life and living, including features of the individual that are not part of a health condition or health states (WHO, 2001).  Polytomous response format: A response format with more than two response options (de Ayala, 2009).  Principal Components Analyses: A technique used to combine a set of items based on their intercorrelations to account for the maximum amount of variance in the data with the fewest mutually independent components/dimensions (Norman & Streiner, 2008).  Rasch Rating Scale Model: A 1-parameter Rasch model used with datasets derived from polytomous response format that assumes that increases in the amount of the construct needed to move up adjacent response options is the same for all items (Andrich, 1978).    xvi Shrinkage: The reduction in the magnitude of R2 upon replication. It is the loss of predictive power of a regression model if the model had been derived from the population, rather than from a sample (Fields, 2009).  Standardized fit statistics (ZSTD):  The t-standardized mean-square infit and outfit statistics that approximate a normal distribution. ZSTD scores outside the range of ?1.96 (or ?2.0 for convenience) indicate misfitting items (Linacre, 2009).  Self-efficacy: The belief individuals have in their ability to perform certain behaviours to achieve desired outcomes (Bandura, 1997).  Self-efficacy with using a manual wheelchair: The belief individuals have in their ability to use their wheelchair in a variety of challenging situations (Rushton, Miller, Kirby, Eng & Yip, 2011).  Unidimensionality: Relating to one dimension. In terms of measurement unidimensionality refers to all items in a measure assessing a single common construct (de Ayala, 2009).  Varimax rotation: A type of orthogonal rotation in which the signs/coordinates are reversed. Has the effect of making large loadings larger and the small loadings smaller on each component/dimension extracted, thereby increasing the interpretability of the components/dimensions (Norman & Streiner, 2008).    	 ? xvii Acknowledgements  I would first like to acknowledge all the individuals who participated in this project. Without their input and desire to help, this research would not have been possible.  Working with a supervisory committee of highly esteemed individuals facilitated the success of this work. A special thanks to Dr. Bill Miller, my thesis supervisor, for his unequivocal support, guidance, encouragement, and belief in my abilities. I am fortunate to have learned from such an incredible person. To Drs. Janice Eng and Catherine Backman, my thesis committee members, I am also thankful for their guidance and support, as well as for their fresh perspectives and insights.   There are many people who worked diligently on this research. I would like to extend my appreciation to the GF Strong Rehabilitation Research Lab coordinator, Elmira Chan, as well as the research assistants whom I was privileged to work with. I also need to thank Dr. Fran?ois Routhier from Universit? Laval, and his research team in Quebec City and Montreal for their solid contribution.  Over the course of my PhD study, I have encountered many talented graduate students and fellows whom I admire and would like to acknowledge including Krista Best, Debbie Field, Ed Giesbrecht, Bita Imam, Dr. Ben Mortenson, Dr. Jeremy Noble, Dr. Paula Rushton, Lisa Simpson, Dr. Ada Tang, and my officemate Dominik Zbogar.   Finally, I would like to acknowledge the Canadian Institutes of Health Research for both funding this research (Operating Grant IAP-107848), and providing personal financial support (Frederick Banting and Charles Best Canada Graduate Doctoral Scholarship).       xviii Dedication    To my wife, Charity, and children, Madeleine, Blake, and Baby.  -and-  To my parents, brothers, sisters, nieces, aunts, and uncles for their unwavering support.                                     1 CHAPTER 1: Introduction   1.1 The wheelchair  The wheelchair is one of the most important forms of assistive technology used by older adults (Mann, Llanes, Justiss, & Tomita, 2004). Wheelchairs assist individuals who have ambulation limitations to become more mobile (Routhier, Vincent, Desrosiers & Nadeau, 2003), and functionally independent (Hoenig, Landerman & Shipp, 2003a; Wee & Lysaght, 2009). For example, Wee and Lysaght (2009) report the wheelchair as one of the most influential factors supporting the performance of activities of daily living in individuals with mobility impairments. Similarly, in a study of 2,368 older, community-dwelling adults with disabilities, Hoenig, Taylor, and Sloan (2003b) demonstrate the utility of assistive technologies (including the use of wheelchairs) at reducing the amount of personal assistance needed.   The wheelchair has also been shown to facilitate social participation. In a pre- post-study evaluating the impact of wheelchair acquisition on social participation among older adults, Rousseau-Harrison et al. (2009) observed significant participation improvements after individuals simply received and started using their wheelchair. Wheelchair acquisition has also improved the social integration of individuals with disabilities in developing countries (Shore, 2008; Shore & Juillerat, 2012). In a pre- post-survey of 519 individuals in India, Chile, and Vietnam, after receiving a simple, depot style wheelchair, and 12-months of its use, participants reported significant improvements to their mobility, employment, and ability to form and maintain social relationships (Shore & Juillerat, 2012). Individuals in this same study also reported less illness, hospitalization, and pain, as well as significantly improved overall health and mood ratings following the 12-months of wheelchair use.   Although wheelchairs are shown to support the functioning, health, and well-being of individuals with disabilities, the benefits are largely a function of the individuals? independence with using the wheelchair. This notion is observed in a study of individuals with a spinal cord injury (Krause, Carter & Brotherton, 2009). Individuals who were independent using their wheelchair reported better participation, health, and well-being outcomes than those who were dependent on  2 others (Krause et al. 2009). More specifically, those independent wheelchair users reported more frequent excursions outside of the home, overnight trips, and daily hours out of bed. They also reported fewer health problems and depressive symptoms, better social engagement, and adjustment scores, as well as rated their overall health higher than individuals who required assistance with using their wheelchair (Krause et al. 2009).  When considering that anything less than independent wheelchair use can lead to compromised health and well-being outcomes, work to address issues with independence is important. This may be especially true for older adults, given estimates that 60.0% of wheelchair users between the ages of 65 and 84, and 76.0% of users 85 years of age and older are not independent with their wheelchair use and require assistance (Shields, 2004; Ganesh et al., 2007). Therefore, as the population ages it is likely that there will not only be an increase in the number of older wheelchair users, but also individuals who lack independence using their wheelchair.    1.2 The wheelchair user  Old age and sex are predisposing characteristics associated with wheelchair use. In Canada, wheelchair users comprised 0.5% of the population aged 45 to 65 in 2000, 1.2% of individuals between the ages of 65 and 74, and 7.20% of the population 85 years of age and older (Shields, 2004). More females (3.1%) reported using a wheelchair than males (2.3%), which reflects the higher proportion of older females than older males in the Canadian population in general (Shields, 2004). Furthermore, Best and Miller (2011) report 62.00% of Canadian wheelchair users 60 years of age and older are female.  A study from the United States reports similar findings, in that only 0.8% of persons between the ages of 18 and 64 reported using a wheelchair, whereas 5.2% of individuals 65 years and older, and 12.3% of persons aged 85 and over reported wheelchair use (LaPlante & Kaye, 2010). Like Canadians, more females (1.8%) were reported to use wheelchairs than males (1.2%) in the United States. Old age and being female are also reported as predisposing characteristics of wheelchair use in a French study (Vignier, Ravaud, Winance, Leoutre & Ville, 2008).    3 Income (Shields, 2004; LaPlante & Kaye, 2010; Best & Miller, 2011) and education (LaPlante & Kaye, 2010; Best & Miller, 2011) are socioeconomic factors associated with wheelchair use. That is, individuals with lower incomes and/or education levels are more likely to report using a wheelchair than individuals with upper-middle to high-incomes, and/or higher educational attainments. In a Canadian study of individuals 60 years of age and older, fewer wheelchair users (46.7%) reported to have graduated from high school than ambulatory individuals (51.7%), and more wheelchair users (81.4%) reported an income of less than $14,999 than their ambulatory counterparts (8.3%) (Best & Miller, 2011). Interestingly, however, in the research of mobility-related assistive technologies in general (e.g. canes, walkers, wheelchairs), authors have observed positive associations between higher incomes and purchases of the technologies (Mathieson, Kronenfeld & Keith, 2002), as well as education and use of mobility technologies, after controlling for health status (Agree, Freedman & Sengupta, 2004). It is suggested that higher incomes facilitate the purchase of assistive technology, and that poor health resulting from lower-socioeconomic status is responsible for assistive technology use, not the individual indicators of socioeconomic status themselves (Clarke, Chan, Lina Santaguida & Colantonio, 2009).  Low health status and high numbers of impairments are commonly reported by wheelchair users (Vignier et al. 2008; LaPlante & Kaye, 2010; Best & Miller, 2011). In Canada, 100% of wheelchair users, 60 years of age and older are reported to have a chronic condition, versus 88.0% of ambulatory individuals. In addition, 74.2% of these wheelchair users are reported to be in poor/fair health (compared to good/very good/excellent), whereas 18.1% of those individuals who do not use of assistive devices for mobility are reported to be in poor/fair health (Best & Miller, 2011). In the United States, 75.0% of community-dwelling wheelchair users reported to be in fair or poor health, compared to 14.0% of the general population (LaPlante & Kaye, 2010). Similarly, wheelchair users in France reported a mean of 2.7 impairments (e.g. motor, visceral or metabolic, or cognitive), whereas non-wheelchair users reported a mean of 0.7 impairments (Vignier et al., 2008). Not surprising, nearly all individuals (96.0%) attribute their wheelchair use to difficulties with ambulation (Shields, 2004; LaPlante & Kaye, 2010).    4 In Canada, wheelchair use reported by the youngest age group (i.e. 12 to 44 years) is mostly due to injury (Shields, 2004). However, with aging the primary reasons for wheelchair use shifts from injury-related, to disease/illness, to natural aging among individuals in the oldest age category (i.e. ?85 years) (Shields, 2004). In the United States, the leading conditions associated with wheelchair use are stroke (11.1%), arthritis (10.4%), multiple sclerosis (5.0%), lower extremity amputation (3.7%), and spinal cord injury (3.6%) (Kaye, Kang & LaPlante, 2000).   1.3 Wheelchair use is increasing   The number of wheelchair users is increasing around the world. In the United States, there was a 112% increase in wheelchair usage among community-dwelling individuals between 1990 and 2005, and a 91.0% increase among individuals 65 years of age and older during the same time period (LaPlante & Kaye, 2010). In 1990, the percent of the older adult population using a wheelchair was below 3.0%. In 2005, the percentage grew to 5.1%, and accounted for 1.8 million older wheelchair users (LaPlante & Kaye, 2010).  Reports from England indicate an almost 100% increase in wheelchair use between 1986 (360,000 users) and 1996 (710,000 users) (Sapey, Stewart & Donaldson, 2005). A more recent government report from England suggests further growth, with an estimate of 1.2 million wheelchair users in 2000 (Sedgwick, Frank, Kemp & Gage, 2005).   A French study estimated the prevalence of community-dwelling wheelchair users to be 35 individuals per 10,000 inhabitants, or 206,000 wheelchair users in 1999 (Vignier et al. 2008). The overall prevalence of both community and nursing home wheelchair users was estimated at 62/10,000, or 361,500 individuals, which represents a more than doubling in wheelchair usage relative to 1991 estimates that were reported to range between 105,000 and 160,000 (Vignier et al. 2008).   Although these reports are from the late 1990 and early to mid-2000s, there was consensus among the studies? authors that the number of wheelchair users will continue increasing over time for several reasons. Population aging is the most likely reason for increasing numbers of  5 wheelchair users. It is well documented that older adults are the largest consumers of wheelchairs. In fact, reports indicate that older adults are more than four times as likely to use a wheelchair than younger adults (Shields, 2004; Clarke & Colantonio, 2005). A key reason for this occurrence may be due to ambulation difficulties, as suggested by LaPlante and Kaye (2010), and the fact that both the prevalence and severity of such difficulties increase with aging (Iezzoni, McCarthy, Davis & Seibens, 2001).   Population aging, however, does not explain the increasing proportion of older adults who use wheelchairs observed in the United States (i.e. 2.1% increase between 1990 and 2005 as discussed above) (LaPlante & Kaye, 2010). It is speculated that unmet needs for wheelchairs may be declining due to improved financing of wheelchairs, and improved standards of living (LaPlante & Kaye, 2010). That is, individuals who required a wheelchair but previously could not afford one are now able to purchase a wheelchair. Other considerations include more accessible community environments, as well as a changing attitude towards individuals with disabilities (Sapey et al, 2005; LaPlante & Kaye, 2010). It may be that wheelchair use is becoming more socially acceptable, and that individuals are more comfortable being seen in them (Sapey et al., 2005).  1.4 Population aging and wheelchair use in Canada  There are no Canadian time-related wheelchair use trends available. However, given the above findings from other industrialized countries, it is plausible that both the number and proportion of wheelchair users are increasing for similar reasons.  In Canada, the number of adults aged 65 years and older is projected to increase from 4.2 to 9.8 million between 2005 and 2036 (Statistics Canada, 2006). By 2056, an estimated 11.5 million older adults will represent 27.2% of the Canadian population (Statistics Canada, 2006), and 93.0% will live in the community (Statistics Canada, 2006).   In 1995-1996, there were an estimated 88,300 older, community-dwelling wheelchair users in Canada, representing 4.6% of all older adults in the community (Clarke & Colantonio, 2005). In  6 applying this estimate of older wheelchair users to the population aging projections noted above, there would have been 179,676 community-dwelling older wheelchair users in 2005 (i.e. 4.2 million older adults x 93.0% who are community living x 4.6% who will use a wheelchair), and will be 419,244 wheelchair users in 2036, and 491,970 in 2056, many of whom will be manual wheelchair usersi (Kaye et al., 2000). Despite these approximations being unsubstantiated, they seem appropriate, and perhaps conservative, when considering the 2005 point-estimate from the United States (i.e. 5.1% = 1.8 million older, community-dwelling wheelchair users), the demographic profile of wheelchair users (e.g. older, female, ambulatory difficulties), along with the time-related wheelchair use trends discussed above.   1.5 Subjective quality of life of manual wheelchair users  Although manual wheelchairs have been shown to support the well-being of individuals with mobility limitations, studies also demonstrate wheelchair users reporting low subjective quality of life (e.g. satisfaction with life) relative to non-wheelchair users (Patrick, Kinne, Engelberg & Pearlman, 2000; Riggins, Kankipati, Oyster, Cooper & Boninger, 2011). For example, in a study comparing the subjective quality of life of wheelchair users, non-wheelchair users, and individuals with and without chronic conditions, the wheelchair user group reported the lowest quality of life scores using the Perceived Quality of Life Scale (Patrick et al., 2000). The low scores were reported despite the group of wheelchair users reporting a better self-rated health status than individuals with chronic conditions, terminal cancer, AIDS, stroke, and those living in nursing homes (Patrick et al., 2000). Similarly, in a retrospective study of 1,826 individuals one-year post spinal cord injury, Riggins et al. (2011) investigated the changes in subjective quality of life, using the Satisfaction with Life Scale, due to changes in mobility status (Riggins et al., 2011). The authors report that individuals who transitioned from being ambulatory to using a wheelchair had statistically significant lower scores than both individuals who transitioned from using a wheelchair to becoming ambulatory, and those who maintained their mobility status of wheelchair use or ambulation after one year (Riggins et al. 2011). Furthermore, of the 	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?i Reports from the United States (Kaye et al., 2000) indicate that there are 8 times the number of community-dwelling manual wheelchair users than there are power wheelchair users.  7 individuals who maintained their mobility status over the study duration, the ambulatory group reported better satisfaction with life scores than the wheelchair user group (Riggins et al. 2011). Evidence that wheelchair users have low subjective quality of life is concerning given the projected increase of this population over the next several years. The evidence is particularly concerning for older individuals when considering that increasing age is a risk factor for wheelchair use (Shields, 2004; LaPlante & Kaye, 2010), and that low quality of life may be most prevalent upon transitioning from walking to needing a wheelchair for mobility (Riggins et al. 2011). There are many implications, including a plethora of older, community-dwelling manual wheelchair users with low subjective quality of life. Therefore, developing a greater understanding of the key indicators of quality of life in older, wheelchair users is justified.   1.6 Participation and mobility of older, community-dwelling manual wheelchair users  The ability to participate and be mobile in daily life are important indicators of subjective quality of life (Baum, 2011; Netuveli, Wiggins, Hildon, Montgomery & Blane, 2006; Webber, Porter & Menec, 2010), and this is no different for wheelchair users (Tonack et al., 2008; McVeigh, Hitzig & Craven, 2009; Routhier et al., 2003). According to the World Health Organization (WHO), participation is the involvement in life situations, and mobility is conceptualized as an activity, which is the execution of a task or action (WHO, 2001). Because both participation and mobility have quality of life implications, they are considered to be important elements of rehabilitation (Heinemann, 2006). That there is clinical interest is particularly true with evidence indicating the potential for individuals to achieve or maintain their desired participation regardless of impairments and functional limitations (Cardol et al. 2002; Jette, Haley & Kooyoomjian, 2005), and wheelchair mobility as a basic and necessary activity that supports the ability of individuals to engage in desired situations (Routhier et al., 2003). An equally compelling reason for the clinical appeal is that individuals with less than optimal participation and mobility may be at higher risk for acquiring illnesses or negative symptoms, such as depression (Wilkie, Peat, Thomas & Croft, 2007).  Despite an understanding of the importance of participation and mobility, it is common for older, community-dwelling manual wheelchair users, to report low frequency of participation (e.g. in  8 physical activity, and social and personal role participation) (Best & Miller, 2011; Sakakibara, Routhier, Lavoie & Miller, 2013c), and issues with using their wheelchair for mobility (Shields, 2004; Ganesh et al., 2007). Moreover, because there is little research on the determining factors of participation and mobility in this population, not only are there unanswered questions pertaining to the reasons for the low participation and issues with mobility, but our knowledge on how to address the issues is lacking.   Although there is a void in our knowledge on the participation and mobility of older, community-dwelling manual wheelchair users, research evidence from other wheelchair using populations may inform our understanding. Below is an overview of the correlates of various forms of participation and mobility of wheelchair users, mostly derived from younger individuals residing in the community, and older individuals living in long-term care settings.   1.6.1   The International Classification of Functioning, Disability, and Health  The WHO?s International Classification of Functioning, Disability, and Health (ICF) is an internationally recognized multifaceted biopsychosocial model of health (WHO, 2001). It is comprised of three interrelated functioning and disability domains, including those at the body (body functions and structures), person (activity), and societal (participation) levels. Whereas body functions are defined as physiological functions of body systems that include psychological functions such as confidence, optimism, and motivation, activity is defined as the execution of a task or action, and participation is the involvement in life situations (WHO, 2001). Activity is further qualified as an individual?s ability to execute a task or action, and participation is qualified as performance, which describes what an individual does in his/her environment (WHO, 2001).   The ICF emphasizes functioning and disability as a dynamic and complex interaction between health conditions, and environmental (i.e. the physical, social and attitudinal environments in which people live and conduct their lives), and personal (i.e. the background of an individual?s life, including features of the individual that are not part of a health condition) contextual factors (WHO, 2001).  The ICF framework is used to organize the existing evidence of participation and  9 mobility correlates. Figure 1.1 presents the ICF framework along with the variables identified below and those assessed in this research. Tables 1.1 and 1.2 provide an overview of the participation and mobility research studies discussed below.  Figure 1.1:  Study variables organized by the International Classification of Functioning, Disability and Health framework 	 ?                 1.6.2   Correlates of participation in manual wheelchair users  Health conditions: There is some evidence relating the health condition to various forms of participation in wheelchair users. Both qualitative and quantitative evidence suggests that increasing numbers of comorbid conditions results in lowered social participation (?=-0.14, p?0.10) (Hoenig et al., 2003a; Barker, Reid & Cott, 2004). Similarly, in studies of individuals with spinal cord injuries (SCI) who use wheelchairs, injury severity is reported as a correlate of leisure time physical activity (Martin-Ginis et al., 2010; Martin-Ginis et al., 2012). Those with Source: WHO, 2001  10 incomplete injuries experience more leisure-time physical activity participation than individuals with complete injuries (?=0.54, p?0.01) (Martin-Ginis et al., 2012).  Body function: Individuals who use manual wheelchairs are subject to shoulder pain due to overuse injuries from wheelchair propulsion (Jain et al., 2010). In individuals with a SCI, the prevalence of shoulder pain due to manual wheelchair use is estimated to range between 38.0% and 67.0% (Jain, Higgins, Katz & Garshick, 2010). Lower rates of social participation are attributed to chronic shoulder pain, as demonstrated by Kemp et al. (2011) in their study of younger manual wheelchair users with a SCI. In this study, the authors observed that reductions in shoulder pain after a 12-week intervention were related to statistically significant increases in social participation (Kemp et al., 2011).   Depression and cognition in older wheelchair users living in long-term care settings have also been reported to be associated with participation frequency in social and personal roles (Mortenson, Miller, Backman & Oliffe, 2012). Through the use of path analyses, Mortenson et al. (2012) determined a negative association between depressive symptoms (?=-0.23, p?0.001), and participation frequency, and a positive association with cognition (?=0.29, p?0.001).   Contrary to the above findings, in a study of individuals with a SCI that examined predictors of leisure-time physical activity participation, neither depression (?=-0.03) nor pain (?=-0.02) reached significance in the multiple regression model (Martin-Ginis et al., 2012). Two reasons are likely for the pain discrepancy. First, Kemp et al. (2011) only studied individuals with chronic shoulder pain, whereas shoulder pain was not an inclusion criteria in the study by Martin-Ginis et al. (2012), and second, Kemp et al. (2011) evaluated pain specific to the shoulder, and Martin-Ginis et al. (2012) used the SF-36 pain subscale that assesses overall pain. The discrepancy regarding the importance of depression again may have to do with measurement. Whereas Martin-Ginis et al. (2012) administered the Patient Health Questionnaire-9, which is typically used to screen for major depressive disorder, Mortenson et al. (2012) used a questionnaire to account for symptom severity. Nonetheless, it is plausible that associations exist between pain, depression and participation in older, community-dwelling manual wheelchair users, however, this has yet to be investigated.  11  Activity: Wheelchair mobility has been shown to be associated with participation. For example, in a qualitative study of older, community-dwelling individuals with stroke, Barker et al. (2003) revealed that limitations with propelling a wheelchair led to issues with community participation by means of decreased independence. Similarly, in a study of community-dwelling individuals who were new wheelchair users, Hoenig et al. (2003a) established that more limitations with wheelchair mobility (e.g. push the wheelchair, mobility inside the home) were associated with fewer excursions outside the home (Odds ratio=0.71, p<0.05), and in the study by Mortenson et al. (2012), wheelchair mobility, conceptualized as life-space diameter, was a positive predictor of participation frequency in social and personal roles (?=0.20, p?0.001). In another study using the life-space diameter as a measure of mobility, younger community-dwelling individuals with greater life-space diameters reported significantly more community destinations successfully reached than individuals with smaller life-space diameters (p?0.05) (Meyers, Anderson, Miller, Shipp & Hoenig, 2002). Furthermore, objective measures of wheelchair mobility such as average speed travelled have also been shown to have positive correlations with community participation (Spearman correlation (rs)=0.84, p=0.019) and socialization (rs=0.77, p=0.042) (Cooper, Ferretti, Oyster, Kelleher & Cooper, 2011).  The ability to use a wheelchair is consistently reported as an important predictor of participation, as are functional abilities to perform activities of daily living. In two studies of younger individuals with SCI, wheelchair skill was reported as a statistically significant factor related to social (?=0.45, p?0.01) (Kilkens, Post, Dallmeijer, van Asbeck & van der Woude, 2005) and community (Hosseini, Oyster, Kirby, Harrington & Boninger, 2012) participation. In the study by Mortenson et al. (2012), wheelchair skill was observed to have both a direct effect (?=0.18, p?0.01) on frequency of social and personal participation, and a mediated effect through the mobility variable. In another study of individuals with SCI examining predictors of social participation using path analyses, van Leeuwen et al. (2012) demonstrated the importance of functional abilities, measured using the motor score of the Functional Independence Measure, on social participation measured using the social dimension of the Sickness Impact Profile-68 (?=-0.43, p?0.05; i.e. higher scores reflect poorer participation).    12 Environmental factors: The wheelchair, physical, and social environments have been shown to correlate with the participation of various wheelchair using populations, and may similarly be associated with the participation of older, community-dwelling manual wheelchair users. Despite the positive benefits associated with wheelchair use, the wheelchair itself is often considered an environmental barrier to participation (Barker et al., 2003; Chaves et al., 2003; Best & Miller, 2011). For example, in a study of younger individuals with a SCI, the wheelchair was the most frequently reported factor limiting participation in the community (Chaves et al., 2003), with specific issues related to wheelchair seating and fit. Fit issues, however, were not evident in another study of older individuals in long-term care (Mortenson et al., 2012). The reason for this discrepancy is likely due to the age and activity differences between the samples. The sample of younger community-dwelling individuals with SCIs likely engaged in activities more often than those older wheelchair users in long-term care, and thus may have been more aware of issues with their wheelchair seating and fit.  The physical environment is commonly reported to be associated with participation among wheelchair users (Barker et al., 2003; Hoenig et al., 2003a; Laliberte-Rudman, Hebert & Reid, 2006; Meyers et al., 2002; Mortenson et al., 2012; Rosenberg, Huang, Simonovich & Belza, 2013; Wee & Lysaght, 2009). For example, Hoenig et al. (2003a) reported more home and community barriers to negatively correlate with the community participation among new wheelchair users (?=-0.32, p?0.01). Similarly, qualitative evidence indicates that access issues in the community limits participation among individuals with stroke (Barker et al., 2003; Laliberte-Rudman et al., 2006) and other disabilities who use wheelchairs (Rosenberg et al., 2013). On the other hand, individuals in long-term care who perceived more physical environmental barriers also reported more frequent participation (?=0.20, p?0.001) (Mortenson et al., 2012). The authors speculate that individuals with more frequent participation are more likely to encounter a greater number of physical barriers than those who have less frequent participation.   In terms of the social environment, Wee and Lysaght (2009) report that support from family and friends facilitate participation in individuals who use assistive technology (including wheelchairs) for their mobility. There is also qualitative evidence indicating negative social  13 attitudes towards people with disabilities limits community participation of older wheelchair users (Barker et al., 2003; Laliberte-Rudman et al., 2006).  Personal factors: There is less information on personal factors associated with participation, however, older age (?=-0.12, p?0.05), female sex (?=0.09, p?0.05), and having more years post-injury (?=-0.12 to -0.18, p?0.05) are reported to be associated with lower leisure-time physical activity participation (Martin-Ginis et al., 2010; Martin-Ginis et al., 2012). Warner, Basiletti and Hoenig (2010) also report associations between education, personal assistance, living status, and leisure-time physical activity participation. That is, individuals with less than high school education (?=-0.42, p?0.05), more personal assistance (?=0.80, p?0.05), and who live alone (?=-0.27, p?0.05) report lower levels of leisure-time physical activity, than individuals with more education, less personal assistance, and who live with others (Warner et al., 2010). 14 Table 1.1:  Correlates of participation in manual wheelchair users Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Barker et al., 2006 Canada Qualitative N=10 Stroke survivors; Mean age = 75.5 years;  Sex = 8 males;  Manual wheelchair users = 8 To examine the perceptions stroke survivors hold regarding their wheelchair, what it means to them, and how the wheelchair affects their lives.  In-depth interviews (community and social participation)  The wheelchair was an enabler of community participation (e.g. going to the race track, movies, religious activities, shopping).  Issues with propelling the wheelchair led to dependence issues, which limited community participation.   Environmental barriers included physical structures, access, wheelchair fit, and social attitudes.  Best & Miller, 2011 Canada Survey N=8301 General Canadian population; Mean age (wheelchair users) = 76.4 years; Sex (wheelchair users = 92 males;  Manual wheelchair users = not reported To examine the association between wheelchair use and physical and leisure activity participation in older Canadians Canadian Community Health Survey Wheelchair use was a risk factor for reduced participation in physical (OR=44.71) and leisure (OR=10.83) activity. Chaves et al. 2004 USA Cross-sectional N=70 Individuals with SCI; Mean age = 41.0 years; Sex = 55 males; Manual wheelchair users = 54 To investigate factors related to participation in activities.  1) Moving around inside the home;  2) Leaving the home;  3) Transportation  Descriptive/frequency data only. The wheelchair was most often cited as limiting participation, in each of the three areas, followed by physical impairment and environment, and then wheelchair seating, pain, fatigue, and illness.   15 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Cooper et al. 2011 USA Cross-sectional N=16   Individuals with SCI; Mean age = 49.1 years; Sex = 15 males; Manual wheelchair users = 7  To investigate the correlations between wheelchair activity and community participation. Participation Survey/ Mobility (community participation)   Data logger parameters (distance, speed, bouts drive time) Correlation between average speed traveled and community participation (rs=0.84); socialization (rs=0.77)  Hoenig et al. 2003a USA Cross-sectional N=153 New wheelchair users; Mean age = 64.8 years; Sex = 140 males;  Manual wheelchair users =  not reported To identify factors associated with participation restrictions in the community. Dependent variable: non-medical visits and medical visits  (i.e. personal role/ community participation)  # of comorbidities (?=-0.14), # of mobility limitations (?=-0.28), and environmental barriers (?=-0.32) were predictors of number of non-medical visits.  Income (?=0.44), amputation (?=-0.82), and environmental barriers (?=-0.21) were predictors of number of medical visits.  Each model accounted for 39.0% of the participation variance. Hosseini et al. 2012 USA Cross-sectional N=214   Individuals with SCI; Mean age = 38.8 years; Sex = 170 males; Manual wheelchair users = 214 To determine the association between wheelchair skills and community integration. Craig Handicap Assessment and Reporting Technique  (community participation)  Wheelchair Skills Test  After controlling for education and employment, wheelchair skills was a statistically significant predictor of community participation.   Wheelchair skills accounted for 12.0% of the participation variance.  16 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Kemp et al. 2011 USA RCT N=58 Individuals with SCI; Mean age = 45.0 years;  Sex = not reported; Manual wheelchair users = 58  To examine changes in social interaction after an exercise treatment for shoulder pain. Social Interaction Inventory (SII) (social participation)  Wheelchair User Shoulder Pain Index (WUSPI) From the baseline to the end of treatment, repeated-measures ANOVA revealed a significant interaction between WUSPI and SII scores, F(1,25)= 28.78, p<0.001.   Kilkens et al. 2005 Netherlands Cross-sectional N=81 Individuals with SCI; 1 year post discharge from inpatient rehabilitation; Mean age = 39.3 years; Sex = 56 males; Manual wheelchair users = 81  To evaluate manual wheelchair skill performance and participation in persons with SCI. 68-item Sickness Impact Profile (SIP) (community participation)  Higher scores on the SIP-68 reflect poorer participation  Wheelchair Circuit comprising three scores: ability, performance time, and physical strain) After controlling for age, gender, education level, lesion level, and motor completeness:  In Model 1: wheelchair ability increased the explained variance by 5.0%. Age and ability accounted for 26.0% of the variance. In Model 2: performance time increased explained variance by 6.0%. Age and performance time accounted for 25.0% of the variance. In Model 3: physical strain increased explained variance by 20.0%, age was the only other predictor variable, accounting for 20.0%.       17 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Laliberte Rudman et al. 2006 Canada Qualitative N=16 Individuals with stroke who were experienced wheelchair users; Mean age = 76.0 years;  Sex = 12 males; Manual wheelchair users = 16 To gain an understanding of: 1) how wheelchair use facilitated and constrained participation; and 2) how contextual factors enabled and constrained wheelchair use and participation In-depth interviews (occupation participation) Range of occupation participation narrowed after wheelchair use.  There was increased dependence resulting from difficulties using the wheelchair. Support from caregiver enabled participation in occupation.  Access and physical barriers in the community limited socialization. The wheelchair was considered an enabler and barrier to participation.  Martin-Ginis et al. 2010 Canada Cross-sectional N=695 Individuals with SCI;  Mean age = 47.1 years; Sex = 531 males; Manual wheelchair users = 389 To identify demographic and injury-related characteristics associated with leisure time physical activity participation Physical Activity Recall Assessment for people with Spinal Cord Injury (Para-SCI)  (leisure-time physical activity participation) Sex (? =0.09), age (? =-0.12), years post injury (? =-0.12), injury severity (? =-0.11--0.12), and mode of mobility (? =-0.10--0.14) each were predictors.  Total model accounted for 9.0% of the participation variance.  Martin-Ginis et al. 2011 Canada Cross-sectional N=160 Individuals with SCI;  Mean age = 47.4 years; Sex = 118 male; Manual wheelchair users = not reported To examine Social Cognitive Theory variables as predictors of leisure time physical activity. Para-SCI (leisure-time physical activity participation) Self-regulation was a statistically significant, direct predictor (? =0.72). Self-regulatory efficacy (? =0.71), and outcome expectations (? =0.36), were mediated by self-regulation.  Total model accounted for 39.0% of the participation variance.  18 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Martin-Ginis et al. 2012 Canada Cross-sectional N=695 Individuals with SCI;  Mean age = 47.0 years; Sex = 453 males; Manual wheelchair users = 309 at 18 months follow up To identify predictors of leisure time physical activity using the ICF. Para-SCI (leisure-time physical activity participation) At 18 months (n=173): Injury severity (? =0.54); Occupation (i.e. hours of paid work) (? =0.23); and years post injury (? =-0.18 were predictors of participation.  Total model accounted for 21.0% of the participation variance.     Meyers et al. 2002 USA Longitudinal (1 month) N=28 Experienced wheelchair users; Mean age = 47 years; Sex = 15 males; Manual wheelchair users = 18 To measure wheelchair users experiences of reaching and failing to reach destinations, extent of facilitators and barriers Destinations reached/unreached, barriers overcome/not overcome. (community participation)  Older individuals (>50) had fewer successful destinations reached, fewer unsuccessful destinations reached, fewer barriers overcome, more barriers not overcome, and fewer facilitators than younger individuals. Mortenson et al. 2012 Canada Cross-sectional N=264   Individuals in long-term care; Mean age: 84.0 years; Sex = 82 males; Manual wheelchair users = 240 To explore the association between wheelchair-related factors, mobility, and participation  Late Life Disability Instrument (social and personal role participation)   Cognition (? =0.29), depression (? =-0.23), mobility (? =0.20), perceive environmental barriers (? =0.20), and wheelchair skills (? =0.18) were statistically associated with participation frequency.  The total model accounted for 53.0% of the participation variance.  19 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Rosenberg et al. 2013 USA Qualitative N=35  Individuals who use assistive devices for mobility; Mean age = 67.0 years; Sex = 9 males; Manual wheelchair users = 7 To examine how built environments impact neighborhood-based physical activity. In depth interviews (physical activity participation) A variety of built environment barriers and facilitators are reported, including:  Curbcuts and condition, sidewalks, hills, lighting, ramps, crosswalks, resting places and shelter on streets, paved or smooth walking paths, safety, and traffic.   van Leeuwen et al. 2012 Netherlands Cross-sectional N=143  Individuals with SCI; Mean age=45.3 years; Sex = 102 males; Manual wheelchair users = not reported   To examine relationships between activities, participation, and mental health 68-item Sickness Impact Profile (SIP) (community participation)  Higher score on the SIP reflect poorer participation          Functional status (? =-0.43), and neuroticism (? =0.58), were statistically related to participation.  The total model accounted for 49.0% of the participation variance.  20 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Warner et al. 2010 USA Cross-sectional N=123 New wheelchair recipients; Mean age = 64.8 years; Sex = 113 males; Manual wheelchair users = not reported To examine intrinsic and extrinsic conceptual factors affecting leisure time physical activity participation Craig Handicap Assessment and Reporting Technique (community participation) For individuals aged 26?64: education (? =?0.42) and unpaid personal assistance (? =?0.80) were significant predictors.  For individuals ?65: living alone (? = ?0.27) was statistically significant.  The total models accounted for 11% (?65 group) to 34.0% (26-64 group) of the participation variance. Wee & Lysaght 2009 Canada Mixed-methods N=24 Individuals who use assistive devices for mobility; Mean age = 63.5 years; Sex = 9 males Manual wheelchair users = not reported To identify factors affecting participation Participation Scale (community participation) Most frequently reported factors to influence participation: personality, accessibility, impairment, wheelchair, transportation, income, services, family/friends.    21 1.6.3   Correlates of mobility in manual wheelchair users  In the wheelchair use literature, wheelchair mobility is often measured subjectively using life-space diameter self-report assessments (e.g. Nursing Home Life-Space Diameter, Life-Space Assessment) and/or objectively with data-loggers and accelerometers (e.g. speed, distance). The majority of studies that operationalize mobility using life-space diameter assessments are with older wheelchair users (using both manual and power wheelchairs) living in long-term care settings (Bourbonniere, Fawcett, Miller, Garden & Mortenson, 2007; Mortenson, Miller, Backman & Oliffe, 2011; Mortenson et al., 2012).   In terms of body function variables, depression is the only significant correlate of life-space mobility that has been identified (?=-0.10, p?0.05) (Mortenson et al., 2012). Both wheelchair skills and functional abilities are activity type factors reported to have significant positive associations with life-space mobility. In the studies by Mortenson et al. (2011; 2012), wheelchair skills had the strongest association with life-space mobility, after controlling for wheelchair-related and personal factor variables (?=0.62, p?0.01; ?=0.42, p?0.05).   More perceived physical barriers (?=0.10, p?0.05) (Mortenson et al., 2012), social visits from friends and family (?=0.17, p?0.05) (Mortenson et al., 2011), and lower need for a seating intervention (?=-0.30, p?0.05) (Bourbonniere et al., 2007), are environmental factor variables reported to be associated with better life-space in older wheelchair users living in long-term care settings. In one study of community-dwelling wheelchair users, Meyers et al. (2002) observed individuals with larger life-space diameters reported reaching significantly more desired destinations (p?0.001), and overcoming more barriers to reach the destinations (p?0.025), than individuals with smaller life-space diameters.  Studies have also investigated correlates of objective mobility parameters (i.e. speed, distance). Wheelchair skills (r=0.36, p?0.01), and age (r=-0.51, p?0.01) are reported to have moderated correlations with daily distance travelled in a sample of younger individuals with a SCI (Lemay, Routhier, Noreau, Phang & Martin-Ginis, 2011). Another study of younger, community-dwelling manual wheelchair users identified individuals who were employed to be faster, and travel longer  22 distances in their wheelchair than unemployed individuals (Oyster et al., 2011). In this same study, age was also observed to have a significant negative correlation with average speed travelled per day (r=-0.23, p?0.01).                  23 Table 1.2: Correlates of mobility in manual wheelchair users  Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Bourbonniere et al. 2007 Canada Cross-sectional N=99 Individuals in long-term care; Mean age = 84.3 years; Sex = 36 males; Manual wheelchair users = 56 To examine the relationships between the need for wheelchair and seating intervention and life-space mobility. Nursing Home Life-Space Diameter  Need for a seating intervention was a significant correlate of life-space mobility (? =-0.30, (p?0.05) after controlling for facility, sex, wheelchair intervention, # of comorbidities, and age.  The total model accounted for 19.0% of the life-space variance. Lemay et al. 2012 Canada Cross-sectional N=54  Individuals with SCI;  Mean age = 47.0 years; Sex = 41 males; Manual wheelchair users = 54 To examine the relationships between wheelchair skills, mobility, and injury levels Data logger parameters (distance, speed) over 7 days Distance moderately correlated with: wheelchair skills (r=0.36) and age (r=-0.51).  There were no correlations with speed. Meyers et al. 2002 USA Longitudinal (1 month) N=28 Experienced wheelchair users; Mean age = 46.7 years; Sex = 15 males; Manual wheelchair users = 18     To examine the frequency of facilitators and barriers of life-space mobility Study specific life space diameter (0-3 scale) Individuals with larger life-space diameters reached significantly more desired destinations (p?0.001) and overcame more barriers to reach the destinations (p?0.025) than individuals with smaller life-space diameters.  24 Author Year Country Study type Sample size Sample characteristics  Purpose Methods/Outcome measures Results Mortenson et al. 2011 Canada Cross-sectional N=268   Individuals in long-term care; Mean age: 84.0 years; Sex =  83 males; Manual wheelchair users = 243 To identify predictors of life-space mobility  Nursing Home Life-Space Diameter Wheelchair skills (? =0.42), use of a power chair (? =0.14), functional ability (FIM motor score) (? =0.08), and >4 weekly visits (? =0.17) were predictors of life-space mobility.   The total model accounted for 48.0% of the life-space mobility variance. Mortenson et al. 2012 Canada Cross-sectional N=264   Individuals in long-term care; Mean age: 84.0 years; Sex = 82 males; Manual wheelchair users = 240 To examine the association between wheelchair-related factors, mobility, and participation  Nursing Home Life-Space Diameter Wheelchair skills (? =0.62), depression (? =-0.10), and perceived environmental barriers (? =0.10) were associated with life-space mobility.  The total model accounted for 46.0% of the life-space mobility variance. Oyster et al. 2011 USA Cross-sectional N=132  Individuals with SCI; Mean age = 39.4 years; Sex = 106 males; Manual wheelchair users = 132 To explore the associations between demographics and wheelchair mobility Data logger parameters (distance, speed, time in wheelchair) over 14 days   Speed significantly correlated with age (r=-0.23, p=0.01).  Married persons accumulated significantly (p?0.05) more minutes per day compared to single individuals.   Employed persons traveled significantly further (p?0.001), faster (p?0.01), and for more minutes per day (p?0.01) compared to unemployed people.  25 1.6.4    Predictive models of participation and mobility in manual wheelchair users  Much of the participation and mobility research of wheelchair users has been on younger, community-dwelling individuals with SCIs, and older individuals living in long-term care settings. The research described above identifies several factors that may similarly be associated with the participation and mobility of older, community-dwelling manual wheelchair users. Knowledge of these factors is important, as is the amount of variance the factors account for in predictive participation and mobility models. Nine predictive models of various forms of participation were found (Hoenig et al., 2003a; Kilkens et al., 2006; Martin-Ginis et al., 2010, 2011, 2012; Warner et al., 2010; Hosseini et al., 2012; Mortenson et al., 2012; van Leeuwen et al., 2012) and three models of life-space mobility (Bourbonniere et al., 2007; Mortenson et al., 2011, 2012).   The amount of participation variance that has been accounted for ranges between 9.00% (Martin-Ginis et al., 2010) and 53.00% (Mortenson et al., 2012), and from 19.00% (Bourbonniere et al., 2007) to 48.00% (Mortenson et al., 2011) for the life-space mobility. Although the maximum amount of participation and mobility variance explained to date is respectable, the amount of unaccounted variance (i.e. as much as 47.00% for participation, and 52.00% for mobility) is reason to believe that current research has yet to identify and study other important and influential variables.  1.7 Self-efficacy  Self-efficacy is the belief individuals have in their ability to perform certain behaviours to achieve desired outcomes (Bandura, 1997). It is a psychological factor that for several reasons may have important implications on the participation and mobility of older, community-dwelling manual wheelchair users, however, has yet to receive adequate research attention. Self-efficacy influences choices and decisions, in addition to efforts, perseverance, and motivation (Bandura, 1997). Individuals with high perceived self-efficacy are more likely to set challenging goals, have positive outcomes, and recover more quickly after setbacks, than individuals with low self-efficacy (Bandura, 1997). They will also perceive barriers as surmountable, and expend greater  26 effort to overcome the barriers, and perseverance to reach their goals. For these reasons, self-efficacy is an important factor to consider in rehabilitation because it may influence an individual?s adherence to rehabilitation programs, goal setting, efforts and persistence. Furthermore, self-efficacy is of interest to health researchers because it has been shown to have strong predictive abilities of behaviour (Bandura, 1997). Moreover, low self-efficacy appears to be a remediable condition (Bandura, 1997). However, because self-efficacy is situation specific, and up until now there has not been an adequate tool to measure self-efficacy with using a manual wheelchair, our knowledge of its relevance and importance in wheelchair related research is at present minimal.   1.7.1   Self-efficacy with using a manual wheelchair  Some past literature exists on self-efficacy with wheelchair use (Hoenig et al., 2005; Roelands, Van Oost, Depoorter & Buysse, 2002). For example, both Hoenig et al. (2005), and Roelands et al. (2002) developed study specific self-efficacy measures pertaining to wheelchair use. Hoenig et al. (2005) developed a 4-item measure to assess self-efficacy with using a wheelchair in different environments, and Roelands et al. (2002) developed a 3-item measure to assess an individual?s self-efficacy to use assistive technology (i.e. not wheelchair specific). Although useful for their study purposes, the generalization of the studies? findings to the broader group of wheelchair users is limited, primarily due to the fact that the measures used lacked evidence supporting their reliability and validity.  More recently, two new measures of self-efficacy with using a manual wheelchair have been developed (Fleiss-Douer, van der Woude & Vanlandewijck, 2011; Rushton, Miller, Kirby, Eng & Yip, 2011). The 10-item Self-efficacy in Manual Wheeled Mobility scale (SEWM) was developed using an item pool from other self-efficacy measures (i.e. the Generalized Perceived Self-Efficacy scale, and the SCI Exercise Self-Efficacy Scale) (Fleiss-Douer et al., 2011), and validated among younger (mean age = 33 years) athletes with a SCI (Fleiss-Douer, Vanlandewijck & van der Woude, 2012). The 65-item Wheelchair Use Confidence Scale (WheelCon) was developed through interviews with wheelchair users and rehabilitation professionals, and a multi-stage Delphi and Think Aloud processes (Rushton et al., 2011). In this  27 measure, self-efficacy with using a manual wheelchair is conceptualized as the belief individuals have in their ability to use their wheelchair in a variety of challenging activities and physical environments (Rushton et al., 2011). The WheelCon assesses the strength of self-efficacy in six conceptual areas including: 1) the physical environment; 2) activities performed; 3) knowledge and problem solving; 4) advocacy; 5) social situations; 6) and emotions (Rushton et al., 2011). Because the WheelCon has recently been validated in a sample of community-dwelling manual wheelchair users from the general population (age range = 31 to 60 years) (Rushton, Miller, Kirby & Eng, 2013), there is more measurement evidence supporting its use with older individuals than the SEWM, which was developed for use with individuals with a SCI, and validated using a sample of younger elite and recreational athletes. Furthermore, the WheelCon was developed using robust methods. As a result, it appears to have greater content validity than the SEWM, thereby providing further support for the use of the WheelCon in studies of older, community-dwelling wheelchair users.  1.8 Aging and self-efficacy  Self-efficacy tends to decrease as people enter into older adulthood (Bandura, 1997). Bandura attributes this to false assessments of abilities, which act to lower the associated self-efficacy (Bandura, 1997). Although aging is related to declining capacity of body systems, individuals tend to overestimate their reduced capacity, resulting in artificially lower self-efficacy. Furthermore, Bandura argues that gains in knowledge, skills, and expertise obtained through experiences over time, compensate for physical declines due to aging, and as a result, older individuals should have higher self-efficacy estimates than what are reported (Bandura, 1997). Fortunately, because low self-efficacy is remediable condition, efforts to foster realistic perceptions of ability may help prevent the negative consequences of low perceived self-efficacy.   1.8.1   Aging and self-efficacy with using a manual wheelchair  The prevalence of lowered self-efficacy with using a manual wheelchair, estimated as a score less than 80 on the WheelCon, in a sample of volunteers, 50 years of age and older, is estimated  28 to be 39.0% (95% CI = 29.0%, 49.0%) (Miller, Sakakibara & Rushton, 2012). Moreover, 27.0% of individuals are reported to have discordant self-efficacy beliefs and wheelchair skills, with several displaying a lower self-efficacy-high wheelchair skills profile (Miller et al, 2012). When considering the projected increase in wheelchair use due to population aging, the number of individuals with lowered self-efficacy with using a manual wheelchair may increase.  1.9 Predictive ability of self-efficacy  According to Social Cognitive Theory (Bandura, 1997), self-efficacy is the most important predictor of behaviour, even more so than actual ability. In Bandura?s seminal study on self-efficacy (Bandura, 1977), adults with severe phobias were tested on both their perceived self-efficacy to perform various tasks confronting their phobia, and their actual performance. Findings revealed that subjects who successfully completed initial tasks over the course of study varied on their performance when tested on subsequent tasks (Bandura, 1977). Prior performance (i.e. successful ability to confront their phobia) therefore had limited value in predicting performance of subsequent tasks. Self-efficacy, on the other hand, predicted the individuals? performance 92.0% of the time (Bandura, 1977). That is, individuals with higher self-efficacy were more likely to successfully complete the task of confronting their phobia, than individuals with lower self-efficacy. As a result, self-efficacy was determined to be a superior predictor of behaviour than actual ability (Bandura, 1977).   In examining the application of self-efficacy to rehabilitation, various forms of self-efficacy have been used to predict different outcomes among older adults. For example, in a one-year intervention study, baseline self-efficacy related to exercise significantly predicted changes in exercise among individuals with risk of cardiovascular disease (Meland, Gunnar Maeland & Laerum, 1999). Those with higher self-efficacy at baseline were observed to increase their amount of exercise over time, more so than individuals with lower self-efficacy. Self-efficacy has also been shown to predict self-reported disability, and functional limitations among older adults (Rejeski, Miller, Foy, Messier & Rapp, 2001). In a two and a half year study of older adults with chronic knee pain, low baseline self-efficacy related to stair climbing significantly predicted higher self-reported disability related to difficulties in performing activities of daily  29 living, and less physical ability to climb stairs over time. Similarly, Seeman, Unger, McAvay and Mendes de Leon (1999) demonstrated higher self-efficacy related to instrumental activities of daily living to predict fewer perceived functional declines in a two and a half year cohort study of older adults. Furthermore, self-efficacy related to performing activities of daily living without falling, was a significant and independent predictor of balance and mobility performance, and a more important predictor of accidental falls than both balance and mobility (Pang & Eng, 2008). Finally, self-efficacy in performing daily activities without losing balance and falling were significant and independent predictors of participation (Anaby, Miller, Eng, Jarus, Noreau & PACC Research Group, 2009; Miller, Speechley, Deathe & Koval, 2001) and satisfaction with community reintegration (Pang, Eng & Miller, 2007), respectively. Given the predictive importance of various forms of self-efficacy established in several areas of health and rehabilitation, it is plausible that the construct specific to wheelchair-use may have similar implications on the participation and mobility of older, community-dwelling manual wheelchair users.   1.9.1   Predictive ability of self-efficacy with using a manual wheelchair  Preliminary evidence indicates that self-efficacy with using a manual wheelchair, measured using the WheelCon, has important implications on the frequency of participation in social and personal roles in older, wheelchair users (mean age = 59 years) (Sakakibara, Miller, Eng, Backman & Routhier, 2013a). In this study, self-efficacy was observed to be a statistically significant and positive predictor of participation after controlling for age, sex, and wheelchair skills (?=0.83, p=0.002). The R2 change associated with the self-efficacy construct was 10.0% (p=0.02) (Sakakibara et al., 2013a). Although these findings are positive, the sample size was small (n=54), which limited the number of variables entered into the regression model. More robust research with larger samples is needed to investigate the effect of self-efficacy with using a manual wheelchair on participation frequency, as well as life-space mobility.      30 1.10 Modifiable nature of self-efficacy  According to Bandura, four sources of information exist that act to modify self-efficacy: performance accomplishment, vicarious learning, verbal persuasion, and physiological/affective states (Bandura, 1997). Research has also demonstrated the utility of targeted interventions that apply the sources of information at enhancing self-efficacy. For example, in a randomized controlled trial of individuals with arthritis (Barlow, Turner & Wright, 2000), those in the intervention group who received an arthritis-related self-efficacy intervention experienced significant self-efficacy improvements relative to the control group. Randomized controlled trials in individuals with stroke (Salbach et al., 2005) and mobility-impaired (Sanford et al., 2006) populations have also shown interventions targeting balance self-efficacy, and self-efficacy related to performing activities of daily living, resulting in greater self-efficacy improvements relative to control groups. Therefore, applying targeted interventions to raise lowered self-efficacy with using a manual wheelchair may result in improved participation and mobility.   1.10.1   Modifiable nature of self-efficacy with using a manual wheelchair  In a small pilot randomized controlled trial of older, inexperienced wheelchair users (i.e. individuals with ambulatory ability who had no prior experience with using a manual wheelchair), individuals in the intervention group who received 2-hours of wheelchair skills training demonstrated statistically significant improvements to their self-efficacy, compared to the control group who received a socialization contact [F(1,17) = 10.9, p = 0.004, partial eta squared = 0.39] (Sakakibara, Miller, Souza, Nikolova & Best, 2013b). These results suggest that improvements to self-efficacy with using a manual wheelchair, measured using the WheelCon, may be accomplished clinically, with little expense and time. Research to further establish the effect of self-efficacy on important clinical rehabilitation outcomes is warranted, and is the purpose of this thesis.     31 1.11 Research purpose  Self-efficacy with using a manual wheelchair (i.e. WC self-efficacy) has recently been conceptualized as the belief individuals have in their ability to use their wheelchair in a variety of challenging activities and physical environments (Rushton et al., 2013). This construct may have important implications for older wheelchair users because a lack of self-efficacy may be a barrier to participation and mobility regardless of ability to use the wheelchair. It is a new construct that to date has received minimal research attention, largely due to the fact that up until now there has not been an adequate tool to capture relevant information. The overall purpose of this research is to develop a better understanding of the WC self-efficacy construct using the WheelCon.   Findings from this research are presented in Chapters 2 to 5. In Chapters 2 and 3, the direct and mediated effects of WC self-efficacy on participation frequency, and life-space mobility, respectively, are investigated in older, community-dwelling wheelchair users. In Chapter 4, predictors of WC self-efficacy are explored in the same population. Finally, because evidence from previous chapters supports the notion that the WC self-efficacy construct is important clinically and in research, and because the WheelCon is a recently developed measure, in Chapter 5, the WheelCon?s measurement properties are investigated using Rasch analyses in a sample of adult wheelchair users (?19 years old). The following outlines each of the four chapters:   Chapter 2:  Direct and mediated self-efficacy effects on social and personal role participation Purpose:  To quantify the association between WC self-efficacy and participation frequency in social and personal roles after controlling for important interaction effects and confounding variables, and to investigate the mediated effect through multiple variables. Hypotheses: i) WC self-efficacy is an independent predictor of participation frequency in older, community-dwelling manual wheelchair users after controlling for confounding variables; and ii) that the association is mediated by multiple functioning/ disability variables.   32 Contribution: This study illustrates the association between WC self-efficacy and participation frequency. The results may be used to inform future intervention studies, as well as other research on the participation of older, community-dwelling manual wheelchair users.  Chapter 3:  Direct and mediated self-efficacy effects on life-space mobility Purpose: To quantify the association between WC self-efficacy and life-space mobility after controlling for important interaction effects and confounding variables, and to investigate the mediated effect through wheelchair skills. Hypotheses: i) WC self-efficacy is an independent predictor of life-space mobility after controlling for confounding variables; and ii) that the association is mediated by wheelchair skills. Contribution: This study illustrates the association of WC self-efficacy on life-space mobility. The results may be used to inform future intervention studies, as well as other research on the mobility of older, community-dwelling manual wheelchair users.  Chapter 4:  Health, personal, and environmental predictors of self-efficacy with using a manual wheelchair Purpose: To investigate health-related, personal and environmental factor predictors of WC self-efficacy. Hypothesis: Health-related, personal and environmental factor variables will each independently predict WC self-efficacy. Contribution: This study identifies indicators of WC self-efficacy that clinicians and researchers may easily access and use to identify individuals at risk of lowered WC self-efficacy.   Chapter 5:  Rasch analyses of the Wheelchair Use Confidence Scale Purpose: To evaluate the measurement properties of the WheelCon using Rasch analyses, and in doing so to compare the functioning of the WheelCon?s 101-point response format to shortened response formats; examine the dimensionality of the WheelCon; identify items not conforming to the Rasch model that could be  33 considered for elimination; identify the overall item content of the WheelCon, and redundant items that could be considered for elimination; convert the raw ordinal total scores into standardized interval total scores; and determine the standard errors of measurement and reliability for the entire range of standardized scores, as well as the test?s internal consistency reliability. Hypotheses: i) That a shortened response format will function better than the original response format; and ii) that the WheelCon with a shortened response format will result in more than one dimension with fewer items; good internal consistency reliability (Cronbach alpha?0.70) throughout the range of total scores; and a good to excellent correlation magnitude (Pearson r?0.75) with the original WheelCon.  Contribution: The results of this study provide evidence in support of the reliability and validity of the WheelCon using an item response approach. The findings indicate that the WheelCon is comprised of two dimensions that assess self-efficacy related to mobility and self-management. The 21-item WheelCon short form, and the 13-item mobility efficacy and 8-item self-management subscales with a recoded 0 to 10 response scale have good reliability, and provide accurate and precise measurements of different forms of self-efficacy with wheelchair use. 	 ?	 ?	 ?            34 CHAPTER 2: Direct and mediated self-efficacy effects on participation frequency  2.1 Introduction  Participation, or the involvement in life situations (WHO, 2001), is an important focus in the rehabilitation of older adults because not only is it used as an indicator of quality of life, which is a broad concept that encompasses many aspects of health, and well-being (Dijkers, 1999), research also shows strong statistical associations between the two constructs (Ravenek, Ravenek, Hitzig & Wolfe, 2012). Mobility limitations are a leading cause of disability among community-dwelling individuals (Iezzoni et al., 2001), and are the primary reason for participation restrictions in individuals 50 years and older (Wilkie, Peat, Thomas & Croft, 2006). Older individuals with mobility limitations are often prescribed wheelchairs to overcome participation restrictions; however, these individuals commonly report low participation levels, with rates as low as 8.3% in the frequency and duration of physical activity participation, for example, compared to 48.8% reported by ambulatory individuals (Best & Miller, 2011).  There is little evidence explaining the low participation frequency in older, community-dwelling manual wheelchair users. Shields (2004) notes, however, that older individuals are more likely to lack independence with using their wheelchair than younger individuals, and LaPlante and Kaye (2010) report difficulties with wheeled mobility increases with age. Although there is a void in our knowledge on the participation of older, community-dwelling manual wheelchair users, the existing evidence from other populations of wheelchair users may inform our understanding. For example, because predictive models of participation developed with younger, community-dwelling manual wheelchair users (Hosseini et al., 2012; Kilkens et al., 2005), and older wheelchair users residing in nursing homes (Mortenson et al., 2012) indicate ability to use a wheelchair is a predictor of participation regardless of age or living arrangement, it may plausibly impact the participation of older, community-dwelling manual wheelchair users. It is also plausible that variables such as depression (Mortenson et al., 2012), life-space mobility (Mortenson et al., 2012), and various contextual factors (Kilkens et al., 2005; Martin-Ginis et al., 2010) may be important, but this is not established.    35 Although existing evidence may contribute to the development of models predicting the participation frequency of older, community-dwelling manual wheelchair users, the variables considered to date have explained between 9.0% (Martin-Ginis et al., 2010) and 53.0% (Mortenson et al., 2012) of the variance of various forms of participation. This indicates that there is much more to be investigated to enhance our knowledge about the participation of wheelchair users in order to sufficiently address areas for improvement. When considering reports that the proportion of older American wheelchair users has been increasing by 4.3% per year (LaPlante & Kaye, 2010), and evidence that old age is a risk factor for wheelchair use (Shields, 2004; LaPlante & Kaye, 2010), there is clear need for more research.  Because studies consistently report that ability to use a wheelchair is an important factor of participation, a person?s belief in their ability (i.e. self-efficacy (Bandura, 1997)), may similarly provide important explanatory value, as has been demonstrated in many areas of health. For example, self-efficacy has been shown to be an important predictor of leisure and physical activity participation in several populations, including individuals with a lower extremity amputation (Miller et al., 2001), and older adults with chronic conditions (Anaby et al., 2009). The construct, however, has yet to receive adequate investigation in wheelchair users.   Self-efficacy with using a manual wheelchair is the belief individuals have in their ability to use their wheelchair in a variety of challenging situations (Rushton et al., 2011). There is prevalence data suggesting 39.0% (95% CI=29.0, 49.0%) of older, community-dwelling individuals have lowered WC self-efficacy (defined as a score of ?80 on the WheelCon) (Miller et al., 2012). Some evidence also indicates a statistically significant positive association between self-efficacy and participation frequency in older, community-dwelling manual wheelchair users (Sakakibara et al., 2013a), and that the construct is modifiable (Sakakibara et al., 2013b). These preliminary data suggest WC self-efficacy may be of clinical interest, however, more robust research is needed.  Social Cognitive Theory postulates self-efficacy has both direct and indirect effects on behavior (Bandura, 1997), therefore, the objective of this study is to investigate the direct effect of self-efficacy on participation frequency, and test the indirect effect via multiple mediators. The  36 International Classification of Functioning Disability and Health (ICF) (WHO, 2001) framework guided our investigation of the hypotheses that: i) WC self-efficacy (conceptualized as a body function) is an independent predictors of participation frequency in older, community-dwelling manual wheelchair users after controlling for important contextual variables; and ii) that the association between WC self-efficacy and participation frequency is mediated by functioning/disability variables at the body function, person, and societal levels.   2.2 Methods  2.2.1   Participant eligibility  Community-dwelling, manual wheelchair users who were 50 years of age and older, living in either British Columbia or Quebec, Canada, were enrolled in this cross-sectional study. Participants had at least 6 months experience with using a manual wheelchair on a daily basis, and communicated in English or French. Individuals with a Mini Mental State Examination score of 23 or less (Folstein, Folstein & McHugh, 1975), and/or an acute illness were excluded from study.  A sample size of 122 was determined using G*Power to have 80.0% power to detect significance in a model with 9 independent variables using an alpha of 0.05, and a moderate effect size (?2=0.14). An effect size was calculated using the R2 increaseii reported in a previous study of wheelchair users that regressed participation frequency on self-efficacy, after controlling for wheelchair skills, age, and sex (Sakakibara et al., 2013a).   2.2.2   Recruitment  Community therapists recruited participants in British Columbia. Advertisements about the study were also posted at community and senior centers, and sent to disability advocacy groups. In 	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?ii ?2=RAB2 ?RA2/1- RAB2; where RAB2 = variance accounted for by the control variables and self-efficacy, and RA2 = variance accounted for by control variables (Cohen, 1988).   37 Quebec, participants were recruited from two local rehabilitation centers in Quebec City and Montreal. The ethics boards from the relevant institutions approved this study.  2.2.3   Outcome measures  Variables/measures were selected based on either empirical or conceptual rationale, and are organized by ICF domain in Chapter 1, figure 1.1 (page 9). The measurement properties of all measures used in this study have been evaluated with wheelchair users, older adults, and/or individuals with mobility limitations. An overview of the measures are presented in appendix A, and copies of each are found in appendices B ? O. All measures have been translated into French.  Dependent variable: Participation frequency was measured using the 16-item Late-Life Disability Instrument (LLDI) (Jette et al., 2002). In this measure, participants rate their participation frequency in social and personal roles using a response scale ranging from 1 (never) to 5 (very often). Item responses are summed to derive raw total scores, which are then standardized into scores ranging from 0 to 100 (Jette et al., 2002). Higher scores indicate more frequent participation. Scores of 51.4 and less are considered low participation frequency in older adult populations (Keysor et al., 2010). Validity testing found the LLDI total scores to differentiate between older adults assigned to four functional levels (Jette et al., 2002), and to moderately correlate with the London Handicap Scale (r=0.47) (Dubuc, Haley, Ni, Kooyoomjian & Jette, 2004). Recent evidence supports the test-retest reliability of measures derived with older, wheelchair users (ICC=0.86, 95% CI=0.76-0.93) (Sakakibara et al., 2013c).  Independent variable of interest: Self-efficacy with using a manual wheelchair was measured using the 65-item Wheelchair Use Confidence Scale (WheelCon) (Rushton et al., 2011). This measure assesses the strength of self-efficacy in six conceptual areas including maneuvering around the physical environment, performing activities, knowledge and problem solving, social situations, advocacy, and emotions. Items are rated on a 0 to 100 scale. A mean score is calculated with higher scores indicating more self-efficacy (Rushton et al. 2011). In a recent methodological study of community-dwelling manual wheelchair users 19 years of age and older  38 (median age = 50.0, interquartile range = 31.0-60.0), who were experienced with using their wheelchair (median years of experience = 13.0, interquartile range = 4.0-28.0), the internal consistency reliability (Cronbach alpha) of the WheelCon measures was 0.92, and the 1-week-retest intraclass correlation coefficient was 0.84 (95% bootstrapped CI = 0.7, 0.9) (Rushton et al., 2013). The study also provides evidence in support of the validity of the measurements through hypothesizing associations with relevant outcomes including wheelchair skills (Spearman correlation (rs)=0.52), activities of daily living (rs=0.32), depression (rs=-0.43), and life-space mobility (rs=0.38) (Rushton et al., 2013).  Potential confounding variables: According to Kleinbaum, Sullivan & Barker (2007), a confounder must be a risk factor, cannot be an intervening/mediating variable, and must be associated with the key independent variable of interest. In this study, potential confounders were considered to be health, personal, and/or environmental contextual factors.   The socio-demographic information form collected health (e.g. diagnosis), personal factor (e.g. age, sex, experience with using a wheelchair), and environmental factor variables such as geographic location that were considered to potentially confound the association between self-efficacy and participation frequency.  Data were also collected for number of comorbidities, need for a seating intervention, social support, and barriers in the home and community using the following measures:   The 18-item Functional Comorbidity Index was used to collect data on number of comorbidities (Groll, To, Bombardier & Wright, 2005). Individuals are asked if they have been diagnosed with any of 18 health conditions (yes=1, no=0). Total scores range from 0 to 18 with higher scores indicating more comorbidity.   The 11-item Seating Identification Tool was used as a measure of need for a seating intervention (Miller, Miller, Trenholm, Grant & Goodman, 2004). Five areas are assessed including: pressure; discomfort behaviours; mobility; positioning; and stability. Each item is scored using a dichotomous (yes=1, no=0) scale. After adjusting for the weighted items (items 1, 2, 4, and 10)  39 total scores range from 0 to 15 with higher scores indicating a greater need for a seating intervention. A score of 2 or more is reported to indicate a need for a seating intervention (Miller et al., 2004). In a sample of older wheelchair users, the SIT?s measures were found to have good test-retest (ICC=0.83), and inter-rater reliability (ICC=0.83) (Miller et al., 2004).   Perceived social support was measured with the 6-item Interpersonal Support and Evaluation List (Cohen & Hoberman, 1983). Total scores range from 0 to 18 with higher scores indicating more social support. The measurement properties of this measure have been established in the general population including older adults (Cohen & Hoberman, 1983), and have demonstrated construct validity with the Community Integration Measure (r=0.42) (McColl, Davies, Carlson, Johnson & Minnes, 2001) and the Sense of Support Scale (r=0.78) (Dolbier & Steinhardt, 2000).  Finally, the amount of physical environmental barriers in the home and community were gathered with the Home and Community Environment Instrument (Keysor, Jette & Haley, 2005). Scores in the 9-item home subscale range from 0 to 10, and scores in the 5-item community subscale range from 0 to 5. Higher scores indicate more barriers. Measures from both subscales have support for their reliability and validity in adults (?21 years) with mobility limitations (Keysor et al., 2005). For the items referring to ?walking areas? in the community subscale, the wording was replaced with ?areas you go to? to prevent ambiguity.  Potential mediating variables: Mediators in this study were selected on the basis of existing evidence demonstrating the variables to both be influenced by self-efficacy, and to influence participation.   The perceived participation limitations variable was quantified using the 16-item limitations dimension in the LLDI (Jette et al., 2002). For each item, individuals rate their perceived limitations with doing various activities using a 1 (completely) to 5 (not at all) response scale. Similar to the LLDI frequency dimension discussed above, item responses are summed to derive raw total scores, which are then standardized into scores ranging from 0 to 100 (Jette et al., 2002). Higher scores indicate fewer participation limitations. Measurements have been observed  40 to have good test-retest reliability in older, wheelchair users (ICC=0.87, 95% CI=0.75-0.93) (Sakakibara et al., 2013c).  The life-space mobility of wheelchair users was evaluated using the Life-Space Assessment (Baker, Bodner & Allman, 2003). This 20-item assessment captures the life-space mobility of individuals in five areas: 1) within the home; 2) around the home; 3) in the neighbourhood; 4) in town; and 5) outside of town. Subjects are asked at what frequency they moved in each of the five areas over the past four weeks, and if any assistance (from other persons (multiplier of 1) with equipment (multiplier of 1.5) or no assistance (multiplier of 2)) was used. The composite scores have excellent test-retest reliability in both community-dwelling older adults (ICC=0.96, 95% CI=0.95, 0.97) (Baker et al., 2003), and power wheelchair users (ICC=0.87, 95% CI=0.69, 0.92) (Auger et al., 2009). In terms of validity, Baker et al. (2003) reports moderate correlations (r=-0.41 to 0.60) in the expected directions with measures of physical performance, activities of daily living, and depression, in older individuals. Because all individuals used manual wheelchairs in this study, the total composite score ranged from 0 to 90, and not 0 to 120 range for individuals who do not use assistive technology.  Ability to use a manual wheelchair was assessed with the 32-item Wheelchair Skills Test ? Questionnaire (Wheelchair Skills Training Program (WSTP) manual, 2008). Individuals are asked if they are able to complete a specific skill using their wheelchair. If individuals express that they are able to complete the skill, they are then asked how they would perform the skill to ensure that they are doing it in a safe manner. Ratings include pass, fail (which includes completing a skill in an unsafe manner), or not applicable if the wheelchair does not have the component. Total percentage scores are derived by dividing the number of individual skills passed by the total number of applicable skills. Higher scores indicate more wheelchair skills. Measurements derived using the Wheelchair Skills Test-Questionnaire have demonstrated high correlations with measurements from the performance based Wheelchair Skills Test version 4.1 (rs=0.89) (Rushton, Kirby & Miller, 2012).  The 10-item Barthel Index-postal version was used as a measure of functional independence related to performing activities of daily living (Gompertz, Pound & Ebrahim, 1994). Items  41 pertain to bathing, climbing stairs, dressing, mobility, transfers, feeding, toilet use, grooming, and bowel and bladder control. Responses are scored from dependent (0) to independent (3). Summing all responses derives a total score ranging from 0 to 20. Higher scores indicate greater functional independence, and have demonstrated moderate reliability using ? statistics ranging from 0.49-0.90 for the 10 items (Gompertz et al., 1994).  Severity of depression and anxiety symptoms were assessed using the 14-item Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983). The depression and anxiety subscales are each made up of 7 items. Responses are on a 4-point scale (0 = not at all, 3 = very often indeed) and based on reflections of the previous week. Total scores for each dimension range from 0 to 21 and higher scores are indicative of more severe symptoms. A score of 8 or more is indication of severe anxiety and/or depression symptoms (Sakakibara, Miller, Orenczuk, Wolfe & SCIRE Research Team, 2009). The depression and anxiety subscale measurements have evidence in support of their reliability and validity with spinal cord, psychiatric, primary care, and general populations (Bjelland, Dahl, Tangen Haug & Neckelmann, 2002; Sakakibara et al., 2010).  The 15-item Wheelchair User Shoulder Pain Index was used to measure the intensity of shoulder pain associated with doing functional activities (e.g. transfers, propulsion, self-care, and general activities) (Curtis et al., 1995a). Each item is scored using a visual analog scale, with a minimum score of 0 (no pain) and a maximum score of 10 (worst pain ever experienced). Summing the responses to each item derive total scores ranging from 0 to 150, with higher scores indicating more intense pain. Measurements from the index have demonstrated high test-retest reliability (ICC=0.99) for use with experienced wheelchair users, and validity testing found significant negative correlations with shoulder abduction (r=-0.49), flexion (r=-0.48), and extension (r=-0.30) (Curtis et al., 1995b).   2.2.4   Study protocol  After completing the socio-demographic information form, the Mini Mental State Examination, and the Wheelchair Use Confidence Scale, participants were administered the remaining  42 measures in a random sequence to minimize response bias. All data were collected during face-to-face interviews.  2.2.5   Data analyses  Data from British Columbia and Quebec were combined for analyses because the mean difference in the dependent variable was less than measurement error (i.e. less than the LLDI?s 95% minimal detectable change of 7.18 calculated with a sample of older wheelchair users (Sakakibara et al., 2013c)), thereby indicating no difference in participation frequency. Descriptive statistics were used to characterize the sample. Results from categorical variables were calculated as percentages, and from continuous variables as means and standard deviations. Income was collapsed into three categories using $30,000 as the cutpoint to indicate a low-income household or not (based on 2 person household in 2012) (Statistics Canada, 2013), in addition to the prefer not to answer category. The following variables were dichotomized, and coded as -0.5 (no) or 0.5 (yes) (Norman & Streiner, 2008): diagnosis (neurological condition); education (high school graduate); formal wheelchair skills training; require assistance with the wheelchair (e.g. supervision, wheelchair set-up, transfers); married or common law; and employed and/or volunteer. Regression modeling was used to establish the direct and mediated self-efficacy effects on participation frequency.  2.2.5.1   The direct effect of self-efficacy on participation frequency  To establish a valid and precise estimate of the direct effect, Kleinbaum?s 3-stage modeling strategy was followed (Kleinbaum & Klein, 2010). Validity-based regression strategies are typically used for research questions inquiring about the importance or association of one predictor on an outcome while accounting for additional variables that may impact the association (i.e. confounders). In using this type of strategy, after specifying variables to be included in the model, interaction terms of interest are evaluated first, followed by an assessment of confounding and precision (Kleinbaum & Klein, 2010).    43 Stage 1: Variable specification In the first stage, potential confounding variables and interaction terms were specified for modeling. To limit the number of variables for entry, only those continuous variables with a fair relationship (r?0.25) (Portney & Watkins, 2009) with participation frequency, and/or those categorical variables with a mean difference in participation frequency that exceeded the LLDI?s 95% minimal detectable change (Sakakibara et al., 2013c) were included in the model. To minimize collinearity all continuous variables were mean centered. However, when potential collinearity was identified (i.e. r?0.70 between independent variables, and with a variation inflation factor value greater than 10 (Kleinbaum, Kupper, Nizam & Muller, 2008), the measure with the highest correlation with the dependent variable was selected, unless there was theoretical rationale to choose one variable over another, or to retain both. Scatterplots of the bivariate data were examined for potential outliers. Data points greater than 1.5 times the variable?s interquartile range were considered cases that could influence the correlation (Portney & Watkins, 2009). These cases were removed from this modeling stage only, and the data reanalyzed. Finally, two interaction terms were included to determine if the relationship between self-efficacy and participation frequency differs by sex, and age, which is in accordance with Social Cognitive Theory (Bandura, 1997) and existing evidence (Sakakibara et al., 2013a).   Regression assumptions: After specifying variables and interaction terms for inclusion, the residuals were examined to detect outliers and violations of regression assumptions. Cases with standardized residual values exceeding ?3 were potential outliers, and excluded from analyses if any Cook?s distance values exceeded 1 (Kleinbaum et al., 2008). In addition to checking for collinearity, as described above, the independence assumption was tested using the Durbin-Watson statistic. A value greater than 2 was deemed problematic (Kleinbaum et al., 2008). The residual plots were also inspected to assess for normality, and homoscedasticity. Data were considered homoscedastic if the width of the plot was around zero for all values of the dependent variable (Kleinbaum et al., 2008).   Stage 2: Interaction assessment The second stage of the modeling strategy evaluated the importance of the interaction terms (Kleinbaum & Klein, 2010). An interaction is said to occur when the relationship of interest is  44 different at different levels of another variable (Kleinbaum et al., 2008). After forcing the self-efficacy variable into the model, the interaction terms were entered in addition to their lower order components to keep the model hierarchically well-formulated (Kleinbaum & Klein, 2010). Forward selection and backward elimination regression approaches were then used to identify and remove non-significant interaction terms (Kleinbaum & Klein, 2010). The result was then considered the crude model for the next modeling stage.  Stage 3: Confounding and precision assessment Confounding was assessed for in the final stage of model development. Confounding refers to the relationship of interest having a meaningful different interpretation when variables are ignored or included in the model (Kleinbaum et al., 2008). The assessment of confounding requires a comparison of the self-efficacy estimate in the crude model with the estimate in the adjusted model (i.e., the crude model plus the addition of the potential confounding variables). According to Kleinbaum et al. (2008), confounding is present when the beta coefficient of the independent variable of interest is meaningfully different when the potential confounding variables are entered into the model. The assessment of confounding therefore is subjective in terms of defining a meaningful difference. In this study, a change in the beta coefficient that exceeded the Wheelchair Use Confidence Scale?s measurement error (i.e., minimal detectable change ?16.40% that was derived with community-dwelling manual wheelchair users (Rushton et al., 2013)) was considered to be indicative of confounding.   If the adjusted model indicated confounding, subsequent analyses were performed to identify possible subsets of the confounding variables that provide equivalent control (Kleinbaum & Klein, 2010). Reducing the number of confounding variables is beneficial for two reasons: 1) developing a parsimonious model; and 2) improving precision in estimating the relationship of interest. Precision refers to the size of the estimator?s variance; the smaller the variance, the higher the precision (Kleinbaum et al., 2008). The precision of each confounding subset was evaluated by examining the width of the 95% confidence interval around the self-efficacy estimate. A narrowing of the confidence interval indicated improved precision (Kleinbaum et al., 2008). The model with equivalent control of confounding, relative to the adjusted model, and the  45 greatest precision was deemed to provide the most valid and precise estimate of the direct effect of self-efficacy on participation frequency. This model was then used in the mediator analyses.  2.2.5.2   The mediated effect of self-efficacy on participation frequency  Because multiple variables were hypothesized as mediators, a single multiple mediation model was tested (Preacher & Hayes, 2008) in lieu of separate simple models. Path c in figure 2.1a depicts the direct effect of self-efficacy on participation frequency. Figure 2.1b represents the mediate effects via the 7 possible mediators. Mediators had to have at least a fair correlation magnitude (i.e. r?0.25) with participation frequency to be included in the analyses. Specific mediated effects are defined as the product of the two unstandardized paths (i.e. aibi, i?7) linking self-efficacy to participation (Preacher & Hayes, 2008). The total mediated effect is the sum of the specific effects. A bias corrected bootstrapping method was used to derive the point estimates and 95% confidence intervals (Preacher & Hayes, 2008). The proportion of the direct effect accounted for by the mediators was calculated as ?i(aibi)/c.  SPSS version 19.0 (SPSS Inc., Chicago, IL), G*Power version 3.1.3 (G*Power, http://www.psycho.uni-duesseldorf.de/abteilungen/aap/gpower3/), and the INDIRECT macro (Preacher & Hayes, 2008) were used for the analyses.                    46 Figure 2.1:  The direct and mediated paths of self-efficacy on participation frequency    2.3 Results  2.3.1   Sample characteristics  Seventy-four individuals from British Columbia and 50 individuals from Quebec were enrolled. The mean age of the total sample was 59.67 (SD=7.49), and 74 (59.7%) were male. Fifty-nine (47.6%) were married or in a common-law relationship, 46 (37.1%) were either employed or do volunteer work, and 110 (89.4%) were high school graduates.   Body	 ?Function:	 ?Self-??efficacy	 ? Participation:	 ?Frequency	 ?Health,	 ?Personal	 ?and	 ?Environment	 ?factors:	 ?Confounding	 ?variables	 ?c	 ?Figure	 ?2.1a:	 ?Direct	 ?effect	 ?Body	 ?Function:	 ?Depression,	 ?Anxiety,	 ?Shoulder	 ?Pain	 ?Participation:	 ?Limitations	 ?Activity:	 ?Life-??space	 ?mobility,	 ?Wheelchair	 ?skills,	 ?Functional	 ?independence	 ?Body	 ?Function:	 ?Self-??efficacy	 ? Participation:	 ?Frequency	 ?Health,	 ?Personal	 ?and	 ?Environment	 ?factors:	 ?Confounding	 ?variables	 ?ai	 ?Figure	 ?2.1b:	 ?Mediated	 ?effect	 ?bi	 ? 47 The majority of the participants reported having a neurological condition (78.2%), with just under half reporting a spinal cord injury (48.4%). Sixteen (12.9%) individuals reported having multiple sclerosis, 12 (9.7%) had a stroke, and others reported Parkinson?s disease, cerebral palsy, lower extremity amputation, post-polio, and arthritis. Participants had a mean of 2.69 comorbidities (SD=2.40) out of 18 possible comorbidities asked about, and low severity of depressive and anxiety symptoms. The sample?s mean Barthel Index score was 14.37 (SD=2.79) out of 20, which is indication of moderate dependence to perform activities of daily living. The sample reported very low levels of shoulder pain on the Wheelchair User Shoulder Pain Index (mean=19.12, SD=28.21, out of 150).  Individuals had a mean 22.31 years (SD=16.05) experience with using a wheelchair, used their wheelchair for a mean 12.30 hours (SD=4.29) a day, and reported few physical barriers in the home and community. Thirty-nine (31.5%) individuals required some form of assistance with using their wheelchair (e.g. mobility, transferring, set-up), and 22 (17.7%) received training to use their wheelchair outside of rehabilitation. The sample?s mean Seating Identification Tool score was 1.98 (SD=1.69), which is just below the critical value of 2 indicative of a need for a seating intervention (Miller et al. 2004).   The mean Wheelchair Skills Test-Questionnaire score was 75.49 (SD=14.89), and the mean Life-Space Assessment score in this sample was 46.99 (SD=17.84). This sample had a mean self-efficacy with using a manual wheelchair score of 78.38 (SD=19.19) out of 100, and a mean participation frequency score of 50.66 (SD=7.85), which is considered low (Keysor et al. 2010). Descriptive statistics of the sample are presented in table 2.1.  2.3.2   The direct effect of self-efficacy on participation frequency  Stage 1: Variable specification Pearson correlation coefficients between the continuous and dependent variables are shown in table 2.1, along with the mean change scores in the dependent variable for the dichotomized variables. The correlation matrix between all continuous variables is presented in appendix P. Variables initially specified for inclusion into the regression model included age, number of  48 comorbidities, perceived social support, and the age and sex interaction terms. The variables specified for inclusion based on the magnitude of their bivariate correlation with the dependent variable did not change after rerunning the analyses without outliers. All model assumptions were met.                              49 Table 2.1:  Descriptive statistics and correlations with/mean differences in participation frequency  Variable Total Participation frequency  mean?sd/frequency (%) r/(mean difference) Participation: Frequency (0-100) Limitations (0-100)  50.66?7.85 63.68?11.71  1 0.54* Activity: Life-space mobility (0-120) Wheelchair skills (0-100) Functional independence (0-20)  46.99?17.84 75.49?14.89 14.37?2.79  0.55* 0.51* 0.22 Body functions: Self-efficacy (0-100) Depression (0-21) Anxiety (0-21) Pain (0-150)  78.38?19.19 3.79?3.13 5.09?3.87 19.12?28.21  0.54* -0.24 -0.10 -0.03 Health condition: Comorbidities (0-18) Neurological condition: Spinal cord injury Multiple sclerosis Stroke Other (e.g. Cerebral palsy) Non-neurological condition: Amputation Polio Arthritis  Other    2.69?2.40 97 (78.20) 60 (48.40) 16 (12.90) 12 (9.70) 9 (9.30) 27 (21.80) 9 (9.30) 5 (4.03) 4 (3.23) 9 (9.30)  -0.31* (1.70) Personal factors: Age Male Education (high school graduate) Income:?  <$30,000 Prefer not to answer Married (yes) Employed/volunteer (yes) Wheelchair Years experience Daily use (hours) Formal training (yes) Wheelchair assistance (yes)  59.67?7.49 74 (59.68) 110 (89.40 )  43 (34.68)  21 (16.94) 59 (47.60) 46 (37.10)  22.31?16.05 12.30?4.29 22 (17.70) 39 (31.50)  -0.28* (1.71) (-0.70)   (-1.91) (2.78) (0.36) (-2.94)  0.10 0.17 (-2.09) (4.81) Environmental factors: Wheelchair Need for seating intervention (0-15) Social Social support (0-18) Physical Home barriers (0-10) Community barriers (0-5)   1.98?1.69  14.48?3.71  1.10?1.22 1.06?0.85   -0.13  0.39*  -0.05 -0.13 *included for modeling; ?mean difference from ?30,000; n=124 50 Stage 2: Interaction assessment After forcing the self-efficacy variable into the model, neither the age nor sex interaction term reached statistical significance. The crude model (table 2.2) therefore included the self-efficacy variable, and accounted for 29.1% of the participation frequency variance.  Stage 3: Confounding and precision assessment In the adjusted model (table 2.2), the self-efficacy estimate was confounded by 18.6% after controlling for age, number of comorbidities, and perceived social support. This model accounted for 41.5% of the participation frequency variance (17.2% by the self efficacy variable). It was also deemed the most valid estimate of the self-efficacy effect on participation frequency because the six other subsets of confounders provided neither an equivalent control of confounding nor a more precise self-efficacy estimate.   2.3.3   The mediated effect of self-efficacy on participation frequency  Three mediators were identified for analyses, including life-space mobility, wheelchair skills, and participation limitations. Although the correlation between wheelchair skills and self-efficacy (r=0.84) indicated potential collinearity, both were retained in the model for theoretical reasons (Bandura, 1997), and because the variation inflation factor value was below 10 (VIF=3.78). The total mediated effect was statistically significant (point estimate=0.14, 95% bootstrapped CI 0.04,0.24) (table 2.3). Examination of the specific indirect effects, however, indicates that the wheelchair skills variable does not contribute to the total indirect effect above and beyond life-space mobility, and participation limitations. The total mediated effect accounted for 78.0% of the direct effect on participation frequency, and the total model accounted for 55.0% of the variance.         51 Table 2.2: The direct effect of self-efficacy on participation frequency   Crude model Adjusted reduced model Factor b SE 95% CI ? b SE 95% CI ? Self-efficacy (path c) 0.22 0.03 0.16, 0.28 0.54 0.18 0.03 0.12, 0.24 0.44 Age     -0.12 0.08 -0.27, 0.03 -0.12 Comorbidities     -0.40 0.24 -0.87, -0.08 -0.12 Social support     0.66 0.15 0.36, 0.96 0.31 adj R2 29.10% 41.50% b=unstandardized coefficients; SE=standard error; ?=standardized coefficients; CI=confidence interval; adj=adjusted; n=124    Table 2.3: Multiple mediator effects of self-efficacy on participation frequency  Factor path ai path bi aibi 95% CI* Life-space mobility 0.36 0.12 0.04 0.02, 0.08 Wheelchair skills 0.63 0.04 0.03 -0.05, 0.10 Participation limitations 0.29 0.23 0.07 0.02, 0.13 Total ?i(aibi)   0.14 0.04, 0.24 ?i(aibi)/c 0.78 adj R2 55.00% CI=confidence interval; adj=adjusted; n=124 *1000 bootstrap samples    52 2.4 Discussion  In this study, the association between WC self-efficacy and participation frequency was investigated in older community-dwelling manual wheelchair users. Individuals in this research were experienced wheelchair users, who used their wheelchair daily. Compared to another study of younger wheelchair users (Rushton et al., 2013), individuals in this sample reported fewer wheelchair skills, and lower WC self-efficacy estimates. These observations are not surprising when considering evidence that functional limitations increase, and self-efficacy diminishes with aging (Bandura, 1997). Furthermore, individuals were healthy in that they reported few comorbidities, and low levels of depression and anxiety symptoms. Despite being in good health, the sample also reported a low mean participation frequency, which is consistent with another study of older, community-dwelling manual wheelchair users (Sakakibara et al., 2013c). That this sample?s mean LLDI score is comparable to scores reported by ambulatory adults who are 10 (Keysor et al., 2010), and 15 (Jette et al., 2002) years older, further indicates that wheelchair users have low participation frequency.  This study?s findings provide evidence in support of the hypothesis that after examining and controlling for important confounding effects, WC self-efficacy is an independent predictor of participation frequency of older, community-dwelling manual wheelchair users. Social Cognitive Theory explains that self-efficacy is at the foundation of human motivation and action (Bandura, 1997). Using this theory to interpret the present findings, if people believe they can produce desired effects by their actions while using their wheelchair, they have greater incentive to participate in personal and social roles more frequently. When considering that the mean participation frequency score is this study?s sample may be considered low, improvements to WC self-efficacy in older individuals may result in notable participation and quality of life outcomes.  The results of this study substantiate preliminary findings that illustrated the importance of the self-efficacy construct on participation frequency (Sakakibara et al., 2013a). However, the findings contrast in that there was no difference in the magnitude of the association by sex in this study. This disagreement may be due to the larger sample size used in this study that allowed for  53 more robust analyses, such as controlling for additional confounders. Nonetheless, the evidence suggests that strategies to improve lowered self-efficacy may have beneficial effects on participation frequency regardless of sex, but the rate of change in participation frequency may be greater for men (Sakakibara et al., 2013a). More research is needed to investigate the possible differences by sex.  Finding that the WC self-efficacy term remained a statistically significant explanatory variable of participation frequency after examining and controlling for interaction effects and confounding variables, has both clinical and research implications. Because lowered self-efficacy may present as a barrier to participation frequency, clinical trials are justified to develop and test WC self-efficacy enhancing interventions. According to Bandura (1997), low self-efficacy is an amenable condition influenced by a variety of social cognitive means. In fact, in a pilot study of older individuals who were inexperienced with using a wheelchair, the researchers demonstrated a significant positive effect of wheelchair skills training on WC self-efficacy (Sakakibara et al., 2013b).   In a recent study, Phang, Martin-Ginis, Routhier & Lemay (2012) examined WC self-efficacy as a mediator in the relationship between wheelchair skills and participation in leisure-time physical activity in younger manual wheelchair users (i.e. mean age?50years) with spinal cord injuries. Contrary to this study?s findings, they found an absence of an association between WC self-efficacy and participation after controlling for skills, and therefore demonstrated no mediating effect (Phang et al., 2012). A reason for this discrepancy relative to the findings in this study is likely in how self-efficacy was measured. Their study assessed the self-efficacy construct with items in the Wheelchair Use Confidence Scale only pertaining to moving around the physical environment. Our findings may reflect the multifaceted nature of participation being accounted for by the different conceptual areas comprising the entire scale that was used in this study. There were also differences in the modeling approach. Whereas they investigated self-efficacy as a mediator, in this study wheelchair skills in addition to life-space mobility and participation restrictions were specified to mediate the association between self-efficacy and participation. The use of situation specific self-efficacy measures is in accordance with theory, as is the functional form of the model developed in this study (Bandura, 1997).  54 This study?s mediation findings also support the hypothesis that the association between self-efficacy and participation frequency is mediated by multiple functioning/disability variables. The analyses suggests a potential causal direction in which higher self-efficacy with using a manual wheelchair may act to improve life-space mobility, and perceptions about participation limitations, which in turn may lead to more frequent participation. Therefore, efficacy-enhancing interventions targeted towards improving all of life-space mobility, and participation limitations, may be more beneficial at improving participation frequency than unilateral approaches. It is interesting that after controlling for life-space mobility, and participation limitations, that wheelchair skills was not a statistically significant mediator. A reason for this finding is likely due to the covariation between wheelchair skills and life-space mobility. More research on this association is warranted, as are clinical trials investigating the causal nature of self-efficacy on participation frequency to confirm the observations made in this study.   Limitations Although causality cannot be established due to the cross-sectional nature of the data, our findings are in agreement with both theory and a large body of research demonstrating the beneficial effects of enhanced self-efficacy on various outcomes, and the expected relationships between our hypothesized mediators and participation. Next, only two-way interaction terms were considered in order to keep the models hierarchically well-formulated. Although lower order interactions minimize collinearity, our sample size limited our ability to evaluate possible higher-order interactions that included province of residence. In addition, the mean age of our adult sample may be considered young. However, our results show that the association between self-efficacy and participation frequency do not differ in individuals 50 years of age and older. Furthermore, the self-report nature of our data from a volunteer sample may be influenced by recall bias and/or social desirability. As a result, the data may not accurately represent the population as a whole. It is also possible that because all data were collected using self-report questionnaires during face-to-face interviews, that the observed covariation may be biased due to the use of common methods of measurement (Podsakoff, MacKenzie, & Podsakoff, 2012). The use of a questionnaire to gauge wheelchair skills may confound the interpretation of the results because it may be that the questionnaire is eliciting another dimension of self-efficacy with wheelchair use (i.e. level of self-efficacy). If this is the case, the implication would be that self- 55 efficacy with wheelchair use is being examined as both a predictor, via the WheelCon, and a mediator, with the use of the Wheelchair Skills Test-Questionnaire. Although this is a possibility, it is equally possible that the Wheelchair Skills Test-Questionnaire is in fact measuring what it is intended to measure, given research evidence indicates a high correlation between the questionnaire and the performance-based versions of the Wheelchair Skills Test (Rushton et al., 2012). Finally, although this research examines the direct and mediated effects of WC self-efficacy, it is also plausible that self-efficacy could mediate the associations between life-space mobility, participation limitations, and participation frequency as per the tenet of triadic reciprocalism in Social Cognitive Theory (Bandura, 1997).  2.5 Conclusion Self-efficacy with using a manual wheelchair has both direct effects on the participation frequency of older, community-dwelling manual wheelchair users, as well as mediated effects through life-space mobility, wheelchair skills, and participation limitations. Employing efficacy-enhancing strategies as part of the rehabilitation process may affect health outcomes. Self-efficacy is an important construct to consider in the study of wheelchair users, and the development of interventions to address lowered self-efficacy with using a manual wheelchair is warranted.               56 CHAPTER 3: Direct and mediated self-efficacy effects on life-space mobility   3.1 Introduction  Wheelchairs are reported to be the most important form of assistive technology for mobility among older adults with mobility limitations (Mann et al., 2004). Research from the United States shows an increasing rate in the usage of wheeled mobility equipment (including the use of manual wheelchairs) (LaPlante & Kaye, 2010), exceeding the rates of both population growth and aging (Shrestha & Heisler, 2011; LaPlante & Kaye, 2010). Evidence indicates that between 1990 and 2005, the use of wheeled mobility equipment increased by 112%, representing an annualized growth rate of 4.6% overall, and 4.3% for adults 65 years and older (LaPlante & Kaye, 2010). This may indicate a reduction in unmet needs for wheelchairs to account for the increase in usage due to better accessibility, and financial resources, and declining stigma (LaPlante & Kaye, 2010).  Mobility is a basic and necessary activity that supports the ability of individuals to engage in desired activities and social roles (Jette et al., 2003; Rousseau-Harrison et al., 2009). Mobility limitations on the other hand have been shown to compromise independence (Hirvensalo, Rantanen & Heikkinen, 2000; Rubenstein, Powers & MacLean, 2001), increase the risk of institutionalization (von Bonsdorff, Rantanen, Laukkanen, Suutama & Heikkinen, 2006), and mortality (Hirvensalo et al., 2000), and decrease quality of life (Rubenstein et al., 2001; Riggins et al., 2011). In fact, mobility limitations are a leading cause of disability in community-dwelling individuals (Iezzoni et al., 2001), and are the primary reason for both the onset and persistence of participation restrictions in individuals 50 years of age and older (Wilkie, Thomas, Mottram, Peat & Croft, 2008). Although the use of wheelchairs for mobility by older adults has been shown to be positively associated with their social participation (Barker et al., 2004; Rousseau-Harrison et al., 2009), independence (Hoenig et al., 2003b), and movement through various life-spaces (Meyers et al., 2002; Auger et al., 2010), many older, community-dwelling individuals report limitations with using their manual wheelchair (Shields, 2004; Ganesh et al., 2007). Because the severity of wheeled mobility limitations are positively associated with aging (Shields, 2004; LaPlante & Kaye, 2010), it is plausible that there will be more older individuals  57 experiencing limitations with their wheeled mobility over time given increasing wheelchair usage (Sapey et al., 2005; LaPlante & Kaye, 2010), and population aging.  Most of the evidence on the wheeled mobility of manual wheelchair users focuses on individuals living in long-term care settings (Bourbonniere et al., 2007; Mortenson et al., 2011; Mortenson et al., 2012). In these studies, mobility is assessed using measures of life-space, which is defined as movement extending from within one?s home to movement beyond one?s town or geographic region (May, Nayak & Isaacs 1985). The predictive models in these studies identify several statistically important factors, including, ability to use a wheelchair (Mortenson et al., 2011; Mortenson et al., 2012), need for a seating intervention (Bourbonniere et al., 2007), depression (Mortenson et al., 2012), environmental barriers (Mortenson et al., 2012), functional ability (Mortenson et al., 2011), and social support (Mortenson et al., 2011). Despite knowledge of these important factors, they are specific to individuals in long-term care settings, and account for less than half of the life-space mobility variance. There remains a lack of evidence describing the mobility of older, community-dwelling manual wheelchair users, and questions as to the associations of other important mobility variables not yet investigated.  Self-efficacy (Bandura, 1997) with using a manual wheelchair is emerging as a new and potentially important clinical/research construct. It is conceptualized as the belief individuals have in their ability to use their wheelchair in varying situations (Rushton et al., 2011). Because self-efficacy has been shown to have important mobility implications in mobility limited populations including older individuals with lower extremity functional limitations (McAuley et al., 2007), lower extremity amputation (Miller et al., 2001), and knee osteoarthritis (Maly, Costigan & Olney, 2007), reasoning by analogy suggests that it may similarly influence the mobility of older wheelchair users. In fact, 39.0% of older Canadian wheelchair users are speculated to have lowered WC self-efficacy (Miller et al., 2012) and moreover, WC self-efficacy has been shown to have both direct and mediated effects on participation frequency after controlling for confounding variables, as illustrated in Chapter 2. Therefore, it seems plausible for the construct to have similar effects on the life-space mobility of manual wheelchair users, but this has not yet been established.   58 In this study, the association between WC self-efficacy and life-space mobility is examined using the International Classification of Functioning, Disability, and Health (ICF) (WHO, 2001) framework to guide the analyses. The direct effect of the self-efficacy construct on life-space mobility is first estimated. The mediating effect of ability to use a wheelchair is then examined because Social Cognitive Theory indicates that self-efficacy influences behavior both directly and by its influence on the other determinants (Bandura, 1997). It is hypothesized that self-efficacy with using a manual wheelchair is an independent predictor of life-space mobility after controlling for confounding variables, and that the association between self-efficacy and life-space mobility is mediated by wheelchair skills.  3.2 Methods  3.2.1   Participants and recruitment  The same cohort of wheelchair users described in Chapter 2 (page 36) participated in this study. Subject inclusion and exclusion criteria were unchanged, and the study protocol and strategies used to recruit the volunteer sample also remained the same.   A sample size of n=123 was calculated with G*Power version 3.1 (available at http://www.psycho.uniduesseldorf.de/abteilungen/aap/gpower3/), using a moderate effect size (?2=0.15), and an alpha of 0.05, power of 0.80, for regression modeling with up to 11 independent variables entered into the model.  The ethics boards from all participating institutions approved this study.  3.2.2   Outcome measures  Variables/measures were selected based on the ICF framework (WHO, 2001), and published empirical research. Chapter 2 provides a detailed description of the measures used in this study, organized by ICF domain (pages 37-41). The measurement properties of all measures used in this study have been evaluated with wheelchair users, older adults, and/or individuals with  59 mobility limitations, and are tabulated in appendix A. The variables/measures used in this study are briefly presented below.  Dependent variable: Life-space mobility was measured using the Life-Space Assessment (Baker et al., 2003). The Life-Space Assessment is a 20-item questionnaire that assesses the frequency (1=less than once a week, to 4=daily) that individuals move in five areas over the past month: 1) within the home; 2) around the home; 3) in the neighbourhood; 4) in town; and 5) outside of town. In addition, independence to move between life-spaces  (e.g. 1=assistance from other persons, 1.5=with equipment, or 2=no assistance) is queried. A composite score is derived by first multiplying each life-space level by the weekly frequency, and then by the level of independence. The products from each life-space level are then summed to derive a composite score that ranges between 0 and 120, or 0 to 90 for individuals who use assistive devices such as those who participated in this study. Higher scores represent more life-space mobility. The composite score has excellent test-retest reliability in both community-dwelling older adults (ICC=0.96, 95% CI=0.95,0.97) (Baker et al., 2003), and power wheelchair users (ICC=0.87, 95% CI=0.69, 0.92). In terms of validity, Baker et al. (2003) reports moderate correlations (r=-0.41 to 0.60) in the expected directions with measures of physical performance, activities of daily living, and depression, in older individuals.   Independent variable of interest: Self-efficacy with using a manual wheelchair was measured using the 65-item Wheelchair Use Confidence Scale (WheelCon) (Rushton et al., 2011. Individuals are asked to rate the strength of their confidence in performing tasks and activities using a 0 to 100 response scale. A mean score is calculated with higher scores indicating greater strengths of self-efficacy. There is evidence in support of the validity and reliability of the WheelCon?s measurements with manual wheelchair-users (Rushton et al., 2013).  Potential confounding variables: The socio-demographic information form collected health related variables such as diagnosis (neurological condition or not), and personal factor variables such as age, sex, education (high school graduate), marital status (married/not married), employment/volunteer status (employed/volunteer or not employed/volunteer), and income (<$30,000/?$30,000/prefer not to answer). The form also gathered data on personal factor  60 variables related to wheelchair use, and physical environments. Wheelchair related variables included years of experience, hours of daily use, formal wheelchair training (yes/no), and assistance needed with using the wheelchair (yes/no), and the physical environment variable included, geographic location (British Columbia/Quebec).  Number of comorbidities, need for a seating intervention, and perceived social support were also considered as potential confounders, and captured using the Functional Comorbidity Index (Groll et al., 2005), the Seating Identification Tool (Miller et al., 2004), and the Interpersonal Support and Evaluation List-6 (Cohen & Hoberman, 1983). Physical environmental barriers were assessed using the Home and Community Environment Instrument (Keysor et al., 2005).   Mediating variable: Ability to use a manual wheelchair was assessed with the 32-item Wheelchair Skills Test-Questionnaire (WSTP manual, 2008)). Individuals are asked if they are able to complete a specific skill using their wheelchair. Ratings include pass, fail, or not applicable if the wheelchair does not have the component. Total percentage scores are derived by dividing the number of individual skills passed by the total number of applicable skills. Higher scores indicate more wheelchair skills. Measurements made using the Wheelchair Skills Test-Questionnaire have demonstrated a high correlation with measure from the performance based Wheelchair Skills Test version 4.1 (Spearman correlation (rs)=0.89) (Rushton et al., 2012).  3.2.3   Study protocol  After completing the socio-demographic information form, the Mini Mental State Examination, which was used as a cognitive screen (i.e. individuals with scores <23 were excluded) (Folstein et al., 1975), and the WheelCon, participants were administered the remaining measures in a random sequence to minimize response bias.   3.2.4   Data analyses  Descriptive statistics were used to characterize the sample. Results from categorical variables were calculated as percentages, and from continuous variables as means and standard deviations.  61 Multiple regression analyses were used to test the study hypotheses. All dichotomous variables were coded as either -0.5 or 0.5 (Norman & Streiner, 2008). To estimate the most valid and precise direct effect of WC self-efficacy on life-space mobility, the three-stage modeling strategy detailed by Kleinbaum and Klein (2010) was utilized. The following analyses are similar to those reported in Chapter 2 (pages 42-45) where more details on each modeling stage may be found.  3.2.4.1   The direct effect of self-efficacy on life-space mobility  Stage 1: Variable specification In the first modeling stage, variables and interaction terms were specified for entry. Only variables with published and plausible evidence of a relationship with the dependent variable were considered. Data for 19 variables were collected. To reduce the number of variables for entry, only those continuous variables with a bivariable correlation to the dependent variable with a fair magnitude of at least 0.25 (Portney & Watkins, 2009), and/or those categorical variables with a significant mean difference (tested using independent sample t-tests or one-way ANOVA) in the life-space mobility variable were included in the model. Scatterplots of the bivariate data were examined for potential outliers. Data points greater than 1.5 times the variable?s interquartile range were considered cases that could influence the correlation (Portney & Watkins, 2009). These cases were removed from this modeling stage only, and the correlation coefficients recalculated. Potential collinearity was identified by an intercorrelation of 0.70 between independent variables, and a variance inflation factor value greater than 10 (Kleinbaum et al., 2008). To minimize collinearity all continuous variables were mean centred, however, when collinearity was identified the variable with the highest correlation with the dependent variable was entered into the model, unless there was theoretical rationale to inform a decision. This process led to seven candidate variables for regression analysis.  Geographic location and age interaction terms were also specified for entry. Because there was a significant difference in life-space mobility between individuals in British Columbia and Quebec, a geographic location interaction was examined term to determine if the association between self-efficacy and life-space mobility is different between individuals in British Columbia and Quebec. An age interaction term was also evaluated because evidence illustrates  62 that self-efficacy diminishes with aging (Bandura, 1997). A significant age interaction term was indicative that the association between self-efficacy and life-space mobility differs between people of different ages.  Regression assumptions: After specifying variables and interaction terms for regression modeling, the residuals were inspected for outliers, and to verify the regression assumptions. Cases with standardized residual values above 3 or below -3 were potential outliers, and were excluded from regression analyses if any Cook?s distance values exceeded 1 (Kleinbaum et al., 2008). In addition to assessing for collinearity as discussed above, the data was examined for independence, normality, and homoscedasticity (Kleinbaum et al., 2008).   Stage 2: Interaction assessment In the second modeling stage the statistical significance of the interaction terms was assessed. The self-efficacy variable was forced into model, followed by the geographic location and age variables, and then the interaction terms. Forward selection, and backward elimination regression approaches were used to eliminate any non-significant interaction term (Kleinbaum et al., 2008). If an interaction term was retained in the model, its lower order components were also retained to ensure the model was hierarchically well-formulated (Kleinbaum et al., 2008).   Stage 3: Confounding and precision assessment In the final stage of model development, an assessment for confounding was done by comparing the self-efficacy beta estimate in the crude model (i.e. the model resulting from the second modeling stage) with the estimate in the adjusted model (i.e. the crude model plus the addition of other potential confounding variables). A change in the WC self-efficacy beta coefficient in the adjusted model of at least 16.4% was indicative of confounding because this represents the WheelCon?s previously reported minimal detectable change (Rushton et al., 2013).   When confounding was present, subsequent analyses were performed to identify possible subsets of confounding variables that provided equivalent control of confounding, but with greater precision. Precision was evaluated by the width of the 95% confidence interval of the self-efficacy variable?s unstandardized regression coefficient (Kleinbaum et al., 2008). A narrowing  63 of the confidence interval was indicative of greater precision. A backward elimination approach was used to reduce the model. The variables that remained in the model after the backward elimination were considered to comprise the first confounder subset. Each subsequent subset examined included one of the backward removed variables to the first subset. The subset of variables that provided equivalent control of confounding relative to the adjusted model, but with greater precision was deemed to provide the most valid estimate of the direct effect (i.e. path c in figure 3.1a) of self-efficacy on life-space mobility. This model was then used in the subsequent mediator analysis.  3.2.4.2   The mediated effect of self-efficacy on life-space mobility  The mediator analysis used the product-of-coefficients approach combined with bootstrapping methods guided by the work of Preacher and Hayes (2008). This approach does not focus on the individual paths in the mediation model but rather on the total indirect effect derived as the product of the regression coefficients linking the independent variable to the mediator (i.e. path a), and the mediator to the dependent variable (i.e. path b) (Preacher & Hayes, 2008). A bias corrected bootstrapping method was used to derive the point estimate of the mediated effect and 95% confidence interval (Preacher & Hayes, 2008). The proportion of the direct effect accounted for by the hypothesized mediator was calculated using the formula ab/c. The adjusted R2 values are reported for all models. Figures 3.1a and 3.1b present the direct and mediated paths.  SPSS version 19.0 (SPSS Inc., Chicago, IL) and the INDIRECT macro (Preacher & Hayes, 2008) were used for the analyses.          64 Figure 3.1:  The direct and mediated paths of self-efficacy on life-space mobility                                  	 ?	 ?	 ?	 ?Body Function: Self-efficacy Activity: Life-space mobility Health, Personal and Environment factors: Confounding variables c	 ?Figure 3.1a: Direct effect Activity: Wheelchair skills Body Function: Self-efficacy Activity: Life-space mobility Health, Personal and Environment factors: Confounding variables ai Figure 3.1b: Mediated effect bi  65 3.3 Results  The mean age of the total sample (n=124) was 59.67 years (SD=7.49), 74 (59.7%) participants were from British Columbia, and 74 (59.7%) were male. Ninety-seven (78.2%) participants had a neurological condition, and the mean number of comorbidities was 2.69 (SD=2.40) out of a possible 18 conditions. Thirty-nine (31.5%) individuals required some form of assistance with using their wheelchair (e.g. mobility, transferring, set-up), and 22 (17.7%) received training to use their wheelchair outside of rehabilitation. The sample?s mean need for a seating intervention score was 1.98 (SD=1.69). The mean self-efficacy with using a manual wheelchair score was 78.38 (SD=19.19) out of 100, and the mean Life-Space Assessment score was 46.99 (SD=17.84) out of 120 (or 90 for individuals who use assistive technology for their mobility). Sample characteristics are discussed in depth in Chapter 2 (pages 46-47), and presented in table 3.1.  3.3.1   The direct effect of self-efficacy on life-space mobility  Stage 1: Variable specification Pearson correlation coefficients between the continuous and dependent variables are shown in table 3.1, along with the mean differences in the dependent variable for the categorical variables. The correlation matrix between all continuous variables is presented in appendix Q. Variables initially meeting the modeling inclusion criteria included the age and geographic location interaction terms, geographic location, sex, number of comorbidities, formal training to use a wheelchair, assistance with using the wheelchair, and employment/volunteer status. Confounders specified for inclusion based on the magnitude of their bivariate correlation with the dependent variable did not change after outlier analyses.   Regression assumptions: None of the regression assumptions were violated. Potential collinearity was noted by a correlation of 0.84 between the WC self-efficacy and wheelchair skills variables but both were kept in the mediated model because the former was the primary explanatory variable to be investigated and the latter was the main hypothesized mediating variable, and because the variance inflation factor values were well below 10.00 (range=1.03-3.77).  66 Table 3.1:  Descriptive statistics and correlations with/mean differences in life-space mobility   Variable Total Life-space mobility  mean?sd/frequency (%) r/(mean difference) Activity: Life-space mobility (0-120) Wheelchair skills (0-100)  46.99?17.84 75.49?14.89  1 0.49* Body functions: Self-efficacy (0-100)  78.38?19.19  0.47* Health condition: Comorbidities (0-18) Neurological condition: Spinal cord injury Multiple sclerosis Stroke Other (Parkinson?s Cerebral palsy, brain injury) Non-neurological condition: Amputation Polio Arthritis  Other    2.69?2.40 97 (78.20) 60 (48.40) 16 (12.90) 12 (9.70) 9 (9.30)  27 (21.80) 9 (9.30) 5 (4.03) 4 (3.23) 9 (9.30)  -0.34*  (0.16) Personal factors: Age Male Education (high school graduate) Income:?  <$30,000 Prefer not to answer Married (yes) Employed/volunteer (yes) Wheelchair Years experience Daily use (hours) Formal training (yes) Wheelchair assistance (yes)  59.67?7.49 74 (59.68) 110 (89.40)  43 (34.68)  21 (16.94) 59 (47.60) 46 (37.10)  22.31?16.05 12.30?4.29 22 (17.70) 39 (31.50)  -0.24 (11.99)* (-4.06)  (6.47)  (9.59)  (3.28) (-7.46)*  0.13 0.06 (-9.19)* (13.91)* Environmental factors: Wheelchair Need for seating intervention (0-15) Social Social support (0-18) Physical British Columbia Home barriers (0-10) Community barriers (0-5)   1.98?1.69  14.48?3.71  74 (59.70) 1.10?1.22 1.06?0.85   -0.07  0.24  (8.59)* -0.07 -0.10  *included for modeling; ?mean difference from ?30,000; n=124  67 Stage 2: Interaction assessment Neither the age nor the geographic location interaction terms were statistically significant. The crude model shown in table 3.2 which included only the WC self-efficacy variable accounted for 22.2% of the life-space mobility variance. Because the association between self-efficacy and life-space mobility did not differ by geographic location, the datasets were combined for further analyses.  Stage 3: Confounding and precision assessment The adjusted model that included the remaining variables specified in stage 1 resulted in confounding of the WC self-efficacy estimate by 56.9%. The backward elimination regression procedure resulted in the first variable subset of the sex, number of comorbidities, geographic location, and assistance with using the wheelchair variables that confounded the association between WC self-efficacy and life-space mobility by 48.5%. The confounding of this variable subset also resulted in improved precision of the self-efficacy estimate. This model provided the most valid estimate of WC self-efficacy on life-space mobility because evaluation of other subsets provided neither equivalent control of confounding nor greater precision. Table 3.2 details the adjusted model that includes the sex, number of comorbidity, geographic location, and assistance with using a wheelchair variables. After controlling for the confounders, the WC self-efficacy variable remained significant and accounted for 3.9% of the life-space mobility variance. Overall, the model accounted for 37.1% of the variance, and was used in the subsequent mediator analysis.   3.3.2   The mediated effect of self-efficacy on life-space mobility   The mediated effect of WC self-efficacy on life-space mobility through wheelchair skills was significant (point estimate = 0.21, 95% bootstrapped CI = 0.05, 0.43). The wheelchair skills variable accounted for 91.0% of the direct effect of self-efficacy on life-space mobility. This model accounted for 39.0% of the life-space mobility variance. The bootstrapped estimates and confidence intervals are shown in table 3.3.     68 Table 3.2:  The direct effect of self-efficacy on life-space mobility   Crude model Adjusted model Factor b SE 95% CI ? b SE 95% CI ? Self-efficacy (path c) 0.45 0.07 0.30, 0.59 0.48 0.23 0.08 0.07, 0.39 0.25 Sex     -5.83 2.82 -11.41, -0.26 -0.16 Number of comorbidities     -1.64 0.56 -2.75, -0.53 -0.22 Geographic location     -10.78 2.72 -16.17, -5.38 -0.30 Wheelchair assistance     -7.78 3.36 -15.03, -1.73 -0.22 adj R2 22.20% 37.10% b=unstandardized coefficients; SE=standard error; ?=standardized coefficients; CI=confidence interval; adj=adjusted; Male sex, British Columbia, and no need for wheelchair assistance were coded as -0.50; n=124       Table 3.3:  The mediated effect of self-efficacy on life-space mobility  Factor path a path b ab 95% CI* Ability to use a wheelchair 0.58 0.37 0.21 0.05, 0.43 ab/c 0.91 adj R2 39.00% CI=confidence interval; adj=adjusted; n=124 *1000 bootstrap samples      69 3.4 Discussion  In this study, the association between WC self-efficacy and life-space mobility was investigated. Participants in this research were older community-dwelling adults with a variety of diagnoses, and many years of experience with using a wheelchair. The sample reported fewer wheelchair skills, and lowered WC self-efficacy than those described in a study of younger, manual wheelchair users (median age = 50.0, interquartile range = 31-60) (Rushton et al., 2013). In addition, this sample?s mean life-space mobility is larger than reports of older individuals who use powered mobility devices (Auger et al., 2010), however, it is smaller than that of younger wheelchair users (Rushton et al., 2013), and ambulatory older adults (Peel et al. 2005; Bentley et al., 2012). That this sample?s life-space mobility is lower than that reported by ambulatory older adults, which is not surprising because the highest possible total score in the Life-Space Assessment measure for individuals who use assistive devices for their mobility is 30 points lower than the highest possible score for individuals who do not use assistive devices (Baker et al., 2003). Nonetheless, the life-space mobility of this sample seems reasonable when compared to other samples of wheelchair users (Auger et al., 2010; Rushton et al., 2013), and may be low when considering that the mean Life-Space Assessment score is just above midpoint of all possible scores for individuals who use assistive devices for mobility.   After specifying and controlling for important confounding variables, evidence was found to substantiate the hypothesis that WC self-efficacy is an independent predictor of life-space mobility. These findings complement the existing knowledge of the WC self-efficacy construct in other areas of rehabilitation such as participation in social and personal roles (Sakakibara et al., 2013a). In the existing literature, more than half of the life-space mobility variance of wheelchair users remains unexplained (Mortenson et al., 2011). Predictive models, however, have not yet considered the WC self-efficacy variable. Therefore, the examination of the self-efficacy construct in addition to important explanatory variables identified in other studies may lead to a more complete understanding of life-space mobility in older wheelchair users.   The results from the mediation analysis also support the hypothesis that the association between WC self-efficacy and life-space mobility is mediated by wheelchair skills. The findings suggest  70 that the association is almost entirely explained by wheelchair skills. Therefore, higher WC self-efficacy may lead to better wheelchair skills, which in turn may increase life-space mobility. Although causality cannot be determined due to the study?s cross-sectional nature, Social Cognitive Theory corroborates the results in postulating that self-efficacy has both direct and indirect effects on behaviour (Bandura, 1997).   The interpretation that changes to WC self-efficacy may occur prior to wheelchair skills reinforces observations from a previous experimental study. In this pilot trial, two 1-hour wheelchair skills training sessions based on the Wheelchair Skills Training Program (WSTP manual, 2008) led to statistically significant improvements in WC self-efficacy, but not in wheelchair skills, measured using the performance-based Wheelchair Skills Test (Kirby et al., 2004a). This finding, contrary to much evidence illustrating the value of the Wheelchair Skills Training Program at improving wheelchair skills (Coolen et al., 2004; Kirby et al., 2004b; MacPhee, Kirby, Coolen, Smith & MacLeod, 2004; Best, Kirby, Smith & MacLeod, 2005), was attributed to the fewer number of hours of skills training administered (2 hours) relative to 3 to 9 hours in the other investigations. It was therefore speculated that one-on-one skills training has efficacy-enhancing effects prior to improving ability. This progression is congruent with Social Cognitive Theory wherein improved self-efficacy functions as an influential regulator of an individual?s behaviour (Bandura, 1997). Further experimental research is warranted to substantiate the finding of a possible causal path between WC self-efficacy, wheelchair skills, and life-space mobility.  Overall, the findings contribute to the growing body of evidence indicating a need for the development and testing of targeted efficacy-enhancing interventions for those older, community-dwelling manual wheelchair users who may have lowered WC self-efficacy. Considering that the life-space mobility of the individuals in this study?s sample is mostly occurring within the home and local neighbourhood, as indicated by the low mean Life-Space Assessment score, improvements to WC self-efficacy may result in more travel to the community, if and when desired, and thus community involvement. Improved WC self-efficacy may also result in more frequent excursions outside of the home, and/or less personal assistance with life-space mobility.  71 In theory, self-efficacy is amenable to information sourced from performance accomplishments, vicarious learning, verbal persuasions, and/or physiological/affective states (Bandura, 1997). Moreover, research illustrates the value of targeted theoretically based interventions resulting in enhanced self-efficacy in mobility impaired populations (Salbach et al., 2005; Sanford et al., 2006). In fact, pilot evidence from the aforementioned experimental study demonstrates the modifiable nature of the WC self-efficacy construct in older individuals who did not have any experience using a wheelchair (Sakakibara et al., 2013b). The Wheelchair Skills Training Program may be administered with the use of community built structures (Best et al., 2005), therefore, the program offers an efficient and cost effective clinical solution to improve lowered WC self-efficacy via performance accomplishments in inexperienced wheelchair users. Furthermore, because in this study there was not a statistically significant age interaction term, improvements made to lowered WC self-efficacy may result in similar changes in life-space mobility regardless of older age cohort.   However, because this study?s sample was comprised of experienced wheelchair users, modifying their efficacy judgments may be more complex than doing so with inexperienced wheelchair users (Bandura, 1997). Thus, interventions comprising performance accomplishment in addition to other sources of efficacy-enhancing information may be superior at improving WC self-efficacy than any element alone. Furthermore, the emphasis of each component of the intervention may differ depending on the person?s strength of WC self-efficacy relative to their ability to use a wheelchair. Interestingly, evidence illustrates that 27.0% of manual wheelchair users have discordant WC self-efficacy beliefs and wheelchair skill capacity, with the majority having higher WC self-efficacy beliefs and lower level of skill (Miller et al., 2012). Conversely, a profile of lower WC self-efficacy beliefs combined with higher levels of skill may be more common among adults in older age groups. According to Bandura, self-efficacy issues for these older individuals center on misappraisals of declining ability that in turn negatively affect their self-efficacy (Bandura, 1997). This suggests that age may interact with WC self-efficacy to influence the association with wheelchair skills. Therefore, the social cognitive strategies used to elicit optimal self-efficacy improvements may differ depending on the individual and/or sample and require both age and situation specific approaches. Research to investigate this is needed.   72 Limitations This study has several limitations. Although this research examines the direct and mediated effects of WC self-efficacy, it is also plausible that self-efficacy could mediate the associations between wheelchair skills and life-space mobility as per the tenet of triadic reciprocalism in Social Cognitive Theory (Bandura, 1997). Next, in addition to limited claims on causality due to study design, the correlation between self-efficacy and ability to use a wheelchair may have introduced collinearity into the mediator model, although the variance inflation factor did not indicate a need for corrective action. Furthermore, the self-report nature of our data from a volunteer sample may be influenced by recall bias and/or social desirability. This might be especially true for the use of the questionnaire version of the Wheelchair Skills Test. As a result, the data may overestimate ability to use a wheelchair, and therefore not accurately represent the sampled participants? skills, and the population as a whole. Another limitation due to the use of the Wheelchair Skills Test-Questionnaire is related to the belief that this questionnaire may be assessing a similar construct as the WheelCon, albeit a different dimension of WC self-efficacy (i.e. level of self-efficacy). If this is the case, then WC self-efficacy is being examined as both the predictor variable and mediator. Despite this consideration, it still remains that WC self-efficacy is a statistically significant and independent predictor of life-space mobility. In addition, common methods biases may have contributed to an overestimation of the amount of variance accounted for (Podsakoff et al., 2012).  3.5 Conclusion  Self-efficacy with using a manual wheelchair has important direct and mediated associations with the life-space mobility in older, community-dwelling wheelchair users. The findings contribute to existing evidence suggesting the clinical relevance of the self-efficacy construct. Although evidence from this study suggests interventions targeted toward improving lowered self-efficacy may lead to improvements in both wheelchair skills and life-space mobility, more age and situation specific research is needed.  73 CHAPTER 4: Health, personal, and environmental predictors of self-efficacy with using a manual wheelchair  4.1 Introduction  Although research on self-efficacy with using a manual wheelchair is in its infancy, evidence is mounting in support of it as an important consideration in the rehabilitation and research of older, community-dwelling manual wheelchair users. For example, Chapters 2 and 3 illustrate the statistically significant associations between the self-efficacy construct, and both frequency of participation in social and personal roles, and life-space mobility. These findings build upon the already published literature reporting the positive effects of self-efficacy with using a wheelchair on participation in smaller samples (Phang et al., 2012; Sakakibara et al., 2013a). That the construct is important both clinically and in research is further supported by well-established evidence on other forms of self-efficacy, such as those particular to balance (Miller et al., 2001), exercise (Meland et al., 1999), and mobility (Sanford et al., 2006), with important implications on participation and activity outcomes in populations with mobility limitations. Moreover, when considering low self-efficacy is a contributing factor to the onset of depressive symptoms (Bandura, 1997), and that lowered WC self-efficacy may be a barrier to participation and mobility, research is warranted to identify individuals at risk of having lowered self-efficacy, and who may be in greatest need of efficacy-enhancing interventions.  According to Social Cognitive Theory, self-efficacy develops and evolves throughout an individual?s lifespan (Bandura, 1997). Although self-efficacy is influenced by a number of modifiable variables, such as functional abilities, it is knowledge of the predisposing contextual factors (e.g. health condition, personal, and environmental factors) that will help clinicians and researchers to identify individuals and subgroups of older wheelchair users most at risk of having lowered self-efficacy.   The International Classification of Functioning, Disability, and Health (ICF) (WHO, 2001), indicates that disability/functioning variables are influenced by all of health conditions, personal, and environmental contextual factors. When considering that WC self-efficacy is conceptualized  74 as a body function it is plausible that these factors may impact self-efficacy. For example, unpredictable health issues such as the onset of chronic conditions or disability may alter beliefs about certain abilities, or create new beliefs about new abilities, such as using a wheelchair. This however, has yet to be investigated. In terms of personal factors, Bandura (1997) theorizes aging influences self-efficacy. Because declines in health and physical functioning are associated with aging, older individuals tend to report lowered beliefs in their ability (Bandura, 1997). There is some empirical evidence on the association between age and WC self-efficacy that both corroborates and conflicts with theory. For example, discrepancies exist as to the importance of age. Whereas Rushton et al. (2013) observed a non-significant negative correlation of low magnitude between age and WC self-efficacy in a sample (n=83) of mostly male (70.0%) community-dwelling wheelchair users (median age=50, interquartile range = 31-60), Fleiss-Douer et al. (2011) established a statistically significant association between age and the Self-Efficacy in Wheeled Mobility scale in a mostly male sample of elite and recreational athletes (mean age=38.20, SD=13.90). Studies of different populations have similarly shown age to have a statistically significant, negative association with other forms of self-efficacy (Wilcox, Bopp, Oberrecht, Kammermann & McElmurray, 2003; Wilcox, Sharpe, Hutto & Granner, 2005). The association between age and WC self-efficacy, therefore, requires further investigation.   Sex is another personal factor that may be associated with self-efficacy. Although both Rushton et al. (2013) and Fleiss-Douer et al. (2012) reported no difference in WC self-efficacy by sex, their findings are contrary to studies of non-wheelchair users where such differences are common observations (Lirgg, 1991; Wilcox et al. 2005). Lirgg (1991) speculates that differences by sex are reported more frequently when the self-efficacy is specific to tasks that are physically challenging or perceived to be masculine in nature. That is, the more the task is perceived to be masculine, the more likely females will report lower self-efficacy than males. Tasks perceived as sex neutral, however, have been shown to elicit similar levels of the construct by both males and females (Lirgg, 1991). Because wheelchair use may be perceived as a sex neutral activity may be a reason why no differences have been observed in self-efficacy with using a manual wheelchair in previous studies. This, however, needs to be established.    75 Personal factors related to wheelchair use may also be associated with WC self-efficacy. For example, Rushton et al. (2013) observed a statistically significant correlation between WC self-efficacy and years of wheelchair use (Spearman correlation (rs)=0.32, p=0.003). Years of wheelchair use and other variables related to experiences with wheelchair use may be associated with WC self-efficacy in multivariate analyses, however, this also needs to be established.  Environmental factors (wheelchair-related, social, and physical) may also contribute to the shaping of one?s self-efficacy. For example, research indicates that individuals who perform better due to better sporting equipment (e.g. tennis racket) also have higher sport-related self-efficacy (Pellet & Lox, 1998). The same may be true with respect to wheelchair seating and fit. In terms of the social environment, as individuals age their social network shrinks due to retirement, and deaths of family and friends for example. As a result, perceived self-efficacy may diminish due to losses of social support (Bandura, 1997). Furthermore, barriers or facilitators in the physical environment are also postulated to influence various forms of self-efficacy (Bandura, 1997). It is therefore plausible that wheelchair users in non-accessible physical environments may have lower perceived WC self-efficacy relative to wheelchair users in accessible environments.   Although some research exists on contextual factor correlates of WC self-efficacy, the evidence to date has focused on only a few factors, and there is some disagreement as to the statistical significance of several variables, with some being contrary to theory. Furthermore, because most of the evidence is from younger samples (Fleiss-Douer et al., 2012; Rushton et al., 2013), or specifically on individuals with a spinal cord injury (Fleiss-Douer et al., 2012), it is not readily generalizable to older, community-dwelling manual wheelchairs users. Moreover, because there are no predictive models of WC self-efficacy, clinicians and researchers are limited in their ability to identify individuals who may be more prone to lowered WC self-efficacy, and benefit the most from efficacy-enhancing interventions.    The purpose of this study is to identify health condition, personal, and environmental factor predictors of self-efficacy with using a manual wheelchair. The ICF (WHO, 2001) was used to help define and categorize variables, and guide the analyses. It is hypothesized that: i) health  76 condition variables would independently predict self-efficacy with using a manual wheelchair; ii) personal factor variables would predict self-efficacy, after controlling for health condition(s); and iii) environmental factor variables (e.g. wheelchair, social, or physical) would predict self-efficacy, after controlling for health condition, and personal factor variables, in older, community-dwelling manual wheelchair users.   4.2 Methods  4.2.1   Participants and recruitment  The same cohort of wheelchair users described in Chapter 2 (page 36) participated in this study. Subject inclusion and exclusion criteria were unchanged, and the study protocol and strategies used to recruit the volunteer sample also remained the same.   The ethics boards from all participating institutions approved this study.  4.2.2   Outcome measures  Chapter 2 provides a detailed description of the measures used in this study, organized by ICF domain (pages 37-39). All measures have established measurement properties of reliability and validity in samples of wheelchair users, older adults, and/or individuals with mobility limitations. A table summarizing all of the measures is provided in appendix A. Figure 1.1 (page 9) also presents the health condition, personal, and environment factor variables evaluated in this study by ICF domain. A brief overview of the variables/measures used in this study is presented below.  Dependent variable: Self-efficacy with using a manual wheelchair was assessed with the 65-item Wheelchair Use Confidence Scale (WheelCon) (Rushton et al., 2011). In this self-report measure, individuals rate the strength of their self-efficacy in performing tasks and activities while using their manual wheelchair using a 0 to 100 response scale. A mean score is calculated with higher scores indicating more self-efficacy. There is evidence in support of the validity and reliability of the WheelCon?s measurement in manual wheelchair users (Rushton et al., 2013).  77 Health condition: Self-efficacy has been positively associated with better health status outcomes in a range of rehabilitation settings and conditions including multiple sclerosis (Motl & Snook, 2008), arthritis (Brekke, Hjortdahl & Kvien, 2001), and spinal cord injury (Horn, Yoels, Wallace, Macrina & Wrigley, 1998). The socio-demographic information form was used to collect data on the subjects? primary diagnosis, which was dichotomized to distinguish between individuals with a neurological condition or not. The 18-item Functional Comorbidity Index (FCI) (Groll et al., 2005) was also used to collect data on each participant?s number of comorbidities, which was used as an indication of health status.  Personal variables: In addition to evidence already discussed about age and sex, different forms of self-efficacy in older adults has been shown to be influenced by socio-economic variables such as income and education (Horn et al., 1998; Wilcox et al., 2005), as well as marital and employment status (Horn et al., 1998). Data on age, sex, income, level of education, and marital and employment/volunteer statuses was collected using the socio-demographic information form. Data was also collected on years of wheelchair use experience, daily hours of wheelchair use, formal training to use a wheelchair, and assistance needed with using a wheelchair.  Environmental variables: Variables in each of the wheelchair (i.e. products and technology), social (i.e. support and relationships), and physical (i.e. natural/human made) environments were examined, as per the ICF (WHO, 2001).   The variable related to the wheelchair environment was the need for a seating intervention. This variable was assessed using the 11-item Seating Identification Tool (SIT) (Miller et al. 2004). The SIT was dichotomized using a cut score of 2. Scores of 2 or more are indicative of a need for a seating intervention (Miller et al., 2004).  The social and physical environments have also been reported to influence self-efficacy. The perceived social support variable was assessed using the 6-item Interpersonal Support and Evaluation List (ISEL) (Cohen & Hoberman, 1983), and barriers in the physical environment were assessed using the home and community subscales in the Home and Community  78 Environment instrument (HACE) (Keysor et al., 2005). Data was also collected on the subject?s geographic location (i.e. British Columbia or Quebec).  4.2.3   Study protocol  After completing the socio-demographic information form, participants were administered the Mini Mental State Examination, which was used as a cognitive screen (i.e. individuals with scores <23 were excluded) (Folstein et al., 1975), and the WheelCon. Participants completed the remaining measures in a random sequence to minimize an ordering effect response bias.   4.2.4   Data analyses  Results from categorical variables were calculated as frequencies and percentages, and from continuous variables as means and standard deviations. Hierarchical multiple regression analyses were used to develop the self-efficacy prediction model, and test our hypotheses. Data from British Columbia and Quebec were combined for analyses because there was no significant mean difference in the self-efficacy dependent variable, as shown in table 4.1.  In this study?s analyses, a maximum model (i.e. the maximum number of variables to be included in the model) was identified, in addition to the variables for entry. The regression assumptions were then tested, followed by a regression modeling approach (Kleinbaum et al., 2008), as discussed below.   4.2.4.1   Maximum model  Using a moderate effect size (?2=0.15), and an alpha of 0.05, G*Power version 3.1 (available at http://www.psycho.uniduesseldorf.de/abteilungen/aap/gpower3/), determined that a sample size of n=123 would be necessary to model up to 11 independent variables, with a statistical power of 0.80.    79 Model inclusion criteria: Only variables with plausible relationships with the WC self-efficacy dependent variable were considered for entry into the model. To reduce the number of variables for entry to a maximum of 11, only those categorical variables with a significant mean difference in the dependent variable, determined using independent sample t-tests or one-way ANOVAs, were included for regression analyses, as well as those variables with at least a fair correlation (i.e. r?0.25 (Portney & Watkins, 2009) with the self-efficacy variable. If data were collected for only one variable in any particular domain, that variable was entered into the model (e.g. social support in the social environment), regardless of the magnitude of the correlation, or mean difference. If no variable met the entry criteria for any particular domain, one was selected for entry based on which had a stronger correlation with the WC self-efficacy variable, in order to test our hypotheses (e.g. physical barriers in the community in the physical environment). Furthermore, potential collinearity was identified by an intercorrelation of 0.70 between independent variables, and a variance inflation factor value greater than 10 (Kleinbaum et al., 2008). To minimize collinearity all continuous variables were mean centred, however, when collinearity was identified, the variable with the highest correlation with the dependent variable was entered into the model. Scatterplots of the bivariate data were examined for potential outliers. Data points greater than 1.5 times the variable?s interquartile range were considered cases that could influence the magnitude of the correlations (Portney & Watkins, 2009), or the significance of the mean difference. These cases were removed from this screening stage only, and the data reanalyzed to identify variables for inclusion into the model. All regression assumptions were tested.  4.2.4.2   Regression modeling   Variables were categorized according to the ICF framework, and a chunkwise hierarchical regression modeling strategy was used (Kleinbaum et al., 2008). The order of variable entry is consistent with other research examining the association of proximal factors first, followed by increasing distal factors (Martin-Ginis et al., 2012). More specifically, the health condition variables were entered first into the model (Model 1), followed by the personal factor variables (Model 2), and then the environmental factor variables, including those variables related to the wheelchair, social, and physical environments (Model 3). In each modeling stage, both forward  80 selection and backward elimination procedures were used to develop a robust model. The adjusted R2 is reported for each stage of modeling.  4.3 Results  One-hundred twenty four individuals were enrolled in this study. The mean age of the total sample was 59.67 years (SD=7.50), and 74 (59.7%) were male. The sample had 22.31 mean years (SD=16.05) of wheelchair use experience, and used their wheelchair 12.30 (SD=4.29) mean hours per day. Thirty-nine (31.5%) individuals required some form of assistance with using their wheelchair (e.g. mobility, transferring, set-up), and 22 (17.7%) received training to use their wheelchair after a rehabilitation program. The sample?s mean need for a seating intervention score was 1.98 (SD=1.69). The mean number of comorbidities was 2.69 (SD=2.40) out of a possible of 18 conditions. The mean self-efficacy with using a manual wheelchair score was 78.38 (SD=19.19) out of 100. Sample characteristic are further detailed in Chapter 2 (pages 46-47). Table 4.1 presents the descriptive statistics.  4.3.1   Maximum model  Health condition variables: Neither primary diagnosis nor the number of comorbidity variables met the modeling inclusion criteria. Therefore, the number of comorbidities variable was selected for entry into Model 1 because it was deemed a better indication of health status than an individual?s primary diagnosis.  Personal factor variables: Age and sex were included in Model 2, as were three variables related to wheelchair use, including:  daily hours using a wheelchair; formal training to use a wheelchair; and assistance needed with using a wheelchair. There was no significant mean difference in WC self-efficacy between individuals with different education levels, incomes, marital status, and employment/volunteer status. Therefore, these variables were not included in the model.    81 Environmental factor variables:  Wheelchair environment: The need for a seating intervention was the only variable related to the wheelchair environment, and therefore included in Model 3.   Social environment: Perceived social support was the only social environment variable, and therefore included in Model 3.   Physical environment: The barriers in the home and community environment variables had correlations below the minimum inclusion criteria, and the mean difference in self-efficacy by province was not significant. Because the magnitude of the correlation between the barriers in the community environment variable with self-efficacy was stronger than the barriers in the home environment variable, the former was selected for entry into Model 3. The physical barriers in the community variable was also selected because it was deemed to have greater generalizability than the geographic location variable.  Table 4.1 presents the correlations with, and mean differences in the WC self-efficacy dependent variable for the entire sample. The correlation matrix between the continuous independent and dependent variables is presented in appendix R. Reanalyzing the data after removing outlying data points had no influence on selection of variables for entry into the regression model. Furthermore, there was no violation of any regression assumption.             82 Table 4.1: Descriptive statistics and correlations with/mean differences in self-efficacy with using a manual wheelchair  Variable Total Self-efficacy  mean?sd/frequency (%) r/(mean difference) Body functions: Self-efficacy (0-100)  78.38?19.19  1 Health condition: Comorbidities (0-18) Neurological condition: Spinal cord injury Multiple sclerosis Stroke Other (Parkinson?s Cerebral palsy, brain injury) Non-neurological condition: Amputation Polio Arthritis  Other    2.69?2.40 97 (78.20) 60 (48.40) 16 (12.90) 12 (9.70) 9 (9.30)  27 (21.80) 9 (9.30) 5 (4.03) 4 (3.23) 9 (9.30)  -0.23*  (7.25)  Personal factors: Age Male Education (high school graduate) Income:?  <$30,000 Prefer not to answer Married Employed/volunteer Wheelchair Years experience Daily use (hours) Formal training Wheelchair assistance  59.67?7.49 74 (59.68) 110 (89.40)  43 (34.68)  21 (16.94) 59 (47.60) 46 (37.10)  22.31?16.05 12.30?4.29 22 (17.70) 39 (31.50)  -0.30* (14.75)* (-0.61)   (5.61) (4.16)  (0.80) (-5.28)  0.23 0.27* (-14.67)* (20.19)* Environmental factors: Wheelchair Need for seating intervention Social Social support (0-18) Physical British Columbia Home barriers (0-10) Community barriers (0-5)   68 (54.84)  14.48?3.71  74 (59.70) 1.10?1.22 1.06?0.85   (-10.70)*  0.12*  (1.78) 0.06 -0.14*  *included for modeling; ?mean difference from ?30,000; n=124 83 4.3.2   Regression modeling  Overall, nine variables were specified for regression modeling. In Model 1, WC self-efficacy was regressed on health condition. After running the stage-specific forward selection and backward elimination procedures, results indicated a significant negative association with number of comorbidities. Individuals with less comorbid conditions have higher WC self-efficacy than individuals with more comorbidity. The R2 increase was 5.0% (F1,121=6.78, p=0.01). Table 4.2 presents the regression results for all three models.  Next, the personal factor variables were evaluated in Model 2. Both age and sex remained in the model after running the regression procedures. Male sex and younger age were significantly associated with higher WC self-efficacy after controlling for the health condition variable. In addition, the three variables related to wheelchair use (i.e., daily hours using a wheelchair, formal training to use a wheelchair, and assistance needed with using a wheelchair) were statistically significant predictors of WC self-efficacy. The R2 increase resulting from these personal factor variables was 32.0% (F6,116=15.53, p<0.00).  In the evaluation of the environmental factor variables in Model 3, both the forward selection and backward elimination procedures confirmed that neither the social support nor physical barriers in the community variables contributed to the prediction of WC self-efficacy. These two variables were then dropped from the model. The variable related to the wheelchair environmental (i.e., need for a seating intervention) however, was statistically significant after controlling for health condition and personal factor variables, and further contributed to the explanation of the WC self-efficacy variance by 2.0% (F7,115=14.84, p<0.00).   The final model  (Model 3) included 1 health condition, 5 personal factors and 1 environmental factor (wheelchair-related). This model accounted for 44.0% of the WC self-efficacy variance.    84 Table 4.2:  Regression modeling to identify predictors of self-efficacy with using a manual wheelchair   Model 1 Model 2 Model 3  b SE ? 95% CI b SE ? 95% CI b SE ? 95% CI (Constant) 78.37 1.69  75.02, 81.72 77.75 1.90  73.98, 81.52 78.29 1.87  74.58, 82.00 Health Condition             FCI -1.84 0.71 -0.23 -3.24, -0.44 -0.09 0.60 -0.01 -1.27, 1.10 0.18 0.59 0.02 -1.00, 1.36 Personal Factors             Age     -0.48 0.19 -0.19 -0.86, -0.11 -0.46 0.19 -0.18 -0.82, -0.09 Sex     -10.85 2.81 -0.28 -16.41, -5.28 -10.02 2.77 -0.26 -15.50, -4.54 Daily hours in WC     0.87 0.32 0.20 0.25, 1.49 0.89 0.31 0.20 0.28, 1.50 Training with WC     9.56 3.55 0.19 2.54, 16.59 9.80 3.47 0.20 2.93, 16.66  Assistance with WC     -14.35 3.08 -0.35 -20.44, -8.26 -14.05 3.01 -0.34 -20.01, -8.10 Environmental Factors             SIT         -6.81 2.70 -0.18 -12.16, -1.47 adj R2 0.05 0.37 0.44 Bold=p?0.05; b=unstandardized coefficients; SE=standard error; ?=standardized coefficients; CI=confidence interval; adj=adjusted WC=wheelchair; FCI=Functional Comorbidity Index; SIT=Seating Identification Tool; Male sex, no training to use a wheelchair, no assistance with wheelchair, and British Columbia were coded as -0.50; n=124.           85 4.4 Discussion  In this study, health (Model 1), personal factor (Model 2), and environmental factor (Model 3) variables were examined that clinicians and researchers may easily access, and use to identify wheelchair users with lowered self-efficacy with using a manual wheelchair.  In Model 1, a statistically significant negative association between number of comorbidities, and self-efficacy was found. This evidence supports the hypothesis that health condition independently predicts WC self-efficacy. The finding is also corroborated by Bandura (1997) who posits decreasing abilities due to diminishing health status negatively affects self-efficacy. Although some research on predictors of self-efficacy in individuals with spinal cord injury demonstrates associations between health condition and self-efficacy that are similar to our findings, there is also evidence suggesting otherwise. For example, in a study examining predictors of self-efficacy in performing activities of daily living in individuals with a spinal cord injury (n=104), Horn et al. (1998) observed less severe neurological impairment at the onset of injury to significantly predict higher self-efficacy at a 12-month follow-up. Conversely, Pang et al. (2009) found no association between neurological severity and disease-management self-efficacy in another study of community-dwelling individuals with a spinal cord injury. One possible reason for this discrepancy may be due to differences in the type of self-efficacy assessed. In the 7-item measure used by Horn et al. (1998), all items inquired about self-efficacy to perform activities requiring physical ability, whereas Pang et al. (2009) administered the 6-item Self-efficacy for Managing Chronic Disease measure (Lorig et al., 1996), in which only one item inquires about self-efficacy to do tasks and activities. It is plausible that the self-efficacy measure used by Pang et al. (2009) was not specific to the ability limitations arising from the severity of the neurological impairment to see an association. The extent that health condition is associated with self-efficacy may therefore depend on the type of limitations arising from the health condition, and how specific the self-efficacy measure is to those limitations. The use of self-efficacy measures that are situation specific result in greater predictability, and is consistent with theory (Bandura, 1997). In this study, the items in the WheelCon assess an individual?s self-efficacy pertaining to activities requiring physical and social abilities, and are specific to the physical and social limitations arising from comorbid conditions, demonstrated by our findings.  86 In Model 2, evidence was found in support the hypothesis that after controlling for health condition variables, personal factor variables predict self-efficacy with using a manual wheelchair. More specifically, after controlling for number of comorbidities, older age, being female, requiring assistance with wheelchair use, no formal training to use a wheelchair, and minimal hours of daily wheelchair use were found to be associated with lowered WC self-efficacy. In addition, the number of comorbidities variable failed to remain statistically significant after entering the age and sex variables into the model. There are several reasons for these findings. In terms of age, individuals who are older may have more health issues and physical limitations (Statistics Canada, 2006) than younger individuals, thus resulting in lower beliefs in their abilities (Bandura, 1997). In addition, minor issues such as aches and pains are more likely to be attributed to perceived declines in abilities by older than younger individuals, which may also lead to lowered self-efficacy attributed to old age (Bandura, 1997). Findings from this research are consistent with previous studies. For example, in the study of individuals with a spinal cord injury, Horn et al. (1998) demonstrate older individuals are more likely to report lower self-efficacy to perform activities of daily living than younger individuals. Similarly, in a large study of community-dwelling individuals (n=703), at least 40 years of age, with a variety of health conditions, physical activity self-efficacy was shown to decline with older age (Smith Anderson-Bill, Winett, Wojcik & Williams, 2011). That is, individuals in their 60s were observed to report significantly lower self-efficacy than individuals in their 40s and 50s.   The association between aging and declining health is likely the reason why the number of comorbidities variable failed to remain statistically significant after entering the personal factor variables into the model. The slightly stronger bivariable correlation between the number of comorbidities variable and age than the correlation between number of comorbidities and self-efficacy supports our interpretation. Other reports showing chronic conditions, such as arthritis, rheumatism, and high blood pressure, to be more prevalent in older adults than younger individuals (Statistics Canada, 2006), is indicative of declining health with aging, and also supports our reasoning. Therefore, countering the negative effects of diminishing health in older wheelchair users is important because health problems that create functional limitations potentially threaten participation and mobility. Self-efficacy may be a protective factor (Bandura,  87 1997). Individuals with higher WC self-efficacy may work harder and persevere when faced with difficulties to overcome limitations brought on by older age.  Sex also significantly predicted WC self-efficacy, with being female predictive of lower self-efficacy relative to being male. This finding is in contrast with Fleiss-Douer et al. (2012) in their reporting of no difference in self-efficacy with wheeled mobility between male and female paralympians. Rushton et al. (2013) also described no difference in self-efficacy with using a wheelchair by sex, but do report a significant age-by-sex interaction. This interaction may suggest self-efficacy diminishes with aging at a faster rate in females than in males.   In other studies, the extent to which self-efficacy differs by sex has to do with the specific form of self-efficacy in question. Those inquiring about beliefs having to do with physical abilities or tasks that are perceived to be more masculine in nature than sex neutral, the larger the difference in self-efficacy, with males reporting more self-efficacy than females (Lirgg, 1991). Therefore, because this study?s findings indicate that males have higher WC self-efficacy than females, wheelchair use may be perceived as a more masculine task requiring greater amounts of physical ability. However, the notion that the differences disappear when females have a high ability to use their wheelchair (Fleiss-Douer et al., 2012), is indication that females would benefit from wheelchair skills training. In fact, previous research demonstrates the efficacy of wheelchair skills training at improving both WC self-efficacy (Sakakibara et al., 2013b), and wheelchair skills (Coolen et al., 2004; Kirby et al., 2004b; MacPhee et al., 2004; Best et al., 2005).  Two other studies have examined self-efficacy, using the WheelCon, in younger, healthy, community-dwelling wheelchair users (Phang et al., 2012; Rushton et al., 2013). The individuals in those studies reported higher self-efficacy with using a wheelchair than individuals in this study. That the samples in those studies were younger, and had a higher proportion of male subjects than the sample used in this study may provide additional support for our findings that younger age and male sex are predictive of higher WC self-efficacy than older age and being female.    88 Also in Model 2, three personal factor variables related to wheelchair use were statistically significant predictors of WC self-efficacy. More daily use of the wheelchair, and having received training to use the wheelchair are predictive of higher WC self-efficacy, whereas requiring assistance to use the wheelchair is associated with lowered self-efficacy. These findings are not surprising when considering the focus of the dependent variable on self-efficacy specific to wheelchair use. It seems intuitive that individuals who spend more time using their wheelchair, and received formal training to use their wheelchair would have higher WC self-efficacy. The more time spent using the wheelchair, and training to use the wheelchair properly may lead to more positive experiences, which are then reflected upon to enhance self-efficacy. Results from this study also suggest that assistance needed with using the wheelchair is significantly associated with lowered self-efficacy. It is not unexpected for individuals who have difficulties with using their wheelchair to also report lower WC self-efficacy than individuals who have no such difficulties.   In the final model (Model 3), one variable related to the wheelchair was a statistically significant predictor of WC self-efficacy. This result provides evidence in support of the hypothesis that environmental factor variables predict self-efficacy, after controlling for health condition, and personal factor variables, in older, community-dwelling manual wheelchair users. Interpretation of this finding indicates that individuals who have a need for a seating intervention are at risk of lowered WC self-efficacy than individuals who do not have such a need. It also seems intuitive that individuals who do not have a need for a wheelchair seating intervention may have higher WC self-efficacy than individuals with such a need. For example, those who experience discomfort from sitting in their wheelchair, or have been at risk of tipping their wheelchair due to an improper set-up may have lowered WC self-efficacy due to uncertainties related to their wheelchair. Therefore, a better fitting wheelchair may lead to more positive experiences, and thus higher WC self-efficacy. Research on sports-related self-efficacy supports this speculation. Modifications made to sporting equipment, such as using a tennis racket with a larger head (Pellet & Lox, 1998) or lowering the height of a basketball net (Chase, Ewing, George & Lirgg, 1994), to facilitate improved performance have been shown to enhance self-efficacy.     89 It is interesting that years of wheelchair use experience, perceived amount of social support, and physical barriers in the community were not predictors of self-efficacy with using a manual wheelchair. The fact that this study?s sample is older may explain the low association between the years of experience variable and WC self-efficacy, which is contrary to another report of younger wheelchair users (Rushton et al., 2013). For example, years of experience may be important for the self-efficacy in younger individuals, but when individuals begin experiencing functional declines due to aging, the perception of declining abilities may have differing effects on self-efficacy. That is, not all individuals will perceive their ability issues the same (Bandura, 1997). For some, diminishing abilities will have a profound effect on self-efficacy, especially if declines are unexpected, and for others, the effect will be minimal and viewed as the natural course of life. Therefore, the association between years of wheelchair use experience and WC self-efficacy may become more random with aging.   The lack of an association between WC self-efficacy and perceived social support may be due to the global measure of social support used in this study. The 6-item ISEL captures both emotional (e.g. people to talk to), and physical (e.g. people to help with daily activities) forms of support. Whereas physical support may limit opportunities to perform tasks that individuals are capable of, which then act to compromise self-efficacy, it is likely that emotional support has positive influences (Pang et al., 2009; Rushton et al., 2013). Because both forms of support have opposing effects on self-efficacy may be the reason why no association was found.  Finally, the physical barriers in the community variable did not significantly predict self-efficacy, and is contrary to other research on the effects of community environments on self-efficacy (Morris, McAuley & Motl, 2008). The reason for this finding may be that individuals reported few physical barriers in the community, along with a wider range of WC self-efficacy scores. Therefore, it is plausible that community accessibility is improving (Laplante & Kaye, 2010), and that there are other more important variables influencing self-efficacy with using a manual wheelchair. However, further investigation on the association between community environments and WC self-efficacy with using a manual wheelchair is warranted.    90 Limitations This study has several limitations. First, the sample was comprised of volunteers and therefore it may under- or over-represent particular groups within the population. In addition, through the use of a volunteer sample, there is a lack of information on potential reasons why individuals chose not to participate. Although a random sampling technique would have been preferred, volunteers were recruited to maximize our sample size. This sampling technique may limit the generalizations of the study?s results. Next, the findings are also limited to individuals 50 years and older who have at least 6 months experience with using a manual wheelchair. In addition, due to the study?s cross-sectional research design, conclusions cannot be made regarding the direction of the observed associations or causality. However, because the purpose of the study was to identify older individuals in greatest need of self-efficacy enhancements, and not on determinants or consequences of self-efficacy, the research design was appropriate. Furthermore, the use of self-report measures may have been affected by a social desirability bias, and/or led to a common methods bias. As a result, individuals may have reported artificially high self-efficacy estimates, which may have biased the results. Finally, there may be health, personal factor, and/or environmental factor variables that were left out of the analyses. Despite this, the modeled variables accounted for 44.0% of the WC self-efficacy variance.  4.5 Conclusion  In this study, health, personal, and environmental factor variables were identified that clinicians and researchers may use to help identify wheelchair users either at risk of or with lowered self-efficacy with using a manual wheelchair. Manual wheelchair users of older age, and female sex, with comorbid conditions, who require assistance to use their wheelchair, use their wheelchair for less hours throughout the day, received no formal training to use a wheelchair, and have a greater need for a seating intervention are at risk of lowered WC self-efficacy. These subgroups of older wheelchair users may have the greatest need for targeted interventions to enhance their WC self-efficacy, which may lead to improved participation, and better life-space mobility.    91 CHAPTER 5: Rasch analyses of Wheelchair Use Confidence Scale  5.1 Introduction  In previous chapters, the positive associations between self-efficacy with using a manual wheelchair and both participation frequency and life-space mobility in older, community-dwelling wheelchair users, was illustrated, as were the underlying mechanisms through which WC self-efficacy operates. In addition, predisposing health, personal, and environment contextual factors were examined and identified that may help to identify individuals at risk for having lowered WC self-efficacy, and who may benefit the most from targeted self-efficacy interventions. These findings provide a foundation for clinical research on the development and testing of efficacy-enhancing interventions for those individuals in greatest need. However, because the best research and clinical evidence is predicated upon accurate measurement, and because the Wheelchair Use Confidence Scale (WheelCon) has only recently been developed, further work to establish its properties using contemporary measurement methods is warranted.  5.1.2   The Wheelchair Use Confidence Scale  The 65-item WheelCon version 3.0 is a newly developed measureiii that assesses the strength of self-efficacy with using a manual wheelchair in in six conceptual areas including: 1) the physical environment; 2) activities performed; 3) knowledge and problem solving; 4) advocacy; 5) social situations; 6) and emotions (Rushton et al., 2011). The measure?s items were developed through interviews with wheelchair users and rehabilitation professionals, and its content validity was established through a multi-stage Delphi process also involving wheelchair users and clinicians (Rushton et al., 2011). For each item, individuals are asked, ?As of now, how confident are you ?.? Each item is rated on a 0 (not confident) to 100 (completely confident) point response scale. To obtain an overall score the responses are summed and divided by the number of items. Higher scores indicate a greater strength of self-efficacy. Measurements derived using the WheelCon are both reliable and valid among manual wheelchair users (Rushton et al., 2013). 	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?iii The terms ?measure? and ?test? are used interchangeably.  92 Despite evidence supporting the WheelCon?s measurement properties, and its effective use in research settings (Phang et al., 2011; Sakakibara et al., 2013a; Sakakibara et al., 2013b), further evaluation of the WheelCon?s measurement properties is necessary for several reasons. First, during the development of the WheelCon, items were created that inquire about beliefs pertaining to both physical (e.g. move over carpet) and non-physical (e.g. solve how to get to destination when there is an unexpected detour) abilities to use a wheelchair. Although these items are conceptually different they are used together to derive a total score. Quantitative investigation into the WheelCon?s dimensionality has yet to be performed, however, it may reveal that more than one form of self-efficacy is being assessed. By identifying different dimensions in the WheelCon and deriving separate subscale scores may result in improved measurements. Next, because the WheelCon measures have high internal consistency reliability (Cronbach alpha=0.92) (Rushton et al., 2013), there may be redundant items that could be eliminated. Identifying and eliminating redundant items will reduce the administrative and responded burden, which may in turn facilitate the measure?s use, both clinically and in research. Furthermore, the use of a 101-point response format with the WheelCon may be resulting in bias because evidence shows individuals, especially older adults, have difficulties using response formats with many response options (Lachman et al., 1998; Arnadottir, Lundin-Olsson, Gunnarsdottir & Fisher, 2010). An inability to differentiate between adjacent options leads to respondent bias arising from increases in subjectivity (Embretson & Reise, 2000). A shortened response format therefore may improve measurement precision. Finally, the WheelCon was developed using principles of Classical Test Theory (CTT). A discussion of the limitations of CTT in relation to Item Response Theory (IRT) follows, which provides further rationale for more research into the measurement properties of the WheelCon (Streiner & Norman, 2008; de Ayala, 2009).  5.1.3   Important differences between CTT and IRT  Although CTT provides a useful framework for test development and evaluation (Konicki Di lorio, 2005), contemporary methodological approaches, such as IRT, and in particular Rasch analyses, are increasingly being used in rehabilitation research to address some of limitations with CTT (Konicki Di lorio, 2005; Streiner & Norman, 2008; de Ayala, 2009).   93 5.1.3.1   CTT sample dependency versus IRT (Rasch) invariance  The circular dependency between the test and the sample being administered the test is an inherent limitation with CTT. That is, the test scores depend on the ability/attribute of the respondents, while the amount of the respondents? ability/attribute is dependent upon their performance on the test (Streiner & Norman, 2008). As a result, the properties of a test change when administered to different samples, and the ability/attribute of the samples change when administered different tests measuring the same thing (Streiner & Norman, 2008). Conversely, a key property of Rasch models is ?sample invariance?iv. This refers to the ability of the model to separate the item/test parameters (i.e. difficulty of the measure) from the sample completing the assessment (Streiner & Norman, 2008). Therefore, when the test is administered to equivalent samples from the same population, the item parameters should be the same (Streiner & Norman, 2008). Similarly, the ability/attribute of the sample should be the same when administered different items of equivalent difficulty (Streiner & Norman, 2008). Because measurement tests should be independent of what they are measuring, sample invariance is a key property of Rasch models.   5.1.3.2   CTT test versus IRT item focus  Classical Test Theory focuses primarily on providing information at the test level. As a result, the raw scores are used to establish test-level estimates for both reliability and the standard error of measurement (SEM). These single estimates are assumed to be constant for the entire range of scores, which in many instances is not true. For example, if the sample scores are normally distributed, the SEM would be smallest around the mean and increase with higher and lower scores (Streiner & Norman, 2008), and the inverse would be true for reliability estimates throughout the range of scores. In IRT, because the focus is at the item-level, SEM and reliability coefficients may be estimated for all scores along the construct?s continuum (Streiner & Norman, 2008). As a result, specific content areas on a measurement test may be identified that require improvements, and/or populations may be identified that the test assesses most reliably. Also, 	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?iv Note: some believe that only the 1-parameter Rasch model holds the property of invariance, and that by evaluating additional parameters (i.e. discrimination, and guessing) with the use of 2- and 3-parameter IRT models, the property of sample invariance is lost. For purposes of this paper the terms IRT and Rasch are used synonymously.  94 because of the item-level focus in IRT, measurement tests may be shortened to improve efficiency, without a loss in reliability or increases in error. In fact, through the use of IRT, measurement tests may be shortened with improved reliability (Velozo, Seel, Magasi, Heinemann & Romero, 2012; Del Toro et al., 2011).    5.1.3.3   CTT ordinal versus IRT interval measurement  In measurement tests developed using principles of CTT, the level of measurement is assumed to be interval or ratio, when in fact they are ordinal in nature (Streiner & Norman, 2008). This has implications on the interpretation and meaning of total scores. For example, it cannot be certain that the distance between raw scores is consistent throughout the entire range of total scores; a change in score from 35 to 40 may not necessarily mean the same thing as a 5 point change from 70 to 75. On the other hand, through the use of IRT, raw scores are converted into an interval level log-of-odds (i.e. logit) continuum with a mean of 0 and a standard deviation of 1 (Streiner & Norman, 2008). As a result, there is equidistance between changes in logits by the same amount throughout the entire range of scores. Because the range of logit scores may be converted into standardized scores, the ordinal raw scores derived from the response format may be directly transformed into interval level scores ranging from 0 to 100.  Overall, IRT (in particular Rasch models due to the property of sample invariance) offer several advantages over CTT in developing and evaluating the measurement properties of self-report tests. Therefore, the purpose of this study is to evaluate the measurement properties of the WheelCon using Rasch analyses. In doing so, we will compare the functioning of the WheelCon?s 101-point response format to shortened response formats; examine the dimensionality of the WheelCon; identify items not conforming to the Rasch model that could be considered for elimination; identify the overall item content of the WheelCon, and redundant items that could be considered for elimination; convert the raw ordinal total scores into standardized interval total scores; and determine the standard errors of measurement and reliability for the entire range of standardized scores, as well as the test?s internal consistency reliability.   95 It is hypothesized that i) a shortened response format will function better than the original response format. It is also hypothesized that ii) the WheelCon with a shortened response format will result in more than one dimension with fewer items; good internal consistency reliability (Cronbach alpha?0.70) throughout the range of total scores; and a good to excellent correlation magnitude (Pearson r?0.75) with the original WheelCon.   5.2 Methods  5.2.1   Participants  Community-dwelling manual wheelchair users, 19 years of age and older, were targeted to participate in this study. Individuals had at least 6 months experience with using a manual wheelchair on a daily basis, and communicated in English or French.   The cross-sectional data from this study (study 1) were combined with data from another study (study 2 = Rushton et al., 2013) with the same inclusion/exclusion criteria. The samples from both studies were drawn from British Columbia (BC), Ontario, Quebec, and Nova Scotia, Canada between 2010 and 2012.   Participants were recruited using letters of information sent by clinicians and/or wheelchair vendors. Study information was also provided to advocacy and community groups. In Quebec, participants were recruited from two rehabilitation institutes in Quebec City and Montreal. Furthermore, individuals also had the option to complete the English questionnaires online. The ethics boards from all relevant institutions approved this study.  5.2.2   Data analyses  Descriptive statistics were used to characterize the sample on demographic variables common to both studies. Results from categorical variables were calculated as percentages, and from continuous variables as means and standard deviations. The Barthel Index (Gompertz et al. 1994) was used to assess functional independence to perform activities of daily living. The descriptive  96 statistics from both studies were visually inspected as a means to identify differences between samples.   Different versions of the WheelCon were used in each of the studies. The 65-item version 3.0 was used in study 1 (sample recruited from BC and Quebec), and the 63-item version 2.4, was used in study 2 (sample recruited from BC, Ontario, and Nova Scotia (Rushton et al., 2011)). Both versions of the WheelCon are presented in appendices M and N. As part of the version 2.4 revision, three items were removed, and five new items were added to improve content coverage. Furthermore, of the 60 similar items in both versions, some items in version 3.0 were modified to provide more contextual detail.   To ensure the responses from each of the 60 items could be combined for analyses, the test scores from 20 individuals who participated in both studies 1 and 2 were examined. After examining a scatterplot of total scores from both studies, the data from one individual was deemed to be an outlier due to a self-efficacy score that had more than doubled (48.68 vs. 98.42) between WheelCon administrations. After removing these data, a paired-sample t-test was used to compare the mean total scores derived from the 60 similar items, and a Pearson correlation was used to derive the magnitude of the association between the total scores. A non-significant mean difference [mean difference = 2.67 (SD=9.00); t(18) = 1.29, p=0.21)], and the excellent strength of the association between the scores (r=0.80) (Portney & Watkins, 2009) were indicative that the tests are measuring the same construct, and that the datasets could be combined for analyses. Subsequent analyses only included the WheelCon version 3.0 data from the 20 individuals who participated in both studies, and on the 60 similar items up until principal components analysis discussed below. Figure 5.1 displays the overall analytical process.        97 Figure 5.1: Analytical process used to perform Rasch analyses on the WheelCon                      Study	 ?1	 ?	 ?(n=146)	 ?65-??item	 ?WheelCon	 ?v3.0	 ?Study	 ?2	 ?	 ?(n=74)	 ?63-??item	 ?WheelCon	 ?v2.4	 ?	 ?WheelCon	 ?v2.4	 ?&	 ?3.0	 ?	 ?60	 ?similar	 ?items 	 ?	 ?	 ?	 ?WheelCon	 ?v3.0	 ?5	 ?new	 ?items	 ?	 ?Principal	 ?Components	 ?Analysis	 ?	 ?	 ?Stage	 ?1:	 ?(n=220)	 ?	 ?60	 ?items	 ?	 ?Stage	 ?2:	 ?(n=146)	 ?5	 ?items	 ?	 ?	 ?Mobility	 ?ef?cacy	 ?46-??items	 ?Self-??management	 ?ef?cacy	 ?25-??items	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?Rasch	 ?Rating	 ?Scale	 ?Model	 ?(n=220)	 ?	 ?Out?it	 ?analysis:	 ?22	 ?items	 ?removed	 ?	 ?	 ?	 ?	 ?Out?it	 ?analysis:	 ?8	 ?items	 ?removed	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?Rasch	 ?Rating	 ?Scale	 ?Model	 ?(n=220)	 ?	 ?In?it	 ?analysis:	 ?10	 ?items	 ?removed	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?In?it	 ?analysis:	 ?9	 ?items	 ?removed	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?Rasch	 ?Rating	 ?Scale	 ?Model	 ?(n=220)	 ?	 ?Item	 ?redundancy:	 ?1	 ?item	 ?removed	 ?	 ?	 ?	 ?Item	 ?redundancy:	 ?0	 ?items	 ?removed	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?Mobility	 ?ef?cacy	 ?13-??items	 ?Self-??management	 ?ef?cacy	 ?8-??items	 ?WheelCon	 ?Short	 ?Form	 ?21-??items	 ? 98 5.2.2.1   Rasch Rating Scale Model  Rasch models are 1-parameter (item difficulty) logistic model (Rasch, 1980). In Rasch analyses, the probability of an item response is derived from both the ability/attribute of the person (i.e. total scores converted into logits) and difficulty of the item (de Ayala, 2009). That is, the more self-efficacy a person reports relative to the self-efficacy the item is targeting, the higher the probability the person will respond with a high self-efficacy, and vice-versa. Item difficulty and person ability/attribute parameters are both located on the interval logit continuum discussed above. Plus/minus 3.00 logits represents 99.7% of the construct of interest (Streiner & Norman, 2008). Items located closer to +3.00 logits represents more difficult items, or rather items that many people respond to with lower self-efficacy estimates. Similarly, individuals with abilities/attributes located closer to +3.00 logits represents people who will most likely respond to each item with higher self-efficacy than individuals located near -3.00 logits.   Because the WheelCon uses the same response format for each item, the Rasch Rating Scale Model (Andrich, 1978) was used. This model is used with datasets derived from polytomous ordered response scales (i.e. response scales with more than two response options that are ordered by difficulty), and assumes that increases in the amount of the construct needed to move up adjacent response options is the same for all items (Andrich, 1978). The Joint Maximum Likelihood Estimation method was used to estimate the item difficulty parameters using the Winsteps version 3.75 software (Linacre, 2009).   5.2.2.2   Response format category collapsing   In Bandura?s guidelines for developing self-efficacy measures, the use of either 0 to 100 or 0 to 10 point response formats are advocated (Bandura, 2006). In order to remain consistent with the guidelines, the 0 to 100 point response format, of which was used to collect data, was compared to 0 to 10 point response formats. Furthermore, because both the 101- and 11-point response formats share similar metric properties, these comparisons were made instead of response formats with fewer options.   99 The original 101-point format was collapsed into two 11-point response formats for testing. In the first modified format, zero was retained as a distinct option, and the remaining 100 categories were evenly grouped (i.e.1 to 10 = 1, 11 to 20 = 2, and so on). In the second modified format, 100 was retained as a distinct option, and the remaining options evenly grouped (i.e. 0 to 9 = 0, 10 to 19 = 1, and so on). Zero and 100 were examined as unique categories in each of the collapsed formats because both response options have conceptually distinct meanings (i.e. 0 = no self-efficacy, and 100 = complete self-efficacy). The collapsing methods were also selected because of their ease for clinical interpretation. The original dataset was then rescored according to the modified response formats, and the functioning of each analyzed individually.  5.2.2.3   Response format functioning   Two criteria were used to assess the functioning of each response format:  1) The Rasch Rating Scale Model?s item-by-response option outfit statistics (mean squares) were examined. Fit statistics provide information on how much the observed responses deviate from the model?s expected responses (de Ayala, 2009). Outfit statistics provide information on unexpected responses to items with self-efficacy level ?far? (i.e. outlier) from an individual?s estimated self-efficacy (Linacre, 2009). Fit statistics follow a chi-square distribution with a range from 0 to infinity, with an expectation of 1. Values substantially greater than 1 reflect more variability than expected, and values less than 1 reflect less variability than expected (de Ayala, 2009). In following Linacre?s guide to optimizing rating scale effectiveness (Linacre, 2002), outfit statistics greater than 2.0 were considered indicative that the response option was being used unexpectedly.   2) The ordering of the response options was also inspected to ensure higher scores equate to higher self-efficacy estimates (Linacre, 2002). Each response option?s average measure value was examined to determine appropriate ordering. The average measure is calculated as the mean ability/attribute of the respondents in a particular response option averaged for all items (Linacre, 2002). A larger value for increasing response options is indication that higher options manifest higher self-efficacy estimates.  100 The response format with the fewest poor-fitting and unordered response options was deemed the most appropriate to use for further analyses.   5.2.2.4   Evaluating Rasch assumptions   Three assumptions underlie Rasch models: 1) unidimensionality; 2) local independence; and 3) monotonicity.  1) Unidimensionality means that all items in a measure are assessing the same construct (de Ayala, 2009). To check for unidimensionality of the WheelCon, Principal Components Analysis (PCA) was used to identify the maximum amount of variance in the data that is explained with the fewest independent dimensions (Norman & Streiner, 2008). Principal Components Analysis is a technique used to categorize items in a measure according to their strength of relationship with other items (Norman & Streiner, 2008). Strongly correlated items are grouped together into ?dimensions? that represents the construct the measurement test is capturing (Norman & Streiner, 2008).   Principal Components Analysis was used in two stages. The first was to identify factors resulting from the 60 similar items from the two versions of the WheelCon, and the second on the 5 new items in WheelCon version 3.0. Prior to conducting PCA, the presence of at least one dimension was ensured by examining the correlation matrix of raw data, and confirming more than 20% of the inter-item correlations were above 0.30 (Norman & Streiner, 2008). The Kaiser-Meyer-Olkin Measure of Sampling Adequacy value >0.70 was also used as a statistical indicator of adequate inter-item correlations (Norman & Streiner, 2008).   Items were specified to load onto a dimension only if they had a minimal loading value of 0.35. This loading value was calculated using the formula 5.15/?(n-2) , where 5.15 is twice the 1% level of significance assuming a normal distribution (Norman & Streiner, 2008). The 1% level of significance is doubled because the dimension loadings are up to twice the regular correlations (Norman & Streiner, 2008). An item with a loading value less than 0.35 was indication that it does not share a significant amount of variance with any dimension. Varimax orthogonal rotation  101 was then used to simplify the interpretation of loadings (e.g. reducing factor complexity, and clarifying the magnitude of the loadings) (Norman & Streiner, 2008). Items not meeting the minimal loading value, or that loaded onto a dimension that did not make conceptual sense were considered for elimination (Norman & Streiner, 2008). Furthermore, items that were factorially complex, (i.e. items loading on more than one dimension with the strongest loading on the dimension with the least conceptual sense) (Norman & Streiner, 2008) were retained in all dimensions for Rasch analyses.   After running PCA on the 60 similar items, the measure was considered to be unidimensional if the first dimension explained at least 40% of the variance (Norman & Streiner, 2008), and if no other dimension had an eigenvalue greater than 3 (Linacre, 2009). The scree plot was visually examined to help determine the number of dominant dimensions extracted.   The second PCA examined the dimensionality of the 5 new items with data from individuals in study 1 (i.e. n=146 individuals who completed the 65-item WheelCon version 3.0). These items were then allocated to the dimension(s) resulting from the first PCA based on both statistical and conceptual rationale.  2) Local independence means that a person?s response to an item is only due to his/her ability/attribute that is being measured (de Ayala, 2009). If this assumption were not true, then something other than self-efficacy would be influencing the responses. In such cases, the test would be multi-dimensional. Therefore, satisfying the unidimensional assumption was one criteria used to confirm local independence (de Ayala, 2009). The correlations of the residuals between each item were also examined. Any inter-residual correlation greater than 0.20 was indication that responses to one item may be determined by responses to another item, and thus a violation of local independence (Velozo et al., 2012).   3) The monotonicity assumption means that the proportion of people selecting a particular option on a response format increases for those with higher ability/attribute levels (de Ayala, 2009). This assumption was considered satisfied after identifying the best functioning response format, as discussed above.   102 Upon meeting the Rasch assumptions the dataset was entered into Winsteps version 3.75 (Linacre, 2009) software to develop the Rating Scale Model.   5.2.2.5   Assessing item fit and item elimination  Item infit and outfit statistics were used to assess the fit between the Rasch Rating Scale Model and data (i.e. how well the model predicts the item responses) (Linacre, 2009). Infit statistics provide information on unexpected responses to items that have a difficulty level ?near? one?s reported self-efficacy estimate, whereas outfit statistics provide information on unexpected responses to items with self-efficacy levels ?far? from one?s reported self-efficacy, as described above in section 5.2.2.3 (Linacre, 2009).  	 ?A range of acceptable infit and outfit statistics was calculated based on the sample size, using the formulas 1?2/?N and 1?6/?N, respectively (de Ayala, 2009). As an additional criterion, the standardized fit statistics (ZSTD) were also examined to identify misfitting items. Standardized fit statistics are the t-standardized mean-square fit statistics that approximate a normal distribution (Linacre, 2009). Therefore, items with infit and outfit statistics outside the ranges of 0.87 to 1.13, and 0.60 to 1.40, respectively, and/or ZSTDs greater than ?2 were considered for elimination.   In following Linacre?s guidelines for the evaluation of fit statistics (Linacre, 2009), items with outfit statistics outside the acceptable range were first examined. Because these items elicit unexpected outlying responses (e.g. unexpectedly high self-efficacy ratings to difficult items by individuals with lower self-efficacy), they were considered for elimination first. After removing the items with misfitting outfit statistics, the analyses were re-ran, and the infit statistics of the remaining items were examined. Similar to items with misfitting outfit statistics, items with infit statistics outside the acceptable range were considered for elimination.  After assessing the item fit, item redundancy was examined for using the item-difficulty hierarchy. Items were considered redundant when there was overlap in both item difficulty ?  103 SEM and conceptual content. When there was redundancy, one item was considered for elimination after discussion with the WheelCon developers (Rushton et al., 2011).   After eliminating the misfitting and redundant items, another Rating Scale Model was developed using the remaining items. Because it is common for items to misfit in the more model-fitting context, Linacre (2009) advises not to eliminate items mechanically or else there will be no items left. Therefore, for the final model, item infit and outfit statistics in the range of 0.50 and 1.50 were considered to be adequate. This range is also considered productive for measurement (Linacre, 2009). For items in the final model, the difficulty parameters, SEMs, as well as the 99% confidence interval around the item difficulty parameters were calculated.  5.2.2.6   Reliability and validity   Both the SEM and internal consistency reliability were estimated for the entire range of total scores. In Rasch analyses, the SEM are based on different scales (logits) than those estimated in CTT (raw score), but the coefficients are interpreted the same (Wilson, Allen & Corser, 2006). The SEM derived via Rasch analyses may be converted into coefficients similar to Cronbach alpha by using the formula, 1/1+logit SEM2 (M?sse, Heesch, Eason & Wilson, 2006). Cronbach alpha estimates <0.70 were considered insufficient.  At the test level, the item separation statistics were derived. Item separation indicates how well the items can be differentiated by respondents (Wright & Stone 1999). Separation statistic values of at least 3 are indication of good item spread (Linacre, 2009). The Cronbach alpha, and the Pearson correlation coefficient of the shorted measures with the original WheelCon were also calculated. The mean difficulty of the measure was also compared to the mean strength of the WC self-efficacy of the sample. For ease of interpretation, the logit scale was transformed into a 0 to 100 standardized scale.      104 5.3 Results  5.3.1   Participants  Samples from two independent studies were used in the analyses. Table 5.1 presents the sample characteristics from each study, in addition to the combined sample (n=220). Visual inspection of the data indicates that the sample from study 1 was older and had more experience with using a wheelchair than the sample from study 2. In addition, more individuals in study 1 were married and university graduates. The mean age of the combined sample was 54.20 (SD=13.00) years, and the mean years of experience with using a wheelchair was 17.90 (SD=14.70). One-hundred-thirty-nine (63.2%) individuals were male, and 115 (52.23%) individuals had a spinal cord injury, 25 (11.4%) had multiple sclerosis, 19 (8.6%) with a lower extremity amputation, and 17 (7.7%) with stroke.   5.3.2   Response format functioning   In the original 101-point response format, eight (7.92%) response options had an outfit statistic greater than 2.00, and 34 response options were out of order as indicated by the average measure value.   In both 11-point response formats, one (9.09%) response option had an outfit statistic exceeding 2.00, and all response options were in order, thereby indicating that increasing response options manifest higher self-efficacy estimates throughout the scale. Therefore, both 11-point response formats were determined to function better than the 101-point format. Furthermore, the 11-point format with 100 as a unique category was determined to function better than the other 11-point format with 0 as a unique category because the misfitting outfit statistic was slightly closer to the critical value of 2.00 (2.56 vs. 2.62, respectively). All data were then recoded using the 11-point format with 100 as a unique category for subsequent analyses. Table 5.2 shows the outfit statistics, and average measure values for both 11-point response formats.    105   Table 5.1:  Sample characteristics  Study 1 (n=146) Study 2  (n=74) Combined (n=220)  mean?sd/frequency (%) Age 57.07?10.41 48.46?15.49 54.2?13.0 Wheelchair use experience (years)  20.43?15.74  12.99?10.90  17.9?14.7 Functional independence  (0-20)  14.32?2.94  14.09?5.36  14.6?2.8* Sex: Male Female  89 (60.96) 57 (39.04)  50 (67.57) 24 (32.43)  139 (63.18) 81 (36.82) Diagnosis: Spinal cord injury Multiple sclerosis Stroke Lower extremity amputation Other (spina bifida, Cerebral Palsy, Parkinson?s disease, arthritis)  77 (52.74) 17 (11.64) 14 (9.59) 9 (6.16)   29 (19.86)  38 (51.35) 8 (10.81) 3 (4.05) 10 (13.51)   15 (20.27)  115 (52.27) 25 (11.36) 17 (7.73) 19 (8.64)   44 (20.00) Married/Common law 69 (47.26) 29 (39.18) 98 (44.55) Education: Some high school Graduated high school Some university Graduated university Post-graduate studies Other  15 (10.27) 26 (17.81) 32 (21.92) 61 (41.78) 9 (6.16) 3 (2.05)  5 (6.76) 29 (39.19) 0 (0.00) 17 (22.97) 4 (5.41) 19 (25.68)  20 (9.10) 55 (25.00) 32 (14.55) 78 (35.45) 13 (5.91) 22 (10.00) *n=217  Table 5.2:  11-point response format outfit statistics and average measure values  0 unique category 100 unique category Response option Average measure Outfit statistic Average measure Outfit statistic 0 -0.23 2.62* -0.26 2.56* 1 -0.19 1.53 -0.21 1.65 2 -0.06 1.59 -0.08 1.53 3 -0.02 1.07 -0.06 0.87 4 0.04 1.12 0.02 1.15 5 0.17 1.26 0.14 1.24 6 0.21 0.95 0.18 0.94 7 0.31 0.70 0.30 0.75 8 0.43 0.63 0.41 0.71 9 0.62 0.79 0.63 0.79 10 1.01 1.00 1.03 0.98 *response option being used unexpectedly  106 5.3.3   Evaluating Rasch assumptions   1) Unidimensionality (60 items) ? Examination of both the item correlation matrix, and the Kaiser-Meyer-Olkin Measure of Sampling Adequacy (MSA = 0.92) indicated sufficient inter-item correlations such that PCA would produce at least one dimension.   Principal Components Analysis resulted in 11 dimensions with eigenvalues greater than 1. The first dimension explained more than 40.0% of the variance, however, the first and second dimensions had eigenvalues greater than 3, thereby indicating multiple dimensions. Examination of the scree plot also suggested the presence of two dominant dimensions. After running another PCA and forcing two dimensions with varimax rotation, the amount of variance explained was 31.3% and 20.8% in each dimension, with eigenvalues of 18.80 and 12.47, respectively. Upon finding two dimensions, it was decided to run separate Rasch analyses on each dimension, and develop two independent subscales in addition to combining the subscales to develop an overall WheelCon short form.  The first dimension was titled ?mobility efficacy? because the items tended to represent beliefs in ability that require movement of the wheelchair (e.g. how confident are you that you can move your wheelchair up a curb cut?). Because most of the items in the second dimension were representative of beliefs in ability to perform self-management skills (e.g. problem solving, decision making, action planning, and self-tailoring (Lorig & Homan, 2003), this dimension was named ?self-management efficacy?.   Items 3, 4, 10, 12, and 42 were factorially complex and therefore retained in both dimensions for Rasch analyses. Item 2 asks about self-efficacy related to moving around furniture in the home, and only loaded on the self-management efficacy dimension. Therefore, this item was deleted from further analyses because it loaded on the unexpected conceptual dimension.  In the second PCA of the five new items (items 37, 53, 54, 58, 61), item 37 loaded onto one dimension, and items 58 and 61 loaded onto the second dimension. These loadings were expected because item 37 asks about moving the wheelchair along a flat dirt path, and items 58  107 and 61 inquire about asking for help and know what to do after falling out of the wheelchair, respectively. Items 53 and 54 ask about negotiating unusual obstacles, moving in anxious/nervous situation, respectively, and were factorially complex. Therefore, these two items were included in both the mobility and self-management efficacy dimensions for Rasch analyses. Appendices S and V present both dimensions and their items.   2) Local independence ? Confirming unidimensionality in both dimensions was an indication that the responses to each item were only being influenced by a single construct. Furthermore, all inter-item residual correlations were below 0.20, thereby, indicating that item responses were independent of all other items in both dimensions.   3) Monotonicity ? Examination of the 11-point response format?s average measure values in table 5.2 confirmed monotonically increasing response options. This indicates that advancing response options are each measuring higher WC self-efficacy.    5.3.4   Assessing item fit, reliability, and validity  Mobility efficacy subscale ? Forty-six items were used to develop the mobility efficacy Rasch model. After removing the 22 items with misfitting outfit statistics, the remaining 24 items were used to develop another Rasch model. Of these remaining items, 14 items had infit statistics within the acceptable range. Appendices S and T show the infit and outfit statistics for these models. There was redundancy in items: 39 and 29; 34 and 33; 20,18 and 32; and 4 and 1. Item 39 inquired about self-efficacy moving through snow along a sidewalk, was statistically the most difficult item, and had content coverage that overlapped with item 29 that inquired about moving through snow and then up a curb cut. Because item 29 inquired about moving through snow and then up a curb cut, and was conceptually deemed to be more difficult than moving through snow along a sidewalk, item 39 was removed from the subscale.     108 The mobility efficacy subscale was finally determined to be comprised of 13 itemsv using a 0 to 10 response format. The item fit statistics were all within the 0.50 to 1.50 acceptable range for the final model as shown in table 5.3. Table 5.3 also presents the 13 items organized by increasing difficulty, the SEMs, and 99% confidence intervals around the item difficulty estimates. Figure 5.2 presents the items ordered by difficulty along the self-efficacy construct continuum.  Summing the recoded responses from each item derives a total raw score ranging from 0 to 130 (-3.51 to 4.10 in logits), with higher scores meaning higher mobility efficacy. Appendix U shows the raw score conversion to standardized scores ranging from 0 to 100, in addition to the SEM and reliability estimates for the entire range of scores. With the exception of the two lowest and highest recoded scores, all other recoded scores (97% of all scores) have reliability ?0.70. Raw recoded scores between 5 and 124 have reliability estimates ?0.90.  The item separation statistic was 8.44 indicating that individuals are adequately able to distinguish the difficulty of each item. The standardized mean item difficulty of this subscale was 46.13 (SD=5.22), whereas the sample?s mean self-efficacy was 59.57 (SD=16.51). The Cronbach alpha was 0.96, and the Pearson correlation with the original longer form from which this subscale was derived was 0.97.         	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?v This subscale includes item 4, which was factorially complex. Other factorially complex items that were eventually determined (by fit statistics) to be part of the longer mobility-efficacy dimension includes items 3, 10 and 42.  109 Table 5.3: Mobility efficacy subscale items in order by difficulty and fit statistics   Mobility efficacy Infit Outfit  Items (n=13) Logits (SE) 99% CI Std score (SE) Mnsq (ZSTD) Mnsq (ZSTD) 1 ?move over carpet?  -0.60 (0.06) -0.75, -0.45 38.23 (0.75) 1.13  (0.80) 1.01 (0.10) 4 ?move in small spaces, such as a bathroom? -0.57 (0.06) -0.72, -0.42 38.64 (0.73) 1.27 (1.60) 1.10 (0.60) 13 ?open, go through, and close a standard door? -0.44 (0.05) -0.57, -0.31 40.35 (0.67) 1.32 (2.00) 1.00 (0.10) 21 ?move down a curb cut? -0.26 (0.05) -0.39, -0.13 42.67 (0.61) 1.06 (0.50) 0.94 (-0.30) 32 ?press the crosswalk button and cross the street? -0.10 (0.04) -0.20, 0.00 44.83 (0.55) 1.15 (1.10) 0.84 (-1.00) 18 ?move down a dry steep slope? -0.03 (0.04) -0.13, 0.07 45.76 (0.54) 1.27 (2.00) 1.16 (1.00) 20 ?move up a curb cut?  -0.02 (0.04) -0.12, 0.08 45.93 (0.54) 1.06 (0.50) 0.90 (-0.60) 33 ?cross a street at a crosswalk with no traffic lights?  0.09 (0.04) -0.01, 0.19 47.32 (0.51) 1.22 (1.70) 1.40 (2.40) 34 ?move across flat, freshly mowed, dry grass? 0.10 (0.04) 0.00, 0.20  47.40 (0.51) 1.18 (1.50) 1.18 (1.20) 23 ?move down a curb cut then over drainage grate? 0.19 (0.04) 0.09, 0.29 48.57 (0.49) 1.06 (0.50) 0.90 (-0.70) 22 ?move over a drainage grate, then up a curb cut?  0.26 (0.04) 0.16, 0.36 49.56 (0.48) 0.88 (-1.10) 0.73 (-2.00) 28 ?move down a curb cut then through snow?  0.63 (0.04) 0.53, 0.73 54.35 (0.47) 1.09 (0.80) 1.08 (0.60) 29 ?move through snow then up a curb cut?  0.76 (0.04) 0.66, 0.86 56.12 (0.47) 1.03 (0.30) 0.96 (-0.20)  Mean  0.00 (0.04) -0.10, 0.10 46.13 (0.56) 1.13 (0.90) 1.02 (0.10) Std=standardized; SE=standard error; CI=confidence interval       110 Figure 5.2: Mobility efficacy items along the self-efficacy continuum                    ?move over carpet? ?move in small spaces? ?open, go through, and close a standard door? ?move down a curb cut? ?press the crosswalk button and cross the street? ?move up a curb cut? ?move down a dry steep slope? ?cross a street at a crosswalk with no traffic lights? ?move across flat, freshly mowed, dry grass? ?move down a curb cut then over a drainage grate? ?move over a drainage grate, then up a curb cut? ?move down a curb cut then through snow? ?move up a curb cut then through snow? 	 ?30	 ? 	 ?	 ?40 	 ?	 ?50 	 ?	 ?60 Easier items Difficult items  111 Self-management efficacy subscale ? Twenty-five items were used to develop the self-management-efficacy Rasch model. After removing the 8 items with misfitting outfit statistics, the 17 remaining items were used to develop another Rasch model. Of these 17 items, 8 items had infit statistics within the acceptable range. Appendices V and W show the infit and outfit statistics for these models. There was redundancy in items: 50, 52, 48, and 64; and 52, 48, 64, and 47. All items were retained however because each was conceptually different.   The self-management efficacy subscale was finally determined to be comprised of 8 itemsvi using a 0 to 10 response format. The item fit statistics were all within the 0.50 to 1.50 acceptable range for the final model, as shown in table 5.4. Table 5.4 also presents the 8 items organized by increasing difficulty, the SEMs, and 99% confidence intervals around the item difficulty estimates. Figure 5.3 presents the items ordered by difficulty along the self-efficacy construct continuum.  Summing the recoded responses from each item derives a total raw score ranging from 0 to 80 (-3.32 to 3.87 in logits), with higher scores meaning higher self-management efficacy. Appendix X shows the raw score conversion to standardized scores ranging from 0 to 100, in addition to the SEM and reliability estimates for the entire range of scores. With the exception of the two lowest and highest recoded scores, all other recoded scores (95% of all scores) have reliability ?0.70. Raw recoded scores between 5 and 73 have reliability estimates ?0.90.  The item separation statistic for this subscale was 3.30. The standardized mean item difficulty of this subscale was 45.49 (SD=2.73), whereas the sample?s mean self-efficacy was 68.47 (SD=18.99). The Cronbach alpha was 0.90, and the Pearson correlation with the original longer form from which the subscale was derived was 0.96.   	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?vi This subscale includes item 54, which was factorially complex. Other factorially complex items that were eventually determined (by fit statistics) to be part of the longer self-management efficacy dimension includes items 12, and 54.  112 Table 5.4: Self-management efficacy subscale items in order by difficulty and fit statistics  Self-management efficacy Infit Outfit  Items (n=8) Logits (SE) 99% CI Std score (SE) Mnsq (ZSTD) Mnsq (ZSTD) 57 ?tell someone how to move your wheelchair if it gets stuck?  -0.32 (0.06) -0.47, -0.17 41.00 (0.88) 1.11 (0.70) 1.03 (0.20) 47 ?use strategies, such as humour, that will help people feel comfortable if they are unsure how to act because you use a wheelchair?  -0.14 (0.06) -0.29, 0.01 43.56 (0.78) 1.00 (0.00) 1.10 (0.60) 64 ?advocate for your needs at work or school?  -0.10 (0.05) -0.23, 0.03 44.03 (0.76) 1.25 (1.60) 1.31 (1.70) 48 ?correct others? mistaken beliefs about people who use wheelchairs? -0.01 (0.05) -0.14, 0.12 45.32 (0.72) 0.91 (-0.60) 0.98 (-0.10) 52 ?solve how to get to your destination when there is an unexpected situation, such as detours?  0.00 (0.05) -0.13, 0.13 45.50 (0.72) 1.25 (1.70) 1.23 (1.40) 50 ?present yourself as you wish to be seen when you are in public and feel people are watching you?  0.03 (0.05) -0.10, 0.16 45.90 (0.71) 0.84 (-1.10) 0.69 (-2.10) 51 ?present yourself as you wish to be seen when you want to impress others?  0.16 (0.05) 0.03, 0.29 47.73 (0.66) 0.85 (-1.10) 0.91 (-0.50) 54 ?move in situations making you anxious/nervous? 0.38 (0.05) 0.25, 0.51 50.84 (0.76) 1.31 (1.80) 1.48 (2.40)  Mean 0.00 (0.05) -0.13, 0.13 45.49 (0.75) 1.06 (0.40) 1.09 (0.50) Std=standardized; SE=standard error; CI=confidence interval        113 Figure 5.3: Self-management efficacy items along the self-efficacy continuum                     	 ?30	 ? 	 ?	 ?40 	 ?	 ?50 	 ?	 ?60 Easier items Difficult items ?tell someone how to move your wheelchair if it gets stuck? ?use strategies to help people feel comfortable if they are unsure how to act?? ?advocate for you needs at work or school? ?correct others? mistaken beliefs about people who use wheelchairs? ?solve how to get to your destination when there is an unexpected situation? ?present yourself as you wish to be seen when you are in public?? ?present yourself as you wish to be seen when you want to impress others?? ?move in situations making you anxious/nervous?  114 The WheelCon short form ? The strength of the correlation between the mobility and self-management efficacy subscales was r=0.65. After combining the items from both subscales and rerunning the Rating Scale Model, one item had an infit statistic greater than 1.50, and 4 items had outfits statistics greater than 1.50, as shown in table 5.5. Table 5.5 also presents the items organized by increasing difficulty, the SEMs, and 99% confidence intervals around the item difficulty estimates. Figure 5.4 presents the items ordered by difficulty along the self-efficacy construct continuum.  Summing the recoded responses from each item derives a total raw score ranging from 0 to 210 (-3.73 to 4.29 in logits), with higher scores meaning higher self-efficacy with using a manual wheelchair. Appendix Y shows the raw score conversion to standardized scores ranging from 0 to 100, in addition to the SEM and reliability estimates for the entire range of scores. With the exception of the two lowest and highest recoded scores, all other recoded scores (98% of all scores) have reliability ?0.70. Raw recoded scores between 5 and 204 have reliability estimates ?0.90.  The item separation statistic for this short form was 6.97. The standardized mean item difficulty was 46.49 (SD=3.81), whereas the sample?s mean self-efficacy was 59.04 (SD=14.32). The Cronbach alpha was 0.96, and the Pearson correlation with the 65-item original WheelCon was 0.97.            115 Table 5.5: WheelCon short form items in order by difficulty and fit statistics  WheelCon short form  Infit Outfit  Items (n=21) Logits (SE) 99% CI Std score (SE) Mnsq (ZSTD) Mnsq (ZSTD) 57 ?tell someone how to move your wheelchair if it gets stuck?  -0.43 (0.05) -0.56, -0.30 41.16 (0.68) 1.06 (0.40) 1.36 (1.70) 1 ?move over carpet?  -0.37 (0.05) -0.50, -0.24 41.89 (0.64) 1.08 (0.50) 0.90 (-0.40) 4 ?move in small spaces, such as a bathroom? -0.34 (0.05) -0.47, -0.21 42.21 (0.63) 0.99 (0.00) 0.80 (-1.00) 47 ?use strategies, such as humour, that will help people feel comfortable if they are unsure how to act because you use a wheelchair?  -0.29 (0.05) -0.42, -0.16 42.87 (0.60) 1.22 (1.40) 2.12 (4.50)* 64 ?advocate for your needs at work or school?  -0.27 (0.05) -0.40, -0.14 43.18 (0.59) 1.51 (3.00)* 2.48 (5.70)* 13 ?open, go through, and close a standard door? -0.24 (0.05) -0.37, -0.11 43.53 (0.57) 0.92 (-0.50) 0.71 (-1.70) 48 ?correct others? mistaken beliefs about people who use wheelchairs? -0.20 (0.04) -0.30, -0.10 44.03 (0.55) 1.25 (1.70) 1.65 (3.00)* 52 ?solve how to get to your destination when there is an unexpected situation, such as detours?  -0.19 (0.04) -0.29, -0.09 44.16 (0.55) 0.93 (-0.50) 0.81 (-1.00) 50 ?present yourself as you wish to be seen when you are in public and feel people are watching you?  -0.17 (0.04) -0.27, -0.07 44.42 (0.54) 1.28 (1.90) 1.03 (0.20) 21 ?move down a curb cut? -0.10 (0.04) -0.20, 0.00 45.30 (0.52) 0.94 (-0.40) 0.88 (-0.70) 51 ?present yourself as you wish to be seen when you want to impress others?  -0.07 (0.04) -0.17, 0.03 45.66 (0.51) 1.11 (0.90) 1.05 (0.40) 32 ?press the crosswalk button and cross the street? 0.04 (0.04) -0.06, 0.14 46.95 (0.47) 1.17 (1.30) 0.89 (-0.70) 54 ?move in situations making you anxious/nervous? 0.08 (0.05) -0.05, 0.21 47.55 (0.58) 0.92 (-0.50) 1.07 (0.40) 18 ?move down a dry steep slope? 0.09 (0.04) -0.01, 0.19 47.65 (0.46) 1.02 (0.20) 1.04 (0.30) 20 ?move up a curb cut?  0.10 (0.04) 0.00, 0.20 47.78 (0.46) 0.96 (-0.30) 0.90 (-0.60) 33 ?cross a street at a crosswalk with no traffic lights?  0.19 (0.03) 0.11, 0.27 48.84 (0.43) 1.17 (1.50) 1.45 (2.60) 34 ?move across flat, freshly mowed, dry grass? 0.19 (0.03) 0.11, 0.27  48.90 (0.43) 1.05 (0.40) 1.13 (0.90) 23 ?move down a curb cut then over drainage grate? 0.26 (0.03) 0.18, 0.34 49.79 (0.42) 0.96 (-0.30) 0.86 (-1.00) 22 ?move over a drainage grate, then up a curb cut?  0.33 (0.03) 0.25, 0.41 50.56 (0.41) 0.85 (-1.40) 0.74 (-1.90) 28 ?move down a curb cut then through snow?  0.62 (0.03) 0.54, 0.70 54.26 (0.40) 1.03 (0.30) 1.01 (0.10) 29 ?move through snow then up a curb cut?  0.73 (0.03) 0.65, 0.81 55.62 (0.40) 1.01 (0.20) 1.08 (0.60)  Mean 0.00 (0.04) -0.10, 0.10 46.49 (0.52) 1.07 (0.50) 1.14 (0.50) Std=standardized; SE=standard error; CI=confidence interval; *misfitting items   116 Figure 5.4: WheelCon short form items along the self-efficacy continuum                      	 ?40	 ? 	 ?	 ?50 	 ?	 ?55 Easier items Difficult items ?tell someone how to move your wheelchair if it gets stuck? ?move over carpet? ?move in small spaces, such as bathrooms? ?use strategies to help people feel comfortable if they are unsure how to act?? ?advocate for your needs at work or school? ?open, go through, and close a standard door? ?correct others? mistaken beliefs bout people who use wheelchairs? ?solve how to get to your destinations when there is an unexpected situation? ?present yourself as you which to be seen in public?? ?move down a curb cut? ?present yourself as you wish to be seen when you want to impress others?? ?press the crosswalk button and cross the street? ?move in situations making you feel anxious/nervous? ?move down a dry steep slope? ?move up a curb cut? ?cross a street at a crosswalk with no traffic lights? ?move across flat, freshly mowed, dry grass? ?move down a curb cut then over a drainage grate? ?move over a drainage grate, then up a curb cut? ?move down a curb cut then through snow? ?move through snow then up a curb cut?  117 5.4 Discussion  The purpose of this study was to evaluate the measurement properties of the WheelCon using Rasch analyses. In doing so, evidence was first found in support of the hypothesis that a shorter response format using recoded data would function better than the original 101-point format. Response formats are intended to capture different degrees of a construct from each individual completing the measurement (Linacre, 2002). Nunnally states the advantage is always with using more than fewer response options, up until the point where the number of options begin to confuse subjects (Nunnally, 1967). In the evaluation of the WheelCon?s original 101-point response format, response options may have been used in unexpected contexts, which likely led to a disordering of response options. For example, if individuals provide a response of 70 to an item of certain difficulty, and a response of 75 to a slightly more difficult item, the response format is not being used as intended because probabilistically the more difficult item should elicit lower responses relative to the easier item. One explanation for this type of discrepancy is that individuals are not differentiating between adjacent response options (Streiner & Norman, 2008), which has been shown to be common among older adults when administered response formats with many options (Lachman et al., 1998; Arnadottir et al., 2010). A consequence is respondent bias arising from increases in subjectivity. Such bias may confound the interpretation of total scores when it leads to response options not manifesting construct levels as intended.   Collapsing the original response format into 11 options and recoding the data resulted in an ordering of the response options such that higher scores equate to higher self-efficacy, unlike the 101-point response format. The results suggest that the 100 option in the original response format be kept as a unique category. This may be an indication that individuals are less likely to report complete self-efficacy when they have higher self-efficacy, and more likely to report no self-efficacy when their self-efficacy is low. In other words, the transition from having some self-efficacy to complete self-efficacy may be greater than the transition from having no self-efficacy to some self-efficacy. This is observed in the standardized self-efficacy scores.   This study?s findings are similar to other studies that have evaluated the functioning of shortened format with self-efficacy measures (Arnadottir et al., 2010; Sakakibara, Miller, & Backman,  118 2011).  For example, others have shown that scores derived using a 5-option response format (reduced from a 101-point format) with the Activities-specific Balance Confidence (ABC) scale have good validity and reliability estimates in individuals with a lower-extremity amputation throughout the confidence continuum (Cronbach alpha = 0.75 to 0.94, between ?3.00 logits) (Sakakibara et al., 2011). Similar results of a 5-option response format used with the ABC-scale were shown in another study of older adults living in Iceland (Arnadottir et al., 2010). Therefore, the use of a shorted 0 to 10 response format with the WheelCon may improve the precision of the self-efficacy estimates, particularly in older wheelchair users. This, however, needs to be established in studies using the WheelCon with the 0 to 10 response format to collect data. The use of a 0 to 10 format remains in accordance with Bandura?s guidelines for developing self-efficacy measures (Bandura, 2006).   Through the use of PCA, two dimensions were found to statistically comprise the 65-items in the WheelCon. In this analysis, the dimension loadings depended primarily on whether the items required self-efficacy related to maneuvering the wheelchair or not. This is consistent with observations made by Rushton et al. (2011) during the development of the WheelCon in that the item groupings were split between activities requiring the direct use of the wheelchair versus those items not requiring such a direct use. Furthermore, because the dimensions moderately correlated with each other supports the development of both distinct subscale scores in addition to an overall total self-efficacy score. Others have similarly developed subscale and total scores during the development and evaluation of measures using Rasch analyses (Jette et al., 2002). Moreover, with the use of a 0 to 10 response format, several items in both dimensions were eliminated due to their model misfit, and redundancy. This resulted in a 13-item mobility efficacy, 8-item self-management efficacy subscales, and the 21-item WheelCon short form. Overall, these results provide evidence in support of the hypothesis that the WheelCon with a shortened response format would result in more than one dimension with fewer items.  Because situation specific self-efficacy measures provide more explanatory and predictive value than global measures of self-efficacy (Bandura 2006), the mobility and self-management subscales may provide more accurate self-efficacy estimates for certain clinical situations or research questions than the 21-item WheelCon short form in its entirety. For example, for  119 outcomes related to mobility or physical abilities, the use of the 13-item mobility efficacy subscale may be more predictive than the 8-item self-management efficacy subscale. Similarly, the reverse may true for outcomes related to problem solving, advocacy, or other self-management tasks. Although the 21-item short form does not function as well as the individual subscales, as indicated by the misfitting items, it may still be of value particularly with multifaceted outcomes such as participation in social and personal roles, or physical activity.   The total scores from both subscales and the overall short form exceed ?3.00 logits, which is an indication that each measure is assessing more than 99.7% of an individual?s strength of self-efficacy. Contrary to the hypothesis that the entire range of recoded scores would have reliability estimates ?0.70, the recoded scores at the lower and upper ranges for each measure are not as reliable or precise as the scores in the middle range for each measure. Nonetheless, more than 95% of all possible recoded scores have good reliability. By offering reliable and valid situation specific subscale measurements along with a more global measurement may facilitate the assessment of self-efficacy with using a wheelchair both clinically and in research.  Both subscales and the 21-item WheelCon are adequate representations of the longer forms from which they were derived. This is demonstrated by the high magnitude of the correlation coefficients between the original and the reduced measures, which supports the hypothesis that the correlation with the original measure would be good to excellent. Furthermore, the hierarchical ordering of the items by difficulty in the subscales is conceptually sound. This is corroborated by item separation statistics exceeding 3.00, which is an indication that respondents are able to adequately differentiate each item by their difficulty, and thus give meaningful responses. Not surprising, the item separation statistic for the self-management efficacy subscale was low relative to the statistic for the mobility-efficacy subscale. Arguably, differentiating items based on beliefs of physical abilities to maneuver a wheelchair is easier than doing so with items pertaining to self-managerial abilities.   Finally, it is interesting to note that the sample?s mean self-efficacy in each of the measures was higher than mean item difficulties (i.e. the area where the measures are assessing self-efficacy most reliably). This is indication that the sample was responding to many of the items with  120 higher self-efficacy. The shortened measures therefore may assess the self-efficacy constructs more reliably among wheelchair users with lower self-efficacy than the individuals in this sample. For example, older, female wheelchair users or other groups at risk for having lowered self-efficacy with using a wheelchair as described in Chapter 4.   Limitations This study has several limitations. First, the modifications made to the response format and subsequent analyses were completed using the same dataset. As a result, the reliability estimates are based on recoded measurements. The results, therefore, only appear to support the reliability of the measurements derived using the shortened version of the WheelCon, and the subscales with a 0 to 10 response format.   Next, a primarily statistical approach was used to evaluate the dimensionality of the WheelCon. Through the use of PCA and varimax rotation, two uncorrelated independent dimensions resulted, which may be considered unrealistic when considering that both dimensions are assessing self-efficacy related to wheelchair use. Despite using sound methodological procedures, had a more theoretical approach been followed to evaluate the dimensionality (e.g. to confirm the 6 conceptual areas of the WheelCon), along with principal axis factoring or the use of an oblique rotation, not only would the dimensions have been correlated, the number of resulting dimensions may have changed. Regardless, the use of PCA to account for the maximum amount of variance with the smallest number of dimensions is consistent with the recommendations advocated by Norman and Streiner (2008).  Furthermore, the sample size may be considered small for Rasch analyses, thereby producing less precise and robust estimates, and having less power for fit analyses. In general, the more item and response option parameters estimated the larger the sample size required. At present, however, there are no guidelines to orient researchers to sample size requirements (Velozo et al., 2013). In addition, Wang and Chen (2005) have shown through the use of Monte Carlo simulations that the Rating Scale Model produces negligible item parameter bias in measures with at least 20 items with samples of at least 100 individuals.  Moreover, for stable item estimates within ?0.50 logits, Linacre (1994) suggests a sample size of at least 150 to have 99%  121 confidence. Therefore, this study?s sample size of 220 is likely large enough to provide accurate item parameter and fit statistic estimates. Next, the analyses on a combination of data from two datasets using slightly different WheelCon versions may have biased the results. However, the investigation of the similarities between the datasets prior to their combination likely reduced the potential for bias. Furthermore, the analyses provide only validation of the internal structure of the measure. Prospective studies with a priori hypotheses relative to other constructs are necessary. In addition, a consequence of eliminating items to create shortened measure is that information may be lost in terms of identifying the specific areas requiring improvement. Although this has important clinical implications, the shortened measures are accurate representations of the longer versions from which they were derived, as shown by the magnitude of the correlations, and provide information on the general areas that may require treatment. The shortened measures also offer less administrative and respondent burden, which may facilitate the clinical assessment of WC self-efficacy. Finally, because the shortened WheelCon includes items pertaining to use of the wheelchair in snow, the results are limited to individuals with similar characteristics of the sample used in this study, residing in similar geographic regions.   5.5 Conclusion  The WheelCon is comprised of two dimensions that assess self-efficacy related to mobility and self-management. The measurements from the 21-item WheelCon short form, and the 13-item mobility efficacy and 8-item self-management subscales with a recoded 0 to 10 response scale appear to have good reliability, and may provide accurate and precise measurements of different forms of WC self-efficacy. The usefulness of the subscales and/or the entire short form depends on the context in which they are being considered. Research to establish the reliability and validity of the measurements using the 0 to 10 response scale is warranted.       122 CHAPTER 6: Discussion and future directions  Self-efficacy is a key determinant of behaviour, and has implications on whether or not individuals achieve their desired outcomes (Bandura, 1997). Higher self-efficacy have positive effects on what people do, is associated with lower health risks, and better overall health. For these reasons, in addition to evidence that self-efficacy has the potential to be modified, I developed this research to learn more about the construct in the context of wheelchair use. Concern over the anticipated increase in numbers of older, community-dwelling manual wheelchair users, along with a lack of knowledge about their participation frequency and life-space mobility also provided impetus for this research on self-efficacy with using a manual wheelchair, a new and novel construct in the wheelchair use literature.   Over the course of this study, I first established the associations between the self-efficacy with using a manual wheelchair construct and both participation frequency and life-space mobility, as well as mediators to help explain the associations. Evidence in support of the hypotheses in Chapters 2 and 3 regarding the explanatory value of self-efficacy on participation and mobility provided rationale to invest more time and resources into investigating the construct itself. Accordingly, I then identified factors and characteristics of individuals at risk for lowered self-efficacy with using a manual wheelchair in Chapter 4, and subsequently evaluated the measurement properties of the WheelCon using Rasch analyses in Chapter 5. The results from each of the studies contribute to a greater understanding of the self-efficacy with using a manual wheelchair construct, which was the overall purpose of this research.  6.1 Self-efficacy, participation, and mobility with using a manual wheelchair  6.1.2   Self-efficacy as a body function  The ICF was used to guide the analyses of the associations between self-efficacy, participation, and mobility. Although it is common for self-efficacy to be conceptualized as a personal factor at the contextual level (e.g. Martin-Ginis et al., 2012), in this research it was conceptualized as a body function at the disability/functioning level.  123 According to the ICF, ?body functions? are the ?physiological functions of body systems (including psychological/mental functions)? (WHO, 2001), and ?personal factors? are the ?particular background of an individual?s life and living, including features of the individual that are not part of a health condition or health states? that may include sex, race, age, fitness, lifestyle, habits, upbringing, coping styles, social background, past and current experience, character style, as well as other psychological assets?? (WHO, 2001). Whereas the ?body function? domain is further elaborated upon and divided into several subsections, the ?personal factor? domain is not. This has led to conceptual confusion about how some seemingly intuitive personal factor variables differ from variables in the body functions chapter, such as confidence, optimism, extraversion, and motivation (Threats, 2007).   A key conceptual difference between personal factor and body function variables is that variables are viewed as a ?body function? when they are considered to be pathological (Threats, 2007). That is, if a health or disabling condition influences the variable, then it would be considered a ?body function?. Conversely, personal factor variables have nothing to do with or are not caused by the health condition. Rather, they are long-standing attributes individuals display over time regardless of health and/or functional status. Therefore, in the context of self-efficacy with using a manual wheelchair, because it is a state, and has the potential to be influenced by a number of events, including health and disability (Bandura, 1997), it was conceptualized it as a body function.   6.1.3    The direct effects of self-efficacy on participation frequency and life-space mobility  After controlling for the confounding effects of age, number of comorbidities, and perceived social support, self-efficacy with using a manual wheelchair was a significant predictor of participation frequency in older, community-dwelling wheelchair users. Whereas the entire model accounted for 41.5% of the participation variance, the independent contribution of WC self-efficacy was 17.2%.    124 Similarly, after controlling for sex, number of comorbidities, geographic location, assistance with using a wheelchair, and education, WC self-efficacy was a significant and independent predictor of life-space mobility. This model accounted for 37.1% of the variance in life-space mobility and the unique contribution of WC self-efficacy was 3.9%.  6.1.4    The mediated effects of self-efficacy on participation frequency and life-space mobility  After quantifying the direct effect of self-efficacy with using a manual wheelchair on both participation frequency and life-space mobility, and deeming the associations worthy of further inquiry, I investigated the mediated self-efficacy effects. Mediators establish how or why an independent variable causes the dependent variable, and represents the mechanism through which the independent variable operates (Frazier, Tix & Barron, 2001). In other words, mediating variables explain the association between independent and dependent variables (Frazier et al., 2001).   According to Bandura (1997), self-efficacy effects are realized through four mechanism: cognitive, motivational, affective, and selection processes. Self-efficacy beliefs influence cognitive thought patterns that potentially improve or undermine behaviour. Such thought patterns are also important for the development of motivation (Bandura, 1997). For example, individuals with higher self-efficacy set challenging and rewarding goals, and have positive outcome expectations (Bandura, 1997). Self-efficacy beliefs also act through affective processes. People with lower self-efficacy have difficulties exerting control over their lives and environments, which lead to feelings of futility, anxiety, and depression (Bandura, 1997). Furthermore, individuals select the activities and environments in which they live their lives. People with higher efficacy are more likely to pursue challenging activities and display greater levels of perseverance to succeed, than individuals with lower self-efficacy (Bandura, 1997).   In this research, the association between self-efficacy, and participation frequency was observed to be simultaneously mediated by both selection (i.e. life-space mobility), and cognitive (i.e. perceived participation limitations) processes. The mediators accounted for 78.0% of the direct  125 association between self-efficacy and participation frequency. Moreover, the mediated model accounted for 55.0% of the participation frequency variance.   In terms of the association between self-efficacy and life-space mobility, the mediated effect through a selection process (i.e. wheelchair skills) was statistically significant, and accounted for 91.0% of self-efficacy?s direct effect on life-space mobility. This mediated model accounted for 39.0% of the life-space mobility variance.  6.1.5   The need for self-efficacy enhancing interventions   Existing interventions for wheelchair users have focused on factors mainly in the environment (e.g. wheelchair technology (Cooper et al., 2006), and prescription (Hoenig et al., 2005)) and activity (e.g. wheelchair skills (WSTP manual, 2008)) ICF domains (WHO, 2001). Although interventions at the environment level have resulted in good outcomes, the outcomes themselves have not been shown to improve participation frequency or mobility. For example, research illustrates improvements to wheelchair technology in terms of weight savings, and improved safety, and durability (Cooper et al., 2006). In many instances, however, the wheelchair itself is a commonly reported factor limiting home and community participation (Barker et al., 2004; Chaves et al., 2004). Furthermore, in a study of community-dwelling veterans (mean age = 65.0, SD = 13.7), those who were prescribed a wheelchair by an experienced clinician with expert knowledge of wheelchairs reported more frequent wheelchair use during the study than individuals who received their wheelchair by usual care (i.e. by a clinician without expert knowledge of wheelchairs) (Hoenig et al., 2005). These findings suggest that an appropriately set-up wheelchair may lead to more of its use. However, research on the implications of wheelchair seating on mobility and participation frequency is mixed. For example Bourbonniere et al. (2007) observed that a lower need for a wheelchair seating intervention among residents in long-term care settings significantly predicted more life-space mobility, whereas in more recent studies in a similar population, associations were not observed with either life-space mobility (Mortenson et al., 2011) or participation frequency (Mortenson et al., 2012).    126 In terms of interventions at the activity level, there is evidence demonstrating the utility of the Wheelchair Skills Training Program (WSTP manual, 2008) at improving wheelchair skills (Coolen et al., 2004; Kirby et al., 2004b; MacPhee et al., 2004; Best et al., 2005). Evidence also indicates the statistically significant, positive associations between wheelchair skills, and mobility (Mortenson et al., 2011; Mortenson et al., 2012), and participation (Kilkens et al., 2005; Hosseini et al., 2012; Mortenson et al., 2012). Although this evidence is positive, Bandura (1977) has demonstrated that ability alone will not lead to desired outcomes if individuals lack the belief in their ability. In fact, Social Cognitive Theory explains that self-efficacy is a more important predictor of behaviour than actual ability (Bandura, 1997), and empirical evidence supports this notion (Bandura, 1977; Pang & Eng, 2008; Schmid et al., 2012). Therefore, individuals who report high ability to use their wheelchair, but low participation and/or mobility, may benefit from improvements to their self-efficacy with using a wheelchair. Interestingly, recent reports indicate discordant self-efficacy beliefs and ability to use a wheelchair in older wheelchair users with several individuals reporting lower self-efficacy and higher wheelchair skills (Miller et al., 2012).  Implication: Self-efficacy with using a manual wheelchair has both direct and mediated effects on the participation frequency, and life-space mobility in older, community-dwelling manual wheelchair users. The development of efficacy-enhancing interventions is warranted, and may result in improved health outcomes for those individuals in need, with lowered self-efficacy.  6.2 Health, personal, and environmental predictors of self-efficacy  After establishing the importance of self-efficacy with using a manual wheelchair on participation and mobility, and recommending the development and testing of efficacy-enhancing interventions, I then conducted a study to identify individuals who may be at risk of lowered self-efficacy, and benefit the most from such interventions. This was done because clinically, lowered self-efficacy may compromise participation frequency and life-space mobility, which in turn could result in lower quality of life, and/or the onset of depressive symptoms.   127 Health, personal and environmental contextual factor variables were selected for evaluation because of their ease of access, and use for an efficient appraisal of WC self-efficacy.   In the first model, the effect of health condition on WC self-efficacy was assessed. Results indicated a significant negative association between number of comorbidities and self-efficacy with using a manual wheelchair. Importantly, this finding supports our conceptualization of WC self-efficacy as a variable in the ICF?s body function domain.  In the second and third models, personal, and environmental factor variables were entered, respectively. Both age and sex significantly predicted self-efficacy. That is, older age and female sex are risk factors for lowered self-efficacy. When considering that older age and being female are predisposing characteristics associated with wheelchair use, as described in Chapter 1, it is plausible that over the next several years there will be a disproportionate increase in wheelchair users who have lowered self-efficacy with using a wheelchair due population aging. In addition, four variables resulting from wheelchair use were significant predictors of WC self-efficacy. The final model accounted for 44.0% of the WC self-efficacy variance, and indicated that assistance with using a wheelchair was the most important predictor of self-efficacy, followed by sex, daily hours of wheelchair use, training to use a wheelchair, need for a seating intervention, and age.  Implication: Manual wheelchair users of older age, and female sex, with comorbid conditions, who require assistance with using their wheelchair, use their wheelchair minimally throughout the day, received no formal training to use a wheelchair, and have a greater need for a seating intervention are at risk of lowered self-efficacy. These subgroups of wheelchair users may be in greatest need of targeted interventions to enhance their self-efficacy with using a wheelchair.  Although the development and testing of efficacy enhancing interventions is advocated for those individuals at risk of having lowered WC self-efficacy, it would be remiss to assume all individuals or groups of wheelchair users with lowered self-efficacy are in need of such intervention. In fact, enhancing WC self-efficacy to a level that exceeds an individual?s ability with wheelchair use may put him/her at risk of injury. Despite it being theorized that  128 enhancements to self-efficacy may contribute to the development of better ability, higher self-efficacy will not automatically produce new or better abilities (Bandura, 1997). Because individuals with higher self-efficacy are more likely to do things, regardless of ability, than those with lower self-efficacy, those with a higher WC self-efficacy-lower WC skill profile may take greater risks, and thus have an increased chance of incurring accidents while using their wheelchair. Therefore, caution must be taken when identifying individuals for treatment because those individuals who may appear to be in greatest need for efficacy enhancements may first require development of their abilities to use a wheelchair. It is prudent to fully understand the reasons why individuals or groups of wheelchair users have lowered WC self-efficacy, and to plan treatments accordingly. 	 ?6.3 Rasch analyses of the Wheelchair Use Confidence Scale  Given that lowered self-efficacy with using a wheelchair may be a barrier to participation frequency and life-space mobility, and that the population of individuals most at risk to use a wheelchair, and have lowered WC self-efficacy is growing at a rate faster than any other group of wheelchair users suggests that self-efficacy may be a potentially important clinical area to address. Clinicians administering wheelchair services make decision based on the existence of supporting research evidence. Evidence-based practice as defined by Sackett, Straus, Richardson, Rosenberg & Haynes (2000) refers to the integration of best research evidence, clinical expertise, and patient values. Best research evidence in turn is predicated upon accurate and precise measurement. Therefore, since all measurements are approximations, especially for unobserved phenomena, much work must be done to ensure measurements are as accurate and precise as possible. This entails assessing the measurement properties of reliability, and validity, and is especially important for new measures such as the Wheelchair Use Confidence Scale (WheelCon). Moreover, when considering the growing evidence that self-efficacy with using a manual wheelchair is an explanatory variable of the participation frequency and life-space mobility of older wheelchair users, additional investment of time and resources into evaluating the measure?s properties is justified.    129 Through the use of contemporary measurement methods, in Chapter 5 I assessed the: dimensionality of the items using Principal Components Analysis, as well as the measure?s internal structure, item reliability using Rasch analyses, and the functioning of the WheelCon?s response format. The results indicated that through the use of a better functioning 0 to 10 response format, relative to the original 0 to 100 format, the WheelCon is comprised of two dimensions assessing mobility efficacy and self-management efficacy. The results from the Rasch analyses further revealed items that could be eliminated from each of the dimensions due to redundancy, and their non-compliance with the Rasch model. In the end, a 13-item mobility efficacy subscale, and an 8-item self-management efficacy subscale were developed each with their own scoring algorithm. In addition, by combining the subscales, and recalibrating items, a scoring system for an overall 21-item WheelCon was developed.   Both subscales in addition to the overall 21-item WheelCon have their own set of measurement properties. With the exception of the extreme lower and upper range of scores for each measure, all other scores have good reliability (i.e. Cronbach alpha ?0.70) and precision. The development of two subscales offers greater measurement accuracy and precision, and the fewer number of items may facilitate the assessment of self-efficacy with using a wheelchair both clinically and in research. The use of either subscale or the combined 21-item measure, however, depends on the clinical purpose and/or research questions.   Furthermore, the results indicate that the shortened measures may provide more reliable assessments of the self-efficacy specific constructs among wheelchair users with lower self-efficacy than the individuals participating in this study. The subgroups of individuals identified in Chapter 4 that are at risk for lowered self-efficacy with using a wheelchair may provide measurements with the highest reliability and precision, however, this needs to be established.  By reducing the number of items, and simplifying the response format, the mobility and self-management subscales and/or the WheelCon short form may be more acceptable by clients and research participants, in terms of easiness to complete, and feasible by clinicians and researchers to administer and score (Fitzpatrick, Davey, Buxton & Jones, 1998; Andresen, 2000). In fact, Fitzpatrick et al. (1998) states that the length and number of items should be reduced in measures  130 to improve patient acceptability when there are no compromises to the measurement properties (Fitzpatrick et al., 1998). Similarly, in establishing criteria for the evaluation of outcome measures, Andresen (2000) gives excellent marks to measures that can be completed in less than 15 minutes (Andresen, 2000). As a result of the analyses, I have both shortened the amount of time to complete the measure, by way of reducing item content, and developed evidence in support of the subscales? and short form?s measurement properties. In addition, I have created a standardized scoring system to help clarify the interpretation of raw scores. The interpretability of scores is another important element in the selection of patient-based outcome measures (Fitzpatrick et al., 1998).  Implication: The WheelCon is comprised of two dimensions that assess self-efficacy related to mobility and self-management. The 21-item WheelCon short form, and the 13-item mobility efficacy and 8-item self-management efficacy subscales with a 0 to 10 response scale have good reliability, and provide accurate and precise measurements of different forms of self-efficacy with wheelchair use. The use of the subscales and/or the entire short form depends on the context in which they are being considered. The measures may assess the self-efficacy constructs with the highest reliability and greatest precision in individuals at risk for having lowered self-efficacy with using a wheelchair.  6.4 Strengths and limitations  This dissertation provides the most comprehensive research on self-efficacy with using a manual wheelchair to date. The results provide new knowledge that will serve as a foundation and rationale for future exploratory and clinical trial research on the self-efficacy, participation, and mobility of older, community-dwelling manual wheelchair users. Furthermore, by examining the measurement properties of the WheelCon using Rasch analyses, I provide a modified version that is reliable, has greater measurement accuracy, and less respondent and administrative burden than the original.   131 The recruitment of participants from both British Columbia and Quebec is both a strength and limitation of this research. By recruiting individuals in both provinces that differ in terms of culture, language, and geography increases the generalizability of the results. This recruiting strategy also increased the studies? sample size, which in turn allowed for more robust statistical analyses. A limitation to this strategy, however, has to do with the use of primarily self-report measures and on combining the data from individuals in British Columbia with those data from individuals in Quebec. To minimize this concern, the datasets from both provinces were analyzed and compared to ensure that there was no reason not to combine the data for analyses. The recruitment of volunteers in both provinces is another study limitation. As a result, data may under- or over-represent particular groups within the population, and there is a lack information on potential reasons why individuals who received study information chose not to participate. Therefore, the non-random sampling technique used in this study may limit the generalizations of the results.   The use of the ICF to guide the analyses also represents strengths and limitations of this research. Because the ICF is an internationally recognized biopsychosocial model of health, its use ensures the utilization of health/disability language that is common around the world. Therefore, the results of this dissertation have a greater chance of being accurately interpreted and understood than if I had used a model not as well known. The use of the ICF is a limitation, however, because not all of its domains have a conceptually clear definition. Therefore, some may argue for variables (e.g. self-efficacy) to belong in one domain, whereas others may argue for the same variable to belong elsewhere. Because the allocation of variables to certain domains has analytical implications, the data analyses and results may have changed had self-efficacy been conceptualized as a personal factor.   Next, because the characteristics and composition of the individuals who participated in this study are different than those of the typical older wheelchair user described in Chapter 1 (pages 2-4), the results are limited to those individuals who have similar characteristics and health conditions of this study?s sample. Whereas the evidence used to report on the characteristics of older wheelchair users in Chapter 1 was on individuals 60 years and older, many individuals who participated in this study were between the ages of 50 and 60. Arguably, the evidence used to  132 characterize older wheelchair users will change as the baby-boom generation ages. Although the results of this dissertation may be limited in generalizing to the older wheelchair user of today, it does provide valuable information on self-efficacy, participation, and mobility of the baby-boom generation that will soon comprise the majority of older wheelchair users.  The use of self-report measures for variables that could have been assessed with performance-based measures or indicators (e.g. wheelchair skills, mobility) is a limitation that may have produced biased results due to social desirability. Furthermore, the use of a self-report measure for wheelchair skills in particular may have contributed to its high correlation with the self-efficacy measure, which in turn may have introduced multicollinearity into the regression models. An explanation for the high correlation between the wheelchair skills and WC self-efficacy measurements may be that the Wheelchair Skills Test-Questionnaire (WST-Q) is similarly assessing self-efficacy with wheelchair use, albeit from a different perspective. According to Bandura (2006), individuals rate their strength of self-efficacy using a response scale such as that in the WheelCon. Level of self-efficacy, on the other hand, refers to the number of activities individuals judge themselves capable of performing (Bandura, 2006). Therefore, it may be that the WST-Q is assessing level of self-efficacy because the questionnaire asks individuals if they are capable of performing certain wheelchair skills. If both the WheelCon and WST-Q are assessing different dimensions of self-efficacy with wheelchair use, then the interpretation of the results in Chapter 3 may be inaccurate. However, the finding that WC self-efficacy measured using the WheelCon is a statistically significant independent predictor of life-space mobility still holds true. Despite acknowledging the possibility that both the WheelCon and WST-Q may be assessing self-efficacy, it also quite possible that the WST-Q is in fact measuring wheelchair skills when considering evidence that a high correlation exists between the questionnaire and performance-based versions of the Wheelchair Skills Test (Rushton et al., 2012). Nonetheless, results in Chapter 3 should be interpreted with caution.  Finally, there may be concerns associated with common method bias in that all measures administered utilized the same method of measurement (i.e., self-report questionnaires) (Podsakoff, MacKenzie, & Podsakoff, 2012). Podsakoff et al. (2012) indicate that the issue with measuring different constructs using the same method has to do with the possibility that some of  133 the observed covariation may be due the common method of measurement. Had common method bias been controlled for (e.g. through the use of temporal separation of the dependent and independent variables) the amount of participation frequency (Chapter 2), life-space mobility (Chapter 3), and WC self-efficacy (Chapter 4) variance accounted for may be lower than what is reported. Therefore, these findings should also be interpreted with caution.  6.5 Future directions  The research findings may inform the development of a variety of exploratory and clinical trial research. Because the findings indicate that self-efficacy with using a manual wheelchair is an important predictor of participation frequency and life-space mobility in older, wheelchair users, a natural next step would be to develop predictive models of the participation and mobility constructs that include the self-efficacy construct. Given that much of the participation and mobility variance in wheelchair users remain unexplained, the evaluation of self-efficacy with using a wheelchair in addition to the already established important predictors of participation and mobility may result in a more complete understanding. The use of structural equation modeling in particular would be especially informative by simultaneously exploring the associations between participation, and mobility, along with the independent variables. Furthermore, evaluating the associations between WC self-efficacy, and performance-based measures of wheelchair skills and mobility would also provide valuable information from a different context.  Although this research examines the direct and mediated effects of WC self-efficacy, it is also plausible that self-efficacy could mediate the associations between wheelchair skills, participation frequency, and life-space mobility. According to Social Cognitive Theory, the tenet of triadic reciprocalism states that the association between self-efficacy and other determinants of behaviour are bidirectional (Bandura, 1997). In the context of variables examined in this dissertation, it may be that WC self-efficacy influences wheelchair skill, and the performance of such skills can, in turn, influence future WC self-efficacy. It is also plausible that self-efficacy may moderate the associations between wheelchair skills, participation frequency, and life-space mobility. That is, the associations may differ at different degrees of WC self-efficacy. For example, the magnitude of the association between wheelchair skills and participation frequency  134 may be small for individuals with lowered WC self-efficacy, and large for those with higher WC self-efficacy. Investigation into the hypotheses of WC self-efficacy as a mediator or moderator will provide important information using a conceptualization of the self-efficacy construct different from that used in this study.   Testing the predictive ability of the self-efficacy model developed in Chapter 4 in different samples of wheelchair users is important. Therefore, research to corroborate this study?s findings, and/or to identify new predictors will help to accurately identify individuals at risk for lowered self-efficacy with using a wheelchair. When considering the expected increase in wheelchair-use due to population aging, the identification and treatment of individuals with or at risk of lower WC self-efficacy may help to improve the lives of many individuals, in addition to reducing burden on health care systems. It will also be sensible to identify those predictors of WC self-efficacy using a social cognitive framework, and to determine those sources of information most effective at modifying WC self-efficacy in older, wheelchair users.  From a measurement perspective, it will be important to further validate the 13-item mobility and 8-item self-management efficacy subscales, in addition to the shortened 21-item WheelCon, in different subgroups of wheelchair users. The use of prospective studies with a priori hypotheses regarding the associations between self-efficacy and other relevant variables will serve to validate the measures, and may contribute to their more widespread use, and a better understanding of the measures? limitations. Evaluating each of the measures? test-retest reliability will also be beneficial, as would inquiry into the measures? responsiveness and clinically important differences, which will be particularly informative for intervention studies. Next, although the functioning of the recoded 0 to 10 response format is demonstrated to outperform the original 0 to 100 format, the use of even fewer response options may prove to be more ideal. Therefore, research to develop and test the optimal response format for use with the WheelCon may be worthwhile. It may also be worthwhile to develop thresholds to differentiate between low and high WC self-efficacy. Although in some cases the use of such thresholds may help to identify individuals requiring further evaluation, in other cases the use of thresholds may lead to false beliefs of need because as previously mentioned, not all individuals with lowered self-efficacy require efficacy enhancements.   135 Not only does Social Cognitive Theory (Bandura, 1997) identify self-efficacy as the key determinant of behaviour and behaviour change, it also provides mechanisms on how to modify self-efficacy (i.e. performance accomplishment, vicarious learning, verbal persuasion, and interpretation of physiological/affective states). Therefore, the development and testing of efficacy-enhancing interventions utilizing tenets of Social Cognitive Theory is a key area of future investigation. For example, in terms of performance accomplishment, an individual?s experience of success may improve their self-efficacy, and disappointments may reduce it. For older individuals, dividing a task into its subskills, establishing several small reachable goals, and allowing for gradual progress may be particularly useful at enhancing self-efficacy (Lachman et al., 1997). Through the use of vicarious learning, individuals who are uncertain about their abilities to use their wheelchair may benefit from observing others? (model) performances (Bandura, 1997). The characteristics of the models, however, should be taken into account because the information they convey is likely to be highly influential, especially for older individuals. Models with comparable characteristics such as age, sex, socio-economic status, health condition, and lifestyles to that of the observer will provide more valuable information than models with dissimilar features (Bandura, 1997). Providing realistic, verbal positive appraisals and/or encouragement may also act to enhance self-efficacy. Important to this type of strategy, however, is that the verbal encouragement is from a trusted and credible source, because unrealistic encouragement from a non-credible source may act to diminish self-efficacy (Bandura, 1997). Another important means to modify self-efficacy is by correcting negative interpretations of physiological responses to various behaviours, with more positive explanations (Bandura, 1997). Importantly, future intervention research should focus on subgroups of wheelchair users at risk of lowered self-efficacy, and who would benefit the most from treatment. Furthermore, because the subgroups at risk of lower WC self-efficacy may respond differently to different treatments, it may then be important to develop and test different strategies or a combination of strategies in order to realize the greatest impact.   Finally, after demonstrating the efficacy of treatment strategies at improving belief?s in ability with using a wheelchair, it will be important to evaluate the impact of such change on all of participation, mobility, wheelchair skills, and perceived participation limitations. Such investigations will either substantiate or challenge my findings on the importance of self-efficacy  136 with using a wheelchair, and the mechanisms through which it operates to influence participation and mobility in older, community-dwelling manual wheelchair users.                              137 REFERENCES  Agree, E.M., Freedman, V.A., & Sengupta, M. (2004). Factors influencing the use of mobility technology in community-based long-term care. Journal of Aging and Health, 16, 267-307. Anaby, D., Miller, W.C., Eng, J.J., Jarus, T., Noreau, L., & the PACC Research Group. (2009). Can personal and environmental factors explain participation of older adults? Disability and Rehabilitation, 31(15), 1275-1282.  Andresen, E.M. (2008). Criteria for assessing the tools of disability outcomes research. Archives of Physical Medicine and Rehabilitation, 81(S2), S15?S20. Andrich, D. (1978). A rating formulation for ordered response categories. Psychometrika, 43, 561-573. Arnadottir, S.A., Lundin-Olsson, L., Gunnarsdottir, E.D., & Fisher, A.G. (2010). Application of Rasch analysis to examine the psychometric aspects of the Activities-Specific Balance Confidence Scale when used in a new cultural context. Archives of Physical Medicine and Rehabilitation, 91, 156-163. Auger, C., Demers, L., Gelinas, I., Routhier, F., Jutai, J., Guerette, C., & Deruyter F. (2009). Development of a French-Canadian version of the Life-Space Assessment (LSA-F): Content validity, reliability and applicability for power mobility device users. Disability and Rehabilitation Assistive Technology, 4(1), 31-41. Auger, C., Demers, L., Gelinas, I., Miller, W.C., Jutai, J.W., & Noreau, L. (2010). Life-space mobility of middle-aged and older adults at various stages of usage of power mobility devices. Archives of Physical Medicine and Rehabilitation, 91, 765-773. Baker, P.S., Bodner, E.V., & Allman, R.M. (2003). Measuring life-space mobility in community-dwelling older adults. Journal of the American Geriatrics Society, 51(11), 1610-1614.  Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change. Psychological Review, 84(2), 191-215. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Bandura, A. (2006). Guide for creating self-efficacy scales. In: Pajares, F., Urdan, T. Editor (Eds.). Self-efficacy beliefs of adolescents. Greenwich, CT: Information Age Publishing.  138 Barker, D.J., Reid, D., & Cott, C. (2004). The experience of senior stroke survivors: factors in community participation among wheelchair users. Canadian Journal of Occupational Therapy, 73, 18-25. Barlow, J.H., Turner, A.P., & Wright, C.C. (2000). A randomized controlled trial of the Arthritis Self-Management Programme in the UK. Health Education Research, 15(6), 665-680. Baum, C.M. (2011). Fulfilling the promise: Supporting participation in daily life. Archives of Physical Medicine and Rehabilitation, 92, 169-175. Bentley, J.P., Brown, C.J., McGwin, Jr. G., Sawyer, P., Allman, R.M., & Roth, D.L. (2012). Functional status, life-space mobility, and quality of life: a longitudinal mediation analysis. Quality of Life Research, [Epub ahead of print], PMID:23161329. Best, K.L., Kirby, R.L., Smith, C., & MacLeod, D.A. (2005). Wheelchair skills training for community-based manual wheelchair users: A randomized controlled trial. Archives of Physical Medicine and Rehabilitation, 86, 2316-2323. Best, K.L. & Miller,W.C. (2011). Physical and Leisure Activity in Older Community-Dwelling Canadians Who Use Wheelchairs: A Population Study. Journal of Aging Research, 2011, 147929.  Bourbonniere, M.C., Fawcett, L.M., Miller, W.C., Garden, J., & Mortenson, W.B. (2007). Prevalence and predictors of need for seating intervention and mobility for persons in long-term care. Canadian Journal on Aging, 26(3), 195-204. Bjelland, I., Dahl, A.A., Tangen Haug, T., & Neckelmann, D. (2002). The validity of the Hospital Anxiety and Depression Scale: an updated literature review. Journal of Psychosomatic Research, 52, 69-77. Brekke, M., Hjortdahl, P., & Kvien, T.K. (2001). Involvement and satisfaction: A Norwegian study of health care among 1024 patients with rheumatoid arthritis and 1509 patients with chronic noninflammatory musculoskeletal pain. Arthritis Care and Research, 45, 8-15.  Cardol, M., de Jong, B.A., van den Bos, G.A.M., Beelen, A., de Groot, I.J.M., & de Haan, R.J. Beyond disability: perceived participation in people with a chronic disabling condition. (2002). Clinical Rehabilitation, 2002, 16, 27-35. Chase, M.A., Ewing, M.E., George, T.R., & Lirgg, C.D. (1994). The effects of equipment modification on children?s self-efficacy and basketball shooting performance. Research Quarterly for Exercise and Sport, 65(2), 159-168.  139 Chaves, E.S., Boninger, M.L., Cooper, R., Fitzgerald, S.G., Gray, D.B., & Cooper, R.A. (2004). Assessing the influence of wheelchair technology on perception of participation in spinal cord injury. Archives of Physical Medicine and Rehabilitation, 85, 1854-1858. Clarke, P., & Colantonio, A. (2005). Wheelchair use among community-dwelling older adults: Prevalence and risk factors in a national sample. Canadian Journal on Aging, 24(2), 191-198. Clarke, P., Chan, P., Lina Santaguida, P., & Colantonio, A. (2009). The use of mobility devices among institutionalized older adults. Journal of Aging and Health, 21, 611-626. Cohen, S. & Hoberman, H.M. (1983). Positive events and social supports as buffers of life change stress. Journal of Applied Social Psychology, 12, 99-125. Cohen, J. (1988). Statistical power analysis for the behavioural sciences 2nd Ed. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Coolen, A.L., Kirby, R.L., Landry, J., MacPhee, A.H., Dupuis, D., Smith C, Best, K.L?MacLeod, D.A. (2004). Wheelchair skills training program for clinicians: a randomized controlled trial with occupational therapy students. Archives of Physical Medicine and Rehabilitation, 85, 1160-1167. Cooper, R.A., Boninger, M.L., Spaeth, D.M., Ding, D., Guo, S., Koontz, A.M., & Fitzerald, S.G?Collins, DM. (2006). Engineering better wheelchairs to enhance community participation. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 14(4), 438-455.  Cooper, R.A., Ferretti, E., Oyster, M., Kelleher, A., & Cooper, R. (2011). The relationship between wheelchair mobility patterns and community participation among individuals with spinal cord injury. Assistive Technology, 23, 177-183. Curtis, K.A., Roach, K.E., Applegae, E.B., Amar, T., Benvow, C.S., Genecco, T.D,? Gualano, J. (1995a). Development of the Wheelchair-user?s Shoulder Pain Index (WUSPI). Paraplegia, 33(5), 290-293. Curtis, K.A., Roach, K.E., Applegate, E.B., Amar, T., Benbow, C.S., Genecco, T.D., & Gualano J. (1995b). Reliability and validity of the Wheelchair User?s Shoulder Pain Index (WUSPI). Paraplegia, 33(10), 595-601. de Ayala, R.J. (2009). The theory and practice of item response theory. New York, NY: The Guildford Press.  140 del Toro, C.M., Bislick, L.P., Comer, M., Velozo, C., Romero, S., Gonzalez Rothi, L.J., & Kendall, D.L. (2011). Development of a short form of the Boston Naming Test for individuals with aphasia. Journal of Speech, Language, and Hearing Research, 54, 1089-1100. Dijkers M. (1999). Measuring quality of life: Methodological issues. American Journal of Physical Medicine and Rehabilitation, 78(3), 286-300. Dolbier, C.L., & Steinhardt, M.A. (2000). The development and validation of the Sense of Support Scale. Behavioral Medicine, 25, 169-179.  Dubuc, N., Haley, S.M., Ni, P., Kooyoojian, J.T., Jette, A.M. (2004). Function and disability in late life: comparison of the Late-Life Function and Disability Instrument to the Short-Form-36 and the London Handicap Scale. Disability and Rehabilitation, 26(6), 362-370.  Embretson, S.E., & Reise, S.P. (2000). Item Response Theory for Psychologists, 2nd Ed. Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Field, A. (2009). Discovering Statistics Using SPSS 3rd Ed. Thousand Oaks,CA:SAGE Publications, Inc. Fitzpatrick, R., Davey, C., Buxton, M.J., & Jones, D.R. (1998). Evaluation of patient-based outcome measures for use in clinical trials. Health Technology Assessment, 2, 1-74. Fleiss-Douer, O., van der Woude, L.H., & Vanlandewijck, Y.C. (2011). Development of a new scale for perceived self-efficacy in manual wheeled mobility: a pilot study. Journal of Rehabilitation Medicine, 43(7), 602-608. Fleiss-Douer, O., Vanlandewijck, Y.C., & van der Woude, L.H. (2013). Reliability and validity of perceived self-efficacy in wheeled mobility scale among elite wheelchair dependent athletes with a spinal cord injury. Disability and Rehabilitation, 35(10), 851-859. Frazier, P.A., Tix, A.P., & Barron, K.E. (2004). Testing moderator and mediator effects in counseling psychology research. Journal of Counseling Psychology, 51(1), 115-134. Folstein, M.F., Folstein, S.E., & McHugh, P.R. (1975). ?Mini-mental state?: a practical method for grading the state of patients for the clinician. Journal of Psychiatric Research, 12, 189-198. Ganesh, S., Hayter, A., Kim, J., Sanford, J., Sprigle, S., & Hoenig, H. (2007). Wheelchair use by veterans newly prescribed a manual wheelchair. Archives of Physical Medicine and Rehabilitation, 88, 434-439.  141 Gompertz, P., Pound, P., & Ebrahim, S. (1994). A postal version of the Barthel Index. Clinical Rehabilitation, 8, 233-239.  Groll, D.L., To, T., Bombardier, C., & Wright, J.G. (2005). The development of a comorbidity index with physical function as the outcome. Journal of Clinical Epidemiology, 58(6), 595-602.  Heinemann, A.W. (2006). Putting outcome measurement in context: a rehabilitation psychology perspective. Rehabilitation Psychology, 50(1), 6-14. Hirvensalo, M., Rantanen, T., & Heikkinen, E. (2000). Mobility difficulties and physical activity as predictors of mortality and loss of independence in the community-living older population. Journal of the American Geriatrics Society, 48, 493-498. Hoenig, H., Landerman, L.R., & Shipp, K.M. (2003a). Activity restriction among wheelchair users. Journal of the American Geriatrics Society, 51, 1244?1251. Hoenig, H., Taylor, D.H., & Sloan, F.A. (2003b). Does assistive technology substitute for personal assistance among the disabled elderly? American Journal of Public Health, 93(2), 330-337. Hoenig, H., Landerman, L.R., Shipp, K.M., Pieper, C., Richardson, M., Pahel, N., & George, L. (2005). A clinical trial of a rehabilitation expert clinician versus usual care for providing manual wheelchairs. Journal of the American Geriatrics Society, 53, 1712-1720. Horn, W., Yoels, W., Wallace, D., Macrina, D., & Wrigley, M. (1998). Determinants of self-efficacy among persons with spinal cord injury. Disability and Rehabilitation; 20(4), 138-141. Hosseini, S.M., Oyster, M.L., Kirby, R.L., Harrington, A.L., & Boninger, M.L. (2012). Manual wheelchair skills capacity predicts quality of life and community integration in persons with spinal cord injury. Archives of Physical Medicine and Rehabilitation, 93, 2237-2243.  Iezzoni, L.I., McCarthy, E.P., Davis, R.B., & Seibens, H. (2001). Mobility difficulties are not only a problem of old age. Journal of General Internal Medicine, 16, 235-243. Inkpen, P., Parker, K., & Kirby, R.L. (2012). Manual wheelchair skills capacity versus performance. Archives of Physical Medicine and Rehabilitation, 93, 1009-1013. Jain, N.B., Higgins, L.D., Katz, J.N., & Garshick, E. (2010). Association of shoulder pain with the use of mobility devices in persons with chronic spinal cord injury. Physical Medicine and Rehabilitation, 2, 896-900.  142 Jette, A.M., Haley, S.M., Coster, W.J., Kooyoomjian, J.T., Levenson, S., Heeren, T.,?Ashba, J. (2002). The Late Life Function and Disability Instrument: I. Development and evaluation of the disability component. Journal of Gerontology A Medical Sciences, 57A(4), M209-M216. Jette, A.M., Haley, S.M., & Kooyoomjian, J.T. (2003). Are the activities and participation dimensions distinct? Journal of Rehabilitation Medicine, 35, 145-149. Kaye, H.S., Kang, T., & LaPlante M.P. (2000). Mobility device use in the United States. Disability Statistics Report (14). Washington, DC: US Department of Education, National Institute on Disability and Rehabilitation Research. Kemp, B.J., Bateham, A.L., Mulroy, S.J., Thompson, L., Adkins, R.H., & Kahan, J.S. (2011). Effect of reduction in shoulder pain on quality of life and community activities among people living long-term with SCI paraplegia: a randomized control trial. Journal of Spinal Cord Medicine, 34(3), 278-284. Keysor, J.J., Jette, A.M., & Haley, S.M. (2005). Development of the Home and Community Environment (HACE) instrument. Journal of Rehabilitation Medicine, 37, 37-44. Keysor, J.J., Jette, A.M., Valley, M.P., Lewis, C.E., Torner, J.C., Nevitt, M.C?..the Multicenter Osteoarthritis (MOST) group. (2010). Community environmental factors are associated with disability in older adults with functional limitations: The MOST study. Journal of Gerontology A Medical Sciences, 65A(4), 393-399. Kilkens, O.J.E., Post, M.W.M., Dallmeijer, A.J., van Asbeck, F.W.A., & van der Woude, L.H.V. (2005). Relationship between manual wheelchair skill performance and participation of persons with spinal cord injuries 1 year after discharge from inpatient rehabilitation. Journal of Rehabilitation and Research Development, 42(3), 65-74.  Kirby, R.L., Dupuis, D.J., MacPhee, A.H., Coolen, A.L., Smith, C., Best, K.L?Bonaparte, J.P. (2004a). The Wheelchair Skills Test (version 2.4): measurement properties. Archives of Physical Medicine and Rehabilitation, 85, 794-804. Kirby, R.L., Mifflen, N.J., Thibault, D.L., Smith, C., Best, K.L., & MacLeod, D.A. (2004b). The manual wheelchair handling skills of caregivers and the effect of training. Archives of Physical Medicine and Rehabilitation, 85, 2011-2019. Kleinbaum, D.G., Sullivan, K.M., & Barker, N.D. (2007). A pocket guide to epidemiology. New York,NY: Springer Science+Business Media.  143 Kleinbaum, Kupper, Nizam, & Muller. (2008). Applied regression analysis and other multivariate methods 4th Ed. Pacific Grove, CA: Duxbury Pr. Kleinbaum, D.G. & Klein, M. (2010). Logistic regression, statistics for biology and health 3rd Ed. New York, NY: Springer Science+Business Media. Konicki Di lorio, C. (2005). Measurement in health behaviour: Methods for research and evaluation. San Francisco, CA: Jossey-Bass. Krause, J., Carter, R.E., & Brotherton, S. (2009). Association of mode of locomotion and independence in locomotion with long term outcomes after spinal cord injury. Journal of Spinal Cord Medicine, 32(3), 237-248. Lachman, M., Jette, A., Tennstedt, S., Howland, J., Harris, B., & Peterson, E. (1997). A cognitive?behavioural model for promoting regular physical activity in older adults. Psychology Health and Medicine, 2, 251?261. Lachman, M.E., Howland, J., Tennstedt, S., Jette, A., Assmann, S., & Peterson, EW. (1998). Fear of falling and activity restriction: the Survey of Activities and Fear of Falling in the Elderly (SAFE). Journals of Gerontology B Psychological Sciences Social Sciences, 53, 43-50. Laliberte Rudman, D., Hebert, D., & Reid, D. (2006). Living in a restricted occupational world: The occupational experiences of stroke survivors who are wheelchair users and their caregivers. Canadian Journal of Occupational Therapy, 73(3), 141-152. LaPlante, M.P. & Kaye, H.S. (2010). Demographic trends in wheeled mobility equipment use and accessibility in the community. Assistive Technology, 22, 2-17.  Lemay, V., Routhier, F., Noreau, L., Phang, S.H., & Martin-Ginis, K.A. (2012). Relationship between wheelchair skills, wheelchair mobility and level of injury in individuals with spinal cord injury. Spinal Cord, 50, 37-41. Linacre, J.M. (1994). Sample size and item calibration stability. Rasch Measurement Transactions, 7, 4. Linacre, J.M. (2002). Optimizing rating scale category effectiveness. Journal of Applied Measurement, 3, 85-106. Linacre, J.M. (2009). A User?s Guide to WINSTEPS. Chicago, IL: MESA Press. Lirgg, C.D. (1991). Gender differences in self-confidence in physical activity:  A meta-analysis of recent studies. Journal of Sport and Exercise Psychology, 13, 294-310.  144 Lorig, K., Stewart, A., Ritter, P., Gonzalez, V., Laurent, D., & Lynch, J. (1996). Outcome measures for health education and other health care interventions. Thousand Oaks:Sage Publication. Lorig, K.R., & Holman, H.R. (2003). Self-management education: History, definition, outcomes, and mechanisms. Annals of Behavioral Medicine, 26(1), 1-7. MacPhee, A.H., Kirby, R.L., Coolen, A.L., Smith, C., & MacLeod, D.A. (2004). Wheelchair skills training program: a randomized clinical trial on wheelchair users undergoing initial rehabilitation. Archives of Physical Medicine and Rehabilitation, 85, 41-50. Maly, M.R., Costigan, P.A., & Olney, S.J. (2007). Self-efficacy mediates walking performance in older adults with knee osteoarthritis. Journals of Gerontology A Biological Sciences Medical Sciences, 62, 1142-1146. Mann, W.C., Llanes, C., Justiss, M.D., & Tomita, M. (2004). Frail older adults? self-report of their most important assistive device. OTJR: Occupation, Participation and Health, 24(1), 4-12. Martin-Ginis, K.A., Latimer, A.E., Arbour-Nicitopoulos, K.P., Buchholz, A.C., Bray, S.R., Craven, C.,?Wolfe, D.L. (2010). Leisure time physical activity in a population-based sample of people with spinal cord injury part I: Demographic and injury related correlates. Archives of Physical Medicine and Rehabilitation, 91(5), 722-728.  Martin-Ginis, K.A., Latimer, A.E., Arbour-Nicitopoulos, K.P., Bassett, R.L., Wolfe, D.L., & Hanna, S.E. (2011). Determinants of physical activity among people with spinal cord injury: A test of Social Cognitive Theory. Annals of Behavioral Medicine, 42, 127-133. Martin-Ginis, K.A., Arbour-Nicitopoulos, K.P., Latimer-Cheung, A.E., Buchholz, A.C., Bray, S.R., Craven, B.C?Horrocks, J. (2012). Predictors of leisure time physical activity among people with spinal cord injury. Annals of Behavioral Medicine, 44, 104-118. M?sse, L.C., Heesch, K.C., Eason, K.E., & Wilson, M. (2006). Evaluating the properties of a stage-specific self-efficacy scale for physical activity using classical test theory, confirmatory factor analysis and item response modeling. Health Education Research, 21(S1), i33-i46.  Mathieson, K.M., Kronenfeld, J.J., & Keith, V.M. (2002). Maintaining functional independence in elderly adults: The roles of health status and financial resources in predicting home modifications and use of mobility equipment. The Gerontologist, 42, 24?31.  145 May, D., Nayak, U.S.L., & Isaacs, B. (1985). The life-space diary. A measure of mobility in old people at home. International Rehabilitation Medicine, 7, 182?186. McAuley, E., Morris, K.S., Doerksen, S.E., Motl, R.W., Hu, L., White, S.M?Rosengren, K. (2007). Effects of change in physical activity on physical function limitations in older women: Mediating roles of physical function performance and self-efficacy. Journal of the American Geriatrics Society, 55, 1967-1973. McColl, M.A., Davies, D., Carlson, P., Johnston, J., & Minnes, P. (2001). The Community Integration Measure: Development and preliminary validation. Archives of Physical Medicine and Rehabilitation, 82, 429-434.  McVeigh, S.A., Hitzig, S.L., & Craven, B.C. (2009). Influence of sport participation on community integration and quality of life:  A comparison between sport participants and non-sport participants with spinal cord injury. Journal of Spinal Cord Medicine, 32(2), 115-124. Meland, E., Gunnar Maeland, J., Laerum, E. (1999). The importance of self-efficacy in cardiovascular risk factor change. Scandinavian Journal of Public Health, 1, 11-17. Meyers, A.R., Anderson, J.J., Miller, D.R., Shipp, K., & Hoenig, H. (2002). Barriers, facilitators, and access for wheelchair users: substantive and methodologic lessons from a pilot study of environmental effects. Social Science and Medicine, 55, 1435-1446. Miller, W.C., Speechley, M., Deathe, A.B., & Koval, J. (2001). The influence of falling, fear of falling and balance confidence on prosthetic mobility and social activity among individuals with a lower extremity amputation. Archives of Physical Medicine and Rehabilitation, 82, 1238-44.  Miller, W.C., Miller, F., Trenholm, K., Grant, D., & Goodman, K. (2004). Development and preliminary assessment of the measurement properties of the Seating Identification Tool (SIT). Clinical Rehabilitation, 18, 317-325.  Miller, W.C., Sakakibara, B.M., & Rushton, P.W. (2012). The prevalence of low confidence with using a wheelchair and its relationship to wheelchair skills. The Gerontologist, 52(S1), 205. Morris, K.S., McAuley, E., & Motl, R.W. (2008). Neighborhood satisfaction, functional limitations, and self-efficacy influences on physical activity in older women. International Journal of Behavioral Nutrition and Physical Activity, 5, 13.  146 Mortenson, W.B., Miller, W.C., Backman, C.L., & Oliffe, J.L. (2011). Predictors of mobility among wheelchair using residents in long-term care. Archives of Physical Medicine and Rehabilitation, 92, 1587-1593. Mortenson, W.B., Miller, W.C., Backman, C.L., & Oliffe, J.L. (2012). Association between mobility, participation, and wheelchair-related factors in long-term care residents who use wheelchairs as their primary means of mobility. Journal of the American Geriatrics Society, 60, 1310-1315.  Motl, R.W., & Snook, E.M. (2008). Physical activity, self-efficacy and quality of life in multiple sclerosis. Annals of Behavioral Medicine, 35(1), 111-115.  Netuveli, G., Wiggins, R.D., Hildon, Z., Montgomery, S.M., & Blane, D. (2006). Quality of life at older ages: evidence from the England longitudinal study of aging (wave 1). Journal of Epidemiology and Community Health, 60, 357-63.  Sedgwick, M., Frank, A.O., Kemp, P., Gage, P. (Eds). (2005). Improving services for wheelchair users and carers ? Good Practice Guide: Learning from the Wheelchair Services Collaborative. London: DH. Norman, G.R. & Streiner, D.L. (2008). Biostatistics: The bare essentials. 3rd ed. Hamilton, ON: BC Decker Inc. Nunnally, J.C. (1967). Psychometric Theory. New York, NY: McGraw Hill. Oyster, M.L., Karmarkar, A.M., Patrick, M., Schmidt Read, M., Nicolini, L., & Boninger, M.L. (2011). Investigation of factors association with manual wheelchair mobility in persons with spinal cord injury. Archives of Physical Medicine and Rehabilitation, 92, 484-490. Pang, M.Y.C., Eng, J.J., Miller, W.C. (2007). Determinants of satisfaction with community reintegration in older adults with chronic stroke: Role of balance self-efficacy. Physical Therapy, 87(3), 282-291. Pang, M.Y.C., & Eng, J.J. (2008). Fall-related self-efficacy, not balance and mobility performance, is related to accidental falls in chronic stroke survivors with low bone mineral density. Osteoporosis International, 19, 919-927. Pang, M.Y., Eng, J.J., Lin, K-H., Tang, P-F., Hung, C., Wang, Y-H. (2009). Association of depression and pain interference with disease-management self-efficacy in community-dwelling individuals with spinal cord injury. Journal of Rehabilitation Medicine, 41, 1068-1073.  147 Patrick, D.L., Kinne, S., Engelberg, R.A., & Pearlman, R.A. (2000). Functional status and perceived quality of life in adults with and without chronic conditions. Journal of Clinical Epidemiology, 53, 779-785. Peel, C., Sawyer-Baker, P., Roth, D.L., Brown, C.J., Bodner, E.V., & Allman, R.M. (2005). Assessing mobility in older adults: The UAB Study of Aging Life-Space Assessment. Physical Therapy, 85, 1008-1019. Pellett, T.L., & Lox, C.L. (1998). Tennis racket head-size comparisons and their effect on beginning college players' achievement and self-efficacy. Journal of Teaching in Physical Education, 17, 453-467. Phang, S.H., Martin-Ginis, K.A., Routhier, F., & Lemay, V. (2012). The role of self-efficacy in the wheelchair skills-physical activity relationship among manual wheelchair-users with spinal cord injury. Disability and Rehabilitation, 34(8), 625-632.  Podsakoff, P.M., MacKenzie, S.B., & Podsakoff, N.P. (2012). Sources of method bias in social science research and recommendations on how to control for it. Annual Review of Psychology, 63, 539-569. Portney, L.G. & Watkins, M.P. (2009). Foundations of clinical research: Applications to practice 3rd Ed. Upper Saddle River, NJ: Prentice-Hall Inc. Preacher, K.J. & Hayes, A.F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879-891.  Rasch, G. (1980). Probabilistic models for some intelligence and attainment tests. Chicago, IL: The University of Chicago Press. Ravenek, K.E., Ravenek, M.J., Hitzig, S.L., & Wolfe, D.L. (2012). Assessing quality of life in relation to physical activity participation in persons with spinal cord injury: A systematic review. Disability and Health Journal, 5, 213-223.  Rejeski, W.J., Miller, M.E., Foy, C., Messier, S., & Rapp, S. (2001). Self-efficacy and the progression of functional limitations and self-reported disability in older adults with knee pain. Journals of Gerontology B Psychological Sciences Social Sciences, 56(5), S261-S265.  148 Riggins, M.S., Kankipati, P., Oyster, M.L., Cooper, R.A., & Boninger, M.L. (2011). The relationship between quality of life and change in mobility 1 year postinjury in individuals with spinal cord injury. Archives of Physical Medicine and Rehabilitation, 92, 1027-1033. Roelands, M., Van Oost, P., Depoorter, A.M., Buysse, A. (2002). A social cognitive model to predict the use of assistive technology for mobility and self-care in elderly people. The Gerontologist, 42(1), 39-50. Rosenberg, D.E., Huang, D.L., Simonovich, S.D., & Belza, B. (2013). Outdoor built environment barriers and facilitators to activity among midlife and older adults with mobility disabilities. The Gerontologist, 53(2), 268-279. Rousseau-Harrison, K., Rochette, A., Routhier, F., Dessureault, D., Thibault, F., & Cote, O. (2009). Impact of wheelchair acquisition on social participation. Disability and Rehabilitation Assistive Technology, 4(5), 344-352. Routhier, F., Vincent, C., Desrosiers, J., & Nadeau, S. (2003). Mobility of wheelchair users: a proposed performance assessment framework. Disability and Rehabilitation, 25(1), 19-34. Rubenstein, L.Z., Powers, C.M., & MacLean, C.H. (2001). Quality indicators for the management and prevention of falls and mobility problems in the vulnerable elderly. Annals of Internal Medicine, 135, 686-693. Rushton, P.W., Miller, W.C., Kirby, R.L., Eng, J.J., & Yip, J. (2011). Development and content validation of the Wheelchair Use Confidence Scale: a mixed-methods study. Disability and Rehabilitation Assistive Technology, 6(1), 57-66. Rushton, P.W., Kirby, R.L., & Miller, W.C. (2012). Manual wheelchair skills: objective testing versus subjective questionnaire. Archives of Physical Medicine and Rehabilitation, 93(12), 2313-2318. Rushton, P.W., Miller, W.C., Kirby, R.L., & Eng, J.J. (2013). Measure for the assessment of confidence with manual wheelchair use (WheelCon-M) version 2.1: Reliability and validity. Journal of Rehabilitation Medicine, 45(1), 61-67.  Sackett, D.L., Straus, S.E., Richardson, W.S., Rosenberg, W., & Haynes, R.B. (2000). Evidence based medicine: how to practice and teach EBM 2nd Ed., London: Churchill Livingstone. Sakakibara, B.M., Miller, W.C., Orenczuk, S.G., Wolfe, D.L., & the SCIRE Research Team. (2009). A systematic review of depression and anxiety measures used with individuals with spinal cord injury. Spinal Cord, 47(12), 841-851.  149 Sakakibara, B.M., Miller, W.C., & Backman, C.L. (2011). Rasch analyses of the Activities-specific Balance Confidence scale with individuals 50 years and older with lower-limb amputations. Archives of Physical Medicine and Rehabilitation, 92, 1257-1263. Sakakibara, B.M., Miller, W.C., Eng, J.J., Backman, C.L., & Routhier, F. (2013a). Preliminary examination of the relationship between participation and confidence in older manual wheelchair-users. Archives of Physical Medicine and Rehabilitation, 94(4), 791-794. Sakakibara, B.M., Miller, W.C., Souza, M., Nikolova, V., & Best, K.L. (2013b). Wheelchair skills training to improve confidence with using a manual wheelchair among older adults: A pilot study. Archives of Physical Medicine and Rehabilitation, 94(6), 1031-1037.  Sakakibara, B.M., Routhier, F., Lavoie, M-P, & Miller, W.C. (2013c). Reliability and validity of the French-Canadian version of the Late Life Function and Disability Instrument (LLFDI-F) in older, community-living wheelchair-users. Scandinavian Journal of Occupational Therapy, in-press. Salbach, N.M., Mayo, N.E., Robichaud-Ekstrand, S., Hanley, J.A., Richards, C.L., & Wood-Dauphinee, S. (2005). The effect of a task-oriented walking intervention on improving balance self-efficacy poststroke: A randomized, controlled trial. Journal of the American Geriatrics Society, 53, 576-582. Sanford, J.A., Griffiths, P.C., Richardson, P., Hargraves, K., Butterfield, T., & Hoenig, H. (2006). The effects of in-home rehabilitation on task self-efficacy in mobility-impaired adults: A randomized controlled clinical trial. Journal of the American Geriatrics Society, 54, 1641-1648. Sapey, B., Stewart, J., & Donaldson, G. (2005). Increases in wheelchair use and perceptions of disablement. Disability and Society, 20(5), 489-505. Schmid, A.A., Van Puymbroeck, M., Altenburger, P.A., Dierks, T.A., Miller, K.K., Damush, T.M., Williams, L.S. (2012). Balance and balance self-efficacy are associated with activity and participation after stroke: A cross-sectional study in people with chronic stroke. Archives of Physical Medicine and Rehabilitation, 93(6), 1101-1107. Seeman, T.E., Unger, J.B., McAvay, G., & Mendes de Leon, C.F. (1999). Self-efficacy beliefs and perceived declines in functional ability: MacArthur studies of successful aging. Journals of Gerontology B Psychological Sciences Social Sciences, 54B(4), 214-222. Shields, M. (2004). Use of wheelchairs and other mobility devices. Health Reports, 15, 37-41.  150 Shrestha, L.B., & Heisler, E.J. (2011). The changing demographic profile of the United States. Congressional Research Service, 27-5700, RL32701. Shore, S.L. (2008). Use of an economical wheelchair in India and Peru: Impact on health and function. Medical Science Monitor: International Medical Journal of Experiment and Clinical Resarch, 14(12), 71-79. Shore, S.L., & Juillerat, S. (2012). The impact of a low cost wheelchair on the quality of life of the disabled in the developing world. Medical Science Monitor: International Medical Journal of Experiment and Clinical Resarch, 18(9), 533-542. Smith Anderson-Bill, E., Winett, R.A., Wojcik, J.R., & Williams, D.M. (2011). Aging and the social cognitive determinants of physical activity behavior and behavior change: Evidence from the Guide to Health trial. Journal of Aging Research, 2011, 1-11. Statistics Canada. (2006). A portrait of seniors in Canada. Ottawa: Statistics Canada. Statistics Canada (2013). Low income lines, 2011-2012. Ottawa: Statistics Canada. Streiner, D.L., & Norman, G.R. (2008). Health Measurement Scales: A Practical Guide to Their Development and Use, 4th Ed. Oxford: Oxford University Press. Threats, T. (2007). Access for persons with neurogenic communication disorders: Influences of personal and environmental factors of the ICF. Aphasiology, 21(1), 67-80.  Tonack, M., Hitzig, S.L., Craven, B.C., Campbell, K.A., Boschen, K.A., & McGillivray, C.F. (2008). Predicting life satisfaction after spinal cord injury in a Canadian sample. Spinal Cord, 46, 380-385. van Leeuwen, C.M., Post, M.W., Westers, P., van der Woude, L.H., de Groot, S., Sluis, T?Lindeman, E. (2012). Relationships between activities, participation, personal factors, mental health, and life satisfaction in persons with spinal cord injury. Archives of Physical Medicine and Rehabilitation, 93, 82-89.  Velozo, C.A., Seel, R.T., Magasi, S., Heinemann, A.W., & Romero, S. (2012). Improving measurement methods in rehabilitation: Core concepts and recommendations for scale development. Archives of Physical Medicine and Rehabilitation, 93(8 Suppl 2), S154-S163. Vignier, N., Ravaud, J-F., Winance, M., Lepoutre, F-X., & Ville, I. (2008). Demographics of wheelchair users in France: Results of national community-based handicaps-incapacites-dependance surveys. Journal of Rehabilitation Medicine, 40, 231-239.  151 von Bonsdorff, M., Rantanen, T., Laukkanen, P., Suutama, T., & Heikkinen, E. (2006). Mobility limitations and cognitive deficits as predictors of institutionalization among community-dwelling older people. Gerontology, 52, 359-365. Wang, W.C., & Chen, C.T. (2005). Item parameter recovery, standard error estimates, and fit statistics of the Winsteps program for the family of Rasch models. Educational and Psychological Measurement, 65, 376-404. Warner, G., Basiletti, M., & Hoenig, H. (2010). Age differences in leisure physical activity by adult wheelchair users. Physical and Occupational Therapy in Geriatrics, 28(4), 410-421. Webber, S.C., Porter, M.M., & Menec, V.H. (2010). Mobility in older adults: A comprehensive framework. The Gerontologist, 50(4), 443-450. Wee, J., & Lysaght, R. (2009). Factors affecting measures of activities and participation in persons with mobility impairment. Disability and Rehabilitation, 31(20), 1633-1642. Wheelchair Skills Program Manual. Version 4.1. (2008). Retrieved February 9, 2013, from www.wheelchairskillsprogram.ca.  Wilcox, S., Bopp, M., Oberrecht, L., Kammermann, S.K., & McElmurray, C.T. (2003). Psychosocial and perceived environmental correlates of physical activity in rural and older African American and white women. Journals of Gerontology B Psychological Sciences Social Sciences, 58, 329-337. Wilcox, S., Sharpe, P.A., Hutto, B., & Granner, M.L. (2005). Self-efficacy for exercise questionnaire in a diverse sample of men and women. Journal of Physical Activity and Health, 3, 285-297. Wilkie, R., Peat, G., Thomas, E., & Croft, P. (2006). The prevalence of person-perceived participation restriction in community-dwelling older adults. Quality of Life Research, 15, 1471-1479. Wilkie, R., Peat, G., Thomas, E., & Croft, P. (2007). Factors associated with participation restriction in community-dwelling adults aged 50 years and over. Quality of Life Research, 16, 1147-1156.  Wilkie, R., Thomas, E., Mottram, S., Peat, G., & Croft, P. (2008). Onset and persistence of person-perceived participation restriction in older adults: a 3-year follow up study in the general population. Health and Quality of Life Outcomes, 6, 92-102.  152 Wilson, M., Allen, D.D., & Corser Li, J. (2006). Improving measurement in health education and health behavior research using item response modeling: introducing item response modeling. Health Education Research 21(S1), i4-i18. World Health Organization. (2001). International Classification of Functioning, Disability and Health. Geneva, CH: Author. Wright, B., & Stone, M. (1999). Measurement Essentials, 2nd Ed. Wilmington, DE: Wide Range, Inc. Zigmond, A.S., & Snaith, R.P. (1983). The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scandinavica, 67, 361?370.             	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ?	 ? 153 APPENDICES                                    154 Appendix A:   Variables/measures organized by the International Classification of Functioning, Disability and Health 	 ?ICF domain Variable Measure # of items Focus and Scoring Studies providing measurement evidence Participation  Frequency Late Life Disability Instrument (Jette et al., 2002) 16 The frequency of participating in two role domains (social and personal). Response scale: 1 (never) to 5 (very often).   0 to 100: Higher scores indicate greater frequency. Older adults (Jette et al., 2002; Dubuc et al., 2004)  Older wheelchair users (Sakakibara et al., 2013)   Limitations  Late Life Disability Instrument (Jette et al., 2002)  16  The extent of limitation in performing tasks in two role domains (instrumental and management). Response scale: 1 (completely) to 5 (not at all).   0 to 100: Higher scores indicate fewer limitations.  Older adults (Jette et al., 2002) Older wheelchair users (Sakakibara et al., 2013) Activity  Life-space mobility Life-Space Assessment (Baker et al., 2003) 20 Questions pertain to the frequency of movement in five areas (within the home, around the home, in the neighbourhood, in town, and outside of town) over the past four weeks, and if any assistance (from persons or with equipment) was used.   0 to 120: Higher scores indicate more life-space mobility.  Older adults (Baker et al., 2003)   Wheelchair skills Wheelchair Skills Test ? Questionnaire (WSTP manual, 2008) 32 Participants are asked if they can safely complete a wheelchair skill. Responses are given a pass or fail.   0 to 100: Higher scores indicate more skills. Wheelchair users (Rushton et al., 2012)   Functional independence  Barthel Index ? postal version (Gompertz et al., 1994)  10  Ability to perform activities of daily living is rated. Response scales differ for each item.  0 to 20: Higher scores indicate more independence.   Individuals with stroke (Gompertz et al., 1994)  155 ICF domain Variable Measure # of items Focus and Scoring Studies providing measurement evidence Body functions  Self-efficacy Wheelchair Use Confidence Scale (Rushton et al., 2011) 65 Self-efficacy in six areas including maneuvering around the physical environment, performing activities, knowledge of the wheelchair and solving problems, social situations, advocacy, and emotions. Response scale: 0 (low) to 100 (high).   0 to 100: Higher scores indicate more self-efficacy. Wheelchair-users (Rushton et al., 2011; Rushton et al., 2013)   Depression/ Anxiety  Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983)  7/7  Depression and anxiety symptoms experienced during the past week are rated on a scale from 0 (not at all) to 3 (very often indeed).   0 to 21: Higher scores indicate more severe symptoms on both subscales.  General population (Zigmond & Snaith, 1983; Bjelland et al., 2002) Individuals with spinal cord injury (Sakakibara et al., 2010)   Pain  Wheelchair Users? Shoulder Pain Index (Curtis et al., 1995a)  15  The degree of shoulder pain experienced while performing various activities are identified on a 10cm visual analog scale. Participants also had the option to select ?item not performed?.   0 to 150: Higher scores indicate more pain.  Wheelchair users (Curtis et al., 1995a; Curtis et al., 1995b) Health condition  Comorbidities Functional Comorbidity Index (Groll et al., 2005) 18 Participants respond as either yes or no when asked if a doctor has diagnosed them with any of the 18 health conditions.  0 to 18: Higher scores indicate more comorbidity.  Individuals with spinal cord injury (Groll et al., 2005)  Diagnosis     Socio-demographic information form 1 Neurological condition or not (yes/no).   156 ICF domain Variable Measure # of items Focus and Scoring Studies providing measurement evidence Personal factors  Age Socio-demographic information form  1    Sex Socio-demographic information form  1 Male or Female.   Education Socio-demographic information form  1 High school or less (yes or no).   Income Socio-demographic information form  1 <$30,000/?$30,000/prefer not to answer.   Marital status Socio-demographic information form  1 Married or common-law (yes/no).   Employment/ volunteer status  Socio-demographic information form 1 Employed or volunteer (yes/no).   Experience Socio-demographic information form  1 Years of experience with using a wheelchair.   Daily use   Socio-demographic information form 1 Hours of daily wheelchair use.   Formal training Socio-demographic information form  1 Received training outside of rehabilitation (yes/no).   Assistance Socio-demographic information form 1 Require assistance with using a wheelchair (e.g. transfers, set-up, supervision) (yes/no).      157 ICF domain Variable Measure # of items Focus and Scoring Studies providing measurement evidence Environmental factors Wheelchair-related  Need for a seating intervention Seating Identification Tool (Miller et al., 2004) 11 Participants respond as either yes or no when asked questions about pressure, discomfort behaviours, mobility, positioning, and stability.  0 to 15: Higher scores indicate higher need for a seating interventions.  Wheelchair users (Miller et al., 2004) Personal environment Social support Interpersonal Support and Evaluation List-6 (Cohen & Hoberman, 1983)  6 Perceived social support is rated on scale ranging from 0 (definitely false) to 3 (definitely true).  0 to 18: Higher scores indicate more social support. Various populations (Cohen & Hoberman, 1983; Dolbier et al., 2000; McColl et al., 2001)   Physical environment  Barriers ? home/ community  Home and Community Environment Instrument (Keysor et al., 2005)  9/5  Amount of physical barriers in the home and community are quantified.  0 to 10: Higher scores indicate more home barriers. 0 to 5: Higher scores indicate more community barriers.  Adults with mobility limitations (Keysor et al., 2005)      	 ?        158 Appendix B:   Barthel Index    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________  Directions: These are some questions about your ability to look after yourself. They may not seem to apply to you.   Please answer them all.  Tick one box in each section.  1. Bathing  In the bath or shower, manage on your own? do you:   need help getting in and out?       need other help?         never have a bath or shower?       need to be washed in bed?   2. Stairs  Do you climb stairs  without any help?     at home:      with someone carrying your frame?       with someone encouraging you?       with physical help?      not at all?          don?t have stairs?   3. Dressing  Do you get dressed? without any help?      just with help with buttons?      with someone helping you most of the time?             159  4. Mobility  Do you walk indoors: without any help apart from a frame?      with one person watching over you?      with one person helping you?      with more than one person helping?       not at all?      or do you use a wheelchair  independently?   5. Transfer   Do you move from  on your own?  bed to chair:     with a little help from one person?      with a lot of help from one or  more people?      not at all   6. Feeding  Do you eat food:  without any help?      with help cutting food or spreading     butter?      with more help?   7. Toilet Use  Do you use the toilet without any help? or the commode:     with some help but can do something?      with quite a lot of help?             160  8. Grooming  Do you brush your  without help? hair and teeth,  wash your face   with help? and shave:      9. Bladder  Are you incontinent  never? of urine?     less than once a week?      less than once a day?      more often?      or do you have a catheter managed     for you?   10. Bowels  Do you soil yourself? never?      occasional accident?       all the time?      Or do you need someone to give  you an enema?                                              161 Appendix C:   Demographic Information Form    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   1) Year of Birth _______________ 2) Age:(years) ______________ 3) Gender:  1 = Male    2 = Female  4) Current Marital Status:  1 = single/never married 2 = married 3 = common-law  4 = divorced 5 = widowed  5) Highest Education Level: 1 = primary or elementary school 2 = some high school 3 = graduated from high school 4 = some college or university 5 = graduated from college or university 6 = post-graduate degree 7 = other    6) Current Employment Status: (circle all that apply) 1 = employed, part-time paid 2 = employed, full-time paid 3 = unemployed, seeking employment 4 = unemployed, not seeking employment 5 = volunteer, part-time 6 = volunteer, full-time 7= student, part-time 8 = student, full-time 9 = retired  7) What is your annual household income? 1 = <14 999 2 = 15 000 ? 29 999 3 = 30 000 ? 44 999 4 = 45 000 ? 59 999 5 = 60 000 ? 74 999 6 = >75 000 7 = Prefer not to answer    8) What is your primary diagnosis accounting for manual wheelchair use?   ________________________________________________________________  9) Years with diagnosis?___________________________________________   10) Type of wheelchair (i.e. make, model, with/without footrests).  ________________________________________________________________   162  11) Time using a manual wheelchair (years/months)?___________________ 12) Have you received any formal wheelchair skills training? 1 = No 2 = Yes   13) Where do you use your wheelchair? (circle all that apply) 1 = home 2 = work 3 = school 4 = community 5 = recreation and leisure 6 = other  14) How many hours are you in your wheelchair everyday, on average?____  15) What assistance do you require with your wheelchair? (circle all that apply) 1 = no assistance required 2 = supervision (someone nearby, just in case) 3 = verbal assistance (instructions, reminders) 4 = assistance with wheelchair set-up  5 = assistance with transfers 6 = other:_____________  16) Is this paid assistance?   1 = No     2 = Yes                   163 Appendix D:   Functional Comorbidity Index    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   The questions contained in this questionnaire will be concerning a variety of health conditions.  When I read off the health condition, I will ask that you indicate yes or no if you have or had this particular condition.  That is, has a doctor told you that you have this health condition?  Item number Disease Yes No 1 Arthritis (rheumatoid and OA)   2 Osteoporosis   3 Asthma   4 Chronic Obstructive Pulmonary Disease (COPD), acquired respiratory distress syndrome(ARDS), or emphysema   5 Angina   6 Congestive heart failure (or heart disease)   7 Heart attack (myocardial infarct)   8 Neurological disease (such as multiple sclerosis or Parkinson?s)   9 Stroke or TIA   10 Peripheral vascular disease   11 Diabetes type I and II   12 Upper gastrointestinal disease (ulcer, hernia, reflux)   13 Depression   14 Anxiety or panic disorders   15 Visual impairment (such as cataracts, glaucoma, macular degeneration)   16 Hearing impairment (very hard of hearing, even with hearing aids)   17 Degenerative disc disease (back disease, spinal stenosis or severe chronic back pain)   18 Obesity and/or body mass index > 30 (weight in kg/height in meters2)   TOTAL    ?yes? = 1 ?no? = 0      164 Appendix E:   Home and Community Environment Instrument    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   Home Mobility The following questions are about your home and community.  There are no right or wrong answers. I just want your thoughts based on your current situation.     Let?s begin with a few questions about your home. Your ?home? is the place where you live, sleep, and eat.  If you live in a building (for example, an apartment building, an assisted living center, or a nursing home), your home would include both the building and the rooms in which you live.  Do you have any questions before we begin?  1. What type of home do you live in? Single Family Multi-Family  Apartment Building or Condominium Complex Congregate Housing or Assisted Living Nursing/ Rest Home Other    1    2    3    4     5     6  2. How many steps are at the main entrance of your home? (Probe, by main entrance we mean the entrance the respondent uses or is likely to use.  A side entrance is acceptable)  None  One or two  Several 10 or more  1 (Skip to #03)  2  3  4   2b.  Is there a railing at the steps?  No Yes  2  1  3. Is there a ramp at the main entrance? (Probe for main entrance is in question #2) No Yes  2  1  4.  Does the door at the main entrance of your home open electronically or is someone available to open the door? (Probe for main entrance is in question #2)  No Yes  2  1        165 5.  (IF RESPONDENT LIVES IN SINGLE FAMILY HOME, DO NOT ASK THE NEXT TWO QUESTIONS)  How many steps are there from the main entrance of your building to your main living areas.  By main living area, we mean the rooms in which you live, sleep, and eat.  None One or two  Several 10 or more    1    2    3    4  6. Is there a chairlift or elevator inside your building?  No  Yes  2  1  7. How many steps are there in your main living area.  By main living area, we mean the rooms in which you live, sleep, and eat. None One or two Several 10 or more    1    2    3    4  8. Is there a chairlift or elevator inside your main living area? By main living area, we mean the rooms in which you live, sleep, and eat. No  Yes  2  1   Community Mobility Now I would like to learn about your local community.  By ?local community? I mean the neighborhood you live in.   Using the answers listed on the card please tell me whether your community has ?a lot,? ?some,? or ?not at all? of what I describe.  If you don?t know, please answer ?don?t know.?  Do you have any questions?   To what extent does your local community have? A lot  Some Not at all Don?t Know 1. Uneven sidewalks or other areas that you go to 1 2 3 4 2. Parks and areas that you go to that are easy to get to and easy to use 1 2 3 4 3. Safe parks or areas that you go to 1 2 3 4 4. Places to sit and rest at bus stops, in parks, or in other places where people walk 1 2 3 4 5. Curbs with curb cuts. By curb cuts we mean little ramps at sidewalk and street corners that make it easy for wheel chairs to move through these areas. (rephrased to improve agreement) 1 2 3 4     166 Appendix F:   Hospital Anxiety and Depression Scale    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   This questionnaire is designed to help us know how you feel.  Read each item and circle the number corresponding to the response that comes closest to how you have been feeling in the past week.  Don?t take too long over your replies.  Your immediate reaction to each item will probably be more accurate than a long thought out response.     A I feel tense or 'wound up':   Most of the time  3  A lot of the time  2  From time to time, occasionally  1  Not at all  0   D I still enjoy the things I used to enjoy:   Definitely as much  0  Not quite so much  1  Only a little  2  Hardly at all  3   A  I get a sort of frightened feeling as if something awful is about to happen:   Very definitely and quite badly 3  Yes, but not too badly 2  A little, but it doesn't worry me 1  Not at all 0   D I can laugh and see the funny side of things:   As much as I always could 0  Not quite so much now 1  Definitely not so much now 2  Not at all 3      167 A Worrying thoughts go through my mind:   A great deal of the time 3  A lot of the time 2  From time to time, but not too often 1  Only occasionally 0   D I feel cheerful:   Not at all 3  Not often 2  Sometimes 1  Most of the time 0   A I can sit at ease and feel relaxed:   Definitely 0  Usually 1  Not Often 2  Not at all 3    D I feel as if I am slowed down:   Nearly all the time 3  Very often 2  Sometimes 1  Not at all 0    A  I get a sort of frightened feeling like 'butterflies' in the stomach:   Not at all 0  Occasionally 1  Quite Often 2  Very Often 3              D I have lost interest in my appearance:   Definitely 3  I don't take as much care as I should 2  I may not take quite as much care 1  I take just as much care as ever 0   168 A I feel restless as I have to be on the move:   Very much indeed 3  Quite a lot 2  Not very much 1  Not at all 0    D I look forward with enjoyment to things:   As much as I ever did 0  Rather less than I used to 1  Definitely less than I used to 2  Hardly at all 3    A I get sudden feelings of panic:   Very often indeed 3  Quite often 2  Not very often 1  Not at all 0   D I can enjoy a good book or radio or TV program:   Often 0  Sometimes 1  Not often 2  Very seldom 3                     169 Appendix G:   Interpersonal Support and Evaluation List ? 6 Item    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   This scale is made up of a list of statements each of which may or may not be true about you. For each statement check "definitely true" if you are sure it is true about you and "probably true" if you think it is true but you are not absolutely certain.  Similarly, you should check "definitely false" if you are sure the statement is false and "probably false" if you think it is false but are not absolutely certain.   definitely true probably true probably false definitely false 1. When I feel lonely, there are several people I can talk to. r? r? r? r? 2. I often meet or talk with family or friends. r? r? r? r? 3. If I were sick, I could easily find someone to help me with my daily activities. r? r? r? r? 4. When I need suggestions on how to deal with a personal problem, I know someone I can turn to. r? r? r? r? 5. If I had to go out of town for a few weeks, it would be difficult to find someone who would look after my house or apartment (the plants, pets, garden, etc.). r? r? r? r? 6. There is at least one person I know whose advice I really trust. r? r? r? r?                 170 Appendix H:   Late Life Disability Instrument    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   Disability Questions How often do you? To what extent do you feel limited in??  Very Often Often Once in a while Almost Never  Never Not at All A little Somewhat A lot Completely D1. Keep (Keeping) in touch with others through letters, phone or email. 5 4 3 2 1 5 4 3 2 1 D2. Visit (Visiting) friends and family in their homes. 5 4 3 2 1 5 4 3 2 1 D3. Provide (Providing) care or assistance to others. This may include providing personal care, transportation, and running errands for family members or friends. 5 4 3 2 1 5 4 3 2 1 D4. Take (Taking) care of the inside of your home. This includes managing and taking responsibility for home making, laundry, room cleaning and minor household repairs. 5 4 3 2 1 5 4 3 2 1 D5. Work (Working) at a volunteer job outside your home. 5 4 3 2 1 5 4 3 2 1 D6. Take (Taking) part in active recreation. This may include bowling, golf, tennis, hiking, jogging, or swimming. 5 4 3 2 1 5 4 3 2 1        171 Disability Questions, continued How often do you? To what extent do you feel limited in?.?  Very Often Often Once in a while Almost  Never Never Not at All A little Somewhat A lot Completely D7. Take (Taking) care of household business and finances. This may include managing and taking responsibility for your money, paying bills, dealing with a landlord or tenants, dealing with utility companies or governmental agencies. 5 4 3 2 1 5 4 3 2 1 D8. Take (Taking) care of your own health. This may include managing daily medications, following a special diet, scheduling doctor?s appointments. 5 4 3 2 1 5 4 3 2 1 D9. Travel (Traveling) out of town for at least an overnight stay. 5 4 3 2 1 5 4 3 2 1 D10. Take (Taking) part in a regular fitness program. This may include walking for exercise, stationary biking, weight lifting, or exercise classes. 5 4 3 2 1 5 4 3 2 1 D11. Invite (Inviting) people into your home for a meal or entertainment. 5 4 3 2 1 5 4 3 2 1 D12. Go (Going) out with others to public places such as restaurants or movies. 5 4 3 2 1 5 4 3 2 1           172  Disability Questions, continued How often do you? To what extent do you feel limited in?.?  Very Often Often Once in a while Almost  Never Never Not at All A little Somewhat A lot Completely D13. Take (Taking) care of your own personal care needs. This includes bathing, dressing, and toileting. 5 4 3 2 1 5 4 3 2 1 D14. Take (Taking) part in organized social activities. This may include clubs, card playing, senior center events, community or religious groups. 5 4 3 2 1 5 4 3 2 1 D15. Take (Taking) care of local errands. This may include managing and taking responsibility for shopping for food and personal items, and going to the bank, library, or dry cleaner. 5 4 3 2 1 5 4 3 2 1 D16. Prepare (Preparing) meals for yourself. This includes planning, cooking, serving, and cleaning up. 5 4 3 2 1 5 4 3 2 1                  173 Appendix I:   Life-Space Assessment    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________  These questions refer to your activities just within the past month. Life Space Level Frequency Independence Score During the past four weeks, have you         been to ? How often did you get there? Did you use aids or equipment? Did you need help from another person? Level x Frequency x Independence Life Space Level 1 Other rooms of your home besides the room where you sleep?  Yes  1  No  0 Less than 1/wk 1 1-3 times/wk 2 4-6 times/wk 3  Daily  4 1    = personal assistance 1.5 = equipment only 2    = no equipment or personal            assistance   ___________ Level 1 Score Life Space Level 2 An area outside your home such as your porch, deck or patio,  hallway (of an apartment building) or garage, in your own yard or driveway?    Yes  2   No  0  Less than 1/wk  1  1-3 times/wk  2  4-6 times/wk  3   Daily   4  1    = personal assistance 1.5 = equipment only 2    = no equipment or personal   assistance     ___________ Level 2 Score Life Space Level 3  Places in your neighborhood, other than your own yard or apartment building?  Yes  3  No  0 Less than 1/wk 1 1-3 times/wk 2 4-6 times/wk 3  Daily  4 1    = personal assistance 1.5 = equipment only 2    = no equipment or personal assistance   ___________ Level 3 Score Life Space Level 4  Places outside your neighborhood, but within your town?  Yes  4  No  0 Less than 1/wk 1 1-3 times/wk 2 4-6 times/wk 3  Daily  4 1    = personal assistance 1.5 = equipment only 2    = no equipment or personal assistance   ___________ Level 4 Score Life Space Level 5 Places outside your town? Yes  5 No   0 Less than 1/wk 1 1-3 times/wk 2 4-6 times/wk 3 Daily  4 1    = personal assistance 1.5 = equipment only 2    = no equipment or personal assistance   ___________ Level 5 Score Total score              174 Appendix J:   Mini Mental State Examination    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   Score 1 for every correct answer:  1. What year is it?         _____  2. What season are we in?        _____   3. What month are we in?        _____  4. What is today?s date?        _____  5. What day of the week is it?       _____  6. What country are we in?        _____  7. What province are we in?        _____    8. What city are we in?        _____  9. What hospital are we in?        _____  10.  What floor of the hospital are we on?      _____  Name three objects (?Ball,? ?Car,? ?Man?).  Take a second to pronounce each word.  Then ask the patient to repeat all 3 words.  Take into account only correct answers given on the first try.  Repeat these steps until the subject learns all the words.  11.  Ball?          _____  12.  Car?          _____  13. Man?          _____  Either ?please spell the word WORLD and now spell it backwards? or ?Please count from 100 subtracting 7 every time?  14. ?D? or 93          _____  15. ?L? or 86          _____      175 16. ?R? or 79          _____  17. ?O? or 72          _____  18. ?W? or 65          _____  What were the 3 words I asked you to remember earlier?  19. Ball?          _____  20. Car?          _____  21. Man?          _____  Show the subject a pen and ask: ?Could you name this object??  22. Pen.          _____  Show the subject your watch and ask: ?Could you name this object??  23. Watch          _____  Listen and repeat after me:   24.  ?No ifs, ands, or buts.?        _____  Put a sheet of paper on the desk and show it while saying: ?Listen carefully and do as I say.?  25. Take the sheet with your left/right (unaffected) hand.    _____  26. Fold it in half.         _____  27. Put in on the floor.         _____   Show the patient the visual instruction page directing him/her to ?CLOSE YOUR EYES? and say:  28.  Do what is written on this page.       _____  Give the subject a blank sheet and a pen and ask:  29.  Write or say a complete sentence of your choice.    _____         176 Give the patient the geometric design page and ask:  30.  Could you please copy this drawing?      _____   Total Score: (/30) _____                     177 Appendix K:   Seating Identification Tool    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________  Within the last four (4) weeks: Yes No 1. Has the individual had red areas on their bottom?   2. Has the individual had an open pressure sore on their bottom?   3. Has the individual had red areas on their back?   4. Has the individual had an open pressure sore on their back?   5. Has the individual reported or demonstrated behaviours that indicate they could be in discomfort or pain while sitting for any length of time? (such as moaning, grimacing, or agitation?   6. Has the individual had difficulty propelling their wheelchair?  (if the individual does not propel their wheelchair circle 0)   7. Has the individual required repositioning as a result of sliding or leaning?   8. Has an anti-slide device such as a foam bolster, pommel, roll bar, posture pal, or posey restraint been used?   9. Have rolled blankets, pillows or homemade devices been used to prevent leaning?   10. Has the individual not been using a wheelchair seat cushion? (do not include linens, pillows, incontinence pads, or home made foam cushions)   11. Has the individual tipped their wheelchair or been at risk of tipping their wheelchair?    Overall Score          178 Appendix L:   Wheelchair Skills Test ? Questionnaire 4.1    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________    Tester: _______________        Time start: _________               Time finish:  _________  For about the next 15 minutes, I will be asking you questions about a number of different skills that you might perform in your wheelchair. We want to see if you can perform the skill properly and safely. We do not expect that you are able to perform every skill. If you do not understand the question, feel free to ask for a further explanation. Do you have any general questions now, before we begin??  For each skill in the following table, ask subjects whether he/she believes themselves to be capable of performing each wheelchair skill and, if so, how he/she would perform the skill.  E.g. ?Can you roll forward 10 meters??, ?How would you do it??, ?Anything else??                                179 Individual Skills Skill  Safety Comments ?? ??  ?? ??  NT  1 Rolls forward 10m        2 Rolls forward 10m in 30s        3 Rolls backward 5m        4 Turns 90? while moving forward L&R        5 Turns 90? while moving backward L&R        6 Turns 180?in place L&R        7 Maneuvers sideways L&R         8 Gets through hinged door in both directions        9 Reaches 1.5m high object        10 Picks object from floor       11 Relieves weight from buttocks       12 Transfers from WC to bench and back       13 Folds and unfolds wheelchair      No Part q?  14 Rolls 100m       15 Avoids moving obstacles L&R       16 Ascends 5? incline        17 Descends 5? incline        18 Ascends 10? incline        19 Descends 10? incline        20 Rolls 2m across 5? side-slope L&R        21 Rolls 2m on soft surface       22 Gets over 15cm pot-hole       23 Gets over 2cm threshold       24 Ascends 5cm level change       25 Descends 5cm level change       26 Ascends 15cm curb       27 Descends 15cm curb       28 Performs 30s stationary wheelie       29 Turns 180? in place in wheelie position L&R       30 Gets from ground into wheelchair        31 Ascends stairs        32 Descends stairs         Total Percentage Scores      (Total passed skills/Total applicable) skills)      180 Appendix M:   Wheelchair Use Confidence Scale for Manual Wheelchair Users v3.0    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   Instructions:  A number of situations are described below that can challenge confidence when using a manual wheelchair.  Please rate how confident you are as of now for each of the situations described using the following scale:        0      10     20     30     40     50     60     70     80     90      100       Not                     Completely                 confident                             confident         For example, a person may be 82% confident they can memorize a grocery list of 5 items, but only 63% confident they can memorize a grocery list with 10 items.  For items requiring physical ability, rate your confidence in performing the activity in a safe manner. For this assessment, confidence refers to your belief in your ability to perform each item independently.  Answer all items even if it is not a situation you would normally experience. If you have never experienced the situation, please rate your confidence as if you had to safely attempt it today.    Some questions include measurement, such as 5cm. Please refer to the ruler on the last page of this assessment if you are uncertain about these measurements.                	 ?  181 As of now, how confident are you that you:  Confidence (0-100) 1) can move your wheelchair over carpet?  2) can move your wheelchair around furniture in your own home?  3) can move your wheelchair over thresholds, such as between rooms?  4) can manoeuvre your wheelchair in small spaces, such as a bathroom?  5) can transfer from your wheelchair to your bed?  6) can transfer from your wheelchair to your toilet?  7) can transfer from your wheelchair into your bathtub (including use of bathseats) OR using your commode to get into your shower stall?  8) can transfer from the floor to your wheelchair by yourself?  9) can transfer from your wheelchair to your vehicle?  10) can make a light meal while using your wheelchair?  11) can carry a hot drink while moving in your wheelchair?  12) can move your wheelchair through a door that opens automatically?  13) can open, go through, and then close a standard 81cm (32?) lightweight door?  14) can open and go through a spring loaded door, such as a door at your local mall?  15) can move your wheelchair up a standard ramp, built to code (5? incline)?  16) can move your wheelchair down a standard ramp, built to code (5? incline)?   17) can move your wheelchair up a dry steep slope (> 5? incline)?   	 ?  182 As of now, how confident are you that you:  Confidence (0-100) 18) can move your wheelchair down a dry steep slope (> 5? incline)?  19) can move your wheelchair down a dry steep slope (> 5? incline) and stopping as soon as you are off the slope?  20) can move your wheelchair up a curb cut?  21) can move your wheelchair down a curb cut?  22) can move your wheelchair over a drainage grate and then up a curb cut?  23) can move your wheelchair down a curb cut then over a drainage grate?  24) can move your wheelchair through a puddle then up a curb cut?   25) can move your wheelchair down a curb cut then through a puddle?  26) can move your wheelchair through slush then up a curb cut?  27) can move your wheelchair down a curb cut then through slush?  28) can move your wheelchair down a curb cut then through 5cm (2?) snow?  29) can move your wheelchair through 5cm (2?) snow then up a curb cut?  30) can move your wheelchair up a standard height curb 15cm (6?) without a curb cut?  31) can move your wheelchair down a standard height curb 15cm (6?) without a curb cut?  32) can manoeuvre your wheelchair to press the crosswalk button and cross the street before the traffic light changes?  33) can cross a street with light traffic at a crosswalk with no traffic lights?  	 ?  183 As of now, how confident are you that you:  Confidence (0-100) 34) can move your wheelchair across 3m (10ft) of flat, freshly mowed, dry grass?  35) can move your wheelchair through a pothole that is wider than your wheelchair and 5cm (2?) deep?  36) can move your wheelchair along a paved sidewalk that is cracked and uneven?  37) can move your wheelchair along a flat dirt path or trail with some tree roots and rocks? New item 38) can move your wheelchair across 3m (10ft) of flat, unpacked gravel?  39) can move your wheelchair along a sidewalk with 5cm (2?) of snow?  40) can move your wheelchair through a crowd of people without hitting anyone?  41) can ask people to move out of your way while moving in your wheelchair?  42) can move your wheelchair down a store aisle that has just enough room for your wheelchair without knocking items over?  43) can manage all toileting activities while in an accessible public bathroom?  44) can use public transportation in your town?  45) can do your chosen leisure activities in your manual wheelchair?  46) can transport items in a backpack that is on the back of your wheelchair?  47) can use strategies, such as humour, that will help people feel comfortable if they are unsure how to act because you use a wheelchair?  48) can correct others? mistaken beliefs about people who use wheelchairs?  	 ?  184 As of now, how confident are you that you:  Confidence (0-100) 49) can present yourself as you wish to be seen while in your wheelchair around acquaintances, colleagues, or peers?  50) can present yourself as you wish to be seen while in your wheelchair when you are in public and feel people are watching you?  51) can present yourself as you wish to be seen while in your wheelchair when you want to impress others, such as during a job interview?   52) can problem solve how to get to your destination when there is an unexpected situation, such as construction detours on a sidewalk?  53) can figure out how to negotiate a challenging, and unusual physical obstacle? New item 54) can continue to move your wheelchair in a situation that is making you feel anxious or nervous? New item 55) know when your wheelchair is not working properly?  56) know what your wheelchair can and can?t do, separate from your own abilities? For example, a wheelchair can go down stairs but many individuals do not go down stairs with their wheelchair due to their inability to do so.   57) can tell someone how to move your wheelchair if it gets stuck?  58) can ask someone for help? New item 59) can tell a cab driver how to fold/unfold your wheelchair, making sure all parts are taken off and put back on properly?  	 ?  185 As of now, how confident are you that you:  Confidence (0-100) 60) can tell a stranger how to help you safely get back into your wheelchair if you tip over?  61) know what to do if you fall out of your wheelchair? New item 62) can advocate for changes you want made to your wheelchair, such as a different cushion to be more comfortable?  63) can advocate for changes you want in your home, such as doorways widened or a ramp installed?  64) can advocate for your needs at work or school, such as modifications in the bathroom?  65) can advocate for changes in your community, such as having a curb cut added in your neighborhood to improve your accessibility?     Measurement Scale 0-5cm   0 cm       3 cm      5 cm                   186 Appendix N:   Wheelchair Use Confidence Scale for Manual Wheelchair Users v2.4    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________   Date: ___________________      Subject No: ___________ Instructions: This questionnaire will assess your level of confidence when using your wheelchair to do different activities. For this assessment, confidence refers to your belief in your ability to do the activities itself.    For example, an answer to the question ?How confident are you that you can lift a 5 lb. box?? might be 82%, whereas ?How confident are you that you can lift a 10 lb box?? might be 48%.   Answer all items even if they are activities you would not normally do or are unsure about.    Rate how confident you are by recording a number from 0 to 100 using the scale below:           0     10    20    30      40       50         60        70        80       90       100          Not                  Completely              confident                                                                                            confident                   187 WheelCon- Physical Environment  As of now, how confident are you moving your wheelchair:  Confidence Level 1) Around furniture in your own home?  2) Over carpet?  3) Over thresholds, such as between rooms?  4) Over freshly mowed grass?  5) Through snow?  6) Along a paved sidewalk that is cracked and uneven?  7) Through a pothole on a sidewalk?  8) Along a level path with unpacked gravel?  9) Across a street with light traffic, at a crosswalk without traffic lights?  10) Across a street with traffic, at a crosswalk with traffic lights?  11) Up a standard ramp?  12) Down a standard ramp?   13) Up a steep slope?     188 As of now, how confident are you moving your wheelchair:  Confidence Level 14) Down a steep slope?  15) Down a steep slope and stopping as soon as you are off the slope?  16) Through a crowded mall?  17) Through a store with just enough space between the aisles for your wheelchair?  18) In tight spaces, such as elevators?  19) Through an elevator door?  20) Up a curb cut?  21) Down a curb cut?  22) Over a drainage grate and then up a curb cut?  23) Down a curb cut then over a drainage grate?  24) Through a puddle then up a curb cut?   25) Down a curb cut then through a puddle?  26) Through slush then up a curb cut?    189 As of now, how confident are you moving your wheelchair:  Confidence Level 27) Down a curb cut then through slush?  28) Down a  curb cut then though snow?  29) Through snow then up a curb cut?  30) Up a standard height curb (15cm) without a curb cut?  31) Down a standard height curb without a curb cut?  32) Through a doorway you have just opened and then closing the door behind you?  33) Through a doorway with a spring loaded door?  Physical Environment Subscore   WheelCon- Activities Performed While Using a Manual Wheelchair  As of now, how confident are you in doing the following activities:  Confidence Level 1) Moving from your wheelchair to your bed?  2) Moving from your wheelchair to your toilet?  3) Moving from your wheelchair to your bathtub or shower?    190 As of now, how confident are you in doing the following activities:  Confidence Level 4) Moving from your wheelchair to your car?  5) Moving from the floor to your wheelchair by yourself?  6) Making a meal while using your wheelchair?  7) Moving your wheelchair on or off public transportation?  8) Carrying a hot drink while moving in your wheelchair?  9) Managing toileting activities in public bathroom facilities?  10) Doing leisure activities, such as going to church or clubs?  11) Carrying items, such as groceries, while using your wheelchair?  Activities Performed while using a Manual Wheelchair Subscore                  191 WheelCon- Knowledge and Problem Solving  As of now, how confident are you that you:  Confidence Level 1) know what your wheelchair can and can?t do, separate from your own abilities? For example, a wheelchair with rear anti-tips in place is not able to go up a standard curb, even if the individual is able to do this skill.   2) can recognize a maintenance problem with your wheelchair, such as low tire pressure?  3) can figure out how to move your wheelchair in new situations?  4) can tell someone how to move your wheelchair if it gets stuck?   5) can tell a stranger how to help you safely get back into your wheelchair if you tip over?  6) can tell a cab driver how to fold/unfold your wheelchair, making sure all parts are taken off and put back on properly?  Knowledge and Problem Solving Subscore   WheelCon- Advocacy  As of now, how confident are you that you can:  Confidence Level 1) advocate for changes to your wheelchair, such as a different cushion to 7be more comfortable?  2) advocate for changes in your community, such as having a curb cut added in your neighborhood to improve your accessibility?  3) advocate for changes in your home, such as doorways widened or a ramp installed?    192 As of now, how confident are you that you can:  Confidence Level 4) advocate for changes to your place of work or school, such as modifications in the bathroom?  Advocacy Subscore   WheelCon- Social Situations  As of now, how confident are you that you can:  Confidence Level 1) put others at ease if they are uncomfortable around a person who uses a wheelchair?  2) comfortably correct other people?s wrong assumptions about  people who use wheelchairs?  3) ask people to move out of your way while moving in your wheelchair?  4) present the self-image you want others to see while doing regular daily activities in public?  5) present yourself as you wish to be seen while doing challenging activities in public?  6) present the image you want in situations where there is a desire or need to impress others, such as during a job interview?   Social Situation Subscore            193 WheelCon- Emotions  As of now, how confident are you that you can:   1) manage any anxious or nervous feelings you may have when moving your wheelchair in new environments?  2) manage any anxious or nervous feelings you may have when trying new or more difficult wheelchair skills?   3) stay calm in stressful situations, such as if your wheelchair were to get stuck or your tire were to blow out?  Emotions Subscore  WheelCon Total Score                194 Appendix O:   Wheelchair User Shoulder Pain Index    Date:_______________________________  	 ?	 ? 	 ? 	 ? 	 ? 	 ? 	 ? 	 ? Subject	 ?ID:____________________________________  Place an ?X? on the scale to estimate your level of pain with the following activities. Check box at right if the activity was not performed               in the past week.  Based on your experiences in the past week, how much shoulder pain do you experience when:           Transferring from a bed to a wheelchair?                                         Not Performed      No pain                                                                                 Worst pain ever experienced   [   ]  Transferring from a wheelchair to a car?  No pain                                                                                 Worst pain ever experienced   [   ]  Transferring from a wheelchair to the tub or shower?  No pain                                                                                 Worst pain ever experienced   [   ]  Loading your wheelchair into a car?  No pain                                                                                 Worst pain ever experienced   [   ]    Pushing your chair for 10 minutes or more?         No pain                                                                                 Worst pain ever experienced   [   ]   195 Pushing up ramps or inclines outdoors?          Not Performed      No pain                                                                                 Worst pain ever experienced   [   ]  Lifting objects down from an overhead shelf?        No pain                                                                                  Worst pain ever experienced   [   ]  Putting on pants   No pain                                                                                  Worst pain ever experienced   [   ]  Putting on a t-shirt or pullover?  No pain                                                                                 Worst pain ever experienced   [   ]  Putting on a button down shirt?  No pain                                                                                  Worst pain ever experienced   [   ]  Washing your back?  No pain                                                                                  Worst pain ever experienced   [   ]   Usual volunteer and/or daily activities?  No pain                                                                                Worst pain ever experienced   [   ]  Driving?  No pain                                                                                 Worst pain ever experienced   [   ]  196 Performing household chores?           Not Performed  No pain                                                                                 Worst pain ever experienced   [   ]  Sleeping?  No pain                                                                                 Worst pain ever experienced   [   ]  197 Appendix P:   Late Life Disability Instrument correlation matrix        LLDI-freq WheelCon FCI Age SIT Years of WC exp. Hours of daily WC use ISEL HACE-Home HACE-Comm HADS-D HADS-A WUSPI LSA WST-Q BI LLDI-limit LLDI-freq 1 .54 -.31 -.28 -.13 .10 .17 .39 -.05 -.13 -.24 -.10 -.03 .55 .51 .22 .54 WheelCon .54 1 -.23 -.30 -.29 .23 .27 .12 .06 -.14 -.25 -.24 -.19 .47 .84 .31 .48 FCI -.31 -.23 1 .24 .23 -.25 -.12 -.20 .11 .19 .28 .32 .18 -.34 -.20 -.04 -.25 Age -.28 -.30 .24 1 .05 -.02 -.11 -.01 .11 .14 .05 -.12 -.08 -.24 -.34 -.03 -.08 SIT -.13 -.29 .23 .05 1 .03 -.01 -.05 -.12 .11 .27 .23 .25 -.07 -.27 -.38 -.33 Years WC exp. .10 .23 -.25 -.02 .03 1 .29 .00 -.16 .01 -.15 -.02 .04 .13 .26 -.04 .18 Hours of daily WC use .17 .27 -.12 -.11 -.01 .29 1 .06 -.07 -.09 .00 -.11 -.08 .06 .21 -.05 .12 ISEL .39 .12 -.20 -.01 -.05 .00 .06 1 .12 -.11 -.27 -.20 -.07 .24 .09 .06 .13 HACE-Home -.05 .06 .11 .11 -.12 -.16 -.07 .12 1 .16 -.01 -.06 -.14 -.07 .08 .29 .00 HACE-Comm -.13 -.14 .19 .14 .11 .01 -.09 -.11 .16 1 .15 .09 .10 -.10 -.07 .17 -.16 HADS-D -.24 -.25 .28 .05 .27 -.15 .00 -.27 -.01 .15 1 .58 .33 -.11 -.15 .06 -.44 HADS-A -.10 -.24 .32 -.12 .23 -.02 -.11 -.20 -.06 .09 .58 1 .47 -.10 -.13 .01 -.35 WUSPI -.03 -.19 .18 -.08 .25 .04 -.08 -.07 -.14 .10 .33 .47 1 -.01 -.08 -.07 -.23 LSA .55 .47 -.34 -.24 -.07 .13 .06 .24 -.07 -.10 -.11 -.10 -.01 1 .49 .15 .27 WST-Q .51 .84 -.20 -.34 -.27 .26 .21 .09 .08 -.07 -.15 -.13 -.08 .49 1 .42 .40 BI .22 .31 -.04 -.03 -.38 -.04 -.05 .06 .29 .17 .06 .01 -.07 .15 .42 1 .23 LLDI-limit .54 .48 -.25 -.08 -.33 .18 .12 .13 .00 -.16 -.44 -.35 -.23 .27 .40 .23 1 Bold = p?0.05    198 Appendix Q:   Life-Space Assessment correlation matrix    LSA WheelCon FCI Age SIT Year WC exp. Hours of daily WC use ISEL HACE-Home HACE-Comm WST-Q LSA 1 .47 -.34 -.24 -.07 .13 .06 .24 -.07 -.10 .49 WheelCon .47 1 -.23 -.30 -.29 .23 .27 .12 .06 -.14 .84 FCI -.34 -.23 1 .24 .23 -.25 -.12 -.20 .11 .19 -.20 Age -.24 -.30 .24 1 .05 -.02 -.11 -.01 .11 .14 -.34 SIT -.07 -.29 .23 .05 1 .03 -.01 -.05 -.12 .11 -.27 Years WC exp. .13 .23 -.25 -.02 .03 1 .29 .00 -.16 .01 .26 Hours of daily WC use .06 .27 -.12 -.11 -.01 .29 1 .06 -.07 -.09 .21 ISEL .24 .12 -.20 -.01 -.05 .00 .06 1 .12 -.11 .09 HACE -Home -.07 .06 .11 .11 -.12 -.16 -.07 .12 1 .16 .08 HACE - Comm -.10 -.14 .19 .14 .11 .01 -.09 -.11 .16 1 -.07 WST-Q .49 .84 -.20 -.34 -.27 .26 .21 .09 .08 -.07 1  Bold = p?0.05            199 Appendix R:   WheelCon correlation matrix   WheelCon FCI Age SIT Years WC exp Hours of daily WC use ISEL HACE -Home HACE-Comm WheelCon 1 -.23 -.30 -.29 .23 .27 .12 .06 -.14 FCI -.23 1 .24 .23 -.25 -.12 -.20 .11 .19 Age -.30 .24 1 .05 -.02  -.01 .11 .14 SIT -.29 .23 .05 1 .03 -.01 -.05 -.12 .11 Years WC exp. .23 -.25 -.02 .03 1 .29 .00 -.16 .01 Hours of daily WC use .27 -.12 -.11 -.01 .29 1 .06 -.07 -.09 ISEL .12 -.20 -.01 -.05 .00 .06 1 .12 -.11 HACE Home .06 .11 .11 -.12 -.16 -.07 .12 1 .16 HACE Comm -.14 .19 .14 .11 .01 -.09 -.11 .16 1  Bold = p?0.05               200 Appendix S:   Fit statistics of the items in the Mobility efficacy dimension Mobility efficacy   Infit Outfit  Items (n=46) Mnsq (ZSTD) Mnsq (ZSTD) 1 ?move over carpet?  0.89 (-0.60) 0.72 (-1.40)* 3 ?move over thresholds, such as between rooms?^ 1.20 (1.00) 0.86 (-0.60)* 4 ?move in small spaces, such as a bathroom?^ 0.86 (-0.80) 0.67 (-1.70)* 5 ?transfer to your bed? 1.56 (3.00) 1.53 (2.30) 6 ?transfer to your toilet? 1.64 (3.80) 1.58 (2.70) 7 ?transfer into your bathtub? 1.74 (4.80) 1.55 (2.80) 8 ?transfer from the floor to your wheelchair? 2.71 (9.90) 3.02 (8.60) 9 ?transfer to your vehicle? 1.77 (5.00) 2.47 (6.20) 10 ?make a light meal?^ 1.65 (3.70) 1.68 (2.90) 11 ?carry a hot drink? 1.08 (0.70) 1.02 (0.20)* 12 ?move through a door that opens automatically?^ 1.36 (1.70) 0.70 (-1.40)* 13 ?open, go through, and close a standard door? 0.89 (-0.70) 0.65 (-2.00)* 14 ?open, go through a spring loaded door? 0.61 (-3.80) 0.54 (-3.40) 15 ?move up a standard ramp? 0.91 (-0.60) 0.65 (-2.10) 16 ?move down a standard ramp? 0.89 (-0.60) 0.57 (-2.40) 17 ?move up a dry steep slope? 0.72 (-2.90) 0.61 (-3.00) 18 ?move down a dry steep slope? 0.87 (-1.00) 0.86 (-0.80)* 19 ?move down a dry steep slope and stopping? 0.86 (-1.20) 0.67 (-2.30) 20 ?move up a curb cut?  0.83 (-1.40) 0.89 (-0.60)* 21 ?move down a curb cut? 0.91 (-0.60) 0.87 (-0.70)* 22 ?move over a drainage grate, then up a curb cut?  0.82 (-1.70) 0.73 (-1.90)* 23 ?move down a curb cut then over drainage grate? 0.94 (-0.50) 0.83 (-1.10)* 24 ?move through a puddle then up a curb cut?  0.69 (-3.00) 0.59 (-3.00) 25 ? move down a curb cut then through a puddle?  0.74 (-2.30) 0.59 (-2.80) 26 ?move through slush then up a curb cut?  0.73 (-2.90) 0.71 (-2.20) 27 ?move down a curb cut then through slush?  0.78 (-2.20) 0.72 (-2.10) 28 ?move down a curb cut then through snow?  0.77 (-2.40) 0.76 (-1.70)* 29 ?move through snow then up a curb cut?  0.78 (-2.30) 0.76 (-1.80)* 30 ?move up curb without curb cut? 1.99 (6.20) 2.52 (5.90) 31 ?move down curb without curb cut?  1.73 (5.90) 1.53 (3.20) 32 ?press the crosswalk button and cross the street? 0.95 (-0.30) 0.74 (-1.60)* 33 ?cross a street at a crosswalk with no traffic lights?  0.94 (-0.50) 0.85 (-1.00)* 34 ?move across flat, freshly mowed, dry grass? 0.87 (-1.10) 0.80 (-1.30)* 35 ? move through a pothole? 0.97 (-0.20) 0.87 (-0.90)* 36 ?move along a sidewalk that is cracked/uneven? 0.70 (-2.90) 0.63 (-2.60) 37 ?move along a flat dirt path with tree roots/rocks? 0.76 (-2.10) 0.85 (-0.80)* 38 ?move across flat, unpacked gravel?  1.00 (0.10) 1.09 (0.70)* 39 ?move along a sidewalk with snow?  0.98 (-0.10) 1.05 (0.40)* 40 ?move through a crowd without hitting anyone? 1.24 (1.80) 3.00 (7.80) 42 ?move in store aisle without knocking items over?^ 1.17 (1.30) 1.03 (0.20)*   201 Appendix S:   Fit statistics of the items in the Mobility efficacy dimension cont? 43 ?manage toileting activities in a public bathroom?  1.54 (3.80) 1.69 (3.50) 44 ?use public transportation in your town?  1.97 (6.80) 2.47 (6.80) 45 ?do your chosen leisure activities?  1.56 (3.40) 1.18 (1.00)* 46 ?transport items in a backpack? 1.61 (3.80) 1.16 (0.90)* 53 ?negotiate a challenging and unusual obstacle?^ 1.11 (0.70) 1.55 (2.20) 54 ?move in situations making you anxious/nervous?^ 0.91 (-0.50) 0.97 (-0.10)* *items outfit statistics within acceptable range of 0.60 and 1.40; ^factorially complex items  202 Appendix T: Fit statistics of the Mobility efficacy dimension after eliminating items with   misfitting outfit statistics  Mobility efficacy   Infit Outfit  Items (n=24) Mnsq (ZSTD) Mnsq (ZSTD) 1 ?move  over carpet?  0.94 (-0.30)* 0.83 (-0.90) 3 ?move over thresholds, such as between rooms?^ 1.19 (1.00) 0.87 (-0.60) 4 ?move in small spaces, such as a bathroom?^ 1.03 (0.20)* 0.75 (-1.30) 11 ?carry a hot drink while in your wheelchair? 1.28 (2.20) 1.24 (1.50) 12 ?move through a door that opens automatically?^ 1.49 (2.30) 0.74 (-1.20) 13 ?open, go through, and close a standard door? 1.07 (0.50)* 0.78 (-1.20) 18 ?move down a dry steep slope? 1.05 (0.50)* 1.09 (0.60) 20 ?move up a curb cut?  1.01 (0.10)* 1.08 (0.50) 21 ?move down a curb cut? 0.98 (-0.10)* 1.09 (0.60) 22 ?move over a drainage grate, then up a curb cut?  0.95 (-0.40)* 0.84 (-1.10) 23 ?move down a curb cut then over drainage grate? 1.05 (0.50)* 0.94 (-0.40) 28 ?move down a curb cut then through snow?  0.91 (-0.80)* 0.89 (-0.80) 29 ?move through snow then up a curb cut?  0.95 (-0.50)* 0.94 (-0.40) 32 ?press the crosswalk button and cross the street? 1.11 (0.90)* 0.85 (-0.90) 33 ?cross a street at a crosswalk with no traffic lights?  1.04 (0.40)* 1.03 (0.20) 34 ?move across flat, freshly mowed, dry grass? 0.94 (-0.50)* 0.83 (-1.10) 35 ? move through a pothole? 1.14 (1.30) 1.36 (2.40) 37 ?move along a flat dirt path with tree roots/rocks? 0.76 (-1.90) 0.80 (-1.10) 38 ?move across flat, unpacked gravel?  1.18 (1.70) 1.34 (2.30) 39 ?move along a sidewalk with snow?  1.12 (1.20)* 1.26 (1.80) 42 ?move in store aisle without knocking items over?^ 1.29 (2.0) 1.27 (1.50) 45 ?do your chosen leisure activities?  1.85 (4.80) 1.33 (1.70) 46 ?transport items in a backpack? 1.88 (5.20) 1.37 (1.90) 54 ?move in situations making you anxious/nervous?^ 1.15 (0.90) 1.26 (1.20) *items infit statistics within acceptable range of 0.87 and 1.13; ^factorially complex items            203 Appendix U:   13-item Mobility efficacy subscale score conversion, SEM and reliability Raw Std. SEM Rel.  Raw Std. SEM Rel.  Raw Std. SEM Rel. 0 0.00 22.83 0.25  47 42.18 1.58 0.99  94 51.46 1.81 0.98 1 13.46 11.63 0.56  48 42.37 1.58 0.99  95 51.71 1.82 0.98 2 20.12 7.64 0.75  49 42.56 1.57 0.99  96 51.97 1.84 0.98 3 23.54 5.90 0.83  50 42.74 1.57 0.99  97 52.23 1.86 0.98 4 25.73 4.90 0.88  51 42.93 1.56 0.99  98 52.50 1.88 0.98 5 27.3 4.24 0.91  52 43.12 1.56 0.99  99 52.77 1.90 0.98 6 28.52 3.78 0.92  53 43.30 1.56 0.99  100 53.05 1.93 0.98 7 29.51 3.44 0.94  54 43.48 1.55 0.99  101 53.33 1.95 0.98 8 30.35 3.18 0.95  55 43.67 1.55 0.99  102 53.63 1.98 0.98 9 31.07 2.97 0.95  56 43.85 1.55 0.99  103 53.93 2.01 0.98 10 31.70 2.81 0.96  57 44.03 1.55 0.99  104 54.24 2.04 0.98 11 32.27 2.67 0.96  58 44.21 1.55 0.99  105 54.56 2.07 0.98 12 32.79 2.55 0.97  59 44.40 1.54 0.99  106 54.89 2.11 0.98 13 33.26 2.45 0.97  60 44.58 1.54 0.99  107 55.24 2.15 0.98 14 33.71 2.37 0.97  61 44.76 1.54 0.99  108 55.60 2.19 0.97 15 34.12 2.29 0.97  62 44.94 1.55 0.99  109 55.97 2.24 0.97 16 34.51 2.23 0.97  63 45.12 1.55 0.99  110 56.36 2.29 0.97 17 34.88 2.17 0.97  64 45.31 1.55 0.99  111 56.77 2.34 0.97 18 35.23 2.12 0.98  65 45.49 1.55 0.99  112 57.20 2.41 0.97 19 35.56 2.08 0.98  66 45.67 1.55 0.99  113 57.65 2.48 0.97 20 35.89 2.04 0.98  67 45.85 1.55 0.99  114 58.13 2.55 0.97 21 36.20 2.00 0.98  68 46.04 1.56 0.99  115 58.64 2.64 0.96 22 36.49 1.96 0.98  69 46.22 1.56 0.99  116 59.19 2.74 0.96 23 36.78 1.93 0.98  70 46.41 1.57 0.99  117 59.79 2.85 0.95 24 37.06 1.90 0.98  71 46.60 1.57 0.99  118 60.44 2.98 0.95 25 37.34 1.88 0.98  72 46.79 1.58 0.99  119 61.14 3.13 0.95 26 37.60 1.85 0.98  73 46.97 1.58 0.99  120 61.93 3.30 0.94 27 37.86 1.83 0.98  74 47.17 1.59 0.99  121 62.81 3.51 0.93 28 38.11 1.81 0.98  75 47.36 1.59 0.99  122 63.81 3.76 0.92 29 38.36 1.79 0.98  76 47.55 1.60 0.99  123 64.97 4.06 0.91 30 38.60 1.77 0.98  77 47.75 1.61 0.99  124 66.35 4.45 0.90 31 38.84 1.76 0.98  78 47.95 1.61 0.99  125 68.03 4.96 0.87 32 39.07 1.74 0.98  79 48.14 1.62 0.99  126 70.16 5.66 0.84 33 39.30 1.72 0.98  80 48.35 1.63 0.99  127 73.04 6.70 0.79 34 39.52 1.71 0.98  81 48.55 1.64 0.99  128 77.33 8.46 0.71 35 39.74 1.70 0.98  82 48.76 1.65 0.98  129 85.21 12.44 0.53 36 39.96 1.68 0.98  83 48.96 1.66 0.98  130 100.00 23.55 0.24 37 40.17 1.67 0.98  84 49.18 1.67 0.98      38 40.38 1.66 0.98  85 49.39 1.68 0.98      39 40.59 1.65 0.98  86 49.61 1.69 0.98      40 40.8 1.64 0.99  87 49.83 1.70 0.98      41 41.00 1.63 0.99  88 50.05 1.72 0.98      42 41.20 1.62 0.99  89 50.27 1.73 0.98      43 41.40 1.61 0.99  90 50.50 1.74 0.98      44 41.60 1.60 0.99  91 50.74 1.76 0.98      45 41.79 1.60 0.99  92 50.98 1.77 0.98      46 41.99 1.59 0.99  93 51.22 1.79 0.98      Raw=raw score; Std=standardized score; SEM=standard error of measurement; Rel=reliability      204 Appendix V:   Fit statistics of the items in the Self-management efficacy dimension Self-management efficacy   Infit Outfit  Items (n=25) Mnsq (ZSTD) Mnsq (ZSTD) 3 ?move over thresholds, such as between rooms?^ 1.26 (1.3) 1.08 (0.40)* 4 ?move in small spaces, such as a bathroom?^ 1.12 (0.80) 0.94 (-0.20)* 10 ?make a light meal?^ 1.84 (4.40) 1.76 (3.20) 12 ?move through a door that opens automatically?^ 1.42 (1.90) 0.87 (-0.50)* 41 ?ask people to move out of your way?  1.20 (1.20) 0.90 (-0.40)* 42 ?move in store aisle without knocking items over?^ 1.11 (0.90) 2.31 (5.50) 47 ?use strategies, such as humour, that will help people feel comfortable if they are unsure how to act because you use a wheelchair?  0.91 (-0.50) 1.12 (0.70)* 48 ?correct others? mistaken beliefs about people who use wheelchairs? 0.83 (-1.20) 1.35 (1.70)* 49 ?present yourself as you wish to be seen around acquaintances, colleagues, or peers?  0.72 (-1.90) 0.56 (-2.60) 50 ?present yourself as you wish to be seen when you are in public and feel people are watching you?  0.82 (-1.30) 0.79 (-1.10)* 51 ?present yourself as you wish to be seen when you want to impress others?  0.81 (-1.40) 1.38 (1.90)* 52 ?solve how to get to your destination when there is an unexpected situation, such as detours?  0.92 (-0.50) 1.04 (0.30)* 53 ?negotiate a challenging and unusual obstacle?^ 0.64 (-2.60) 1.23 (1.10)* 54 ?move in situations making you anxious/nervous?^ 0.86 (-0.90) 1.37 (1.60)* 55 ?know when your wheelchair is not working properly? 1.17 (1.00) 1.25 (1.20)* 56 ?know what your wheelchair can and can?t do, separate from your own abilities?  1.43 (2.90) 1.80 (3.60) 57 ?tell someone how to move your wheelchair if it gets stuck?  0.80 (-1.20) 0.92 (-0.30)* 58 ?ask someone for help? 0.59 (-1.90) 0.53 (-1.90) 59 ?tell a cab driver how to fold/unfold your wheelchair?  1.31 (1.80) 1.24 (1.20)* 60 ?tell a stranger how to help you safely get back into your wheelchair if you tip over?  1.17 (1.10) 0.95 (-0.20)* 61 ?know what to do if you fall out of your wheelchair? 1.61 (2.90) 1.58 (2.20) 62 ?advocate for changes to your wheelchair?  1.10 (0.60) 1.15 (0.70)* 63 ?advocate for changes in your home? 1.07 (0.50) 1.50 (2.20) 64 ?advocate for your needs at work or school?  0.99 (0.00) 1.22 (1.10)* 65 ?advocate for changes in your community?  1.01 (0.20) 1.42 (2.20) *items outfit statistics within acceptable range of 0.60 and 1.40; ^factorially complex items    205 Appendix W: Fit statistics of the Self-management efficacy dimension after eliminating items with misfitting outfit statistics  Self-management efficacy   Infit Outfit  Items (n=17) Mnsq (ZSTD) Mnsq (ZSTD) 3 ?move over thresholds, such as between rooms?^ 1.31 (1.50) 1.33 (1.40) 4 ?move in small spaces, such as a bathroom?^ 1.17 (1.00) 1.03 (0.20) 12 ?move through a door that opens automatically?^ 1.46 (2.10) 0.99 (0.00) 41 ?ask people to move out of your way?  1.30 (1.70) 0.96 (-0.10) 47 ?use strategies, such as humour, that will help people feel comfortable if they are unsure how to act because you use a wheelchair?  0.96 (-0.20)* 1.45 (2.10) 48 ?correct others? mistaken beliefs about people who use wheelchairs? 0.87 (-1.00)* 1.19 (1.00) 50 ?present yourself as you wish to be seen when you are in public and feel people are watching you?  0.89 (-0.70)* 0.80 (-1.10) 51 ?present yourself as you wish to be seen when you want to impress others?  0.89 (-0.80)* 1.20 (1.10) 52 ?solve how to get to your destination when there is an unexpected situation, such as detours?  1.01 (0.10)* 1.09 (0.60) 53 ?negotiate a challenging and unusual obstacle?^ 0.69 (-2.10) 1.39 (1.70) 54 ?move in situations making you anxious/nervous?^ 0.91 (-0.50)* 1.37 (1.70) 55 ?know when your wheelchair is not working properly? 1.31 (1.70) 1.40 (1.80) 57 ?tell someone how to move your wheelchair if it gets stuck?  0.87 (-0.70)* 1.13 (0.70) 59 ?tell a cab driver how to fold/unfold your wheelchair?  1.41 (2.30) 1.15 (0.80) 60 ?tell a stranger how to help you safely get back into your wheelchair if you tip over?  1.25 (1.60) 0.96 (-0.10) 62 ?advocate for changes to your wheelchair?  1.21 (1.20) 1.16 (0.80) 64 ?advocate for your needs at work or school?  1.12 (0.80)* 1.73 (3.2) *items infit statistics within acceptable range of 0.87 and 1.13; ^factorially complex items      	 ?	 ?	 ?	 ?  206 Appendix X: 8-item Self-management efficacy subscale score conversion, SEM and reliability  Raw Std. SEM Rel.  Raw Std. SEM Rel.  Raw Std. SEM Rel. 0 0.00 23.59 0.26  27 40.72 2.07 0.98  54 49.03 2.32 0.98 1 13.40 12.10 0.57  28 41.02 2.05 0.98  55 49.42 2.36 0.97 2 20.32 8.20 0.75  29 41.32 2.04 0.98  56 49.82 2.40 0.97 3 24.08 6.49 0.83  30 41.61 2.03 0.98  57 50.24 2.44 0.97 4 26.59 5.46 0.87  31 41.91 2.02 0.98  58 50.67 2.49 0.97 5 28.43 4.76 0.90  32 42.19 2.01 0.98  59 51.12 2.55 0.97 6 29.87 4.25 0.92  33 42.48 2.01 0.98  60 51.59 2.61 0.97 7 31.04 3.87 0.93  34 42.77 2.00 0.98  61 52.09 2.68 0.97 8 32.02 3.57 0.94  35 43.05 2.00 0.98  62 52.61 2.75 0.96 9 32.86 3.33 0.95  36 43.34 2.00 0.98  63 53.16 2.83 0.96 10 33.60 3.14 0.95  37 43.62 2.01 0.98  64 53.75 2.92 0.96 11 34.27 2.98 0.96  38 43.91 2.01 0.98  65 54.37 3.02 0.96 12 34.87 2.84 0.96  39 44.20 2.02 0.98  66 55.04 3.13 0.95 13 35.42 2.73 0.97  40 44.49 2.02 0.98  67 55.76 3.25 0.95 14 35.93 2.63 0.97  41 44.78 2.03 0.98  68 56.55 3.39 0.95 15 36.40 2.55 0.97  42 45.07 2.04 0.98  69 57.40 3.55 0.94 16 36.85 2.47 0.97  43 45.37 2.05 0.98  70 58.34 3.73 0.93 17 37.27 2.41 0.97  44 45.67 2.07 0.98  71 59.39 3.95 0.93 18 37.68 2.35 0.97  45 45.98 2.08 0.98  72 60.57 4.21 0.92 19 38.06 2.30 0.98  46 46.29 2.10 0.98  73 61.92 4.53 0.91 20 38.43 2.26 0.98  47 46.60 2.12 0.98  74 63.51 4.93 0.89 21 38.78 2.22 0.98  48 46.93 2.14 0.98  75 65.42 5.46 0.87 22 39.13 2.18 0.98  49 47.26 2.16 0.98  76 67.81 6.18 0.84 23 39.46 2.15 0.98  50 47.59 2.19 0.98  77 70.98 7.26 0.79 24 39.79 2.13 0.98  51 47.94 2.22 0.98  78 75.65 9.11 0.70 25 40.11 2.10 0.98  52 48.29 2.25 0.98  79 84.14 13.35 0.53 26 40.42 2.08 0.98  53 48.66 2.28 0.98  80 100.00 25.23 0.24 Raw=raw score; Std=standardized score; SEM=standard error of measurement; Rel=reliability                207 Appendix Y:   21-item WheelCon short form score conversion, SEM and reliability Raw Std. SEM Rel.  Raw Std. SEM Rel.  Raw Std. SEM Rel. 0 0.00 21.95 0.24  47 40.39 1.19 0.99  94 44.84 1.05 0.99 1 13.28 11.36 0.55  48 40.50 1.18 0.99  95 44.93 1.05 0.99 2 20.10 7.61 0.73  49 40.61 1.17 0.99  96 45.02 1.05 0.99 3 23.71 5.95 0.81  50 40.72 1.17 0.99  97 45.11 1.05 0.99 4 26.07 4.96 0.86  51 40.83 1.16 0.99  98 45.19 1.05 0.99 5 27.78 4.29 0.90  52 40.93 1.15 0.99  99 45.28 1.05 0.99 6 29.09 3.81 0.91  53 41.04 1.15 0.99  100 45.37 1.05 0.99 7 30.13 3.44 0.93  54 41.15 1.14 0.99  101 45.46 1.05 0.99 8 31.00 3.14 0.94  55 41.25 1.13 0.99  102 45.55 1.05 0.99 9 31.73 2.91 0.95  56 41.35 1.13 0.99  103 45.64 1.05 0.99 10 32.36 2.71 0.95  57 41.45 1.12 0.99  104 45.72 1.05 0.99 11 32.92 2.55 0.96  58 41.55 1.12 0.99  105 45.81 1.05 0.99 12 33.41 2.41 0.97  59 41.65 1.11 0.99  106 45.90 1.06 0.99 13 33.85 2.29 0.97  60 41.75 1.11 0.99  107 45.99 1.06 0.99 14 34.26 2.19 0.97  61 41.85 1.10 0.99  108 46.08 1.06 0.99 15 34.62 2.10 0.97  62 41.95 1.10 0.99  109 46.17 1.06 0.99 16 34.96 2.02 0.98  63 42.04 1.10 0.99  110 46.26 1.06 0.99 17 35.28 1.94 0.98  64 42.14 1.09 0.99  111 46.35 1.06 0.99 18 35.57 1.88 0.98  65 42.24 1.09 0.99  112 46.44 1.06 0.99 19 35.84 1.82 0.98  66 42.33 1.09 0.99  113 46.53 1.07 0.99 20 36.10 1.77 0.98  67 42.42 1.08 0.99  114 46.63 1.07 0.99 21 36.35 1.72 0.98  68 42.52 1.08 0.99  115 46.72 1.07 0.99 22 36.58 1.68 0.98  69 42.61 1.08 0.99  116 46.81 1.07 0.99 23 36.80 1.64 0.98  70 42.70 1.07 0.99  117 46.90 1.08 0.99 24 37.01 1.60 0.98  71 42.80 1.07 0.99  118 47.00 1.08 0.99 25 37.21 1.57 0.98  72 42.89 1.07 0.99  119 47.09 1.08 0.99 26 37.40 1.54 0.99  73 42.98 1.07 0.99  120 47.18 1.08 0.99 27 37.59 1.51 0.99  74 43.07 1.06 0.99  121 47.28 1.09 0.99 28 37.77 1.48 0.99  75 43.16 1.06 0.99  122 47.37 1.09 0.99 29 37.94 1.45 0.99  76 43.25 1.06 0.99  123 47.47 1.09 0.99 30 38.10 1.43 0.99  77 43.34 1.06 0.99  124 47.56 1.10 0.99 31 38.27 1.41 0.99  78 43.43 1.06 0.99  125 47.66 1.10 0.99 32 38.42 1.39 0.99  79 43.52 1.06 0.99  126 47.76 1.10 0.99 33 38.58 1.37 0.99  80 43.61 1.05 0.99  127 47.86 1.11 0.99 34 38.72 1.35 0.99  81 43.70 1.05 0.99  128 47.95 1.11 0.99 35 38.87 1.33 0.99  82 43.79 1.05 0.99  129 48.05 1.12 0.99 36 39.01 1.32 0.99  83 43.88 1.05 0.99  130 48.15 1.12 0.99 37 39.15 1.30 0.99  84 43.96 1.05 0.99  131 48.25 1.12 0.99 38 39.28 1.29 0.99  85 44.05 1.05 0.99  132 48.36 1.13 0.99 39 39.41 1.28 0.99  86 44.14 1.05 0.99  133 48.46 1.13 0.99 40 39.54 1.26 0.99  87 44.23 1.05 0.99  134 48.56 1.14 0.99 41 39.67 1.25 0.99  88 44.32 1.05 0.99  135 48.67 1.14 0.99 42 39.79 1.24 0.99  89 44.40 1.05 0.99  136 48.77 1.15 0.99 43 39.92 1.23 0.99  90 44.49 1.05 0.99  137 48.88 1.15 0.99 44 40.04 1.22 0.99  91 44.58 1.05 0.99  138 48.98 1.16 0.99 45 40.15 1.21 0.99  92 44.67 1.05 0.99  139 49.09 1.16 0.99 46 40.27 1.20 0.99  93 44.76 1.05 0.99  140 49.20 1.17 0.99 Raw = raw score; Std = standardized score; SEM = standard error of measurement        208 Appendix Y:   21-item WheelCon short form cont? Raw Std. SEM Rel.  Raw Std. SEM Rel. 141 49.30 1.18 0.99  187 57.01 1.92 0.98 142 49.42 1.18 0.99  188 57.31 1.97 0.98 143 49.54 1.19 0.99  189 57.63 2.02 0.98 144 49.65 1.20 0.99  190 57.96 2.08 0.97 145 49.77 1.20 0.99  191 58.32 2.14 0.97 146 49.88 1.21 0.99  192 58.70 2.21 0.97 147 50.00 1.22 0.99  193 59.10 2.28 0.97 148 50.12 1.22 0.99  194 59.54 2.36 0.97 149 50.24 1.23 0.99  195 60.00 2.46 0.96 150 50.36 1.24 0.99  196 60.51 2.56 0.96 151 50.49 1.25 0.99  197 60.06 2.68 0.96 152 50.61 1.26 0.99  198 61.66 2.81 0.95 153 50.74 1.27 0.99  199 62.33 2.96 0.95 154 50.87 1.27 0.99  200 63.07 3.14 0.94 155 51.00 1.28 0.99  201 63.92 3.35 0.93 156 51.13 1.29 0.99  202 64.88 3.60 0.92 157 51.27 1.30 0.99  203 66.01 3.91 0.91 158 51.41 1.31 0.99  204 67.35 4.29 0.90 159 51.55 1.32 0.99  205 68.99 4.78 0.87 160 51.69 1.33 0.99  206 71.09 5.46 0.84 161 51.83 1.35 0.99  207 73.91 6.46 0.79 162 51.98 1.36 0.99  208 78.10 8.15 0.70 163 52.13 1.37 0.99  209 85.77 11.93 0.52 164 52.28 1.38 0.99  210 100.00 22.45 0.24 165 52.43 1.39 0.99      166 52.59 1.41 0.99      167 52.75 1.42 0.99      168 52.91 1.44 0.99      169 53.08 1.45 0.99      170 53.25 1.47 0.99      171 53.43 1.48 0.99      172 53.61 1.50 0.99      173 53.79 1.52 0.99      174 53.97 1.54 0.99      175 54.17 1.56 0.99      176 54.36 1.58 0.98      177 54.57 1.60 0.98      178 54.77 1.62 0.98      179 54.99 1.65 0.98      180 55.21 1.67 0.98      181 55.44 1.70 0.98      182 55.68 1.73 0.98      183 55.92 1.77 0.98      184 56.18 1.80 0.98      185 56.44 1.84 0.98      186 56.72 1.88 0.98      Raw=raw score; Std=standardized score; SEM=standard error of measurement; Rel=reliability        209 Appendix Z:    Regression syntax  CHAPTER 2: Direct and mediated self-efficacy effects on participation frequency  Interaction assessment syntax: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE ZPP   /CRITERIA=PIN(.05) POUT(.1)   /NOORIGIN    /DEPENDENT LLDI_freq_std   /METHOD=BACKWARD Age_inter Sex_inter   /METHOD=ENTER WheelCon_center Age_center Sex   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3)   /SAVE COOK.  DATASET ACTIVATE DataSet2. REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE   /CRITERIA=PIN(.10) POUT(.11)   /NOORIGIN    /DEPENDENT LLDI_freq_std   /METHOD=ENTER WheelCon_center   /METHOD=ENTER Age_center Sex   /METHOD=FORWARD Sex_inter Age_inter   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3)   /SAVE COOK.  Confounding assessment syntax: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS R ANOVA COLLIN TOL CHANGE   /CRITERIA=PIN(.05) POUT(.10)   /NOORIGIN    /DEPENDENT LLDI_freq_std   /METHOD=ENTER WheelCon_center   /METHOD=ENTER Age_center FCI_center ISEL_center.   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3).   /SAVE COOK.   210 Mediation with bootstrapping syntax: DATASET ACTIVATE DataSet1. /* Written by Andrew F. Hayes */. /* http://www.afhayes.com */. /* Version 4.2 */. DEFINE INDIRECT (y = !charend('/')/x = !charend('/')/m = !charend('/')/c=!charend('/')      !default(xxxxx)/   boot =!charend('/') !default(1000)/conf = !charend('/') !default(95)/percent = !charend('/')      !default(0)/bc = !charend('/')   !default(1)/bca = !charend('/') !default(0)/normal = !charend ('/') !default(0)/contrast =      !charend ('/') !default(0)/iterate = !charend('/') !default(10000)/converge =    !charend('/') !default(.0000001)). PRESERVE. SET LENGTH = NONE. SET MXLOOPS = 10000001. SET SEED = RANDOM. SET PRINTBACK = OFF. MATRIX. get dd/variables = !y !x !m/names = nm/MISSING = OMIT. compute temp = ncol(dd). get dd2/variables = !y !x Mobility_center WCskills_center LLDI_limit_centre/MISSING = OMIT. compute nc = ncol(dd)-ncol(dd2). compute ovals = ncol(design(dd(:,1))). do if (ovals = 2).    compute omx = cmax(dd(:,1)).    compute omn = cmin(dd(:,1)).    compute dd(:,1) = (dd(:,1) = omx).    compute rcd = {omn, 0; omx, 1}. end if. compute nm = t(nm). compute outv = t(nm(1,1)). compute n = nrow(dd). compute nv = ncol(dd). compute con = make(n,1,1). compute dat2 = dd. compute dat = dd. compute bzx = make(nv-2-nc,1,0). compute bzxse = make(nv-2-nc,1,0). compute b=make((nv-1-nc),(nv-1-nc),0). compute resid = make(n,(nv-nc),0). compute info = make((2*(nv-nc-2)+1),(2*(nv-nc-2)+1),0). compute imat = make(ncol(info),4,1). compute imat(1:(nv-nc-2),1)=t({2:(nv-nc-1):1}). compute imat(1:(nv-nc-2),3)=t({2:(nv-nc-1):1}). compute imat((nv-nc-1):(ncol(info)-1),2)=t({2:(nv-nc-1):1}). compute imat((nv-nc-1):(ncol(info)-1),4)=t({2:(nv-nc-1):1}). compute imat((nv-nc-1):(ncol(info)-1),1)=make((nv-nc-2),1,(nv-nc)). compute imat((nv-nc-1):(ncol(info)-1),3)=make((nv-nc-2),1,(nv-nc)). compute imat(ncol(info),:)={(nv-nc),1,(nv-nc),1}. compute cname=     {"C1";"C2";"C3";"C4";"C5";"C6";"C7";"C8";"C9";"C10";"C11";"C12";"C13";"C14";"C15";"C16";"C17"}.   211 compute cname=     {cname;"C18";"C19";"C20";"C21";"C22";"C23";"C24";"C25";"C26";"C27";"C28";"C29";"C30";"C31"}. compute cname=     {cname;"C32";"C33";"C34";"C35";"C36";"C37";"C38";"C39";"C40";"C41";"C42";"C43";"C44";"C45"}. compute p0=-.322232431088. compute p1 = -1. compute p2 = -.342242088547. compute p3 = -.0204231210245. compute p4 = -.0000453642210148. compute q0 = .0993484626060. compute q1 = .588581570495. compute q2 = .531103462366. compute q3 = .103537752850. compute q4 = .0038560700634. compute conf = rnd(!conf). compute lowalp = 0.5*(1-(conf/100)). compute upalp = 0.5*(1+(conf/100)). compute zbca = {lowalp; upalp}. do if (!boot > 999).    compute btn = trunc(!boot/1000)*1000.    compute lpmax = n+1+btn.    else.    compute btn = 1.    compute lpmax = 1. end if. compute blowp = trunc(lowalp*btn). do if (blowp < 1).   compute blowp = 1. end if. compute bhighp = trunc((upalp*btn)+1). do if (bhighp > btn).   compute bhighp = btn. end if. compute indeff = make(n+1+btn,nv-1-nc,-9999). compute bdbp = 0. loop #d = 1 to lpmax.    do if (#d = (n+2)).     compute dat = dat2.     compute con = make(n,1,1).   end if.   do if (#d > 1 and #d < (n+2)).     do if (#d = 2).       compute con = make((n-1),1,1).       compute dat = dat2(2:n,:).     else if (#d = (n+1)).       compute dat = dat2(1:(n-1),:).     else.       compute dat = {dat2(1:(#d-2),:);dat2((#d:n),:)}.     end if.   end if.   do if (#d > (n+1)).     loop.     compute v=trunc(uniform(n,1)*n)+1.     compute dat(:,1:nv) = dat2(v,1:nv).   212     compute dat3 = {con,dat(:,2:ncol(dat))}.     compute rk = (rank(dat3)=ncol(dat3)).     compute bdbp = bdbp+(1-rk).     end loop if (rk = 1).   end if.   compute x = dat(:,2).   compute m = dat(:,3:(nv-nc)).   compute y = dat(:,1).   compute xz = dat(:,2:nv).   compute xo = {con,x}.   do if (nc > 0).     compute c = dat(:,(nv-nc+1):nv).     compute xo = {xo, c}.   end if.   loop #k = 3 to (nv-nc).      compute ytmp = dat(:,#k).      compute bzxt = inv(t(xo)*xo)*t(xo)*ytmp.      compute bzx((#k-2),1)=bzxt(2,1).      do if (#d = 1).        compute resid(:,#k-1) = ytmp-(xo*bzxt).        compute mse=csum((ytmp-(xo*bzxt))&**2)/(n-2-nc).        compute olscm=(mse*inv((t(xo)*xo))).        compute bzxse((#k-2),1)=sqrt(olscm(2,2)).      end if.   end loop.    do if (#d = 1).     do if (nc > 0).       compute cnt = dd(:,(nv-(nc-1)):nv)).       compute xo = {con,x,cnt}.     else.       compute xo = {con,x}.     end if.    do if (ovals = 2).    compute pt2 = make(nrow(y),1,(csum(y)/nrow(y))).    compute pt1 = make(nrow(y),1,0.5).    compute bt1 = make(ncol(xo),1,0).    compute LL1 = 0.    loop jjj = 1 to !iterate.     compute vt1 = mdiag(pt1&*(1-pt1)).     compute byx = bt1+inv(t(xo)*vt1*xo)*t(xo)*(y-pt1).     compute pt1 = 1/(1+exp(-(xo*byx))).     compute itprob = csum((pt1 < .00000000000001) or (pt1 > .99999999999999)).     do if (itprob = 0).     compute LL = y&*ln(pt1)+(1-y)&*ln(1-pt1).     compute LL2 = -2*csum(ll).     end if.     do if (abs(LL1-LL2) < !converge).       compute vt1 = mdiag(pt1&*(1-pt1)).       compute varb = inv(t(xo)*vt1*xo).       compute olscm = diag(varb).       break.     end if.     compute bt1 = byx.     compute LL1 = LL2.     end loop.     compute byx = byx(2,1).     compute byxse = sqrt(olscm(2,1)).   213     do if (jjj > !iterate).      compute itprob = 2.     end if.   end if.     do if (ovals <> 2).     compute byx = inv(t(xo)*xo)*t(xo)*y.     compute mse=csum((y-(xo*byx))&**2)/(n-2-nc).     compute olscm=(mse*inv((t(xo)*xo))).     compute byxse = sqrt(olscm(2,2)).     compute byx = byx(2,1).     end if.   end if.   compute xzo = {con,xz}. do if (ovals = 2). compute pt2 = make(nrow(y),1,(csum(y)/nrow(y))). compute LL3 = y&*ln(pt2)+(1-y)&*ln(1-pt2). compute LL3 = -2*csum(LL3). compute pt1 = make(nrow(y),1,0.5).   compute bt1 = make(ncol(xzo),1,0).   compute LL1 = 0.   loop jjj = 1 to !iterate.     compute vt1 = mdiag(pt1&*(1-pt1)).     compute byzx = bt1+inv(t(xzo)*vt1*xzo)*t(xzo)*(y-pt1).     compute pt1 = 1/(1+exp(-(xzo*byzx))).     compute itprob = csum((pt1 < .00000000000001) or (pt1 > .99999999999999)).     do if (itprob = 0).     compute LL = y&*ln(pt1)+(1-y)&*ln(1-pt1).     compute LL2 = -2*csum(ll).     end if.     do if (abs(LL1-LL2) < !converge).       compute vt1 = mdiag(pt1&*(1-pt1)).       compute varb = inv(t(xzo)*vt1*xzo).       compute olscm = diag(varb).       break.     end if.     compute bt1 = byzx.     compute LL1 = LL2.   end loop.   compute byzx2 = byzx(3:(nv-nc),1).   do if (nc > 0).       compute bcon = byzx((nv-nc+1):nv,1).       compute bconse = sqrt(olscm((nv-nc+1):nv,1)).     end if.     compute cprime = byzx(2,1).     compute cprimese = sqrt(olscm(2,1)).     compute byzx2se = sqrt(olscm(3:(nv-nc),1)).      do if (#d = 1).     compute pi = (exp(xzo*byzx)/(1+exp(xzo*byzx))).     compute resid(:,ncol(resid))=((y-pt1)/abs(y-pt1))&*sqrt(-2*(LL)).     end if. do if (jjj > !iterate).    compute itprob = 2. end if. end if.   do if (ovals <> 2).   compute byzx = inv(t(xzo)*xzo)*t(xzo)*y.   compute byzx2 = byzx(3:(nv-nc),1).   214   do if (#d = 1).      compute mse=csum((y-(xzo*byzx))&**2)/(n-nv).     compute resid(:,ncol(resid))=y-(xzo*byzx).     compute covmat=mse*inv(t(xzo)*xzo).     compute olscm=diag(covmat).     compute sse = mse*(n-nv).     compute sst = csum((y-(csum(y)/n))&**2).     compute r2 = 1-(sse/sst).     compute ar2 = 1-(mse/(sst/(n-1))).     compute fr = ((n-nv)*r2)/((1-r2)*ncol(xz)).     compute pfr = 1-fcdf(fr,ncol(xz),(n-nv)).     do if (nc > 0).       compute bcon = byzx((nv-nc+1):nv,1).       compute bconse = sqrt(olscm((nv-nc+1):nv,1)).     end if.     compute byzx2se = sqrt(olscm(3:(nv-nc),1)).     compute cprime = byzx(2,1).     compute cprimese = sqrt(olscm(2,1)).   end if.   end if.   compute indeff2 = (bzx&*byzx2).   compute zs = (bzx&/bzxse)&*(byzx2&/byzx2se).   compute temp = t({csum(indeff2); indeff2}).   compute indeff(#d,:) = temp.   do if (#d = 1).     compute vs = nm(1:(nv-nc),1).     print/title = "*****************************************************************".     print/title = "Preacher and Hayes (2008) SPSS Macro for Multiple Mediation".     print/title = "Written by Andrew F. Hayes, The Ohio State University".     print/title = "http://www.comm.ohio-state.edu/ahayes/".     print/title = "For details, see Preacher, K. J., & Hayes, A. F. (2008). Asymptotic".     print/title = "and resampling strategies for assessing and comparing indirect effects".     print/title = "in multiple mediator models. Behavior Research Methods, 40, 879-891.".     print/title = "*****************************************************************".     print vs/title = "Dependent, Independent, and Proposed Mediator Variables:"/rlabels = "DV ="      "IV = " "MEDS = "/format a8.     do if (nc > 0).       compute vs = nm((nv-nc+1):nv,1).       print vs/title = "Statistical Controls:"/rlabels = "CONTROL="/format a8.     end if.     print n/title = "Sample size"/format F10.0.     do if (ovals = 2).     compute nmsd = {outv, "Analysis"}.     print rcd/title = "Coding of binary DV for analysis:"/cnames = nmsd/format = F9.2.     end if.     compute nms = nm(3:(nv-nc),1).     compute te = bzx&/bzxse.    215     compute df = n-2-nc.     compute p = 2*(1-tcdf(abs(te), df)).     compute bzxmat = {bzx, bzxse,te,p}.     compute b(2:(nv-1-nc),1)=bzx.      compute se2 = bzxse&*bzxse.     print bzxmat/title = "IV to Mediators (a paths)"/rnames = nms/clabels "Coeff" "se" "t"      "p"/format f9.4.     compute te = byzx2&/byzx2se.     compute df = n-nv.     do if (ovals <> 2).     compute p = 2*(1-tcdf(abs(te), df)).     compute byzx2mat={byzx2, byzx2se, te, p}.     print byzx2mat/title = "Direct Effects of Mediators on DV (b paths)"/rnames = nms/clabels      "Coeff" "se" "t" "p"/format f9.4.     end if.     do if (ovals = 2).       compute wald = te&*te.       compute p = 2*(1-cdfnorm(abs(te))).       compute byzx2mat={byzx2, byzx2se, te, p, Wald}.       print byzx2mat/title = "Direct Effects of Mediators on DV (b paths)"/rnames = nms/clabels      "Coeff" "se" "Z" "p" "Wald"/format f9.4.     end if.     compute te = byx&/byxse.     compute df = n-2-nc.     compute xnm = nm(2,1).     do if (ovals <> 2).     compute p = 2*(1-tcdf(abs(te), df)).     compute byxmat = {byx, byxse, te, p}.     print byxmat/title = "Total Effect of IV on DV (c path)"/rnames = xnm/clabels "Coeff" "se"      "t" "p"/format f9.4.     end if.     do if (ovals = 2).     compute wald = te&*te.     compute p = 2*(1-cdfnorm(abs(te))).     compute byxmat = {byx, byxse, te, p, Wald}.     print byxmat/title = "Total Effect of IV on DV (c path)"/rnames = xnm/clabels "Coeff" "se"      "Z" "p" "Wald"/format f9.4.     end if.     compute te = cprime&/cprimese.     compute df = n-nv.     do if (ovals <> 2).     compute p = 2*(1-tcdf(abs(te), df)).     compute cprimmat = {cprime, cprimese, te, p}.     print cprimmat/title = "Direct Effect of IV on DV (c' path)"/rnames = xnm/clabels "Coeff"      "se" "t" "p"/format f9.4.     end if.     do if (ovals = 2).     compute wald = te&*te.     compute p = 2*(1-cdfnorm(abs(te))).     compute cprimmat = {cprime, cprimese, te, p, Wald}.     print cprimmat/title = "Direct Effect of IV on DV (c' path)"/rnames = xnm/clabels "Coeff"    216     "se" "Z" "p" "Wald"/format f9.4.     end if.     do if (nc > 0).       compute df = n-nv.       compute nms = nm((nv-nc+1):nv,1).       compute te = bcon&/bconse.       do if (ovals <> 2).       compute p = 2*(1-tcdf(abs(te), df)).       compute bconmat = {bcon, bconse,te,p}.       print bconmat/title = "Partial Effect of Control Variables on DV"/rnames = nms/clabels      "Coeff" "se" "t" "p"/format f9.4.       end if.       do if (ovals = 2).       compute wald = te&*te.       compute p = 2*(1-cdfnorm(abs(te))).       compute bconmat = {bcon, bconse,te,p, Wald}.       print bconmat/title = "Partial Effect of Control Variables on DV"/rnames = nms/clabels      "Coeff" "se" "Z" "p" "Wald"/format f9.4.       end if.     end if.     do if (ovals <> 2).     compute dvms = {r2, ar2, fr, ncol(xz), (n-nv), pfr}.     print dvms/title = "Model Summary for DV Model"/clabels "R-sq" "Adj R-sq" "F" "df1" "df2"      "p"/format F9.4.     end if.    do if (ovals = 2).    compute LLdiff = LL3-LL2.    compute mcF = LLdiff/LL3.    compute cox = 1-exp(-LLdiff/n).    compute nagel = cox/(1-exp(-(LL3)/n)).    compute pf = {LL2, LLdiff, mcF, cox, nagel, n}.    print pf/title = "Logistic Regression Summary for DV Model"/clabels = "-2LL" "Model LL"      "McFadden" "CoxSnell" "Nagelkrk" "n"/format F10.4.    end if.     do if (!normal <> 0 and nc = 0 and ovals <> 2).       compute bmat = make((nv-nc),(nv-nc),0).       compute bmat(2:(nv-nc-1),1) = bzx.       compute bmat((nv-nc),2:(nv-nc-1))=t(byzx2).       compute bmat((nv-nc),1) = cprime.       compute imbinv = inv(ident(ncol(bmat))-bmat).       compute imbtinv=inv(ident(ncol(bmat))-t(bmat)).       compute resid(:,1)=x-(csum(x)/(n)).       compute psi = sscp(resid)/(n).       compute invpsi = inv(psi).       compute ibpsiib = imbinv*psi*imbtinv.       loop ic = 1 to ncol(info).       loop ic2 = 1 to ncol(info).       compute info(ic,ic2)=(n-1)*((imbinv(imat(ic2,4),imat(ic,1))*imbinv(imat(ic,2),imat(ic2,3)))+     (ibpsiib(imat(ic2,4),imat(ic,2))*invpsi(imat(ic,1),imat(ic2,3)))).       end loop.       end loop.       compute varcov = inv(info).    217       compute varcov = varcov(1:(2*(nv-nc-2)),1:(2*(nv-nc-2))).       compute ses = diag(varcov).       compute avar = ses(1:nrow(bzxse),1).       compute bvar = ses((nrow(bzxse)+1):nrow(ses),1).       do if ((nv-nc-2) > 1 and (!contrast = 1)).         compute prws=make(((nv-nc-2)*(nv-nc-3)/2),1,0).         compute prwse=prws.         compute kk=1.         loop ic = 1 to (nv-nc-3).         loop ic2 = (ic+1) to (nv-nc-2).         compute vf2 = ((byzx2(ic,1)**2)*varcov(ic,ic))-(2*byzx2(ic,1)*byzx2(ic2,1)*(varcov(ic,ic2)))     .         compute vf2=vf2+((byzx2(ic2,1)**2)*varcov(ic2,ic2))+((bzx(ic,1)**2)*(bvar(ic,1))).         compute vf2=vf2-(2*bzx(ic,1)*bzx(ic2,1)*covmat((2+ic),(2+ic2)))+((bzx(ic2,1)**2)*(bvar(ic2,     1))).         compute cnt = indeff2(ic,1)-indeff2(ic2,1).         compute prws(kk,1)=cnt.         compute prwse(kk,1)=sqrt(vf2).         compute kk=kk+1.         end loop.         end loop.         compute cnam2 = cname(1:(kk-1),1).       end if.       compute dermat = {byzx2;bzx}.       compute totse = sqrt(t(dermat)*varcov*dermat).       compute specse = sqrt((byzx2&*byzx2)&*(avar)+(bzx&*bzx)&*(bvar)).       compute specse = {totse; specse}.              compute specz = {csum(indeff2);indeff2}&/specse.       compute ind22 = {csum(indeff2);indeff2}.       compute nms = {"TOTAL";nm(3:(nv-nc),1)}.       do if ((nv-nc-2) > 1 and (!contrast = 1)).         compute ind22 = {ind22;prws}.         compute specse = {specse;prwse}.         compute specz = {specz;(prws&/prwse)}.         compute nms = {nms;cnam2}.       end if.       compute pspec= 2*(1-cdfnorm(abs(specz))).       compute spec = {ind22, specse, specz, pspec}.       print/title = "******************************************************************".       print/title = "           NORMAL THEORY TESTS FOR INDIRECT EFFECTS".       print spec/title = "Indirect Effects of IV on DV through Proposed Mediators (ab "+     "paths)"/rnames = nms/clabels "Effect" "se" "Z" "p"/format = f9.4.     end if.   end if. end loop. do if (btn > 1).   compute nms = {"TOTAL"; nm(3:(nv-nc),1)}.   do if ((nv-nc-2) > 1 and (!contrast = 1)).     compute crst = make((n+1+btn),((nv-nc-2)*(nv-nc-3)/2),0).     compute kk=1.   218     loop ic = 2 to (nv-nc-2).       loop ic2 = (ic+1) to (nv-nc-1).         compute crst(:,kk)=indeff(:,ic)-indeff(:,ic2).         compute kk=kk+1.       end loop.     end loop.     compute indeff = {indeff,crst}.     compute cnam2 = cname(1:(kk-1),1).     compute nms = {nms;cnam2}.   end if. compute lvout = indeff(2:(n+1),:). compute tdotm = csum(lvout)/n. compute tm = (make(n,ncol(lvout),1))*mdiag(tdotm). compute topa = csum((((n-1)/n)*(tm-lvout))&**3). compute bota = 6*sqrt((csum((((n-1)/n)*(tm-lvout))&**2)&**3)). compute ahat = topa&/bota. compute indsam = t(indeff(1,:)). compute boot = indeff((n+2):nrow(indeff),:). compute mnboot = t(csum(boot)/btn). compute se = (sqrt(((btn*cssq(boot))-(csum(boot)&**2))/((btn-1)*btn))). save boot/outfile = indirect.sav/names = nms. loop #e = 1 to ncol(indeff).   compute boottmp = boot(:,#e).   compute boottmp(GRADE(boot(:,#e))) = boot(:,#e).   compute boot(:,#e) = boottmp. end loop. compute xp = make((nrow(mnboot)+2),1,0). loop i = 1 to (nrow(mnboot)+2).   do if (i <= nrow(mnboot)).     compute pv = (boot(:,i) < indsam(i,1)).     compute pv = csum(pv)/btn.   else.     compute pv = zbca((i-nrow(mnboot)),1).   end if.   compute p = pv.   do if (pv > .5).     compute p = 1-pv.   end if.   compute y5=sqrt(-2*ln(p)).   compute xp(i,1)=y5+((((y5*p4+p3)*y5+p2)*y5+p1)*y5+p0)/((((y5*q4+q3)*y5+q2)*y5+q1)*y5+q0).   do if (pv <= .5).     compute xp(i,1) = -xp(i,1).   end if. end loop. compute bbb = nrow(mnboot). compute zz = xp(1:bbb,1). compute zlo = zz + ((zz+xp((bbb+1),1))&/(1-t(ahat)&*(zz+xp((bbb+1),1)))). compute zup = zz + ((zz+xp((bbb+2),1))&/(1-t(ahat)&*(zz+xp((bbb+2),1)))). compute ahat = 0. compute zlobc = zz + ((zz+xp((bbb+1),1))&/(1-t(ahat)&*(zz+xp((bbb+1),1)))). compute zupbc = zz + ((zz+xp((bbb+2),1))&/(1-t(ahat)&*(zz+xp((bbb+2),1)))). compute zlo = cdfnorm(zlo). compute zup = cdfnorm(zup). compute zlobc = cdfnorm(zlobc). compute zupbc = cdfnorm(zupbc). compute blow = trunc(zlo*(btn+1)).   219 compute bhigh = trunc(zup*(btn+1))+1. compute blowbc = trunc(zlobc*(btn+1)). compute bhighbc = trunc(zupbc*(btn+1))+1. compute lowbca = make(nrow(blow),1,0). compute upbca = lowbca. loop i = 1 to nrow(blow).   do if (blow(i,1) < 1).     compute blow(i,1) = 1.   end if.   compute lowbca(i,1)=boot(blow(i,1),i).   do if (bhigh(i,1) > btn).     compute bhigh(i,1) = btn.   end if.   compute upbca(i,1)=boot(bhigh(i,1),i). end loop. compute lowbc = make(nrow(blow),1,0). compute upbc = lowbca. loop i = 1 to nrow(blowbc).   do if (blowbc(i,1) < 1).     compute blowbc(i,1) = 1.   end if.   compute lowbc(i,1)=boot(blowbc(i,1),i).   do if (bhighbc(i,1) > btn).     compute bhighbc(i,1) = btn.   end if.   compute upbc(i,1)=boot(bhighbc(i,1),i). end loop. print/title = "*****************************************************************". print/title = "           BOOTSTRAP RESULTS FOR INDIRECT EFFECTS". compute res = {indsam, mnboot,(mnboot-indsam), t(se)}. print res/title = "Indirect Effects of IV on DV through Proposed Mediators (ab paths)"/rnames =      nms/clabels "Data" "Boot" "Bias" "SE"/format f9.4. compute lowperc = boot(blowp,:). compute upperc = boot(bhighp,:). compute ci = {lowbca, upbca}. do if (!bca <> 0).   print ci/title = "Bias Corrected and Accelerated Confidence Intervals"/rnames = nms/clabels      "Lower" "Upper"/format F9.4. end if. do if (!bc <> 0).   compute ci = {lowbc, upbc}.   print ci/title = "Bias Corrected Confidence Intervals"/rnames = nms/clabels "Lower"      "Upper"/format F9.4. end if. do if (!percent <> 0).   compute ci = {t(lowperc), t(upperc)}.   print ci/title = "Percentile Confidence Intervals"/rnames = nms/clabels "Lower" "Upper"/format      F9.4. end if. print/title = "*****************************************************************".   220 print conf/title = "Level of Confidence for Confidence Intervals:". print btn/title = "Number of Bootstrap Resamples:". end if. do if ((nv-nc-2) > 1 and (!contrast = 1) and ((!normal = 1 and nc = 0) OR btn > 999))). print/title = "*****************************************************************". print/title = "  INDIRECT EFFECT CONTRAST DEFINITIONS: Ind_Eff1 MINUS Ind_Eff2". compute kk=1. compute prwsv=make(((nv-nc-2)*(nv-nc-3)/2),2,0).  loop ic = 1 to (nv-nc-3).         loop ic2 = (ic+1) to (nv-nc-2).           compute prwsv(kk,1)=nm(ic+2,1).           compute prwsv(kk,2)=nm(ic2+2,1).           compute kk=kk+1.        end loop. end loop. compute prwsv = {cnam2, prwsv}. print prwsv/title = " "/clabels = "Contrast" "IndEff_1" "IndEff_2"/format A9. end if. Print/title = "********************************* NOTES **********************************". do if (btn = 1). Print/title = "Bootstrap confidence intervals are preferred to normal theory tests for inference "+     "about indirect effects". Print/title = "See Hayes, A.F.(2009). Beyond Baron And Kenny: Statistical mediation analysis in "+     "the new millennium". Print/title = "Communication Monographs, 76, 408-420". end if. do if (bdbp > 0). print/title = "*****************************************************************". print/title = "WARNING: SOME BOOTSTRAP MATRICES WERE SINGULAR". print/title = "SINGULAR MATRICES WERE REPLACED DURING RESAMPLING". print bdbp/title = "Number of singular bootstrap samples replaced:". end if.    do if (ovals = 2).    print/title = "*****************************************************************".    print/title = "NORMAL THEORY TESTS NOT AVAILABLE FOR MODELS WITH DICHOTOMOUS OUTCOMES".    do if (!boot = 0).    print/title = "To obtain indirect effects, request bootstrapping".    end if.    end if.    do if (nc > 0 and !normal = 1).    print/title = "NORMAL THEORY TESTS NOT AVAILABLE IN MODELS WITH COVARIATES".    do if (!boot = 0).    print/title = "To obtain indirect effects, request bootstrapping".    end if.    end if.   221 END MATRIX. RESTORE. !ENDDEFINE. INDIRECT y = LLDI_freq_std/x = WheelCon_center/m = Mobility_center WCskills_center      LLDI_limit_centre Age_center FCI_center ISEL_center/boot = 1000/conf = 95/normal = 0/contrast =      1/percent = 0/bc = 1/bca = 0.                                            222 CHAPTER 3: Direct and mediated self-efficacy effects on life-space mobility  Interaction assessment syntax: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE ZPP   /CRITERIA=PIN(.05) POUT(.1)   /NOORIGIN    /DEPENDENT LSA_composite   /METHOD=BACKWARD Age_inter Prov_inter   /METHOD=ENTER WheelCon_center Age_center Province   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3)   /SAVE COOK.  REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE   /CRITERIA=PIN(.10) POUT(.11)   /NOORIGIN    /DEPENDENT LSA_composite   /METHOD=ENTER WheelCon_center   /METHOD=ENTER Age_center Province   /METHOD=FORWARD Age_inter Prov_inter   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3)   /SAVE COOK.  Confounding assessment syntax: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA CHANGE   /CRITERIA=PIN(.05) POUT(.10)   /NOORIGIN    /DEPENDENT LSA_composite   /METHOD=ENTER WheelCon_center   /METHOD=ENTER Province Sex Assistance_with_WC FCI_center Employed_volunteer_student Training_with_WC   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3).   /SAVE COOK.     223 Subset of confounders with equal control syntax: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE ZPP   /CRITERIA=PIN(.05) POUT(.1)   /NOORIGIN    /DEPENDENT LSA_composite   /METHOD=BACKWARD Province Sex Assistance_with_WC Employed_volunteer_student FCI_center  Training_with_WC   /METHOD=ENTER WheelCon_center   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3)   /SAVE COOK.  Mediation with bootstrapping syntax: DATASET ACTIVATE DataSet1. /* Written by Andrew F. Hayes */. /* http://www.afhayes.com */. /* Version 4.2 */. DEFINE INDIRECT (y = !charend('/')/x = !charend('/')/m = !charend('/')/c=!charend('/')      !default(xxxxx)/   boot =!charend('/') !default(1000)/conf = !charend('/') !default(95)/percent = !charend('/')      !default(0)/bc = !charend('/')   !default(1)/bca = !charend('/') !default(0)/normal = !charend ('/') !default(0)/contrast =      !charend ('/') !default(0)/iterate = !charend('/') !default(10000)/converge =    !charend('/') !default(.0000001)). PRESERVE. SET LENGTH = NONE. SET MXLOOPS = 10000001. SET SEED = RANDOM. SET PRINTBACK = OFF. MATRIX. get dd/variables = !y !x !m/names = nm/MISSING = OMIT. compute temp = ncol(dd). get dd2/variables = !y !x WCskills_center/MISSING = OMIT. compute nc = ncol(dd)-ncol(dd2). compute ovals = ncol(design(dd(:,1))). do if (ovals = 2).    compute omx = cmax(dd(:,1)).    compute omn = cmin(dd(:,1)).    compute dd(:,1) = (dd(:,1) = omx).    compute rcd = {omn, 0; omx, 1}. end if. compute nm = t(nm). compute outv = t(nm(1,1)). compute n = nrow(dd). compute nv = ncol(dd). compute con = make(n,1,1). compute dat2 = dd.   224 compute dat = dd. compute bzx = make(nv-2-nc,1,0). compute bzxse = make(nv-2-nc,1,0). compute b=make((nv-1-nc),(nv-1-nc),0). compute resid = make(n,(nv-nc),0). compute info = make((2*(nv-nc-2)+1),(2*(nv-nc-2)+1),0). compute imat = make(ncol(info),4,1). compute imat(1:(nv-nc-2),1)=t({2:(nv-nc-1):1}). compute imat(1:(nv-nc-2),3)=t({2:(nv-nc-1):1}). compute imat((nv-nc-1):(ncol(info)-1),2)=t({2:(nv-nc-1):1}). compute imat((nv-nc-1):(ncol(info)-1),4)=t({2:(nv-nc-1):1}). compute imat((nv-nc-1):(ncol(info)-1),1)=make((nv-nc-2),1,(nv-nc)). compute imat((nv-nc-1):(ncol(info)-1),3)=make((nv-nc-2),1,(nv-nc)). compute imat(ncol(info),:)={(nv-nc),1,(nv-nc),1}. compute cname=     {"C1";"C2";"C3";"C4";"C5";"C6";"C7";"C8";"C9";"C10";"C11";"C12";"C13";"C14";"C15";"C16";"C17"}. compute cname=     {cname;"C18";"C19";"C20";"C21";"C22";"C23";"C24";"C25";"C26";"C27";"C28";"C29";"C30";"C31"}. compute cname=     {cname;"C32";"C33";"C34";"C35";"C36";"C37";"C38";"C39";"C40";"C41";"C42";"C43";"C44";"C45"}. compute p0=-.322232431088. compute p1 = -1. compute p2 = -.342242088547. compute p3 = -.0204231210245. compute p4 = -.0000453642210148. compute q0 = .0993484626060. compute q1 = .588581570495. compute q2 = .531103462366. compute q3 = .103537752850. compute q4 = .0038560700634. compute conf = rnd(!conf). compute lowalp = 0.5*(1-(conf/100)). compute upalp = 0.5*(1+(conf/100)). compute zbca = {lowalp; upalp}. do if (!boot > 999).    compute btn = trunc(!boot/1000)*1000.    compute lpmax = n+1+btn.    else.    compute btn = 1.    compute lpmax = 1. end if. compute blowp = trunc(lowalp*btn). do if (blowp < 1).   compute blowp = 1. end if. compute bhighp = trunc((upalp*btn)+1). do if (bhighp > btn).   compute bhighp = btn. end if. compute indeff = make(n+1+btn,nv-1-nc,-9999). compute bdbp = 0. loop #d = 1 to lpmax.   225    do if (#d = (n+2)).     compute dat = dat2.     compute con = make(n,1,1).   end if.   do if (#d > 1 and #d < (n+2)).     do if (#d = 2).       compute con = make((n-1),1,1).       compute dat = dat2(2:n,:).     else if (#d = (n+1)).       compute dat = dat2(1:(n-1),:).     else.       compute dat = {dat2(1:(#d-2),:);dat2((#d:n),:)}.     end if.   end if.   do if (#d > (n+1)).     loop.     compute v=trunc(uniform(n,1)*n)+1.     compute dat(:,1:nv) = dat2(v,1:nv).     compute dat3 = {con,dat(:,2:ncol(dat))}.     compute rk = (rank(dat3)=ncol(dat3)).     compute bdbp = bdbp+(1-rk).     end loop if (rk = 1).   end if.   compute x = dat(:,2).   compute m = dat(:,3:(nv-nc)).   compute y = dat(:,1).   compute xz = dat(:,2:nv).   compute xo = {con,x}.   do if (nc > 0).     compute c = dat(:,(nv-nc+1):nv).     compute xo = {xo, c}.   end if.   loop #k = 3 to (nv-nc).      compute ytmp = dat(:,#k).      compute bzxt = inv(t(xo)*xo)*t(xo)*ytmp.      compute bzx((#k-2),1)=bzxt(2,1).      do if (#d = 1).        compute resid(:,#k-1) = ytmp-(xo*bzxt).        compute mse=csum((ytmp-(xo*bzxt))&**2)/(n-2-nc).        compute olscm=(mse*inv((t(xo)*xo))).        compute bzxse((#k-2),1)=sqrt(olscm(2,2)).      end if.   end loop.    do if (#d = 1).     do if (nc > 0).       compute cnt = dd(:,(nv-(nc-1)):nv)).       compute xo = {con,x,cnt}.     else.       compute xo = {con,x}.     end if.    do if (ovals = 2).    compute pt2 = make(nrow(y),1,(csum(y)/nrow(y))).    compute pt1 = make(nrow(y),1,0.5).    compute bt1 = make(ncol(xo),1,0).    compute LL1 = 0.    loop jjj = 1 to !iterate.     compute vt1 = mdiag(pt1&*(1-pt1)).     compute byx = bt1+inv(t(xo)*vt1*xo)*t(xo)*(y-pt1).   226     compute pt1 = 1/(1+exp(-(xo*byx))).     compute itprob = csum((pt1 < .00000000000001) or (pt1 > .99999999999999)).     do if (itprob = 0).     compute LL = y&*ln(pt1)+(1-y)&*ln(1-pt1).     compute LL2 = -2*csum(ll).     end if.     do if (abs(LL1-LL2) < !converge).       compute vt1 = mdiag(pt1&*(1-pt1)).       compute varb = inv(t(xo)*vt1*xo).       compute olscm = diag(varb).       break.     end if.     compute bt1 = byx.     compute LL1 = LL2.     end loop.     compute byx = byx(2,1).     compute byxse = sqrt(olscm(2,1)).     do if (jjj > !iterate).      compute itprob = 2.     end if.   end if.     do if (ovals <> 2).     compute byx = inv(t(xo)*xo)*t(xo)*y.     compute mse=csum((y-(xo*byx))&**2)/(n-2-nc).     compute olscm=(mse*inv((t(xo)*xo))).     compute byxse = sqrt(olscm(2,2)).     compute byx = byx(2,1).     end if.   end if.   compute xzo = {con,xz}. do if (ovals = 2). compute pt2 = make(nrow(y),1,(csum(y)/nrow(y))). compute LL3 = y&*ln(pt2)+(1-y)&*ln(1-pt2). compute LL3 = -2*csum(LL3). compute pt1 = make(nrow(y),1,0.5).   compute bt1 = make(ncol(xzo),1,0).   compute LL1 = 0.   loop jjj = 1 to !iterate.     compute vt1 = mdiag(pt1&*(1-pt1)).     compute byzx = bt1+inv(t(xzo)*vt1*xzo)*t(xzo)*(y-pt1).     compute pt1 = 1/(1+exp(-(xzo*byzx))).     compute itprob = csum((pt1 < .00000000000001) or (pt1 > .99999999999999)).     do if (itprob = 0).     compute LL = y&*ln(pt1)+(1-y)&*ln(1-pt1).     compute LL2 = -2*csum(ll).     end if.     do if (abs(LL1-LL2) < !converge).       compute vt1 = mdiag(pt1&*(1-pt1)).       compute varb = inv(t(xzo)*vt1*xzo).       compute olscm = diag(varb).       break.     end if.     compute bt1 = byzx.     compute LL1 = LL2.   end loop.   compute byzx2 = byzx(3:(nv-nc),1).   227   do if (nc > 0).       compute bcon = byzx((nv-nc+1):nv,1).       compute bconse = sqrt(olscm((nv-nc+1):nv,1)).     end if.     compute cprime = byzx(2,1).     compute cprimese = sqrt(olscm(2,1)).     compute byzx2se = sqrt(olscm(3:(nv-nc),1)).      do if (#d = 1).     compute pi = (exp(xzo*byzx)/(1+exp(xzo*byzx))).     compute resid(:,ncol(resid))=((y-pt1)/abs(y-pt1))&*sqrt(-2*(LL)).     end if. do if (jjj > !iterate).    compute itprob = 2. end if. end if.   do if (ovals <> 2).   compute byzx = inv(t(xzo)*xzo)*t(xzo)*y.   compute byzx2 = byzx(3:(nv-nc),1).   do if (#d = 1).      compute mse=csum((y-(xzo*byzx))&**2)/(n-nv).     compute resid(:,ncol(resid))=y-(xzo*byzx).     compute covmat=mse*inv(t(xzo)*xzo).     compute olscm=diag(covmat).     compute sse = mse*(n-nv).     compute sst = csum((y-(csum(y)/n))&**2).     compute r2 = 1-(sse/sst).     compute ar2 = 1-(mse/(sst/(n-1))).     compute fr = ((n-nv)*r2)/((1-r2)*ncol(xz)).     compute pfr = 1-fcdf(fr,ncol(xz),(n-nv)).     do if (nc > 0).       compute bcon = byzx((nv-nc+1):nv,1).       compute bconse = sqrt(olscm((nv-nc+1):nv,1)).     end if.     compute byzx2se = sqrt(olscm(3:(nv-nc),1)).     compute cprime = byzx(2,1).     compute cprimese = sqrt(olscm(2,1)).   end if.   end if.   compute indeff2 = (bzx&*byzx2).   compute zs = (bzx&/bzxse)&*(byzx2&/byzx2se).   compute temp = t({csum(indeff2); indeff2}).   compute indeff(#d,:) = temp.   do if (#d = 1).     compute vs = nm(1:(nv-nc),1).     print/title = "*****************************************************************".     print/title = "Preacher and Hayes (2008) SPSS Macro for Multiple Mediation".     print/title = "Written by Andrew F. Hayes, The Ohio State University".     print/title = "http://www.comm.ohio-state.edu/ahayes/".     print/title = "For details, see Preacher, K. J., & Hayes, A. F. (2008). Asymptotic".     print/title = "and resampling strategies for assessing and comparing indirect effects".     print/title = "in multiple mediator models. Behavior Research Methods, 40, 879-891.".   228     print/title = "*****************************************************************".     print vs/title = "Dependent, Independent, and Proposed Mediator Variables:"/rlabels = "DV ="      "IV = " "MEDS = "/format a8.     do if (nc > 0).       compute vs = nm((nv-nc+1):nv,1).       print vs/title = "Statistical Controls:"/rlabels = "CONTROL="/format a8.     end if.     print n/title = "Sample size"/format F10.0.     do if (ovals = 2).     compute nmsd = {outv, "Analysis"}.     print rcd/title = "Coding of binary DV for analysis:"/cnames = nmsd/format = F9.2.     end if.     compute nms = nm(3:(nv-nc),1).     compute te = bzx&/bzxse.      compute df = n-2-nc.     compute p = 2*(1-tcdf(abs(te), df)).     compute bzxmat = {bzx, bzxse,te,p}.     compute b(2:(nv-1-nc),1)=bzx.      compute se2 = bzxse&*bzxse.     print bzxmat/title = "IV to Mediators (a paths)"/rnames = nms/clabels "Coeff" "se" "t"      "p"/format f9.4.     compute te = byzx2&/byzx2se.     compute df = n-nv.     do if (ovals <> 2).     compute p = 2*(1-tcdf(abs(te), df)).     compute byzx2mat={byzx2, byzx2se, te, p}.     print byzx2mat/title = "Direct Effects of Mediators on DV (b paths)"/rnames = nms/clabels      "Coeff" "se" "t" "p"/format f9.4.     end if.     do if (ovals = 2).       compute wald = te&*te.       compute p = 2*(1-cdfnorm(abs(te))).       compute byzx2mat={byzx2, byzx2se, te, p, Wald}.       print byzx2mat/title = "Direct Effects of Mediators on DV (b paths)"/rnames = nms/clabels      "Coeff" "se" "Z" "p" "Wald"/format f9.4.     end if.     compute te = byx&/byxse.     compute df = n-2-nc.     compute xnm = nm(2,1).     do if (ovals <> 2).     compute p = 2*(1-tcdf(abs(te), df)).     compute byxmat = {byx, byxse, te, p}.     print byxmat/title = "Total Effect of IV on DV (c path)"/rnames = xnm/clabels "Coeff" "se"      "t" "p"/format f9.4.     end if.     do if (ovals = 2).     compute wald = te&*te.     compute p = 2*(1-cdfnorm(abs(te))).     compute byxmat = {byx, byxse, te, p, Wald}.   229     print byxmat/title = "Total Effect of IV on DV (c path)"/rnames = xnm/clabels "Coeff" "se"      "Z" "p" "Wald"/format f9.4.     end if.     compute te = cprime&/cprimese.     compute df = n-nv.     do if (ovals <> 2).     compute p = 2*(1-tcdf(abs(te), df)).     compute cprimmat = {cprime, cprimese, te, p}.     print cprimmat/title = "Direct Effect of IV on DV (c' path)"/rnames = xnm/clabels "Coeff"      "se" "t" "p"/format f9.4.     end if.     do if (ovals = 2).     compute wald = te&*te.     compute p = 2*(1-cdfnorm(abs(te))).     compute cprimmat = {cprime, cprimese, te, p, Wald}.     print cprimmat/title = "Direct Effect of IV on DV (c' path)"/rnames = xnm/clabels "Coeff"      "se" "Z" "p" "Wald"/format f9.4.     end if.     do if (nc > 0).       compute df = n-nv.       compute nms = nm((nv-nc+1):nv,1).       compute te = bcon&/bconse.       do if (ovals <> 2).       compute p = 2*(1-tcdf(abs(te), df)).       compute bconmat = {bcon, bconse,te,p}.       print bconmat/title = "Partial Effect of Control Variables on DV"/rnames = nms/clabels      "Coeff" "se" "t" "p"/format f9.4.       end if.       do if (ovals = 2).       compute wald = te&*te.       compute p = 2*(1-cdfnorm(abs(te))).       compute bconmat = {bcon, bconse,te,p, Wald}.       print bconmat/title = "Partial Effect of Control Variables on DV"/rnames = nms/clabels      "Coeff" "se" "Z" "p" "Wald"/format f9.4.       end if.     end if.     do if (ovals <> 2).     compute dvms = {r2, ar2, fr, ncol(xz), (n-nv), pfr}.     print dvms/title = "Model Summary for DV Model"/clabels "R-sq" "Adj R-sq" "F" "df1" "df2"      "p"/format F9.4.     end if.    do if (ovals = 2).    compute LLdiff = LL3-LL2.    compute mcF = LLdiff/LL3.    compute cox = 1-exp(-LLdiff/n).    compute nagel = cox/(1-exp(-(LL3)/n)).    compute pf = {LL2, LLdiff, mcF, cox, nagel, n}.    print pf/title = "Logistic Regression Summary for DV Model"/clabels = "-2LL" "Model LL"      "McFadden" "CoxSnell" "Nagelkrk" "n"/format F10.4.    end if.     do if (!normal <> 0 and nc = 0 and ovals <> 2).   230       compute bmat = make((nv-nc),(nv-nc),0).       compute bmat(2:(nv-nc-1),1) = bzx.       compute bmat((nv-nc),2:(nv-nc-1))=t(byzx2).       compute bmat((nv-nc),1) = cprime.       compute imbinv = inv(ident(ncol(bmat))-bmat).       compute imbtinv=inv(ident(ncol(bmat))-t(bmat)).       compute resid(:,1)=x-(csum(x)/(n)).       compute psi = sscp(resid)/(n).       compute invpsi = inv(psi).       compute ibpsiib = imbinv*psi*imbtinv.       loop ic = 1 to ncol(info).       loop ic2 = 1 to ncol(info).       compute info(ic,ic2)=(n-1)*((imbinv(imat(ic2,4),imat(ic,1))*imbinv(imat(ic,2),imat(ic2,3)))+     (ibpsiib(imat(ic2,4),imat(ic,2))*invpsi(imat(ic,1),imat(ic2,3)))).       end loop.       end loop.       compute varcov = inv(info).        compute varcov = varcov(1:(2*(nv-nc-2)),1:(2*(nv-nc-2))).       compute ses = diag(varcov).       compute avar = ses(1:nrow(bzxse),1).       compute bvar = ses((nrow(bzxse)+1):nrow(ses),1).       do if ((nv-nc-2) > 1 and (!contrast = 1)).         compute prws=make(((nv-nc-2)*(nv-nc-3)/2),1,0).         compute prwse=prws.         compute kk=1.         loop ic = 1 to (nv-nc-3).         loop ic2 = (ic+1) to (nv-nc-2).         compute vf2 = ((byzx2(ic,1)**2)*varcov(ic,ic))-(2*byzx2(ic,1)*byzx2(ic2,1)*(varcov(ic,ic2)))     .         compute vf2=vf2+((byzx2(ic2,1)**2)*varcov(ic2,ic2))+((bzx(ic,1)**2)*(bvar(ic,1))).         compute vf2=vf2-(2*bzx(ic,1)*bzx(ic2,1)*covmat((2+ic),(2+ic2)))+((bzx(ic2,1)**2)*(bvar(ic2,     1))).         compute cnt = indeff2(ic,1)-indeff2(ic2,1).         compute prws(kk,1)=cnt.         compute prwse(kk,1)=sqrt(vf2).         compute kk=kk+1.         end loop.         end loop.         compute cnam2 = cname(1:(kk-1),1).       end if.       compute dermat = {byzx2;bzx}.       compute totse = sqrt(t(dermat)*varcov*dermat).       compute specse = sqrt((byzx2&*byzx2)&*(avar)+(bzx&*bzx)&*(bvar)).       compute specse = {totse; specse}.              compute specz = {csum(indeff2);indeff2}&/specse.       compute ind22 = {csum(indeff2);indeff2}.       compute nms = {"TOTAL";nm(3:(nv-nc),1)}.       do if ((nv-nc-2) > 1 and (!contrast = 1)).         compute ind22 = {ind22;prws}.         compute specse = {specse;prwse}.         compute specz = {specz;(prws&/prwse)}.   231         compute nms = {nms;cnam2}.       end if.       compute pspec= 2*(1-cdfnorm(abs(specz))).       compute spec = {ind22, specse, specz, pspec}.       print/title = "******************************************************************".       print/title = "           NORMAL THEORY TESTS FOR INDIRECT EFFECTS".       print spec/title = "Indirect Effects of IV on DV through Proposed Mediators (ab "+     "paths)"/rnames = nms/clabels "Effect" "se" "Z" "p"/format = f9.4.     end if.   end if. end loop. do if (btn > 1).   compute nms = {"TOTAL"; nm(3:(nv-nc),1)}.   do if ((nv-nc-2) > 1 and (!contrast = 1)).     compute crst = make((n+1+btn),((nv-nc-2)*(nv-nc-3)/2),0).     compute kk=1.     loop ic = 2 to (nv-nc-2).       loop ic2 = (ic+1) to (nv-nc-1).         compute crst(:,kk)=indeff(:,ic)-indeff(:,ic2).         compute kk=kk+1.       end loop.     end loop.     compute indeff = {indeff,crst}.     compute cnam2 = cname(1:(kk-1),1).     compute nms = {nms;cnam2}.   end if. compute lvout = indeff(2:(n+1),:). compute tdotm = csum(lvout)/n. compute tm = (make(n,ncol(lvout),1))*mdiag(tdotm). compute topa = csum((((n-1)/n)*(tm-lvout))&**3). compute bota = 6*sqrt((csum((((n-1)/n)*(tm-lvout))&**2)&**3)). compute ahat = topa&/bota. compute indsam = t(indeff(1,:)). compute boot = indeff((n+2):nrow(indeff),:). compute mnboot = t(csum(boot)/btn). compute se = (sqrt(((btn*cssq(boot))-(csum(boot)&**2))/((btn-1)*btn))). save boot/outfile = indirect.sav/names = nms. loop #e = 1 to ncol(indeff).   compute boottmp = boot(:,#e).   compute boottmp(GRADE(boot(:,#e))) = boot(:,#e).   compute boot(:,#e) = boottmp. end loop. compute xp = make((nrow(mnboot)+2),1,0). loop i = 1 to (nrow(mnboot)+2).   do if (i <= nrow(mnboot)).     compute pv = (boot(:,i) < indsam(i,1)).     compute pv = csum(pv)/btn.   else.     compute pv = zbca((i-nrow(mnboot)),1).   end if.   compute p = pv.   do if (pv > .5).     compute p = 1-pv.   end if.   compute y5=sqrt(-2*ln(p)).   232   compute xp(i,1)=y5+((((y5*p4+p3)*y5+p2)*y5+p1)*y5+p0)/((((y5*q4+q3)*y5+q2)*y5+q1)*y5+q0).   do if (pv <= .5).     compute xp(i,1) = -xp(i,1).   end if. end loop. compute bbb = nrow(mnboot). compute zz = xp(1:bbb,1). compute zlo = zz + ((zz+xp((bbb+1),1))&/(1-t(ahat)&*(zz+xp((bbb+1),1)))). compute zup = zz + ((zz+xp((bbb+2),1))&/(1-t(ahat)&*(zz+xp((bbb+2),1)))). compute ahat = 0. compute zlobc = zz + ((zz+xp((bbb+1),1))&/(1-t(ahat)&*(zz+xp((bbb+1),1)))). compute zupbc = zz + ((zz+xp((bbb+2),1))&/(1-t(ahat)&*(zz+xp((bbb+2),1)))). compute zlo = cdfnorm(zlo). compute zup = cdfnorm(zup). compute zlobc = cdfnorm(zlobc). compute zupbc = cdfnorm(zupbc). compute blow = trunc(zlo*(btn+1)). compute bhigh = trunc(zup*(btn+1))+1. compute blowbc = trunc(zlobc*(btn+1)). compute bhighbc = trunc(zupbc*(btn+1))+1. compute lowbca = make(nrow(blow),1,0). compute upbca = lowbca. loop i = 1 to nrow(blow).   do if (blow(i,1) < 1).     compute blow(i,1) = 1.   end if.   compute lowbca(i,1)=boot(blow(i,1),i).   do if (bhigh(i,1) > btn).     compute bhigh(i,1) = btn.   end if.   compute upbca(i,1)=boot(bhigh(i,1),i). end loop. compute lowbc = make(nrow(blow),1,0). compute upbc = lowbca. loop i = 1 to nrow(blowbc).   do if (blowbc(i,1) < 1).     compute blowbc(i,1) = 1.   end if.   compute lowbc(i,1)=boot(blowbc(i,1),i).   do if (bhighbc(i,1) > btn).     compute bhighbc(i,1) = btn.   end if.   compute upbc(i,1)=boot(bhighbc(i,1),i). end loop. print/title = "*****************************************************************". print/title = "           BOOTSTRAP RESULTS FOR INDIRECT EFFECTS". compute res = {indsam, mnboot,(mnboot-indsam), t(se)}. print res/title = "Indirect Effects of IV on DV through Proposed Mediators (ab paths)"/rnames =      nms/clabels "Data" "Boot" "Bias" "SE"/format f9.4. compute lowperc = boot(blowp,:). compute upperc = boot(bhighp,:). compute ci = {lowbca, upbca}. do if (!bca <> 0).   233   print ci/title = "Bias Corrected and Accelerated Confidence Intervals"/rnames = nms/clabels      "Lower" "Upper"/format F9.4. end if. do if (!bc <> 0).   compute ci = {lowbc, upbc}.   print ci/title = "Bias Corrected Confidence Intervals"/rnames = nms/clabels "Lower"      "Upper"/format F9.4. end if. do if (!percent <> 0).   compute ci = {t(lowperc), t(upperc)}.   print ci/title = "Percentile Confidence Intervals"/rnames = nms/clabels "Lower" "Upper"/format      F9.4. end if. print/title = "*****************************************************************". print conf/title = "Level of Confidence for Confidence Intervals:". print btn/title = "Number of Bootstrap Resamples:". end if. do if ((nv-nc-2) > 1 and (!contrast = 1) and ((!normal = 1 and nc = 0) OR btn > 999))). print/title = "*****************************************************************". print/title = "  INDIRECT EFFECT CONTRAST DEFINITIONS: Ind_Eff1 MINUS Ind_Eff2". compute kk=1. compute prwsv=make(((nv-nc-2)*(nv-nc-3)/2),2,0).  loop ic = 1 to (nv-nc-3).         loop ic2 = (ic+1) to (nv-nc-2).           compute prwsv(kk,1)=nm(ic+2,1).           compute prwsv(kk,2)=nm(ic2+2,1).           compute kk=kk+1.        end loop. end loop. compute prwsv = {cnam2, prwsv}. print prwsv/title = " "/clabels = "Contrast" "IndEff_1" "IndEff_2"/format A9. end if. Print/title = "********************************* NOTES **********************************". do if (btn = 1). Print/title = "Bootstrap confidence intervals are preferred to normal theory tests for inference "+     "about indirect effects". Print/title = "See Hayes, A.F.(2009). Beyond Baron And Kenny: Statistical mediation analysis in "+     "the new millennium". Print/title = "Communication Monographs, 76, 408-420". end if. do if (bdbp > 0). print/title = "*****************************************************************". print/title = "WARNING: SOME BOOTSTRAP MATRICES WERE SINGULAR". print/title = "SINGULAR MATRICES WERE REPLACED DURING RESAMPLING".   234 print bdbp/title = "Number of singular bootstrap samples replaced:". end if.    do if (ovals = 2).    print/title = "*****************************************************************".    print/title = "NORMAL THEORY TESTS NOT AVAILABLE FOR MODELS WITH DICHOTOMOUS OUTCOMES".    do if (!boot = 0).    print/title = "To obtain indirect effects, request bootstrapping".    end if.    end if.    do if (nc > 0 and !normal = 1).    print/title = "NORMAL THEORY TESTS NOT AVAILABLE IN MODELS WITH COVARIATES".    do if (!boot = 0).    print/title = "To obtain indirect effects, request bootstrapping".    end if.    end if. END MATRIX. RESTORE. !ENDDEFINE. INDIRECT y = LSA_composite/x = WheelCon_center/m = WCskills_center FCI_center Province Sex      Assistance_with_WC/boot = 1000/conf = 95/normal = 0/contrast = 0/percent = 0/bc = 1/bca = 0.                             235 CHAPTER 4:  Health, personal, and environmental correlates of self-efficacy with using manual wheelchair  Health Condition: REGRESSION   /MISSING LISTWISE   /STATISTICS COEFF OUTS R ANOVA   /CRITERIA=PIN(.05) POUT(.10)   /NOORIGIN    /DEPENDENT WheelCon   /METHOD=BACKWARD FCI_center.  REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS R ANOVA   /CRITERIA=PIN(.1) POUT(.11)   /NOORIGIN    /DEPENDENT WheelCon   /METHOD=FORWARD FCI_center.  Personal Factor: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS R ANOVA   /CRITERIA=PIN(.1) POUT(.11)   /NOORIGIN    /DEPENDENT WheelCon   /METHOD=ENTER FCI_center   /METHOD=FORWARD Sex Age_center Training_with_WC Assistance_with_WC Hours_center.  REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS R ANOVA   /CRITERIA=PIN(.05) POUT(.1)   /NOORIGIN    /DEPENDENT WheelCon   /METHOD=BACKWARD Sex Age_center Training_with_WC Assistance_with_WC Hours_center   /METHOD=ENTER FCI_center.  Environmental Factor: REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS R ANOVA   /CRITERIA=PIN(.05) POUT(.1)   236   /NOORIGIN    /DEPENDENT WheelCon   /METHOD=BACKWARD SIT_2orMore ISEL_center Comm_center   /METHOD=ENTER FCI_center Sex Age_center Training_with_WC Assistance_with_WC Hours_center.  REGRESSION   /MISSING PAIRWISE   /STATISTICS COEFF OUTS CI(95) R ANOVA COLLIN TOL CHANGE   /CRITERIA=PIN(.1) POUT(.11)   /NOORIGIN    /DEPENDENT WheelCon   /METHOD=ENTER FCI_center   /METHOD=ENTER Age_center Sex Training_with_WC Assistance_with_WC Hours_center   /METHOD=FORWARD SIT_2orMore ISEL_center Comm_center   /SCATTERPLOT=(*ZRESID ,*ZPRED)   /RESIDUALS DURBIN HISTOGRAM(ZRESID) NORMPROB(ZRESID)   /CASEWISE PLOT(ZRESID) OUTLIERS(3)   /SAVE COOK.   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074105/manifest

Comment

Related Items