UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The determinants of late life exercise in women over age 70 Cousins, Sandra 1993

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

Item Metadata


831-ubc_1993_fall_phd_cousins_sandra.pdf [ 14.45MB ]
JSON: 831-1.0064461.json
JSON-LD: 831-1.0064461-ld.json
RDF/XML (Pretty): 831-1.0064461-rdf.xml
RDF/JSON: 831-1.0064461-rdf.json
Turtle: 831-1.0064461-turtle.txt
N-Triples: 831-1.0064461-rdf-ntriples.txt
Original Record: 831-1.0064461-source.json
Full Text

Full Text

THE DETERMINANTS OF LATE LIFE EXERCISE IN WOMEN OVER AGE 70 by  SANDRA O'BRIEN COUSINS B.P.E., The University of British Columbia, 1971 M.P.E., The University of British Columbia, 1977  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF EDUCATION Department of Administrative Adult and Higher Education  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA April 1993 © Sandra O'Brien Cousins, 1993  In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  (signat  Department of  4, A H, 1  -  ,  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  Aiiplif  Pi/ 1990  Administrative Adult and Higher Education  ii ABSTRACT Too many elderly women suffer rapid aging decline, frailty and hypokinetic disease simply because of inadequate levels of physical activity. While the biopsychosocial benefits of regular exercise are now well-known, explanations are lacking for the reluctance of aging Canadian females to take up, or keep up, healthful forms of leisure-time physical activity. The purpose of this study was to examine and explain the variability of participation in health-promoting forms of exercise in elderly women. Several health behavior theories and personal attributes have shown promise in explaining exercise behavior, and thus, a second purpose of the study was to test the utility of a composite theoretical model. The composite model included ten personal and situational attributes as well as five cognitive beliefs about physical activity adapted from Social Cognitive Theory and a belief about personal control over one's health from Health Locus of Control Theory. A city-wide sample of 327 Vancouver women aged 70 and 98 years filled out survey questionnaires providing information on the 16 model variables in addition to kilocalorie estimates of exercise in the past week. Multiple regression analysis was used to explain late life exercise in three stages: 1) regression on the ten personal and situational attributes; 2) regression on the six cognitive beliefs; and 3) combined regression on all the significant predictors. From the life situational variables, health, childhood movement confidence, school location, and age were significant factors explaining 18% of the variability seen in current exercise level. From the cognitive variables, current self-efficacy to exercise and current social support to engage in physical activity were the only significant predictors (R 2 = 22%). A full regression model was tested by including the four statistically important situational variables  111 and the two cognitive variables from the previous analyses. The utility of the Composite Model was supported in that both situational variables and selfreferent beliefs played significant and independent roles in explaining late life exercise (R 2 = 26%). The main reasons that older women were physically active were: 1) they perceived high levels of social support to exercise (b = .239, p < .01); 2) they felt efficacious for fitness-types of activities (b = .185, p < .01), 3) they had satisfactory health (b = .174, p < .01), and 4) they were educated in foreign countries (b = -.125, p < .01). Health locus of control offered some explanation but was not able to demonstrate significance alongside other cognitive beliefs (b = -.106, p < .06). Education, socioeconomic status, work role, family size, and marital status were not able to explain late life exercise. This study found that health difficulties do indeed interfere with women's activity patterns. However, women are also influenced by perceptions of declining social support, lower levels of movement confidence, and chronological age, to reduce their physical activity. Thus, regardless of their health situation, the explanation of exercise involvement in older women rests to a large degree on the amount of social encouragement they perceive from family, friends and physicians, their self-efficacy for fitness activity, as well as perceptions of age-appropriate behavior. Older women who were educated as children outside of Canada, Britain and the U.S. appear to be culturally advantaged for late life physical activity participation. Moreover, childhood movement confidence stands as a significant predictor among the situational variables. These findings suggest that participation in physical activity, and positive beliefs about exercise in late oo , are rooted in competencies and experiences acquired in childhood.  iv Perceptions of inadequate encouragement appear to be limiting females, from childhood on, to develop and sustain confidence in their physical abilities that would promote a more active lifestyle into their oldest life stage.  ^  v  TABLE OF CONTENTS  ^ABSTRACT  ii  TABLE OF CONTENTS ^  v  LIST OF TABLES ^  xvi  LIST OF FIGURES ^  xv  ACKNOWLEDGEMENTS ^  xviii  DEDICATION ^ PROLOGUE  v xix  CHAPTER I. INTRODUCTION Background to the Problem ^  1  Statement of the Problem ^  8  Research Questions ^  10  Rationale and Significance of the Study ^ 10 Definition of Key Terms ^  14  Delimitations ^  18  Limitations ^  18  Organization of the Remaining Chapters ^ 20 CHAPTER II. THE REVIEW OF THE LITERATURE Introduction ^  22  The Poor Aging of Women ^  24  The Activity Socialization of Females Born Before 1921 . ^ 25 A Demographic Profile of Women Born Before 1921 ^ 28 Size of the Aging Population ^  29  ^  vi Marital Status, Family Size and Socioeconomic Status ^  31  Education ^  33  Ethnicity ^  34  Health Status ^  35  Institutionalization Elderly Women's Exercise Patterns ^  36 37  Known Exercise Patterns of Older Women ^ 37 Known Benefits of Exercise Participation ^ 40 Introduction ^  40  The Benefits of Exercise ^  41  Immediate Benefits ^  44  Long-term Benefits ^  46  Known Risks of Exercise Participation ^ Introduction ^  51 51  Known Risk of Sudden Death in the Elderly ^ 52 Known Risk of Injury in the Elderly ^ 54 Risk of Over-Exertion or Exhaustion ^ 55 Risk of Provoking Ill Health ^  56  Summary ^  57  The Determinants of Exercise Behavior: Useful Theoretical Approaches ^ Introduction to Determinants ^ Social Epidemiological Perspectives ^  57 57 59  Social Epidemiology ^  59  Age and Exercise ^  mi  Education and Exercise ^  62  ^  vii Subjective Health and Exercise ^  63  Health Symptoms and Exercise Behavior ^ 67 Marital Status and Exercise ^  68  Motherhood, Children and Leisure-time Exercise . . ^ 69 Work Role, Employment and Exercise ^ 72 Socioeconomic Status and Exercise ^ Socialization Theory ^  74 75  The Significance of Socialization for Childhood Physical Activity ^  78  Childhood Situation: Opportunity, Mastery, Movement Competence ^  81  Behavioral Health Psychology Perspectives ^ 85 The  Health  Belief  Model  85  Health Locus of Control Theory ^  89  Theory of Reasoned Action ^  91  Social Cognitive Theory ^  92  Social Learning Theory ^  92  The Constructs of Social Cognitive Theory ^ 93 The Cognitive Determinants of Exercise: SCT ^ 96 Incentives or Motives to Exercise ^ Health as Incentive to Act ^  97 99  Outcome Expectations ^  100  Self-Efficacy and Movement Confidence ^  101  Introduction: Self-Efficacy ^  101  Barriers to Perceived Efficacy to Exercise . . . ^103 The Role of Habit and Previous Physical Activity . 105 The Role of Efficacy on Exercise Behavior ^ 106  viii The Role of Exercise on the Development of Efficacy ^  109  Environmental^Cues for Physical^Activity:^Social^Support  .^.  Ageist Practice in Communities Social^Support  113  ^  Defining and Measuring Social^Support  114 ^  115  ^  116  Physical Activity as Social^Networking Role of Group Cohesion Role of Companionship  113  117 ^  The Role of Spousal^Support The Role of the Physician Role of Friends and Family  118  ^  ^ ^  118 120 122  CHAPTER III. THEORETICAL FRAMEWORK Introduction  ^  124  Selection of the Most Suitable Theoretical Perspective ^  124  The Application of Social^Cognitive Theory^• •^.^.  126  ^  The Synthesis of Theory: The Composite Model Life Situational^Variables  ^  The Cognitive Variables Health Incentive  128 131 133  ^  Outcome Expectations of Late Life Exercise  133 ^  135  Perceived Benefits and Risks of Exercise Participation ^  137  Defining Perceived Risks of Exercise Participation ^  139  Age and Risk Perception  ^  143  ^ ^  ix Movement Confidence (Self-Efficacy)  143  Social Support to Exercise ^  147  Health Locus of Control ^  148  Summary of the Theoretical Framework ^ 149  CHAPTER IV. DESIGN OF THE STUDY Survey Questionnaire Construction ^ Description ^  150 150  The Older Adult's Exercise Status Inventory ^ 152 Description ^  152  Reliability of the OA - ESI ^  154  Validity of the OA- ESI ^  155  Use of the OA - ESI ^  156  Interpretive Variables ^  160  Life Situational Measures ^  162  Introduction ^ Age  162 162  Culture  162  Socioeconomic Status ^  163  Marital Status ^  164  Education  164  Work  165  Role  Family Size: Number of Children  165  Health Variables ^  165  Self-Rated Physical  Health  165  Symptoms  166  ^  x Perceived Well-Being ^ Medications  166 166  Composite Health Index ^ Body  Mass  167  167  Childhood Movement Confidence ^  168  Childhood Social Support to Exercise ^ 171 The Cognitive Variables ^  172  Health Incentive ^  172  Adult Movement Confidence ^  173  Adult Social Support ^  175  Perceived Risks and Benefits of MCN Exercises . . ^ 175 Health Locus of Control ^  176  The Pilot Study ^  178  Selection of the Sample ^  182  The Population ^  182  Sampling Procedure ^ The Strategic Sample ^  183 183  Second Sampling Procedure: Convenience Sample ^ 188 Data Collection Protocol ^ Data Preparation ^  193 198  Missing Data ^  198  Cleaning the Data ^  198  Outliers ^  199  Statistical Analysis ^  199  CHAPTER V. RESULTS Description of the Sample ^  202  ^  xi Descriptive Results of the Situational Variables . . . ^ 204 Age  204  Marital Status ^  204  Family Size: Number of Children ^ 204 Work Role ^ Cultural Origin: School Location  204 205  Education ^  205  Socioeconomic Status ^  205  Medical Symptoms ^  206  Self-Rated Health ^  207  Childhood Social Support ^  207  Childhood Movement Confidence ^  207  Lifelong Status ^  208  The Typical Week ^  208  Total Exercise Sessions in the Past Week ^ 208 Hours of Activity in the Past Week ^ 209 Change of Activity in the Past Five Years ^ 209 Descriptive Results of the Cognitive Variables ^ 209 Health Incentive ^  210  Adult Social Support to Exercise ^ 211 Adult Movement Confidence to Exercise ^ 211 Outcome Expectations ^  211  Health Locus of Control ^  212  Descriptive Results of Late Life Exercise ^ 212 Energy Expended in the Past Week ^ 212 Statistical Findings ^  214  ^  xii  Correlations ^  214  Multiple Regression Analysis ^  218  Analysis 1: Life Situational Variables ^ 219 Analysis 2: Cognitive Variables ^ 220 Analysis 3: Best Situational and Cognitive Variables ^  221  Post-Hoc Analyses: Interactions ^ 226 CHAPTER VI. SUMMARY AND DISCUSSION Summary of the Findings ^  228  Descriptive Findings ^  228  Theoretical Findings ^  229  Discussion of the Methodology ^  231  Self-Report Interviews ^ Statistical  Analysis  231 231  Movement Confidence and Social Support Scales . . ^ 232 The Older Adult - Exercise Status Inventory . . . ^ 234 Discussion of the Findings ^  235  Discussion of the Criterion Variable ^ 235 Life Situational Variables ^  237  Cognitive Variables ^  238  Discussion of the Model: Theoretical Significance . . ^ 241 Chief Limitations and Suggestions for Future Research ^  244  Self-Selection Bias and Survivorship ^ 244 Variables With Poor Response Rate ^ 246 Generalizability ^ Quality of the Data ^  248 249  Unanswered Questions: Gaps in the Present Work ^ 252 Missing Constructs ^  252  Identifying the Perceived Risks of Exercise . . . ^ 252 The Significance of Domestic Activity ^ 253 The Significance of Being a Turn-of-the-Century Tomboy ^ ^ Policy Implications of the Findings  255 257  Creating Social Support for Physical Activity . . . 257 Increasing Self-Efficacy and Incentive for Physical ^ Activity  259  Implications for Professional Practice in Adult ^ Education 260 CHAPTER VIII. REFERENCES APPENDIX A:  The Questionnaire  264 311  APPENDIX B: Ethics Approval Certificate ^ 336 APPENDIX C: Agency Approval Forms ^  340  APPENDIX D: Subjects' Letters and Comments ^ 343 APPENDIX E: Post-Hoc Analyses: Missing Data ^ 355 APPENDIX F: Residual Error ^  359  APPENDIX G: The Measurement of Physical Activity The Outcome Variable ^  361  xiv  LIST OF TABLES Table 2.0^Demographic Structure of Elderly Women ^ 30 Table 2.1^The Known Benefits of Exercise ^ 42, 43 Table 4.0^Metabolic Units of 38 Activities ^ 159 Table 4.1^Movement Skills Estimating Movement Confidence and Past Experience 174 Table 4.2^Correlation Coefficients and Significance Levels of Test-Retest Data ^  181  Table 4.3^Geographic Clusters and the Randomly Selected Sites ^  187  Table 4.4^Descriptive Characteristics of the Two Research Samples and the Pilot Sample^. . . . 190 Table 4.5^Results of T-Tests on Key Variables Between the Strategic Sample and Convenience Sample^. 192 Table 4.6^Chronological Distribution and Return of the Questionnaire by Research Site ^ 196 Table 4.7^Distribution of Questionnaires by Program Type Table 5.0^Descriptive Characteristics of the Sample, Vancouver and Canadian Data  197  203  Table 5.1^The Descriptive Statistics for the Cognitive Variables 210 Table 5.2^Correlation Matrix  of the Fxplanatnry Variables  216  XV  Table 5.3  Table 5.4  Table 5.5  Unstandardized Regression Coefficients and Standardized Errors for the Regression of Late Life Exercise on the Situational Variables  222  Unstandardized Regression Coefficients and Standardized Errors for the Regression of Late Life Exercise on the Cognitive Variables  223  Unstandardized Regression Coefficients and Standardized Errors for the Regression of Late Life Exercise on the Significant Variables of the Full Model  224  Table E^Comparison of the Missing Data Group with the Completed Data Group  358  xvi  LIST OF FIGURES Figure 2.0^Basic Elements of the Health Belief Model ^. . . 86 Figure 2.1^Basic Elements of Social Learning Theory ^ 93 Figure 2.2^Bandura's Concept of Reciprocal Determinism  95  Figure 3.0^The Composite Model of Elderly Female Physical Activity  130  Figure 4.0^City of Metropolitan Vancouver: The 18 Research Sites  185  Figure 5.0^Activity Levels of Vancouver Women Aged 70+ ^. 213 Figure 5.12^Proposed Model of Explanation of Elderly Women's Physical Activity  217  Figure F^Probability Plot of Residual Error ^ 360  xvii  ACKNOWLEDGEMENTS A project of this magnitude comes at a cost. I am grateful to my children, Catherine and Kristina O'Brien who waited and watched Mommy working on her dissertation for almost five years. They have provided patience and understanding beyond their years. My mother, Gladys Hartley, was invaluable in filling out the original survey questionnaire. Her comments and suggestions significantly altered the form of the data collection and ultimately improved the quality of the research. David Cousins, my husband, is "the wind beneath my wings." He has been my head coach on the computer and empowered me to become confident with many forms of software. Furthermore, David advocates a feminist perspective, and a critical perspective, which I am sure has affected the way I went about this study. As I write this, those following the 1992 Summer Olympics in Barcelona are witnessing the basketball "dream team" from the United States. It is also a suitable label for my doctoral committee: Dr. Patricia Vertinsky, Dr. Doug Willms and Dr. Kjell Rubenson. Not only are all three stellar in their individual professionalism making them "the best", they are able to capitalize on each others strengths and work together as a graduate student's "dream team". I thank them for their carefully thought out advice all along the way. To Patricia, in particular, I must acknowledge exceptional support. The provision of financial assistance and access to her files and journals were beyond the call of duty. Until 1988, I had never met any academic who I wanted to emulate. She immediately earned my respect and I hope that we'll continue to work on collaborative projects into the future. Finally, I am indebted to the 17 women of the U of Agers gymnastics team and the 327 women who gave of their valuable time to be part of the study. Unfortunately, most of these women may never fully appreciate the contribution they have made to the limited field on knowledge about women's activity patterns over the life course and the determinants of them.  DEDICATION Dad, this is for you. I remember your advice when I was 17 years old. You said, "Why don't you go to University just for one year so you will know what it is like?" Well, here is the outcome of what happened after that one year. Twenty-eight years later, I am in my twelfth year of university studies. And the most important thing that I have learned in all this time is that I will never stop learning.  xix  PROLOGUE In 1985, I attended Dr. Ernst Jokl's keynote speech at the Physical Activity, Aging and Sport Conference at Westpoint Military Academy. I was intrigued by something he said. Play bestows and conveys the attitudes of youth. An old person who plays turns young. A young person who cannot play turns old. Ernst Jokl, 1985. The importance of physical recreation at all life stages is a concept that many physical educators endorse, but the lifelong significance of physical activity, exercise and sport has as yet to be scientifically determined. Dr. Jokl's speech started me thinking about the role of physical activity in healthy aging. Shortly after this conference, I began to teach fitness classes to adults aged 50 to 75. Many of the participants were truly unfit, and among the women in particular, were some who could not do even one modified pushup or situp. Others in the group demonstrated excellent physical abilities -- physical fitness and skills which compared favourably to adults 30 to 50 years younger. Within a few months, the unfit were significantly fitter, and the already fit had started taking up new physical challenges, namely, gymnastics! This was an intriguing phenomenon: elderly women venturing beyond the challenges of maintaining physical fitness, forging new skills in a young girls' sport, and then enjoying their efforts in public performances. There was a visible enthusiasm to test their physical limits and the commitment to training was obvious. Why were these women so eager now to participate in more vigorous physical activity?  xx The research literature is replete with evidence about the increasing biological, psychological and social heterogeneity that accompanies aging; it seems that human variability is no better demonstrated than in the exercise patterns and physical fitness measures of aging adults, some of whom have become functionally frail while others appear to successfully counter aging declines with high levels of physical activity and sustained physical fitness. Such visible evidence of these different outcomes for people led me to wonder why older people exercise as they do. Which of the many biological, social and psychological forces are operating to create such diverse lifestyles and health outcomes by late life? The fitness class developed into a performing gymnastics team called the U of Agers. Since May of 1986, the twenty-five women and men on "the team" have been seen live and on national television news broadcasts on a number of occasions. They are the documentary centrepiece of the National Film Board's (1990) "Age is No Barrier." They appear frequently on CBC's "The Best Years" and are solicited to perform at major events across Canada. These older adults have become a remarkable social phenomenon mainly because the Canadian public has so underestimated the physical capabilities and interests of older adults. This social perspective, which finds elderly athleticism incongruous, drives the sociological and psychological aspects of this research. Never before has society been better able to offer leadership, supportive technology and diverse opportunities for the physical education and recreation of aging adults. While a minority of elderly are ready for, and interested in, physical activity, the majority, do not appear to be interested enough to become participants. Curiosity about what leads to highly active versus highly sedentary living in older adulthood, has brought my research into focus.  1  I. INTRODUCTION  "Much more information is needed on how the determinants of physical activity change with age, with particular reference to factors influencing the participation of children and of middle-aged and elderly people." (Bouchard, Shephard, Stephens, Sutton & McPherson, 1990, p.10) Background to the Problem Differences in the way individuals age have intrigued scientists as well as lay people for years. The contemporary search for perpetual health and immortality has failed to find any "fountain of youth," but we know that people do age with remarkable variation (Nelson & Dannefer, 1992). Probably, the spectrum of aging possibilities are best seen in people's day-to-day lifestyle behaviors and the outcomes of these behaviors over the life-span. For at least three decades, gerontologists have proposed that there are two ways of growing old: "usual" aging and "successful" aging (Butler, 1988; Dermody, Saxon & Sheer, 1986; Havighurst, 1963; Meusel, 1991; Nowlin, 1985; Palmore, 1979; Rowe & Kahn, 1987). In recent decades, increased longevity has illuminated the problems of "usual" aging, clarified what is normal and abnormal aging, and compelled societies to "rethink how we age" (Prado, 1986). Evidence is rapidly accumulating that regular and moderate forms of physical activity are a resource to adults which foster opportunities for improved survival, life quality and more successful aging (O'Brien & Vertinsky, 1991; Stewart & King, 1991). Although knowledge about how to age better is improving, population disability levels are climbing due to the relative aging and inactivity of the population (Ramlow, Kriska & Laporte, 1987; WHO Health Education Unit, 1986).  2 Enough scientific support exists to suggest that moderate and frequent exercise may be the "best preventive medicine" for old age. Exercise is being prescribed, first to prevent premature aging (Spirduso, 1986), and second, to prevent premature disease by controlling hypertension, heart disease, bowel and breast cancer, the immune response, osteoporosis, obesity, arthritis, diabetes, insomnia, and depression (see Chapter 3 for a complete review). Exercise scientists are beginning to understand the significant role which exercise can play in controlling aging decline and delaying mortality (Blair, Kohl,III, Paffenbarger, Jr., Clark, Cooper, & Giddons, 1989; Donahue, Abbott, Reed & Yano, 1988; Grand, Grosclaude, Bocquet, Pous, & Albarede, 1990; Kaplan, Seeman, Cohen, Knudsen & Guralnik, 1987; Linsted, Tonstad, & Kuzma, 1991; Rakowski & Mor, 1992). Physiologists estimate that up to half of what we currently know as usual aging is a phenomenon of disuse (Berger, 1989; Bortz, 1982; DeVries, 1970, 1974; DeVries & Adams, 1972; Shephard, 1989a; Smith, 1981) -- disuse which hurries more women than men toward experiences of hypokinetic disease (Abdellah, 1985; Butler, 1968; Heckler, 1984; Ostrow, 1989; Verbrugge, 1990; Vertinsky, 1991). Government health promotion. programs (Don't Take It Easy, 1983; Choosing Wellness, 1988) are conspicuous public health campaigns addressing the issue of unfit aging in Canada - an issue particularly targeting women who appear to have much to gain from increased participation in physical activity. Dozens of scientifically-controlled exercise interventions, particularly since the 1980's, have provided substantial evidence that elderly individuals can positively affect their mobility, endurance, strength and balance by first, reversing the circulatory and neurological limitations they have acquired through sedentary living (MacRae, 1989; Spirduso, 1986), and second, by elevating their functional capacities to the level of adults decades younger than themselves  3 (Conger & O'Brien, 1989; Dummer, Clarke, Vaccaro, VanderVelden, Goldfarb & Sockler, 1985; Fiatarone, Marks, Ryan, Meredith, Lipsita, & Evans, 1990). Lee (1991) has reviewed the exercise intervention studies pertaining to middle-aged and older women and concludes that "older women have the potential to benefit from exercise to much the same degree as men" (p. 133). Furthermore physical activity is known to be more than just a preventive and controlling measure; at certain intensities, sustained human activity is healthpromoting and can lead to a "high-level wellness" (Dunn, 1961; Teague, 1987). Exercise elevates function to levels which guarantee more years of independent living (Health & Welfare Canada, 1989). Regular physical activity places controls on aging decline which, in "usual" aging, contributes to physiological losses of about one percent per year in most body systems (deVries, 1979). A consensus is forming that the health of most adults, including the able elderly, can best be promoted with brisk walking at 40 to 75% of one's maximal heart rate (MHR = 220 age)(Bouchard et al., 1990). Until recently, however, little attention has been placed on the mechanisms of women's aging and activity patterns. Frail elderly women, especially, have been virtually invisible in feminist, sociological, and gerontological literature (Evers, 1985, p. 86). By many accounts, most women can expect to live to the ripe age of 80 years or better, but are not likely to age very "successfully" (Dulude, 1978; Lewin & Oleson, 1985; Posner, 1980; Quinlan, 1988; Verbrugge, 1990a; 1990b; Verbrugge & Wingard, 1987). As women age, they are said to slow down, and for many, up to 55% of their body mass becomes infiltrated with fat (Young, Blondin, Tensuan & Fryer, 1963). Significant declines in strength, endurance and aerobic fitness are apparent even by middle-age (Alexander, Ready, & Fougere Mailey, 1985; Cinque, 1990). Rarely are older women found participating -  4 in the more vigorous forms of sport and recreation (Cauley, LaPorte, Black Sandler, Schramm, & Kriska, 1987). Indeed, among older women, such activity is almost nonexistent except for the 12% who over a four-week period undertake from time to time a walk of two miles or more. (Abrams, 1988, p.32) Health promoters are concerned, not only about older women, but about much younger females too. Only 24% of Canadian girls aged 15 to 19 could achieve the recommended levels of aerobic fitness on a recent Canada Fitness Survey (Stephens & Craig, 1990). Other research has found that 20% of Canadian children are considered to be obese and that 80 to 85% of those children remain obese as adults. Despite public awareness of the benefits of exercise at every life stage, as yet, females at all ages are less active than their male counterparts (Stephens, Craig & Ferris, 1986), and they are seen to become progressively less active as they get older (Alexander, Ready, & Fougere-Mailey, 1985). By late life, only a small minority are adequately active to benefit their health and well-being (Blair, Brill & Kohl, 1988; Lee, 1991; Stephens & Craig, 1990; Teague & Hunnicutt, 1989). Statistics Canada (1990) reports that only 10% of women over the age of 45 are considered to be "active" compared with one in three males. In the U.S., only 1% of adult women regularly performed more than one vigorous activity (Sallis, Haskell, Wood, Fortmann, Rogers, Blair & Paffenbarger, Jr. (1985). In the eight extra years that women, on average, outlive their male counterparts, too many of them endure poor mental health (Grau, 1988), over onethird are frail and physically limited (Charette, 1988), and almost one quarter of women over age 65 use sleeping pills (Health & Welfare Canada, 1989). The aging difficulties of women have prompted researchers to examine differences in le r lifestyle. both medical and social professions would probably agree; the  5 one factor that is most likely to exacerbate the emotional and physical difficulties of very aged women is their inadequate leisure-time exercise -often a lifelong deficiency of vigorous and strength-promoting physical activity (Biddle & Smith, 1992; Verbrugge, 1990a; Vertinsky, 1991, Work, 1989). While "doctor's orders" might activate up to 25% of women, almost 60% of older women have said that "nothing would persuade them to increase their physical activity" (Shephard, 1986, p.136). Of interest is the fact that where other health behaviors are concerned, women generally exhibit better life habits than men; physical activity is the  only positive health behavior that is pursued  by men more than women (Stephens, 1985; Stephens & Craig, 1990). While public health campaigns about the risk of heart disease has spurred many men into joining health-promoting exercise programs (Davidson & Sedgewick, 1978), simply raising the issue of heart disease may have frightened aging women away. If there is a persuasive force to mobilize aging women to more activity, it might be "figure improvement" (Davidson & Sedgwick, 1978) and stress reduction (Duda & Tappe, 1989). "Feeling better" and "looking better" are important reasons why women have been physically active in the past (Canada Fitness Survey, 1983). But this reasoning is problematic. Almost half of Canadian females over age 45 are at risk of health problems due to obesity - a group who could most benefit from increased physical activity but who are at risk by doing so. Those women who are more inclined to be active are already leaner and healthier, and thus, the women who least need to increase activity levels are the most likely to do so (Dishman, 1990; Sallis, Haskell, Wood et al., 1985; Stephens & Craig, 1990). Possibly concerned about their health and safety, women prefer to exercise in groups under expert leadership. About 50% of active women over age 65 say they exercise in public places and are more likely than men to be in supervised  6 activity settings (Stephens & Craig, 1990). Female propensity for public participation does pose a problem for bigger and older women. Recent research has found that overweight women perceive social disapproval for their body size; unfortunately they also perceive disapproval and experience embarrassment in the exercise setting where they most anticipate rewards for participating (Bain, Wilson, & Chaikland, 1989). Evidently, older women who are active are already somewhat comfortable with their physique in public settings; the heavier women, who most want to look and feel better as they age, are unfortunately less likely to participate in, or adhere to, the kinds of programs which might help them succeed. But more important than female appearance must be female health. Just as many women will die of heart disease as will men (Statistics Canada, 1986), and yet women's risks have not been a focus of heart health campaigns (Nachitall & Nachitall, 1990). While women's spines and abdomens "take a tremendous beating in pregnancy and childbirth" (Davidson & Sedgewick, 1978, p.27), "older women who perform aerobic exercise for the sake of improved health are generally viewed somewhat suspiciously" (p. 27). Without sufficient exercise, older women exhibit a level of muscle weakness that places them in a category of "functionally disabled" (Branch & Jette, 1981; Work, 1989). Social scientists have argued convincingly that women have merely learned their social roles well; passive behavior is thought to be the outcome of a lifelong experience of female disempowerment and learned helplessness (Fedorak & Griffin, 1986; Myers & Huddy, 1985; Schulz, 1980; Zinberg & Kaufman, 1963). Others point to women's chronic stress from poverty (Labonte & Penfold, 1981), their commitment to caregiving for others (Robinson, 1988; Thomas, S.P., 1990), and fatigue from "daily hassles" (Mishler, Amarasingham, Hauser, Liem, & Others,  7 1981; Spacapan & Oskamp, 1989). Aging individuals apparently live up to the "self-fulfilling prophecy" of social expectation that labels older people, and women in general, as less physically competent (Kuypers & Bengston, 1973). Evers (1985) proposes that elderly women, more than men, continue to live at home with disabilities because "women are simply expected to be able to put up with limiting disabilities to a greater extent than are men" (p.89). Indeed it is a paradox that one of the main reasons given in surveys of elderly women for not being more physically active is their declining health and the perception that they are "too old," while at the same time scientific research increasingly demonstrates that one of the certain benefits of physical activity is health improvement. It is a further paradox that, while women have proven more durable than men from a physiological standpoint, they have done so in a culture which has, until recently, encouraged them to take on the characteristics of aging too readily. (Vertinsky, 1991, p.8) Elderly women, suggest some critics, impose more on the health care system, collect social security benefits and receive government assistance longer than men, and are most at risk of living out their last decade of life with severely diminished capacities (Statistics Canada, 1990 Women in Canada; Wilkins, Murb & Adams, 1983). These phenomena concern government and health care systems (Eriksson, Mellstrom & Svanborg, 1987), especially since older women are the fastest growing segment of the population. The women who reach the age of 90 outnumber their male counterparts by almost 3:1 (Statistics Canada, 1990). There are, however, examples of remarkably athletic elderly women with limited resources who have not shied away from vigorous involvements with their own serious and sometimes multiple health conditions (Drinkwater, 1988; Dummer, Clarke, Vaccaro, Vandervelden, Goldfarb, & Sockler, 1985; Gandee, Campbell, Knierim et al., 1989; National Film Board, 1990; Ruder, 1989; Starischka & Bohner, 1986; Wilmore, Miller & Pollock, 1974). While some females at all ages  • •^-^I'^.  V.  1 es y e, insu icien par icipation, especially in  8 vigorous play and sport, is more characteristic of the female life course from adolescence on (Vertinsky, 1992) and serves to highlight the heterogeneity of this social group. This heterogeneity is aptly described by Eric Pfeiffer as follows: In my considerable contacts with elderly persons, both clinically and socially, I have run into not only the lonely and the despaired and disabled elderly. I have also met some very, very exciting older people. Older people who were intellectually and socially stimulating and exciting, who were physically active and who obviously seemed to have made a successful adaptation to their growing years. Yet as I observed one after another aging person with whom I came in contact, there did seem to emerge a set of common characteristics for all or almost all of these persons. It struck me as though the successfully aging person was someone who somewhere along the way had decided to stay in training. He or she had decided to stay in training physically, intellectually and emotionally, and socially. (Pfeiffer, 1973, P. 3 ) This holistic view of human aging accounts for physical, intellectual, emotional and social developments and provokes the conception of biopsychosocial models that could better guide research, have clinical utility, and provide more comprehensive understanding (Engel, 1980; Levy, Derogatis, Gallagher & Gatz, 1980). McPherson (1986a) advocates the interdisciplinary approach in aging and sport research: ...there could very well be greater levels of explanation achieved concerning aging phenomena and the elderly if sport scientists from different disciplines were to pool their expertise. Specifically, greater attention needs to be directed to possible interactions among social, psychological, biological and physiological variables. (McPherson, 1986a, p. 8)  Statement of the Problem Too many older women are at heightened risk of suffering hypokinetic diseases and rapid aging decline simply because they are insufficiently physically active. Yet, the reasons for the deficient physical activity patterns of women in their seventh, eighth and ninth decade of life are virtually  9 unexplored. The paradox, that those aging women who could best improve their well-being through regular exercise, may often be the least likely to do so, suggests that certain barriers may be operating. These barriers to more active lifestyles need to be identified if women are to age with better life quality, less chronic disease, and full independence. Until the main barriers to older women's physical activity involvement are more clearly identified, social and educational programs cannot be designed with clear objectives for change. As evidence mounts regarding the significant health-promoting role of exercise in mental, social and physical well-being, social scientists ask, "Why are so few older women taking advantage of the 'best preventive medicine?'"  To  address this question, some researchers have focussed on people's beliefs about physical activity -- beliefs which are thought to be socially learned and internalized. Other research suggests that people are victims of their circumstances, and that gender, age, health, education and financial means are more likely to be the limiting forces affecting individual lifestyle behavior. Little research has attempted to mesh these two theoretical perspectives into a single study even though both perspectives appear to have merit in the explanation of why some older women are highly active while many more fall far short of adequate physical activity. This study brings together the most promising explanations of late life exercise for older women by merging the cognitive beliefs of Social Cognitive Theory and Health Locus of Control Theory with ten personal and situational attributes. Guided by a composite theoretical approach, this study aims to capture the most important influences on the physical activity behavior of a community-wide sample of women over age 70.  10 In particular, I am interested in the answers to these research questions: 1.  What are the participation patterns of older women in leisure-time physical activity?  2.  To what extent can life situation explain variability in leisure-time physical activity behavior in women over age 70?  3.  To what extent can cognitive beliefs explain variability in leisure-time physical activity behavior in women over age 70?  4.  Does a composite theoretical model have utility in explaining late life exercise?  Rationale and Significance of the Study The research presented here explores needed and promising explanations for older women's exercise behavior. This study integrates potentially useful constructs from several theoretical perspectives; altogether there are 16 variables with biological, psychological, sociological, or environmental origins which have been found to predict health behavior and physical activity behavior. McPherson has asserted that: ...future work concerning socialization in a sport context needs to abandon the almost exclusive use of the functionalist perspective and become more theoretically integrated. This does not imply that eclecticism should prevail, but rather that conscious attempts to examine the process from a merged theoretical perspective are needed to advance knowledge... In short, not only is there a need for greater use of microlevel theories but also for the integration and synthesis of theories within sociology and between sociology, psychology, and related disciplines. (McPherson, 1986b, p.116) To assist in understanding the barriers to more active lifestyles, the Health Promotion Survey (Charette, 1988) and the recent Campbell's Survey (Stephens & Craig, 1990) provide substantial information on the physical activity of Canadian adults of all ages. These government-funded studies have assessed activity behavior in representative populations and have begun to identify  11 various situational and psychological barriers to exercise by age and gender groupings. But to date, little is known about the relative importance of these barriers in explaining exercise behavior. More importantly, much of the research has searched for explanation without the guidance of human behavior theory. When behavioral theories have been used to predict exercise, they are often partially applied using only one or two key constructs. Moreover, current behavioral theories do not account for past situations and former attributes which may be important to the explanation of older adult behavior. Behavioral theories have tended to focus on the prospective perceptions, attitudes, and beliefs of adults, (all of which are considered to be alterable), with little regard for their social settings and personal circumstances which may make behavior modification difficult. Thus there is inadequate information about the relative influence of social and environmental factors  versus individual beliefs in explaining late  life exercise. Yet identifying the main biopsychosocial  determinants' of active  lifestyles and their relative significance and interactions would assist a range of professionals and agencies on how best to help older women age more successfully. This study will enrich Canadian data concerning the epidemiology of women's exercise patterns in old age - information that may assist physical educators and health practitioners in promoting the physical abilities and interests of older women. The specific nature and scope of the physical activity patterns of Canadian women over age 70 is poorly understood and data are lacking on the particular activities in which older women are engaged. As well, normative data are lacking on the amount of energy expended on weekly exercise by older women.  The term determinant is used according to the definition of Dishman (1990), that is to denote a reproducible association or predictive relationship other than cause and effect.  12 The development of the Older Adult Exercise Status Inventory for the purpose of this study will provide a potential solution to the problem of assessing the weekly participation of the elderly in exercise pursuits. The inventory, based upon a seven-day recall design in combination with the Canada Fitness Survey form, was designed specifically for older adult activity assessment. This kind of inventory is considered to be "a pragmatic approach for large populations for which direct observation or objective monitoring cannot be implemented" (Dishman, 1990, p.94). The inventory is easy to use, yet it generates a great deal of information about the frequency, intensity, duration and specific type of activities engaged in by older adults. The instrument has demonstrated reliability, concurrent validity with two contemporary field instruments and is built upon instruments that have demonstrated criterion validity. While much health promotion literature is aimed at finding ways to increase the physical well-being of the elderly, the age cohort of women under study (age 70 and older) has largely been neglected in previous research in terms of understanding the factors related to lifelong vigor and exercise participation. This study examines women, rather than men, because of their unique and often limited experiences with vigorous forms of physical activity, because of their relatively poor participation rate in late-life physical activity, and the fact that women are long-living, more chronically ill, and under-researched. Unique to this study is the self-assessment of the perceived risks and perceived benefits of six different fitness-related activities. In addition to Likert rating scales, subjects were asked open-ended questions about what they perceived to be the benefits and risks of six specific exercises commonly found in adult fitness classes. These data appear to be the first of their kind to  13 elicit responses from older women about their fears and hopes about the expected outcomes of participating in a contemporary exercise setting. This study initiates an important line of enquiry exploring the life situations and personal factors which lead to movement confidence and the perceived ability to be physically involved at different life stages. Recent work in the epidemiology of exercise participation stresses the importance of a life course perspective in forming and maintaining health promoting behavioral patterns - patterns which have the potential to be socially and environmentally influenced positively or negatively at any point in life span. The present research takes a retrospective and prospective look at female involvement in vigorous physical activity at two life stages: memories from girlhood in the early years of the 20th Century, and at late adulthood in recent years. If it can be confirmed that participation in school or organized sports as a youth leads to a more physically active adulthood, then appropriate changes in policy should be vigorously pursued. At the moment, however, we have little hard data with which to support such recommendations. (Powell & Dysinger, 1987, p.281) To date, little research has been done to begin to understand the past and present factors which would explain the reluctance of females, from girlhood on, to take up, or keep up, healthful levels of physical recreation, sport and play. This study will begin to explore, in retrospective fashion, possible childhood sources of physical competency for activity in later life. This study is unique for its inclusion of several historical-situational variables: childhood social support, childhood movement confidence for six physically challenging skills, and country of main schooling as a child. The merging of current exercise behaviors and beliefs with perceptions of childhood opportunity and capability for skilled physical activity, promises to generate new  14 understandings about the attitudes and experiences of elderly women toward exercise and how these may be a lifelong consequence of previous experience. The study will provide information that can inform policies concerned with public health and physical education for the elderly. Policy makers and health promoters, however, need specific information on the enabling elements of society that influence whether late life will be vigorously active or relatively sedentary. Physical educators can better design and instruct exercise programs if the attitudes and perceptions of the elderly toward fitness activities are known. Vancouver, with its mild winter climate, provides a particularly appropriate arena for this kind of study. The city is one of the "retirement" headquarters of Canada, and Western Canadians are known to be more active than other Canadians (Stephens, 1988). Therefore the women in this study are thought to be found in the "best of environments" for regular physical activity.  Definition of Key Terms The following definitions will assist the reader in the interpretation of the theoretical framework and for the review of literature.  Efficacy Expectation is the conviction that one can successfully execute the behavior required to produce desired outcomes (Bandura, 1977a,b). Efficacy refers to personal judgements of how well one can organize and implement patterns of behavior in situations that may contain novel, unpredictable, and stressful elements (Bandura & Schunk, 1981). Perceived efficacy can affect one's choice of activities and activity environments. Persons who continue to shun activities  15 out of self-doubts preclude opportunities for skill development and thereby remain inefficacious (Schunk & Carbonari, 1984).  Exercise is a subset of physical activity that is planned, structured, repetitive and has as an objective the improvement or maintenance of physical fitness (Caspersen, Powell & Christenson, 1985). Although "exercise" is used interchangeably with "physical activity," researchers recognize that "it has characteristics that separate it from many other physical activities" (Powell & Paffenbarger, 1985, p.118).  Health is "a human condition with physical, social, and psychological dimensions, each characterized on a continuum with positive and negative poles. Positive health is associated with a capacity to enjoy life and to withstand challenges; health is not merely the absence of disease. Negative health is associated with morbidity and, in the extreme, with mortality" (Bouchard, Shephard, Stephens, Sutton & McPherson, 1990, p.6).  Health Incentive (Motive) is defined as the behavioral incentive, instrumental value, or motivation to participate in health promoting behavior. In terms of older adult exercise participation, motivation to live a long and healthy life was the incentive considered in this study.  Health Locus of Control is defined as one's perceived control over one's health. An external locus of control describes perceptions that a health event is due to chance or the actions of powerful others. Internal locus of control refers to perceptions  that a health event is due to one's own personal actions (Kist-Kline  16 & Lipnickey, 1989; Wallston, Wallston, Kaplan & Maides, 1976; Wallston, Wallston & DeVellis, 1978).  Health Benefits are defined as the degree of perceived advantage or positive health outcomes from participation in physical fitness activity.  Health Risks are defined as the degree of perceived personal harm or negative health outcomes from participation in physical fitness activity.  Movement Confidence is a combination of personal efficacy and personal experience which represent a person's perception of assurance of success in physical activity and performance settings.  Outcome Expectation (Perceived Risk or Benefit) is defined as a person's estimate that a given behavior will lead to certain outcomes (Bandura, 1977a,b). The expected outcomes of exercise participation may be viewed as positive or negative (beneficial or harmful) to health and well-being.  Physical Activity is defined as any bodily movement produced by skeletal muscles which results in energy expenditure. The energy expenditure can be measured in kilocalories (Caspersen, Powell & Christenson, 1985).  Physical Competence describes the perceived mastery of complex motor skills in a particular movement situation as judged by an external authority. Such mastery may require a certain amount of physical fitness, but in addition, skilled  17 movement may require precision, strength, balance and coordination which requires substantial learning and practice.  Physical Fitness is a set of functional attributes that are health- and/or skillrelated which can be measured with specific performance tests (Caspersen et al., 1985). Tests of fitness aim to measure physiological attributes (such as muscular strength, muscular power, cardiovascular endurance or flexibility).  Physical Fitness Activity refers to self-regulated participation in fitnessenhancing exercise activities such as sport, dance, vigorous walking, home or community exercise programs. Optimal benefits are thought to be achieved with regular sweat-inducing participation at least three times per week, and the activity must have the potential to contribute in some way to development or maintenance of aerobic fitness, joint mobility, muscle strength and endurance, posture or balance. For the purposes of this study, housework was not considered in this definition.  Physical Efficacy is the strength of an individual's perceived self-confidence or belief that she or he can successfully complete a physical task through the expression of movement ability (Brody, Hatfield & Spalding, 1988, p.32). See also "self-efficacy to exercise" and "physical competence".  Self-efficacy to Exercise refers to the strength of an individual's perceived self-confidence or belief that one can successfully complete a task through the expression of physical ability (Bandura, 1977a). In some of the literature, •  physical^al  ly i  s referred to as  physical competence" although a distinction  18 should be made that competence in physical situations is often judged by others and is not necessarily a personal judgement. For the purpose of this study selfefficacy to exercise is considered to be a reflection of self-perceived performance ability. Self-efficacy to exercise is potentially mediated by actual performance knowledge and the known judgements of others.  Social Support refers to the endorsement, approval, advocacy or encouragement by significant others of an individual's behavior (in physical activity).  Delimitations The study is limited to women born in 1921 or earlier (presently age 70 or older) who are currently attending community programs in Central Vancouver or who reside within its proximity. The sample is all female, predominantly Caucasian, Canadian educated and middle class.  Limitations Following the example of other prominent studies, the present research was limited to a seven-day recall assessment of leisure-time physical activity, exercise, and sport. Thus the domestic physical activities of women have been omitted from this study. The significant role which physical work in the home may add to the overall physical activity patterns of women was considered at the outset of the study. However, a conscious decision was made that the study would focus on voluntary exercise normally found in leisure time. Thus gardening, but not other domestic work was admitted to the activities assessed in this survey. The reader could argue that women's domestic work is often conducted in their  l eraure  -  time,  and may be considered to be voluntary and enjoyable. In  19 retrospect, I may have erred in assuming that most women view domestic labour to be a work experience, devoid of choice and pleasure. A further limitation exists with the weekly assessment of energy expended on leisure-time or voluntary activity. Adults are known to overestimate their activity levels in self-report situations. Furthermore, the assessment is only an estimate of energy expended on exercise. While the assessment is detailed, the actual calculation of kilocalories is based on approximated MET units for each particular activity reported. A critic of this research would probably take issue with the retrospective data collected on childhood efficacy and childhood social support. These were "recall" measures and thus are vulnerable to memory loss and altered perceptions over the years. However, these measures have demonstrated satisfactory reliability in a pilot study, and although validity is not guaranteed, the perceptions of early efficacy and support are really what matter in providing the  foundation for the attitudes and beliefs of older women toward exercise. Validity for the childhood recall measures was, however, found in this study as discussed in Chapter 6. Geographic community sampling improved the prospects of reaching older women who would be representative of all segments of society. However, non-random sampling is susceptible to selection bias. Less represented in this study were older women who were house-bound as caregivers, or at home with their own physical limitations. Although senior's lodges and residences were represented in the sample, women in institutional settings or hospitals were not accessible to this study. Volunteer subjects who fill out exacting surveys are, in many ways, except luludl indivi duals. They are more likely to be highly educated, at least  20 middle class, in good health and highly mobile. Furthermore, the oldest women in this study represent the survivors of their generation - by longevity alone they have outlived most of their birth cohort, and thus may represent an array of exceptional qualities attributable to genetics, biological resilience, positive coping strategies and other skills for adaptation. Therefore the results of the study can only be generalized to mobile elderly women of metropolitan centers with similar ages and social context.  Organization of the Remaining Chapters The review of literature following next in Chapter 2 presents the key constructs of the main theories which guide the explanation of human behavior and surveys the literature with regard to their application to the explanation of physical activity. From the review of literature, the theoretical model guiding this study is drawn. The theoretical framework integrates key constructs of previous research into a Composite Model and thus a separate chapter is justified and follows immediately in Chapter 3. Chapter 4 explains the methodological design of the study from data collection to statistical analyses. Chapter 5 reports on the descriptive results of the study and provides the outcomes of the regression analyses. Chapter 6 includes a summary and discussion of the findings, identifies policy implications and discusses unanswered questions. The references are displayed in alphabetical order in Chapter 7 followed by appendixed materials as described below. Appendix A contains the survey questionnaire; Appendix B contains the approval forms of the Ethics Committee of the University of British Columbia; Appendix C holds the form used to obtain agency approval, and also instructions  to th e univerbi  y students for oral administration of the questionnaire.  21 Voluntary letters from subjects who offered interesting comments and points of view are found in Appendix D. Appendix E contains post-hoc analyses conducted on the subjects who provided missing data. The residual error of the final regression equation is plotted and discussed in Appendix F. Finally, Appendix G contains a critical review of literature pertaining to the reliability and validity of the outcome variable, "exercise in the past week".  22  II. REVIEW OF THE LITERATURE  Introduction The multidisciplinary nature of this study, as well as the number of variables involved, demands parsimonious selection and careful organization of the relevant research literature. Over 2000 articles were reviewed in the initial planning of this study. Dozens of research papers were found on many of the independent variables; despite this, few studies were found which specifically related to the explanation of exercise behavior of aging adults, namely older women. Therefore this chapter will focus on the literature most pertinent to the study at hand, namely that which is known about the theoretical variables driving the study and their known relationships to the dependent variable, weekly exercise in late life. The chapter begins with a brief overview of the problems facing today's elderly women -- problems which are considered to be outcomes of their past socialization and experiences as younger females. Because of studies which indicate that early activity habits may be maintained throughout the life course, the review explores historical perspectives and the activity socialization of females born before 1921. Next, their contemporary characteristics are presented in the form of a demographic profile of women born before 1921. In this section the following topics are addressed: size of the aging population, marital status, family size and socioeconomic status, education, ethnicity, health status, and institutionalization rates. The profile sets the stage for understanding the women involved in this research by highlighting the significant contextual features  of  their  generation.  23 Next comes a review of the literature about the known exercise and physical activity patterns of older women. Recent findings confirm that in the past ten years, activity patterns of Canadians have actually decreased slightly, and the majority of older adults are still insufficiently active. At the same time, the enormous benefits of exercise have become known, and enough evidence has accumulated to know that the risks of participating in late life exercise are very minimal. A thorough review of the known benefits and risks of exercise participation are discussed in this chapter. Thus the logical conclusion is that significant barriers must be operating which thwart the involvement of older women in health-promoting physical activity. To examine these barriers, the second half of the chapter turns to theoretical explanations for understanding the determinants of late life exercise in women. Two theoretical perspectives which are used to generally guide health behavior research are introduced: 1) social epidemiology and socialization which focus upon the personal characteristics and environmental situation of the individual; and 2) cognitive beliefs of an individual representing a psychobehavioral perspective. The chapter introduces the Health Belief Model, Health Locus of Control Theory, the Theory of Reasoned Action, and finally Social Cognitive Theory. The cognitive determinants of exercise are well-articulated in Social Cognitive Theory, and the research literature related to the self-efficacy construct is strong. The chapter concludes with a focus on the relationship of exercise and the main constructs of Bandura's Social Cognitive Theory: Incentive, Self-Efficacy, Environmental Cues (social support), and Outcome Expectations.  24  The Poor Aging of Women Aging for women is, in many ways, a "survival of the unfittest" (Isaacs, Livingstone & Neville, 1972) because a female's extra life span is usually accompanied by significantly high levels of reported illness (Verbrugge, 1987). Simply too many women are encountering early onset of preventable chronic diseases. Part of the explanation is that women may be more attentive to their symptoms of ill health, may perceive that society accepts, if not expects, them to report health problems, and may find it easier to visit doctors due to flexible work patterns (Waldron, 1982). Others blame the health difficulties of women on their more passive lifestyle - a lifestyle reinforced, in North America, at least, by public policy which limits their participation in health promoting physical activities over much of the life span (Boutilier & SanGiovanni, 1983). The social sanctions imposed on women in sport contexts, for example, have been likened to the marginal social status of the disabled whereby both have been viewed as helpless and are reinforced by society to compound this helplessness (Mastro, Hall & Canabal, 1988). The penalties of lifelong inactivity are ultimately seen in frailty, depression and inability to carry out the simple activities of daily living - factors which guarantee institutionalization for almost half of all women over the age of 85 (Fletcher & Stone, 1982; Gee & Kimball, 1987). In the following section, historical perspectives are presented about the socialization of women born before 1921. The limited socialization of females into vigorous forms of physical activity in the first half of the 20th Century helps to explain why their aging is often accompanied by sedentary lifestyles and health  difficulties.  25  The Activity Socialization of Females Born Before 1921 The social incentives and rewards for participating in vigorous exercise and sport have historically been lacking for girls and women (Csizma, Wittig, & Schurr, 1988; Greendorfer, 1983). The 1909 Board of Education's Syllabus of Physical Exercises for the Public Elementary Schools portrays females demonstrating a number of static postures in the gym. In the same syllabus, boys are depicted climbing ropes, and in more dynamic situations showing movement and strength (Board of Education, 1909). Adding to this lack of curricular support for girls to be as active as boys have been medical notions of female fragility (Vertinsky, 1988), social devaluation and invisibility (Gee & Kimball, 1987). Sport socialization research describes a clear picture of the gender constraints on turn-of-the-century, middle-class females undertaking the more vigorous forms of exercise (Lucas & Smith, 1978; Morrow, Keyes, Simpson, Cosentino & Lappage, 1989; Verbrugge, 1990a). One conspicuous limitation was the heavy, multi-layered and cumbersome attire worn by females of all ages which placed women in greater danger than any physical exertion (Bolotin, 1987, 1980; Heisch, 1988; McCrone, 1988). The corset, for example, was not only uncomfortable, it deformed the ribs, and caused abdominal organs to be permanently displaced. The wasp-waist corset bound the lower rib-cage so tightly and restricted breathing so seriously that women easily fainted. Judged by their clothing, women of this generation were indeed physically limited and helpless. As significant as fashion in dictating the physical abilities and activities of girls and women was their socialization into a particular feminine role -that of mother and caregiver for her family (McPherson, Curtis, & Loy, 1989). Childbearing was extremely important -- a fart which allowed medical authority  26 to have a crucial role to play in prescribing appropriate behaviors and activities for females. Early adolescence was the focus of most medical concern, for this was the stage of maximum female growth for sexual maturation. Doctors believed that all physical energies had to be conserved for the critical development of reproductive maturity (Vertinsky, 1990). The complex physiology of women seemed to overtax the understanding of an all-male medical profession; physicians many times prescribed even more passive behavior for females who were having physical or psychological difficulties with their lifestyles. Males, at all ages, were universally judged to be better physical specimens than females - a phenomenon that seems to have been rooted in their more physically aggressive play patterns in childhood. Tolerance for aggressive behavior, even fighting, was the social context for boys and men, while the labelling of active girls as "tomboys" and "bicycle faces" (Heisch, 1988) was probably only a small part of the larger socio-environmental forces limiting the physical opportunities of females (Dishman & Dunn, 1988; Espenshade, 1969; Gilman, 1911). Tomboy style of play was considered by many to be rude and vulgar (Guttman, 1988). Once into the mothering role, a woman would have scant time and energy to take on sportive or other recreative activity outside the context of her family (Gee, 1987, 1986a, 1986b). Clearly deterrents to be as strong, as fast and as physically able as the average male were operating in the early twentieth century and undermined many young female's motivations to be physically competent and able-bodied in strenuous undertakings. The negative reactions of the female body and mental health to this narrow social role were explained at the time as further evidence of the inferiority of women. For those women who did ignore society's xpectations, L.umpe erIL les were acquired in skilled activities which no doubt  27 enhanced their physical well-being, but which may have taken a psychological toll on their status as they stepped outside of conventional female roles. There are some historical accounts and women still alive to tell us that, by necessity or choice, they did not adopt the passive role that society expected of them. While middle and upper class women were socially constrained, there were many working women who had to labor intensively in order to survive. Photographic evidence exists that show many immigrant women acting as "horse-teams" on the prairies, pulling ploughs when oxen were not available (Bolotin, 1987), while others combed the fields, literally on their hands and knees, at harvest time. Thousands of rural women, mostly immigrants, did intense and difficult work carving out a pioneer existence and building homesteads in western North America. The physical challenge and skill of horse-back riding and ranching cattle were also part of the lifestyle for many of these women. Just over one hundred years ago, young women were admonished for their reckless attempts to learn how to ride bicycles (Lucas & Smith, 1978). By the mid-1890's, a number of city women were riding bicycles for transportation and pleasure (Harmond, 1984). Some women found physical challenge in permissible forms of dance such as ballet training and tap dancing. Social types of dance were highly popular with both men and women and "marathon" dancing was in vogue. Others developed interest in tennis (Danzig, 1928; Heathcote, 1894), swimming, (Shea, 1986), golf (Nickerson, 1987), basketball (Smith, 1984), and figure skating (Cruikshank, 1921). In the early decades of the twentieth century, the more wealthy women were beginning to experience the sport club scene in golf and tennis. While middleclass women were still being socialized into more passive roles than males, at  ti mes,  necesbi y required that working class women contribute resilience,  28 endurance and sweat alongside the physical labour of men. This dialectic between the roles of women according to their social class is perhaps part of cultural answer in explaining the heterogeneity in physical activity which accompanies women's aging. In the next two sections, I present information which describes the present life quality and lifestyle status of today's elderly women. First, a demographic profile touches on the size of the aging female population, followed by a brief examination of their cohort features such as socioeconomic status, marital status, family size, education, ethnicity, health status, and rates of institutionalization. Second, the participation patterns of elderly women in physical activity, exercise and sport are presented. These two sections summarize the  life situation of older women as it is known for the pre-1921 cohort.  A Demographic Profile of Women Born Before 1921 Each generation is accompanied by contextual features which help to make that cohort unique. Reviewing these contextual features may help to set the stage for understanding the physical activity and sport patterns of today's older women in Western Canada. Although one's life circumstances do not remain stable over the lifespan, one's short-term situation is often uncontrollable and irreversible from an individual viewpoint. Thus, the prospects for rapidly improving women's quality of aging by knowing how past and present circumstances create barriers or opportunities for exercise involvement may be limited.  29 Size of the Aging Population Over 1.25 million Canadian women are over the age of 65, and of these, 13% reside in British Columbia. In the 1986 Census, 12.1% of the total population of B.C. was over the age of 65 (Sources, 1991) and 5% were over the age of 75. The Vancouver Metropolitan Area Population Forecast, 1986 - 2011 (1988) predicted that in 1991, 1.5 million people of all ages would reside in Central Metropolitan Vancouver, and of these, 11.5% would be over the age of 70. Almost 6,000 individuals were estimated to be over the age of 90! Canadian women over the age of 65 began to outnumber men in 1961 (Stone & Fletcher, 1980). In Vancouver, there are proportionally more older women in the 65+ population than is found in B.C. or in Canada as an average. Women over the age of 65 in Vancouver now represent about 60% of the total seniors population (Table 2.0). By age 85, women outnumber their male peers by 2:1 and by age 90, by almost 3:1 (Statistics Canada, 1984). The City of Vancouver is home to approximately 28,000 women over the age of 70, or about 1.5% of the general population (City of Vancouver, 1986 Census). The 327 women involved in this study represent over 1% of these 28,000 women.  30 Table 2.0 Demographic Structure of Elderly Women CANADA'  BRITISH COLUMBIA'  VANCOUVER'  65+ FEMALE POPULATION  1,269,440  166,340  98370  65+ MALE POPULATION  1,010,850  131,830  68480  % FEMALE OF 65+ POP.  55.7%  55.8%  59.0%  MARITAL STATUS  Married  41.4%  34.5%  39.0%  Widowed  50.5%  57.1%  48.4%  Single  7.2%  6.4%  6.0%  Divorced  1.1%  2.1%  4.3%  6.0% (1981) 4.6% (1985)  5.4%  4.9%  < Grade 9  41.7%  36.6%  31.0%  Gd. 9 to 12  48.3%  38.6%  41.4%  22.0%  23.7%  2.9%  3.8%  63.7%  66%  German  7.9%  8%  Scandinay.  4.7%  Chinese  2.9%  12%  20.8%  24%  VARIABLE  EMPLOYMENT  EDUCATION  Post-Second. Univ. Deg. ETHNIC STATUS  British  50%  Other SELF-RATED HEALTH  Excellent'  20%  Good  42%  Fair  30%  Poor  8%  2  Profiles Part II. (1988). Census Tracts. Ottawa: Ministry of Supply & Services. Also data came from a special run on Census data from Statistics Canada. 3  Seniors' Resources and Research Society of B.C. (1991). ^Profile of seniors in British Columbia. - 1185 West Georgia St., Vancouver, B.C., V6E 4E6. Also: Ministry of Health. (1988). Disease 9(8), 188-218.  Sources, #503 Surveillance, 4  Profiles - Vancouver Part 2  (1988). Census Tracts. City of Vancouver (1986). Ottawa: Minister of Supply  & Services. 5  Statistics Canada. (1987).  Health and Social Support, 1985.  Ottawa: Minister of Supply& Services Canada.  31 Marital Status, Family Size and Socioeconomic Status Census data indicate that in Vancouver, about two-thirds of women over age 65 are widowed, single or divorced (Table 2.0). After age 85, four out of five Canadian women are widowed (Statistics Canada, The Elderly in Canada, 1984). Statistics Canada reports in Women in Canada (1990) that the most significant group of persons living alone is that composed of females aged 65 and over. Thus by late life, the majority of Canadian women are without partners and many are without pensions: a predicament with serious economic implications for quality of life and life choices. The statistics on the financial status of older women are shocking. The traditional social roles of women have meant that marriage has usually provided a woman a degree of financial security, at least while her spouse was alive. But being widowed, living alone, and living with minimal finances are predictable outcomes for the majority of aging women. In Canada, 60% of women over age 65 are regarded as poor, and 80% of these women are widows. The National Advisory Council on Aging reports that some 43% of seniors received the Guaranteed Income Supplement in 1990, meaning that almost half of Canada's seniors have only a marginal monthly income. Housing represents 29% of the expenses of female seniors compared to 16% of the expenses of the total population (NACA Precis No. 1, 1991). In B.C., 56% of female seniors live in detached family homes while another 33% live in their own apartments. Even when mortgages have been paid off, property taxes and house repairs can be a major financial burden to an older woman living on her own. Even those women who had procured satisfactory employment in their younger years, had often earned the lower wages of women throughout the 20th Century, meaning that poverty often ac  LAJImpanies thei  rdying. More than 20% of Canada's  32 poor are senior citizens who are living on fixed incomes. Yet despite such limited finances, less than five percent of women over age 65 report any current employment income (Statistics Canada, 1986). In 1981, the average income of B.C. females aged 65 to 69 was $8,478 while that for same-age males was $16,802 (Sources, 1991). In 1988, the average annual income of families headed by a person aged 65 and older was $37,462, but only $16,316 for singles aged 65 and over (National Advisory Council on Aging, 1991). In 1986, more than one-third of senior's incomes came from Old Age Security and the Guaranteed Income Supplement. The high poverty rates among older females, no doubt, have some degree of impact on the health and activity behaviors of those affected. Limited financial resources place extra stress on the older woman, and ultimately limits her solutions to health problems that accompany aging. Seniors in the upper-middle income category are more than twice as likely (59%) to rate their health as excellent as those in the very poor category (28%) (Sources, 1991). Family size may have contributed to the financial and health burden of women. Older women tended to raise larger families than women do today. In 1920 to 1924, the U.S. Bureau of the Census reported there were 2,701 children for every 1,000 ever-married women. Although many North American women at the turn of the century remained single and up to 20% of all women were childless (U.S. Bureau of the Census (1975), between 1910 to 1920, 11 to 16% of all women aged 15 to 44 years gave birth in any single year.  33 Education The Canada Fitness Survey states that "by most definitions of active leisure, there is a direct relationship between amount of education and the probability of being active" (Stephens & Craig, 1988, p.4). Recognition of a relationship between education and economic development and of the subsequent improvement of personal and social life, thus provides a further economic argument for a radical change in the organization of education, since education, economic development and improved quality of life are intimately connected. (Cropley, 1977, p.25) Lack of education in Canada is significantly related to poorer health, physical limitations and less happiness. Canadian data reports that about 50% of adults over the age of 70 had less than Grade 9 education while less than 5% received a University degree (Statistics Canada, The Elderly in Canada, 1984). In the 65 and over age group, 34% of those with only an elementary education report fair to poor health, compared to 7% of those with post-secondary education (Health & Welfare, Canada, 1989). According to the 1981 Census, 61.4% of B.C.'s seniors reported completing nine or more years of schooling. This figure is increasing over time (from 50% in 1971) and may have positive implications for the future educational programs seniors will attend. People in B.C. who have less than secondary education are less likely to have plans to improve their health than those with higher education. In Vancouver, over 25% of the women in the age-group studied have postsecondary education and almost 4% have university degrees. But having higher education doesn't guarantee women the same financial status as males. For example, Canadian women aged 65 to 74 with University degrees had an average annual income of $14,500, half that of the same-age and qualified male income of  34 $27,900. Women with some university education had an annual income at the same level as males who had less than Grade 9 (about $9000). This data suggests that education may not be a good substitute variable for socioeconomic status of women. Rather education may be important to examine as lending skills for information seeking about health knowledge, as well as extending opportunities to female students to participate in active recreation and sport further into their adult years.  Ethnicity  In Canada, the predominant ethnic groups among the elderly are of British descent, who make up half of the elderly population, and those of French descent, who account for 25%. These ethnic features reflect Canada's demographic trends over many decades in the past, when birth rates among the French were higher than those of the total population and when immigration of British persons was particularly high. Almost 17% of the Jewish population in Canada are over the age of 65 while the Native peoples over the age of 65 account for less than 4% of their total population (Statistics Canada, 1984). In British Columbia, 64% of the elderly are of British descent with other ethnic groups represented in quite small proportions (<5% each). In Vancouver, English is the mother tongue for 66% of citizens, while French is represented by 1.6%. Almost one third of Vancouver residents speak a non-official language. Chinese is the most prominent non-official language, and is spoken by 43% of the non-official group (Burrard Health Unit, 1990).  35 Health Status General population health practices in Western Canada are considered to be slightly better than in the rest of the country. The B.C. Ministry of Health (1988) reports that B.C. is third behind the Yukon and Alberta for reporting regular exercise (66%, 63% and 61% respectively). Obesity (BMI = 28.6 in women) applied to 14% of B.C. women and 16% of B.C. men. Twenty-one percent of the women were underweight while only six percent of the men were underweight. As with other Canadian provinces, about 90% of British Columbians of all ages perceived their health to be good, very good, or excellent. However, people of low socio-economic level and the elderly were two groups less likely to report good health. Not only was health poorer in these groups but plans to improve health were less common (B.C. Ministry of Health, 1988). While the average person visited a physician five times a year, persons between 65 and 74 make 7.4 visits, and those over 75 make 8.2 visits (Schick, 1982). Functional problems rise after the age of 80 and this is the point where need for support systems greatly increase (Stone & Frenken, 1988). The Health and Activity Limitation Survey (Statistics Canada, 1988) uses the World Health Organization's definition of disability, which is ...any restriction or lack (resulting from an impairment) of ability to perform an activity in the manner or within the range considered normal for a human being." (Statistics Canada, 1988). "  In this survey, disabled persons were defined as those who indicated some difficulty in performing any of the 17 activities, such as "Do you have any trouble walking up and down stairs?" or "Are you limited in the kind or amount of activity you can do because of a long-term emotional, psychological or mental health condition?" Older Canadians rate their health quite positively considering that about 30% of them have activity limitations, and 6% of these  36 limitations are considered to be severe (Health & Welfare, Canada, 1989). In 1987, there were 494,340 disabled males over the age of 65 compared to 727,655 disabled older women in Canada (Statistics Canada, 1988). Older women also rate their health less positively than men. Only 17% of women over age 65 rate their health as excellent compared to 24% of men; 29% of older women rate their health as poor compared to 24% of men; 37% of women report activity limitations compared to 31% of men; 32% of older women say life is fairly or very stressful compared to 27% of men; 9% of older women say they are not too happy compared to 6% of men (Health & Welfare, Canada, 1989).  Institutionalization The proportion of Canadian seniors over age 85 who are residents of institutions has risen since 1976 from 36.5% to 40.5% for women and 25.2 to 28.4% for men (Government of Canada, 1988). Below the age of 85, however, the number of institutionalized adults is declining slightly, and at age 70 to 74, only 3.2% of women and 2.9% of men are living in institutions. About 55% of people confined to nursing homes suffer from chronic mental conditions or senility (Schick, 1986). In British Columbia, 56% of seniors live in houses, 33% live in apartments and 8% of the senior population live in nursing homes, old age homes or chronic care institutions (Sources, 1991). In the age group 75 to 84, the proportion of institutionalized elderly increases to 21% and then becomes as much as 55% of all those over age 85. These numbers will likely continue for some years to come. In the U.S., women comprise 75% of all nursing home residents aged 65 and over (United Nations Office at Vienna, 1990) and yet families are still the lar•est re  I'^III^•^II  •!  37 that the majority of informal caregivers are women, and many appear to be midlife women who are caring for older women. Meta-analysis recently conducted on gender differences in caregiving contradicts this finding (Miller & Cafasso, 1992), but in the majority of studies documented, most caregivers were women with female children claiming caregiver roles in 79% of the cases (Miller, 1990). The family unit as a caregiving resource for the elderly appears to be in jeopardy. In Canada, Fletcher and Stone (1982) claim that the increasing incidence of childlessness or one-child families, the increasing rate of divorce, the high rate of mobility among young adults and recent increases in the labour force participation of women, all point to the probability that an elderly person in the future will have less access for family support than present and previous generations of older persons. This means that more older adults need to be maintaining their strength and mobility in order to live out their remaining years with dignity and independence.  Elderly Women's Exercise Patterns Known exercise patterns of older women In reviewing the literature on the known activity patterns of older women, there are inconsistent findings. Part of the confusion arises because there are different methods used to quantify physical activity and there have been dynamic shifts in participation trends over a period of a few years. While activity levels seem to be generally on the rise since 1981, Stephens (1988) warns that many adult Canadians are unaware of how much exercise is adequate. Most epidemiological literature supports an age-related decline in participation, with women exhibiting less activity and less physical fitness at  38 every age group (Alexander, Ready, & Fougere-Mailey, 1985). While post-adolescent girls demonstrate a greater decline in fitness than at any other time, the next most critical period in terms of physical fitness decline is considered to be age 40 to 49 (Alexander et al., 1985). Over 1200 women across six Canadian regions were interviewed in the 1985 General Social Survey where active physical exercise was defined as "exercise which made one perspire or breathe more heavily than normal" (General Social Survey, 1987, p. 59). From this survey, it was estimated that only 27% of the adult Canadian population over the age of 15 were active enough to anticipate health benefits which may include additional years of life. The survey notes that physical activity declines sharply after age 24, and again after age 44. In addition, among adults over age 65, almost 40% are identified as sedentary. Only 14% of the 65 and over women in this survey were in the "active" category, and 36% were in the "sedentary" category. The report Changing times: Women and physical activity (Fitness and Amateur Sport: Women's Program, 1984) stated that 53% of Canadian females over age 60 were active in their leisure time (an average of at least three hours per week over nine months of the year). Next, 21% of the women were moderately active while 24% were sedentary. Only 39% of the women studied achieved the recommended level of cardiovascular fitness. Therefore, the agency concluded that "women need to increase the intensity of their participation" (p.2). But the report Women in Canada: A Statistical Report (1990) using data from the 1985 General Social Survey conducted a few years earlier, classified only 6.3% of Canadian women over age 65 as "active". The Campbell's Survey on the well-being of Canadians reports that 50% of men and 30% o  ,,111'1^•^-  ivi y  39 for at least 30 minutes every other day (Stephens and Craig, 1990). The report suggests that women are just as active as men in time spent, but do less intense physical activities. The physical recreation activities which have the largest number of participants are walking (77% of all-age women), gardening (55% of allage women) and swimming (48% of all-age women). With increasing age, the number of adults walking and gardening increases, while the number who are cycling, swimming, and dancing decreases. While almost half of women over 65 claim to be exercising regularly (The  Active Health Report on Seniors, 1989), as few as 15 percent of older women are exercising at an intensity that would foster better health (Stephens, 1985). The Campbell's Survey (Stephens & Craig, 1990) notes that there is a general decline in participation with age, but since 1981, adults of all ages, and notably older adults, have made significant efforts to become more physically active. Still the most recent document on the physical activity of Canadians  (Health Promotion, 1993) suggests that little change at all has occurred in the past ten years. The deficient activity patterns of adults of all ages, and the particular reluctance of women to participate in the more vigorous activities, suggests that the benefits of involvement in moderately intense exercise are not well known or are of little consequence to the average citizen and the risks of exercise may appear to be too great. On the contrary, the reality is that the benefits of regular exercise are almost overwhelming to document as can be seen in the quite sizable section following. In addition, the negligible risks of adult participation are also presented.  40  Known Benefits of Exercise Participation Introduction Health, in its broadest sense, is the topic of central interest and is the primary reason for most research on physical activity. (Kohl, Blair, Paffenbarger, Jr., Macera & Kronenfeld, 1988, p. 1229) Evidence is rapidly accumulating that physical mobility has become a survival need for the elderly, and that society must change its attitudes toward older women and their physical capabilities and requirements (Milde, 1988). An array of studies from a number of disciplines provide support for a long list of biopsychosocial benefits from involvement in physical activity throughout the entire life span: increased longevity, improved physical and mental health, as well as independence and improved quality of life in very old age (O'Brien & Vertinsky, 1991). Some seniors appear to be aware of these benefits; 30% of older men and women with heart disease identified lack of exercise as a personal lifestyle problem (Clark, Janz, Becker, et al., 1992). Not only do seniors want to avoid institutionalization, but they also seek opportunities for social involvement and personal growth  (Toward A Better Age,  1989). However, recent research indicates  that health promotion programs for older employees may have an insignificant impact on increasing their exercise behavior, even when the employees have good intentions to become more active. Sharpe & O'Connell (1992) failed to find any increase in exercise behavior of university faculty and staff after one year of health promotion activities such as participation in walking groups, one-to-one counselling, and work-site exercise programs. Predictors of intention to exercise were level of education, gender, self-efficacy, outcome expectancies, perceived barriers, and baseline exercise frequency. However, at the end of the study, only  41 exercise at baseline was predictive of current exercise level. Thus work-based initiatives in promoting activity may not succeed if older adults are not convinced of the personal benefits of exercise, or if other life circumstances interfere in their intentions to exercise.  The Benefits of Exercise By the 1980's, the benefits of exercise for older women were becoming evident in the scientific literature; presently, physical activity is consistently addressed as one of the most significant health interventions in the lives of the elderly. The preventive and restorative benefits of physical activity are recognized by medical and sport science research in most serious health threats. Table 2.1 outlines some of the exercise outcomes that have been found with aging women in relation to exercise intervention research. Although at least half a dozen studies have found no significant change in their dependent measure of exercise, it is because the intensity of exercise, the duration of the training program, or the type of exercise chosen are often inadequate to produce benefits. Overall the evidence seems to favor aerobic activity for both psychological and physiological impact. However, frail elderly women appear to be able to make gains in functional status even with mild mobility programs.  42  Table 2.1 The Known Benefits of Exercise HEALTH BENEFIT  SCIENTIFIC SOURCES  Aerobic fitness, increased maximum oxygen uptake  Badenhop, Cleary, Schell, Fox, & Bartels, 1988; Barry et al., 1966; Blumenthal, Emery, Madden, Schniebolk, Walsh-Riddle, George et al., 1991; Blumenthal, Schocken, Needels, & Hindle, 1982; Cobb & Coleman, 1991; Foster, Hume, Byrnes, Dickinson & Chatfield, 1989; Hopkins, Murrah, Hooger, & Rhodes, 1990; Johannessen, Holly, Lui & Amsterdam, 1986; Kilbom^1971; Suominen, Heikkinen, & Parkatti, 1977; Tonino & Driscoll, 1988.  Management and control of chronic obstructive pulmonary disease  Atkins, Kaplan, Timons, Reinsch, & Lofback, 1984; Swerts, Kretzers, Terpstra-Lindeman, Verstappen, & Wouters, 1990, Webster, 1988.  Prevention and control of heart disease  McCunnev, 1987; Morev et al., 1989; Posner et al., in press; Sedo wick, Taplin, Davidson, & Thomas,^1988; Sidney & Shephard, 1978; Terpstra-Lindeman, Verstappen, & Wouters, 1990; Upton, Hagan, Rosentswieg, & Gettman, 1983.  Control of obesity, lower cholesterol  Blumenthal, et al., 1989; Shepherd, 1986b; Sidney, Shephard, & Harrison, 1977;  Therapeutic value for arthritis  Ellert, 1985; Harcom, Lampman, Banwell, & Castor, 1985; Lewis, 1984; Minor, Hewett, Webel, Anderson, & Kay, 1989.  Incidence of cancer, heightened immune response  Fiatarone, Morley, Bloom, et al., 1989;  Prevention and control of osteoporosis  Block^Smith, Freidlander, & Genant, 1989; Evans & Smith, 1987; Rikli & McManis, 1990; Smith, Reddan, & Smith, 1981; Snow-Harter, 1987; Snow-Harter & Marcus, 1991.  Social support and encouragement for exercise  Andrew et al., 1981; Dishman, 1986; Gray, 1987; Hauoe, 1973; Perusse, LeBlanc, & Bouchard, 1988; McPherson, 1982; Powell, Spain, Christenson, & Mollencamp, 1986; Snyder & Spreitzer, 1973; Spreitzer & Snyder, 1973; Stephens & Craig, 1990; Tait & Dobash, 1986; Wankel, & Berger, 1991; Wechsler, Levine, Idelson, Rohman & Taylor, 1983.  continued ^  43 Table 2.1 continued... The Known Benefits of Exercise HEALTH BENEFIT  SCIENTIFIC SOURCE  Control of hypertension  Barry, Daly, Pruett, et al., 1966; Emes, 1979; Richardson, & Rosenberg, 1989;^Shephard, Corey, & Cox, 1982; Vaccaro, Ostrove, Vandervelden, Goldfarb, & Clarke, 1984; Weber, Barnard & Roy, 1983;  Control and management of diabetes  Cantu, 1982; Shepherd, 1984.  Improved functional status  Sulmar & Wilkinson, 1989.  Cognitive processing speed  Baylor & Spirduso, 1988; Dustman, Ruhling, Russell et al., 1984; Emery, 1991; ^Powell, 1974; Stacey, Kozma, & Stones, 1985.  Balance  Burger, 1989; Hopkins, Murrah, Hoeger, & Rhodes, 1990; Lichenstein, Shields, Shiavi, & Burger, 1989; Rikli, & Edwards, 1991.  Positive self-concept, improved body image, and control of depression  Bolla-Wilson & Bleeker, 1989; Mittleman Crawford, Holliday, Gutman, & Bhakthan, 1989; Perri, II, & Templar, 1984-85;^Sidney & Shephard, 1976.  Joint mobility  Karl, 1982; Morey, Cowper, Feusner, et al., 1898.  Muscular strength and endurance  Brown, & Holloszy, 1991; Gueldner & Spredley, 1988; Meredith, Frontera, Fisher, et al., 1989; Naso, Carner, Blankfort-Doyle, et al., 1990; Petersen, Petersen, Raymond, et al., 1991; Sinaki & Grubbs, 1989.  Gait velocity  Judge, Underwood, Gennosa, 1991.  Posture and increased height  Ball, VanderVeen, Johnson, & Lukert, 1991.  No benefit or change found  In balance (Emes, 1979; Clarke, Wade, Massey, & VanDyke, 1975); In cognitive speed and aerobic fitness (Molloy, Richardson, & Crilly, 1988); In physical and psychological measrues (Blankfort-Doyle, Waxman, Coughey et al., 1989; In bone mineral content (Nelson, Fisher„ Dilmarian, et al., 1991); In hematological, anthropometric and metabolic comparisons (Nieman, Pover, Segebartt et al., 1990.  ,  44 Research studies have begun to develop a profile of those elderly women who are physically active. Whether she has been active all her life, or is a recent convert to exercise, the physically active woman is likely to be one or two decades younger physiologically than her sedentary contemporary (Drinkwater, 1988). Master's athletes in their seventies can match performances of sedentary 20 year old individuals (Vaccaro, Dummer & Clarke, 1981). In brief, knowledge is available about two types of benefits to be expected: first, within weeks, short term enhancement of physical, social and emotional well-being; and second, over years, long term contributions to prolonged good health, resistance to illness, optimization of self-care and functional independence, reduced mortality risk and overall increased quality of life.  Immediate Benefits Many health and fitness benefits of regular exercise participation are felt immediately by the older adult. For example, those who are new to physical activity often report "feeling better" right away (Fitness and Aging, 1982; Feel  Better, 1980). Such feelings include the perception of doing something good for oneself (Dowall, Bolter, Flett & Kammann, 1988) as well as a sense of achievement (Lutter, Merrick, Steffen, Jones, & Slavin, 1985). Participants entering supervised programs generally find themselves in a social group with others of their age and the potential then exists to widen their social network. Social support and interaction are thought to be among the most important factors in adherence and enjoyment in activity programs, although research is lacking in this regard (Lee & Markides, 1990; Wakat & Odom, 1982). A quite strenuous seniors' 100 day cycling tour under the scientific scrutiny of Mittleman,  Crawford, Holliday, Gutman and Bhaktan (1989) pruved instead to be a test of the  45 social relations of older adults with 6 of the 33 cyclists dropping out only one week from the conclusion of the tour. Feelings toward others and social intolerance were cited as reasons. More study is obviously needed to determine the social benefits and social risks of exercise in a variety of contexts. A number of inter-related psychosocial and physiological parameters have been positively linked with short-term exercise participation. Possibly among the most important outcome of physical activity is stress reduction (DeVries, 1975) since coping with stress is probably linked to other benefits such as better sleep (Griffin & Trinder, 1978; Osis, 1986), muscle relaxation (Berger, 1989), positive mood states (Bolla-Wilson & Bleeker, 1989; Monahan, 1986), improved self-image (Paige, 1987) and self-concept (Perri & Templar, 1984-85). Overall, exercise appears to act as a buffer in many stress-illness relationships (Eichner, 1987; Eisdorfer & Wilkie, 1977) possibly through biochemical interactions linking mind and body (Haug, Ford & Sheafor, 1985). For example, increased levels of beta-endorphins accompanying high-intensity exercise may explain some individual's enhanced perceptions of increased coping and relaxation (DeVries, 1981). However, it is doubtful that many older women exercise at this level of intensity. Exercising individuals demonstrate higher levels of self-efficacy (Atkins, Kaplan, Reinsch, & Lofback, 1984; Hogan & Santomier, 1984), internal locus of control (Perri & Templar, 1984-85) or sense of life control (Rodin, 1986). No less important to psychological health is the opportunity created by exercise to socialize, to play and have fun with peers, form new friendships and develop a community spirit with other elderly (Eckert, 1986; Langlie, 1977; Wakat & Odom, 1982).  46 Possibly the most significant short-term benefit for females of all ages is the potential to gain same day improvement in joint mobility (Burgess, 1992; Hartley-O'Brien, 1980). Mobilizing and lubricating major joints of the body through a variety of stretching and relaxation regimes seems to have great therapeutic merit by contributing to better motor control and dynamic balance (Manchester, Woollacott, Zederbauer-Hylton & Marin, 1989). However, joint mobilization, on its own, has not been researched for its potential in easing the strain of everyday living.  Long-Term Benefits The broader health benefits of regular physical activity can be realized months or years later (Suominen, Heikkinen & Parkatti, 1977). Rikli and Edwards (1991) found significant improvements in motor function and cognitive processing speed continuing throughout a three year period of exercise in older women, while declines were evident in a non-exercising control group. Bortz (1980) claims that there is no medicine that can compete with the range of pathology for which exercise has been prescribed: obesity, depression, diabetes, arthritis, hypertension, coronary heart disease, menstrual cramps, migraine, smoking cessation and many other states. The most tantalizing prospect is the ultimate extension of life and several new studies support the prospect of reduced mortality, at least for males, stemming from regular participation in physical activity (Blair, Kohl,  III,  Paffenbarger, Clark, Cooper & Gibbons, 1989; Grand, Grosclaude, Bocquet, Pous, & Albarede, 1990; Kaplan, Seeman, Cohen, Knudsen, & Guralnick, 1987; Karvonen, Klemola, Virkajarvi & Kellonen, 1974; Linsted, Tonstad, & Kusma, 1991;  rger, Hyde, W iry,  & Hsled, 1986;  vattenbarger & Hale, 1975; Paffenbarger,  47 Wing, & Hyde, 1978). Reuben, Siu, and Kimpau (1992) have found that measures of physical performance can predict mortality over a two-year period. Rakowski and Mor (1992) recently reported that less activity/ exercise was associated with a higher risk of mortality for each of four questions relating to activity compared to their peers, having a regular exercise routine, getting enough exercise, and days walking a mile per week. For individuals with one or more impairments in the activities of daily living, walking was associated with lower mortality. This investigation supports literature on the importance of maintaining physical activity into older adulthood, and suggests that clinicians should attend to reports of activity level by their patients as one of the broader psychosocial domains of patient care. (Rakowski & Mor, 1992, p. M122) Added to the rapidly accumulating mortality data is an extensive list of claimed long-term benefits of exercise such as postponement of cardiac diagnoses (Marti, Pekkanen, Nissenen, Ketola, Kivela, Punsar, & Karvonen, 1989; Posner, Gorman, Prouty Sands, Gitlin, Kleban, Windsor & Shaw, in press); reduced mortality risk through mediating factors such as effective weight control (Evans & Meredith, 1989; Upton, Hagan, Rosenswieg & Gettman, 1983) and not smoking (Heydon & Fodor, 1988); lower blood pressure (Adams & deVries, 1973; Vaccaro, Ostrove, Vandervelden, Goldfarb & Clarke, 1984; Weber, Barnard & Roy, 1983); lowered cholesterol levels (Evans & Meredith, 1989; Blumenthal, Emery, Madden, George, Coleman, Riddle, McKee, Reasoner & Williams, 1989); reduced risk of colon cancer (Gerhardsson, Norell, Kiviranta, Pederson, & Ahlbom, 1986); and increased aerobic fitness (DeVries, 1979; Buskirk & Hodgson, 1987; Seals, Hagberg, Hurley, Ehsani & Hollowszy, 1984). Some have argued that the prospects for life extension are effectively very small (Heyden & Fodor, 1988; Waterbor, Cole, Delzell, & Andjelkovich, 1988) and others claim that life span changes are affected more by genetics  (Johnson,  1988), or environment (Bourliere, 1973), or cultural factors (Waldron, 1976). At  48 least one animal study has revealed a possible age-threshold for mortality benefits; exercise initiated in older rats was found to actually reduce their survival rates (Edington, Cosmas, & McCafferty. 1972). Palmore (1989) reviewed the evidence and suggested that healthier people do have reduced mortality, and are naturally bound to be more active; hence reduced mortality may just be an association with healthier individuals who self-select for physical activity. A nine- to twelve-month exercise program of walking and/or jogging at 80% of maximal heart rate improved fat distribution patterns in 60 to 70-year-old men and women (Kohrt, Obert, & Holloszy, 1992). Older adults lost 3 to 4% of their body weight over the course of the intervention, all of the weight lost was fat weight, and furthermore, the fat lost occurred in the truncal area indicating "a preferential loss of fat from the central regions of the body (Kohrt, et al., 1992, p. M99). This study provides evidence that one mechanism by which exercise operates to reduce risk of disease may be the control of abdominal obesity. Evidence is accumulating that physically fit elderly adults experience less profound declines in cognitive performance than their less-fit contemporaries (Chodzko-Zajko, 1991). Other significant findings accompanying exercise participation include quicker reaction time (Baylor & Spirduso, 1988; Rikli & Edwards, 1991; Spirduso, 1975, 1980), improved joint flexibility (Frekany & Leslie, 1978; Munns, 1978; Rikli & Edwards, 1991), muscular strength and endurance (Rikli & Edwards, 1991; Shephard, 1978; Work, 1989), increased muscle mass (Meredith, Frontera, Fisher, Hughes, Herland, Edwards & Evans, 1989), and the retardation of osteoporosis (Oyster, Morton & Linnell, 1984; Sidney, Shephard & Harrison, 1977) or even increased bone mineralization (Blumenthal et al., 1989; Rikli & McManis, 1990; Smith, Reddan & Smith, 1981) after only weeks of exercise (Beverly, Rider. Evans & cmitb, 1987).  49 A balanced program of nutrition, exercise, and stress reduction appear to benefit all postmenopausal women (Davidson, 1986). A recent study of 3,110 retired Florida residents, average age of 73, concluded that walking one mile at least three times per week offered protection from bone fractures (Sorock, Bush, Golden, Fried, Breuer & Hale, 1986). Peterson and associates recently found that 59 healthy women, ages 36 to 67, increased muscular strength over 12 months, but not bone mass (Peterson, Peterson, Raymond, Gilligan, Chechovich & Smith, 1991). Aerobic exercise intervention, more than a basic calisthenics program, appears to be important in enhancing the cognitive skills of older adults. While research supports short-term neurophysiological improvements in measures of memory, intelligence and cognitive speed (Stacey, Kozma & Stones, 1985), a nonaerobic three month exercise program was ineffective in elevating neuropsychological attributes in elderly institutionalized women (Molloy, Delaquerriere Richardson, & Crilly, 1988). Thus the level of intensity of an exercise program does appear to matter, but not whether the program is conducted at home or in a supervised setting (Miller, Haskell, Berra, & DeBusk, 1984). Aerobic efforts may not have the expected aerobic effect in very old age. In women averaging over 80 years of age, active women were walking briskly over 100 minutes per week, while an inactive group was averaging about 5 minutes of walking per week. Exercise histories showed that the active group had followed their present exercise program of walking for an average of 28 years. These differences in activity level did not translate into significant differences in aerobic fitness in old-old age, a finding that is partially explained by inadequate sample sizes (Nieman, Pover, Segebartt, Arabatzis, Johnson & Dietrich, 1990). The active women had higher aerobic capacities, less body fat, and lower -4^-  is ica signs icance was  50 not reached. However in other studies on cardiovascular responses with small sample sizes, the exercise effects have been powerful enough to obtain 30% increases in aerobic capacity (Seals, Hagberg, Hurley, Ehsani, & Holloszy, 1984). Most studies find that physical conditioning has significant positive effects on the older adult. Adults suffering from osteoarthritis and rheumatoid arthritis can also obtain important improvement in aerobic capacity, walking time, depression, anxiety, and increased habitual activity after a 12 week walking or aquatic exercise program (Minor, Hewett, Webel, Anderson, & Kay, 1989). Researchers are exploring the possibility that exercise might prevent falls by improving balance, but so far, the results are not promising (Emes, 1979). Twelve weeks of light physical activity is not apparently long enough, or intense enough to reverse losses in proprioceptive function. A prospective study attempted to reduce falls and injury in the elderly using stand-up exercises from sitting in a chair, and step-ups onto a 6 inch high stool (Reinsch, MacRae, Lachenbruch, & Tobis, 1992). The researchers concluded that the exercise program had merit, but was too light in intensity to reduce falls significantly. Many of the activities which seem to appeal to the interests of women are also lacking adequate intensity and vigor. While old-timer hockey, golf and slowpitch are attracting large numbers of middle-aged and older men, the pursuit of skilled exercise and team sport activities do not seem to be favoured by many elderly women; rather individual, expressive, socially cooperative and self-paced activities are more popular offerings of community programs for older women (e.g. Tai Chi, Yoga, Line dance, Keep fit, Aquacise). Older women seem to prefer supervised and low skill programs close to home. An important activity that is fitnecc  enhancing, age appropriate, suitable tu both meh and women and selt-paced  51 is walking, and of course, gardening. Not surprisingly, walking is the most popular form of exercise with older adults (Fitness and Aging, 1982). Despite the remarkable benefits just outlined, the declining physical activity involvement as women get older suggests that there must be significant barriers operating which undermine their participation. The fact that elderly women are rarely, if ever, seen in vigorous activities such as running and team sports, suggests that older women avoid high exertion settings. In this next section, the known risks of exercise involvement are explored as possible barriers for older women.  Known Risks of Exercise Participation Introduction Although the real risks of engaging in physical activity are becoming known, risk assessment is often strictly based on personal beliefs, and not always is attached to personal experience. While susceptibility to risk of illness and disease in general is under-rated by most people (Weinstein, 1984), and furthermore is likely very much under-rated as an outcome of insufficient exercise, perceptions about susceptibility for harm from exercise participation is common (Heitmann, 1982; Monahan, 1986; Waller, B., 1985). Conrad (1976) observed that the elderly tend to exaggerate the risks of exercise, overestimate the benefits of irregular physical activity, underestimate their exercise capabilities and believe that the need for regular exercise decreases with age. Women, especially, appear to have greater fears about exercise risk even though incidents of sudden death are almost universally a male phenomenon (Ragosta, Crabtree, Sturner & Thomason  •:,  .^111 41 11..  52 influenced by the memory of past events and the imagination of future events. There is even the possibility that simply reminding people that exercise prevents CV disease may alert them to their personal vulnerability to heart attack and make them even more cautious (Slovic, 1986)! Further discussion about the perceived risks of exercise is presented in Chapter 3. In the following section, the known risks of sudden death, injury and ill-health are presented.  Known Risk of Sudden Death in the Elderly Serious complications in supervised exercise programs for even heart disease patients are rare. Van Camp and Peterson (1986) observed one fatality per 750,000 patient hours of supervised exercise and nine cardiac arrests per million patient hours of exercise - a mortality figure no different than would be expected in non-exercising patients in this age group. Cardiac risk assessment in the elderly is considered to be essential protocol in identifying "silent ischemia" (Smith, 1988) and regular exercise is recommended in order to provide older adults with early warning of symptoms related to disease and deficiency in the cardiovascular system (Gottlieb & Gerstenblith, 1988). There is no medical evidence that physical activity - even strenuous exercise - is harmful to the healthy cardiovascular system. However a person with structural cardiovascular disease, even if asymptomatic, is at an increased risk for sudden death during vigorous forms of physical activity and therefore supervised and graded exercise for these individuals becomes mandatory (Van Camp, 1988). Submaximal and maximal exercise stress assessment permits a differentiation of changes in heart rhythm, heart rate, systolic blood pressure  53 and ECG manifestations if myocardial ischemia is present (Bruce & McDonough, 1969). Van Camp and Petersen (1986) obtained data from 167 randomly selected cardiac rehabilitation programs via mailed questionnaires reporting on over 50,000 cardiac patients and over 2 million hours of exercise between 1980 and 1984. Twenty-one cardiac arrests (18 in which the patient was successfully resuscitated and three fatal) and eight nonfatal myocardial infarctions were reported. The 1.3 fatalities per 784,0000 patient-hours of exercise was considered to be a normal mortality rate. The findings suggested that supervised programming along with heart-rate monitoring provided low risk health promoting exercise opportunities for cardiac patients. Ewart and Taylor (1985) claim that the biggest barrier to recovery in individuals experiencing a cardiac event is an unrealistic fear arising from inaccurate self-perceptions about one's physical abilities. Nearly half of the men under 70 years of age who survive three weeks after a myocardial infarction are physically capable of resuming their normal activities within 12 weeks of the acute event. Many of these individuals become physically over-cautious while a few become overzealous and are apt to exercise too strenuously. These researchers make a case for the role of self-efficacy assessment in identifying individuals who are deficient in perceived competence as well as those who are over-confident in their self-perceived ability to resume normal activity levels. McKelvie (1986) claims there is no medical evidence that super-marathon type running protects one from coronary heart disease; rather, there is evidence that extreme endurance events, especially in middle age, place people with unknown problems at increased risk. People add to their own risk by ignoring symptoms of vague or definite chest pain during activity - warning signs that should be  54 immediately addressed in order to reap the benefits of early detection. As population age increases, progressively larger numbers of aged adults will be affected by serious pathologies where vigorous exercise is contraindicated. In such cases a reconditioning program at a steady heart rate of 100 to 120 beats per minute provides an effective stimulus for senior citizens, meaning that for older individuals new to exercise, "walking, recreational swimming, dancing, lawn bowling and even chair exercises have training value" (Shephard, 1986a, p.227). Shephard summarizes the available evidence by stating: Given the strong probability that moderate, progressive activity improves the quality of life, such findings are no reason to prohibit physical activity in an asymptomatic senior citizen. (Shephard, 1986a, p.227)  Known Risk of Injury in the Elderly To consider "exercise" a health-promoting practice already implies that we have prejudged its benefits as outweighing its risks. Indeed, tens of millions of people have already made this judgement, as they regularly engage in a variety of aerobic exercise activities. However, if health professionals are to promote exercise objectively, they need to provide consumers with a balanced view". (Koplan, Siscovick & Goldbaum, 1985, p.190) While risk of musculoskeletal injury may be high, especially in sport activities requiring training for competition, such risk may not exist at other activity levels. Kavanagh and Shephard (1978) found that in the early months of geriatric exercise programs, about 50% of the participants had encountered muscular injury. Yet, a ten week, five day-a-week aerobic program of brisk walking and jogging elicited not a single injury in sedentary middle-aged women according to Johannessen, Holly, Lui, and Amsterdam (1986). Koplan, Siscovick & Goldbaum (1985) suggest that the risks of exercise injury are linked to the specific characteristics of the type of activity  intensit y, duration acid 8eyuenLy)  dS  well as the attributes of the participant  55 and the exercise environment. In this respect, incidence information is unavailable on the actual specific risks of exercise to specific individuals in specific situations. In summary, there is little data on adults of any age group with which to objectify and quantify specific exercise risk. Moreover the long-term effects of exercise, such as relationships to osteoarthritis, do not exist. Adding to the problem of insufficient short and long-term data are inadequate definitions. Comparisons across the exercise research on risk require standard definitions of injury, of the characteristics of various groups of participants, of the nonparticipants (or drop-outs), of the time interval or the main purpose of the participation etc. (Koplan et al., 1985). Without longitudinal and randomized cohort studies, such challenges may be difficult to overcome. In summary, the literature advises the pursuit of moderate exercise in a variety of physical activities allowing for adequate recovery, and this makes good sense until more is known about incidence relationships (Bruce, 1984; Munnings, 1988).  Risk of Over-Exertion or Exhaustion "In those callings that require great physical qualifications, old age is decisive." (de Beauvoir, The Coming of Age, 1972:385) Clearly one has to be realistic about the aging process and admit that physical declines are part of the natural aging process no matter how heroic one is in terms of disciplined participation in fitness activity. With even the most ambitious older adult, declining activity levels are likely to accompany old age as one "often sinks into physical weariness, general fatigue and indifference" (de Beauvoir, 1972, p.404). Moreover, fitness for the future is no longer important as "our eager spring towards the future is broken" (de Beauvoir, 1972, p.404). Cooperating with one's finitude may in the long run be the more healthy  56 route, and acceptance to adapt where necessary is better than acceptance of helplessness. The only general statement that can be made about older adults is that one cannot generalize!  Risk of Provoking Ill Health Each year, one in five aged women suffer an injury requiring medical attention and their injury rate surpasses that of elderly men. Moreover, older women are more often ill and experience more days of bed-disability than men. The bones of elderly women are more vulnerable to fracture because they are smaller and more porous, and while exercise is certain to help maintain bone strength, activity itself is an inherent risk. Older women are twice as likely as men to have arthritis and 13 percent of elderly women are limited by arthritis alone in their daily activities (Haug, Ford & Sheafor, 1985). But exercise is not considered to be the threat to this disease as much as is the lack of it. In fact, Burckhardt (1988) reports that passive recreation pursuits are significantly more important to women with arthritis; she notes that particularly for the over 70 year old women, quality of life and dissatisfaction with active recreation and personal fulfilment in general were significantly affected. Musculoskeletal fatigue, soreness, joint stiffness and delayed recovery is particularly a risk for those who are unaccustomed to exercise and who do not initiate exercise in very gradual and low intensity stages (Kasper, 1990). The first experience with an exercise program for an elderly women may turn out to be a painful experience, and one that is soon learned to be avoided.  57 Summary To this point in the chapter, the health difficulties and poor aging of women have been addressed as likely outcomes of the past socialization and present lifestyle of women born before 1921. The benefits of exercise have been well-documented, while the known risks of participation appear to be almost insignificant. The reluctance of so many older women to reap the many benefits from physical activity demands explanation, and this explanation is explored in the coming section presenting prominent theoretical approaches.  The Determinants of Exercise Behavior: Useful Theoretical Approaches Introduction to Determinants The known determinants of physical activity can be categorized as past and present personal attributes, past and present environments, and physical activity itself (Dishman, 1990). Dishman (1990) uses the term  determinant to "denote a  reproducible association or predictive relationship other than cause and effect"(p. 78).^Personal attributes are defined as demographic variables, biomedical^status, past and present behaviors, activity history, and psychological states and traits associated with physical activity. Dishman claims that determinants residing or originating in the individual are important because they can identify personal variables or population segments that may be targets for interventions to increase physical activity, or conversely, can describe impediments or people resistive to physical activity interventions. However,  summarizing the associations of personal variables and exercise behavior poses a challenge.  58 The absence of uniform standards for defining and assessing physical activity and its determinants and the diversity of the variables, population segments, time periods, and settings sampled in published studies make it difficult to interpret and compare results. (Dishman, 1990, p.78) The complexity of predicting a health behavior such as physical activity has been aptly described as "a web of causation" (Sallis, & Hovell, 1990; Thomas, 1984). Addressing this complexity are two major theoretical perspectives, representing two different research approaches: 1) social epidemiology, and 2) behavioral health psychology. An understanding of these two approaches is essential to the study of exercise behavior and theory development because: ...the degree to which the true origin of the determinants resides in the person or the environment remains to be determined. (Dishman, 1990, p.84) In the present study, personal attributes have been divided into two categories: 1)  "situational" determinants, or the personal and socio-environmental  circumstances of each older woman representing a social epidemiological perspective, and ... 2) "cognitive" determinants, or the self-referent beliefs of the older woman, representing a psychobehavioral perspective. The material immediately following introduces social epidemiology and socialization theory as perspectives which attend to the personal characteristics and environmental situation of the individual. The situational characteristics reviewed are age, education, health status, marital status, family size, work role in mid-life, socioeconomic status, ethnicity, childhood socialization and childhood movement confidence.  59  SOCIAL EPIDEMIOLOGICAL PERSPECTIVES Social Epidemiology The social epidemiological perspective focuses upon personal characteristics and the environmental situation of the individual (Berkman, 1980). The personal qualities and social circumstances of individuals have important associations with health and activity behavior and thus are considered to useful predictors (Belloc & Breslow, 1972; Berkman & Breslow, 1983; Dishman, 1989). Such predictors infer that people are not the creators of their behaviors, but rather are victims of circumstance. Bandura (1989) calls this "environmental determinism" or the study of human behavior in terms of "mechanical agency" (Bandura, 1989, p. 1175). In this view, internal events are mainly products of external ones devoid of any causal efficacy on the part of individuals. For example, Sidney, Niinimaa and Shephard (1983) found that both active and inactive senior citizens had equally positive attitudes toward physical activity. They wondered why there were discrepancies between attitudes and behavior, and concluded that "there must be other factors, perhaps more important than attitudes, which influence behaviour" (p. 207). As another example, one's life occupation can alter possibilities for active behavior later on. Svanborg (1988) reports on a longitudinal study that found previously sedentary workers were more disabled in activities of daily living than those whose work had been strenuous. At the level of the individual, personal attributes and life situations are not easily altered, and therefore are not entirely suitable to social intervention and health promotion. Even so, epidemiological approaches are useful because they identify specific social groups that can be targeted for particular assistance. Theoretically, however, demographic variahlps nn their  own, are  60 deficient because they only provide association, not explanation. Once descriptive associations are found, however, hypotheses can be developed to explain the findings which then can be tested in further research. For example, if never-married women are found to be significantly more active in late life, hypotheses would then be generated and tested as explanations for this finding. McPherson (1986a) emphasized that demographic characteristics can interact or confound one another. For example, intracohort (age) differences, which could explain activity involvement, can vary dramatically by education, marital status, health status, economic status, degree of mobility, employment status, and social network. Nixon II, (1990) recently claimed that sport socialization research "needs to address these issues of 'contextualization'" (p. 35). In Sweden, understanding contextualization was essential to the development of a comprehensive health promotion intervention of 1200 adults over age 70 (Eriksson, Mellstrom, & Svanborg, 1987). For the Swedish study, a "life-style" hypothesis was proposed: that the kind of everyday life led by the individual had consequences not only for social performance but also for functional well-being. We are only beginning to understand why some people are physically active and others are not. The behavior is determined, at least in part, by characteristics of the person, the environment, and the activity itself. (Powell & Paffenbarger, 1985, p. 120) Ten life situational variables appear to have significance as potential determinants for late life exercise. These ten variables are reviewed in the coming section for their relevance in explaining late life exercise.  Age and Exercise Age is not just a chronological variable but also a social construct that defines social behaviour at specific points in the life cycle. Age is an important form of social differentiation that can result in social inequality because of ageism. (McPherson, 1984, p.223)  61 There is a well-documented and universal pattern of declining physical activity and sport participation by age, especially in the early twenties and again after age 65. These sharp decreases in participation have been tied to two major life events - leaving highschool or entering the work force, and leaving the work force (McPherson, 1984). The pattern of declining involvement with age appears to be more pronounced among the less educated, those with lower incomes, those in rural and smaller communities, among females and blue collar workers and among those who live in countries where sport participation is not highly valued or promoted (Stephens & Craig, 1990). McPherson (1978) makes the point that, after peak performance age in many sports, incentives are lacking for adult participation. Facilities and coaching time tend to be allocated to high performance children. With only young role models present in a sport, adults readily assume that these sports events are for the young. The most recent Canadian data states that 42% of men over 65+ and 23% of women over age 65 are active (spending 3+ kilocalories per kg. per day on exercise) which exceeds that of middle-aged groups (Stephens & Craig, 1990). In the age group of 45-64, only 30% of men and 20% of women are classified as active. Adults are at their most sedentary in the years just approaching retirement (Stephens & Craig, 1990). Little is known how physical abilities and skilled motor patterns actually deteriorate over the years, although disuse, muscle wasting, muscle deactivation and neural decline are thought to be inter-related (Shephard, 1989a; Smith & Serfass, 1981). There is evidence that simple daily motor patterns of adults over the life course are developmentally altered in adaptation to age-related change  62 (Van Sant, 1989). More likely, however, individual behavior changes, such as reducing the more vigorous activities with age, are just as responsible for developmental changes as are actual maturational processes.  Education and Exercise The phenomenon that Canadians are less physically active as they age may be determined, in part, by lower levels of education (Rudman, 1986b). Education has strong associations with physical activity level in both free-living and supervised exercise settings (Dishman, 1990). Many older individuals had only a few years of schooling and thus may lack knowledge or habits related to physical activity in public settings. For example, in 1911, only 80% of those aged ten to fourteen were attending Canadian schools (Harrigan, 1990). Boys and girls attended about equally, although young people often withdrew in the their teens to work (Harrigan, 1990). Today, about 60% of Canadians over the age of 65 claim to have Grade 9 or better education. The Canada Fitness Survey (Stephens & Craig, 1990) reports that 52% of adults with incomplete highschool education are inactive compared to 33% of those with a university degree. Furthermore, 89% of university educated adults, 65 years and older, spend over 3 hours per week on physical activity in their leisure-time compared to 71% of those who did not complete highschool. The interest in advanced education and participation in sports activities also increase with economic status and level of education. This makes the following very clear: The course for successful ageing is set predominantly in childhood and youth. (Meusel, 1991, p.16) If this is true, then the challenges of activating individuals in adulthood will continue for some time. Recent studies reveal that contemporary lifestyles of children are predominantly sedentary (Simons-Morton, O'Hara, Parcel, Huang,  63 Baranowski, & Wilson, 1990) and yet hyperactive behavior in school settings is believed to be the most common problem referred to child-guidance clinics in the United States (Alexander, 1990). Baecke, Burema and Frijters (1982) reported highly significant relationships between level of education and leisure-time physical activity in younger males (r = .38; p < .001). Godin & Shephard (1986) examined psychosocial factors influencing intentions to exercise in a group of individuals ranging from 45 to 74 years of age. Education influenced intention to exercise by interacting with "subjective norm," a construct representing a subject's perceptions about social expectations. Less educated subjects were influenced by social norms, and more educated people tended to exercise independently of external influences.  Subjective Health and Exercise Self-assessed health is considered perhaps the most important variable likely to explain late life exercise behavior. Individuals may simply not feel well enough to exercise. Yet the process by which a person comes to understand and evaluate personal health is, in itself, poorly understood. Self-rating of health is thought to be a multidimensional construct which encompasses a global sense of well-being (Zautra & Hempel, 1984). There is surprising statistical support for such a simple and subjective scale. Maddox and Douglas (1973) found "self- and physician-ratings of health are predominantly congruous" (p.59). In fact several studies have found subjective ratings of health to be superior to objective measures of health in terms of predicting well-being, happiness, morale, and life satisfaction (Cockerham, Sharp, & Wilcox, 1983; Zautra & Hempel,  1984).  64 Mossey and Shapiro (1982) followed over 3500 randomly selected Manitoban residents aged 65 and over and found a risk of early mortality almost three times greater for individuals who had rated their health as "poor" only two years earlier. Idler and Kasl (1991) studied mortality in over 2800 older adults with the mean age for females being 74.9 years. About 12% of the women rated their health as "excellent", 46.5% as "good", 33.7% fair, and 8.4% as "bad" or "poor". At the four year follow-up, Idler and Kasl found that "the odds of death increased at every lower level of self-evaluation of health" (Idler & Kasl, 1991, p. S60). Women who ranked their health as "bad" or "poor" were over three times more likely to die within the four year period as were women who had rated their health as "excellent". Idler and Kasl (1991, p. S64) concluded, The knowledge that expressions of subjective health status are sensitive indicators of survival length should engender new respect among health professionals for what people, especially the elderly people they treat, are saying about their health. Larson (1978) suggested that while physician ratings should provide the most objective evidence of the severity of illness in absolute terms, they may not accurately reflect the extent to which an individual's physical condition is actually debilitating. Furthermore, there is evidence to suggest that while physician's ratings may be age-biased in favour of younger adults, the "old-old" category (75+) of adults have been found to rate their health more positively than "young-old" groups (Ferraro, 1980). Gender bias and age bias may be operating since older groups are predominantly women, and older adults may have reduced expectations for optimal health. Thus deteriorating health may not necessarily be reflected in the subjective self-rating. A majority of inactive elderly adults perceive themselves to have good to excellent health and do not believe that they need more exercise (Gunter &  65 Kolanowski, 1986). These beliefs persist in spite of the fact that the prevalence of illness and disability increases with age and is significantly greater for women than for men (Vallbona & Baker, 1984). Charette (1988) states that only 25% of inactive Canadians actually have an activity limitation, and of these, more than half think that exercise will improve their health either moderately or a great deal, regardless of their activity limitation status. What can explain why older adults, who are physically inactive, and who state that exercise could improve their health, are still inactive anyway? In most of the literature relating late life exercise behaviors to health outcomes, positive relationships are found. Exercise participation is related to better health, and better health is associated with increased levels of physical activity. The difficulty in the interpretation is which comes first? Do people exercise more because they have better health to start with, or do people who exercise actually create and/or perceive, better health? The problem of causality is partially answered in large population demographic studies such as that of Belloc and Breslow (1972) and the longitudinal study on college males by Paffenbarger, Hyde, Wing and Hsied (1986) which have linked habitual exercise in the lifestyles of large populations to favourable mortality outcomes. More answers and confirming evidence need to be sought, especially for women who, with a life-span advantage over males of 7 to 8 years, are possibly more concerned about health outcomes and quality of life than extending their life span. Perceptions about one's health may be the germinating force leading to specific kinds of self-protecting behaviors. Possibly the prospects of chronic illness provoke certain women to action, while convincing others that it is time to slow down. Both strategies can be considered health protective even though  66 the behaviors are oppositional lifestyle choices. Such a dichotomy needs further exploration since women, in general, exceed males in all other personal health care behaviors  except exercise (Verbrugge & Wingard, 1987) .  A plausible explanation for the reluctance of females to be diligent about promoting their health through exercise is that the way girls and women have been socialized over the life course has lessened their advocacy and belief in exercise and sport as valued behaviors (Csizma, Wittig, & Schurr, 1988). By late life, vigorous physical pursuits are not only seen as socially inappropriate, but also viewed as high-exertion (Winborn, Meyers & Mulling, 1988), and therefore potentially life and health-threatening. Evidence is accumulating that adults who perceive their health as poor are more reluctant to adopt exercise than those who perceive good personal health. Morgan and colleagues (1984) studied an unspecified age group of General Foods employees and found that while male participants who enrolled in the fitness program perceived good health and positive beliefs about exercise, women associated exercise with  poorer health. Furthermore, female exercise adopters did  not improve their perceptions of health at retest, while male adopters did. In short, Morgan's team found that the exercise and health relationship differ for men and women. In a randomized walking exercise intervention on older women, those who adhered to the two year exercise program were, at base-line, of lighter weight, already more active, and non-smokers (Kriska, Bayles, Cauley, LaPorte, Black Sandler, & Pambianco, 1986). However, the variable that best differentiated between compliers and noncompliers was the frequency of reported illness over the two year period. Women who adhered to the exercise program reported significantly 1  ' ^'  II II I -  a II^  i  67 in this population may be quite different from factors limiting physical activity in the young" (Kriska et al., 1986, p.562).  Health Symptoms and Exercise Behavior Although women are outliving their male counterparts by seven or eight years, aged women are vulnerable to one or multiple chronic conditions through much of the period of this extended life. A survey of Canadians in the mid-1980's found 55% of adults over 65 reporting arthritis/rheumatism, 39% reporting hypertension, 26% reporting heart trouble and 24% with respiratory problems (General Social Survey, 1985). "Normal aging" for elderly women typically follows this profile: almost half are physically limited in daily activity; 60% of women over 65 were screened out of random public physical fitness testing for reasons of health risk (Canada Fitness Survey, 1983); and 46% are institutionalized by age 85. Many are truly unfit and cannot complete even modified fitness tests of basic strength (O'Brien & Conger, 1988). This lack of basic strength is blamed for the majority of the falls experienced by one third of all adults over age 65 (Blake, Morgan, Bendell, Dallasso, Ebrahim, Arie, Fentem & Bassey, 1988; Frontera & Meredith, 1989). One-third of women aged 65 years and up will have one or more vertebral fractures. As women survive into their eighties, one third are expected to experience a hip fracture (Nelson, Fisher, Dilmanian, Dallal, & Evans, 1991). With widowhood, poverty and declining health as the norm for about half of all older women (Arendell & Estes, 1987), women over the age of 65 are more likely to report stress than men. Contributing to this stress are psychosocial  68 and physiological effects of motor-sensory deprivation due to physical inactivity (Winget & Derosha, 1986).  Marital Status and Exercise An active life partner has been hypothesized to have a strong influence on the activity patterns of their mate. In 1976, Spreitzer and Snyder advocated a social learning perspective and suggested that the acquisition of the sport role resulted from exposure to role models and reinforcement from significant others. Using self-administered questionnaires with a systematic probability sample on 264 adults under the age of 61, these researchers found that female involvement appeared to be determined more by their spouses's degree of involvement than the extent to which women participated in their youth. Having a spouse who is indirectly involved in sports tends to reinforce earlier encouragement from one's parents and to interact with perceived ability partly to explain the degree to which one is involved in sport as an adult. (Spreitzer & Snyder, 1976, p. 244) Tait and Dobash (1986) claim that women consciously or unconsciously marry a male whose orientations in lifestyle are similar to their own. They suggest that "women who take part in sport perceive a very high degree of support from their nominated or significant male" (Tait & Dobash, 1986, p.268). The relationship between marital status and cardiovascular risk behaviors was the focus of a study on 7,849 midwestern men and women (Venters, Jacobs, Jr., Pirie, Luepker, Folsom & Gillam, 1986). Separated or divorced persons reported higher levels of relaxation-enhancing behaviors such as smoking, drinking and higher levels of physical activity. Married men showed lower mortality rates over single men, but married women were not advantaged in this way over single women.  69 For women, "never having been married" was the most favourable status with respect to educational attainment and reported history of heart attack and stroke. Being married, or over the age of 40, were situations that were accompanied by less physical activity (Rudman, 1986). Ishii-Kuntz (1990) studied the formal activities of elderly women and the determinants of their participation in senior's centers. Using a nation-wide probability sample of 1,051 women over the age of 65 (data collected in 1981), this research categorized variables as: "predisposing" (age, race, education, and marital status), "enabling" (income, employment status, health status, and transportation) and "need" (loneliness and living arrangement). The average age of the women were 73.2 and 62% were widowed. The major findings were that age, race and health status were influencing participation in voluntary organizations and senior centers. Elderly widows were more likely to participate in voluntary organizations than married women and loneliness had a positive impact on senior centre participation.  Motherhood, Children and Leisure-time Exercise Motherhood and grandmotherhood are the most enduring social roles with which women identify (Moen & Huntington, 1991). Yet little research has examined the role of motherhood on women's exercise patterns in middle age and beyond. The number of children born, number of children raised in the household, the spacing of children, and the health of children cared for are plausible factors affecting women's leisure and physical activity patterns over much of the adult life course (Henderson, Bialeschki, Shaw, & Freysinger, 1990). The impact of family size on women's exercise patterns is likely to be a reflection of available leisure-time,  70 availability of financial resources and a mother's interest in being physically active in the play patterns of her family. Leisure for women has been, and largely still is, home-based. Since home is also a place, if not the place, of work for women, it is not surprising that work and leisure activities are often intertwined and indistinguishable (Henderson, et al., 1990, p. 10). A Canadian time budget study by Shaw (1985) examined the distribution of leisure of 60 married women for a 48 hour period. While over 70% of gardening and animalcare chores were considered by the women to be leisure, only 4% or less of home chores and laundry were defined as leisure. Henderson and colleagues (1990) argue that women have typically been oppressed in most aspects of their lives, including leisure. Allen and Chin-Sang (1990) studied the meaning of leisure and work for 30 aging black women. When asked how their definition and experience of leisure had changed over the years, most women said they "had no leisure in the past" (p.737). Housework was clearly the predominant feature of their lives, but gardening was classified by many women to be a leisure activity. Even though housework may be the predominant physical activity for many women, Verbrugge (1986) has reported that homemakers are not particularly enthusiastic about their work compared to employed men and women. Furthermore, employed women usually liked housework less than their jobs. An important example of family leisure is the holiday, at least for those who can afford it. Yet the family holiday is "often a breeding ground for arguments and family conflicts and where the domestic labour for women may actually increase..." (Deem, 1982, p.112). Leisure outside the home has often been viewed as something that mothers should willingly sacrifice. As with today's contemporary women, even if pioneer women had been ysicatty  rul es of Lhild rearing, housekeeping and  71 domestic skills such as cooking and sewing, would have consumed much of their time and energy during their maternal years. Housewives with young children are likely to perceive little freedom in their lives because of constantly being "on call" (Meissner, 1977). Children constrain women's leisure not just because of the considerable physical care required by babies and young children, but also because of their social and emotional needs. The responsibility of child care, which falls disproportionately to women in society, reduces women's leisure options and inhibits a considerable number of leisure activities... (Henderson et al., 1990, p.123) Many pioneer women worked farmland or ranches or provided support systems for their husband's occupational pursuits. Home industry and responsibilities were often were initiated in the adolescent years, no doubt heightened during the depression and war years, so that leisure-time physical activity may have been limited. Certainly the time required for skilled athletic development would not have been highly valued by society as the preferred way for average women to spend their time. Women's time (each day and across their life span) was (and still is) perceived as time that could be interrupted for whatever needs or crises arose, particularly those needs related to the family, while the time of men was respected as private. (Henderson et al., 1990, p.25). Deem (1982) has contended that leisure spaces are particularly difficult for women to find in their own home. When older, most females have established lifestyles without sport skills or habitual fitness activities, and the normal course of action is to taper activity in the later years, not increase it. An interaction effect between income and number of children demonstrates the complexity of developing a simple understanding of exercise behavior and one's family situation (Fishwick & Hayes, 1989). This interaction effect indicates that as number of children increases for lower income persons, participation in physical^  edbeb.  Fur hiyh income persons, participation  72 increases with more children. Women who have adequate financial resources may be better able to afford the time and cost of engaging in activities alongside their children. Children of middle class families may experience more instruction in lifelong activities such as tennis, swimming, skiing, skating, and golf. They, therefore, are in a better situation to participate in sports in which whole families can enjoy. Recent studies show, however, that parents typically pay less for their daughters' sports equipment, instruction, and training than for their sons'.  Work Role, Employment and Exercise Changing patterns of activity involvement are thought to be the result of altered role transitions and altered opportunities across the life cycle (McPherson, 1984). Particularly relevant to leisure-time activity are the demands of an individual's employment and non-paid work. McPherson suggests that "the decisions concerning how to minimize costs and maximize rewards with respect to physical activity involvement are related to commitment, adherence, and the relationship between work and leisure and between work and family responsibilities" (1984, p.223). Neither women's employment, nor domestic work role, have been studied extensively for their role in determining physical activity patterns. Life work is thought to be closely tied to level of education, marital status, number of children raised, health status, social class and so either types of work, as activity-promoting forces for women, are difficult to study in isolation. Morgan (1986) points out that it is physical activity of any kind, not athleticism, that is associated with quality and quantity of life. To date, no rpcearch- on the  73 fitness and health outcomes of women's domestic and/or employment activities have been reported. However, employed women are apparently more physically active. Almost 60% of Canadian women in manager/professional occupations report they are active compared to only 44% of women in blue collar work (Government of Canada, 1984). Recently, there has been information to suggest that employed women carry most of the domestic work load at home in addition to full-time engagement in employed labour. This means that leisure-time opportunities for employed women may be even more severely limited, and that opportunities for exercise may be lacking unless women undertake activity during their normal work day. Fishwick and Hayes (1989) surveyed 401 adults aged 18 to 83 years of age to determine their involvement in recreational activities by age, race, gender and social class. In contrast to much of the literature, they found that women were not under-represented as leisure-time sports participants, but were vastly underrepresented as team sport participants possibly because time constraints made it difficult to schedule practices and games with other adults. In summary, Fishwick and Hayes emphasize that normative expectations channel women into "gender-appropriate" activities such as aerobics. Steinhardt and Carrier (1989) have examined early and continued participation in a corporate work-site health and fitness program (Conoco Inc.). Using a broad array of variables representing socio-environmental factors, physical-behavioral factors and psychological factors, they obtained physiological and questionnaire data on 143 women aged 19 to 60 years old. They found younger employees were more likely to be "starters" in the program while those who adhered claimed more "attitudinal commitment" and perceived the health and fitness program was to be more convenient.  74 McPherson and Kozlik (1987) reviewed studies on Canadian leisure activities by age, and noted that participation rates drop severely after age 19 and again at age 64 - two points regarded as endpoints of labour force activity. At these transition points, men participated in sport activity to a greater degree than women, and rates of participation increased in a linear pattern with income and level of education. More study is needed to understand how paid work facilitates or undermines women's active leisure patterns, and how retirement from employment encourages or discourages future participation.  Socioeconomic Status and Exercise Findings from Canada's Health Promotion Survey (1987) suggest that: Canadians who rate their health as excellent are three times more likely to be in the highest income bracket than those who consider their health to be poor. Canadians in the lowest income bracket are four times as likely to rate their health as only fair or poor as those in the highest income bracket. However, direct information on the role of financial situation and older adult exercise is lacking. McDaniel (1989) points out that economic inequities tend to accumulate in old age and are exacerbated by a pension system that is not workable for many women. More research will be needed to tease out the interwoven elements of socioeconomic status, gender, educational level, occupational level and race which, in various ways, are likely to limit lifestyle choices, activity patterns and outcomes of good health.  75  SOCIALIZATION THEORY In this section, discussion continues on the situational determinants of exercise. Socialization Theory is presented with particular attention paid to the situation of childhood and how experiences and opportunity may lead to early mastery and movement confidence. Because child socialization and child efficacy for physical skill are historical variables, for the purpose of this study, they are considered to be unalterable characteristics or "situational variables". "Social system theories" reflect the dynamic interaction between society and individuals (McPherson, 1990). General sociological theory advocates the importance of social structure, social processes, social roles and the effect of the environment on human behavior (George, 1985).  Socialization is a lifelong process that enables an individual to participate in a society...Socialization is both a process and a product. As a process, socialization involves learning skills, traits, knowledges, attitudes, language, beliefs, norms, values and shared behavioral expectations associated with present or future roles. The process may vary because of such factors such as gender, socioeconomic status, community or ethnic differences, cultural differences, and individual differences in the lifestyle and values of socializing agents. (McPherson, 1990, p. 130) One of the most significant social roles affecting human behavior is that created by gender. A wealth of evidence exists to suggest that males and females are socialized very differently from an early age, particularly with respect to aggressive play and choice of toys. Traditionally, females have had little encouragement to engage in vigorous and challenging forms of physical activity and sport (Lirgg & Feltz, 1989; Mangan & Park, 1987; Vertinsky, 1991). Females, from birth to death, have been socialized to be more passive physically, and in particular, are lured away from aggressive forms of sport (Zoble, 1973). Wakat and Odom (1982) note that "although infant males and females start out with ruuyi  y fi le  same physical capabilities, they soon begin to experience different  76 courses of development, as set down by society according to what is appropriate for little boys and what is appropriate for little girls" (p.34). Media interest and the public popularity of contemporary male professional sport heron attest to important differences in gender support for physical activity which still persist today (Cole, 1991; Kane, 1989). Hall (1976) reports that women's attitudes toward activity are generally favourable, and concludes that socialization and opportunity are therefore most responsible for the inadequate participation of females. Apparently, contemporary females still experience considerable role conflict in certain athletic settings and that the female in sport is still considered to be in man's territory (Csizma, Wittig, & Schurr, 1988). Since the 1970's, sport sociologists recognize that physical activity and sport are Used as an important medium in which males are "masculinized" (DiIorio, 1989; Hall, 1985; Theberge & Donnelly, 1984). Hauge (1973) wrote an early review paper of the influence of the nuclear family on female sport participation. She raised the possibility that a propensity toward tomboyish behavior might be affected by sibling order, family size, parental modelling, and childhood opportunity. As important as the family in socializing young people, is probably their school experience. Vertinsky (1992) presents a thorough summary and discussion of the challenges and opportunities physical educators face in providing gender equity in contemporary school settings. Thus with strong social forces operating at school and in the home, the socializing determinants for physical activity participation are highly likely to be different for men and women. In addition to gender, age is another socializing force affecting human behavior,^y N^aLLivi ty behavior. For example, being old in the  77 1990's is the present identity of a particular social group who have experienced a certain social orientation to life and look upon retirement as "a well-earned rest". Furthermore, social stereotyping or "ageism" dictates late life expectations and behavior (Achenbaum, 1986; Palmore, 1990). Fraser (1989) has found that many disease processes such as hearing loss, vision loss, shortness of breath and joint immobility are simply accepted by many elderly as "normal aging". Socialization theory, therefore, recognizes ageist practice as a social construction whereby adults are reinforced for more passive roles and ageexpected behaviors (McPherson, 1990; Ostrow, 1982; Ostrow & Dzewaltowski, 1986; Teague, 1987), for learning helplessness (Brown & Inouye, 1978) and carrying out a self-fulfilling prophecy of age decline, frAilty and illness (Edgerton, 1986; Waxier, 1980). Chronological age can therefore be hypothesized to play a role in explaining physical activity, exercise and sport behavior. One's personal circumstances and socializing experiences are likely to play an important role in determining how active one would want to be over the life course (Labonte, 1983; Rudman, 1986). Sport socialization theory more specifically recognizes that individuals are socialized differently in physical activity settings often starting at a very early age. A number of studies support the hypothesis that early experiences in childhood physical activity create advantages for adult participation later on (Butcher, 1983; Dishman & Dunn, 1988; Howell & McKenzie, 1987; Spreitzer & Snyder, 1976; Powell & Dysinger, 1987; Sofranko & Nolan, 1972). Adams II and Brynteson (1992) reported that middle-aged adults exercised more frequently, and held higher health value for exercise, if they had simply more hourly exposure to physical activity training as college or  78 participation in highschool increased adult sports involvement for both men and women. At least two studies do not support the early socialization hypothesis (Adams & Brynteson, 1992; Steinhardt & Carrier, 1989). Dishman and Dunn (1988) warn that the "available evidence on the relationship between childhood and adulthood exercise patterns is not compelling" (p. 186) because the associations come exclusively from cross-sectional and retrospective surveys with adults and is limited to sport and physical education experiences.  The Significance of Socialization for Childhood Physical Activity A collection of studies have identified relationships among childhood activity opportunities, perceived physical ability and social support by one's parents, peers, male siblings, teachers (Greendorfer, 1983; Griffin, 1982; Weiss & Knoppers, 1982; Wood & Abernathy, 1989). Research indicates that physical play and recreation during childhood contributes to an awareness of one's physical world and enhances the ability to manipulate and control one's surroundings (Lewin & Olsen, 1985, p. 216). Moore and colleagues (1991) researched the relationship between activity levels of parents and those of their young children aged four to seven. Caltrac accelerometers monitored children, mothers and fathers for more than ten hours per day for about eight days. Children of active mothers were 2.0 times as likely to be active as children of inactive mothers. The relative odds of being active for the children of active fathers was 3.5. When both parents were active, the children were 5.8 times as likely to be active. Possible mechanisms for the relationship between parents' and child's activity levels include the parents' serving as role models, sharing of activitieb by family members, enhancement and support by active parents of their child's participation in physical activity, and genetically  79 transmitted factors that predispose the child to increased levels of physical activity. (Moore, Lombardi, White, Campbell, Oliveria, & Ellison, 1991, p.215) Boutilier and San Giovanni (1985) summarize a number of studies which support the important role of early childhood physical activity in the development of role play, sense of group membership, fitness and motor skill, bodily awareness and improved self-concept. Although sex differences in motives for participation have been identified, with boys placing more emphasis on achievement, rewards and status, generally the motivational differences for boys and girls may no longer be so great (Gill, Gross, & Huddleston, 1983). Powell and Dysinger (1987) reviewed the available literature on the association between childhood and adult physical activity patterns. In summarizing the available studies, they felt that the Harvard alumni study, which connected college sport to current activity patterns and absence of coronary disease (Paffenbarger, Hyde, Wing & Hsied, 1986), provided the strongest evidence supporting an association between youth and adult activity in males. Butcher (1983) examined three categories of variables which influenced 661 adolescent girls: personal attributes, socializing agents, and socialization situations. For competitive interschool teams and intramural activities, certain personal attributes (movement satisfaction and self-confidence, independence, and assertive self-description) were most important. For community-organized activities, socializing agents (parental influence) and socialization situations (socioeconomic status) were most influential, while for the total activities participated in, the amount of sports equipment was crucial. Butcher noted that by Grade 10 there was a noticeable drop in girls' school physical activity participation, but not so in physical activity in the community setting. Three  80 types of social influence were family (parents and siblings), peers, and teachers or coaches. The physical education teacher is better qualified than most to open the path towards successful ageing to individuals and society with their work. Should not the physical education teacher also learn and teach how sport activity is to be adapted not only to the immediate but also to the longterm interests of the individual! (Meusel, 1991, p.16) Spreitzer and Snyder (1976) developed a path analysis model of early sport socialization from self-administered questionnaires on a middle aged population of 110 women and 154 men (mean age = 42 years), and an average education level of 13 years. From seven predictor variables, 46% of the variance in sport involvement in men and 40% of the variance among women was explained. From the results they formed a causal framework stated as follows: Parents (especially the father) who are interested in sports tend to encourage their offspring to participate in sports, which markedly increases the likelihood of a youth's participation in sports. One's participation as a youth markedly affects how one perceives his/her athletic ability. This perception, in turn, has a strong impact on the degree of adult participation in sport. (Spreitzer & Snyder, 1976:244) Howell and McKenzie (1987) found that participation in high school athletic programs was related to team sport activity later in adulthood with some gender differences. The effects of high school varsity sport experience on late life sport involvement was greater for men than for women, a finding similar to that of Spreitzer and Snyder (1976) who obtained a correlation of .25 between youth and adult sport involvement for males, but no relationship at all for females. An interesting finding comes from Steinhardt and Carrier's (1989) study of men and women attending a health and fitness program in a large corporation. Their findings contradict research that indicates youth participation has positive effects on initial participation in a work-site program. Instead they found that the individuals who were among the first to participate in the  81 corporate fitness program appeared to have been sedentary as youth. They suggest "that those individuals who were active as youth may be less attracted to an organized program or perhaps less dependent on an organized program to exercise" (Steinhardt & Carrier, 1989, p. 123).  Childhood Situation: Opportunity, Mastery, Movement Competence Growing evidence suggests that physical mastery, or an individual's selfconcept of physical ability, may be the most important determinant of both affect and expectancy, and therefore, an important mediator of human motivation and movement behavior (Roberts, Kleiber & Duda, 1981). Evidence is lacking, however, at what stage in human development the self-concept of movement ability must be realized to be incorporated into one's identity. Positive and early experiences with physical activity and sport seem to be important in some cultures. Raivio (1986) reports that interviews with Finnish female sport administrators, aged 29 to 60, indicate a similar childhood background; all ten women had mainly participated in outdoor activities with both boys and girls, and all but one had had a parent or close relative encouraging their participation in physical activities. Morgan (1986) found that former male highschool athletes reported significantly more positive attitudes towards activity and their estimation of personal physical ability in young adulthood, yet were not necessarily more active than former non-athletes. He concluded that former athletes appeared to base their subjective judgements of physical ability on an earlier reference point. Furthermore Morgan notes that former athletes still regarded themselves as athletes. Dishman (1990) claims that organized sport experience in one's youth might I I  I^•^I  y in a er  82 years and is amenable to large-scale public intervention" (p. 81). Thus, a causal relationship between skilled participation in childhood and adult activity would have strong implications for public health (Dishman, 1990). Yet no prospective study has been conducted to specifically show this relationship; prospective studies of childhood sport or physical education as a determinant of contemporary adult activity have not been reported. Rather, the numerous cross-sectional and retrospective studies that link youth sport history with contemporary physical activity must be viewed with caution. Thus, the question of whether physical activity determinants for the individual who begins habitual activity at middle age are the same as those of a person who was active since childhood remains unclear. Research is consistently finding that perceived competence in physical skills is an important influence on the participation and motivation of children in sport contexts. Young participants in organized sports are found to have higher perceived competence, more persistence and higher expectations for future success than non-participant children (Roberts, Kleiber & Duda, 1981). Selfefficacy ratings and experience level in gymnastics were significant predictors of the actual success of boys aged 7 to 18 in competitive gymnastics settings. There is preliminary evidence that mastery of high-risk sport skill may even generate increased self-efficacy for other physical and social tasks (Brody, Hatfield & Spalding, 1988). Yet what children say and do in the physical education setting may be highly sex-differentiated and a subtle socialization is part of a "hidden curriculum" (Griffin, 1982, p.84). Boys limit girls' opportunities to learn physical skill, and hence undermine their future confidence by hassling them, and interfere with their own 1 arning by clowning around.  83 Similarly, family members are thought to be powerful determinants of the activity behaviors of children. Parents, in particular, are key role models and may at first support the athletic achievements of a daughter. But "when parents decide that sport achievement is a threat to her social life and eventual marriage, they push the feminine role" (Hauge, 1973, p.19) to be more competitive in the social setting. There is a tendency for like-sexed parent to have more influence on a child's involvement than does the opposite-sexed parent (Snyder & Spreitzer, 1973). Years ago, Koch (1956) reported that girls with brothers more than two years older had a tendency to be tomboyish. Petruzzello & Corbin (1988) also advocate that experience and gender are important determinants of performer confidence. They conducted two studies on college-age males and females which indicated that even on gender-neutral motor tasks, females rate thOselves with significantly lower levels of confidence. They concluded that the greater experiences and social rewards for males in physical activity raises their expectancies for success on new tasks. It becomes increasingly clear that experience affects self-confidence. Successful experiences/mastery attempts can serve to enhance selfconfidence. As such, the more experience one has at a variety of tasks in a variety of physical situations, the greater the possibilities are that self-confidence can be generalized to more situations. (Petruzzello & Corbin, 1988, p.182) The construct  movement confidence was employed for predicting children's  physical performance and play decisions. Griffin and Keogh (1982) developed a working model which describes "movement confidence as both a consequence and a mediator in an involvement cycle" (p.213). Movement confidence was defined as an individual feeling of adequacy in a movement situation. The confidence or assurance with which an individual approaches a movement situation should be an important determinant of what an individual will choose to do and how adequate the movement performance will be. (Griffin & Keogh, 1982, p.213)  84 The Griffin and Keogh Model of Movement Confidence assumes that a cognitive evaluation of self in relation to the perceived demands of a task is an antecedent to confidence. Utilizing a Movement Confidence Inventory (MCI), Griffin, Keogh and Maybee (1984) studied perceptions of movement confidence of 450 college-age students for performing 12 different movement and sport related tasks. In that study it became clear that movement competence was not a lone predictor of performance confidence; rather, it was accompanied by perceptions of potential pleasant and unpleasant movement experiences. A playground movement confidence inventory (PMCI) and a stunt movement confidence inventory (SMCI) were developed to identify children who may be in need of special assistance in learning sport and physical activity (Crawford & Griffin, 1986; Griffin & Crawford, 1989). Griffin'and Crawford (1989) noted that level of confidence varied according to the nature of the specific task and context. The Stunt Movement Confidence Inventory was able to reliably discriminate between high- and low-confidence children aged 9 to 11 on six "stunting" tasks. The next major section to be addressed in this literature review introduces four prominent theories which have demonstrated utility in predicting health behaviors. The four theories are: The Health Belief Model, Health Locus of Control Theory, the Theory of Reasoned Action, and Social Cognitive Theory. The chapter concludes with the literature pertaining to the specific constructs of Social Cognitive Theory and their relationship to exercise behaviors.  85  BEHAVIORAL HEALTH PSYCHOLOGY PERSPECTIVES Behavioral research in health psychology is concerned with attitudes and behaviors of adults regarding their health and independence. Some health professionals suggest that individual knowledge and beliefs about one's world are the main controlling determinants of one's behavior. Research is needed on how health attitudes and behaviours are acquired, are affected by daily living, change as people grow older, and can be modified as scientific knowledge advances. (Health & Welfare Canada, 1990, p. 4) Rosenstock's Health Belief Model is prominent in representing this second stance in many contemporary health behavior change studies  6  .  THE HEALTH BELIEF MODEL The Health Belief Model (HBM) of Rosenstock, Strecher and Becker (1988) defines health behavior as "any activity undertaken by a person who believes him/herself to be healthy for the purpose of preventing disease". The model highlights cognitive mediating processes through an emphasis of the role of subjective beliefs or expectations held by the individual. According to Rosenstock (1974) and Becker (1974), health-related action depends on individual perceptions about perceived susceptibility to a specific disease and perceived severity of the disease, environmental modifying factors such as cues to act, and likelihood of action based on perceived barriers and perceived benefits of taking preventive action (Figure 2.0).  6 e e  is,  A review of the role Becker, & Rosenstock, 1986).  v or c ange is provided by Strecher,  86  Figure 2.0 Basic Elements of the Health Belief Model  ^ ^ INDIVIDUAL ^MODIFYING ^ LIKELIHOOD OF PERCEPTIONS FACTORS ACTION Perceived benefits of preventive action Demographic variables (age, sex, race, ethnicity, etc.) Sociopsychological variables  --Jo  Perceived bathos to preventive action  •  V Perceived susceptability to Disease "X" ••■•••■••►  Perceived Seriousness (Severity) of Disease "X"  minus  erceived Threat of Disease "X"  1.■•••■■■11•11411111i1  cum to &lion Mass Media Campaigns  Advice from others Reminder postcard from physician or dentist Illness of family member or friend Newspaper or magazine article  Likelihood of Taking Recommended Preventive Health Action  87 The Health Belief Model has served as a conceptual core for many compliance studies and has predicted health, illness, and sick-role behavior. Health beliefs have been positively associated with exercise adherence in older adults when there is more knowledge about the exercise regimen (Tirrell & Hart, 1980) and when individuals are knowledgeable about their actual health situation (Rakowski, 1984). Janz and Becker (1984) report on a comprehensive review of 29 health-related investigations utilizing the Health Belief Model. Summary results provide substantial support for the HBM, with "perceived barriers" and "perceived benefits" proving to be the most powerful HBM dimensions in demonstrating associations with behavior. There are a number of conceptual difficulties and inconsistencies associated with the Health Belief Model. First, health beliefs are understood to be a reflection on how knowledgeable people are about the consequences of their health behaviors. Blumenthal (1983) has acknowledged that the sole basis of evaluating people's beliefs is to assess their biomedical knowledge. Thus health beliefs based on personal experience, or beliefs based on social, psychological, and cultural foundations may be good predictors of behavior but are rarely examined. Kirscht (1974) was troubled by the fact that "supporting evidence for the utility of the Health Belief Model has come primarily, but not exclusively, from retrospective studies, and with reference to preventive behaviors (Kirscht, 1974, p.455). Yet, the Health Belief Model rests on the notion of beliefs causing behavior -- a time order sequence that is not often addressed in the research. The Health Belief Model hinges on an individual's perceived susceptibility to a disease occurrence (Rosenstock, 1974). Yet some research suggests that health beliefs arc more predictive of health protective behaviur in the well  88 elderly who perceive lower susceptibility to risk or health threats (Lindsay-Reid & Osborn, 1980; Segall & Chappell, 1989). These views do not support the Health Belief Model's readiness to act because of recognition of an illness threat (perceived susceptibility). Perceived severity of a health problem is another belief assessed by the  Health Belief Model; the original HBM has a disease-avoidance orientation. Individuals who are emotionally aroused by the provision of knowledge about the severity of their condition are predicted to initiate or adhere to preventive health practice. There are, however, potential shortcomings in using threat or knowledge of threat to promote health practice. Gintner, Rectanus, Achord, and Parker (1987) found that attendance at a blood pressure screening was cut in half when an illness.-threat format was used to motivate participants. Participants were more likely to attend the screening when a wellness appeal was presented. The structure and reliability of eight health beliefs were examined by Jette, Cummings, Brock, Phelps, and Naessens (1981). They identified six general health threats, perceived severity of five conditions, four perceived barriers to taking medications, four questions measuring general health concern, three items assessing trust in physicians, three items of perceived susceptibility, three items on perceived health status, and two questions about health locus of control. Factor analysis revealed that condition-specific measures of perception of susceptibility and severity and situation-specific measures of perceived barriers were empirically distinct from general measures of these beliefs. Their findings supported the theoretical assumption that HBM dimensions were sufficiently distinct to be considered deferent beliefs, but warned against mixing specific and general questionnaire items within the same index.  89 Norman (1985) reviewed over 600 research articles in an effort to determine what is known about people's health habits and practices. He states, No one factor has been found to provide a sufficient basis for predicting health behaviour. People's health habits and practices are often daily actions which have been influenced by a host of cultural, social, psychological and biological factors. (Norman, 1985, p.3) Norman claims that while knowledge and beliefs about the health consequences of behavior may have some impact on the way in which people behave, they do not, on their own, provide a strong basis for preventive health activities. This is because "it is an enormously complex task to present people with information in a way that will lead them to change their health beliefs" (p.3). Norman suggests that health education initiatives may well be futile considering the discrepancy between people's health beliefs and their actual behavior. His scepticism about altering people's beliefs in the hopes of altering their behavior mirrors the findings of Haefner and Kirscht (1970) who found that merely changing the participants' beliefs about health was not enough to alter behavior. To date, little theoretical guidance has been given to the prediction of older adult exercise behavior, but two prominent theories have been explored (Dzewaltowski, 1989a, 1989b; Dzewaltowski, Noble & Shaw, 1990).  HEALTH LOCUS OF CONTROL THEORY The third theory in behavioral health psychology which is potentially able to predict late life exercise behavior is the Health Locus of Control Theory. Rotter (1954, 1966) contends that a person learns or is conditioned operantly on the basis of his or her history of positive or negative reinforcement. The person also develops a sense of internal or external locus of control. Those with an internal locus of control are more likely Lu belt inilidte change, whereas those -  90 who are externally controlled are more likely to be influenced by others. Therefore, locus of control refers to an individual's perceived influence in regulating outcomes. Waliston and Waliston (1978) proposed that an individual's sense of control varies by domains of experience and actions, such as health experiences. Health  Locus of Control (HLC) is defined as perceived control over one's health. "Internal" locus of control refers to perceptions that a health outcome is due one's personal actions, while "external" locus of control refers to perceptions that a health outcome is due to chance, external factors or the actions of others (Kist-Kline, 1989; Wallston & Wallston, 1978). According to Health Locus of Control Theory, individuals exhibiting more control over their health (internals) would be more likely to be participating in health-promoting behaviors such as physical activity. Many studies have found that participants who claim to be more physically active tend to be more internally controlled (Bonds, 1980; Carlson & Petti, 1989; Kleiber & Hemmer, 1981; Lumpkin, 1985; Moore, 1980; O'Connell & Price, 1982; Perri & Templar, 19841985; Sonstroem & Walker, 1973). Internality may be found especially if they are in team sports (Lynn, Phelan, & Kiker, 1969; Kleiber & Hemmer, 1981). Health Locus of Control has only modest associations with exercise (Calnan, 1988; McCready & Long, 1985), specifically free-living physical activity (Dishman & Steinhardt, 1990). Sometimes health locus of control is not predictive at all of health behaviors (Calnan, 1988) or exercise behavior (Blair, Kohl, Pate, Blair, Howe, Rosenberg, & Parker, 1980; Dishman & Gettman, 1983; Kaplan, Atkins & Reinsch, 1984; Laffrey & Isenberg, 1983). People who believe in chance tend to be older (Calnan, 1988; female (Calnan, 1988^less educated Bo 1  91 1988). Recent research suggests that as soon as people learn that their health has deteriorated, they also exhibit more external perceptions of control (Waller & Bates, 1992). This loss of a sense of personal control for both success and failure is associated with depression (Mirowski & Ross, 1990).  THEORY OF REASONED ACTION One prominent theory is Ajzen and Fishbein's Theory of Reasoned Action (1977, 1980), recently revised as the Theory of Planned Behavior (Ajzen, 1985; Ajzen & Madden, 1986). The formation of "intention" to act is central to this theory. According to this model, an individual's intention to perform a given behavior is a function of attitude toward the behavior, and normative beliefs about what relevant others think one should do, weighted by personal motivation to comply with those relevant others. Behavioral intention is viewed as a type of expectancy and is indicated by the person's subjective probability that she will perform the behavior in question. In regard to predicting exercise intentions, the theory has been more useful in explaining exercise intentions in males (Godin & Shephard, 1987), in explaining exercise behavior compared to Kenyon's Attitudes Towards Physical Activity inventory (Godin & Shephard, 1986), but "has not identified a predominant cognitive profile of those who intend to exercise" (Godin, Shephard, & Colantonio, 1986). Riddle (1980) obtained success with Fishbein's model obtaining a high correlation (r = .82) between jogging behavior and further intention to exercise. However, Riddle noted that her most important finding was that "joggers had stronger positive evaluations of the beneficial consequences of regular jogging" while nonexercisers "were not as convinced" (p. 673). The  92 importance of these perceptions about the "consequences of exercise" is that "consequences" is not a theoretical element in the theory of reasoned action, but rather matches a component called outcome expectations in Social Cognitive Theory. Furthermore, recent attempts have been made, with some success, to improve the Fishbein-Ajzen model by adding the self-efficacy component of Bandura's Social Cognitive Theory (de Vries, Dijkstra & Kuhlman, 1988; Wurtele & Maddux, 1987). This new work has found that "self-efficacy has also a direct effect on behavior after controlling for intention" (de Vries, et al., 1988, p.273).  SOCIAL COGNITIVE THEORY Social Learning Theory Although Social Cognitive Theory arose from a behavioral psychology perspective, the theory incorporates important features related to socialization. Perry, Baranowski and Parcel (1990) trace the 50 year history of social learning theory noting that Miller and Dollard (1941) originally introduced Social Learning Theory to explain imitation of behavior among animals and humans. Rotter first applied early social learning principles to clinical psychology (1954), which in turn led to his development of the idea of "generalized expectancies of reinforcement" (1966). Building on Rotter's social learning theory, Bandura is credited for the contemporary development of Social Cognitive Theory (1977a, 1981, 1982, 1986, 1989; Bandura & Cervone, 1986; Bandura & Schunk, 1981). Social Cognitive Theory continues to evolve as a broad conceptual domain that incorporates many theoretical ideas and is employed by many areas of practice. Thus, with such  93 diversity, abuse is possible. Perry and colleagues (1990) point out a theoretical pitfall of research using SCT; that pitfall is "one concept was often explored while the others were excluded completely" (Perry et al., 1990, p. 180). For example, the concept of internal and external locus of control dominated social learning research at one time. Bandura stated that self-efficacy may be the single most important factor in promoting behavioral change. The emphasis on a single variable oversimplifies reality but is "a reflection of the structure of experimental research, which usually permits analysis of only a few variables at a time" (Perry et al., 1990, p. 180). Clearly, research which examines all of the main constructs of Social Cognitive Theory should be encouraged.  The Constructs of Social Cognitive Theory Originally, social learning theory emphasized the role of self-referent beliefs or subjective expectancies held by the subject. Beliefs or subjective expectancies about the possibilities and consequences of personal action were considered to be the key mediating forces between a person and a specific behavior (Figure 2.1).  Figure 2.1 Basic Elements of Social Learning Theory  PERSON ^ > SELF-REFERENT THOUGHTS ^ > BEHAVIOR  94 Recent years have witnessed a resurgence of interest in the study of selfreferent phenomena. Bandura (1989) points to several reasons why self-processes have come to pervade many domains of psychology. Self-generated activities lie at the very heart of causal processes. They not only contribute to the meaning and valence of most external influences, but they also function as important determinants of motivation and action. (Bandura, 1989, p.1175) According to SCT, individuals' beliefs of self-efficacy are central to their decision to participate in physical activity. Efficacy expectations are defined as a person's judgements of their capability to organize and execute their skills and resources and that of the environment to perform an action that will lead to a designated outcome (Bandura, 1977, 1986). Specifically Bandura's theory claims thg•human action is guided by a core set of four beliefs: motivation to obtain a goal (Incentive to Act), beliefs that a certain behavior will be beneficial in reaching a goal (Outcome Expectations), a belief in one's ability to perform the action (Self-Efficacy), and finally, a perception that the action will be endorsed or "positively reinforced" (Social-Environmental Cues). Thus the four expectancies encompass internal and external factors that may affect individual behavior. In contrast, perceived behavioral control (in the Fishbein-Ajzen model) and perceived barriers (in the Health Belief Model) are assumed to reflect external factors (availability of time, facilities etc). Exercise behavior, using a social cognitive perspective, is predicted to occur when an individual: 1) Highly values the outcomes of physical activity, 2) Perceives that specific forms of physical activity will lead to health benefits and that harmful outcomes are not likely, y, an  95 4) Perceives that they will be socially reinforced for participating. To the extent that individuals "learn" what to value, what is risky for them, how competent they really are, and how much endorsement society will offer for their activity, SCT beliefs indirectly reflect the socio-environmental milieu, cultural learning, and past experiences of the individual. But individuals are not considered to be passive recipients of environmental influence. Inherent in SCT is the idea that people self-regulate their environment as well as their actions. Bandura has described the nature and function of human agency as a "conceptual model of triadic reciprocal causation" (1986, p. 1175), also known as "reciprocal determinism" (Perry et al., 1990, p. 165). Self-functioning is viewed as a continuous interaction between environmental factors, beliefs and behaviors (Bandura, 1986) (Figure 2.2). The interaction is such that a change in one has implications for the others. According to SCT, the environment provides the social and physical situation within which the person must function and thus also provides the incentives and disincentives (expectancies) for the performance of behavior.  Figure 2.2 Bandura's Concept of Reciprocal Determinism  Environment  Incentives Self—Efficacy Environmental Cues Outcome Expectations  Person  -4------m-----...—  Behavior  96  The Cognitive Determinants of Exercise: SCT Self-referent perspectives are important to understand because attitudes, opinions and beliefs are formative and modifiable (Dishman, 1990). Health promotion initiatives and program experiences can profoundly and quickly transform individual perspectives and behavior and are claimed to have a significant impact on mortality outcomes even after age 70 (Kaplan, Seeman, Cohen, Knudsen & Guralnik, 1987). Conceptual models to understand and explain the diverse determinants for participation in physical activity and exercise in the elderly are relatively undeveloped. Motivational theories which have had some success are now being combined for further strength in prediction (Sharpe & O'Connell, 1992). Incentives, attitudes, beliefs, expectancies, perceived barriers, cues to act, and self-perceptions are among the most common constructs used in contemporary theoretical models. Even in younger adult groups where there has been "a remarkable growth in applied interest about exercise adherence, the development of conceptual models leading toward a motivational theory of habitual physical activity has lagged far behind" (Dishman, 1988). In the forthcoming section, key studies pertaining to the cognitive elements of Social Cognitive Theory as applied to exercise behavior are reviewed. Earlier in this chapter, theoretical models which utilize these cognitive elements are addressed in detail. In Chapter 3, justification is made for the utility of Social Cognitive Theory in the explanation of older adult exercise behavior.  97  Incentives or Motives to Exercise Developing an understanding about the mechanisms underlying exercise participation at every life stage poses a challenge. Many studies on motivation and exercise have concentrated on adherence, or the problem of keeping participants exercising once they have started. Less has been studied on the reasons why people start exercising, and little information is available on the motives behind the initiation or resumption of exercise regimens of older adults. Despite a remarkable growth in applied interest about exercise adherence, the development of conceptual models leading toward a motivational theory of habitual physical activity has lagged far behind. (Sonstroem, 1988, p.123) Sonstroem (1978) advanced one of the first models for prediction of exercise involvement. Key to his theoretical model, was the notion of "self-esteem". In predicting exercise participation, his model posits that self-perceptions of physical ability (Estimation) influence an individual's interest in physical activity (Attraction) and that Attraction provides the greater influence on exercise participation. Although Sonstroem's model has enjoyed only limited success in explaining the activity patterns of high school boys, it provides some important information: that is, people who are motivated to be more active will likely turn to those activities to which they are attracted, in which they feel competent and confident and through which their self-esteem is likely to be maintained. But there are many participatory motives that could apply with older adults. An individual may seek exercise as a way to relieve boredom or stress and get out of the house, to socialize, to have fun, to maintain or learn physical skills, to follow doctor's orders, to demonstrate self-discipline, to experioncP  98 competition or self-measurement, to promote beauty and/or to obtain health and fitness benefits. Some research suggests that motives may differ by gender. Finkenberg (1991) used Kenyon's Attitudes Toward Physical Activity Scale and found, in college students, that males were significantly more motivated to exercise for competition, while females were more motivated to exercise for health and fitness. Similarly, O'Brien and Conger (1991) have found that older men and women participating in the Alberta Seniors Games seemed to be motivated by different expected outcomes. Male participants admitted they enjoyed maintaining a degree of public acclaim for sustaining their physical prowess into old age. In contrast, older women said that sport participation promoted their health and personal independence so that they could sustain more active caregiver roles within their families. Godin, Shephard and Colantonio (1986) focused their study on middle-aged employees who expressed a willingness to exercise but actually did not exercise. The overall findings identified surprisingly little difference between cognitive profiles of inactive and active adults. Individuals who had intended to exercise, but didn't exercise during the two month period of the study, differed from fellow exercisers in perceiving a problem with "lack of time", and believing that exercise required more effort and provided less health value. The researchers suggested that motivation might not be the limiting factor for many inactive people; sedentary intenders might simply be confronted with more social and environmental constraints than those who are active. However, lack of leisure time as a barrier to exercise, does not seem to rank high among many aging adults, most of whom are retired.  99 Health as Incentive to Act Of interest to this research is the expectancy of gaining health from participation in physical activity. The value of an expectancy is an important construct in Social Cognitive Theory and is seen as a key motive or incentive in explaining human behavior. Thus, health as an expected outcome of participation in physical activity must be valued in order to be considered to be an incentive. That is, those individuals who place a high value on maintaining their health, and who understand the role of regular moderate exercise in maintaining health, would be hypothesized to be more likely to be exercising. There is scant evidence to support this hypothesis. The Campbell's Survey (Stephens & Craig, 1990) reports that, by age 65, less than half of Canadian adults judged physical activity as very important to their health. In regard to their personal health promotion, females attach more importance than males to all factors, especially body weight, a good diet, and rest and sleep, with only one exception: regular physical activity. (Craig & Stephens, 1990, p.43) Karen Altergott, editor of Daily life in later life (1988), includes a chapter titled "Life course and the daily lives of older adults in Canada" written by Zuzanek and Box. They claim that, as a result of retirement, older Canadians gain approximately 38 extra hours of disposable time per week. Paradoxically, although older adults possess greater amounts of free time, their rates of participation in leisure activities, and the number of leisure activities they engage in, decline after retirement. (Zuzanek & Box, 1988, p.153) Substantial participation declines are noted to occur in the post-retirement period in sports, sport spectatorship, and outdoor recreation (especially for women), while activities affected little, or even slightly expanded, are visiting, reading, radio listening, watching television, playing cards, hobbies, pleasure driving, and physically less demanding forms of outdoor activities such  100 as walking. These, then, are the activities which appear to most interest older adults. Upon scanning the main activities chosen by seniors, few appear to have been chosen for their "health value". Rather, the recreational activities of the older Canadians often seem to have been selected instead for their social and entertainment value. Thus one might question the assumption that health value acts as an "incentive" in explaining older adult involvement in physical activity. First, many older adults may not be aware of the health-promoting effects of exercise participation. Second, older adults may see less relevance to sustaining or taking up healthpromoting activities as age advances, particularly if they believe that such efforts are only likely to contribute to short-term gains. They may indeed feel that their "life time" is running out and they are too late to realize any significant gains from serious participation in physical activities (O'Brien & Conger, 1991). They may feel that the finite life-span may cut short the fruits of long-term gains from an exercise program, and thus undermine perceptions of "health value" to be gained from regular exercise participation. Perhaps there comes a point in the life course when people stop doing "the right thing", because they perceive that life will soon end anyway. In those circumstances, older adults may be motivated to undertake activities primarily for pleasure and entertainment.  Outcome Expectations Expectations about positive or negative consequences (benefits or risks) are important to the actions taken by individuals according to both Social Cognitive Theory and the Health Belief Model. The ability to envision the likely outcomes  101 of prospective actions is one way in which anticipatory mechanisms regulate human motivation and action. "People strive to gain anticipated beneficial outcomes and to forestall aversive ones" (Bandura, 1989, p.1180). However, the effects of outcome expectancies on performance motivation are partly governed by selfbeliefs of efficacy. In activities in which the level of competence dictates the outcomes, the types of outcomes people anticipate depend largely on their beliefs of how well they will be able to perform in given situations. The association of outcome expectations and efficacy are such that "when variations in perceived self-efficacy are partialed out, the outcomes expected for given performances do not have much of an independent effect on behavior" (Bandura, 1989, p.1180). Positive outcome expectations, in physical activity settings, would require that the individual would have to believe that the outcomes of participation would be personally beneficial and the risks of participation would be reasonably low. According to Bandura's interpretation, perceptions of risk would then be predicted to be higher for individuals who have low self-efficacy for physical activity, and perceptions of benefits would be predicted to be higher for those who exhibit high self-efficacy for physical efficacy. More information and discussion about the perceived benefits and risks of exercise are provided in Chapter 3.  Self-Efficacy and Movement Confidence Introduction: Self-Efficacy Self-efficacy is the most studied component of social cognitive theory and it has received a good deal of interest from researchers attempting to understand the social-psychology of exercise behavior. Perceived self-efficacy appears to  102 play an influential role in ways that affect motivation and "intention" for involvement in physical activity and sport. Since physical activity is often conducted in public settings, and performance is visible to any observer, confidence to perform may be particularly important in the exercise setting. Bandura (1977a) has defined self-efficacy as the strength of an individual's perceived self-confidence or belief that he or she can successfully complete a task through the expression of ability. Although those who view themselves as having high ability for a task are also apt to feel efficacious for performing it, simply possessing the ability to perform a task does not guarantee a high degree of self-efficacy. In addition to efficacy perceptions, Schunk and Carbonari (1984) claim that "ability" involves effort, luck and task difficulty three additional elements that can explain the success or failure of personal actions. Furthermore, competent behavior is unlikely to occur if social, psychological or structural barriers exist, or if there are inadequate incentives (Bandura, Adams & Beyer, 1977). Without aspirations and active involvement in activities, people are unmotivated, bored, and uncertain about their capabilities. Life without elements of challenge can be rather dull. (Bandura & Cervone, 1986, p. 111) Self-referent perceptions of efficacy are at least partly responsible for the kinds of challenges which people choose to undertake, how much effort they will spend on an activity, and how long they will persevere in the face of difficulties (Bandura, 1986). Social cognitive theory claims that discrepancies between performance feedback and personal standards lead to self-dissatisfaction which then serves as a powerful motivational inducement for enhanced effort. Those who distrust their capabilities are easily discouraged by failure, whereas those who feel  •  I^II  103 their performances fall short, and persevere until they succeed (Bandura & Cervone, 1986). Dzewaltowski (1989b) therefore suggests that a measure of efficacy in the exercise setting should "examine individuals' efficacy toward coping with difficult situations and still adhere to an exercise program" (p. 254). Recent research suggests that task-specific efficacy measures are superior predictors of behavior over general efficacy measures (Bandura & Cervone, 1986). Thus self-efficacy in exercise research has been defined in many ways such as self-rated confidence to adhere to an exercise program, or confidence to sustain exercise for 60 minutes. The need to be specific in the assessment of efficacy expectations has required researchers to develop their own measures to deal with their particular research question. Consequently, the available research on efficacy and exercise at times appears to be haphazard. Some studies merely verify what Bandura's theory has already demonstrated. Others add some interesting information but direction is lacking.  Barriers to Perceived Efficacy to Exercise The Campbell's Survey (Craig & Stephens, 1990) indicates that barriers and lack of perceived control interferes with the desire of 70% of 65+ Canadian women to regularly exercise. In general, males at all ages felt they had more control over their life situations than did females. The major gender differences in perceived barriers were a greater emphasis by females on family time pressures, lack of energy, and lack of ability. Lack of a partner and lack of ability were important explanations among those who resisted or lapsed in their activity programs.  104 Davis-Berman (1989) has provided evidence that physical self-efficacy in aging women may be undermined by the effects of depression. Thirty percent of the variance in depression scores were explained by a physical self-efficacy score. Research by Kelly (1987) serves to remind us that efficacy for specialized tasks may be undermined by numerous barriers largely outside of individual control. Kelly studied leisure in aging adults and identified two kinds of activity which distinguished adults with the highest levels of life satisfaction: "those providing a context for interaction with valued others and 'high investment' activities" (Kelly, 1987, p.111). High investment activities are those that have been developed over a period of time, require some acquisition of skill, and are most likely to yield outcomes of an enhanced sense of competence, worth, and Orsonal expression" (Kelly, 1987, p. 112) Kelly suggests that the low rates of participation in exercise, sport and outdoor recreation for those age 55 and above reflect the likelihood that real health and physical ability limitations are operative. Exercise and sport settings are highinvestment activities that call for special facilities, supervision and guidance, companions, specialized equipment and high levels of effort and skill requirements and resources which may be harder to fulfil as people reach advanced age. Therefore, social contexts become more limited to family and neighbours and locales of activity become limited to the private residence. Along the same lines, Godin, Shephard & Colantino (1986) found that sedentary adults (average age 39) who had positive intentions to exercise perceived regular exercise to be "tiring", "time-consuming" and placed less value on "being healthy". The major perceived obstacles of intenders to exercise participation was "lack of time" and "perceived exertion". Both of these factors can be linked to one's efficacy to carry out a program of exercise.  105 Waller and Bates (1992) studied self-efficacy beliefs, multidimensional health locus of control and lifestyle behaviors in 57 healthy elderly subjects (mean age 74.7) with a view to determine who could benefit most from health promotion programs. Most of the subjects were characterized by an internal locus of control belief (91.2%), high generalized self-efficacy (57.9%) and good health behaviors. Waller and Bates (1992) suggested that individuals with an internal locus of control and high generalized self-efficacy are more likely to benefit from a health education program than those with an external locus of control and low self-efficacy, but this suggestion was not examined directly in their study. Woodward and Wellston (1987) examined age, desire for control, information and general self-efficacy in 116 adults aged 20 to 99. They found that individuals over 60 years of age desired less health.,related control than did younger adults, and preferred that health professionals make decisions for them. Perceived self-efficacy was lower for individuals over 60 years of age. The findings suggest that "those individuals most at risk for chronic illnesses and hospitalization are also those who are most likely to fail to take an active role in their health care" (p.3).  The Role of Habit and Previous Physical Activity In a random sample of 136 University of Toronto employees, Godin, Valois, Shephard & Desharnais (1987) examined, among other factors, the influence of past behavior on subsequent behavior. Three measures of "habit" were used: Immediate Past Behavior (weekly score using MET units to quantify activity level), Past 4 Months Behavior, and Adulthood Behavior (frequency of getting sweaty during leisure time as an adult). They found that "distal" exercise behavior (three weekc and two months later) was predicted by both intention to exer. i  dud  106 Immediate Past Behavior. The important role of "habit" was highlighted in this study. Godin and colleagues concluded that if a person has never engaged in a particular behavior, it remains uniquely under the control of behavioral intentions. However as the behavior is repeated, the importance of the habit increases, with a corresponding diminution in the importance of behavioral intention. ..the decision to adopt an active life-style, over the previous habit of being sedentary, requires more "girding up of loins" than the decision to continue to exercise for an individual who has a well-established habit of exercising. Consequently, a process of change has to take place, this requiring "will" in order to compete and resist the forces of the old habit in establishing a new habit. (Godin et al., 1987) While the role of "habit" has received only preliminary attention in predicting current exercise behavior, habitual activity may have some bearing on advancing intention into actual action. In this study, the notion of "habit" was accommodated in the variable "movement confidence" which combined self-efficacy  ("I  am sure I can do this") with habitual experience ("I have done this a lot").  The Role of Efficacy on Exercise Behavior A number of contemporary studies are linking physical self-efficacy, perceived movement competence, or self-rated physical ability to predictions of exercise behavior, physical fitness and adherence to fitness programs.  For  example, self-efficacy has been found to be the most powerful and statistically significant correlate of both walking and vigorous exercise among ill and healthy groups alike (Hofstetter, Hovell, Macera, Sallis, Spry, Barrington, Callender, Hackley & Rauh, 1991). Ryckman, Robbins, Thornton, and Cantrell (1982) developed a general Physical Self-Efficacy Scale (PSE) in order to identify an individual's perceived physical self-confidence. While this instrument demonstrated adequate reliability and  107 validity in predicting general self-efficacy in sport, McAuley and Gill (1983) did not find this to be a useful instrument in evaluating female college performer's efficacy for gymnastics performances. Rather, the Perceived Physical Ability (PPA), a sub-scale of the PSE was more situation-specific and offered better prediction of performance outcomes. Duda and Tappe (1989) studied 145 adults aged 25 to 81 years of age with the purpose of examining motivational differences in exercise by gender and age. Sense of physical competence was assessed with the Perceived Physical Ability subscale of the Physical Self-Efficacy Scale (Ryckman et al., 1982). They found that older and younger physically active adults did not differ in perceived physical self-efficacy and health status. However, middle-aged and elderly adults tended to engage in.exercise more for the positive consequences on health status than young adults. Males engaged in exercise more for competition than females while females exercised more for fitness reasons. There was also a trend for females to exercise more for affiliative reasons than males. Women tended to view themselves to be less physically able, perceive greater significant other support for their involvement in exercise, and believe that one's fitness status is primarily of fate or chance occurrences (externality). Duda and Tappe concluded that exercise has different meanings to young, middle-aged and elderly men and women noting that perceptions of efficacy and social support for exercise were the major differences between men and women. Ryckman's Perceived Physical Ability (PPA) scale was used to examine the relationship between perceived physical ability and indices of physical fitness (Thornton, Ryckman, Robbins, Donelli & Biser, 1987). The males in the study (aged 17 to 64) generally were more physically fit relative to the females (aged 18 to 64), yet there was no difference between their PPA ab predicted. They  108 concluded that their findings did not support the predictive utility of perceived physical ability where actual indices of physical ability were involved. Efficacy expectations were originally thought to be specific to particular behaviors and not necessarily generalized to other behaviors. However, evidence exists in one study that efficacy for exercise may be generalizable. Kaplan, Atkins and Reinsch (1984) examined specific versus generalized efficacy expectations for exercise in older patients with chronic obstructive pulmonary disease (COPD). All subjects (mean age was 65 years) were given a prescription to undertake two daily walks. The experimental group of the randomized design received three months of supervised training and advice while the control group received only advice. After three months, the experimental group had significantly increased their activity level, their perceived efficacy for walking and also efficacy expectations for other similar physical behaviors in comparison to the control group. Their results suggest a "bidirectional" type of causation, or a "reciprocal-causal" model meaning that "efficacy and performance attainments may affect each other in reciprocal fashion" (Kaplan et al., 1984, p.239). Dzewaltowski (1989b) conducted a study comparing Bandura's Social Cognitive Theory to Fishbein and Ajzen's Theory of Reasoned Action. The theories' constructs were assessed on 328 physical education students prior to collecting data on the total days exercised over an 8 week period. With all the variables in a regression equation, the Theory of Reasoned Action could only account for 6% of the variance in exercise behavior. When Social Cognitive Theory variables were entered into the equation, they accounted for 14% of the variation. Dzewaltowski (1989a) concluded "it may be that those who exercise are confident  that they sari exerLise despit e unLontrol1 able factors" (p. bb).  109 Self-efficacy has been used to predict over-exertion during programmed exercise in 40 men; The men were recovering CAD patients with an average age of 55 years (Ewart, Stewart, Gillilan, Keleman, Valenti, Manley & Keleman (1986b). Patients' confidence in their ability to jog various distances was measured with a jog self-efficacy (SE) scale before an eight-week group exercise program was begun. Ambulatory heart rate monitoring disclosed significant noncompliance with exercise prescriptions: 33% of patients exceeded their prescribed range of 70 to 85% of maximum treadmill heart rate for at least 10 minutes of the 20 minute exercise bout. Another 25% spent 10 minutes or longer exercising below the prescribed range. "Overachievers" were patients who overestimated their ability to jog, while "underachievers" were those who overestimated their exercise heart rate. Jogging SE proved superior to treadmill performance, depression measures and Type A personality measures in predicting patient adherence to exercise prescription.  The Role of Exercise on the Development of Efficacy The literature generally supports Bandura's theory that efficacy and performance strengthen each other in reciprocal causation (Kaplan et al., 1984). Previous successful performance leads to stronger efficacy expectations, and stronger efficacy perceptions increase the likelihood of successful performance. Marcus and colleagues (1992) found that higher levels of self-efficacy for exercise accompanied higher levels of exercise activity in blue collar employees. Results indicated that employees who had not yet begun to exercise, in contrast to those who exercised regularly, had little confidence in their ability to exercise. Stewart and King (1991) suggest that exercise may enhance a sense of y  g^ h Pr  dl IU11 of  twu IlleL 1c1E1 I S1115 .  110 Regular exercise may provide people with an enhanced sense of their ability to handle problems. Regular exercise may also provide a model of control (e.g. "I obtain improved health by exercising") that may generalize to other life domains. (Stewart & King, 1991, p.113) However, no differences in a sense of control were found in at least one randomized study of a twelve-week aerobic exercise and strengthening program for 15 men and women aged 61 to 86 compared to a social activity control group (N = 15) or a waiting list control group (N = 18) (Emery & Gatz, 1990). One of the best studies done on self-efficacy and exercise behavior was recently reported by McAuley, Courneya & Letturich (1991). Fifty females and 50 males (average age of 54) were examined for the effect of acute and long-term exercise on self-efficacy responses in sedentary adults. Three measures of selfefficacy were employed to determine subject's beliefs in their physical capacities as related to exercise and fitness. Specifically, the efficacy scales represented subject confidence to be able to succeed with 1) increasing numbers of sit-up repetitions in one minute, 2) cycling longer at increasing work loads. and 3) walk-jog successive quarter-mile distances within 4 minute intervals. Subjects participated in a 5 month aerobic exercise program, three times per week in one hour sessions. Both males and females demonstrated significant increases in efficacy following acute exercise. Females, who had demonstrated initially lower self-perceptions than males, made dramatic increases in efficacy during the exercise program, equalling or surpassing those of males. Increased selfefficacy closely accompanied the actual measured physiological gains in performance as expected. McAuley and associates concluded that their results are encouraging since sedentary individuals in their middle years were able to make significant health-related gains through a relatively low-impact activity such as walking.  111 The effects of running the treadmill only three weeks following myocardial infarction (MI) on subsequent physical activity were evaluated in 40 consecutive men with a mean age of 52 years (Ewart, Taylor, Reese & DeBusk, 1983). The men were examined for self-perceived ability to walk/run distances from one block to five miles, climb stairs from several steps to four flights, engage in sexual intercourse from one to 20 or more minutes and handle objects from 10 to 75 pounds. Patients' confidence in their ability to perform these activities were assessed before and after a symptom-limited treadmill test of aerobic fitness. Significant increases in self-efficacy occurred after the treadmill test for activities most similar to the test: walking, stair climbing and running. Another finding supports Bandura's specificity of SE and suggests that physical efficacy perceptions can be quickly improved by successful performance on:a related activity. In another study by Ewart and colleagues (1986b) examined 43 men with coronary artery disease proposing that highly specific estimates of personal capabilities mediate adoption of new or difficult exercise settings. Correlational analyses of self-efficacy in relation to strength and endurance tests strongly supported the contention that the adoption of novel activities is governed by highly specific self-perceptions. The pattern of findings suggest that favourable appraisal of one's athletic ability increases motivation to pursue sport, leading to greater participation, increased skill, more positive self-appraisal, and consequently, higher motivation. Hogan and Santomier (1984) examined the effects of participation in a learnto-swim program on the self-efficacy of older adults. The subjects were 38 volunteers 60 years of age or older. This study was quasiexperimental in that it utilt7ed a non-randomized control group. As such, the study k vulnerable to  112 the confounding effects of self-selection. As expected, significant changes  in  post-test efficacy were found, but more importantly, this efficacy generalized to other performance-related situations. Open-ended questioning indicated  that  other aspects of their lives had been affected such as "now able to handle a trip to China", "my walking has improved" and "chores are more easy". Such generalized efficacy outcomes are not in agreement with Bandura's theory that efficacy expectations are specific. A ten week exercise program containing 69% women aged 55 to 80 was conducted by Howze, DiGilio, Bennett, and Smith (1983). "High-attenders" were those who attended 15 or more sessions out of the 20 two-hour sessions. "Low-attenders" were less confident of their physical abilities and were more worried about injury. Ninety-two percent of the participants said that they "felt better in general" and felt more physically fit after the program. Howze and colleagues suggested building self-confidence by progressive exercise which provided successful participation all along the way. Feltz (1988) has examined gender differences in the causal elements of selfefficacy on a "high-avoidance" motor task (the back dive) in college age students. Feltz proposed a respecified model of Social Cognitive Theory that included both self-efficacy and previous performance (experience) as direct predictors of approach/avoidance behavior on the dive. The diving efficacy scale asked the subject to rate the degree of confidence he or she felt about accomplishing the back dive successfully for each of four board heights. Each rating was made on a 10-point scale from 0 (great uncertainty) to 10 (great certainty). Actual performance was measured by two trained observers using an objectively designed performance evaluation. Females attempting the back dive reported higher levels of state anxiety and autonomic arousal (high heart rates)  113 than males on their first attempts. In perceiving heart rate changes, males tended to underestimate while females tended to overestimate their increases in heart rate. No sex differences were found in self-efficacy scores, and both males and females significantly increased self-efficacy perceptions from Trial 1 to Trial 2. Thus previous performance and self-efficacy measures were both strong predictors of subsequent performance for males and females. The reciprocal nature of efficacy and performance is confirmed in one other study. Barling and Abel (1983) studied self-efficacy beliefs and tennis performance in 40 active males (26.6 years). Three 10-item scales assessed selfefficacy strength ("I can play most of my strokes correctly"). Two judges evaluated actual tennis performance (inter-rater r = .91; p < .001 in all 12 rated skills). Tennis players who had higher•self-efficacy for tennis were rated as the most skilful performers, meaning the relationship between efficacy and performance holds even when performance is evaluated by others.  Environmental Cues for Physical Activity: Social Support The objective of this section to review the known inter-relationships of physical activity, aging and social support systems. Understanding the impact of various social reinforcements on individual and group behavior holds promise for cost-effective, community-level intervention.  Alleist Practice in Communities Ageism, or the explanation of behavior by age considerations alone, is thought to be a powerful social element governing the present active living choices for adults as they age. One major theory of aging rests on this  114 discriminating assumption: disengagement theory endorses the withdrawal from social participation as a natural and healthy course of aging (Cumming & Henry, 1961). Others would argue that disengagement is not at all a choice, but rather is aggressively driven by the political economy of age stratification, with certain privileges being denied, and access to social participation limited, based simply on the age of individuals. Examples of age discrimination in physical activity are: limited access by the elderly to high-demand public sport facilities, few opportunities to receive expert coaching and instruction, disinterest by the media and general public, and a lack of publicly organized events and activity programs representing the broad interests of older adults (Curtis & White, 1984). Ageist practice socializes older adults into a narrow range of "appropriate" activities, even though those activities may not represent those in which older adults have developed lifelong skill or interest (McPherson, 1984). Views that older adults should exert themselves less as they age, and should become less competitive at the activities they do, are consistent with the social forces that are present in much of contemporary society. When these social forces are evident, it becomes clear to older adults that social acceptance is lacking for them to demonstrate many of the varieties of athletic excellence. Ageism is probably the most obstructive form of adult socialization preventing older adult participation in some of the most valued forms of physical activity.  Social Support Of interest to the present research are studies which 1) identify the types of people who lack interest and desire to exercise, and 2) the kinds of physical And cultural environments which promote more physical daivity. Among the  115 prominent relationships between socio-environmental context and optimal health is a powerful construct called "social support." Many studies have already demonstrated with clarity that social support is important to the maintenance of good health (Pilisuk & Minkler, 1985), including reduced psychological distress (Holahan & Moos, 1981), and reduced mortality in elderly populations (Blazer, 1982). Exercise scientists are similarly beginning to appreciate the significance of social support in physical activity settings. Indeed, while physical activity settings may be among the most important  sources  of social contact and support in the lives of older adults, other forms of social support must apparently already exist. For older adults to live actively, some degree of social support may be  an essential prerequisite.  Defining and Measuring Social Support The scientific measurement of social support is a recent phenomenon requiring clear operational definitions. Social support has seen  a rapid  evolution of conceptual meaning ranging from individualized emotional and affective dimensions to large, contextual features of a particular society (Esdaile & Wilkins, 1987). Most measurement to date has addressed social support at the "micro" level. For example, an early definition of social support was given by Cassel (1976) as the gratification of a person's basic social needs (approval, esteem, succorance). Cobb (1976) conceived social support to be information leading a person to believe that they were cared for, esteemed and belonging to a network. The measurement of a multidimensional construct such as social support is difficult, if only because there are now a legion of available instruments which tap intn emotional support, tangible support, informational support and support  116 provided to others at both perceptual and enacted levels. Social support, as a  quantity, can be measured objectively as the number of social contacts one has, or the number of phone contacts or visits from friends in the past week. The  quality of social support can be subjectively assessed with perceptual rating scales about the adequacy of one's support network.  Physical Activity as Social Networking Recreation centres, senior's groups, sport clubs and even shopping malls provide positive settings in which to engage in socializing and physical activity (Graham, Graham & MacLean, 1991). Within these social settings, both formal and informal structures can provide instrumental aid, information and advice. In addition, such settings supply one of the biggest benefits of leisure activity companionship (Ishii-Kuntz, 1990; Tinsley, Teaff, Colbs & Kaufman, 1985). Institutionalized elderly are known to place high priority on the social context of physical activity, although they also hope that exercise will also enhance their health and fitness (Mobily, Lemke, Drube, Wallace, Leslie, & Weissinger, 1987). Women have, in the past, been less likely to take advantage of these existing community networks for participation in physical activity and sport. Even though finding companionship for activity has occurred with more frequency for males (Curtis & White, 1984), women do not necessarily choose to exercise at home alone. From young adulthood on, females are found more often in caregiving and domestic situations which may limit their ability to formalize social networks outside the home environment.  117 According to a national survey on women with disabilities, physical limitations and medical concerns were not considered to be their primary limitations to activity participation. The primary changes that would encourage greater participation in physical activities were accessible facilities that are closer to home, knowledgeable instructors, people with whom to participate, and more available information on programs. (Fitness Canada Women's Program, 1990). Social encouragement from other adults may be lacking for older women with limitations. Almost half of the women surveyed said they alone were responsible for getting involved in activity while family and friends were influential in activating only 13% of respondents. For older adults, ease of transportation to physical activity settings and costs of participation may be practical barriers to obtaining the support they need to be more active. Many elderly women never learned to drive the family car or can no longer afford to maintain a car and therefore limit their activities to whatever is available in their neighbourhood.  Role of Group Cohesion Social support and group cohesiveness have been studied with a view to understanding why people begin physical activity, why some maintain their involvement and why others stop participating altogether. Dishman (1984) has noted that after six months, over half of those who begin an exercise program have already dropped out. As might be expected, people who do not adhere to  a  specific fitness or sport setting are less personally attracted to the group's task and to the group as a social unit (Carron, Widmeyer & Brawley, 1988). The literature underscores the need for an awareness of how the social and physical environment can affect the elderly individual's sustained involvement in group activity, as well as the need for understanding how activity engagement may  118 relate to the individual's evaluation of the environment and of the self (Barris, 1987). In other research, a series of six studies examined the behavioral and cognitive procedures which would enhance adherence to a 3-day-per-week walking/jogging program in sedentary adults (Martin, Dubbert, Katell, et al., 1984). Overall, the results of the studies confirm the importance of social support, including instructor feedback and praise during exercise.  Role of Companionship Lack of social support for older women in the form of sport opportunities and companionship is one of the key findings of a study by Curtis and White (1984). Using a sample of 33,762 native-born Canadians who filled out a nine-page survey questionnaire, they found that older females participated in fewer sport -  activities than younger women, but participated more frequently in the past year. Only 10 percent of women over age 60 had at least one physical recreation which they had pursued one or more times on an annual basis. Problems in finding others with whom to participate was a problem for over 20% of the elderly women, and they were the one age group who had the most problems with finding companionship. Older women had twice as much difficulty as same-age men with finding companionship, yet at the same time, reported that time conflicts in activities were only half the problem that men had experienced. Ishii-Kuntz (1990) reports that widowed women, in particular, are likely to be seeking companionship and social opportunities in senior's centers.  The Role of Spousal Support One hypothesis is that women who have active life companions, active partners or active spouses are more likely to be physically active themselves  119 (Snyder & Spreitzer, 1973). Having a spouse who is indirectly involved in sports is thought to reinforce earlier encouragement from one's parents and to increase one's perceived ability to be involved as an adult (Spreitzer & Snyder, 1976). In analyzing family influence on sports involvement, researchers have claimed that there is considerable similarity of activity patterns between a couple. Evidently, they mutually reinforce one another's interest in this sphere of leisure behavior. Explanations of this finding might lie in the mate selection process where a common interest in sports might serve as an additional inducement for the match; also, the findings might suggest that a strong interest in sports on the part of one spouse is gradually transmitted to the partner. (Snyder & Spreitzer, 1973, p.252) Other research suggests that "women who take part in sport perceive a very high degree of support from their nominated or significant male" (Tait & Dobash, 1986). This evidence notwithstanding, compared with men, women aged 45 and older reported less support from their spouse and experience less encouragement to be active with advancing years (Stephens & Craig, 1990). Furthermore, Hauge (1973) has suggested that "middle class men look outside the home for sport companions almost twice as frequently as the women do" (p.25). One of the few available studies on the role of spousal support to exercise adherence is unfortunately available only in relation to men. Myocardial patients were studied for drop-out rate from an exercise program over a seven year period (Andrew, Oldridge, Parker, Cunningham, Rechnitzer, Jones, Buck, Kavanagh, Shephard, Sutton & McDonald, 1981). Of all the determinants being considered, spouse approval was the most significant finding. The drop-out rate of those with little or no support from their wives was three times greater than in those men with positive spousal encouragement. Ishii-Kuntz (1990) examined how predisposing, enabling and need factors influenced elderly women's participation in voluntary organizations and senior  120 centers.  A nationwide probability sample provided data which indicated that age,  race and health status influenced participation. Elderly widows were more likely to be involved in voluntary organizations than married women, with loneliness being a major factor leading to seniors center participation.  The Role of the Physician The cautionary warnings that one must always consult a physician before taking up any interest in physical activity may be doing more harm than good. Certainly anyone who is doubtful about one's personal state of health should consult a physician. In principle, however, there is less risk in activity than in continuous inactivity. In a nutshell, our opinion is that it is more advisable to pass a careful medical examination if one intends to be sedentary in order to establish whether one's state of health is good enough to withstand the inactivity. (Astrand, 1986, p.4) Evidence suggests that ordinary people will not get extraordinary advice from a physician about how to start an exercise program. Moreover the "see your physician" prescription may prevent many adults from ever getting started since a chain of dependency is then formed (O'Brien Cousins & Burgess, 1992). Becoming more physically active depends on seeing the doctor, and it also depends on what the doctor has to say. Because of time constraints, physicians often do not discuss their attitudes and knowledge about exercise with their patients.  The  dependency continues as one then seeks out an activity program in which exercise needs are met. If the program is good, one may become dependent on a highly prescriptive program and the motivational skills of its leader in order to be certain to adequately exercise. Regular pulse rate checking serves to remind people that they may be at risk of something going wrong with the heart, and consequently individuals may become too anxious to exercise on their own. This scenario is an example of how some  121 forms of social support can backfire and become barriers to individualized and independent involvement in physical activity. Health promoters have begun to examine the interest and competency of physicians to provide encouragement for their patients' activity patterns. It has been noted that physicians who have graduated since the late 1960's are more likely to believe in the importance of regular aerobic exercise, but overall, only about one-quarter of physicians have been found to think engagement in aerobic activity three times a week is very important to health (Wechsler, Levine, Idelson, Rohman, & Taylor, 1983). Internists are more likely to ask about exercise behaviors than general practitioners (53% to 31%) and all physicians are more likely to ask about smoking, alcohol and other drugs, than they are likely to ask about diet, exercise and stress (Wechsler, et al., 1983). Only 3 to 8 percent of physicians thought they were "very successful" in helping patients achieve changes in various health behaviors, but 21 percent were optimistic about their ability to help patients increase exercise. Surveys of physicians in Massachusetts and Maryland indicated that just less than 50 percent of primary care physicians routinely inquire about their patients' exercise practices (Wechsler et al„ 1983). In an exercise intervention study on women aged 55 to 80, only 5% of the participants noted that regular exercise had been recommended to them by a physician (Howze et al., 1983). In another study, only 27% of the physicians felt that exercise was "very important" for the average person. Thus "a large proportion of physicians are not fully convinced of the value of exercise for health" (Powell, Spain, Christenson, & Mollenkamp, 1986). Whether physician inquiries include in-depth questions about intensity, during or frequency of exercise is not known, nor do we know how physicians alter their questioning with younger and older individuals.  122 Current views on the athletic potential of older adults, and older women in particular, are considered to be overly conservative even by health promotion experts and exercise physiologists. Most professionals may be concerned more about the risks of participation for more frail elders and the potential for harm and litigation outcomes than they are for raising activity levels of the entire community (DeLorey, 1989). Added to this professional conservatism, there has been a persistent myth among older adults advocating the scientific concept of "conservation of energy." Retirement, for many, means that it is time to take a long-deserved rest from lifelong physical work demands. The social norm for retired individuals has been rather passive leisure pursuits, and it is difficult to change expectations about activity choices if the participant is perceived to be already more ambitious than others of the same age. Particularly if an adult has been physically inactive in recent years, physicians and friends are unlikely to try to convince him or her that now is the time to start exercising.  The Role of Friends and Family The number of close friends which a person has appears to be significantly associated with the pursuit of general preventive health behaviors such as nonsmoking behavior and exercise (Calnan, 1985). While companionship for physical activity is considered to be a reinforcing factor (Biddle & Smith, 1991), in recent decades, time pressures are evident, at least in middle age families. Inflationary pressures and changing attitudes towards the social roles of men and women have seen the rapid rise of two-career families and what some call the "death of leisure" (Posner, 1991). Over half of the adult female population is  in  the labour  force full-time; this means that there has been a significant  123 change in the workload patterns of women. About one quarter of Americans at work are spending 49 hours or more each week on the job (Kilborn, 1990). It has been estimated that women average 66 to 75 hours per week at combined job and family responsibilities as compared with 42 to 49 hours per week 50 years ago (Edwards & Hill, 1982). The implications of this for women's physical activity patterns is that many women have too little leisure time in which to be physically active, and friendships and social networks may also become more difficult to sustain under these kinds of time pressures. The next chapter identifies the theoretical framework for the study based on the literature that has been outlined in this chapter. The theoretical framework discusses the selection of Social Cognitive Theory as the most suitable starting point and then justifies a system of synthesis of theories. The 16 constructs of importance to a Composite Theory of late life exercise are presented and the next chapter concludes with a visual working model for this study.  124  III. THEORETICAL FRAMEWORK Introduction The review of literature, in the previous chapter, has identified as many as 16 constructs from different theoretical models which have demonstrated predictive relationships with exercise behavior. One purpose of this study was to develop a theoretical framework which would have utility in explaining late life exercise for older women. This chapter begins with an explanation of why the Health Belief Model and Theory of Reasoned Action have not been adopted in this study. Next, the chapter presents a rationale for synthesizing 10 constructs from social epidemiology and Socialization Theory with six cognitive constructs from Social Cognitive Theory and Health Locus of Control Theory. This is not a new model, but rather a more comprehensive application of Social Cognitive Theory, and provides an opportunity to clarify the interpretation of SCT theory. By combining situational variables with cognitive variables, the Composite Model will offer a test of the power of cognitive beliefs to reflect the individual and society from which they form. An illustrated model of the Composite Model is found on page 3-6. Further discussion on each construct of the model is found in the remaining sections of the chapter.  Selection of the Most Suitable Theoretical Perspective Since older women may lack efficacy for physical skills and fitness exercise, assessing their perceptions of physical ability was believed to be important to the prediction of late life exercise behavior. Neither the Theory of Reasoned Action, nor the Health Belief Model assess these perceptions about efficacy. The self-efficacy element of SCT, as well as the theory's other self-  125 referent beliefs, appear to hold the most promise for the explanation of older women's exercise behavior. In addition, health locus of control is a logical addition to a framework which is attempting to explain a health behavior such as late life exercise. Dzewaltowski (1989b, 1990) has twice shown that Bandura's Social Cognitive Theory is a superior predictive model to the Fishbein-Ajzen model in the explanation of exercise in older adults. Dzewaltowski (1989b, 1990) has compared the ability of SCT with the Theory of Reasoned Action to predict exercise behavior over a seven-week period (1989b), and four-week period (1990), in college-age physical education students. The Theory of Reasoned Action explained only 6% (Dzewaltowski, 1989b) and 10% (Dzewaltowski, Noble & Shaw, 1990) of the variance in exercise behavior while two SCT variables predicted 16% of exercise behavior variance (Dzewaltowski, 1989b), and could explain 7% more variance after controlling for all the variance explained by the Theory of Reasoned Action (Dzewaltowski, et al., 1990). In a separate study attempting to explain female exercise behavior, intention scores emerged as the only significant predictor, accounting for only 9% of the variance (Wurtele & Maddux, 1987). Dzewaltowski concluded that "the theory of reasoned action did not account for any unique variance in exercise behavior over the social cognitive theory constructs" (1989b, p. 251), and that "the social cognitive theory constructs were better predictors of physical activity than those from the theory of reasoned action and planned behavior" (Dzewaltowski et al., 1990, p.388). A further reason to reject the Theory of Reasoned Action was that intentions to exercise, and planned behavior to exercise may not be particularly relevant to the elderly who may possess "constricted future expectations" and may •^Oa  126 (Rakowski & Hickey, 1980, p.287). Since the main goal of this study was to identify important determinants of current exercise behavior, Social Cognitive Theory was selected as the more suitable theory.  The Application of Social Cognitive Theory Social Cognitive Theory conceptualizes that environmental events, personal factors, and behavior all function as interacting determinants of each other. People can exert some influence over their life course by their selection of environments and construction of situations (Bandura, 1989). Social cognitive theory subscribes to a model of emergent interactive agency. Persons are neither autonomous agents nor simply mechanical conveyors of animating environmental influences. (Bandura, 1989, p.1175) This perspective of individual activity and reactivity. supports the central thesis that "self-referent thoughts mediate the relationship between knowledge and action" (Schunk & Carbonari, 1984, p.230). However, some experiences and contextual elements of one's life cannot be easily altered and are thought to make significant contributions to one's control beliefs. Therefore one's cultural context and personal situation are thought to alter the subjective value of an expected outcome and the subjective probability (or expectation) that a particular action will achieve that outcome. To date, all of the cognitive mechanisms from Bandura's social cognitive theory (1986) have not been applied in concert to explain exercise motivation (Dzewaltowski, 1989b, p.252). Furthermore, SCT has not been tested directly for its ability to explain exercise behavior alongside the cultural context and personal situation of the individual. A test of this interpretation is attempted in this study, since both situational factors and cognitive variables are all I • • II • I I •^•  •  127 regression analysis permits simultaneous analyses of multiple independent variables. If cognitive determinants account for all of the variance in exercise behavior, and situational variables do not compete for power of explanation, then the interpretation of Schunk and Carbonari (1984), that cognitive beliefs are adequate proxy for personal and environmental circumstances, will be confirmed. If socio-environmental factors offer unique explanation in addition to cognitive variables, then support will exist that human behavior may be affected by significant mechanistic forces. A third possibility exists. Situational variables may over-ride all cognitive explanation, meaning that sociocultural environment and personal situation are the most important controlling determinants of leisure-time physical activity. The latter finding would force a conclusion that there is little potential for behavioral intervention, since the situational environment would then be the only route to improving the activity patterns of elderly women. A theoretical framework was sought which would attend to the cognitive beliefs of the older woman about exercise. If a woman places little value on the health outcomes of exercise, feels unskilled in popular fitness pursuits, has few friends or little social reinforcement to be more active, perceives the risks to be of more import than the benefits, and believes that she cannot improve her health or life outcomes by her involvement, then she would not very likely be a physically active individual. On the other hand, a physically active woman may be one who wants to live a long an healthy old age, who feels that physical activity is a low-risk and sure way to promote this goal, who feels physically confident to participate in fitness pursuits, and who knows that she will be encouraged to do so by others. This is the multi-hypothetical stance governing  this  t  d  128  The Synthesis of Theory: The Composite Model Three main theoretical perspectives - Social Cognitive Theory, Social Epidemiology, and Health Locus of Control Theory provide important elements for construction of a Composite Model of exercise behavior. The model is "a composite" in the sense that the structure combines biological, social and psychological constructs found across the research to be potentially explanatory of a health behavior such as leisure-time physical activity. Moreover, the triangular model of ENVIRONMENT (situational attributes) - PERSON (cognitive beliefs) - BEHAVIOR (leisure-time physical activity in the past week) proposed by Bandura (see Chapter 2-78) is preserved. For ENVIRONMENT, the Composite Model provides a detailed socioenvironmental context; features such as age, health, ethnicity, education, socioeconomic status, family size, marital status, and work role are potential forces which can provide windows of opportunity or overwhelming barriers to human beliefs and behavior (see Chapter 2-39). Representing the PERSON, the Composite Model presents four key constructs of Social Cognitive Theory which have demonstrated predictive ability in health behavior research: 1) the incentive to take action, 2) positive and negative expectations about the outcomes or consequences of the action, 3) socioenvironmental cues which encourage the action to take place, and 4) self-efficacy to successfully undertake the action. In the Composite Model as it is applied to late life exercise behavior, the incentive to take action is the  motive to live  a long and healthy life. Self-efficacy is interpreted as adult movement confidence to undertake six fitness types of exercise. Perceived social support to exercise represents the "environmental cue" or reinforcement for late life  129 exercise. Outcome expectations are interpreted as  perceived risks and perceived  benefits of participation in fitness types of exercise. The model adds an important cognitive construct,  health locus of control, which has demonstrated  success in predicting preventive health behaviors. In addition to eight situational variables, and  childhood movement confidence  childhood social support are included in the model to represent past mastery  experiences and early situations of opportunity. If retrospective childhood measures predict late life exercise, tentative evidence will be provided that older adult exercise participation may be rooted in early development and socialization of people. Adding the early origins of efficacy and environmental cues were thought to theoretically strengthen the Social Cognitive Theoretical model for its application in explaining adult physical activity (Figure 3.0). In the next section, the sixteen constructs of the Composite Model are presented along with acknowledgement of the literature which supports their inclusion as explanatory variables of exercise behavior. The ten life situational variables are presented first, followed by the six cognitive variables. A summary of the theoretical framework concludes the chapter.  130 Figure 3.0 The Composite Model of Elderly Female Physical Activity  Situational Environment Age Health Education Work Role Marital Status Number of Children Cultural Background Socioeconomic Status Childhood Social Support Childhood Movement Confidence  Cognitive Beliefs Health Incentive Risk a Benefit Outcome Expectation Adult Movement Confidence Adult Social Support Health Locus of Control  Behavior Leisure-time physical activity  131  Life Situational Variables An hypothesis advanced to explain women's lack of participation in regular and vigorous exercise is that "personal and societal barriers or obstacles in the lives of women make it difficult for them to exercise" (Yoshida, Allison & Osborn, 1988). Dishman (1990) documents as many as 44 variables that are possible determinants of exercise behavior (Dishman, 1990, p.93). Under "personal attributes," education, white-collar occupation, past exercise participation, and perceived good health are positively linked to current activity behavior. Under "environmental factors," past family influences and school programs are predictive of physical activity. In another study, Yoshida and colleagues (1990) report that lack of time due to work, cost and access to programs, family size and family responsibilities are structural barriers to regular exercise for women (Yoshida, et al. 1988). Chapter 2-37 reviews the ten situational variables most likely to explain late life exercise behavior. In brief, the research literature has found that physically active women tend to:  1) be younger (Brooks, 1988; Ishii-Kuntz, 1990; Mobily, Lemke, Drube, Wallace, Leslie & Weissenger, 1987; Maddox & Eisdorfer, 1962; Ostrow & Dzewaltowski, 1986), not middle aged (Unkel, 1981), -  2) be single or widowed (Altergott, 1988; Brooks, 1988; Canada Fitness Survey, 1983; Fasting & Sisjord, 1985; Ishii-Kuntz, 1990); 3) be better educated (Brooks, 1988; Godin & Shephard, 1987; Health & Welfare Canada, 1988; Sallis, Haskell, Wood, Fortmann, Rogers, Blair & Paffenbarger, 1985; Unkel, 1981; Yoshida et al., 1988);  132 4) be employed or 5) have good income and higher socioeconomic status (Boothby, Tungatt, & Townsend, 1981; Brooks, 1988; Calnan, 1986; Eggers, 1988 (negative association with employment); Gale, Eckhoff, Mogel, & Rodnick, 1984; Health & Welfare Canada, 1988; Stephens, Jacobs & White, 1985);  6) be generally healthier and fitter (Boothby et al., 1981; Burckhurdt, 1988; Ishii-Kuntz, 1990; Kolanowski & Gunter, 1988; Thomas, S.P., 1990), or believe  they are in quite poor health (Morgan, Shephard, Finucane, Schimmelfing & Jazmaji, 1984; Rechnitzer, 1989);  7)be geographically / culturally / racially differentiated (Ishii-Kuntz, 1990; Stephens, Jacobs, Jr., & White, 1985; The Perrier Study, 1979);  8) have smaller family size (Fishwick & Hayes, 1989; Yoshida et al., 1988); and more freedom and time (Fasting & Sisjord, 1985; Yoshida et al., 1988); 9)have lifelong habits and competencies for physical activity (Fishwick & Hayes, 1989; Godin, Valois, & Shephard, 1987; Rikli & Busch, 1986); and, 10) have been encouraged at a younger age in physical activity (Greendorfer, 1983; Greendorfer, Blinde, & Pellegrini, 1986; Greendorfer & Lewko, 1978). Other important explanations for late life exercise are likely to come from psychobehavioral constructs of Social Cognitive Theory and Health Locus of Control Theory. Six cognitive beliefs were chosen for representation in the Composite Model and are discussed below. For more information on Social Cognitive Theory and Health Locus of Control, the reader should consult Chapter 2.  133  The Cognitive Variables In this study, the PERSON is linked to the behavioral variable (the criterion) of LEISURE-TIME PHYSICAL ACTIVITY (exercise level in the past week) through five beliefs about health and exercise: four SCT beliefs and health locus of control (Figure 3.1):  Figure 3.1 The Cognitive Variables related to Social Cognitive Theory and Health Locus of Control Theory  * HEALTH INCENTIVE (Incentive to Act)  * RISKS / BENEFITS (Outcome Expectations) * MOVEMENT CONFIDENCE (Efficacy expectations) * SOCIAL SUPPORT (Environmental cues) * HEALTH LOCUS OF CONTROL  The following section interprets the cognitive constructs as they are applied to this study.  Health Incentive Bandura's "incentive to act," behavioral goal or motive is represented in this study as the incentive to live a long and healthy life (Health Incentive). Setting goals for maintaining or improving health is considered to be a key determinant in explaining why adults might take up health-promoting exercise in late life. For example, individuals who highly value their health, and also value  134 the future effects of physical activity, are hypothesized to make  a significant  effort to maintain or initiate a more active lifestyle. "By representing foreseeable outcomes symbolically, future consequences can be converted into current motivators and regulators of behavior" (Bandura & Cervone, 1983). The capacity to exert self-influence by setting personal challenges and reevaluating one's own goal attainments provides an important cognitive mechanism of motivation (Bandura & Cervone, 1986). "Motivation through pursuit of challenging standards has been the subject of extensive research on goal setting" (Bandura & Cervone, 1986, p. 92). Motivation based on standards involves a cognitive comparison process. When people commit themselves to explicit standards or goals, the perceived negative discrepancies between performance and the standard they seek to attain create self-dissatisfaction that serves as a motivational inducement for enhanced effort. Activation of self-evaluative reactions by internal comparison requires both personal standards and knowledge about one's performance level. (Bandura & Cervone, 1986, p.92, 93). If this study finds that older women with health problems are more active than healthy women, this could be interpreted as a self-dissatisfaction with one's health. Social Cognitive Theory would hypothesize that falling short of one's personal health goals would then act as additional motivation for engaging in health-promoting forms of physical activity. Few studies have examined the hypothesis that values for sustained health may predict exercise behavior. One study has particular relevance to the healthvalue hypothesis. Petersen-Martin and Cottrell (1987) used the Rokeach Values Survey and the Martin Index of Health Behavior with 83 males and female students aged 17 to 49. Twenty-five percent of the sample ranked health as their most important, or second most important value, and 43% of the sample ranked health in their top four values. Petersen-Martin and Cottrell expected that people with higher self-concept might exhibit better health behaviors, but this outcome was  135 not supported. Only one significant difference was found between persons with differing values for health. "Persons who placed a high value on health exercised more than persons who placed a low value on health" (p = .006) (Petersen-Martin & Cottrell, 1987, p.8). For older women, the motivation to live a long and healthy life seems to be a logical prerequisite to pursuing health-maintaining behavior. Logically, women who feel that they have little reason to live much longer, or to expect better health, may have less incentive to achieve health-promoting levels of physical activity. Supporting the association between value for health and health behavior, Kristianson (1985) has found that respondents who reported good preventive health behavior in a mail survey, also valued health more than did those who reported poor preventive health behavior. She,warns that health value is more likely to be predictive of behavior involving a direct, rather than an indirect risk to health, such as drinking and driving, or wearing adequate clothing for the weather.  Outcome Expectations of Late Life Exercise While susceptibility to risk of illness and disease is under-rated by most people (Weinstein, 1984), perceptions about susceptibility to harm from exercise participation is common (Del Monte, 1985; Heitmann, 1982; Lindsay-Reid & Osborn, 1980; Monahan, 1986; Waller, 1985a, 1985b). Women in particular seem to downplay the benefits of physical activity and have heightened anxieties about vigorous exercise even though incidents of sudden death are almost universally a male phenomenon (Ragosta, Crabtree, Sturner & Thompson, 1984). The reader is referred  136 to Chapter 2 for extensive reviews about the known risks and benefits of physical activity for adults of all ages. Exercise is a complex, time-consuming and high effort behavior - one that at times requires discipline and commitment. Active adults may already have developed certain positive expectations about the value of exercise in their lives which is adequate compensation for the effort involved. Similarly, inactive adults may have developed negative outcome expectations that act as barriers to their participation. Social Cognitive Theory hypothesizes that beliefs about the expected positive and negative consequences of exercise would be important determinants of exercise participation. Bandura (1989, p. 1178) states that, people avoid potentially threatening situations and activities, not because they are beset with anxiety, but because they will be unable to cope with situations that they negard as risky. They take self-protective action regardless of whether they happen to be anxious at the moment. (Bandura, 1989. p. 1178) .  Explored in the Composite Model guiding this research study are the importance of the perceived health benefits and the perceived health risks of participating in fitness-type of exercise situations. In this study, older women are asked to evaluate their expected risks and benefits in six exercise settings which are reflective of community program offerings or at home fitness activities. SCT would hypothesize that the perceived risks and benefits, representing negative and positive expected outcomes of exercise, would be important determinants of exercise participation. More specifically, SCT would support the hypothesis that adults who perceive personal benefits from participation in exercise, and who perceive little risk of harm, would be more likely to be physically active. Part of this construct reveals the public and private knowledge that individuals have acquired about these activities and also reflects  their  137 perceived ability to participate safely. Even though the health benefits of regular exercise may be publicly known, several researchers claim that older women may feel particularly vulnerable to injury or exaggerate the risks to health in physical activity settings (Calnan & Johnson, 1985; Heitmann, 1982; Siscovick, LaPorte & Newman, 1985).  Perceived Benefits and Risks of Exercise Participation While there is little evidence to support the idea that older adults may be aware of the benefits of physical activity, participation in Seniors Games and general physical activity is on the rise. Interest in more intense activities seems to be growing as evidenced by various community and regional sport developments in whiCh older women and men are seeking strenuous physical challenges that require months of conditioning, technical skill, first-rate equipment and expert instruction (O'Brien Cousins & Burgess, 1992; U. of Agers, 1990). The perceived benefits of physical activity participation may be distorted for women who have been warned throughout their lifespan about the reproductive consequences that accompany physical exertion on the body (Vertinsky, 1990). The social justification for women to exercise has primarily been for weight control, perhaps not so much for health reasons as for beauty's sake. Traditionally, female beauty has been considered an important precursor to successful partnering with men, and successful partnering with men has been, until recently, the only route to elevated social status for women. Thus a female's greatest perceived benefit may be maintaining her "physical attraction" despite natural aging processes. Believing that her youth cannot be retrieved and with physical strain  138 ever more present, many older women may see little benefit in undertaking vigorous physical activity. Some activities may be perceived to be more beneficial for older adults than others, although it is doubtful that adults are necessarily attracted to programs solely on the basis of these differences. For example, participation in safely implemented high-risk activity is thought by programmers to be superior to low-risk activities in fostering cooperation, confidence, self-esteem and a sense of empowerment, but older adults may not necessarily agree (Alessio, Grier & Leviton, 1989; O'Brien Cousins & Burgess, 1992). Still, recreational programming for older adults has not been well researched in terms of the role of adventure and risk-taking on positive/negative outcomes. As yet, there is no way. to tell if increasing the challenges and interest level of older adults provides comparable increases in perceived benefits that would make participation worthwhile. Data providing information on the perceived (and actual) benefits and risks are nonexistent, even for the six most frequently utilized activities in the United States - walking, jogging, swimming, cycling, calisthenics and racquet sports. Empirical data about the most common activity for all ages, walking, is absent. Furthermore, dose-response, or how much exercise is associated with how much benefit/risk, needs to be explored in more detail. Such effects need to be explored at various points in the life span and the benefits and risks cannot be considered in isolation. It may be necessary to study them separately, but the overall effect of physical activity on the health of the population requires that both be known, both be studied with equal care, and that both be considered dispassionately. The potential overall beneficial impact of physical activity on health will be poorly served if activity patterns are recommended indiscriminately for all groups without regard for the subgroup-specific benefits and risks. (Powell & Paffenbarger, 1985, p.121)  139 Defining Perceived Risks of Exercise Participation According to Giovacchini (1983), "safe behavior" is defined as 'freedom from unreasonable risk or significant injury under reasonable foreseeable conditions of use'. He further defines "risk" as 'the chance of getting hurt, losing, failing, or placing one's self in a dangerous or hazardous condition'. This kind of definition, if applied in the sport and physical activity setting, may be interpreted to mean that every activity has one or more risks associated with it, or at least a situation of zero risk is unlikely in most human undertakings. Risk perceptions are thought to vary depending upon one's point of view, value system and personal priorities. As such, risk is meaningful only as it is self-defined, and therefore risk is as real as is the person's perceptions who is defining it. The problem with such self-defined risk is that highly motivated individuals are often unrealistically optimistic about their personal susceptibility to hazards while sedentary adults are often negatively biased about their personal odds (Weinstein, 1984). These "why it won't happen to me" perceptions appear to be highly resistant to change, even when new scientific information is provided that should counter that view. Estimates about personal risk are likely carved out of past experience and one's perceived competence in a given situation. For example, a physically capable older woman may be wary of carrying out her daily walk in an unfamiliar city while another woman may judge the traffic and pollution to be more detrimental than the benefits of a walk. Still another may worry that getting breathless or provoking angina is her primary risk or may worry about falling on icy pavement if the weather is inclement. These are examples of risk perceptions that create self-defined barriers to a regular walking program.  140 The fact that everyone faces risks and hazards does not mean that we are all accurately evaluating risk and making good decisions about personal safety. On one hand, the hazards confronting people in daily life are often externally controlled, and on the other, various risks are difficult to compare. As an example, just "being old" could be considered to be a risk. "Being old, for instance, is risky in that it introduces potential costs and dangers, but the attitude to 'being old' is likely to be of a different order from attitudes to climbing stairs" (Brearley, Hall, Jeffreys, Jennings & Pritchard, 1982, p.53). The present activities of older adults are thought to be a reflection of their beliefs about expected outcomes and self-perceptions of ability. For example, about 70% of older Canadians claim to be walking and gardening, but less than 20% claim to be swimming or cycling (Stephens & Craig, 1990). Of all age groups, older adults are more likely to exercise alone (about 60%) and at home (40%). The reasons for these findings are not clear. Perhaps older adults are more inclined to walk and garden because their energies would then be applied constructively to no-cost transportation or to accomplishing tasks around their home. A number of possible psychological, social and situational barriers may be operating to prevent many older people from participating in cycling and swimming, or in activities outside of their homes. Many older adults do not own swimsuits or bicycles; for many, lycra swimsuits and 21 speed dirt bikes are too expensive and beyond their needs. Swimming requires convenient pool facilities while cycling requires paved terrain and good weather. In Canada, neither the facilities, nor the weather, are often suited to participation. The elderly probably associate greater risks with these activities. They may fear traffic and the speed needed on roadways and possibly are concerned about their ability to balance safely on bicycles; in swimming, adults must own  141 acceptable attire and be prepared to undress in public places. Older women  are  vulnerable to stares from young people in the locker room and on the pool deck. Further they must be confident about moving in water, even if the water temperature is below their comfort level; they must feel able to swim well if that is the perceived expectation. With severe Canadian winters the norm in most parts of the country, these two activities are not likely to be very inviting for older adults as year-round activities. For older adults, understanding about the need for daily exercise is apparently improving, but overall, specific knowledge about the benefits and risks of different exercise behaviors is lacking. When asked about why they exercise, adults often can only reply that they "feel better" or "look better" (Canada Fitness Survey, 1982). Adults of all ages are warned to consult their physician before increasing activity or taking up any new activity, and therefore the health-promoting message may be drowned out by this acknowledgement of what is likely a very small risk. The conscious effort by individuals to assess the nature and scale of the possible hazards in order to make self-protective decisions is likely made "in an environment composed of all gradations of ignorance and fear" (Zuckerman, 1983, p.v). Physical activity has been found to become less structured and more casual with age (Stephens & Craig, 1990). Ironically people may become more and more independent from supervision in the exercise setting at the very life stages where they may also feel that they are at increased risk in exercise. However, casual participation does guarantee a sense of personal control over the "pace" of an activity and likely reduces perceptions of risk by removing social  pressures to "keep up" (O'Brien & Burgess, 1992).  142 Females participate at about half the rate of males in organized sport, and participate at about twice the rate of males in supervised activity (Stephens & Craig, 1990). This finding suggests that social support for safe participation in activity settings is required more by women while males seek support for activity in higher risk settings. Yet, considering body fat patterns, males are at greater risk than women. Men are much more likely to weigh too much for their height and to have abdominal rather than low-trunk fat which place them at increased risk of cardiovascular disease (Stephens & Craig, 1990). Still, men over the age of 45 are more likely than women to report obtaining encouragement for activity participation and are more likely to report a sense of control over their ability to participate in regular activity. Females, on the other hand, are more likely to report family pressures and a lack of energy as significant barriers to being active (Stephens & Craig, 1990). Sofalvi and Airhihenbuwa (1992) have researched the impact of the media on public beliefs about health issues. The media has been keen to report incidences of sudden death in the exercise setting, such as the death of exercise advocate, James Fixx. Educators concerns for contraindicated exercises combined with media interest on rare fatalities during exercise shows why certified fitness leaders tend to reinforce participation in supervised classes and give reason for reluctant exercisers to avoid physical activity altogether. Behaviors based on risk assessment and behaviors related to health enhancement both seek to avoid personal harm. But while physical activity participation may have biological relevance to self-protection from disease, many individuals participate for entirely different social or psychological reasons (Thuen, Klepp, & Wold, 1992). Clearly specific, scientific information on the Is I •^•^  •!  143 in certain populations vulnerable to joint and cardiac stress, is badly needed. Without this information, misinformed outcome expectations will continue to undermine social support initiatives for increasing activity in aging adults.  Age and Risk Perception A common perception exists that the elderly are afraid or incapable of acting for themselves. They are considered helpless and hesitant with little to offer the community. Sadly, many behave according to ageist expectations and social labels and perpetrate the self-fulfilling prophecy (Edgerton, 1986). Although older people are stereotyped as being more cautious, some research does not bear this out (Brearley, Hall, Jeffreys, Jennings, & Pritchard, 1982). In Littlewood's (1989) study on the elderly, 19% agreed with the statement that "regular exercise can do you more harm than good". While such data only scratches the surface of the risk perceptions of aging adults toward the risks and benefits of exercise, these findings do lend support to the idea that lack of involvement in health-promoting exercise may indeed have much to do with people's overall assessment of benefit and harm. This study will be among the first to specifically assess the perceived risks and benefits of six forms of late life exercise in women as described in the next chapter on methodology.  Movement Confidence (Self-efficacy) Among the mechanisms of personal agency, "none is more central or pervasive than people's beliefs about their capabilities to exercise control over events that affect their lives" (Bandura, 1989, p.1175). Perceived self-efficacy is a  144 cognitive factor which appears to play an influential role in personal agency in ways that affect motivation. As such, self-efficacy, or "people's judgements of their capabilities to organize and execute courses of action required to attain designated types of performances" is the most studied component of social cognitive theory (Bandura, 1966, p.391). Self-efficacy beliefs affect thought patterns in self-aiding or selfhindering ways. A strong sense of efficacy is required to remain task-oriented in the face of judgemental failures. "Those who have a high sense of efficacy visualize success scenarios that provide positive guides for performance" (Bandura, 1989, p.1176). Thus self-referent perceptions of efficacy are at least partly responsible for the kinds of challenges which people choose to undertake, how much effort they w.11 spend on that activity, and how long they will persevere in the face of obstacles (Bandura, 1986, 1989). "When faced with difficulties, people who are beset by self-doubts about their capabilities slacken their efforts or abort their attempts prematurely and quickly settle for mediocre solutions, whereas those who have a strong belief in their capabilities exert greater effort to master the challenge" (Bandura, 1989, p.1176). Maintaining motivation for life pursuits is thought to be fostered by adopting challenges in accordance with one's perceived capabilities and having informative feedback that supports these perceptions of capability. Thus, experiences of mastery strengthen perceptions of efficacy. Even cognitive imagery (mental simulations) in which individuals visualize themselves competently executing an activity can enhance performance. Therefore, perceived self-efficacy and cognitive simulations affect each other reciprocally. A high sense of efficacy fosters cognitive images of effective actions, while successful experiences with efficacious courses of action strengthens self-perceptions of efficacy (Bandura & Adams, 1977).  145 Schunk and Carbonari (1984) acknowledge that competence is conceptually similar to self-efficacy, but ability is only one of the possibilities along with effort, luck and task difficulty that can explain success or failure of personal actions. Although those who view themselves as having high ability for a task are also apt to feel efficacious for performing it, simply possessing the ability to perform a task does not guarantee a high degree of self-efficacy, nor is competent behavior likely to occur without adequate incentives (Bandura, 1977a, 1977b). The "self-efficacy" element of Social Cognitive Theory has been highly successful in the explanation of a host of health and behavioral outcomes even though the tools for assessing efficacy have varied from study to study (Strecher % DeVellis, Becker, & Rosenstock, 1986). People must have a robust sense of personal efficacy to sustain the persevering effort needed to succeed in physical activity, exercise and sport settings. Self-efficacy, the belief that one is able to perform a specific activity, is the most powerful and statistically significant correlate of both walking and vigorous exercise among a number of adult groups (Hofstetter, Hovell, Macera, Sallis, Spry, Barrington, Callender, Hachly & Rauh, 1991). Of interest to this study, perceived selfefficacy in the physical activity setting has predicted:  1) positive mental health, self-esteem, and stress management (Davis-Berman, 1989; Holahan & Holahan, 1987; Holahan, Holahan & Belk, 1984; Rogers, 1987; Ryckman, Robbins, Thornton, & Cantrell, 1982; Sonstroem, 1976; Wells-Parker, Miller, & Topping, 1990);  2)pain management (Litt, 1988);  146 3) over exertion (Ewart, Stewart, Gillilan, Kelemen, Valenti, Manley & Kelemen, -  1986; Ewart, Stewart, Gillilan, & Keleman, 1986) and strenuous performance (Bandura & Cervone, 1986); 4) skilled motor performance (Brody, Hatfield & Spalding, 1988; Crawford &  Griffin, 1986; Feltz, 1988; Griffin & Crawford, 1989; Griffin & Keogh, 1982); 5) competitive sport performance (Barling & Abel, 1983; Feltz, 1988; McAuley &  Gill, 1983; Roberts, Kieiber, & Duda, 1981; Weinberg, Gould, & Jackson, 1979; Weinberg, Yukelson, & Hackson, 1980; Weiss, Wiese, & Klint, 1989); 6) explanations of competitive sport performance (Duncan & McAuley, 1987); 7) eating and other preventive health behavior (Sallis, Pinski, Grossman,  Patterson, & Nader, 1988; Waller & Bates, 1992); 8) attendance at a fitness program (Howze, DiGilio, Bennett, & Smith, 1983); 9) physical fitness and activity behavior (Atkins, Kaplan, Timms, Reinsch, &  Lofback, 1984; Dzewaltowski, 1989; Ewart, Taylor, Reese & DeBusk, 1983; Hofstetter, Novell, Macera, Sallis, Spry, Barrington, Callender, Hackley, & Rauh, 1991; Kaplan, Atkins, & Reinsch, 1984; Marcus, Selby, Niaura, & Rossi, 1992); and 10) indices of physical fitness (Neale, Sonstroem, & Metz, 1970; Thornton,  Ryckman, Robbins, Donelli & Biser, 1987). Furthermore, efficacy in physical activity settings may be directly observable by others (Keogh, Griffin, & Spector, 1981), enhanced by experience and practice (Hogan & Santomier, 1984; Kaplan, Atkins & Reinsch, 1984; McAuley, Courneya & Lettunich, 1991) and even generalizable to other performance related settings (Hogan & Santomier, 1984). Efficacy expectations for exercise situations are stronger for males and younger adults (Duda & Tappe, 1989).  147  Social Support to Exercise Of interest to this study are the socializing forces or cues that might be perceived by an older woman relative to her exercise behavior -- perceptions of endorsement, approval, advocacy or encouragement for physical activity. In broad terms, these socializing forces could reflect the processes of the family and the community, as well as the larger forces of society, in influencing differing roles, age norms, behaviors and lifestyles of men and women (Hobart, 1975; McPherson, Curtis, & Loy, 1989). Expectancies about how others may view a behavior brings into play a socially normative variable called social support affecting whether one might participate in late life physical activity, exercise and sport. In this study, the local environment is considered to be a source of immediate social supports, such as friends and family. Social feedback from friends, family, and significant others provide cues for reinforcement and discrimination (Perry el al., 1990). In this respect, social support can be thought of as a "social efficacy to be physically active." For women, affiliative benefits have been emphasized as important personal incentives for physical activity involvement (Duda & Tappe, 1989). Social support to exercise in late life is thought to be an important "environmental cue" for older women since some of those who plan to meet their fitness requirements with their choice of daily exercise may well experience: 1) disapproval from their spouse (Andrew, et al., 1981; Dishman, 1986; Perusse, LeBlanc, & Bouchard, 1988; Snyder & Spreitzer, 1973; Stephens & Craig, 1990; Tait & Dobash, 1986); 2) lack of peer interest and companionship (Hauge, 1973);  148 3) discouragement by the immediate family (McPherson, 1982; Spreitzer & Snyder, 1973; Spreitzer & Snyder, 1983); and 4) inadequate encouragement from physicians (Dishman, 1986; Gray, 1987; Powell, Spain, Christenson, & Mollenkamp, 1986; Wechsldr, Levine, Idelson, Rohman & Taylor, 1983).  Health Locus of Control Combining Health Locus of Control Theory with Social Cognitive Theory has been advocated as an important theoretical step (McCready & Long, 1985). Despite inconsistent findings in the literature, health locus of control is theoretically important to include in the model guiding this study for this reason: an older woman may be highly motivated to live a long and healthy life, but if she perceives she has little control over her health (external HLC), she is unlikely to take on a health-promoting behavior such as exercise. A fatalistic attitude was uncovered in a survey by Littlewood (1989) in which 69% of the elderly agreed with the statement "there is no point worrying about a heart attack - you can't prevent it". Loss of appetite (43%) and constipation (40%) were as much associated with ageing as was mental illness (42%). If elderly people see certain health events as inevitable functions of aging, they are therefore less likely to act on them. With application to the study of late life exercise behaviour, health locus of control theory would hypothesize that older adults who keenly value their health and longevity, and who believe that they have some degree of positive control over their health by  149 participating in exercise, would be more likely to be found engaging in late life physical activity.  Summary of the Theoretical Framework A Composite Model of explanation has been proposed for this research which explores the situational and cognitive determinants of late life exercise in women over the age of 70. The Composite Model provides some advantages over other models in that: 1) a comprehensive assembly of beliefs derived from Social Cognitive Theory and Health Locus of Control Theory are examined. 2) a comprehensive list of personal and situational variables are explored. 3) several types of social support have been incorporated. A detailed explanation of the application of this model are provided in the next chapter on study design.  150  IV. DESIGN OF THE STUDY  This chapter includes the detailed methodology used to test the Composite Model and includes: survey questionnaire construction; measurement and validation of the outcome variable; discussion of the interpretive variables; measurement and coding of life circumstance variables and cognitive mediating variables; the pilot study; the sampling procedure; the data collection protocol; a second sampling procedure; and procedures used in the data analysis.  Survey Questionnaire Construction Description The survey instrument was a booklet composed of questions which were designed to assess the many constructs of the Composite Model: 1) the ten situational variables (age, marital status, education, economic status, health, school location, work role, number of children, childhood movement confidence and childhood social support for physical activity); 2) the five theoretical elements (health incentive, perceived risks and benefits, adult movement confidence, adult social support, and health locus of control); and 3) measurement of the leisuretime exercise in the past week. Questionnaire construction was guided by Statistics Canada's Development and Design of Survey Questionnaires (Platek, Pierre-Pierre, & Stevens, 1985) with consideration given to the older adult reader wherever possible. For example, a larger than normal reading size font was used to assist readers, some of whom were likely to have visual difficulties. The questionnaire, stapled inside a bright pink cover, included a title page with an explanation of the study, a  151 carbon-backed consent form and 22 pages of questions, well-spaced, to facilitate reading (see Appendix A). The questionnaire conformed to the requirements of the Human Ethics Review Committee of the University of British Columbia (see Appendix B). A number of instruments were included in the questionnaire which have been widely used. Their validity and reliability are reported later in this chapter. Instruments designed for this study have a reported test-retest reliability and concurrent validity value. The outcome variable,  exercise status in the past week, was assessed using  a newly designed instrument. The Older Adult Exercise Status Inventory combines a number of positive attributes from other seven-day recall instruments used. A review and critique of the literature pertaining to these instruments can be found at the end of the Review of Literature. The inventory provides more detailed assessment of physical activity than most instruments and accounts for the unique activities of older adults as recommended by Washburn, Jette, and Janney (1990). To facilitate comparisons among studies, Gordis (1979) recommends uniform wording of questions. Standardized questions such as those used on the Canada Fitness Survey (1983) and the General Health Survey (1985) assessed life  situational measures on items such as age, PARQ health symptoms, number of  medications, self-rated health status, marital status, and education. In addition to these, I designed questions on cultural background:  country of main schooling,  socioeconomic status, employment activity, number of children and childhood social support to be physically active. I assessed childhood movement confidence using the format of a validated stunt movement confidence inventory and creating ment cdllyd "M uvemenL C onfidence as a child.".  152 I designed four of the six instruments used for the cognitive measures of the study: health incentive, perceived risks with exercise, perceived benefits  with exercise, and social support to be physically active. Adapting the format of a stunt movement confidence inventory, I created a new instrument called "Movement Confidence Now" to assess adult movement confidence. Health locus of  control was measured using a validated instrument which has been widely used in health promotion research.  The Older Adult's Exercise Status Inventory Description  Exercise status (ENERGY) in total kilocalories was the criterion measure used for this study. For this study, a seven-day recall instrument is designed to assess the type, duration, frequency and level of intensity of the physical activities of older adults (See Appendix A). The design brings together the strengths of a number of instruments which had been used in prominent epidemiological research projects (Blair, 1984; Paffenbarger, Hyde, Wing & Hsieh, 1986; Canada Fitness Survey, 1988; Taylor, Jacobs, Schucker, Knudsen, Leon & Debacker, 1978). In general, the Older Adult's Exercise Status Inventory (OA-ES') used in this study compromises some instrument brevity for increased detail and rigor than in other studies which have used the seven-day self-report. The OA-ES' is among the first physical activity assessment tools to quantify the specific exercise patterns of adults over age 70. The Canada Fitness Survey (1988) used an inventory called "Physical Activity in Your Spare Time" for recording the activities of individuals over an entire year. This inventory excludes many activities where seniors are active  153 such as curling, line dancing, horseshoes and darts. Another criticism of this inventory is that recall over a full year may be too difficult for adults of any age. Still, the column-row style combined with a seven-day recall format appeared to have merit. Compared to the CFS Inventory and the Physical Activity Index (PAO used by Paffenbarger, Wing and Hyde (1978), the OA-ES' is age-relevant and more comprehensive in documenting types of exercise participation. As with the CFS Inventory, the OA-ESI examined exercise as a form of leisure behavior, and therefore did not include domestic work, nor employed work activity as part of the weekly energy estimate. The work energy of women on domestic tasks has, unfortunately, received little interest by researchers. At least two studies claim that domestic activity accounts for much of women's daily physical activity and must be documented in the future (Cauley, LaPorte, Sandler, Schramm & Kriska, 1987; Mattiasson-Nilo, Sonn, Johannesson, Gosman-Hedstroem, Persson & Grimby, 1990). The OA-ESL is a two-page inventory which prompts subjects with categories organized in columns by the seven days of the week and organized in rows by a list of 38 physical activities. The 38 activities were considered age-appropriate since they were chosen for the list from personal observation and experience with the activities available to older women in Edmonton and Vancouver. The 38 activities provided a comprehensive list of leisure-time physical pursuits likely to closely reflect older adult's types of exercise involvement. Two open categories called "Other" accommodated any other activities that were not already included on the main list. These exercise categories acted as memory prompts and were listed alphabetically from "aerobic fitness class" to "walking (no sweating)". Aquacize activity was subdivided into "vigorous" and "gentle", cycl ing,  gardening, 1v yymy and wdlkiny were subdivided into "sweat-  154 inducing" and "no sweating". The purpose of the sub-categories was to reduce error in estimating the intensity of a particular activity and thereby improve the estimate of the criterion variable, weekly energy spent on exercise. To aid precision, subjects were asked to report the "time spent in minutes" for each activity on each day.  Reliability of the OA-ESI As a test of reliability, the 0A-ESI was administered twice in a four-week period to 16 older women from Edmonton (mean age of 67). Pearson Product Moment correlations (rp)were used for interval and ratio data; Spearman Rank order correlations (r e ) were used for ordinal variables such as school location, marital status, education, and work role. Out of several self-reported activity measures, only mild exercise demonstrated poor reproducibility (r p = .114; n.s.) and undermined the reproducibility of the total amount of exercise reported (r p = .340; p = .198). Moderate exercise was reported more consistently (r p = .756; p <.001) while vigorous exercise was only moderately reliable (r p = .505; p <.05).^(See The Pilot Study). Ironically these inconsistent findings for mild exercise patterns provide support for construct validity. The initial survey and retest were conducted in the four weeks of September during which the pilot sample entered various fall sport and fitness programs. Thus substantial changes in the nature of physical activity occurred over the four-week pilot study. Many of the women in the pilot study were re-initiating participation in structured, supervised and vigorous forms of exercise and forfeiting some of their less structured and milder summer activities at that time of year. In this regard the piloted questionnaire was sensitive to these changes in activity patterns of the 16 women. In doing so,  155 self-reported activity as measured by the OA-ES' appeared to demonstrate weak reproducibility, but, at the same time, validated the known changes in participation. This phenomenon provided a degree of confidence that the sevenday recall had adequate sensitivity to be administered to a larger sample.  Validity of the OA-ES' The Older Adult's Exercise Status Inventory (OA-ES') used in this study integrates the best assessment strategies of the other validated survey instruments for physical activity and thus is considered to retain adequate validity. Moreover, as a simple record of daily exercise involvement, the OA-ES' has demonstrated adequate construct validity in the pilot study by representing what was known to be seasonal changes in exercise patterns (see the above "Reliability of the OA-ESI"). Concurrent validity is demonstrated by examining the correlations of ENERGY in kilocalories with other concurrent activity indicators on the same OA-ES' survey. For example, ENERGY (total of weekly kilocalories based on self-report) had a correlation of rp = .403 with LIFESTAT, a subjective question about lifelong activity status similar to that used by Godin, Valois, and Shephard (1987). This question was worded, "How would you describe your physical fitness activity over your entire life course?" (See page 4 of the questionnaire in Appendix A). The five response choices were reduced to two for dummy coding: 1 = Currently active, 0 = Not currently active. To another subjective question used by Godin & Shephard (1982), "How often did you participate in vigorous physical activities long enough to get sweaty within the past four months?", the present ENERGY score correlated r p = .411 (p <.UUU1).  156 Exercise activity has been shown to decline with age and poorer health status. In this study, ENERGY supported this relationship with an r p = -0.226 (p <.0001) with age and rp = .222 (p <.0001) with a positive health self-rating. These relationships provide evidence for construct validity.  Use of the OA-ESI The F.I.T.T. formula is used in Canada by fitness and lifestyle counsellors to account for frequency, intensity, type of exercise and time spent on exercise. All four variables are considered useful measures of exercise involvement and therefore were considered to be essential measures in this study. Thus the Older Adult's Exercise Status Inventory specifically assessed: type of exercise reported, intensity of exercise reported (MET units), total duration of exercise (hours of activity), and total number of exercise sessions (frequency). The type of exercise reported was used two ways: first to provide descriptive information on the activity preferences of older women, and second, to estimate more accurately the energy expenditure of each type of activity. "Amount of exercise in the past week" was calculated using reported metabolic charts giving MET units for physical activities (Cantu, 1980; Passmore & Durnin, 1955; Taylor, Jacobs Jr., Schucker et al., 1978; Wilson, Paffenbarger, Morris & Havlik, 1986). The MET is the ratio of working metabolic rate to resting metabolic rate and is a convenient method of expressing energy expenditure (Sallis et al., 1985). It can be thought of as the ability of an individual to tolerate multiples of their resting energy level (Astrand, 1992, personal communication). One MET = 1 kcal/kg/hour, an equivalent of one kcal of energy expended by d  60 kg. person sitting for one minute. MET units account for the intensity of  157 the activity, the duration of the activity as well as the body weight of the individual (if they are not too different from 60 kg.). For sake of convenience, many studies assume an average body weight of 60 kg., meaning that the average individual, sitting at rest, spends about 60 kilocalories per hour or 1.0 kilocalorie/kg./hour, or 1.0 MET. From the information provided on the OA-ESI, a number of dependent measures were identified: total energy in kilocalories spent on reported exercise in the past week (TOTKCAL); total kilocalories spent on reported exercise but adjusted for individual body weight (ENERGY); total kilocalories spent in three intensity categories of exercise, from mild (<4 METS), and moderate (4 to 5.9 METS) to vigorous forms of exercise (>6 METS). These three exercise intensity measures were called MILDKCAL, MODKCAL and VIGKCAL. Also the total number of hours of activity (ACTHOURS) and total number of separate exercise sessions (TOTSESS) over the seven days were counted. There is some debate whether the weight of an individual should be used in the calculation of energy expenditure, since the MET unit is meant to be a metabolic ratio, independent of body weight. The work/rest ratio method assumes that a task performed by a heavy person raises metabolic rate to the same extent as the same task performed by a person weighing less, even though the caloric expenditure might be different (Reiff, Montoye, Remington, Napier, Metzner & Epstein, 1967). In the Five City Project, researchers suggested that the measurement of exercise in "kilocalories per kilogram per day was not an acceptable measure for overweight populations" (Sallis et al., 1985, p. 95). Being overweight added to the energy estimate of exercise and was considered to negatively affected the reliability of their self-report data.  158 In the present study, however, the MET units were calculated according to individual body weight (according to personal advice from Dr. R. S. Paffenbarger, Jr., April 19, 1991). Thus the duration of each activity reported in minutes was recorded along with the individuals's body weight (kg) so that total kilocalories spent on exercise accounted for portions of an hour as well as individual differences in body size. Where the reported MET estimates differed in the literature, the more conservative estimate was used. The MET unit was multiplied by the individual's reported participation time in hours over the seven days  for each activity and  adjusted for body weight in kilograms.  ACTIVITY STATUS (KCAL) = Duration x Intensity x Body Weight = X minutes/60 (hours) x MET unit (kcal/kg/hr) x weight (kg) = kilocalories  Totals for mild, moderate and vigorous exercise categories were calculated. All three categories were then summed, thereby providing a seven-day Exercise  Status (ENERGY) measure in the form of kilocalories spent on exercise in the past week.  EXERCISE STATUS (ENERGY) = Reported Mild Activity + Moderate Activity + Vigorous Activity  The MET units used for the 3R activities in this study arc in Table 4.0.  159 Table 4.0  Metabolic Units of 38 Activities Activity Type Aerobic Fitness Class Aquafit/Aquacize Class (Vigorous) Aquafit/Aquacize Class^(Gentle) Badminton Bicycling outdoors Bicycling indoors^(sweat-inducing) Bicycling indoors^(no sweating) Bowling^(any kind) Calisthenics Canoeing or kayaking Curling Dancing^(Square,^tap,^folk) Dancing^(Ballroom,^ballet) Dancing^(Line,^Hawaiian) Darts,^Billiards,^Pool Gardening^(sweat-inducing) Gardening^(no sweating) Golf Gymnastics or rhythmics Hiking hilly terrain Horseshoes Jogging^(sweat-inducing) Jogging (no sweating) Rebounding (trampoline) Rope skipping Rowing (machine) Skating^(ice or roller) Stair climbing for fitness Stretching exercises Swimming^(gentle) Swimming^(non-stop lengths) Table Tennis Tai^Chi Tennis Walking^(sweat-inducing) Walking^(no sweating) Weight Training Yoga  MET Unit' 6.0 7.0 4.0 5.5 4.0 6.0 4.0 3.5 4.5 3.5 3.0 6.0 5.5 5.0 2.5 6.0 4.0 5.0 6.0 7.0 3.5 8.0 6.0 6.0 8.0 6.0 7.0 8.0 3.5 5.0 10.0 4.0 3.5 6.0 4.0 3.0 6.0 3.5  The metabolic estimates were compiled using reported metabolic charts as described on page 4-7.  160  Interpretive Variables A number of questions were asked to provide support and understanding of the criterion variable. These were as follows:  Normal Activity Level (TYPICAL) Subjects were asked "How typical was this past week in terms of your normal activity level?" ( ) more activity than typical ( ) quite typical ( ) less activity than typical These were dummy-coded into "typical week" (0) or "not a typical week" (1).  Change in Activity Level (CHANGE) A second question assessed changes in level of physical activity in the past five years. Response choices were: ( ( ( ( (  ) ) ) ) )  Significantly decreased Somewhat decreased Not changed Somewhat increased Significantly increased  With dummy-coding, responses which indicated no change or an increase in activity were labelled "no decrease" (0) and "decreased" (1).  Participation in the Past 4 Months (PAST4MON) A question from Gaston Godin's (1982) survey instrument (unnamed) from the School of Physical and Health Education at the University of Toronto was included. Since a number of studies have reported stronger relationships between concurrent criteria when subjects self-report more vigorous forms of exercise,  161 it was logical to include a question which captured such vigor. Subjects were asked "How often did you participate in vigorous physical activities long enough  to get sweaty with the past four months?" Response categories were: ( ( ( ( ( (  ) ) ) ) ) )  not at all less than once a month about once a month about 2 to 3 times a month about once a week two or more times a week Godin, Jobin and Bouillon (1986) reported on a concurrent validation study  for this question on 32 male and 29 female volunteers aged 19 to 66 years.  A two-  week test-retest reliability coefficient was .64. Values of maximum oxygen intake, body fat and muscular endurance, expressed in percentiles of appropriate age and sex categories, were used as concurrent criteria validity. Correlation coefficients between reported physical activity in the past four months were r = .38; p <.001 for maximum oxygen intake, r = .43, p <.01 for body fat and r = .54, p<.001 for muscular endurance. This is in agreement with Siconolfi and coworkers who also demonstrated that a simple "sweat" question taken from Paffenbarger's Physical Activity Index Questionnaire had concurrent validity of r = .60 with self-reported physical activity over the past week (Siconolfi, Laseter, Snow, & Carleton, 1985). In the pilot study, no relationship was found between test and retest of this question on 16 older Edmonton women (r s = .048; p= .861). The reliability appeared to have been undermined by substantial changes in exercise patterns due to the time of year.  162  Life Situational Measures Introduction The role of an individual's life circumstances in creating opportunities and incentives, and in creating obstacles and barriers, are recognized as possible determinants of exercise behavior. The literature reviewed in Chapter 2 suggests that habitual physical activity patterns are likely if the individual is male, younger, economically secure, of a higher educational level, of lower body mass and in good health. Less is known about the role of ethnicity and cultural background, marital status, family size and domestic/employment status although they are thought to be circumstances that are likely to be important at certain life stages in explaining activity patterns. In the following section, the contextual variables used in this study and the instruments chosen to measure them are identified along with their eight-character computer label in brackets. Test-retest Pearson correlation coefficients are also included from the pilot study on 17 older women.  Age (AGE) Age was obtained by subtracting the year of birth from 1990 as reported in the question "In what year were you born?" (Test-retest rp (AGE) = .998; p<.0001).  Culture (SCHOOLOC) Cultural background was obtained from  a checklist of countries which  respondents used to reply to the question, "In which country did you complete mnct of your schooling as a child?"  163 (1) Canada (2) United States (3) Britain (4) Japan (5) Germany (6) China (7) Italy (8) Scandinavia (9) Other Culture was dummy-coded as "English-speaking country" (1) which included those schooled in Canada, Britain and the U.S. and "foreign language speaking country" (0) which included all other countries. Test-retest r s (SCHOOLOC) = .926 (p<.0001).  Socioeconomic Status (COMPSES) Three questicins were asked: 1) Do you feel financially secure for the remainder of your life? (SECURE) 2) Are you able to handle unexpected expenses with no worry? (NOWORRY) 3) Which of the following financial assistance do you receive? (INCOME) ( ) None ( ) Guaranteed Income Supplement (GIS) ( ) Spouse's Allowance/ Widowed Spouse's Allowance To the first two questions, respondents answered "yes", "not sure" or "no". These were dummy coded as "Yes" = 1, and "Not sure" or "No" = O. The women reporting receipt of the Guaranteed Income Supplement (GIS) and or Spouse's Allowance were expected to have more marginal financial resources since their economic status would have already been determined by the Government of Canada. Dummy coding for question 3 was: "No GIS" = 1; GIS/Allowance = O. Test-retest reliabilities were: rs (SECURE) = .778; p<.0001; r s (NOWORRY) = .763; p<.001; r s (INCOME) = .562; p<.03.  164 These three economic indicators were standardized and pooled using principal components analysis to provide a composite indicator of socioeconomic status:  COMPSES = .867 (No financial worry) + .866 (feeling secure) + .364 (no GIS) / 1.637 (eigenvalue).  Marital Status (MARITAL) Five standard categories were provided for marital status: single (never married), common-law partner, married, widowed, and separated/divorced. Testretest r s (MARITAL) = 1.0. These were re-coded into dummy variables of "partnered" (1) and "unpartnered" (0).  Education (EDUCATIO) Eight forced-choice categories covered various levels of years and type of schooling from: 1) 2) 3) 4) 5) 6) 7) 8)  No schooling Grade 1 to 4 Grade 5 to 8 Some high school Completed high school Business or trade school Some university or college University or college degree(s)  Test-retest r s (EDUCATION) = .797; p<.0001. These eight levels were dummy-coded into "less than highschool graduation" (0) and "at least highschool graduation" (1)  165  Work Role The survey asked: "What kind of work situation best describes you from age 35 to 65? (Pick only one). (0) No paid employment (1) Part-time or intermittent full-time employment (2) Steady full-time employment The test-retest r s (WORKROLE) = .886; p < .0001. Dummy coding scored neveremployed women a (0), while women reporting part- or full-time employment scored (1).  Family Size: Number of Children (CHILDREN) The question "Which domestic situation best describes your adulthood?" provided choices of "on my own", "homemaker, no children", and "Mother of children". A domestic role variable (not used in the regression analysis) was coded "mother" (1) or "no children" (0). The number of children was the number reported. Test-retest r p (CHILDREN) = 1.0.  Health Variables  Self-Rated Health (HEALTH) In this study respondents were asked to describe their current state of health from (1) "poor", (2) "fair" (3) "good" to (4) "excellent". Ratings ranged from 1.0 to 4.0. Test-retest reliability for self-rated health of 16 older Edmonton women in the pilot study obtained an rp (HEALTH) = .506; p<.046. The scale, while considered reliable and valid in younger populations, may be less useful in the very old. For example, some of older adults might increase activity levels to combat a known disease, while others might significantly  deck-pace Activity to "reserve their strength".  166  Physical Symptoms (SYMPTOMS) The Physical Activity Readiness Questionnaire (PAR-Q) is a yes-no screening device used by Fitness Canada to eliminate high-risk exercisers. Five questions evaluate physical symptoms of "heart trouble", "frequent pains in your heart and chest", "often have spells of severe dizziness", "doctor has said your blood pressure is too high" and "other good physical reasons why you should not exercise...". An individual who scores a "yes" (1) on any single item is considered to be at higher level of risk for fitness exercise participation than a "no" (0) response and is recommended for exclusion from physical fitness testing (Fitness Canada). Symptoms were additive so that total symptoms reported ranged between 0 to 5. The test-retest reliability in the pilot study of older women was .667 (p < .005).  Perceived Well Being (TOTPWB) -  A 14 item, seven point Likert scale type of instrument (strongly agree to strongly disagree) called the Perceived Well-Being Scale (Reker & Wong, 1984) assessed physical and psychological well-being. Scores could range from 14 to 98. The pilot study obtained a test-retest rp (TOTPWB) = .565; p < .02) over a four week period. The scale included two subscales: psychological well-being (PWBPSYCH) and physical well-being (PWBPHYS). Reliability of the subscales were .604 for PWBPSYCH (p <.05) and .590 for PWBPHYS (p <.05).  Medications (MEDICINE) Subjects were asked how many prescription medicines they were taking that required a written prescription by their doctor. Scores could range from 0 to 7  167 (seven medications or more). A test-retest reliability of .857 (p <.0001) was obtained in the pilot study.  The Composite Health Index Since health was represented by four variables (self-rated health, PARQ, number of medications and the Perceived Well-Being Scale), a composite variable was created using weights derived from Principal Components Analysis. The composite health variable correlated r p = .324; p = .0001) with Exercise Level as calculated from the equation:  COMPOSITE HEALTH = .801 Well-Being - .782 PARQ + .749 Self-Rated Health - .740 Medicine/2.362  Body Mass (Quetelet Index) (HEIGHT; WEIGHT; BODYMASS) Body height and weight were self reported and converted to metric for use in a formula to calculate a body mass index. This variable was used in the descriptive data only.^Body mass is only an indicator of proportional or relative weight, not fatness. In sedentary people however, body mass is usually considered an indicator or adiposity. Body mass index was calculated as follows:  BODYMASS  -  v  Weight Height  2  kilograms  meters  2  168  Childhood Movement Confidence (CHILDMOV) This variable was viewed as a measure of predisposing context, rather than a current mediating variable. The 'Movement Confidence As A Child' (MCC) scale retrospectively evaluated  childhood movement confidence as perceived confidence  and experience to engage in six childhood physical skills. Childhood Movement Confidence, in this study, is a measure of self-efficacy which is reinforced by "habit" or at least considerable  experience with successful performance.  The MCC scale was adapted from a contemporary instrument called the 'Stunt Movement Confidence Inventory' (SMCI) which assessed a) perceptions of personal competence for performing a task, b) experience in performing a specific task, and c) perceived potential for physical harm during the performance (Griffin & Crawford, 1989). However, "perceived movement confidence as-a child" (CHILDMOV) is built on Griffin and Crawford's theoretical model with some alteration. The MCC scale used in this study omitted Griffin and Crawford's measures for "perceived enjoyment" and "potential for harm" since they were already represented in the Integrated Model with Bandura's construct "outcome expectations (perceived benefits and risks of exercise). As with Griffin and Crawford's study, the skills were portrayed in picture form (See Appendix A). In this study, for each of the six recalled physical skills, individuals were first asked about their movement confidence in childhood: "How sure are you that you could have done this as a youth?" and this was scored "Very sure" (4), "Pretty sure" (3), "Not very sure" (2) and "I know that I couldn't" (1). Secondly, the survey asked "how many times would you have done this as a youth?" and this was scored "I have done this a lot" (4), "I have done this a few times" (3), "I tried it once" (2), and "I've never done this" (1). Scores for childhood movement confidence and eAperienLe were summed and averaged (Range =  169 1 to 4 for each of six exercise skills). Thus MCC scores could range from 6 to 24. Test-retest reliability over four weeks was examined in the pilot study on older women and the MCC scale produced an extremely strong coefficient of r p (CHILDMOV) = .951; p <.0001. The SMCI has components of  Harter's Perceived Competence Scale for  Children, an instrument which had originated from her factor analysis of various elements of competence (Harter, 1982). She hypothesized that perceived competence ...should be positively related to one's ^intrinsic motivational orientation to prefer challenge, to be curious, and to engage in independent mastery attempts. (Harter, 1982, p. 94) Higher order factoring revealed that perceived cognitive competence was strongly related to preference for challenge (r = .57), to independent mastery ( r = .54), and moderately related to curiosity (r = .33). These four variables formed a distinct factor with high loadings of .76, .87, .80 and .79 respectively. Emphasis was placed on this factorial (construct) validity which remained stable across this grade range. Harter found stable patterns in Grade 3 to 6 children who consistently (over 6 different samples) identified four theoretically meaningful components: cognitive competence, social competence, physical competence and general selfworth. The internal consistency of her physical competence scale across all samples was .77 to .86 while test-retest correlation was .87. Teacher's ratings of physical competence correlated .62 with pupils' own self ratings. Discriminant validity was supported in that participants of school athletic teams scored significantly higher on perceived physical and social competence. That same year Griffin and Keogh (1982) published a significant theoretical ^O...^aloe_^  I^II^•  emen s: se  170 assessed competence, potential for enjoyment and potential for harm. The elements of competence, enjoyment and harm were incorporated into the  Playground  Movement Confidence Inventory (PMCI) published by Crawford and Griffin in 1986. The PMCI was tested with 250 fifth-grade students using a cluster sampling procedure. A test-retest reliability coefficient was .78 with the Grade 5 school children. The PCMI was validated using significant discriminant functions (p <.05) which classified subjects above chance levels into experience/confidence cells on the basis of systematic response variation to the movement confidence model. Classification accuracy ranged from 77.87% to 92.62%. Cross-validation of the PMCI was achieved through splitting the original  n into estimation (60%) and  holdout (40%) samples. The function values obtained from the estimation sample were applied to the holdout sample with the following classification results: a validity coefficient (r = .9768) was obtained with 84.65% classification accuracy or a 49.51% improvement over basic chance. Using the same response format, but different physical skills, the SMCI scale was then designed. Meant for contemporary boys and girls, SMCI used "stick" drawings of people performing risky physical stunts such as skateboarding over a ridge and cycling over a hill (Griffin & Crawford, 1989). Test-retest correlations with the SMCI were .82 for self-report experience and .80 for selfreported confidence with the task. Test-retest correlations for competence, enjoyment and harm were .88, .79 and .85 respectively. For the empirical cross-validation procedures, discriminant analysis was used to determine total scale classification power. Regression weights developed on one group (an estimation group) were applied to a second group (a holdout sample). Application of the regress jun weiyhis derived from the estimation sample  171 to the holdout sample data set resulted in a validity coefficient of r = .98 with an 88.2% classification accuracy. In this study the MCC scale retained the SMCI rope climb and bike riding with slight modification (feet on the rope and no aerial phase on the bike) and replaced the SMCI roller skating, skate boarding, stilts and pogo stick jumping with other challenging play elements that would have been available to some young girls in the early twentieth century. The MCC represented six physical challenge categories of trunk strength, arm strength, aquatic activity, leg power, hip flexibility and balance.  Childhood Social Support to Exercise (CHILDSOC) Four questions with five-point Likert scales were designed to retrospectively examine  childhood social support for exercise. Four recall  questions were asked relative to: 1) family athleticism (FAMSPORT), 2) personal encouragement by a parent, teacher or friend (CHLDHELP), 3) childhood opportunity to participate in vigorous physical activity in one's free time (OPPORTUN) and  4) enjoyment of physical education, exercise and sport at school (PEFUN). The scales represented important theoretical elements in the literature. In the pilot study, the scale for childhood social support in physical activity demonstrated good reliability (rp =.785; p <.0001).  172  The Cognitive Variables The Integrated Model of Exercise Behavior suggests that there are five beliefs or cognitive perceptions that one might consider in deciding whether to engage in exercise in their later years. These are: Health Incentive, Movement Confidence, Social Support, Outcome Expectations (Perceived Risks and Benefits) and Health Locus of Control. These five beliefs were measured as follows:  Health Incentive (MOTIVE) A variable called  Health Incentive represented Bandura's theoretical  element called "incentive to act." This construct assumed that if older adults were exercising, it was for reasons of maintaining or improving health and longevity. Health Incentive was measured by four statements about prolonging health and longevity. These were: "I don't really care how much longer  I live" (LIVECARE)  "I am motivated to avoid illness any way I possibly can" (AVOIDILL) "I am trying to live as long as I possibly can" (LIVELONG) "I am trying to stay healthy as long as I possibly can" (HLTHLONG). Four-point Likert scales were used to score responses: Strongly agree  (4), Agree  (3), Disagree (2), and Strongly disagree (1), except the first statement which was reverse-scored. Scores could range from  4 (low motive to live a long and  healthy life) to 16 (high motive). ^The scale showed adequate test-retest reliability in the pilot study (r s = .559; p <.03).  173  Adult Movement Confidence (ADULTMOV) The 'Movement Confidence Now' (MCN) scale evaluated adult movement  confidence as perceived confidence and experience in six adult physical fitness activities. The MCN scale was similar to the MCC scale but had wording to reflect the present tense, and evaluated whether each of the six specified fitness activities were done in the past year. Scores for adult movement confidence and experience were summed and averaged (Range = 1 to 4). Therefore the MCC and MCN scales would each provide scores between 6 (low confidence + low experience averaged) and 24 (high confidence and high experience averaged). The MCC and MCN scales were approximately matched for representative physical categories of trunk strength, arm strength, aquatic activity, leg power, hip flexibility and balance as described in Table 4.1.  174 Table 4.1  Movement Skills Estimating Movement Confidence and Past Experience  Movement Category  ^  1. Trunk strength 2. Arm strength 3. Aquatic activity 4. Leg power 5. Hip flexibility  MC as a Child'  ^  MC Now b  Swing by the knees  Curl-up 10 times  (KNEECONF; KNEEEXP)  (CURLCONF, CURLEXP)  Rope Climb  Knee push-ups 5 times  (ROPECONF; ROPEEXP)  (PUSHCONF; PUSHEXP)  Dive into deep water  Aquafit class (50 min.)  (DIVECONF; DIVEEXP)  (AQUACONF; AQUAEXP)  Jump from high object  Brisk walking (20 min.)  (JUMPCONF; JUMPEXP)  (WALKCONF; WALKEXP)  Splits  Toe touch  (SPLITCON; SPLITEXP)  (FLEXCONF; FLEXEXP)  6. Endurance, balance^Ride a two-wheel bike (BIKECONF; BIKEEXP)  Cycle for 20 min. (BYCCONF; BYCEXP)  ' Movement Confidence as a Child is a recall measure. Movement Confidence Now is a current measure  b  The MCN scale represented activities which contemporary seniors would likely find in their communities in exercise programming for older adults. Where possible, MCN items attempted to closely represent the MCC child activity in adult form. For example, stationary exercise cycling (as an adult) matched outdoor bike riding (as a child), aquafit exercises (as an adult) matched diving and swimming in deep water (as a child), jumping from a high box (as a child) was replaced by the leg power required for brisk walking (as an adult), and a sitting forward stretch (as an adult) represented flexibility to do the splits (as a  child  175 Validity was not directly tested, because the MCN scale was a modified version of a previously validated instrument (the SMCI). Furthermore, the MCN scale used clear illustrations of the performance requirements of the activities which should have maintained face validity. Test-retest reliability for the MCN scale was r p = 0.799; p <.001 in the pilot study.  Adult Social Support to Exercise (ADULTSOC) Prospectively, four items, matching childhood social support were worded along the same lines to assess adult social support for exercise: 1) family support (FAMSUPP), 2) encouragement by at least one person to be active (ADULTHLP, 3) peer involvement in physical fitness activities (FRENDACT) and  4)  physician endorsement for vigorous activities in late adulthqcd (DOCSUPP). Each question used a forced-choice format ranging from 1 to 5 (strongly disagree to strongly agree). Total scores could range from  4 (low support) to 20 (high  support). In the pilot study, reproducibility was poor for the adult social support scale. It appeared to be undermined by substantial seasonal changes in the physical and social activity of the pilot group from August to September. Consequently the test-retest reliability for adult social support was low and non-significant (r p = .372; p = .156).  Perceived Risks and Benefits of MCN Exercises The fifth cognitive mediator in Social Cognitive Theory was the Outcome  Expectation. In this study, the risks and benefits as outcomes from current exercise participation were rated as perceived outcomes. An example of one of  176 the six items was "Please rate the possible risk to your health of doing a 50 minute aqua-fit class". Five-point continuous Likert scales "(1) low risk...(3) moderate risk...(5) high risk" and "(1)low benefit...(3) moderate benefit...(5)high benefit" accompanied each adult exercise in the MCN instrument. This provided six ratings of perceived risk and six ratings of perceived benefits which were then separately summed to represent "Total Perceived risk" (TOTRISK) and "Total Perceived Benefit" (TOTBENE). Open ended statements concluded each exercise section of the MCN scale with: "The major risk for me would be " and "the major benefit for me would be .. In the pilot study, the perceived benefits of the six adult movement skills were reproduced within a four week period with rp (BENEFITS) = .837; p <.001. This was an encouraging indicator that the scale was highly reliable. However, perceived risks of the six movement skills gave an r p (RISKS) = .266 and did not reach significance. This finding is hard to explain, but may indicate a sensitivity of the perceived risk scale to the seasonal change in activity that the women experienced between test and retest. The mean risk declined from 10.15 to 8.39 at retest - a reduction in perceived risk that may have accompanied their renewed involvement in activity.  Health Locus of Control (TOTHLC) Health locus of control was added to the theoretical model as a fifth cognitive variable because recent studies suggest that individuals would not be likely to undertake health promoting behavior if they felt they were unable to promotetheir  177 The Health Locus of Control Scale (HLOC) of Wallston, Wallston, Kaplan and Maides (1976) uses eleven Likert items (6 points from strongly agree to strongly disagree) to assess internal versus external sense of control over one's health. Internal locus of control is an attributional style whereby an individual perceives that they are in control of an outcome. External locus of control is found in individuals who believe that an outcome is due to fate, chance or luck. The original 1976 scale was used for this study because it was shorter than the 1978 version and more statements were relevant to health and exercise. The HLC scores range from 11 to 66 with lower scores meaning "internality" and higher scores meaning "externality". According to locus of control theory, "internals" would be more likely to take steps to better their personal condition than would externals. Therefore, in a program designed to modify health-related behaviors, one might expect internals to be more successful than externals whose beliefs are leaning toward helplessness. Internality has been associated with higher levels of leisure-time and sport behavior in women (Bonds, 1980; Calnan, 1988; Carlson & Petti, 1989; Dishman & Steinhardt, 1990; Kleiber & Hemmer, 1981; Laffrey & Isenberg, 1983; Lee, 1980; McCready & Long, 1985; Perri & Templer,II, 1984-85;), but externality seems to accompany aging and declining health (Calnan, 1988; Kist-Kline & Lipnickey, 1989; Lumpkin, 1985). Wallston et al. (1976) reported a mean score of 35.57 (sd = 6.2) for college students with internal reliabilities of .72, .54, .50, and .40 on various college and community populations. The scale was reported to have a concurrent validity of r =.33 with Rotter's I-E scale. Moreover the scale was claimed not to discriminate by gender nor reflect a social desirability bias.  178  The Pilot Study In mid-December, 1989, three anonymous women in their late nineties were interviewed using the prototype questionnaire as an interview guide. Next, Gladys Hartley, a retired professional figure skater and dance teacher, critically evaluated the initial questionnaire for readability and comprehension. Finally, the questionnaire was also reviewed by Dr. Gloria Gutman, Director of the Simon Fraser Gerontology Centre, as well as the members of the Research Supervisory Committee. I made numerous revisions according to their advice during the first half of 1990. At the end of August, 1990, the questionnaire was piloted with 18 older women from the "U. of Agers" Women's Gymnastics Team (average age 68) at Hinton, Alberta during the Alberta Seniors Games. Subjects in the pilot study repeated the same questionnaire two to four weeks later in Edmonton and these were mailed back to me at the end of September. A serious car accident forced one woman into hospital so that only 17 women were retested. After the second administration of the questionnaire, statistics on the reliability of the subjects to reproduce the same information were calculated. Test-retest correlation coefficients and probability levels for all variables are reported in Table 4.2. Pearson Product Moment correlations (r e ) were used for interval and ratio variables; Spearman coefficients (r e ) were used for marital status, education, work role, and school location. Women were 100% reliable in reporting marital status and number of children. The majority of measures were highly reliable with many correlations in the range of .70 to 1.0. The criterion variable, moderate exercise, was highly reliable (r p = .756, p < .001). Weak and non-significant correlations were found  179 in five of 24 variables: total amount of exercise; amount of mild exercise reported; adult social support; perceived risks; and health locus of control. Lack of reliability in reported exercise was initially a concern. However reproducing these women's activity patterns during the particular four weeks of the pilot study proved to be difficult under the circumstances. Since the U of Agers fall classes resumed after the summer recess (in the interim period of the administration of the two questionnaires), it is not surprising that these three variables demonstrated an inconsistency. The questionnaires were, however, reflective of these altered activity patterns in the early fall. Indeed, closer examination of the test-retest data for total weekly exercise level indicated that variance more than doubled from test to retest (S  2  = 392.1 to 968.2). Although moderate exercise reported was highly reliable, it was evident from the self-reports that some women engaged in less mild activities and more vigorous activity in late September than in late August. The altered exercise status reported on the second questionnaire coincided with renewed involvement in the fall gymnastics program. The timing of the retest, then, is probably responsible for the decreased and inconsistent reporting in the mild exercise category (r p = .114) and the increased and inconsistent reporting in the vigorous exercise category (rp = .505). Although the total of mild activity reported showed a decrease at the retest, this change was insignificant (t = 0.541; p = .597). Therefore it appears that low test-retest coefficients may be more an issue of real lifestyle changes than a reflection of the quality of the survey instrument. Changing activity patterns may also explain the inconsistent reporting in perceived social support, perceived risk and health locus of control.  Examination of the means indicated no s ignifiLdul cha nge in level of social  180 support from test to retest; however, the variability was reduced noticeably on the second questionnaire (S 1 2 = 21.116 and S 2 2 = 6.729). In other words, at the time of the second questionnaire, the women perceived more similar levels of social support. It is possible that a few women perceived less social support during the summer months when the group was on recess, and that by late September, everyone was back together and perceiving similar social incentives to exercise together. In terms of perceived risk, a similar phenomenon occurred. Perceptions of risk at the retest were significantly reduced (t = 2.570; p = 0.26) and variance reduced (S 1 2 = 5.074, S 22 = 3.912). The reduced risk perceptions accompanying the resumption of seasonal exercise programs reported on the second questionnaire might be related to the resumption of the supervised activity program. Therefore it is possible that reliability in reporting risk perceptions was undermined by real changes in perceived level of risk that occurred over the month of September. The lack of consistency in reporting mild exercise, social support, perceived risk and locus of control were in conspicuous contrast to the high and significant correlations of all the other variables examined and appeared to be due to known seasonal changes in lifestyle.  181 Table 4.2 Correlation Coefficients and Significance Levels for Test-Retest Data TYPE OF MEASURE Age  TEST-RETEST CORRELATION  PROBABILITY LEVEL  r= .998  p<.0001  Marital Status  rs=1.000  p<.0001  Number of Children  r=1.000  p<.0001  Education  rs= s^..792  p<.0001  Work Role  rs= .886  p<.0001  School Location  rs= .999  p<.0001  Height  r= .998  p<.0001  Weight  r= .679  p<.004  Self-rated Health  r= .506  p<.046  PARQ Symptoms  r= .667  p<.005  Perceived Well-Being  r= .565  p<.028  Number of Medications  r= .856  p<.0001  Childhood Social Support  r= .785  p<.0001  Child Mov. Confidence  r= .951  p<.0001  Health Incentive  rs= .660  p<.025  Adult Mov. Confidence  r= .779  p<.001  Adult Social Support  r= .372  n.s.  Perceived Risks  r= .226  n.s.  Perceived Benefits  r= .837  p<.001  Health Locus of Control  r= .472  p<.077  Mild Exercise  r=-.114  n.s.  Moderate Exercise  r= .756  p<.001  Vigorous Exercise  r= .505  p<.046  Total Amount of Exercise  r= .340  p<.198  182  Selection of the Sample The Population Women who were born in 1921 or earlier were the target population of this research. One difficulty in surveying older adults is that a substantial minority are simply not well enough to participate. This study purposely excluded the very ill and institutionalized older women - women who represent about 8 to 10% of this age cohort. Also excluded were women who had adequate health, but were not venturing into their community at the time of the study. Thus, the bias inherent in studying the well-elderly further exaggerates the statement: If gerontology has a central message, it is that those who have survived to old age often represent a special case. (Branch & Jette, 1984, p.1128) Prohibitive costs required that the study be limited further to situations where surveys could be distributed and collected without mailing. Thus the population under study was delimited to women, age 70 and over, who could be found in Metropolitan Vancouver regularly attending community programs. There were a number of difficulties associated with surveying this target population. First, Canadian women in this age group feel vulnerable to exploitation by business groups, and indeed, in the Vancouver area, to two universities who are interested in gerontological research. Many of the women refused to fill out the survey questionnaire because they had "already been researched to death". Second, there were a number of women who had never been involved in survey research before and were immediately suspicious of the intent of the study. Several looked at the survey briefly and could not see any personal value to their involvement in this type of research. Often a second explanation would Lorivince some women to take part, especially it one of the "ring-leaders" of a  183 particular group appeared to endorse the project. Although the survey was anonymous and did not ask them for specific income levels, several women felt that it was "too personal" for them to participate. Almost 20% of the questionnaires were returned empty and eventually "recycled" to other volunteers. Follow-up phone calls to those women who changed their mind about participating revealed the following excuses: the survey was too much bother; they felt simply too busy to participate; the questions were "too personal;" exercise was of no importance at their age; the survey questionnaire was lost or thrown out; the questionnaire was left at the program site; they had forgotten to fill out the survey; they couldn't remember anything about the study; the questionnaire was too long; they came down with an illness or serious accident; their spouse had become seriously ill; they were not exercising and therefore did not feel worthy of participation; they had filled it out and thought it had been returned, and so on.  Sampling Procedure The Strategic Sample The purpose of the survey was to examine and explain the physical activities of elderly women who were mobile in their communities. With this purposive sample, I intended to survey all eligible women who were willing to participate. The sampling procedure first required identification of all community facilities where seniors could be