UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An evaluation of a teacher reported measure for the early identification of selective mutism Martinez, Yvonne Julia 2011

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

Item Metadata

Download

Media
24-ubc_2011_fall_martinez_yvonne.pdf [ 2.31MB ]
Metadata
JSON: 24-1.0072062.json
JSON-LD: 24-1.0072062-ld.json
RDF/XML (Pretty): 24-1.0072062-rdf.xml
RDF/JSON: 24-1.0072062-rdf.json
Turtle: 24-1.0072062-turtle.txt
N-Triples: 24-1.0072062-rdf-ntriples.txt
Original Record: 24-1.0072062-source.json
Full Text
24-1.0072062-fulltext.txt
Citation
24-1.0072062.ris

Full Text

  AN EVALUATION OF A TEACHER REPORTED MEASURE FOR THE EARLY IDENTIFICATION OF SELECTIVE MUTISM   by   Yvonne Julia Martinez   Bachelor of Arts (Honours Psychology), University of Waterloo, 2003 Master of Arts (School Psychology), University of British Columbia, 2007     A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF   DOCTOR OF PHILOSOPHY   in   The Faculty of Graduate Studies   (School Psychology)    THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)    August 2011     © Yvonne Julia Martinez, 2011 ii Abstract Selective mutism (SM) is a childhood disorder characterized by failure to speak in social situations despite there being an expectation to speak and the capacity to do so.  The prevalence of SM among children is estimated to be between approximately 0.5 to 1% (Sharp, Sherman, & Gross, 2007). Children with SM exhibit more symptoms at school than at home, yet current identification and diagnosis of SM relies exclusively on parent reports (McInnes & Manassis, 2005; Schniering, Hudson, & Rapee, 2000).  There is often a 3 to 6 year delay in SM referrals because parents do not recognize symptoms until children begin school.  SM appears to interfere with achievement and social-emotional development in children (Sharp et al., 2007).  The first phase of the study involved the development and validation of the Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV; Tannock, Fung, & Manassis, 2003), to teachers of a clinical sample of 29 children (Kindergarten to Grade 5) referred to three large urban hospitals for SM across Canada.  The second phase was a follow-up validation study that involved the revision of the TTI-SM-DSM-IV into a paper-pencil measure, the TTI-SM-R.  The revised TTI-SM-R was administered to a sample of 30 (Kindergarten to Grade 4) teachers of children in the community with SM and normal controls in western Canada.  The third phase combined the clinical and community samples for a total sample of 59 participants.  The results indicated that the Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R demonstrated evidence of reliability and validity (construct, face, predictive, concurrent, convergent, discriminant) for assessing SM.  Additionally, the results showed that students with SM were found to have more symptoms of social and school anxiety and greater difficulty with daily social participation compared to children without SM.  The analyses also revealed a high concordance rate between parent and teacher ratings of children with SM.  The development and validation of a teacher reported measure for SM may help reduce the iii lag time between symptom onset and treatment referral for students with SM.  The findings may contribute to a better understanding of anxiety symptomatology in students with SM, and the impact of mutism on social behaviours.           iv Preface The first phase of this study was part of a larger, multi-site investigation (at three large, urban cities across Canada) of language and cognitive skills in children with anxiety disorders (ANX) and selective mutism (SM) (Manassis, Tannock, Sloman, Fiskenbaum, & McInnes, 2007).  The study received ethical approval through the Hospital for Sick Children in Toronto, Ontario.  Details concerning the original study, such as the design, participants (number of participants, inclusion/exclusion criteria for participation, sampling and recruitment procedures, etc.), setting (locations of sites), measures (with the exception of the TTI-SM), and procedures are described by Manassis and her colleagues (2007). The first phase of the study (Chapter 3) is based on work I conducted with Drs. Katharina Manassis and Rosemary Tannock.  I was responsible for the conceptualization of the study, data entry, analysis, and writing and preparing the manuscript for submission.  Research related to this first phase has been submitted for publication to a peer reviewed journal in March 2011.  I am first author on this publication. The second phase of the study (Chapter 3) received approval of UBC Research Ethics Boards (Behavioural Research Ethics Board), Certificate Number: H08-02871, dated March 12, 2009.  The study also received approval from five school districts in British Columbia.  I was responsible for the design, methodology, data collection and entry, analysis, and writing.  The third part of the study (Chapter 3) combines the data from samples 1 and 2.  I am the primary investigator of this project.  I was responsible for the design, methodology, data collection and entry, analysis, and writing. v Table of Contents Abstract ........................................................................................................................................... ii Preface ............................................................................................................................................ iv Table of Contents ............................................................................................................................ v List of Tables ................................................................................................................................... x List of Figures ................................................................................................................................ xi List of Abbreviations..................................................................................................................... xii Acknowledgements ...................................................................................................................... xiv Dedication ..................................................................................................................................... xv 1.     Introduction ......................................................................................................................... 1 1.1 Overview ...........................................................................................................................1 1.2  Differential Diagnosis of Selective Mutism (SM) and Social Phobia ..............................3 1.3  Differential Diagnosis of SM and a Communication Disorder ........................................7 1.4  Delay in Clinical Referrals and Intervention ....................................................................8 1.5  Assessment and Diagnosis of Selective Mutism ..............................................................9 1.6 Parent Reports for SM Diagnosis ...................................................................................10 1.7  Importance of Using Multiple Raters in SM Diagnosis ..................................................12 1.8  Multiple Informant Agreement and Concordance Rates for SM ....................................15 1.9  Objectives of Research ...................................................................................................16 1.10 Research Questions .........................................................................................................17 vi 1.11  Summary .........................................................................................................................24 2.  Methodology ......................................................................................................................... 26 2.1  Overview of Phase 1 (Sample 1) and Phase 2 (Sample 2) ..............................................27 2.2  Sample 1: Development and Validation of TTI-SM-DSM-IV with a Clinical Sample ...30 2.2.1  Participants. .............................................................................................................30 2.2.2 General procedures. .................................................................................................31 2.2.3  Teacher measures. ...................................................................................................31 2.2.3.1 Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV; Tannock et al., 2003). ........................................................... 32 2.2.3.1.1 Development and description of the TTI-SM-DSM-IV. ............................ 32 2.2.3.1.2 Mode of administration of the TTI-SM-DSM-IV. ...................................... 36 2.2.3.1.3 Scoring the TTI-SM-DSM-IV. ................................................................... 37 2.2.4  Parent measures. ......................................................................................................38 2.2.4.1 Anxiety Disorders Interview Schedule for DSM-IV: Parent and Child Version (ADIS-C/P; Silverman et al., 2001). ................................................................................. 38 2.2.4.2 Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008). ........... 39 2.2.5  Summary of findings based on Sample 1 (clinical sample). ...................................41 2.3 Sample 2: Development and Validation of TTI-SM-R with a Community-based Sample41 2.3.1  Participants. .............................................................................................................42 2.3.2 General procedures. .................................................................................................42 vii 2.3.3 Teacher data collection procedures. ........................................................................44 2.3.4 Teacher measures. ...................................................................................................45 2.3.4.1 Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-Pencil Version (TTI-SM-R). .................................................................................. 45 2.4.4.1.1 Development and description of the TTI-SM-R (paper-pencil version). ... 46 2.3.4.1.2 Mode of administration of the revised TTI-SM-R (paper-pencil version). 48 2.3.4.1.3 Scoring of the revised TTI-SM-R (paper-pencil version). .......................... 49 2.3.4.2  School Anxiety Scale – Teacher Report (SAS-TR; Lyneham et al., 2008). ... 49 2.3.5 Parent data collection procedures. ...........................................................................50 2.3.6 Parent measures. ......................................................................................................51 2.3.6.1 Anxiety Disorders Interview Schedule for DSM-IV: Parent Version (ADIS-P; Silverman, & Albano, 2004; Silverman et al., 2001). ....................................................... 52 2.3.6.2 Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008). ........... 52 2.3.6.3 Preschool Anxiety Scale (PAS; Spence et al., 2001). ..................................... 52 2.3.7  Summary of Sample 2 (community-based sample). ...............................................54 2.3.8 Differences between Samples 1 and 2. ....................................................................54 2.4 Sample 3: Combining Sample 1 (Clinical) and Sample 2 (Community)........................56 2.5 Nonparametric Statistical Methods (Mann-Whitney U Test) .........................................58 3. Results ................................................................................................................................... 61 3.1 Research Questions .........................................................................................................61 viii 3.2 Description of Samples ...................................................................................................62 3.2.1 Description of Sample 1 (clinical sample). .............................................................62 3.2.2  Description of Sample 2 (community-based sample). ............................................66 3.2.3  Description of the combined sample. ......................................................................67 3.3  Statistical Analyses and Results for Research Questions ...............................................67 3.3.1 Statistical analysis and results for research question 1 . ..........................................67 3.3.1.1 Findings based on Sample 1 (clinical sample). ............................................... 68 3.3.1.2 Findings based on Sample 2 (community-based sample). .............................. 76 3.3.1.3 Findings based on combined sample. .............................................................. 78 3.3.2 Statistical analysis and results for research question 2 . ..........................................79 3.3.2.1 Findings based on professional field-based review. ........................................ 80 3.3.3 Statistical analysis and results for research question 3 . ..........................................80 3.3.3.1 Findings based on Sample 1 (clinical sample). ............................................... 80 3.3.3.2 Findings based on Sample 2 (community-based sample). .............................. 81 3.3.3.3 Findings based on the combined sample. ........................................................ 81 3.3.4 Statistical analysis and results for research question 4 . ..........................................81 3.3.4.1 Findings based on Sample 1 (clinical sample). ............................................... 82 3.3.4.2 Findings based on Sample 2 (community-based sample). .............................. 87 3.3.4.3 Findings based on the combined sample. ........................................................ 88 3.3.5 Statistical analysis and results for research question 5 . ..........................................88 ix 3.3.5.1 Findings based on Sample 2 (community-based sample). .............................. 89 3.3.6 Statistical analysis and results for research question 6 . ..........................................92 3.3.6.1 Findings based on Sample 1 (clinical sample). ............................................... 92 3.3.6.2 Findings based on Sample 2 (community-based sample). .............................. 98 3.3.6.3 Findings based on the combined sample. ........................................................ 99 4. Conclusion and Discussion .................................................................................................. 101 4.1  Summary .......................................................................................................................101 4.2  Discussion of Research Findings ..................................................................................103 4.3 Strengths and Limitations of the Research ...................................................................110 4.4 Future Directions ..........................................................................................................116 References ................................................................................................................................... 120 Appendix A: Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV) ....................................................................................................................... 136 Appendix B: Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-pencil Version (TTI-SM-R) .............................................................................................. 148 Appendix C: School Anxiety Scale – Teacher Report (SAS-TR) ................................................ 152 Appendix D: Selective Mutism Questionnaire (SMQ) ................................................................ 153 Appendix E: Preschool Anxiety Scale (PAS) .............................................................................. 155   x List of Tables Table 2.1: Differences between Samples 1 and 2 ......................................................................... 55 Table 3.1: Descriptives for Samples 1, 2, and Combined Samples  .............................................. 63 Table 3.2: Intercorrelation Matrix for the Mutism subscale  for Sample 1, 2 and Combined Sample ........................................................................................................................................... 70 Table 3.3: Reliability Analyses for Samples 1, 2, and Combined Samples: Internal Consistency and Split Half for the Mutism subscale ....................................................................................... 74 Table 3.4: Mean and Median Scores for the Mutism subscale and SMQ in Samples 1, 2, and Combined ...................................................................................................................................... 83 Table 3.5: Spearman’s Rho Correlations for Mutism subscale Scores and SMQ for Samples 1, 2, and Combined Sample .............................................................................................................. 86 Table 3.6: Mean and Median Scores for SAS-TR and PAS for Sample 2 ..................................... 90 Table 3.7: Spearman’s Rho Correlations for Mutism Subscale Scores to SAS-TR and PAS for Sample 2 ........................................................................................................................................ 91 Table 3.8: Chi-square Analysis for the Behaviour Subscale for Samples 1, 2, and Combined Samples ......................................................................................................................................... 94  xi List of Figures Figure 2.1: Flow Chart for the Development and Validation of the TTI-SM-DSM-IV and TTI-SM- R .................................................................................................................................................... 29 Figure 3.1: Marked to Severe Problems on the Behaviour subscale for Sample 1 (SM vs. ANX) 98 Figure 3.2: Marked to Severe Problems on the Behaviour subscale for Sample 2 (SM vs. NC) .. 99 Figure 3.3: Marked to Severe Problems on the Behaviour subscale for Combined Sample (SM vs. NC) .............................................................................................................................................. 100    xii List of Abbreviations Abbreviation Full Name ADIS-C/P Anxiety Disorders Interview Schedule, Child and Parent Forms (Silverman, Saavedra, & Pina, 2001) ANX  Anxiety disorders APA American Psychiatric Association Behaviour subscale Behaviour subscale of the TTI-SM-DSM-IV and TTI-SM-R * On the original TTI-SM-DSM-IV protocol listed in Appendix A, this subscale is referred to as the “Other” subscale CBCL Child Behavior Checklist (Achenbach, 1991; 1992) CRS Conners’ Rating Scale, Revised (Conners, 1997) DSM-III-R Diagnostic and Statistical Manual of Mental Disorders, 3rd Edition, Revised (APA, 1980) DSM-IV-TR  Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision (APA, 2000) ELL English Language Learners GAD Generalized Anxiety Disorder KSADS-PL Schedule for Affective Disorders and Schizophrenia for School-Age Children – Present and Lifetime Version (Kaufman et al., 1997) MASC Multidimensional Anxiety Scale for Children (March et al., 1997)      xiii Abbreviation Full Name Mutism subscale Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R * On the original TTI-SM-DSM-IV protocol listed in Appendix A, this subscale is referred to as the “SMQ” subscale, but was renamed in the current paper to avoid confusion with the SMQ (Bergman et al., 1999; 2008) NC  Normal control NIMH DISC-IV NIMH Diagnostic Interview Schedule for Children Version IV (Shaffer et al., 2000) OCD  Obsessive-Compulsive Disorder ODD  Oppositional Defiant Disorder PAS Preschool Anxiety Scale (Spence et al., 2001) PCP Principal Components Analysis PTSD  Post-Traumatic Stress Disorder SAD  Separation Anxiety Disorder SAS-TR School Anxiety Scale, Teacher Report (Lyneham et al., 2008) SASC-R Social Anxiety Scale for Children – Revised (La Greca & Stone, 1993) SM Selective Mutism SMQ Selective Mutism Questionnaire (Bergman et al., 1999, 2008) SPSS-18.0 Statistical Package for the Social Sciences, 18.0 Version TTI-SM-DSM-IV Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (Tannock et al., 2003) TTI-SM-R Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-pencil Version (TTI-SM-R) xiv Acknowledgements I want to offer my appreciation and gratitude to Dr. Lynn Miller, my research supervisor, for her continuous support throughout my graduate studies at the University of British Columbia (UBC).  I would also like to thank my committee members, Drs. Kadriye Ercikan and William McKee.  I am so grateful to have such a supportive committee, and appreciate all of the constructive feedback, help, insight, and guidance you offered throughout my studies. Next, I want to extend my thanks to Drs. Rosemary Tannock and Katharina Manassis at the Hospital for Sick Children (Toronto, ON) and the University of Toronto, for giving me the opportunity to volunteer at Sick Kids Hospital in 2008 during a specialty practicum placement, and who helped inspire my research work with children with anxiety disorders and selective mutism. I want to thank my family, soon-to-be husband Mike, and friends, who have provided continuous support and encouragement throughout my graduate studies at UBC.  Also, this research would not have been possible without the help of wonderful peers at UBC, including: Angela Taschuk, Lina Darwich, Amanda Hume, Jessica Trach, Sarah Hussain, and Christine Yu. Thank you for all your time, care and conscientiousness. Finally, thank you to all of the students, families, and school personnel from British Columbia, Ontario, and Quebec who took part in this research. This research was supported in part by the Canadian Institutes for Health Research (CIHR Doctoral Research Award, jointly funded by the Institute of Neurosciences, Mental Health and Addictions), UBC Faculty of Graduate Studies (4-Year Fellowship, and Full Tuition Scholarship), and UBC Faculty of Education (Hampton Fund Research Grant for Humanities and Social Sciences, and the Dean of Education Scholarship). xv Dedication  To all the silent children, and the adults who won’t give up.                1  1.     Introduction 1.1 Overview The Diagnostic and Statistical Manual of Mental Disorders, 4 th  Edition, Text Revision (DSM-IV-TR; American Psychiatric Association [APA], 2000) defines selective mutism (SM) as a childhood disorder that is characterized by a child’s failure to speak in some social situations, despite there being an expectation to speak (Criterion A; American Psychiatric Association [APA], 2000).  This may mean that a child will fail to speak at school, with peers, and/or with family members who are not part of his or her immediate family.  Another characteristic of SM is that the child’s failure to speak interferes with educational achievement or with social communication (Criterion B), for over 1 month, though failure to speak cannot be during the first month of school while the child is still acclimatizing to the new environment (Criterion C).  The child’s failure to speak cannot be a result of a lack of knowledge of or comfort with the spoken language required in the situation (Criterion D).  Finally, the lack of speech cannot be better accounted for by a Communication Disorder (e.g., receptive language disorder) and/or psychological disorder (e.g., Schizophrenia; Criterion E). Prevalence rates for SM vary from study to study (Drewes & Akin-Little, 2002), but generally speaking, SM is reported to affect less than 1% of the population (Sharp et al., 2007). Researchers speculate that the estimated prevalence rates are an underestimate due to factors such as underreporting by parents, and families not recognizing or being aware that there is a problem with their child.  This speculation is supported by prevalence estimates in the school (community) setting being higher than those in clinical samples (Bergman, Keller, Piacentini, & Bergman, 2008; Sharp et al., 2007; Standart & Couteur, 2003).  Another reason prevalence rates differ may be due to the diagnostic criteria used and the age that is sampled.  2  An intriguing finding in prevalence studies is the prevalence of SM among children who are English Language Learners (ELL) and/or new immigrant language minorities.  ELLs are individuals whose first language is not English.  Researchers have found that SM is almost 4 times higher for ELL students who are immigrants, compared to English-speaking residents (2.2% vs. 0.47%, respectively; Elizur & Perednik, 2003; Toppelberg, Patton, Coggins, Alissa, Lum, & Burger, 2005).  Research with children of bilingual backgrounds have found that children with SM are found to have difficulty speaking in social situations in both their first (home) language, and their second language (Toppelberg et al., 2005).  Higher prevalence rates of SM among new immigrant language minorities may be attributed to general issues concerning mental health identification and referrals made for minority children/adolescents, which include stigma, social and language barriers to care, socioeconomic status, barriers in mental health coverage for psychological services, and knowledge of mental health issues.  For example, mental health issues in Chinese, Hispanic, and African American children are underreported compared to Caucasian children, and these ethnic minority families have been found to underutilize mental health services compared to Caucasian children (Chen, Kazanjian, & Wong, 2008; Lung & Sue, 1997; Pumariega, Glover, Holzer, & Nguyen, 1998).  With over 18% of its population being foreign born, Canada is one of the most multicultural and ethnically diverse countries in the world (Statistics Canada, 2003).  The promotion of mental health issues with diverse populations is becoming a more prominent area of research.  The higher rate of SM in ELL populations, paired with the high rate of ethnically diverse families in Canada, highlight the importance of early screening for child mental health problems. Given the low prevalence rate of SM, it remains a challenge to conduct research of SM with a significant sample size for statistical analysis (Drewes & Akin-Little, 2002).  The  3  published research on SM is largely based on single-subject (case-study) designs or small clinical samples.  To obtain a more complete understanding of SM in children, single-subject and small clinical samples need to be complemented by school and community-based research. 1.2  Differential Diagnosis of Selective Mutism (SM) and Social Phobia To date, there continues to be debate about the classification and diagnosis of SM (Sharp et al., 2007).  In the DSM-IV-TR (APA, 2000), SM currently is classified under the category of “Other disorders of infancy, childhood, or adolescence”, and is listed in the final section, which is entitled, “Disorders first diagnosed in infancy, childhood, or adolescence”.  It is commonly understood that disorders that appear in this category of the DSM-IV-TR pose difficulties for researchers and clinicians because these sets of disorders do not neatly fit into the nosology of other more typical pathologies (House, 2002).  Unlike the other diagnostic categories in the DSM-IV-TR (i.e., Anxiety disorders, Mood disorders, Substance-related disorders), the childhood disorders listed in this category are not grouped together because they share similar diagnostic characteristics, but rather, these disorders are grouped together because they are typically diagnosed prior to adulthood (House, 2002).  The debate of where to place SM for diagnostic purposes continues to perplex clinicians and researchers.  The lack of clarity on how to classify SM in the DSM-IV-TR may be due to the low prevalence rates and limited research on this disorder.  It may also be due to crossover symptoms between SM and other disorders.  Criticisms related to the DSM-IV classification system for young children continue to be debated in the field (Egger & Angold, 2006; House, 2002). There is an intriguing relationship between SM and anxiety disorders (Sharp et al., 2007). Anxiety disorders are listed as a defined and distinct psychopathology in the DSM-IV-TR. Anxiety disorders are also clustered under a broader umbrella of internalizing disorders.  4  Internalizing disorders, however, are not listed as a diagnostic category in the DSM-IV-TR, but is a descriptor used to understand the nature of a behavioural problem, such as whether the problem is directed inward (internalizing disorder) or outward (externalizing disorder).  Internalizing disorders are behaviourally difficult to observe due to its largely internal manifestation, as compared to externalizing disorders, which can more readily be observed behaviourally by others.  SM is typically considered to be an internalizing disorder.  However, symptoms of SM are manifested differently compared to other more typical internalizing disorders, such as anxiety and depression.  In fact, symptom presentation in SM is so varied that some researchers have found that a subset of children with SM may also exhibit externalizing behaviours, specifically, oppositional behaviours (Dummitt et al., 1997; Yeganeh, Beidel, & Turner, 2006). Given the internalizing nature of these disorders, behaviours associated with such disorders are largely difficult to observe.  For example, for more typical internalizing disorders, there is a large component of non-observable symptoms, such as physiological (i.e., headaches, stomach aches, tense muscles) and cognitive symptoms (i.e., negative thoughts, etc.) that can only be experienced, detected, and verbally revealed by the individual.  For children with SM, the most obvious or observable behaviour is remaining silent when the expectation is to be vocal, which is not a listed or observed symptom of any of the internalizing disorders.  Additionally, this behaviour may occur only outside the home setting (e.g., school, classroom; Comer & Kendall, 2004), thus making this behaviour more difficult to decipher. Social phobia is one of seven subtypes of the anxiety disorders.  The DSM-IV-TR (APA, 2000) characterizes social phobia as a persistent and excessive fear of social or performance situations around unfamiliar people.  Typically, the individual fears negative social appraisal or scrutiny by others that may result in embarrassment and/or humiliation.  As a result, these  5  situations are avoided or endured with intense anxiety and distress.  The symptoms of social anxiety can lead to an excessive withdrawal and “shrinking” from unfamiliar social contact. Symptoms are also serious enough to interfere with one’s daily routines and development of social relationships.  The prevalence of social phobia is approximately 12% in the general population (Ruscio et al., 2008), making it a common mental health concern.  The average age of onset of social phobia is 10-12 years, which is older than the age of onset for SM.  Although the fear of social situations may result in the inability to speak, social phobia is marked by other, more commonly occurring anxious behaviours (e.g., fear of social or performance situations, shrinking from social situations with unfamiliar people). SM presents with similar diagnostic characteristics and symptoms as social phobia, thus making the two disorders difficult to differentiate (Sharp et al., 2007).  For example, children with SM and social phobia experience fear of being judged in social situations (Anstendig, 1999).  Also, both groups of children attempt to avoid a situation, though the mode of avoidance differs as SM children avoid by not talking (with mutism being the key feature).  Children with social phobia, on the other hand, typically avoid the situation by not attending social situations, or by speaking in a very soft voice or by minimally speaking. Given such similar characteristics between SM and social phobia, SM has been considered by some theorists as existing at the extreme end of the continuum of social phobia where children experience fear of social or performance situations that are so debilitating that they are unable to speak (Anstendig, 1999; Sharp et al., 2007).  Thus, SM is considered by some to be a symptom or variant of social phobia (Black & Uhde, 1992).  A second camp of researchers have suggested that SM is a developmental precursor to social phobia, due to the discrepancy between the age of onset for the two conditions (i.e., social phobia = 10 years; SM =  6  5 years; Bergman, Piacetini, & McCracken, 2002).  A third camp of researchers (Dummit et al., 1997; Yeganeh et al., 2006) has suggested that SM should be a standalone disorder, and sometimes found to be co-morbid with other disorders.  For example, Dummit and his colleagues (1997) revealed that their entire sample of 50 children with SM met the criteria for social phobia or avoidant disorder (a childhood disorder recognized in the DSM-III-R; APA, 1980, but is not included in the later edition of the DSM-IV-TR; APA, 2000) and close to half of these children with SM also met the criteria for an additional anxiety disorder subtype.  Thus, SM appears to be either a variant of social phobia (Black & Uhde, 1992), a prodromal state of social phobia (Bergman et al., 2002), or a co-occurring condition of social phobia (Dummit et al., 1997). It is important for researchers and practitioners to take into account the similarities (and differences) between SM and social phobia when conducting research and practice with both populations.  There are children who are referred for SM but whose symptoms better fit the diagnostic characteristics of social phobia, whereas there are other children who are referred for social phobia but whose symptoms better fit the diagnostic characteristics of SM.  From a treatment perspective, improper diagnosis can have a potentially negative effect on the type of interventions that are recommended for children with SM and social phobia, given that students with the two disorders may respond differently to the same treatments (i.e., cognitive-behavioral and behavioral interventions; Omdal & Galloway, 2008).  From a research perspective, using both SM and social phobia measures to understand the presenting symptoms may help researchers better understand the shared and unique variance of the two disorders.  7  1.3  Differential Diagnosis of SM and a Communication Disorder For a diagnosis of SM, the DSM-IV (APA, 2000) states that a child’s failure to speak cannot be better accounted for by a Communication Disorder (Criterion E).  A Communication Disorder, as defined in the DSM-IV, includes the following five disorders: Expressive Language Disorder; Mixed Receptive-Expressive Language Disorder; Phonological Disorder; Stuttering; and Communication Disorder, Not Otherwise Specified (NOS). Several researchers have found that children with SM are often reported to have co- morbid communication disorders. Kristensen’s 2000 study found that 50% of children with SM had a co-morbid communication disorder, specifically expressive language disorder, mixed- receptive-language disorder, or phonological disorder.  Children with SM are also more likely to have expressive and receptive language delays compared to children without SM (McInnes, Fung, Manassis, Fiksenbaum, & Tannock, 2004; Steinhausen & Juzi, 1996). Interestingly, although co-morbid communication disorders have been found in children with SM, communication disorders are not often found in children with anxiety, thus suggesting a unique diagnostic feature of SM from ANX (McInnes et al., 2004).  Kristensen and Torgersen (2001) have found a different etiology and course for children with SM with a co-morbid communication disorder.  In their study of 160 children, they found children with SM and a co- morbid communication disorder were more emotionally stable and sociable than SM children without a communication disorder, as assessed by the EAS Temperament Survey for Children: Parental Ratings (Buss & Plomin, 1984).  The researchers theorized that failure to speak in SM- communication disordered children was driven by difficulty with communication and language, whereas for children with SM, failure to speak was driven by anxiety and shyness.  Thus, children in the group with a diagnosis of SM and a co-morbid communication disorder were less  8  likely to have difficulty with emotional stability and sociability compared to children who had a diagnosis of SM but no communication disorder.  Further, the group of children who had a diagnosis of SM but no communication disorder scored higher on the Distress, Fear, and Activity scales compared to both the control group and the group of SM-communication disordered children.  Based on the research findings, it appears that not all children with SM necessarily have a co-morbid communication disorder; rather, communication disorders appear to be a challenge for only a subgroup of children with diagnosed SM. Co-morbidity of SM and communication disorders presents a challenge in the diagnosis of SM.  Clinicians must determine whether the communication disorder is the primary disorder and better accounts for the symptoms of SM, or whether the communication disorder is a co- morbid disorder with SM.  Inaccurate diagnosis may have a negative effect on children with SM with a co-morbid communication disorder because interventions, especially school-based interventions, may not target symptoms of both SM and the communication disorder. 1.4  Delay in Clinical Referrals and Intervention Some symptoms of SM usually occur before children begin school (i.e., failure to speak in some social situations), and the age of onset for SM is estimated to be as early as 3 years (Garcia, Freeman, Francis, Miller, & Leonard, 2004; Remschmidt, Mathias, Herpertz-Dahlmann, Hennighausen, & Gutenbrunner, 2001).  Previous longitudinal studies have estimated that the age when a child is referred for SM is between 6.5 to 9 years (Ford, Sladesczek, Carlson, & Kratochwill, 1998; Kumpulainen, Rasanen, Raaka, & Somppi, 1998; Standart & Le Couteur, 2003).  School is often the environment where the child exhibits the most symptoms of SM, and teachers are usually first to notice symptoms of SM shortly after the child first begins school (Sharp et al., 2007).  As most children in North America start school approximately at 5 years of  9  age, a child may wait from 3 to 6 years before they are referred by a teacher for evaluation or treatment (Ford et al., 1998; Kumpulainen et al. 1998; Standart & Le Couteur, 2003). This 3 to 6 year delay in referral is likely because parents do not recognize the symptoms of SM, as the child typically participates in normal conversation with family members at home (Hayden, 1980).  Parents may also underestimate the seriousness of shy behaviour that is observed in social situations outside the home.  This delay in referral for children with SM, compounded with issues concerning the identification of mental health issues in children who are minorities or immigrants, further impacts this delay in referral for children who are also ELL. 1.5  Assessment and Diagnosis of Selective Mutism The diagnosis of SM is typically made from an assessment conducted by a mental health clinician.  The assessment usually consists of a parent interview and child observations (Kronenberger & Meyer, 2001).  Some clinicians may choose to supplement their clinical interview by using parent/teacher questionnaires concerning broad, general child behaviours (i.e., measures of anxiety, behaviour, social skills, and/or adaptive functioning; Kronenberger & Meyer, 2001).  The assessment of SM does not typically include the child (due to their reluctance to speak) or the teacher (Kronenberger & Meyer, 2001; Lucas, & Shaffer, 2010).  In fact, teacher reports of behaviours in children with SM are uncommon in published studies (Cabone et al., 2010; Kristensen, 2001). The most common parent interviews for child mental health diagnosis include: the Anxiety Disorders Interview Schedule for DSM-IV: Parent and Child Version; ADIS-C/P (ADIS for DSM-IV:C/P; Silverman, Saavedra, & Pina, 2001), the Schedule for Affective Disorders and Schizophrenia for School-Age Children – Present and Lifetime Version (KSADS-PL; Kaufman et al., 1997), and the NIMH Diagnostic Interview Schedule for Children Version IV (NIMH DISC-  10  IV; Shaffer, Fisher, Lucas, Dulcan, & Schwab-Stone, 2000).  These clinical interviews are often only administered to parents due to the child’s reluctance to speak.  Of these clinical interviews, the ADIS-C/P is arguably the most commonly used diagnostic clinical interview in the literature for SM diagnosis (Chavira, Shipon-Blum, Hitchcock, Cohan, & Stein, 2007, Manassis, Tannock, Sloman, Fiskenbaum, & McInnes, 2007, Yeganeh, Beidel, Turner, Pina, & Silverman, 2003). Unfortunately, studies that look specifically at SM diagnosis, phenomenology, treatment, and outcome rarely incorporate standardized, well-validated measure of a child’s failure to speak (Bergman et al., 2008).  When standardized measures are used, such as the Child Behavior Checklist (CBCL; Achenbach, 1991), and the Conners’ Rating Scale, Revised (CRS-R; Conners, 1997), these instruments capture a broad range of behaviours but are not specific to symptoms of SM.  Hence, such measures do not address the unique symptoms and functional deficits of children with SM (Bergman et al., 2008).  Currently, there is only one validated measure of parent-reported SM in the field, the Selective Mutism Questionnaire (SMQ; Bergman, Halloway, & Piacentini, 1999; Bergman et al., 2008).  The SMQ is a parent-reported measure of a child’s frequency of failure to speak across various settings (school, home, and other social situations outside of the home and school settings).  The SMQ appears to be a psychometrically sound paper-pencil parent measure of the core features of SM as described in the DSM-IV-TR.  Specific details concerning the psychometric properties of the SMQ are discussed at length in the methodology section. 1.6 Parent Reports for SM Diagnosis Parent reports are critical in the assessment and diagnosis of SM.  One reason for this may be due to the ease of obtaining parent reports of child behaviours.  Another reason may be due to research that has found that depending on the age of the child and the specific mental  11  health problem that is being assessed, parent reports are sometimes more thorough and reliable in reporting complex details about a child compared to reports from teachers or child (self) reports (Schniering, Hudson, & Rapee, 2000; Smith, 2007). Despite the utility of parent reports in the assessment of internalizing disorders and SM, there are several concerns in relying only on parent reports.  One common problem found in parent reports is the over-reporting of child symptoms.  Several researchers have found parent reports of child psychopathology were inflated due to parent psychopathology (Briggs-Gowan, Carter, & Schwab-Stone, 1996; Chi & Hinshaw, 2002; Dadds, James, Barrett, & Verhulst, 2004; Richters & Pellegrini, 1989).  Dadds, James, Barrett, and Verhulst found that parents often inflate reports of anxiety symptoms in their children as a result of their own anxiety and projection of anxiety onto their children.  On the contrary, parents of shy children have also been found to under-report symptoms, and have been found to be anxious or shy themselves, and thus may not recognize shyness in their own child as being excessive (Beidel & Turner, 1997). Another concern with relying on parent reports is due to the internalizing nature and non- observable (physiological and cognitive) symptoms of SM, and/or the observable behaviours associated with SM that may occur only outside the home setting (e.g., school, classroom; Comer & Kendall, 2004).  The somewhat “invisible” nature of the disorder and the limited expression of problematic behaviour across settings may reduce the reliability of parent reports.  Symptoms of SM are often subtle in the home setting (i.e., some children show normal rates of speech at home), which means that parents may interpret child behaviours as shyness but not realize the extent to which these behaviours interfere with social development outside the home (Hayden, 1980).  As a result, parents may not feel it necessary to seek mental health services for their child.  On the other hand, some parents may be reluctant to seek treatment for their child due to  12  the stigma associated with mental health services (Chavira et al., 2007; Kristensen & Torgersen, 2001).  Other problems related to parent reports include: withholding information about the child’s behaviours, having insufficient understanding of normal child development, being misinformed as to how the child is performing in school, and/or having difficulty with communication (Ford et al., 1998).  In sum, there are many challenges associated with working with parents of children with SM, and these challenges are compounded when these children are of ELL background.  The obvious question regards the extent to which parent reports are the most useful source of information for detecting selectively mute behaviour. 1.7  Importance of Using Multiple Raters in SM Diagnosis Researchers have recommended that multiple informants (parent, teacher, and child) be used in the assessment and diagnosis of all childhood internalizing disorders (Grills & Ollendick, 2003; Silverman et al., 2001).  It is especially important to use multiple informants due to the potential limitations of parent reports discussed above, and because symptoms of SM are most pronounced in the school setting (Sharp et al., 2007).  Children do not usually participate as informants due to the typical age of onset and age of referral (thus impacting a child’s insight and ability to read and respond to self-report checklists and questionnaires), and the child’s reluctance to speak to unfamiliar adults (Kronenberger & Meyer, 2001).  Teachers, however, are in an ideal position to help with the identification, assessment, and intervention of children with SM, given the context where this disorder occurs.  A teacher’s perspective can provide information on symptoms and impairments that parents may not notice, and this information may assist in the accurate diagnosis of SM and the development and implementation of school-based interventions.  Further, children with SM have been found to have co-morbid academic and language problems (e.g., receptive language, phonemic awareness, mathematics; Manassis et al.,  13  2007; Nowakowski et al., 2009), which would require the expertise of teachers to identify and remedy. Another reason why it is important to include teacher reports in the assessment of SM is due to teachers’ beliefs about SM and how it affects the way they interact with students, potentially impacting the effectiveness of intervention.  For example, school-based studies on students with SM reveal that a student’s refusal to speak has a strong impact on teachers’ emotions and feelings of teaching efficacy (Cline & Baldwin, 1994).  It is natural for teachers to find SM perplexing and incomprehensible, because teachers may have limited understanding and experience working with children with SM due to its low prevalence rate.  Also, teachers may have a difficult time believing that a child’s refusal to speak is a result of a mental health concern such as anxiety (Black & Uhde, 1995; Cunningham, McHolm, Boyle, & Patel, 2004; Dummit et al., 1997; Kristensen, 2001), and misinterpret SM behaviour as an oppositional or peculiar behaviour.  By asking both parents and teachers to report on a child’s speaking behaviours, clinicians may be able to obtain a rich source of information, and more importantly, improve the accuracy of SM diagnosis. Despite the importance of including teachers in the assessment of SM, teachers continue to be underutilized in the assessment of SM.  One reason why teachers may not be included in the assessment of SM may be because there are no standardized teacher reported measures for SM.  Past studies have relied on teacher reports of more broad behavioural measures such as the CBCL (Achenbach, 1991), and the CRS-R (Conners, 1997) (Elizur & Perednik, 2003; McInnes et al., 2004).  The lack of teacher-reported measures of SM and delay of referrals for children with SM suggest a fundamental need for better screening and identification tools for children with SM as they enter the school system (Garcia et al., 2004).  14  The use of multiple informants and validated parent and teacher measures for SM would help identify, intervene with, and treat children with SM at an early age.  Prevention and early intervention research is crucial for children with SM given the course of the disorder.  Kolvin and Fundudis (1981) found that children with SM are resistant to treatment, and children with SM become more resistant to intervention and treatment the longer they have had SM (Kehle, Madaus, Baratta, & Bray, 1998).  Further, prevention and early intervention research have shown that subsequent psychiatric problems in children can be curtailed and treatment gains have longer-lasting effects when mental health problems are addressed early in a child’s development (Centre for Community Child Health, 2007; Waddell, 2007). Dow, Sonies, Scheib, Moss, and Leonard (1995) recommend a school-based multidisciplinary assessment of SM.  A multidisciplinary assessment would involve teachers, clinicians, and parents, given the presence of symptoms in the school and community settings. McInnes and Manassis (2005) discuss best practices in diagnostic assessment for children with SM, which would include: 1) gathering information from multiple informants; 2) establishing the age of onset of the mutism; 3) obtaining the child’s medical history; 4) collecting developmental history and issues; 5) gathering information on the child’s temperament; 6) gathering information on other co-morbid conditions; 7) gathering information on family history; 8) conducting observations in a clinical setting; 9) gathering information on language development; 10) gathering information on impairments related to SM; 11) gathering information on child non- verbal and verbal communication; and 12) gathering information on interventions that have been attempted in the past.  By using multiple methods of assessment, measurement error may be reduced.  15  1.8  Multiple Informant Agreement and Concordance Rates for SM Despite the advantages of using multiple informants in the assessment and diagnosis of SM, the degree of informer agreement between parents and teachers is unknown.  It is important to understand informer agreement in child diagnosis so that conflicting reports between raters can be better interpreted and understood.  There are few extant studies that specifically investigate parent-teacher informant agreement for SM.  These studies have reported general agreement between teacher and parent ratings on symptoms of SM and anxiety (Cunningham, McHolm, & Boyle; 2006; Kristensen, 2001).  However, correlation coefficients between parent and teacher ratings were not presented in any of these studies, thus it is unknown to what extent parent and teacher observations were confirmatory or in disagreement. Although there is a lack of research concerning multiple informer agreement for SM, there is a substantial body of research concerning multiple informer agreement for childhood internalizing and anxiety disorders, which could provide insight about informer agreement for SM.  Achenbach, McConaughy, and Howell (1987) conducted one of the first studies using multiple informants on paper-pencil standardized assessments of broad behavioural indicators of children and reported low correlations between parent, teacher, and child (self) reports (correlations ranged from r = .20 to .27).  Rather than interpreting these results as reflecting poor reliability and validity, the authors argued that this showed how individual raters were able to provide a different perspective of the same problem across different settings.  For example, some child behaviours may be present only in the classroom setting, such as in the case of SM, in which children do not speak in school and other social situations settings but generally do not have difficulties speaking at home to members of their immediate family.  16  Research continues to show that concordance rates between multiple informants, particularly between teachers and parents, are mixed.  Most findings report low to moderate agreement, particularly in ratings of childhood internalizing disorders, such as anxiety and depression (Barbosa, Manassis, & Tannock, 2002; Kristensen, 2001; Layne, Bernstein, & March, 2003; Mesman & Koot, 2000; Woo et al., 2007).  Overall, stronger agreement has been found between parent, teacher, and child reports for externalizing behaviours than internalizing disorders (DiBartolo, Albano, Barlow, & Heimberg, 1998; Grills & Ollendick, 2003, Rapee, Barrett, Dadds, & Evans, 1994).  This phenomenon is likely due to the nature of externalizing behaviours lending them to be visibly observed, in contrast to internalizing disorders, which often remain undetected by others.  Although children with SM first show symptoms when they enter school, it is unknown whether teachers (or parents) are better (or more accurate) informants of SM.  Overall, it appears that multiple informant agreement for SM and for internalizing and anxiety-related disorders remains inconclusive, and disagreement between raters has been well documented (Kraemer et al., 2003; MacLeod, McNamee, Boyle, Offord, & Friedrich, 1999). 1.9  Objectives of Research A review of the SM literature reveals issues concerning the early identification and proper diagnosis of SM.  There are no validated and standardized teacher-reported measures specifically assessing SM (Garcia et al., 2004).  To address this gap in the field, Tannock and her colleagues (2003) developed a teacher-reported semi-structured interview protocol for the identification of selective mutism, administered orally by phone, the Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV).  However, due to the length and time required for administration of the TTI-SM-DSM-IV and its accessibility to teachers and other school personnel, a decision was made by the author to take one of its  17  subscales, the Mutism subscale, and develop a paper-pencil version of the TTI-SM-DSM-IV, which is referred to as the TTI-SM-R (revised).  The Mutism subscale was intended to assess behaviours associated specifically with SM and possibly be used as a standalone measure. Although the TTI-SM-DSM-IV was developed in 2003, this measure, and its subscales, was never submitted to rigorous reliability and validity trials.  Similarly, the psychometric properties of the newly developed TTI-SM-R had not yet been evaluated.  Thus, the current study has two main goals: 1) to evaluate the psychometric properties of the Mutism subscale of the TTI-SM-DSM-IV (Tannock et al., 2003), and 2) to validate the Mutism subscale of the revised paper-pencil version of the TTI-SM-DSM-IV (the revised TTI-SM-R). Evaluation of the Mutism subscale was guided by standard scale development and test validation techniques, which include presenting evidence for reliability (internal consistency and principal components analysis [PCA]) and validity (criterion, construct, concurrent, discriminant, and face validity; Cronbach, 1970; DeVellis, 1991; 2003).  Thus, the selective mutism items from the TTI-SM-DSM-IV, which are also found, by definition, in the TTI-SM-R, need to be subjected to psychometric scrutiny.  The Mutism subscale, exclusively targeting selective mutism behaviors, will be the focus of attention in this study. 1.10 Research Questions The overarching research question of this validation study was: To what extent does a teacher reported measure for selective mutism (SM) identify children with SM accurately and consistently?  In order to demonstrate that a measure has adequate psychometric properties, researchers must develop a scale based on theory, and additionally provide evidence of reliability (i.e., internal consistency, PCA) and validity (i.e., predictive, convergent, discriminant, concurrent, face; Cronbach, 1970).  DeVellis (1999; 2003) proposed a stepwise approach to  18  guide proper scale development.  Specifically, he outlined 8 steps that researchers should consider when developing a scale: 1) Determine clearly what it is you want to measure; 2) Generate an item pool; 3) Determine the format for measurement; 4) Have initial item pool reviewed by experts; 5) Consider inclusion of validation items; 6) Administer items to a development sample; 7) Evaluate the items; and 8) Optimize scale length.  These steps guided the development of the TTI-SM-DSM-IV, and the process for determining the reliability and validity of the measure under scrutiny.  The application of each of DeVellis’ steps is detailed in the methodology section.  Messick’s (1989) discussion on the unitary concepts of validity was used to guide the interpretation of the validity evidence. A series of sub-questions provided evidence to support the overarching research question. For the research questions below, the “Mutism subscale” refers to the Mutism subscale of the TTI-SM-DSM-IV (telephone protocol) and in the TTI-SM-R.  The study aimed to answer 6 sub- questions: 1) What is the reliability of the Mutism subscale? 2) What is the face validity evidence for the Mutism subscale? 3) What is the predictive validity evidence (i.e., diagnosis of SM) for the Mutism subscale? 4) What is the convergent validity evidence for the Mutism subscale in relation to parent- reported measures of SM? 5) What is the convergent and discriminant validity evidence for the Mutism subscale in relation to other measures of nonspecific anxiety? and, 6) What is the concurrent validity evidence for the Mutism subscale in relation to other classroom behaviours of students with SM?  19  The first research sub-question was: What is the reliability of the Mutism subscale?  The development of the TTI-SM-DSM-IV, by Tannock and clinical team in 2003 was derived from theory and clinical practice (steps 1-5 of DeVellis’ guidelines).  Tannock’s clinical research team is located in a large, urban mental health clinic, and has vast training and experience working with a clinical population of children with SM.  Tannock’s team has a distinguished research record.  Based on their academic and clinical expertise and access to clinical populations, Tannock’s team developed the TTI-SM-DSM-IV, a teacher interview protocol used in the assessment of children with SM. The TTI-SM-DSM-IV (and its subscales), although developed in 2003, was never submitted to rigorous reliability and validity trials.  Thus, the first goal of the current study was to evaluate the psychometric properties of the Mutism subscale of the TTI-SM-DSM-IV.  The TTI-SM-DSM-IV, in its entirety, was administered to a clinical (hospital) sample via phone interviews.  Following the evaluation of the TTI-SM-DSM-IV, the second goal of the study was to take the Mutism subscale of the TTI-SM-DSM-IV and develop a standalone paper-pencil instrument (which is referred to as the TTI-SM-R), administer the TTI-SM-R to a community- based (school) sample (step 6 of DeVellis’ guidelines), and then validate the TTI-SM-R. Evaluation of the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R was determined by standard test construction techniques (Cronbach, 1970), which include presenting evidence for reliability and validity (i.e., criterion, construct, concurrent, discriminant, and face; steps 4-8 of DeVellis’ guidelines).  It was hypothesized that the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R would demonstrate strong internal consistency, and moderate-to-high inter-item correlations.  20  The second research sub-question was: What is the face validity evidence for the Mutism subscale?  This question evaluated the face validity of the Mutism subscale by experts and professionals in the child mental health field (i.e., psychologists, school counselors, and speech and language pathologists; Step 4 of DeVellis’ guidelines).  This question was investigated by asking a panel of child mental health professionals to review the TTI-SM-DSM-IV and the TTI- SM-R, and report whether the two instruments measure what the instruments purport to measure. Positive feedback on the content of the TTI-SM-DSM-IV and TTI-SM-R (specifically, the Mutism subscale within these two instruments) would help provide evidence for face validity.  It was hypothesized that the Mutism subscale would provide evidence of face validity, and reviewers would report that the Mutism subscale would appear to be measuring the construct of SM. The remaining research questions address step 7 of DeVellis’ guidelines (evaluating the items).  The third research sub-question was: What is the predictive validity evidence (i.e., diagnosis of SM) for the Mutism subscale?  This question investigated the predictive validity of the Mutism subscale of the TTI-SM-DSM-IV and revised TTI-SM-R.  Specifically, the question aimed to look at whether the scores on the Mutism subscale on both instruments can predict a clinician diagnosis using the ADIS-P (Silverman & Albano, 2004).  It was hypothesized that there would be a strong relationship found between the Mutism subscale scores and ADIS-P diagnosis.  Specifically, students with a diagnosis of SM using the ADIS-P, the gold standard of a clinical interview, would have lower scores on the Mutism subscale compared to students without a diagnosis of SM.  Lower scores on the Mutism subscale are indicative of difficulty speaking in situations when expected to do so, indicating selective mutism in a child.  This hypothesis was identified because both the Mutism subscale and ADIS-P were developed based on theory and the diagnostic criteria of SM found in the DSM-IV-TR (APA, 2000).  Thus,  21  theoretically, the Mutism subscale and ADIS-P should show a moderate to high correlation. Correlations between the Mutism subscale and ADIS-P diagnosis would provide evidence of consequential validity of the Mutism subscale as a screening measure for SM that can be used in schools, which can assist in the early identification of children with SM.  Children were not interviewed due to the age of the participants, thus the child interview portion of the ADIS-P was not used (ADIS-C). The fourth research sub-question was: What is the convergent validity evidence for the Mutism subscale in relation to parent-reported measures of SM?  This research question investigated the convergent validity of the Mutism subscale to another measure of a related concept, the Selective Mutism Questionnaire (SMQ).  It was hypothesized that teacher reports of SM (specifically, the Mutism subscale) would be moderately to highly correlated to parent reports of SM (SMQ).  The reason for this hypothesis was twofold.  First, the Mutism subscale of the TTI-SM-DSM-IV (and the TTI-SM-R) were based on the 6 original questions of the School factor of the SMQ, and second, both of the measures purport to measure mutism (or failure to speak) in students.  Therefore, the two measures should show a moderate to strong relationship. Aside from providing evidence of convergent validity, the fourth research question additionally investigated concordance rates between parent and teacher reports on SM.  Given that past research findings have shown mixed concordance rates for childhood internalizing disorders, it is important to understand the concordance rate for SM if parents and teachers will act as informants in the assessment and diagnosis of SM.  It was hypothesized that parents and teachers would show moderate agreement because parents are provided information on their child’s school behaviours and performance by teachers, thus parent reports can be viewed as an extension of the teacher reports, which should result in moderate agreement.  A moderate to high  22  correlation between the teacher report on the Mutism subscale and the parent report on the SMQ would provide evidence of concordance between teacher and parent reports of SM. The fifth research sub-question was: What is the convergent and discriminant validity evidence for the Mutism subscale in relation to other measures of nonspecific anxiety?  This research question provided evidence as to whether there is a relationship between the Mutism subscale with other measures of related concepts (anxiety), such as the Preschool Anxiety Scale (PAS; Spence, Rapee, McDonald, & Ingram, 2001), and the School Anxiety Scale, Teacher Report (SAS-TR; Lyneham, Street, Abbott, & Rapee, 2008).  It was hypothesized that there would be a moderate correlation between teacher reported Mutism subscale scores to other measures of related concepts (anxiety), using multiple raters.  Specifically, the teacher-reported Mutism subscale should demonstrate a moderate relationship with the parent-reported PAS Social Anxiety subscale, and the teacher-reported SAS-TR total scale score.  Both the PAS and SAS-TR scales and subscales measured social anxiety, which has been found to be related to SM. Thus, moderate correlations indicated the Mutism subscale was tapping into a construct that was related to SM (anxiety, specifically school and social anxiety), yet was also unique to SM (mutism or reluctance to speak when there is a requirement/expectation to do so). It was further hypothesized that the Mutism subscale should have low correlations with the other subscales of the PAS (i.e., the generalized anxiety disorder [GAD], obsessive compulsive disorder [OCD], Fear of Physical Injury and separation anxiety disorder [SAD] subscales).  The reason for this hypothesis was because the Mutism subscale of the TTI-SM- DSM-IV and in the revised TTI-SM-R were expected to have a weaker relationship with constructs that were not believed to be strongly related to SM, such as other anxiety disorder subtypes (not including social phobia).  These other anxiety disorder subtypes included obsessive  23  compulsive disorder (OCD), generalized anxiety disorder (GAD), separation anxiety disorder (SAD), panic disorder, post traumatic stress disorder (PTSD), and specific phobia.  Low correlations between the Mutism subscale and other PAS subscales (that are not theoretically related to SM) provided evidence for discriminant validity of the TTI-SM-DSM-IV. Given the similarities between SM and social phobia, this research also aimed to complement the literature on SM and anxiety.  The research provided an understanding of the relationship between SM and school-based anxiety (i.e., worries about making mistakes at school, or worries that s/he will do badly at school, etc.).  Findings helped identify similar and unique symptoms that students with SM and social anxiety disorder may face.  As there has been more research on all anxiety disorders compared to SM, understanding common and unique symptoms experienced by students with SM and anxiety disorders may help guide future research for the assessment, intervention, and treatment for both groups of students. The sixth, and final, research question was: What is the concurrent validity evidence for the Mutism subscale in relation to other classroom behaviours of students with SM?  As there are currently no standardized teacher reported measures for SM and a lack of other general descriptive information of children with SM in the classroom (McInnes et al., 2004), it was important to investigate whether the Mutism subscale score demonstrated concurrent validity with other atypical classroom behaviours that are found to be related to SM. To measure these other types of childhood classroom behaviours, 8 additional items were administered as part of the TTI-SM-DSM-IV and the TTI-SM-R, which was referred to as the Behaviour subscale.  The 8-item Behaviour subscale probed for school and classroom social participation, behaviours that have been anecdotally and empirically suggested to be more challenging for students with SM than students without SM in previous research studies (Beidel,  24  Turner, & Morris, 1999; Cunningham et al., 2006).  For example, children with SM and social anxiety have been found to have poorer social skills, have fewer friends, and may avoid typical childhood activities such as joining clubs and groups or going to birthday parties.  Thus, a strong relation between the Mutism subscale and the degree of problems with school and classroom social participation, as determined from the Behaviour subscale, would help provide evidence that the Mutism subscale on both instruments are measuring the construct of SM.  It was hypothesized that students with SM would demonstrate more difficulties with respect to social participation in the classroom compared to students without SM. 1.11  Summary    Prevalence rates for SM vary from study to study, depending on the diagnostic criteria used, the age that is sampled, and whether the sample is clinical or community-based.  There is currently a 3 to 6 year delay in referrals in children with SM (Ford et al., 1998, Kumpulainen et al., 1998; Standart et al., 2003).  The delay in referrals may be due to parents not noticing the symptoms of SM, and other referral issues such as parents’ general lack of knowledge about SM, lack of access to mental health services, parent psychopathology, or stigma of a mental health disorder.  Given the issues that have been found in parent reports for child internalizing disorders, the development of a teacher reported measure of SM may be valuable in identifying children with SM at an early age.  A teacher reported measure would also be useful because SM is typically first noticed after a child begins school, and the school is usually the context in which children with SM have the most difficulty functioning.  With the early identification of SM, the lag time between symptom onset and treatment referral may be significantly reduced.  Another area of discussion in the literature is whether SM should be classified as an anxiety disorder, specifically, an extreme variant of social phobia.  There is emerging evidence  25  showing a link between SM and social phobia, however, there are also distinct differences that have been found between children with SM and social phobia.  For example, children with SM fail to speak when expected to do so, although this is not a criterion of social phobia. Additionally, the age of onset for SM is much earlier than that of social phobia. Multiple informant agreement for internalizing and anxiety-related disorders remains inconclusive.  There is a lack of research concerning the agreement among parents and teachers for SM, and some research for children with internalizing disorders reveals low to moderate agreement between teacher and parent reports.  Although SM is categorized as an internalizing disorder, the main symptom of SM (mutism) is observable; therefore, the concordance rate between parent and teachers may be higher for reports of SM than what has been found in the literature for children with other internalizing disorders. Based on the review and evaluation of the current literature, the overarching goal of the study was to develop a teacher-reported measure of SM, and to investigate its reliability and validity.  A teacher-reported measure of SM may help address some of the issues concerning delay in referrals of children with SM, problems with relying on one informant (typically parents for SM diagnosis), and the utility of teacher-reports for SM diagnosis.  The study additionally aimed to understand parent and teacher concordance on measures of SM, symptoms of school and social anxiety in children with SM, and school and classroom social participation for students with SM versus children without SM. The next chapters of this study will present the methodology (Chapter 2), results (Chapter 3), and conclusion and discussion (Chapter 4).   26  2.  Methodology This chapter begins with a review of the overarching research question that will guide the study, followed by the 6 narrow sub-questions that are specifically addressed in this study.  Next, each of the three samples in the study (clinical, community, and combined sample) is discussed in detail.  Information concerning the participants, general procedures, data collection methods, and parent and teacher measures were presented separately because each sample underwent a different set of methodological design and procedures (which are referred to as “Phases” in this study).  Finally, the chapter will conclude with a discussion and rationale of the statistical methods that were used in this study, namely nonparametric analyses. The overarching research question of this study was: To what extent does a teacher reported measure for selective mutism (SM) identify children with SM accurately and consistently?  This broader research question was addressed by six narrower research questions that helped provide evidence of reliability and validity of the TTI-SM.  For the research questions below, the “Mutism subscale” refers to the Mutism subscale of the TTI-SM-DSM-IV (telephone protocol) and in the TTI-SM-R.  Specifically, the study aimed to answer 6 sub-questions: 1) What is the reliability of the Mutism subscale? 2) What is the face validity evidence for the Mutism subscale? 3) What is the predictive validity evidence (i.e., diagnosis of SM) for the Mutism subscale? 4) What is the convergent validity evidence for the Mutism subscale in relation to parent- reported measures of SM? 5) What is the convergent and discriminant validity evidence for the Mutism subscale in relation to other measures of nonspecific anxiety? and,  27  6) What is the concurrent validity evidence for the Mutism subscale in relation to other classroom behaviours of students with SM? These 6 research sub-questions were based on standard test construction and test validation techniques, which include presenting (and interpreting) evidence for reliability and validity (Cronbach, 1970; DeVellis, 1991; 2003; Messick, 1989).  In particular, DeVellis proposed a stepwise model on scale development that helped guide the development of the TTI- SM-DSM-IV (telephone protocol) and the TTI-SM-R (revised paper-pencil version), beginning with item development (Step 1) to an empirical scrutiny of the measure with specified populations (steps 6 and 7), and finally, optimizing the length of the scale (step 8).  To guide the interpretation of the validity evidence, Messick’s (1989) discussion on the unitary concepts of validity was used. 2.1  Overview of Phase 1 (Sample 1) and Phase 2 (Sample 2) The first phase of this research study involved the development and evaluation of a teacher-reported interview protocol, the TTI-SM-DSM-IV (Tannock et al., 2003).  The TTI-SM- DSM-IV was administered to the first development sample (Sample 1, clinical sample) as part of a larger study by Manassis et al. (2007) on language and cognition.  This first phase involved examining data from a subset of teachers extracted from the larger data set (secondary use of data). The second phase of this research study involved modifying the TTI-SM-DSM-IV telephone interview protocol to a paper-pencil version, the TTI-SM-R, and then evaluating this revised version.  The rationale for developing a paper-pencil version of the telephone interview protocol was an effort to create a more accessible (a measure that does not require administration by a trained clinician), user-friendly (a measure that can be completed via paper-pencil and  28  available for use in schools), and brief (a measure that can be completed in 10 minutes) teacher- reported measure for SM.  The TTI-SM-R was then administered to the second development sample (Sample 2, community-based sample) Finally, the third phase of the study involved combining data from Samples 1 and 2 from the first 2 phases for analysis to produce a larger overall sample (Sample 3).  A justification was made for combining the data for a final analysis.  Figure 2.1 illustrates the flow chart of the current study.   29   Figure 2.1: Flow Chart for the Development and Validation of the TTI-SM-DSM-IV and TTI- SM-R  30  2.2  Sample 1: Development and Validation of TTI-SM-DSM-IV with a Clinical Sample The first phase of this study was part of a larger multi-site investigation of language and cognitive skills in children with diagnosed anxiety and SM in a clinical population at three large urban children’s hospitals in Canada (Manassis et al., 2007).  Although the TTI-SM-DSM-IV was developed in 2003, the psychometric properties of the measure (and its subscales) were never investigated.  Thus, the purpose of Sample 1 of this study was to specifically validate the Mutism subscale of TTI-SM-DSM-IV. Parents of children referred for SM to any of the three anxiety disorders clinics at the hospitals were recruited for participation in this phase of the study.  Parents of the referred children were asked to complete a set of questionnaires, and were additionally asked whether they would consent to having the research team contact the child’s teacher regarding participation in the TTI-SM-DSM-IV validation study.  Each parent of the child participants was administered a clinical interview (using the Anxiety Disorders Interview Schedule for DSM-IV: Parent Version; ADIS-P; Silverman & Albano, 2004) for a diagnosis of an anxiety disorder, SM, another disorder, or no diagnosis.  With consent from the parents enrolled in the clinical sample, trained graduate-level research assistants with experience in school or counseling psychology, contacted classroom teachers of children diagnosed with SM to participate in the research and complete the TTI-SM-DSM-IV via telephone interview. 2.2.1  Participants. There were a total of 91 children in the TTI-SM-DSM-IV validation study (Manassis et al., 2007), but only 33 teachers of the children in the study consented to participation in the TTI- SM-DSM-IV validation study.  Of these 33 children, a total of 19 children met a clinical diagnosis of SM, as determined by the ADIS-P (Silverman & Albano, 2004).  Ten children had a  31  diagnosis of an anxiety disorder (i.e., social phobia, generalized anxiety disorder, specific phobia, separation anxiety disorder, etc.), and four children did not meet diagnostic criteria of SM or an anxiety disorder.  A decision was made to exclude the four children without a diagnosis of SM or anxiety disorder from analyses because the goal of this study was to compare behaviours of a clinical sample of children diagnosed with SM versus children diagnosed with any anxiety disorder.  For that reason, there were a total of 29 Kindergarten to Grade 5 students with a diagnosis of SM (SM group) or an anxiety disorder (ANX group) that made up the clinical sample in the first phase of this study. 2.2.2 General procedures. Details specific to Phase 1 of this study (i.e., development and validation of the Mutism subscale of the TTI-SM-DSM-IV telephone interview protocol), are discussed below.  Other information concerning the larger, multi-site study, such as the design, participants (the 91 child participants, inclusion/exclusion criteria for participation, sampling and recruitment procedures, etc.), settings (locations of sites), measures and procedures are described in more detail by Manassis and colleagues (2007).  Because the current study extends findings from the 2007 Manassis study, some elements of the original study will be reviewed, thus establishing a linear development and validation of the measure under scrutiny, the TTI-SM-DSM-IV. 2.2.3  Teacher measures. The teacher-reported measure of SM that was used in Manassis et al.’s (2007) original study was the TTI-SM-DSM-IV.  This section will include a description of the TTI-SM-DSM-IV, mode of administration, and scoring of the items.  32  2.2.3.1 Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV; Tannock et al., 2003). In response to the lack of validated teacher-reported measures for SM, Tannock and her colleagues (2003) developed a semi-structured (verbal) telephone interview protocol, the Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM- IV).  The TTI-SM-DSM-IV telephone interview protocol queried teachers on behaviours associated with SM, including classroom behaviours, social interactions, communication patterns, and externalizing behaviours.  The telephone interview protocol took approximately 1 hour to administer to teachers and was conducted via telephone interview by trained mental health clinicians and/or researchers.  The purpose of this telephone interview protocol was to assist in the early identification and diagnosis of SM in students, to ease administration of a face- to-face clinical interview (less time intensive for teachers), and to increase response rates (teachers are potentially more likely to respond to a phone interview than meeting face-to-face). The telephone interview protocol is presented in Appendix A. 2.2.3.1.1 Development and description of the TTI-SM-DSM-IV. Below is an overview of how DeVellis’ (1999; 2003) stepwise model was applied to the development (and evaluation) of the TTI-SM-DSM-IV and a description about the items that make up the TTI-SM-DSM-IV. First, Tannock and her colleagues used current theory and research as a guide to their development of the TTI-SM-DSM-IV (i.e., symptoms and impairments of children with SM, current parent measures of SM that are being used, and the diagnostic criteria for SM in the DSM-IV-TR [APA, 2000]).  Tannock’s team also determined that “selective mutism” was the construct they wanted to measure, and they wanted their measure to be reported on by teachers.  33  The procedure used by Tannock’s team was in line with DeVellis’ step 1 of scale development, which was to determine what is to be measured.  Next, Tannock’s team, at their clinical site, developed an item pool (DeVellis’ step 2) to measure SM as reported by classroom teachers. The TTI-SM-DSM-IV is composed of 54 items, which make up 5 subscales.  The 5 subscales are: 1) Selective Mutism subscale (15 items), 2) Verbal and Nonverbal Communication with Teachers subscale (14 items), 3) Verbal and Nonverbal Communication with Peers subscale (10 items), 4) Other School and Classroom Social Participation Behaviours subscale (8 items), and 5) Externalizing behaviours subscale (7 items).  For the purpose of this paper, the Selective Mutism subscale (15 items) will be referred to as the “Mutism subscale”, and the Other School and Classroom Social Participation Behaviours subscale (8 items) will be referred to as the “Behaviour subscale”. As part of the TTI-SM-DSM-IV, there were also qualitative questions about non-SM or non-social behaviour items (e.g., response to psychotropic medication [if the child was taking medication for anxiety], strategies and interventions being used in the classroom, classroom seating arrangements, etc.).  Only the 15-item Mutism subscale and the 8-item Behaviour subscale are discussed in the study because these were the 2 subscales that were of interest in the current validation study. Tannock and her colleagues’ development of the TTI-SM-DSM-IV Mutism subscale was largely based on the School factor items of the Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008).  The SMQ is a validated 23-item parent-report measure (paper-pencil) that probes parents about speaking behaviours in their children across school (6 items), home/family (6 items), and social situations (outside of school; 5 items) and includes an additional 6 items concerning interference/distress.  The SMQ has well established psychometric properties  34  (Bergman et al., 1999; 2008).  Using the 6 SMQ School factor items, Tannock and her colleagues developed an additional parallel item (with permission from the instrument’s author), to probe for the same behaviour but in a different context (i.e., small group versus large groups).  The additional item was added in an effort to develop a more thorough and comprehensive teacher- reported measure of SM that would evaluate child classroom and school behaviours across several contexts.  For example, one of the original SMQ questions was, “He/she talks to selected peers (his/her friends) in the class.”  The parallel item to probe for speaking behaviours in a different setting at school was: “He/she talks to selected peers (friends) during transitions or at recess/lunch.” There was only 1 item of the SMQ in which there was no parallel item developed; this question was context specific (large group context), and thus it was not possible to generate a question probing for speaking behaviours in a different context.  This question was: “When called on by me (the teacher), the student answers.”  The process of developing a parallel item for each of the 6 SMQ School factor items resulted in the development of 11 items in the Mutism subscale.  As part of the Mutism subscale, Tannock and her colleagues were additionally interested in including 4 questions concerning functional impairments in children with SM (2 items) and non-verbal communication (2 items).  As a result, there were a total of 15 items that made up the Mutism subscale of the TTI-SM-DSM-IV. The response option format for the Mutism subscale (DeVellis’ step 3) remained the same as for the original SMQ items, which was a 4-point frequency-type Likert scale.  The response options were: Always (3), Often (2), Sometimes (1), and Never (0).  It is of note that in the original TTI-SM-DSM-IV interview protocol (see Appendix A), the Mutism subscale is coded (named) “SMQ items” because they were derived based on the SMQ (Bergman et al., 1999;  35  2008).  For the purpose of this study, this subscale was renamed “Mutism subscale” to avoid confusion with the SMQ, which is a parent-reported measure that is used in this study. Tannock and her colleagues also developed an 8-item Behaviour subscale, which is made up of items that measure other school and classroom social participation behaviours.  These school and classroom social participation questions were largely derived from theory and anecdotal and empirical information concerning students with SM and included behaviours that have been suggested to be more challenging for students with SM than students without SM. These 8 items are: 1) In general, what is this student’s mood and attitude like when s/he comes into the classroom? 2) What does (Child’s Name) typically do during the National Anthem? 3) When something very funny happens (e.g., student talks about a very funny event), how does (Child’s Name) typically respond? 4) Do you ever have students come up and write on the board or calendar?  If Yes, does this student come up? 5) Can he/she work with any group of peers or only specific peers? 6) In general, how well does (Child’s name) handle transitions within the class, going in and out of the class, and recess times? 7) Does (child’s Name) have any close friends? and 8) In general, how do the other students interact with (Child’s name)?  The response option format for these 8 social participation items (DeVellis’ step 3) was presented on a 3-point Likert type scale (that indicated the severity of a behavioural problem).  Response options for these 8 items were: No problem or only mild problem, occurs occasionally, not impairing (0); Marked problem, occurs often, impairing (1); Severe problem, occurs most or all of the time, very impairing (2). The format of the TTI-SM-DSM-IV interview protocol was ordered by school and classroom context, and began with questions concerning child behaviours during school arrival time, and then followed by child behaviours during large group contexts, small group contexts,  36  individual seat work, and transition points (i.e., recess and lunch).  The items of the Mutism subscale and Behaviour subscale were embedded throughout the TTI-SM-DSM-IV interview protocol and were not presented sequentially in numerical order (please refer to Appendix A). The reason the telephone protocol was organized in this manner, rather than presenting each question in order, was to facilitate an organized and efficient interview method for teachers where questions are categorized by school and classroom context.  For example, Tannock and her colleagues believed it would be easier to administer sets of questions to teachers about behaviours in large group contexts, and then move on to behaviours in small group contexts, and so on and so forth. DeVellis’ fourth step in scale development involved the TTI-SM-DSM-IV (specifically, the Mutism subscale) being reviewed by an expert panel prior to the administration of this instrument to Sample 1.  Tannock’s clinical research team reviewed the items of the TTI-SM- DSM-IV.  Tannock and her team are located across three large, urban mental health clinics, and the team has training and experience working with clinical populations of children with SM.  The team also has a distinguished research record.  Next, the TTI-SM-DSM-IV interview protocol was administered as part of a larger study on language and cognition by Manassis and her colleagues (2007), which is step 6 of DeVellis’ model (administering the items to a development sample). Next, a validity study was conducted in Phase 2 of the study, where the individual items of the TTI-SM-DSM-IV were reviewed (DeVellis’ step 7).  The results of this validity study are presented in the results section.  Finally, the last step (step 8) was to optimize the scale length for both brevity and reliability, which resulted in the development of a revised paper-pencil version of the TTI-SM-R (Phase 2 of the study). 2.2.3.1.2 Mode of administration of the TTI-SM-DSM-IV.  37  In Manassis and her colleagues’ 2007 study, trained graduate-level research assistants and supervisors administered the entire TTI-SM-DSM-IV protocol (54 items) to teachers of participating children in the clinical sample verbatim over the phone.  Interviews lasted approximately 1 hour.  Teachers were informed of the response options (verbatim) for each question on the TTI-SM-DSM-IV.  For the 15-item Mutism subscale of the TTI-SM-DSM-IV, response options were presented orally on a Likert-type scale (that indicated frequency of a behaviour).  Again, the response options were: Always (3), Often (2), Sometimes (1), and Never (0).  For the 8-item Behaviour subscale regarding school and classroom social participation, response options were also presented orally on a Likert-type scale (that indicated the severity of a behavioural problem).  The response options were: No problem or only mild problem, occurs occasionally, not impairing (0); Marked problem, occurs often, impairing (1); or Severe problem, occurs most or all of the time, very impairing (2). 2.2.3.1.3 Scoring the TTI-SM-DSM-IV. For the 15-item Mutism subscale, the 4-point Likert-type scale responses were tallied, and a mean score was then computed for analysis.  The 8-item Behaviour subscale, regarding school and classroom social participation responses, was recomputed into categorical response options (yes or no) for analysis.  The reason for this was because after consultation with Tannock and her colleagues, it was determined that what was of interest was whether a student had any problems with these behaviours (yes or no), rather than the severity/intensity of these behaviours (none, moderate or severe).  To create a categorical response option (or a dichotomous variable), response options “1” and “2” (Marked Problem and Severe Problem, respectively) were grouped together for a response option of “1” (Marked to Severe Problem), and “0” (No problem) was  38  maintained.  As a result, there was no total score that was computed for the 8-item Behaviour subscale. 2.2.4  Parent measures. There were several parent measures that were used in the original 2003 study, including measures of anxiety, language and cognitive skills.  Only the parent measures that are relevant to this current study will be discussed.  Specific details concerning the original study by Manassis and colleagues (2007) are documented elsewhere.  This following section will include a review of the Anxiety Disorders Interview Schedule for Parents (ADIS-P; Silverman & Albano, 2004; Silverman et al., 2001), and the Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008). 2.2.4.1 Anxiety Disorders Interview Schedule for DSM-IV: Parent and Child Version (ADIS-C/P; Silverman et al., 2001). The Anxiety Disorders Interview Schedule for DSM-IV: Parent and Child Version (ADIS- C/P; Silverman, Saavedra, & Pina, 2001) is the most widely used and accepted clinical interview that is used in the field of childhood and adolescent anxiety (Compton, et al., 2004; Summerfeldt & Antony, 2004).  The ADIS-P/C is a measure with established reliability evidence for anxiety disorder symptoms and diagnosis in children using the DSM-IV-TR criteria (Silverman et al., 2001).  Three types of scores are derived from the ADIS-C/P: a diagnosis of child psychopathology, a symptom scale score, and parent/clinician impairment ratings.  The ADIS-P/C has demonstrated high reliability for symptom scale scores (separation anxiety disorder, social phobia, specific phobia, and generalized anxiety disorder), and demonstrated good to excellent reliability for combined diagnoses and for child-only and parent-only interview information (Silverman et al., 2001).  39  The correlation between two clinicians’ diagnoses on the ADIS-P/C ranged from kappa = .65 to .88, which is considered good to excellent (Silverman et al., 2001).  Intraclass correlation (ICC), which measures the reliability of the ADIS-P/C symptoms scale scores, ranged from .81 to .96, or strong.  Finally, correlations for impairment ratings by parents from Time 1 to Time 2 (consistency of ratings) was moderate to strong, with correlation coefficients ranging from r = .56 to .84.  Further, correlations for impairment ratings by clinicians ratings from time 1 to time 2 ranged from r = .84 to 90.  Concurrent and convergent validity evidence has been established between the ADIS-P/C with the Multidimensional Anxiety Scale for Children (MASC; March, Parker, Sullivan, Stallings, & Conners, 1997) and the SMQ (Bergman et al., 1999; 2008) by several teams of researchers (Silverman & Albano, 2004; Wood, Piacentini, Bergman, McCracken, & Barrios, 2004). Although the ADIS-P/C is an integrated parent-child interview (Silverman, & Albano, 2004), the child interview portion of the (ADIS-C) was not administered in this study due to the young age of study participants and reluctance of children with SM to speak to unfamiliar adults. Only the parent version, ADIS-P, was administered as part of the larger study for Manassis and her colleagues (2007).  The ADIS-P clinical interview was used to confirm the presence and/or absence of an anxiety disorder (ANX) and/or SM, or no diagnosis (normal control, NC) with parents.  Group membership (SM, ANX or NC) in the study was therefore based on the ADIS-P diagnosis. 2.2.4.2 Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008). The Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008) was used as the parent reported measure of SM in the current study.  The SMQ is a paper-pencil parent-report measure of a child’s frequency of speaking across various settings (school, home and other social  40  situations).  The SMQ has a total of 23 items, which are rated using a 4-point Likert scale (Always [3], Often [2], Sometimes [1], and Never [0]).  For items 1-18 (questions concerning speaking behaviours at school, home and other social situations), a low score would indicate more problems with respect to failure to speak, whereas a high score would indicate fewer problems with speaking behaviours.  For items 19-23 (questions concerning interference and distress), a low score would be indicative of less interference and distress. The SMQ has demonstrated strong internal consistency and reliability, and good discriminant, concurrent and convergent validity (Bergman et al., 2008).  The SMQ has shown a strong relationship to other validated measures of social anxiety, including the MASC (March et al., 1997) and the Social Anxiety Scale for Children – Revised (SASC-R; La Greca & Stone, 1993).  Specifically, the SMQ has shown a strong (but negative) relationship to the MASC Social Anxiety subscale (r = -.62, p ≤ .001) and the SASC-R Total scale with talking items (r = -.52, p ≤ .001).  The reason for the negative correlations is due to the anchors of the response options being reversed on the SMQ compared to the MASC and SASC-R.  However, the SMQ has not shown a strong relationship to other subscales within the MASC (e.g., Harm Avoidance, Separation Anxiety, Physical Symptoms) nor with subscales of the SASC-R (e.g., items that did not involve speaking), therefore demonstrating discriminant validity.  Finally, there is a strong relationship that has been found between the SMQ and the ADIS-P (r = -.67, p ≤ .001; Silverman & Albano, 2004). Factor analysis revealed a 3-factor solution: School, Home/Family, and Other (Public). Internal consistency for the Total scale was  = .97, School was  = .97, Home/family was  = .88, and Other (Public) was  = .96 (Bergman et al., 2008).  The SMQ also demonstrated good discriminant, concurrent and convergent validity, and sensitivity to treatment response (Bergman  41  et al., 2008).  Thus, the SMQ appears to be a psychometrically sound measure of the core features of SM as described in DSM-IV-TR.  The SMQ has been used for research purposes by Bar-Haim et al. (2004), Chavira et al. (2007), Fung, Manassis, Kenny and Fiksenbaum (2002), Manassis et al. (2003), and Manassis et al. (2007) since it was developed.  A copy of the SMQ is available in Appendix D. 2.2.5  Summary of findings based on Sample 1 (clinical sample). The purpose of Phase 1, using data from parents and classroom teachers of children in the clinical sample, was to develop a teacher-reported measure of SM (TTI-SM-DSM-IV) and then to validate one subscale, the Mutism subscale, in the TTI-SM-DSM-IV.  Findings for this study would provide support as to whether the TTI-SM-DSM-IV, and the Mutism subscale, merits further research.  If the Mutism subscale demonstrates adequate evidence for reliability and validity, the two final steps of DeVellis’ model (step 7 and 8) would be used, which is to evaluate individual items of the measure and then optimize the scale length for both brevity and reliability.  More specifically, these final steps would involve the development of a revised paper-pencil version, the TTI-SM-R (Phase 2), and the administration of this revised measure to a second development sample (Sample 2, community-based sample) for validation. 2.3 Sample 2: Development and Validation of TTI-SM-R with a Community-based Sample The purpose of the second phase was to extend the findings from the first phase.  Thus, the extension of Phase 1 would be to take the same instrument (TTI-SM-DSM-IV; Tannock et al., 2003), and submit it to evaluation with a different population (community-based sample).  Phase 2 of the study examined if teachers, who are the key audience of interest for this study, could accurately evaluate SM not only in children from clinical settings, but also from more typical  42  populations, or the general community.  However, due to some of the challenges associated with administering lengthy telephone interviews to teachers, an additional purpose of the second phase was to extract the Mutism subscale from the TTI-SM-DSM-IV telephone protocol and develop a revised paper-pencil measure, the TTI-SM-R, using this subscale. 2.3.1  Participants. Participants in Sample 2 were independent from Sample 1.  There were a total of 30 Kindergarten to Grade 4 students who made up sample 2.  There were 8 students who had a diagnosis of SM, 19 students with no diagnosis, and three students with an anxiety disorder. Additional details concerning recruitment, procedures, and measures are presented below. 2.3.2 General procedures. In the second phase of the research project, nine school districts in a western Canadian province were invited to participate.  These districts are considered large, with over 3000 students enrolled in Kindergarten to Grade 3, and were specifically targeted for inclusion in this second phase of the study.  Letters to school superintendents were sent to districts, describing the study and research procedures.  Five school districts agreed to participate. A 3-stage multiple-gating procedure was used (originally discussed in Cronbach & Gleser, 1965) to recruit research participants, their parents, and their teachers.  Following school district ethics approval and school superintendent approval, superintendents, school principals, school psychologists, speech and language pathologists, and school counsellors were contacted via telephone, letter, and/or email asking them to recruit students into the study.  These select school personnel (principal, psychologist, speech and language pathologists, and counselors) were asked to distribute parent and teacher consent forms to every student referred for suspected SM or elevated anxiety in the primary grades.  For school personnel who did not have specific  43  SM and anxiety referrals on their caseload, they would ask primary teachers with whom they worked whether there were any children who would meet the criteria of “won’t talk” and are “shy or timid”.  Based on a previous study by Kristensen (2001), the two questions on the Child Behavior Checklist (CBCL; Achenbach, 1991) that differentiated children with SM from typical children were the questions concerning “won’t talk” and “shy or timid”. Teachers and school personnel were additionally asked to nominate children for the study whose family’s first language must be English (not ELL background).  The reason for requesting only families whose first language was English was due to the challenges that may be associated with enrolling ELL children and families into this validation study, which included: language barriers of parents (which would impact the proper assessment and diagnosis of children with SM), and the DSM-IV-TR diagnostic criterion for SM (failure to speak cannot be due to a lack of knowledge of or comfort with the spoken language required in the situation). Once these children (who “won’t talk” and “shy or timid”) were identified, parent consent packets were sent home by the nominating teachers, which included a consent form to participate in the research, a copy of the Preschool Anxiety Scale (PAS; Spence et al., 2001) and the Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008).  A self-addressed stamped envelope, with directions for parents to mail the documents back to the University (consent form, and the PAS and SMQ), were provided.  Upon receipt of the parent consent packet and measures at the University, a trained graduate research assistant contacted the parents to conduct a telephone clinical interview, the ADIS-P, to confirm nominations and ratings. It is of note that for each student with suspected SM participating in this phase of the study, teachers were asked to additionally nominate three “typical” students to be recruited into the study to form a control group.  Parents of non-anxious or mute students were predicted to be  44  less interested in participating in this phase of the study, thus over-enrollment was planned for the matched controls (e.g., three control students for each SM child).  Control students were matched on grade and gender.  The matched peers were students who were considered “typical” students in their class, who were not too shy but not too outgoing.  These control students were then invited to take part in the study.  Interested parents completed the consent forms and parent forms (PAS and SMQ) and mailed these back to the researcher in self-addressed stamped envelopes. 2.3.3 Teacher data collection procedures. Following the receipt of parent consent, teachers were contacted via study recruitment letters to seek consent for participation in the study.  Teachers were mailed a set of teacher materials, which included one set of materials for each student identified in the teacher’s class. Teacher packets included a teacher consent form, one copy of the Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-pencil Version (TTI-SM-R), and the School Anxiety Scale, Teacher Report (SAS-TR; Lyneham et al., 2008) for each student. Instructions were also included on how to complete each of the rating scales. Teachers were asked to mail the materials back in a self-addressed stamped envelope to the researcher.  There were teachers who initially agreed to participate; however, following the receipt of all parent materials, seven teachers from a possible pool of 41 were no longer interested in participating in the study.  One teacher indicated that she did not have time to complete the forms.  Another teacher indicated that the child who was nominated does not have SM (the student was part of our NC group), and was not sure why this child was enrolled in the study.  For the other five teachers, reasons for non participation were not provided.  Three students changed schools and their new teachers were not interested in participating in the study.  45  One parent requested that we not contact the school to collect information from the teacher.  At the conclusion of the study, school principals and teachers were sent a letter to thank them for their participation and to debrief them about the study. 2.3.4 Teacher measures. The teacher measures that were administered in Phase 2 of the study (community-based sample) included the: Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-pencil Version (TTI-SM-R), and School Anxiety Scale, Teacher Report (SAS-TR; Lyneham et al., 2008).  Teacher measures took approximately 15 minutes to complete (per student). 2.3.4.1 Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-Pencil Version (TTI-SM-R). The first step in Phase 2 of this validation study required revising the oral protocol of the TTI-SM-DSM-IV into a shorter paper-pencil version, which is referred to as the “TTI-SM-R” from this point forward.  A shorter version of the TTI-SM-DSM-IV was needed due to the time constraints of teachers, and the feasibility of administering a 1-hour interview to teachers in the “real world”.  Similar to the original TTI-SM-DSM-IV, the revised TTI-SM-R (paper-pencil version) included demographic information concerning the student, and only 2 of the 5 original subscales: 1) Mutism subscale (that probe for SM), and 2) Behaviour subscale (that probe for other school and classroom social participation behaviours).  Other qualitative questions concerning any interventions that have been tried were also included at the end of the measure. Please see Appendix B for a copy of the TTI-SM-R paper-pencil version.  46  2.4.4.1.1 Development and description of the TTI-SM-R (paper-pencil version). Similar to the development TTI-SM-DSM-IV in Phase 1, DeVellis’ stepwise model of scale development (DeVellis, 1991; 2003) was used to guide the modification and validation of the revised TTI-SM-R (paper-pencil version) in Phase 2.  Below is an overview of how DeVellis’ model was applied to the development of the TTI-SM-R, a description about the items that made up the TTI-SM-R, and administration and scoring procedures for the measure. Following the administration of the TTI-SM-DSM-IV telephone protocol to the first development sample (clinical sample), the measure was revised into a paper-pencil version (TTI- SM-R) to help address issues concerning the time involved in administering a lengthy telephone protocol, and to develop a more accessible paper-pencil measure that can be used in schools.  Of the 54 items (comprising 5 subscales) of the TTI-SM-DSM-IV, a decision was made to include only the Mutism subscale (15 items) and Behaviour subscale (8 items) on the revised paper- pencil version, resulting in a total of 23 items in the paper-pencil version.  DeVellis’ final steps of scale development (steps 7 and 8), which involved evaluating the items of the scale and optimizing the length of the measure, helped determine which items would be included in the revised TTI-SM-R paper-pencil version. With the newly developed TTI-SM-R, which included the Mutism subscale and the Behaviour subscale, the measure needed to be reviewed again for accuracy, clarity and content by others from an expert stance.  The review had to be conducted prior to re-administering the measure to a second development sample.  This is an effort to establish face validity (and content validity) of the revised TTI-SM-R paper-pencil version (step 4 of DeVellis’ guidelines, 1999; 2003).  47  To establish face (and content) validity, the TTI-SM-R was distributed to a group of 50 school psychologists, special educators, speech and language pathologists, school counsellors and social workers in British Columbia at a local professional development workshop for school and community mental health employees.  These professionals were selected to review the instrument for feedback and input following a daylong in-service workshop about SM and anxiety disorders in children.  This audience was deemed to be representative of school mental health professionals and have sufficient expertise on SM to comment about the appropriateness of the TTI-SM-R for a community sample.  At the end of this daylong workshop, each participant was asked to provide feedback on the TTI-SM-R (paper-pencil version).  All workshop attendees provided verbal consent to participate and were given a copy of the TTI-SM-R.  Participants were given a construct definition of SM, based on the DSM-IV-TR diagnostic criteria. Next, the mental health professionals were asked to comment on the format, length, clarity, wording, and relevance of the TTI-SM-R scale and its items.  The group of mental health professionals was also asked to suggest additional questions they thought would tap into the construct of “selective mutism”.  Finally, each participant was asked to complete the TTI-SM-R while thinking about a student with whom they currently work (or have worked with in the past) who was described as “shy” and “won’t talk”. In response to the 50 school mental health clinicians’ and researchers’ suggestions, one major change to the TTI-SM-R (paper-pencil version) was the addition of an item (item #16) that probed about the audibility of the child’s speech in the class.  This question was: “His/her speech is sufficiently loud to be easily heard.”  Other minor changes were noted, such as spacing and formatting of the items on the page.  Otherwise, the group of reviewers unanimously agreed that the TTI-SM-R, specifically, the Mutism subscale, appeared to measure what it was purported to  48  measure.  Step 5 of DeVellis’ (1999; 2003) guidelines was to consider the inclusion of validation items.  Based on the feedback from the professional field-based review, the revised TTI-SM-R paper-pencil version included the original items of the Mutism subscale and an additional, 16 th  item. The next step (step 6) involved the administration of the revised TTI-SM-R to a development sample.  This sample was a community-based sample (Sample 2).  After validating the results with the clinical sample (Sample 1), it was important to validate the revised TTI-SM-R paper-pencil version with a community-based sample (Sample 2) using a less stringent format (paper-pencil version). The next step in scale development (step 7) was to evaluate the individual items.  This was done by conducting a second validity study using Sample 2.  DeVellis suggested evaluating various pieces of reliability and validity evidence in determining which items to retain.  DeVellis (1999; 2003) also recommended that one’s evaluation of the reliability evidence should be driven on the type of measure that is created.  For example, measures that are intended for screening, diagnostic purposes, and/or high-stakes, should have higher reliabilities (mid .90s range) compared to low stakes tests.  Finally, the last step (step 8) was to optimize the scale length for both brevity and reliability.  The last step was done by examining the extent of covariation among items and the number of items in the scale, and making a decision about the items that should be included in the final version of the TTI-SM-R. 2.3.4.1.2 Mode of administration of the revised TTI-SM-R (paper-pencil version). The paper-pencil TTI-SM-R was given to teachers to complete as part of the teacher questionnaire packet.  The TTI-SM-R took approximately 10 minutes to complete.  For the Mutism subscale of the TTI-SM-R, response options were presented on a frequency-type Likert  49  scale.  The response options were Always (3), Often (2), Sometimes (1), and Never (0).  For the Behaviour subscale, response options were presented on a Likert-type scale: No problem (0), Marked Problem (1), and Severe Problem (2).  A copy of the TTI-SM-R is available in Appendix B. 2.3.4.1.3 Scoring of the revised TTI-SM-R (paper-pencil version). Scoring for the Mutism subscale and the Behaviour subscale on the revised TTI-SM-R was the same as the scoring procedures used for the TTI-SM-DSM-IV (interview protocol).  To score the Mutism subscale, Likert-type scale responses were tallied, and a mean score was then computed.  For the Behaviour subscale, questions were again recomputed into dichotomous response options because it was determined that what was of interest was whether a student had any problems with these behaviours (yes or no), rather than the intensity/severity of these behaviours (none, moderate or severe).  In effect, response options “1” and “2” (Marked Problem and Severe Problem, respectively) were grouped together for a response option of “1” (Marked to Severe Problem), and response option of “0” (No problem) was maintained.  The scoring procedure of the Behaviour subscale was the same procedure that was used in the TTI- SM-DSM-IV. 2.3.4.2  School Anxiety Scale – Teacher Report (SAS-TR; Lyneham et al., 2008). The SAS-TR (Lyneham et al., 2008) is a 16-item questionnaire to query for anxiety symptoms in children ages 5 to 12.  Each question is answered on a 4-point Likert-type scale.  As a paper-pencil measure, the SAS-TR takes approximately 5-minutes to complete (Lyneham et al., 2008).  The psychometric properties of the scale, as reported by the authors, are quite strong. Internal consistency of the SAS-TR, as measured using Cronbach’s alpha, is  = .93 (p ≤ .001), indicative of a measure with strong reliability.  Internal consistency of each of the 2 subscales  50  was also very high (r = .90 to .92).  Positive correlations were reported with other teacher reported-scales, such as the Strengths and Difficulties Questionnaire (Goodman, 1997), which has sound psychometric properties. The SAS-TR also demonstrated discriminant validity, using a Linear Mixed Models procedure, between children with and without anxiety using both clinical and community samples of children (Lyneham et al., 2008).  The community sample consisted of students between the ages of 5 and 12 years old from 16 different schools.  Factor analysis revealed 2 factors (subscales) in the SAS-TR: social anxiety and generalized anxiety.  Given the more recent publication of this measure (2008), there is a lack of published studies available documenting the use of the SAS-TR in research studies.  A copy of the SAS-TR is available in Appendix C. 2.3.5 Parent data collection procedures. Parents were contacted via telephone following the receipt of their consent form and parent measures (Selective Mutism Questionnaire [SMQ]; Bergman et al., 1999; 2008, and Preschool Anxiety Scale [PAS]; Spence et al., 2001) at the University.  Interview dates and times were scheduled with parents for the administration of the Anxiety Disorders Interview Schedule for DSM-IV: Parent Version (ADIS-P; Silverman & Albano, 2004).  At this time, the researchers obtained verbal informed consent from the parents and reminded them that the interview was audio taped.  A trained graduate level research assistant administered the ADIS-P via telephone. Parent procedures for both groups of children (SM and NC) were the same. All ADIS-P parent interviews were conducted via telephone.  In a previous study by Rohde, Lewinsohn, and Seeley (1997), telephone interviews were found to be a reliable, valid, and cost-effective alternative to face-to-face interviews for the diagnosis of DSM-IV-TR Axis I and II disorders.  Lyneham and Rapee (2005) directly investigated the use of telephone versus  51  face-to-face interviews using the ADIS-P, and their results also show that telephone interviews are a good alternative to face-to-face interviews.  Thus, all parent interviews were conducted via telephone.  Digital audio technology (DAT) was used, which consisted of an Olympus VN- 4100PC Digital Voice Recorder, Olympus TP-7 Telephone Recording Device, and Olympus ME-15 Microphone.  The advantages of DAT are outlined in Modaff and Modaff (2000) including: higher fidelity and more reliable audio recordings, greater portability, and cost effectiveness. The primary researcher and trained graduate level research assistants (RAs) conducted parent interviews using the ADIS-P with each parent.  Supervision was provided by a licensed psychologist in Canada.  The researcher and RAs received ADIS-P/C training by an expert ADIS- P/C trainer.  Interrater agreement between the researcher and RAs on clinical diagnoses was 100%.  Children with a positive diagnosis on the ADIS-P (interference rating of 4 and above were considered to be in the “clinical” range for an anxiety disorder), were contacted by the licensed psychologist for a referral for further assessment and treatment at a local community- based mental health center. 2.3.6 Parent measures. Parent measures for Sample 2 included the: Anxiety Disorders Interview Schedule for DSM-IV: Parent Version (ADIS-P; Silverman, & Albano, 2004), Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008), and the Preschool Anxiety Scale (PAS; Spence et al., 2001). The ADIS-P parent interview took approximately 1 hour to conduct.  Parent measures (SMQ and PAS) took 15-20 minutes to complete (per child).  52  2.3.6.1 Anxiety Disorders Interview Schedule for DSM-IV: Parent Version (ADIS- P; Silverman, & Albano, 2004; Silverman et al., 2001). The ADIS-P (parent version) was administered via telephone interviews to participating parents of Sample 2.  The ADIS-P was used to confirm the presence and/or absence of an anxiety disorder and/or SM.  Group membership (SM or NC) in the study was based on the ADIS-P diagnosis.  ADIS-P diagnosis was the dependent variable.  It is of note that all ADIS-P interviewers were kept blind to the teacher TTI-SM-R scores of Phase 2 participants. 2.3.6.2 Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008). The Selective Mutism Questionnaire (SMQ; Bergman et al., 1999; 2008) was the parent- reported measure of a child’s frequency of speaking across various settings (school, home and other social situations).  Please refer to section 2.2.4.2 (above) for details concerning the psychometric properties and details of the SMQ. 2.3.6.3 Preschool Anxiety Scale (PAS; Spence et al., 2001). The Preschool Anxiety Scale (PAS; Spence et al., 2001) is a 34 item paper-pencil measure to query for symptoms of anxiety in preschool children.  The PAS is appropriate for children ages 3 to 6.5.  The majority of the questions are answered on a 5-point Likert-type scale, ranging from “0” (Not True At All) to “4” (Very Often True).  There is one question (item number 29) that requires a “Yes” or “No” response, and parents are asked to follow-up on this question by providing a brief description.  The PAS takes approximately 5 minutes to complete (Lyneham et al., 2008). The psychometric properties of the PAS are quite strong.  A factor analytic study by Spence and her colleagues (2001) revealed a five-factor structure of the PAS: social phobia, separation anxiety, obsessive-compulsive disorder, fears of physical injury, and generalized  53  anxiety.  Factor loadings for each item on the PAS ranged from r = .41 to .75.  Correlations between each of the five factors were moderate to high and ranged from r = .45 to .87 (Spence et al., 2001).  Construct and concurrent validity was established by comparing the PAS to the CBCL (Achenbach, 1992), a well-validated measure of competencies and problems in children and adolescents through self, teacher, and parent reports.  A copy of the PAS is available in Appendix E. Although the target age range of the population being sampled was 5 to 8 years, there are no parent-reported measures of anxiety for children that include this age range.  Thus, a decision was made to use the PAS, because the PAS captures parent-reported anxiety symptoms in children aged 3 to 6.5).  Other validated parent-reported measures of childhood anxiety are typically recommended for children ages 8 and older (i.e., MASC; March et al., 1997); anxiety symptoms in children younger than 8 may be different.  Another rationale for using the PAS is the five subscales of the PAS (GAD, Social Phobia, OCD, Fears of Injury, and SAD) are anxiety subtypes that are more commonly found in both preschool and elementary school-aged children (Kronenberger & Meyer, 2001).  Thus, the PAS and its subscales would be appropriate in identifying symptoms of anxiety in the sample of children for the current study (ages 5 to 8). Finally, the PAS (and the revised version, PAS-R; Edwards, Rapee, Kennedy, & Spence, 2010) has been used with populations that surpass the recommended age range in several research studies, including with children to age 13 (Benga, Tincas, & Visu-Petra, 2010; Broeren & Muris, 2009; Muris, Mayer, Kramer Freher, Duncan, & van den Hout, 2010).  54  2.3.7  Summary of Sample 2 (community-based sample). The purpose of the second phase of the study was to validate the revised paper-pencil version, the TTI-SM-R.  The TTI-SM-R was administered to a second development sample, a community-based sample.  Other teacher and parent-reported measures were administered as part of this study.  Findings for this study would provide support as to whether the TTI-SM-R demonstrates evidence of reliability and validity, and whether the measure may merit further research with a larger sample. 2.3.8 Differences between Samples 1 and 2. There were several differences between Samples 1 and 2.  First, Sample 1 was drawn from a clinical sample and was part of a larger multi-site investigation, whereas Sample 2 was drawn from a community-based sample.  The TTI-SM-DSM-IV data from Sample 1 was a secondary use of data.  The other difference was the measures that were administered to parents and teachers in both samples (see Table 2.1).  For Sample 1, teachers were administered the TTI- SM-DSM-IV telephone protocol, and for Sample 2, teachers were administered a revised paper- pencil version, the TTI-SM-R.  The TTI-SM-DSM-IV contains 54 items and 5 subscales, whereas the TTI-SM-R contains only 23 items and 2 subscales.  Both versions included the Mutism subscale and the Behaviour subscale.  Please refer to Table 2.1 for a list of how the 2 samples differed from each other in terms of the participants sampled and measures used.       55  Table 2.1: Differences between Samples 1 and 2 Description Sample 1 Sample 2 Participants Clinical sample Community-based (school) sample Sample Size: N = 29 (SM = 19, ANX = 10)  N = 30 (SM = 8, NC = 19, ANX = 3) Data Use Secondary use of data Primary use of data Teacher Measures  Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM- DSM-IV) o 54 items, 8 subscales  Revised Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting paper-pencil version (TTI-SM-R) o 28 items, 2 subscales  School Anxiety Scale – Teacher Report (SAS-TR; Lyneham et al., 2008) Parent Measures  Anxiety Disorders Interview Schedule for DSM-IV: Parent Version (ADIS-P)  Selective Mutism Questionnaire (SMQ)  Anxiety Disorders Interview Schedule for DSM-IV: Parent Version (ADIS-P)  Selective Mutism Questionnaire (SMQ)  Preschool Anxiety Scale (PAS)   56  2.4 Sample 3: Combining Sample 1 (Clinical) and Sample 2 (Community) The purpose of the third phase of this study was to combine the data from the first and second phases, thus having a larger sample for statistical analyses.  It was unknown whether there would be differences in the results for the combined samples than for Samples 1 and 2. After combining Samples 1 from the first phase (n=29) and Sample 2 from the second phase (n=30), there were a total of N = 59 Kindergarten to Grade 5 participants.  Data for all 59 children were available for the Mutism subscale and the SMQ measures.  The majority of the participants (70%) were in Grades 1-3 (ages 6-8).  The age group was slightly younger than what has been found in previous studies, which estimate the typical age of referral being between 6.5 to 9 years for SM (Ford et al., 1998; Kumpulainen et al., 1998; Standart & Le Couteur, 2003). Combining two independent samples is deemed acceptable provided there is a justification provided for combining the samples (National Research Council, 2004).  Data from the clinical and community samples were collapsed into one total sample based on the overlap in age and clinical diagnosis of the participants and the measures that were completed by parents and teachers.  Also, the two samples in the current study were combined to increase the power of the analyses.  By combining the data from two diverse samples, the issue of sampling bias would be examined.  For example, one sample was drawn from a clinical group (parents were referred to anxiety disorders clinics), but the second sample was drawn from a school-based (community) sample of children who were nominated for study by school personnel. Furthermore, the two samples were also drawn from two diverse and geographically distant areas of Canada.  Thus, combining the two samples increased the external validity of the results.  Individual analyses of each sample may perhaps result in an overrepresentation of children from a particular area of Canada, or a particular subgroup of children.  Another rationale  57  for combining Samples 1 and 2 comes from DeVellis (1999; 2003), who recommended that for test construction, measures should be administered to a large sample; however, DeVellis noted that there is no consensus on a minimum number.  SM is a low incidence disorder, thus resulting in low prevalence rates.  As a result, recruiting children with SM was challenging.  A decision was made to combine Samples 1 and 2 for a total sample of N = 59, with 27 of these students having a diagnosis of SM from the two (smaller) samples.  One reason that DeVellis gives for using a larger sample is that smaller sample sizes may impact the stability of the psychometric properties of the Mutism subscale.  For example, items that appear to be contributing to the internal consistency of a measure may be different for one sample compared to another when small development samples are used (DeVellis, 1991; 2003).  Also, Type I and II errors of correlational analyses may also have increased when small development samples are used (DeVellis, 1991; 2003). However, despite the advantages of combining the two samples, there were also concerns and disadvantages of this approach.  For example, the mode of administration for the TTI-SM- DSM-IV (telephone interview protocol) and revised TTI-SM-R (paper pencil version) differed for the two samples; Teachers completed the TTI-SM-DSM-IV via telephone interview protocol in Phase 1 (clinical sample), whereas for Phase 2, teachers from a community-based sample completed the revised TTI-SM-R as a paper-pencil measure.  This means that for the combined sample, it would be difficult to decipher whether the Mutism subscale of this teacher-reported measure of SM were influenced by how it was being measured (telephone vs. paper-pencil questionnaire) versus what was being measured.  A second concern of combining samples is that moderate effects that may be captured with one specific group (i.e., students with an anxiety disorder diagnosis) may not be captured when the samples are combined.  58  2.5 Nonparametric Statistical Methods (Mann-Whitney U Test) Nonparametric analyses were used for group comparison and correlations in the current study.  Parametric methods, on the other hand, were used for reliability analysis, and for exploratory purposes.  Nonparametric analyses were used for group comparisons in the study because specific parameters of the data must be met or assumed (i.e., normality of the data and sample size) otherwise these parametric methods lose power and rigor.  When these assumptions are not met or cannot be assumed, researchers and statisticians recommend that other methods such as nonparametric methods (or “distribution free” methods) be used (Siegel, 1956; Corder & Foreman, 2009). Although parametric analyses are more commonly used in the behavioural sciences (Siegel, 1956; Corder & Foreman, 2009), nonparametric analyses were used to analyze group comparisons for Samples 1 and 2 due to the data not meeting the assumptions of normality, homogeneity of variance, and the small sample size.  Normal probability plots are commonly used to evaluate normality of the variables of interest (Hair, Anderson, Tatham, & Black, 1998). Using histograms and normal probability plots for the PAS, SAS-TR, SMQ, and Mutism subscale (of the TTI-SM-DSM-IV and TTI-SM-R), revealed that the data for Samples 1 and 2 did not adequately meet the assumption of normality.  The data for each of the dependent variables were not normally distributed.  As a result, the slope of the data in the normal probability plot deviated from a linear line.  A second rationale for the use of nonparametric analyses was due to the data violating the assumption of homogeneity of variance.  The Levene’s tests of equality of error variances for each of the dependent variables tested the null hypothesis that the error variance of the dependent variables was equal across the two groups (SM vs. NC).  The Levene’s tests for three of the four dependent variables (Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-  59  R, SAS-TR, and PAS) were not statistically significant, although some of the p-values were marginally significant (e.g., p = .09 for the TTI-SM-DSM-IV, and p = .06 for one of the PAS subscales).  The Levene’s test was statistically significant for one measure, the SMQ.  This indicates that the groups had equal variances in some instances, but unequal variances in others. Thus, nonparametric analyses were appropriate to use in such a circumstance. A third rationale for the use of nonparametric analyses was due to Samples 1 and 2 being small.  The first phase of the research had a total of 29 participants, and the second phase of the research had a total of 30 participants, which was considered to be small for a parametric analysis of substantial power (> .80).  Analysis of small sample sizes result in the population distribution being unknown, but can be addressed using nonparametric methods. Nonparametric methods were also appropriate to use for group comparisons in the current study given that some nonparametric tests can be used with samples that are taken from different populations (Siegel, 1956).  Participants in the first phase were all drawn from a clinical sample, and participants in the second phase were drawn from a school-based sample, thus, nonparametric methods would be appropriate to use.  The two assumptions that need to be met for nonparametric methods are that the observations must be independent of each other, and that the variable that was being measured has some continuity; both assumptions were met with the current data. There are several disadvantages that need to be considered when contemplating the use of nonparametric analyses.  First, if all of the assumptions were met for parametric methods, parametric tests would produce more powerful results.  In this study, however, the assumptions were not met, and thus power from the parametric analysis may be considered reduced.  Another disadvantage to nonparametric methods is that these methods are generally not as widely known  60  by students and researchers in the behavioural sciences (Black, 2010; Siegel, 1956).  Hence, there may be some barriers in researchers understanding and interpreting the results of nonparametric methods or sharing the results of the study in journals.  61  3. Results This section will begin with an overview of the research questions of the study.  Next, a description of each of the two independent samples (Samples 1 [clinical] and 2 [community- based]), and a description of the combined sample will be provided.  Finally, the statistical analyses and results for each of the 6 research sub-questions will be presented, which will be organized by presenting the research question first, and then second, presenting the specific results that were obtained using Samples 1 and 2, and the combined sample for analyses. 3.1 Research Questions The core research goal of this study was to evaluate the reliability and validity of the Mutism subscale (in the TTI-SM-DSM-IV telephone protocol, and the revised TTI-SM-R paper- pencil version).  Specifically, the overarching research question was: To what extent does a teacher reported measure for selective mutism (SM) identify children with SM accurately and consistently?  To answer this broad research question, 6 research sub-questions were investigated: 1) What is the reliability of the Mutism subscale? 2) What is the face validity evidence for the Mutism subscale? 3) What is the predictive validity evidence (i.e., diagnosis of SM) for the Mutism subscale? 4) What is the convergent validity evidence for the Mutism subscale in relation to parent-reported measures of SM? 5) What is the convergent and discriminant validity evidence for the Mutism subscale in relation to other measures of nonspecific anxiety? and, 6) What is the concurrent validity evidence for the Mutism subscale in relation to other classroom behaviours of students with SM?  62  3.2 Description of Samples 3.2.1 Description of Sample 1 (clinical sample). For Sample 1 (clinical sample), the final sample size was N = 29.  There were 12 males and 17 females in the sample, ranging in age from 6 to 11 (Grades K to 5).  Diagnosis from the ADIS-P interview confirmed 19 children who had a diagnosis of SM, and 10 children with a diagnosis of an anxiety disorder (ANX).  Approximately 68% of participants had a co-morbid anxiety disorder.  Diagnostic groups did not differ significantly in age, but there were a larger number of females than males in the SM group.  The higher preponderance in females compared to males in this sample was similar to what has been reported in the literature (Cunningham et al., 2006).  Table 3.1 presents the descriptive results for the sample.              63  Table 3.1: Descriptives for Samples 1, 2, and Combined Samples Sample 1 (n = 29) Variable SM Group ANX Group* TOTAL N % N % N % Gender (total) 19 100.00% 10 100.00% 29 100.00% Male 6 31.58% 6 60.00% 12 41.38% Female 13 68.42% 4 40.00% 17 58.62% Grade K 1 5.26% 0 0.00% 1 3.40% 1 5 26.32% 0 0.00% 5 17.20% 2 4 21.05% 3 30.00% 7 24.10% 3 4 21.05% 5 50.00% 9 31.00% 4 3 15.79% 0 0.00% 3 10.30% 5 2 10.53% 2 20.00% 4 13.80% Age 6 7 36.84% 0 0.00% 7 24.10% 7 3 15.79% 3 30.00% 6 20.70% 8 3 15.79% 5 50.00% 8 27.60% 9 5 26.32% 1 10.00% 6 20.70% 10 0 0.00% 1 10.00% 1 3.40% 11 1 5.26% 0 0.00% 1 3.40%  64   Sample 2 (n = 30) Variable SM Group NC Group** TOTAL N % N % N % Gender (total) 8 100.00% 22 100.00% 30 100.00% Male 1 12.5% 8 36.36% 9 30.00% Female 7 87.5% 14 63.64% 21 70.00% Grade (total) K 2 25.00% 6 27.27% 8 26.67% 1 3 37.50% 13 59.09% 16 53.33% 2 2 25.00% 1 4.55% 3 10.00% 3 1 12.50% 1 4.55% 2 6.67% 4 0 0.00% 1 4.55% 1 3.33% Age 5 3 37.50% 4 18.18% 7 23.33% 6 1 12.50% 13 59.09% 14 46.67% 7 3 37.50% 3 13.64% 6 20.00% 8 1 12.50% 1 4.55% 2 6.67% 9 0 0.00% 1 4.55% 1 3.33% Combined Sample (N = 59) Variable SM Group NC Group TOTAL N % N % N % Gender 27 100.00% 32 100.00% 59 100.00% Male 7 25.93% 14 43.75% 21 35.59% Female 20 74.07% 18 56.25% 38 64.41%  65   Combined Sample (N = 59) Variable SM Group NC Group TOTAL N % N % N % Grade K 3 11.11% 6 18.75% 9 15.25% 1 8 29.63% 13 40.63% 21 35.59% 2 6 22.22% 4 12.50% 10 16.95% 3 5 18.52% 6 18.75% 11 18.64% 4 3 11.11% 1 3.13% 4 6.78% 5 2 7.41% 2 6.25% 4 6.78%  Age 5 3 11.11% 4 12.50% 7 11.86% 6 8 29.63% 13 40.63% 21 35.59% 7 6 22.22% 6 18.75% 12 20.34% 8 4 14.81% 6 18.75% 10 16.95% 9 5 18.52% 2 6.25% 7 11.86% 10 0 0.00% 1 3.13% 1 1.69% 11 1 3.70% 0 0.00% 1 1.69% SM Group = Selective Mutism Group ANX Group = Anxiety Disorder Group NC Group = Normal Control Group    66  Parents of each participant in Sample 1 completed the Selective Mutism Questionnaire (SMQ; Bergman, 1999; 2008).  Participating teachers completed the Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV), in a semi- structured telephone interview. 3.2.2  Description of Sample 2 (community-based sample). For Sample 2 (community-based), the final sample size was 30 students, 8 males and 22 females, ranging in age from 5 to 9 (Grades K to 4).  The sample descriptive statistics are presented in Table 3.1.  There were 8 students who had a diagnosis of SM, 19 students with no SM diagnosis (i.e., normal controls), and 3 students with a diagnosed anxiety disorder.  The students with no SM diagnosis and an anxiety disorder formed the Normal Control (NC) group, thus resulting in 22 students in the NC group.  Similar to the first study, there were a larger number of females than males in the SM group, which corresponds to past research concerning the preponderance of SM in females compared to males (Cunningham et al., 2006). The mode of administration for the TTI-SM-R for Sample 2 was a paper-pencil measure, which was mailed to participating teachers along with the School Anxiety Scale-Teacher Report (SAS-TR; Lyneham et al., 2008).  The TTI-SM-R paper-pencil version that was administered to Sample 2 differed from the original telephone interview protocol, the TTI-SM-DSM-IV that was administered to Sample 1.  Changes to the paper-pencil version (TTI-SM-R) were based on feedback from professionals in the field of child mental health, as discussed in sections 2.4.4.1.1 and 3.2.2.  Most notably, the paper-pencil version had an additional item added to the measure. It is also of note that the items of the TTI-SM-R version were rearranged on the paper-pencil version.  67  3.2.3  Description of the combined sample. Samples 1 (n = 29) and 2 (n = 30) were combined to a sample size combined (N = 59). There were 21 males and 38 females in the total sample, ranging in grade from Kindergarten to Grade 5 (ages 5 to 11).  The majority of the participants in the combined sample were in Grades 1 (35%), 2 (20%), and 3 (17%).  There were 27 students who had a diagnosis of SM (study group), and 32 students in the control group (referred to as the Normal Controls [NC]) who had either a diagnosis of an anxiety disorder or no SM diagnosis.  The second group of ANX and NC students formed one contrast group to the SM group.  The reason the students with anxiety disorders were aggregated with students without a diagnosis was because the goal of the study was to compare students with a diagnosis of SM to all other students without SM.  Table 3.1 presents the descriptive information for the total combined sample. 3.3  Statistical Analyses and Results for Research Questions 3.3.1 Statistical analysis and results for research question 1 . The first research question was: What is the reliability of the Mutism subscale?  To answer this research question, a correlational analysis was computed using the Statistical Package for the Social Sciences, 18.0 Version (SPSS-18.0).  Using data from Samples 1, 2, and the combined sample, an intercorrelation matrix provided item-total and inter-item correlations for the Mutism subscale of the TTI-SM-DSM-IV and in the TTI-SM-R.  Second, reliability analyses were computed to evaluate the internal consistency and split-half reliability of the Mutism subscale of the 2 instruments.  Finally, Principal Components Analysis (PCA) was computed as an exploratory method for investigating possible data reduction and dimensional scaling purposes.  Reliability analysis and PCA provided evidence for the construct validity of the Mutism subscale of the two instruments.  68  3.3.1.1 Findings based on Sample 1 (clinical sample). Intercorrelation matrices (using Spearman’s rho) for the Mutism subscale of the TTI-SM- DSM-IV revealed inter-item correlations that ranged from low (rs = .004, p > .05) to high (rs = .95, p < .001).  Item-total correlations for the Mutism subscale ranged from rs = .04 (p > .05) to .88 (p ≤ .001).  Item analysis revealed that items 4, 4a, 8, and 8a had low inter-item and item- total correlations.  It was hypothesized that items 4 and 4a did not significantly correlate with the other items of the scale and the total score because these items queried about students’ non- verbal participation (e.g., pointing, shrugging), which has not been found to be a significant difficulty for students with SM.  Items 4 and 4a were found to be correlated with each other (r = .59, p ≤ .001). It was hypothesized that items 8 and 8a did not correlate with the other items of the scale and the total score because these items concern behavioural impairment (negative behaviours) rather than behavioural functioning (positive behaviours).  Items 8 and 8a were also positively correlated (r = .73, p ≤ .001). Principal component analysis (PCA) was used as an exploratory method for investigating the number of factors for the TTI-SM.  For the Mutism subscale, the eigenvalues revealed three factors.  The first factor explained 56.3% of the variance, the second factor explained 14% of the variance, and the third factor explained 9.3% of the variance.  Subsequent factors had total eigenvalues less than 1.0.  For the first factor, 4 items had low factor loadings: item 4 had a factor loading of .09, item 4a had a factor loading of .22, item 8 had a factor loading of -.19, and item 8a had a factor loading of -.001.  The remaining 11 items had factor loadings of .62 to .93 to the first component.  69  Due to low inter-item and item-total correlations of items 4, 4a, 8, and 8a and PCA, these 4 items (items 4, 4a, 8 and 8a) were not included in any additional statistical analysis for the Mutism subscale for Sample 1. For the 11 remaining items on the Mutism subscale within the TTI-SM-DSM-IV, correlational analyses were computed.  Item-total correlations of the Mutism subscale ranged from rs = .68 (p ≤ .001) to .88 (p ≤ .001), and inter-item correlations ranged from rs = .33 (p > .05) to .95 (p ≤ .001).  Table 3.2 presents an intercorrelation matrix for the 11 items of the Mutism subscale.  A PCA of the remaining 11 items revealed only one factor, with factor loadings of .63 to .94.   70   Table 3.2: Intercorrelation Matrix for the Mutism subscale  for Sample 1, 2 and Combined Sample Sample Item Number 1 2 3 5 6 7 1a 2a 5a 6a 7a Mean Score 1 1 1.00 .65**  .33 .59**  .83**  .64**  .82**  .74**  .64**  .79**  .69**  .81**  2  1.00 .71**  .73**  .66**  .79**  .81**  .74**  .68**  .75**  .82**  .85**  3   1.00 .51**  .35 .55**  .57**  .46*  .45*  .41*  .76**  .68**  5    1.00 .73**  .79**  .78**  .81**  .95**  .65**  .83**  .81**  6     1.00 .66**  .72**  .80**  .75**  .82**  .61**  .85**  7      1.00 .74**  .75**  .76**  .69**  .81**  .76**  1a       1.00 .78**  .83**  .73**  .88**  .88**  2a        1.00 .75**  .79**  .76**  .88**  5a         1.00 .65**  .83**  .82**  6a          1.00 .60**  .87**  7a           1.00 .85**  M            1.00  71   Sample Item Number 1 2 3 5 6 7 1a 2a 5a 6a 7a Mean Score 2 1 1.00 .90**  .68**  .86**  .78**  .82**  .88**  .86**  .80**  .80**  .80**  .91**  2  1.00 .80**  .88**  .79**  .86**  .82**  .85**  .77**  .83**  .80**  .90**  3   1.00 .87**  .73**  .87**  .75**  .88**  .81**  .86**  .85**  .86**  5    1.00 .78**  .88**  .85**  .90**  .84**  .86**  .83**  .93**  6     1.00 .91**  .89**  .90**  .83**  .89**  .86**  .90**  7      1.00 .93**  .95**  .89**  .94**  .96**  .96**  1a       1.00 .92**  .84**  .87**  .88**  .94**  2a        1.00 .84**  .95**  .94**  .95**  5a         1.00 .79**  .91**  .93**  6a          1.00 .90**  .92**  7a           1.00 .95**  M            1.00  72   Sample Item Number 1 2 3 5 6 7 1a 2a 5a 6a 7a Mean Score Combined 1 1.00 .79**  .57**  .76**  .86**  .79**  .90**  .84**  .75**  .81**  .80**  .88**  2  1.00 .78**  .83**  .76**  .86**  .84**  .82**  .78**  .79**  .84**  .90**  3   1.00 .74**  .57**  .74**  .69**  .68**  .69**  .62**  .82**  .80**  5    1.00 .79**  .88**  .85**  .88**  .94**  .77**  .88**  .90**  6     1.00 .80**  .82**  .85**  .80**  .84**  .76**  .88**  7      1.00 .87**  .88**  .86**  .83**  .91**  .90**  1a       1.00 .88**  .86**  .81**  .90**  .92**  2a        1.00 .84**  .86**  .89**  .93**  5a         1.00 .74**  .89**  .89**  6a          1.00 .77**  .90**  7a           1.00 .93**  M            1.00 *p ≤ .05. **p ≤ .001. Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R   73  Next, internal consistency and item analyses were computed for the 11-item Mutism subscale, which revealed a Cronbach’s alpha of  = .97 (p ≤ .001).  Mean scores for each of the items ranged from M = 1.11 to 2.17 (on a 4-point Likert-type Scale), with a standard deviation (SD) of SD = 0.75 to 1.12. To further examine the reliability of the Mutism subscale, split-half reliability was computed to estimate the correlation of two random halves of the Mutism subscale.  The mean score for the total scale was M = 8.00 (SD = 9.14).  The first half of the Mutism subscale (Part 1) included 6 items, and the mean was M = 4.44 (SD = 4.94), and Cronbach’s alpha was  = .93 (p ≤ .001).  The second half of the Mutism subscale (Part 2) included 5 items, and the mean score was M = 3.56 (SD = 4.30), and Cronbach’s alpha was  = .93 (p ≤ .001).  The Spearman-Brown formula was .98 (p ≤ .001; for unequal lengths).  Table 3.3 presents the reliability analyses for Sample 1.   74   Table 3.3: Reliability Analyses for Samples 1, 2, and Combined Samples: Internal Consistency and Split Half for the Mutism subscale Sample Variable N Mean (SD) Cronbach's  F p 1 Cronbach's Alpha  Entire Form 11* 8.00 (9.14) .97 5.06 ≤ .001  Part 1 6 4.44 (4.94) .93  Part 2 5 3.56 (4.30) .93  Between Forms 11   .96  Spearman-Brown  Unequal Length 11   .98  Guttman Split Half 11   .97  2 Cronbach's Alpha  Entire Form 11* 18.72 (11.73) .98 5.58 ≤ .001  Part 1 6 10.22 (6.61) .97  Part 2 5 8.50 (5.29) .97  Between Forms 11   .94  Spearman-Brown  Unequal Length 11   .97  Guttman Split Half 11   .96  75   Sample Variable N Mean (SD) Cronbach's  F p Combined Cronbach's Alpha  Entire Form 11* 12.29 (11.44) .98 9.80 ≤ .001  Part 1 6 6.76 (6.29) .96  Part 2 5 5.53 (5.27) .96  Between Forms 11   .96  Spearman-Brown  Unequal Length 11   .98    Guttman Split Half 11     .97 *p ≤ .05. **p ≤ .001. Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R   76   3.3.1.2 Findings based on Sample 2 (community-based sample). The TTI-SM-R measure administered for Sample 2 differed from the TTI-SM-DSM-IV for Sample 1 in three modest ways: the measure was administered as a paper-pencil test rather than a telephone interview protocol, the items on the Mutism subscale were reorganized on the paper- pencil version, and an additional item (item 16) was added to the Mutism subscale on the paper- pencil version (which was not on the original TTI-SM-DSM-IV interview protocol). Similar to the analyses for Sample 1, Spearman’s rho correlations were established for all of the items within the Mutism subscale for Sample 2 first, which included 16 items.  The item- total correlations for the 16 item Mutism subscale revealed low (rs = -.03, p = .922) to high (rs = .95, p ≤ .001) correlations.  Inter-item correlations also ranged from being low (rs = -.12, p = .631) to high (rs = .96, p ≤ .001).  Items 4, 4a, 8, and 8a (which are items 5, 8, 14, and 16 on the TTI-SM-R (paper-pencil version) were the items that had very low inter-item and item-total correlations.  There was a moderate correlation detected for item 16 to the TTI-SM-R total mean score. Again, it was hypothesized that low inter-item and item-total correlations for items 4 and 4a was because these two items queried on students’ non-verbal participation.  Non-verbal participation has not been found to be a significant difficulty for students with SM.  It was hypothesized that items 8 and 8a had low inter-item and item-total correlations because these two items concern behavioural impairments (negative behaviours) rather than behavioural functioning (positive behaviours).  For item 16, it was hypothesized that this item had moderate inter-item and item-total correlations because the question queried on the audibility of a child’s voice.  Specifically, item 16 appears to be a measure of the magnitude of the child’s verbal  77  projection (audibility of his/her voice) as opposed to the other 15 questions which measure the frequency of a behaviour. Principal Components Analysis (PCA) was used as an exploratory method for investigating the number of factors for all of the 16 items on the Mutism subscale.  Using all 16 items of the Mutism subscale, the eigenvalues revealed two factors.  The first factor explained 68.86% of the variance, and the second factor explained 12.13%.  Subsequent factors had total eigenvalues less than 1.0.  For the first component, Item 4 had a factor loading of -.38, item 4a had a factor loading of -.34, item 8 had a factor loading of -.82, and item 8a had a factor loading of -.70.  Item 16 had a factor loading of .56 on the first component.  The remaining 11 items of the Mutism subscale had factor loadings of .88 to .97 to the first component. A decision was made to remove items 4, 4a, 8, 8a, and 16 from all remaining analyses of the Mutism subscale of the TTI-SM-R with Sample 2 (and for analyses using the combined sample).  The reason for removing items 4, 4a, and 16 was due to the low inter-item and item- total correlations for Sample 2, and low inter-item and item-total correlations for Sample 1. Although items 8 and 8a did not have low inter-item and item-total correlations (they had negative correlations to the rest of the scale), these two items were removed from the Mutism subscale because these items measured behavioural impairments in students, whereas the other 11 items measured desirable behaviours in students.  Also, items 8 and 8a had low inter-item and item-total correlations for Sample 1 (clinical sample), which may suggest that these two items were measuring a different (but related) construct, depending on the type of sample that was surveyed. Using only the 11 remaining items of the Mutism subscale of the TTI-SM-R, item-total correlations ranged from rs = .86 to .96 (p ≤ .001), and inter-item correlations ranged from rs =  78  .68 to .96 (p ≤ .001).  Eigenvalues for the PCA revealed one factor.  The first factor explained 84.97% of the variance.  Subsequent factors had total eigenvalues less than 1.0.  Similar to the results for Sample 1, the results for Sample 2 suggested that the Mutism subscale of the TTI-SM- R was measuring one unitary concept. Reliability analyses revealed that the 11 questions of the Mutism subscale of the TTI-SM- R demonstrated high internal consistency.  Cronbach’s alpha, based on the 11 standardized items, was  = .98 (p ≤ .001).  Mean scores for each of the 11 items ranged from M = 1.28 to 2.17 (on a 4-point Likert-type Scale), with a standard deviation (SD) of SD = 0.99 to 1.25. Split-half reliability was established by estimating the correlation of the Mutism subscale of the TTI-SM-R by splitting the 11 items into two random halves (6 items and 5 items).  Split- half reliability for part 1 of the form was  = .97 (p ≤ .001) and for part 2 of the form was  = .97 (p ≤ .001).  The correlation between the two forms was  = .94 (p ≤ .001), and Spearman- Brown for unequal lengths of the two halves was  = .97 (p ≤ .001).  Please refer to Table 3.3 for results of the reliability analyses. 3.3.1.3 Findings based on combined sample. For the combined sample, only the 11 items of the Mutism subscale of the TTI-SM-DSM- IV and the TTI-SM-R were used for analyses (without items 4, 4a, 8, 8a, and 16).  Spearman’s rho correlations revealed high item-total correlations, which ranged from rs = .80 to .93 (p ≤ .001).  Inter-item correlations ranged from rs = .57 to .94 (p ≤ .001).  PCA for the 11 items revealed correlations ranging from rs = .79 to .96 (p ≤ .001).  Eigenvalues revealed one factor, which explained 83.23% of the variance.  Similar to the results in Samples 1 and 2, the 11 items of the Mutism subscale of the two instruments appeared to be measuring one unitary concept.  79  The 11 items of the Mutism subscale demonstrated high internal consistency, with a high Cronbach’s alpha  = .98 (p ≤ .001).  Mean scores for each of the 11 items ranged from M = 0.84 to 1.53 (on a 4-point Likert-type Scale), with a standard deviation (SD) of SD = 1.02 to 1.23. Split-half reliability for the first half of the Mutism subscale (Part 1) included 6 items, and the mean was M = 6.76 (SD = 6.29), and Cronbach’s alpha was  = .96 (p ≤ .001).  The second half of the Mutism subscale (Part 2) included 5 items, and the mean score was M = 5.53 (SD = 5.27), and Cronbach’s alpha was  = .96 (p ≤ .001).  The Spearman-Brown formula was .98 (p ≤ .001; for equal lengths).  Please refer to Table 3.3 for results of the reliability analyses for the combined sample. The reliability analyses for Sample 1, Sample 2, and the combined sample showed that the Mutism subscale demonstrated strong internal consistency.  Split-half reliability for two random parts of the Mutism subscale was found to be high.  The analyses also revealed high item-total and inter-item correlations.  Taken together, the 11 items of the Mutism subscale appeared to be measuring one unitary concept. 3.3.2 Statistical analysis and results for research question 2 . The second research question was: What is the face validity evidence for the Mutism subscale?  To answer this research question, the TTI-SM-DSM-IV telephone protocol was first modified into a paper-pencil test (TTI-SM-R), using 2 of the 5 subscales (Mutism subscale and Behaviour subscale) and maintaining the same frequency-type Likert scale response options for the 2 subscales.  Next, using the guidelines for scale development, as suggested by DeVellis (1991; 2008), the researcher consulted with approximately 50 school mental health professionals (e.g., school and/or clinical psychologists, school counselors) in BC for review and feedback following a full-day workshop on anxiety disorders and selective mutism.  80  3.3.2.1 Findings based on professional field-based review. Each workshop participant either provided feedback verbally to the researcher, or by notes and comments written on a sample copy of the TTI-SM-R.  All feedback from the group of participants was reviewed with three professors of Applied Psychology at UBC.  The one significant change to the TTI-SM-R paper-pencil measure was the recommendation to add an additional item (#16) to probe about the audibility of the child’s speech in the class.  This question was: “His/her speech is sufficiently loud to be easily heard.”  Other minor changes were noted, such as spacing and formatting of the items on the page.  The group of mental health professionals unanimously agreed that TTI-SM-R, in particular, the Mutism subscale appeared to measure what it was purported to measure.  The TTI-SM-R thus demonstrated initial evidence of face and content validity. 3.3.3 Statistical analysis and results for research question 3 . The third research question was: What is the predictive validity evidence (i.e., diagnosis of SM) for the Mutism subscale?  To answer this question, correlational analyses were conducted to understand the relationship between the Mutism subscale of the TTI-SM-DSM-IV and the TTI- SM-R, with SM diagnosis.  Additionally, Mann-Whitney U Tests (generally considered the nonparametric equivalent to the t-test; Tabachnick & Fidell, 2001) were computed to investigate whether there were significant differences in Mutism subscale scores for SM versus non-SM students.  Results from this analysis provided evidence of predictive validity. 3.3.3.1 Findings based on Sample 1 (clinical sample). For Sample 1 (clinical sample), the correlation for the Mutism subscale score and SM diagnosis was rs = -.64 (p ≤ .001).  A Mann-Whitney U Test revealed differences in Mutism subscale ratings between students with SM (Mdn = .19) compared to students with ANX (Mdn =  81  1.41; U = 22.00, p ≤ .001, r = .63).  Specifically, students with SM scored significantly lower on the Mutism subscale compared to students with ANX.  Lower scores on the Mutism subscale indicated more difficulty with speaking behaviours. 3.3.3.2 Findings based on Sample 2 (community-based sample). Similar to the results for Sample 1, the results for Sample 2 (community-based sample) revealed a significant correlation between the Mutism subscale score (in the TTI-SM-R) and SM diagnosis (rs = -.82; p ≤ .001).  The Mann-Whitney U Test again revealed differences in Mutism subscale ratings between students with SM (Mdn = 3.5) compared to students without SM (Mdn = 12.5; U = 0, p ≤ .001, r = .80) whereby students with SM scored lower on the Mutism subscale of the TTI-SM-R compared to students without SM. 3.3.3.3 Findings based on the combined sample. For the combined sample, there was a strong correlation between the Mutism subscale score (in the TTI-SM-DSM-IV and the TTI-SM-R) and SM diagnosis (rs = -.74; p ≤ .001). Students with low scores on the Mutism subscale (which is indicative of problems speaking when required to do so), were found to have a positive diagnosis of SM, and thus provided evidence of predictive validity.  The Mann-Whitney U test revealed statistically significant differences in Mutism subscale scores between students with SM (Mdn =.73) compared to students in the NC group (Mdn = 1.63; U = 39.00, p ≤ .001, r = .74). 3.3.4 Statistical analysis and results for research question 4 . Research question 4 was: What is the convergent validity evidence for the Mutism subscale in relation to parent-reported measures of SM?  A correlational analysis (using an intercorrelation matrix) was conducted in order to provide evidence that the Mutism subscale (teacher report) and SMQ (parent report) were measuring the construct of “selective mutism”.  82  The SMQ is a validated parent report measure for SM.  Moderate to high correlations between the Mutism subscale and SMQ provided evidence for convergent validity.  Additionally, the results also provided evidence about the concordance rates between parent and teacher reports of SM. 3.3.4.1 Findings based on Sample 1 (clinical sample). For Sample 1 (clinical sample), the median Mutism subscale score (on the TTI-SM-DSM- IV) for the SM group was Mdn = .18, and the mean score was M = 0.41 (SD = .69).  For the ANX group, the median Mutism subscale score was Mdn = 1.41, and the mean score was M = 1.31 (SD = .67).  Refer to Table 3.4 for a distribution of the scores.   83   Table 3.4: Mean and Median Scores for the Mutism subscale and SMQ in Samples 1, 2, and Combined Sample 1 Variable SM Group  ANX Group  TOTAL Mean (SD) Mdn Min Max N Mean (SD) Mdn Min Max N Mean (SD) Mdn Min Max N TTI-SM-DSM- IV (Mutism subscale) 0.41 .18 .00 3.00 19  1.31 1.41 .36 2.45 10  .72 .45 .00 3.00 29  (.69)      (.67)      (.80) SMQ (m) 1.94 1.79 1.46 3.12 19  3.12 3.29 2.30 3.89 10  2.35 2.09 1.46 3.89 29  (.42)      (.56)      (.74) School (m) 1.34 1.17 1.00 2.67 19  2.88 3.00 1.67 4.00 10  1.87 1.50 1.00 4.00 29  (.48)      (.77)      (.95) Family (m) 2.82 2.60 2.00 3.80 19  3.70 3.90 3.20 4.00 10  3.12 3.40 2.00 4.00 29  (.64)      (.36)      (.69) Other (m) 1.66 1.56 1.00 3.22 19  2.78 3.11 1.67 3.67 10  2.04 1.67 1.00 3.67 29  (.51)      (.69)      (.78) SMQ (f) 1.79 1.73 1.43 2.39 8  2.93 2.98 2.33 3.47 3  2.10 1.86 1.43 3.47 11  (.29)      (.58)      (.64) School (f) 1.29 1.08 1.00 2.67 8  2.67 2.33 2.00 3.67 3  1.67 1.17 1.00 3.67 11  (.57)      (.88)      (.89) Family (f) 2.60 2.50 2.00 3.40 8  3.33 3.20 3.20 3.60 3  2.80 2.80 2.00 3.60 11  (.45)      (.23)      (.52) Other (f) 1.48 1.44 1.11 1.89 8  2.78 3.00 1.78 3.56 3  1.83 1.56 1.11 3.56 11  (.23)      (.91)      (.76)  84   Sample 2 Variable SM Group  ANX Group  TOTAL Mean (SD) Mdn Min Max N Mean (SD) Mdn Min Max N Mean (SD) Mdn Min Max N TTI-SM-R (Mutism subscale) 0.44 .32 .00 1.00 6  2.33 2.50 1.36 2.00 12  1.70 1.73 .00 3.00 18  (.39)      (.62)      (1.07)  SMQ  1.14 1.14 .22 1.96 8  2.41 2.66 .98 3.00 22  2.07 2.22 .22 3.00 30  (.62)      (.57)      (.81) School 0.85 .67 .00 2.17 8  2.36 2.58 .33 3.00 22  1.96 2.17 .00 3.00 30  (.89)      (.72)      (1.01) Family 1.98 2.17 .50 2.83 8  2.76 2.92 2.00 3.00 22  2.55 2.83 .50 3.00 30  (.78)      (.36)      (.60) Other 0.60 .50 .00 1.20 8  2.10 2.30 .40 3.00 22  1.70 1.90 .00 3.00 30  (.50)      (.75)      (.96)  85   Combined Sample Variable SM Group  NC Group  TOTAL Mean (SD) Mdn Min Max N Mean (SD) Mdn Min Max N Mean (SD) Mdn Min Max N TTI-SM-R (Mutism subscale) 0.42 .18 .00 3.00 25  1.87 1.73 .36 3.00 22  1.10 .82 .00 3.00 47  (.63)      (.82)      (1.02)  SMQ  1.70 1.74 .22 3.12 27   2.63 2.76 .98 3.89 32   2.21 2.18 .22 3.89 59   (.60)           (.65)           (.78) School 1.20 1.00 .00 2.67 27   2.53 2.75 .33 4.00 32   1.92 2.00 .00 4.00 59   (.65)           (.76)           (.97) Family 2.57 2.50 .50 3.80 27   3.05 3.00 2.00 4.00 32   2.83 2.83 .50 4.00 59   (.77)           (.57)           (.70) Other 1.34 1.44 .00 3.22 27   2.31 2.40 .40 3.67 32   1.87 1.80 .00 3.67 59   (.70)           (.79)           (.89) Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R SMQ: Selective Mutism Questionnaire    86  For the SMQ, the median total score for the SM group was Mdn = 1.79, and the mean score was M = 1.94 (SD = .42).  For the ANX group, the median SMQ total score was Mdn = 3.29, and the mean score was M = 3.12 (SD = .56).  The correlation between the Mutism subscale mean score and the SMQ mother (school) was rs = .84 (p ≤ .001), and with the SMQ father (school) was rs = .86 (p ≤ .001).  Please see Table 3.5 for correlations between the Mutism subscale and SMQ.  These results provided evidence for convergent validity of the TTI-SM.  Table 3.5: Spearman’s Rho Correlations for Mutism subscale Scores and SMQ for Samples 1, 2, and Combined Sample Sample Variable (Rater)         Subscale Mutism Subscale  Mean Score p N 1 SMQ Total (Mother) .71** ≤ .001   School .84**  ≤.001 29   Family .41* .027 29   Other (Public) .79** ≤ .001 29    SMQ Total (Father) .65* .029   School .86** .002 11   Family .30 .378 11   Other (Public) .65* .029 11  87   Sample Variable (Rater)         Subscale Mutism Subscale  Mean Score p N 2 SMQ Total (Parent) .81**  ≤ .001 18   School .73* .001 18   Family .58* .011 18   Other (Public) .87** ≤ .001 18  Combined SMQ Total (Parent) .81** ≤ .001 47   School .81** ≤ .001 47   Family .14 .361 47   Other (Public) .64** ≤ .001 47 *p ≤ .05. **p ≤ .001. Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R SMQ: Selective Mutism Questionnaire   3.3.4.2 Findings based on Sample 2 (community-based sample). For Sample 2 (community-based sample), the median Mutism subscale score on the TTI- SM-R for the SM group was Mdn = .32, and the mean score was M = .44 (SD = .39).  For the NC group, the median Mutism subscale score was Mdn = 2.50, and the mean score was M = 2.33 (SD = .62).  The median SMQ total score for the SM group was Mdn = 1.14, and the mean  88  score was M = 1.14 (SD = .62).  For the ANX group, the median SMQ total score was Mdn = 2.66, and the mean score was M = 2.41 (SD = .57). The correlation between the Mutism subscale score and the SMQ total score was rs = .81 (p ≤ .001).  The correlation between the Mutism subscale score to the SMQ School factor was rs = .73 (p ≤ .001), to the SMQ Family factor was rs = .58 (p = .011), and to the SMQ Other (Public) factor was rs = .87 (p ≤ .001).  See Table 3.5.  The high correlation between the Mutism subscale and the SMQ School and Public factors provided evidence of convergent validity. 3.3.4.3 Findings based on the combined sample. Finally, for the combined sample, the median Mutism subscale score (on the TTI-SM- DSM-IV and the TTI-SM-R) for the SM group was Mdn = .18, and the mean score was M = .42 (SD = .63).  For the NC group, the median Mutism subscale score was Mdn = 1.73, and the mean score was M = 1.87 (SD = .82).  The median SMQ total score for the SM group was Mdn = 1.74, and the mean score was M = 1.70 (SD = .60).  For the NC group, the median SMQ total score was Mdn = 2.76, and the mean score was M = 2.63 (SD = .65). The correlation between the Mutism subscale mean score to the SMQ total score was rs = .81 (p ≤ .001).  The correlation between the Mutism subscale score to the SMQ School factor was rs = .81 (p ≤ .001), and to the SMQ Other (Public) factor was rs = .64 (p ≤ .001).  The correlation between the Mutism subscale mean score was not significant to the SMQ Family factor (rs = .14, p = .361; Table 3.5).  The high correlation between the Mutism subscale mean score and the SMQ mean score provided evidence for convergent validity of the Mutism subscale. 3.3.5 Statistical analysis and results for research question 5 . The fifth research question was: What is the convergent and discriminant validity evidence for the Mutism subscale in relation to other measures of nonspecific anxiety?  To  89  answer this question, correlational analyses were only computed with Sample 2 because sample 2 was the only sample that was administered the PAS.  The analysis investigated the relationship between scores on the Mutism subscale (of the TTI-SM-R) to scores on the teacher reported measures of anxiety in students (SAS-TR), and parent reported measures of anxiety (PAS).  The results provided evidence for convergent validity. Additionally, a Mann-Whitney U Test was computed to investigate whether groups of students with SM scored differently on other measures of anxiety compared to students without SM.  These results provided some evidence of convergent validity of the Mutism subscale because school and social anxiety are constructs that are related to SM. 3.3.5.1 Findings based on Sample 2 (community-based sample). There was a strong correlation between the Mutism subscale score (on the TTI-SM-R) and the SAS-TR total mean score (rs = -.84, p ≤ .001).  The correlation between the Mutism subscale and the PAS mean score was rs = -.62 (p = .006).  Correlational analyses were also conducted for the Mutism subscale and the individual PAS subscales.  The correlation between the Mutism subscale mean score and the PAS Social Anxiety subscale was rs = -.64 (p = .004), PAS Physical Injury Fears was rs = -.56 (p = .016), and PAS GAD was rs = -.57 (p = .015).  It is of note that the negative correlation was due to the anchors of the response options being reversed on the Mutism subscale (lower scores suggest more problematic behaviours) compared to the SAS-TR and the PAS (higher scores suggest more problematic behaviours).  Please see Table 3.6 for score distributions for the SAS-TR and the PAS, and Table 3.7 for correlations between the Mutism subscale to the PAS and SAS-TR.   90   Table 3.6: Mean and Median Scores for SAS-TR and PAS for Sample 2 Variable SM Group   ANX Group   TOTAL Mean Mdn Min Max N   Mean Mdn Min Max N   Mean Mdn Min Max N SAS-TR 34.00 29.50 29 45 6  8.58 6.00 0 22 12  17.06 15.00 0 45 18  (7.43)      (7.94)      (14.46)  PAS 40.50 40.00 15 70 8  21.50 12.50 6 70 22  26.57 19.00 6 70 30  (22.25)      (19.25)      (21.47) GAD 6.13 5.50 0 16 8  3.05 2.00 0 12 22  3.87 2.50 0 16 30  (5.59)      (3.36)      (4.20) Social Anxiety 16.25 14.50 10 24 8  6.00 4.00 1 20 22  8.73 6.00 1 24 30  (5.04)      (5.47)      (7.00) OCD 2.38 1.50 0 7 8  1.91 1.00 0 7 22  2.03 1.00 0 7 30  (2.88)      (2.37)      (2.47) Phys Injury 9.63 10.50 0 19 8  6.64 4.00 0 27 22  7.43 4.00 0 27 30  (6.99)      (6.64)      (6.75) SAD 6.13 4.50 1 15 8  3.91 2.00 0 14 22  4.50 3.00 0 15 30  (4.79)      (4.19)      (4.38) PAS: Preschool Anxiety Scale SAS-TR: School Anxiety Scale - Teacher Form  91   Table 3.7: Spearman’s Rho Correlations for Mutism Subscale Scores to SAS-TR and PAS for Sample 2 VARIABLE Subscale Mutism SubscaleScore P N SAS-TR -.84** ≤ .001 18  PAS -.62* .006 18 GAD -.56* .015 18 Social Anxiety -.64* .004 18 OCD -.40 .100 18 Phys Injury Fears  -.56* .016 18 SAD -.40 .100 18 *p ≤ .05. **p ≤ .001. Mutism subscale of the TTI-SM-DSM-IV and TTI-SM-R SAS-TR: School Anxiety Scale – Teacher Report PAS: Preschool Anxiety Scale SMQ: Selective Mutism Questionnaire   A Mann-Whitney U test for the SAS-TR revealed that students in the SM group (Mdn = 29.50) scored higher on behaviours that are related to school anxiety compared to students in the NC group (Mdn = 6.00; U = 72.00, p ≤ .001, r = .80).  For the PAS, Mann-Whitney tests revealed that students in the SM group (Mdn = 40.00) did not score higher on the PAS total score compared to the NC group (Mdn = 12.50; U = 136.00, p ≤ .24, r = .41).  There were also no  92  differences detected for the GAD, OCD, Physical Injury Fears, and SAD subscales, with respect to the central tendency.  However, the U test revealed that students in the SM group scored higher (Mdn = 14.50) on the Social Anxiety subscale compared to students in the NC group (Mdn = 4.00; U = 159.50, p ≤ .001, r = .62). 3.3.6 Statistical analysis and results for research question 6 . The sixth, and final, research question was: What is the concurrent validity evidence for the Mutism subscale in relation to other classroom behaviours of students with SM?  To answer this research question, Chi-square analyses were conducted (due to the categorical variables) to investigate whether students with SM score differently on these other classroom behaviours compared to students without SM (Behaviour subscale).  Given that the Behaviour subscale (other classroom behaviours items are theoretically related to students with SM) was administered at the time of the Mutism subscale, the results provided evidence of concurrent validity for the Mutism subscale. First, response options for the Behaviour subscale were recomputed into dichotomous variables given the low sample size.  The Behaviour subscale queried teachers on student’s classroom and school social participation.  Teachers who reported that students had “No problem, or only mild problem (and problem occurs seldom)” were given a score of “0”. Teachers who reported that students had a “Marked problem (occurs often)” and “Severe problem (occurs most or all of the time)” were given a score of “1”. 3.3.6.1 Findings based on Sample 1 (clinical sample). For Sample 1, Chi-square analyses revealed that students with SM had more problems on the Behaviour subscale (behaviours that were considered “Marked Impairment” to “Severe Impairment”) compared to students with anxiety disorders (ANX).  In particular, teachers rated  93  students with SM as having significantly more difficulty: 1) responding during morning announcements (χ2 [1, N = 25] = 4.74, p = .030); and 2) going up to the front of the group to write on the board or calendar (χ2 [1, N = 25] = 4.59, p = .032).  Students also appear to have difficulty responding when something funny happens (e.g., another student talks about a funny event, etc.), although this was not statistically significant (χ2 [1, N = 28] = 3.32, p = .069).  Please see Table 3.8 for Chi-square analyses.  The results of the Chi-square analyses provided evidence for concurrent validity.  Specifically, classroom behaviours that are believed to theoretically be problematic for students with SM appeared to be problematic for students with SM compared to students without SM.  Figure 3.1 presents the percentage of students with SM and ANX with marked to severe problems on the Behaviour subscale.  94   Table 3.8: Chi-square Analysis for the Behaviour Subscale for Samples 1, 2, and Combined Samples Sample 1  Variable Group N x 2  p Eta 1 Mood and attitude when s/he enters the classroom NC 10 .26 .611 .096   SM 18 2 Typical response when National Anthem or morning announcements occur NC 8 4.74 .030 .435   SM 17 3 Response when something funny happens NC 10 3.32 .069 .344   SM 18 4 Ability to come to the front of the group and write on the board/calendar NC 8 4.59 .032 .428   SM 17 5 Ability to work with any group of peers NC 10 .08 .778 .053   SM 18  95   Sample 1  Variable Group N x 2  p Eta 6 Ability to transition within the class and prepare for recess NC 10 .44 .507 .125   SM 18 7 Ability to make close friends NC 10 .00 1.00 .000   SM 18 8 Classroom popularity when interacting with peers NC 10 .26 .61 .096     SM 18 Sample 2  Variable Group N x 2  p Eta 1 Mood and attitude when s/he enters the classroom NC 12 .53 .467 .171   SM 6 2 Typical response when National Anthem or morning announcements occur NC 12 .44 .506 .161   SM 5 3 Response when something funny happens NC 12 1.00 1.00   SM 6  96   Sample 2   Variable Group N x2 p Eta 4 Ability to come to the front of the group and write on the board/calendar NC 12 6.79 .009 .614   SM 6 5 Ability to work with any group of peers NC 12 1.8 .180 .316   SM 6 6 Ability to transition within the class and prepare for recess NC 12 .28 .596 .125   SM 6 7 Ability to make close friends NC 12 6.79 .009 .614   SM 6 8 Classroom popularity when interacting with peers NC 12 7.20 .007 .632     SM 6 Combined Sample  Variable Group N x 2  p Eta 1 Mood and attitude when s/he enters the classroom NC 22 1.18 .277 .16   SM 24  97   Combined Sample   Variable Group N x2 p Eta 2 Typical response when National Anthem or morning announcements occur NC 20 6.45 .011 .392   SM 22 3 Response when something funny happens NC 22 6.32 .012 .371   SM 24 4 Ability to come to the front of the group and write on the board/calendar NC 20 12.92 .000 .548   SM 23 5 Ability to work with any group of peers NC 22 .03 .238 .027   SM 24 6 Ability to transition within the class and prepare for recess NC 22 .03 .857 .027   SM 24 7 Ability to make close friends NC 22 3.42 .064 .273   SM 24 8 Classroom popularity when interacting with peers NC 22 2.49 .115 .233     SM 24 NC = Normal Control Group. SM = Selective Mutism Group Behaviour subscale in the TTI-SM-DSM-IV and TTI-SM-R   98  Figure 3.1: Marked to Severe Problems on the Behaviour subscale for Sample 1 (SM vs. ANX)   3.3.6.2 Findings based on Sample 2 (community-based sample). For Sample 2, Chi-square analyses revealed that students with SM had more problems on the Behaviour subscale (behaviours that were considered “marked impairment” to “severe impairment”) compared to students in the NC group.  In particular, teachers rated students with SM as having significantly more difficulty: 1) ability to go up to the front of the group to write on the board or calendar (χ2 [1, N = 18] = 6.79, p = .009); 2) ability to make close friends (χ2 [1, N = 18] = 6.79, p = .009); and 3) classroom popularity when interacting with peers (χ2 [1, N = 18] = 7.20, p = .007).  Please refer to Figure 3.2.  The results with Sample 2 provided evidence of concurrent validity.   99  Figure 3.2: Marked to Severe Problems on the Behaviour subscale for Sample 2 (SM vs. NC)   3.3.6.3 Findings based on the combined sample. For the combined sample, Chi-square analyses revealed that overall, students with SM had more problems on the Behaviour subscale (behaviours that were considered “marked impairment” to “severe impairment”) compared to students within the NC group.  With all students combined, teachers rated students with SM as having significantly more difficulty: 1) responding during morning announcements (χ2 [1, N = 42] = 6.45, p = .011); 2) responding when something funny happens (e.g., another student talks about a funny event, etc.) (χ2 [1, N = 46] = 6.32, p = .012); and 3) going up to the front of the group to write on the board or calendar (χ2 [1,  100  N = 43] = 12.92, p ≤ .001).  The results provided additional evidence for concurrent validity of the Mutism subscale.  See Figure 3.3.  Correlational analyses between the Mutism subscale mean score to individual items of the Behaviour subscale revealed a strong relationship between the Mutism subscale mean score to 7 out of the 8 other behaviours, with correlations that ranged from r = .30 to -.67 (p ≤ .001).  The only item that was not significantly correlated with the Mutism subscale score was item number 6, How much of a problem to classroom routines or how problematic to him/herself is this student’s ability to transition within the class and prepare for recess?  These results differed from the results obtained for Samples 1 and 2.  Figure 3 3: Marked to Severe Problems on the Behaviour subscale for Combined Sample (SM vs. NC)    101  4. Conclusion and Discussion 4.1  Summary Prevalence estimates for children with selective mutism (SM) is as high as 1%, which means that school and mental health professionals are likely to encounter students with SM in their practice.  Although children with SM exhibit most of their symptoms at school, current assessments are more clinically focused and typically involve parent reports only (Sharp et al., 2007).  However, research has continued to find issues with relying on parent reports for this disorder which include: the internalizing nature and non-observable symptoms of SM, and the observable behaviours associated with SM that occur only outside the home setting (e.g., school, classroom; Comer, & Kendall, 2004).  Relying solely on parent reports may help explain the 3 to 6 year delay in referrals for children with SM.  Given such limitations of parent reports, best practice would indicate that multiple informants (parent and teacher) participate in the assessment and diagnosis of all childhood internalizing disorders (Grills, & Ollendick, 2003; Silverman et al., 2001). The overarching goal of the research was to validate a teacher reported instrument of SM in order to minimize the delay in referrals and treatment of these children.  Specifically, the Mutism subscale of the TTI-SM-DSM-IV (telephone interview) and the revised TTI-SM-R (paper- pencil version).  Given that the school is often the environment where the child exhibits the most symptoms of SM, and teachers are usually first to notice symptoms of SM, a teacher measure may help in the early identification of SM in children, and hence, potentially reduce the lag in these children receiving intervention. There were three areas that were addressed in this study through the validation of the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R.  First, the validation of the Mutism  102  subscale may help contribute to a better understanding of symptoms of anxiety in children with SM, given the continued debate about the classification of SM in the Diagnostic and Statistical Manual of Mental Disorders, 4 th  Edition, Text Revision (DSM-IV-TR; APA, 2000), and whether SM is actually better classified as an anxiety disorder (specifically, a social anxiety disorder) or a standalone disorder.  Next, the relationship between scores on the Mutism subscale (from the TTI-SM-DSM-IV and the TTI-SM-R) and on the Selective Mutism Questionnaire (SMQ; Bergman 1999; 2008) not only provided evidence for convergent validity, but may support better the concordance rate between teacher and parent ratings of SM in students.  Lastly, the current study also provided information about social participation of students with SM in the school and classroom contexts, which is an area that has had very little research. The development and validation of a teacher reported measure for SM (whether it is a telephone interview or paper-pencil measure) may help reduce the lag time between symptom onset and treatment referral for students with SM.  The results of the study may also help us better understand symptoms of school and social anxiety in students with SM, multiple informer agreement for SM, and the impact of mutism on social participation, which may then be linked to school-based interventions. A summary of each of the findings and how these findings build on past literature will be discussed next.  General implications of the findings, including theoretical, research, and practical implications will be discussed to understand better how the findings of the current study contribute new knowledge in this current field.  Finally, the chapter will conclude with a discussion of the strengths and limitations of the study, and suggested future directions in light of the findings.  103  4.2  Discussion of Research Findings This study aimed to answer one overarching research question, which was: To what extent does a teacher reported measure for selective mutism (SM) identify children with SM accurately and consistently?  To answer this large research question, 6 sub-questions were addressed in this research.  These were: 1) What is the reliability of the Mutism subscale? 2) What is the face validity evidence for the Mutism subscale? 3) What is the predictive validity evidence (i.e., diagnosis of SM) for the Mutism subscale? 4) What is the convergent validity evidence for the Mutism subscale in relation to parent-reported measures of SM? 5) What is the convergent and discriminant validity evidence for the Mutism subscale in relation to other measures of nonspecific anxiety? and, 6) What is the concurrent validity evidence for the Mutism subscale in relation to other classroom behaviours of students with SM? Two samples of children were recruited for this validation study.  First, Sample 1 was drawn from a clinical sample and was part of a larger multi-site investigation.  Sample 2 was drawn from a community-based sample.  Parent and teacher measures for the two samples differed.  One notable difference was that teachers in Sample 1 were administered the TTI-SM- DSM-IV telephone protocol, whereas teachers in Sample 2 were administered the TTI-SM-R paper-pencil version, which was developed in this study.  The results of this study show that the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R demonstrated evidence of reliability. These results were established with Samples 1 and 2, and when data were combined for a third analysis (the combined sample).  Across all analyses, the internal consistency and split-half reliability were high (all of which were above .96, p ≤ .001), therefore suggesting that the Mutism subscale is measuring one unitary concept.  Reliability analyses also showed that removing any 1 of these 11 items would result in a lower Cronbach’s alpha.  In other words, it  104  appears that the 11 items of the Mutism subscale strongly contributed to the overall reliability of the scale.  For the combined sample, principal component loadings were .62 or higher, which suggests that there was not a lot of variability between the 11 items, and that the 11 items in combination were measuring one unitary factor. The remaining 5 research questions concerned the validity of the Mutism subscale of the TTI-SM-DSM-IV (telephone protocol) and the TTI-SM-R (revised paper-pencil version).  Various individual facets of validity for the Mutism subscale were investigated in the current study. Messick’s (1989) discussion on the unitary concepts of validity was considered in the interpretation of the validity evidence of the Mutism subscale given that it is often difficult to disentangle the relevance and importance of one type of validity evidence over another.  Thus, each piece of validity evidence was interpreted using a unified concept view of validity. Predictive validity was investigated in the Mutism subscale by investigating the relationship between scores on the teacher-reported Mutism subscale and SM diagnosis (via the ADIS-P diagnostic parent interview), and whether there were significant differences in Mutism subscale scores for SM versus non-SM students.  The analyses showed a moderate to strong correlation between the Mutism subscale and SM diagnosis, with correlations ranging from -.64 (for Sample 1) to -.82 (for Sample 2).  Also, children with diagnosed SM scored lower on the Mutism subscale (indicating reliable mutism symptoms) compared to students without a diagnosis of SM.  In other words, there was evidence that the Mutism subscale of the TTI-SM- DSM-IV (telephone protocol) and the TTI-SM-R (revised paper-pencil version) may be able to predict clinical diagnosis of SM in students. Face validity was investigated by asking professionals in the field of child mental health to review and provide feedback on Mutism subscale items.  Reviewers provided feedback on the  105  format, length, clarity, wording, and relevancy of the TTI-SM-R, specifically, the Mutism subscale.  The reviewers were also asked to suggest additional questions they thought would tap into the construct of “selective mutism”.  Overall, the group of reviewers unanimously agreed that the Mutism subscale of the revised TTI-SM-R paper-pencil version appeared to measure what it was purported to measure, thus providing evidence for face validity. Construct validity, specifically, convergent and discriminant validity were established by comparing the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R to other scales that are theoretically related to SM.  Teacher ratings on the Mutism subscale were compared to well- validated parent report measures (Selective Mutism Questionnaire [SMQ]; Bergman et al., 1999; 2008; and Preschool Anxiety Scale [PAS]; Spence et al., 2004), and teacher report measures (School Anxiety Scale-Teacher Report [SAS-TR]; Lyneham et al., 2008), but that also have unique variance.  Correlational analyses showed that the responses on the Mutism subscale correlated highly to the SMQ.  Additionally, the Mutism subscale showed moderate correlations to measures of school and social anxiety, constructs that are theoretically related to SM.  A moderate correlation would mean that although the Mutism subscale is related to the SAS-TR and the PAS Social Anxiety subscale, there is unique variance between the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R, with other measures of anxiety.  In terms of discriminant validity, the Mutism subscale had weak correlations with other measures of anxiety that are not theoretically related to SM (i.e., obsessive compulsive disorder, and separation anxiety disorder). Weak correlations may indicate that the Mutism subscale is measuring a construct that is different from anxiety, with unique variance. For evidence of concurrent validity, the Mutism subscale was administered along with a set of other classroom behaviours of students with SM (Behaviour subscale).  These classroom  106  behaviours are mainly related to social participation, and should theoretically be problematic for students with SM.  Results of Chi-square analyses indicated that students with SM had more problems on the Behaviour subscale compared to students with anxiety disorders (ANX) and students with no diagnosis.  In particular, teachers reported that students with SM have difficulty: 1) responding during morning announcements (Samples 1 and Combined Sample); 2) going up to the front of the group to write on the board or calendar (Samples 1 and 2 and Combined Sample); 3) responding when something funny happens (e.g., another student talks about a funny event, etc.) (Combined Sample); 4) making close friends (Sample 2); and 5) classroom popularity when interacting with peers (Sample 2).  The results for the combined sample differed from the results that were obtained separately from Samples 1 and 2. An interesting finding to highlight was the specific items on the Behaviour subscale that were found to be more problematic for children in Sample 1 compared to Sample 2.  Overall, children in Sample 1 (clinical sample of SM and ANX groups) had more difficulty across all of the items on the Behaviour subscale compared to Sample 2.  The higher rate of reported difficulties may be due to Sample 1 being drawn from a clinical sample, whereas Sample 2 was drawn from a community-based (school) sample.  For Sample 1, group differences were detected for behaviours related to going up to the front of the group to write on the board or calendar, and responding during morning announcements.  In other words, although the SM and ANX groups in Sample 1 had difficulty across the set of school and classroom behaviours, children with SM seem to have more difficulty with respect to certain behaviours over and above that of children with a diagnosed anxiety disorder. For Sample 2 (community-based sample), there were fewer reported difficulties on the Behaviour subscale.  Only a small percentage of children with SM and NC were reported to have  107  any difficulty with their mood and attitude when entering the classroom, responding during morning announcements, responding when something funny happens, and ability to transition within the class.  However, there were marked difficulties detected between SM and NC groups in their ability to go to the front of the group to write on the board/calendar, ability to make close friends, and classroom popularity.  Taken together, when children with SM were compared to children without SM, children with SM appear to have some unique difficulties with specific school and classroom behaviours compared to children without SM. Another facet that was important to consider was the social validity and social consequences of the validity evidence of the Mutism subscale.  The overarching goal of the research was to validate a teacher reported instrument of SM to minimize the delay in referrals and treatment of these children.  The development of the TTI-SM-DSM-IV and a revised paper- pencil version (TTI-SM-R) was in response to the lack of available teacher reported measures of SM.  These measures also helps bridge assessment to intervention, by including questions about the classroom climate (e.g., seating arrangements, number of students in the class) as well as individual student behaviours that were important to collect in the assessment of children with SM (e.g., SM behaviours, active participation, verbal and non-verbal behaviours, previous interventions that have been tried).  Further, past research has shown that SM is a disorder that is perplexing to parents and teachers and has a strong impact not only on a child’s social and academic functioning, but also on teacher emotions and feelings of teaching efficacy. Over and above this goal of establishing reliability and validity evidence, the findings additionally address the issue of multiple-informer agreement in parent and teacher reports of SM.  Parent and teachers showed high concordance rates on the Mutism subscale of the TTI-SM- DSM-IV and the TTI-SM-R.  These findings are in contrast to some literature for other  108  internalizing disorders (i.e., anxiety disorders), which has revealed low congruence between parent and teacher reports of child behaviours (Youngstrom, Loeber, & Stouthamer-Loeber, 2000; Verhulst & Akkerhuis, 2006). One explanation for such a high concordance rate may be due to the characteristics and symptoms of SM.  The majority of SM symptoms are observable characteristics in children, mainly, a child’s failure to speak in social settings.  Symptoms of SM are different from those of other anxiety disorders, and behaviours are largely context specific.  Thus, although SM is considered to be an internalizing disorder, many of the characteristics of SM are quite observable, which makes the symptoms easier to identify and report. With the first clinical sample, the correlation between teacher ratings on the Mutism subscale of the TTI-SM-DSM-IV to mother reports on the SMQ measure was highly correlated to the School and the Other (Public) factors, and moderately correlated to the Family factor. However, the Mutism subscale was only significantly correlated to father reports on the SMQ measure on the School factor.  This correlation may be due to the lower response rates of fathers. The results show that overall, teachers and parents rate children similarly on speaking behaviours in the school and social setting.  However, there was low concordance between teacher ratings of SM and parent ratings of speaking behaviours in the home, which was expected because teachers do not see children in the home setting. Findings with the second, community sample were similar to the findings with the first study except that only one parent was queried on SM (rather than both mother and father being queried on SM).  Again, the results showed that teacher Mutism subscale scores on the TTI-SM- R were found to be strongly correlated to parent reports on the SMQ School and Other (Public) factors, and moderately correlated to the Family factor.  An interesting finding was that teacher  109  ratings of child anxiety (SAS-TR) were found to be correlated to parent reports of child anxiety (PAS).  Specifically, there was a strong relationship between how a teacher rates a child on school anxiety and how a parent rates a child on generalized anxiety, social anxiety, and fears of physical injury.  The findings in this study contradict the literature showing a low concordance rate between teacher and parent reports of child anxiety. These results suggest that teacher reports in the evaluation of SM in children may be informative in the identification and diagnosis of SM.  Given that past research has found flaws in parent reports of childhood internalizing disorders, it may be important to consider teacher reports of SM in the future.  For example, some studies have found that parents may over or under-report symptoms, which may be due to the low severity of SM symptoms in the home setting and tendency of parents to misinterpret symptoms as shyness.  Also, parents may have anxiety or be shy themselves, thus not seek mental health services of their child. One interesting finding was that children with SM did not score higher on the PAS overall total score compared to the NC group, which suggested that overall anxiety in both groups of students appeared to remain similar.  For example, there were no differences detected between scores on the GAD, OCD, Physical Injury Fears, and SAD subscales.  However, children in the SM group scored higher on the Social Anxiety subscale compared to children in the NC group.  Given that the Social Anxiety subscale is only 1 of 5 subscales on the PAS, the scores on this particular subscale may not have impacted the overall total PAS score.  The finding is consistent with past findings that show that children with SM often have difficulties with social anxiety, but no particular difficulty with respect to other anxieties. The findings of this study complement past findings about the similarity of symptoms found in children with SM and social phobia.  There continues to be debate about the  110  classification and diagnosis of SM with a strong relationship between SM and anxiety disorders. In this sample, 68% of the children diagnosed with SM also had a co-occurring diagnosis of social phobia.  The results show that children with SM also exhibit many symptoms of social phobia, thus lending support that SM may be a variant of social phobia (perhaps an extreme form of social phobia in which fears are so debilitating that a child is unable to speak).  However, due to the small sample size, all of these findings should be viewed with caution. 4.3 Strengths and Limitations of the Research A strength of the study was that children’s teachers and parents participated in the study, thus resulting in a more comprehensive study on child behaviours.  Parents were also administered a diagnostic interview using the ADIS-P, which established a reliable diagnosis of the child. Despite the higher rates of children with SM who are English Language Learners (ELL) and/or new immigrant language minorities (Elizur & Perednik, 2003; Toppelberg et al., 2005), our sample did not include a higher number of children from ELL backgrounds.  The low rates of referrals for children of ELL backgrounds for Sample 1 may be due to similar issues that have been found concerning mental health identification and referrals made for minority children/adolescents (i.e., stigma, social and language barriers to care, and knowledge of mental health issues).  For Sample 2, teachers were specifically asked not to nominate children of ELL backgrounds due to the limitations concerning the proper assessment and diagnosis of SM (i.e., conducing parent diagnostic interviews in English, asking parents to complete rating scales that are in English, and using the DSM-IV diagnostic criterion for SM whereby failure to speak cannot be due to a lack of knowledge of or comfort with the spoken language required in the situation).  It will be important for future research to investigate the effectiveness of mental  111  health assessments and screeners, such as the TTI-SM-R for parents and children of ELL backgrounds.  This is particularly important given the findings that the prevalence of SM is almost 4 times higher in ELL populations. Another limitation of the study is that there were no measures of communication disorder (i.e., expressive or receptive language) that were administered to child participants.  Thus, in this study, it is conceivable that the children diagnosed with SM may actually be suffering from a communication disorder, or may have a communication problem co-occurring with their SM. For a diagnosis of SM, the DSM-IV (APA, 2000) states that a child’s failure to speak cannot be better accounted for by a Communication Disorder.  Given the co-morbidity that has been found between SM and communication disorders, it will be important for future research to include measures of oral communication and language, in order to rule out communication problems and select an empirically “pure” sample of children with SM only. The low prevalence rate of SM, paired with the challenges of recruiting participants in a community-based sample resulted in a low sample size in Samples 1 and 2.  Thus, the low sample size presents one of the major weaknesses of the study.  Although prevalence rates of SM has varied from study to study, it was projected at the outset of the second phase (with the community-based sample) that there would be more students identified and referred for study in this geographic area.  However, the number of students identified and referred to our study did not reflect the projected numbers of children with SM. The reason for such a low referral rate may be due to several factors.  In the first phase (with the clinical sample), one factor that impacted the recruitment was that during the time of the project, there was a job action (i.e., workers’ strike) in the province, which resulted in a lower-than-anticipated response rate from teachers from one Canadian province.  In the second  112  phase (with the community-based sample), the recruitment procedure involved a 3-stage multiple gating procedure (Cronbach & Glesar, 1965), which was believed to be beneficial to the study. However, this procedure created multiple levels of barriers to our recruitment.  At every level of recruitment (approval from the District Superintendant, referrals from school mental health professionals, approval and permission from the School Principals, and consent and completion of questionnaires by teachers and parents), there were additional barriers to enrolment and retention in the study.  As a result, there was a lower response rate than what was anticipated at the outset of the study. In some instances, school districts reported that there were no students with SM in their district, which resulted in a large population of children not being recruited.  Another barrier was the willingness of individual school principals to allow teachers to participate in the study. Teachers’ willingness to spend time to participate in completing questionnaires also posed a barrier, given the increasing responsibilities of teachers today.  Incentives, such as talks at the school on anxiety disorders, telephone support for teachers, and gift cards were offered; however, these strategies proved to be minimally effective in facilitating teacher participation. There were instances in which children were referred to the study, but the teacher did not give consent to participate; therefore, this child could not be included in the study.  There were also a large number of students whose teachers initially agreed to participate.  However, following the receipt of all parent materials, some teachers were no longer interested in participating and completing the teacher measures.  Similar to what was hypothesized by other researchers (Bergman et al., 2008; Sharp et al., 2007; Standart & Couteur, 2003), our response rate was reduced compared to actual prevalence rates of SM.  113  To help address the challenges of conducting analyses with small sample sizes, a decision was made to combine data from Samples 1 and 2 to produce a larger sample for analyses. Combining data from smaller samples for a larger overall sample is deemed acceptable provided there is a justification for combining the samples (National Research Council, 2004).  The main reason for combining the two samples in this study was due to sample similarity; the two samples were similar in age and clinical diagnoses, and data from multiple informants (i.e., teachers and parents) were drawn on similar measures (i.e., SM and anxiety).  The larger sample would increase the overall power of the analyses and reduce sampling bias.  Despite some of the advantages associated with combining the two samples, there were also disadvantages to this approach that needed to be considered (e.g., different modes of administration of the Mutism subscale to teachers, and moderate effects that may not be captured when samples are combined). It was unknown whether combining the two samples would yield different results from what would be found separately for each of the independent samples.  As the individual samples were drawn from two different populations (clinical versus community-based), we believed it would be useful to provide the results for each sample, and the results for the combined sample. By providing all results, any unique differences for each of the two independent samples would be captured, in addition to any differences that may be captured when the samples were combined. One interesting finding from the combined results was for research question 6.  The results suggest that the 2 samples were different in many ways, but similar in others.  Descriptive statistics indicated that behaviours (from the Behaviour subscale) were found to be problematic only when the two samples were combined.  Specifically, in Sample 2 (community-based  114  sample), children with SM were not reported to have problems for 3 out of the 8 behaviours; however, a number of children in Sample 1 (clinical sample) were reported to have moderate to severe problems with these 3 behaviours.  After combining the two samples, the descriptive statistics revealed children with SM had difficulty across all of the 8 behaviours, when in fact, this effect was only present for Sample 1, clinical sample. Another interesting finding was for item number 8 on the Behavioural subscale.  The significant group differences found between children with SM and NC in Sample 2 were no longer detected after the two samples were combined.  Some statistically significant results were no longer obtained after the two samples were combined for analyses.  For items 2, 3, and 8 of the Behavioural subscale, group differences between children with SM and NC were detected when the samples were combined.  Thus, the analysis for the combined sample appeared to capture modest effects that were not found when the samples were analyzed individually. It will be important for future researchers working with small samples to consider the justification for combining small samples, and to take into consideration whether analyses are more informative when small samples are combined into a larger sample, or whether each sample should be analyzed individually.  In the current study, the results for the individual samples were presented along with the results for the combined samples because it was unknown whether or not the results would produce similar or dissimilar results.  Thus, by providing both sets of results, it provided us with a more comprehensive understanding about differences within each of the two smaller independent samples, while at the same time, provided us with information about the samples when they were combined into a larger group for analyses. Another limitation was the different procedures that were used for teachers in collecting teacher-reported data.  This provided challenges in understanding the construct validity evidence  115  of the Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R, and how much of the validity evidence was indeed due to the Mutism subscale capturing the construct of selective mutism and how much of that was due to measurement error.  It was unclear whether the results from the Mutism subscale were affected by the mode of administration rather than the questions that made up the Mutism subscale. The recruitment method also presented as a challenge in terms of the study’s goal in obtaining grade and gender-matched control group students.  Due to the previously mentioned barriers in recruitment and participation, there were an inadequate number of gender and grade matched controls to allow for statistical analyses using matched pairs.  As a result, the effect of gender and grade could not be investigated.  Additionally, teacher ratings of the student pairs (student with SM and student with no SM) could not be controlled. One of the main challenges of having a low sample size was the resultant limitations to generalizability (external validity) of the results of the study.  Also, due to the sample size, more commonly used parametric statistical methods were not used for group comparison analyses. Rather, nonparametric methods were employed for group comparisons in both studies.  Using nonparametric methods was not a particular weakness because this method was actually more statistically powerful compared to parametric methods when sample sizes are small.  Parametric methods were only used for exploratory purposes and for reliability analyses, with the caveats of parametric methods for small samples taken into consideration. Another limitation of the study was the availability of research measures for parent- reported anxiety in children.  Based on the age group selected for study, there were few parent- reported child anxiety scales available.  In the first study, the Multidimensional Anxiety Scale for Children (MASC; March, et al., 1997) and the Social Anxiety Scale for Children (SASC-R; La  116  Greca & Stone, 1993) was used to measure anxiety given the target age of the sample.  These measures could not be used in the second phase with the community-based sample given that the target age of the second phase were younger than in the first phase (clinical sample).  A decision was made to use the PAS as the measure of child anxiety for the current study, instead of a measure normed with older school-age children (age 8 and up). 4.4 Future Directions The Mutism subscale of the TTI-SM-DSM-IV and the TTI-SM-R appears to have adequate psychometric properties and merits further research.  A larger sample of children diagnosed with selective mutism would be useful in providing additional evidence of reliability and validity, and to develop a normative sample for the Mutism subscale.  Administering the measure over several time points would assess the stability of the Mutism subscale.  It will also be useful to administer the measure to a more geographically, linguistically, and culturally diverse population to improve the generalizablity of the measure.  With further research to replicate findings of the current study, these measures could potentially be used as a primary source of information for the identification of SM in school settings.  When modified into a brief paper-pencil version, the TTI-SM-R can potentially be used as a school screener, in which it can be administered by teachers or school mental health professionals when a student shows difficulty speaking to peers and adults at school.  After the child has been identified by the school, the child may be referred for school-based mental health services, or the family may be referred for community-based mental health services. A widely available teacher-reported measure of SM may help reduce the delay in clinical referrals for children with SM in the future, and allow these children to gain earlier access to mental health services.  Students of ELL backgrounds and/or new immigrant language minorities  117  may also be identified early, which is important given the elevated prevalence rates of SM among this population of children.  In clinic settings, the revised TTI-SM-R (paper-pencil version) could be used as a supplementary source of information for the identification and diagnosis of SM. The findings concerning school and classroom behaviours were very interesting, and are an area that may warrant further research.  The preliminary results provided some indication that mutism may be impacting a students’ day-to-day functioning and social-emotional development in the classroom (i.e., going up to the front of the group to write on the board or calendar, responding during morning announcements, making close friends, and being popular in the classroom).  By gaining a better understanding of the specific difficulties which students with SM face, interventions can be better tailored to address these difficulties at school. One area of research that was not addressed in the current research project was the other 3 subscales of the original TTI-SM-DSM-IV, which included: verbal and nonverbal communication with teachers (14 items), verbal and nonverbal communication with peers (10 items), and externalizing behaviours (7 items).  The reason these three subscales were not investigated in this current study was because the goal of the study was to develop a brief and specific paper-pencil questionnaire for the early identification of SM.  However, it would be important in the future to understand better how children with SM communicate both verbally and nonverbally with teachers and peers, especially given the co-morbidity between SM and communication disorders.  It is also important to discover if children with SM exhibit any externalizing behaviours in the school context, as limited past findings that suggest oppositional behaviors in children with SM.  Understanding how students with SM communicate and whether  118  they may exhibit any externalizing behaviours in the classroom context may possibly help inform school-based interventions and accommodations for students with SM. An area of research, beyond the scope of this study, was to examine academic achievement in students with SM.  A metaanalysis of 114 studies, revealed that overall, there is very scant research concerning academic achievement in children with SM (Pionek Stone, Kratochwill, Sladezcek, & Serlin, 2002).  Kumpulainen and colleagues (1998) found that children with SM performed below grade level compared to their peers.  McInnes and Manassis (2005) found that children with SM appeared to have difficulty with academic achievement, which the researchers believed were due to the children’s lack of speech in the classroom and school settings.  These findings, however, are in contrast to findings by Cunningham and colleagues (2004) who found that children with SM did not perform significantly poorer on achievement (reading and math) and teacher-rated classroom cooperative skills compared to normal controls.  Overall, research findings in the area of achievement in students with SM are mixed and warrant further research. Although more direct forms of assessment (i.e., achievement testing) may be useful in measuring a child’s academic skills, it remains a challenge to conduct research with children with SM given their failure to speak in academic settings.  One avenue to measure achievement would be to use performance-based achievement measures and/or language-reduced assessment measures which would not require the child to use oral language while being evaluated.  There is a need to better understand academic achievement in children with SM, so that interventions may focus on both symptoms of SM and, if needed, school achievement. The impact of SM in school children has implications for the practice of School Psychology.  With prevalence rates at 1%, school psychologists are likely to encounter students  119  with SM in their practice. SM is typically first noticed in the schools by teachers, thus, school psychologists are in an ideal position to provide information and support to teachers, schools and families.  School psychologists can also be involved in the consultation, assessment, and intervention (direct and indirect) for students with SM, with the goal being to improve school behaviors and achievement. Given the overlapping symptoms in children with SM and social phobia, it would be also worthwhile to study the effects of school-based interventions for children with SM and social phobia, and whether children with SM respond similarly to interventions targeted for social phobia (i.e., cognitive-behavioral and behavioral interventions).  A review of published intervention literature suggests that cognitive-behavioural and behavioural interventions in the treatment of SM are most effective (Cohan et al., 2008).  However, Pionek-Stone and her colleagues (2002) and Kratochwill and Stoiber (2000), reveal that effectiveness of evidence- based treatments for SM when delivered in the school settings remains unknown.  Early intervention is crucial for children with SM, given that these children are rather resistant to treatment, and their resistance to treatment has been found to be related to the amount of time they have had SM (Kehle et al., 1998; Kolvin & Fundudis, 1981).  Consequently, there is a need to identify, intervene, and treat children with SM at an early age.  Thus, future research on SM should investigate the use of a multimodal approach to assessment and intervention, and bridge school and clinical interventions (as discussed in Cohan et al., 2008) in the treatment of children with SM in school settings.   120  References Achenbach, T. M. (1992). Manual for the Child Behavior Checklist/2-3 and 1992 Profile. Burlington, VT: University of Vermont, Department of Psychiatry. Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT: University of Vermont, Department of Psychiatry. Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems: Implications of cross informant correlations for situational specificity. Psychological Bulletin, 101(2), 213-232. doi: 10.1037//0033-2909.101.2.213 American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4 th  ed. Text Rev). Washington, DC: Author. American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3 rd  ed. Rev). Washington, DC: Author. Anstendig, K. D. (1999). Is selective mutism an anxiety disorder? Rethinking its DSM-IV classification. Journal of Anxiety Disorders, 13(4), 417-434. doi:10.1016/S0887- 6185(99)00012-2 Bar-Haim, Y., Henkin, Y., Ari-Even-Roth, D., Tetin-Schneider, S., Hildescheimer, M., & Muchnik, C. (2004). Reduced auditory efferent activity in childhood selective mutism. Biological Psychiatry, 55(3), 1061-1068. doi:10.1016/j.biopsych.2006.02.020 Barbosa, J., Manassis, K., & Tannock, R. (2002). Measuring anxiety: Parent-children reporting differences in clinical samples. Depression and Anxiety, 15(2), 61-65. doi: 10.1002/da.10022  121  Beidel, D. C., & Turner, S. M. (1997). At risk for anxiety: I. Psychopathology in the offspring of anxious parents. Journal of the American Academy of Child and Adolescent Psychiatry, 26 (7), 918-924. doi: 10.1097/00004583-199707000-00013 Benga, O., Tincas, I., & Visu-Petra, L. (2010). Investigating the structure of anxiety symptoms among Romanian preschoolers using the Spence Preschool Anxiety Scales. Cognition, Brain, Behavior: An Interdisciplinary Journal, 14(2), 159-183. Retrieved from: http://www.britannica.com/bps/additionalcontent/18/52103399/INVESTIGATING-THE- STRUCTURE-OF-ANXIETY-SYMPTOMS-AMONG-ROMANIAN- PRESCHOOLERS-USING-THE-SPENCE-PRESCHOOL-ANXIETY-SCALES Bergman, R. L., Keller, M. L., Piacentini, J., & Bergman, A. J. (2008). The development and psychometric properties of the Selective Mutism Questionnaire. Journal of Clinical Child & Adolescent Psychology, 37(2), 456-464. doi:10.1080/15374410801955805 Bergman, R. L., Piacentini, J., & McCracken, J. T. (2002). Prevalence and description of selective mutism in a school-based sample. Journal of the American Academy of Child and Adolescent Psychiatry, 41(8), 938-946. doi:10.1097/00004583-200208000-00012 Biedel, D. C., Turner, S. M., & Morris, T. L. (1999). Psychopathology of childhood social phobia. Journal of the American Academy of Child and Adolescent Psychiatry, 38(6), 643-650. doi:10.1097/00004583-199906000-00010 Black, B., & Uhde, T. W. (1992). Elective mutism as a variant of social phobia. Journal of the American Academy of Child and Adolescent Psychiatry, 31(6), 1090–1094. doi:10.1097/00004583-199211000-00015 Black, K. (2010). Business statistics: Contemporary decision making. Hoboken, NJ: John Wiley & Sons, Inc.  122  Briggs-Gowan, M. J., Carter, A. S., & Schwab-Stone, M. (1996). Discrepancies among mother, child, and teacher reports: Examining the contributions of maternal depression and anxiety. Journal of Abnormal Child Psychology, 24(6), 749-765. doi: 10.1007/BF01664738 Broeren, S., & Muris, P. (2009). The relation between cognitive development and anxiety phenomena in children. Journal of Child and Family Studies, 18(6), 702-709. doi: 10.1007/s10826-009-9276-8 Buss, A. H., & Plomin, R. (1984). Temperament: Early developing personality traits. Hillsdale, NJ: Erlbaum. Centre for Community Child Health (2007). Child behaviour: Overview of the literature, monograph 3. In A. O’Hanlon, A. Patterson, & J. Parham (Eds.), Promotion, prevention and early intervention for mental health in general practice. Retrieved from http://nla.gov.au/nla.arc-107363-20091002-1309- www.auseinet.com/files/resources/auseinet/child_beh.pdf Chavira, D. A., Shipon Blum, E., Hitchcock, C., Cohan, S., & Stein, M. B. (2007). Selective Mutism and Social Anxiety Disorder: All in the family? Journal of the American Academy of Child and Adolescent Psychiatry, 46(11), 1464-1472. doi:10.1097/chi.0b013e318149366a Chen, A. W., Kazanjian, A., & Wong, H. (2008). Determinants of mental health consultations among recent Chinese immigrants in British Columbia, Canada: Implications for mental health risk and access to services. Journal of Immigrant and Minority Health Volume, 10(6) 529-540. doi: 10.1007/s10903-008-9143-5  123  Chi, T. C., & Hinshaw, S. P. (2002). Mother-child relationships of children with ADHD: The role of maternal depressive symptoms and depression-related distortions. Journal of Abnormal Child Psychology, 30(4), 387-400. doi: 10.1023/A:1015770025043 Cline, T., & Baldwin, S. (1994). Selective mutism in children. San Diego, CA: Singular Publishing Group. Cohan, S. L., Chavira, D. A., Shipon-Blum, E., Hitchcock, C., Roesch, S. C., & Stein, M. B. (2008). Refining the classification of children with selective mutism: A latent profile analysis. Journal of Clinical Child & Adolescent Psychology, 37(4), 770-784. doi: 10.1080/15374410802359759 Comer, J. S., & Kendall, P. C. (2004). A symptom-level examination of parent-child agreement in the diagnosis of anxious youths. Journal of the American Academy of Child and Adolescent Psychiatry, 43(7), 878-886. doi:10.1097/01.chi.0000125092.35109.c5 Compton, S. N., March, J. S., Brent, D., Albano, A. M., Weersing, V. R., & Curry, J. (2004) Cognitive-behavioral psychotherapy for anxiety and depressive disorders in children and adolescents: An evidence-based medicine review. Journal of the American Academy of Child and Adolescent Psychiatry, 43(8), 930-959. doi:10.1097/01.chi.0000127589.57468.bf Conners, C. K. (1997). Conners' rating sales, revised. New York, NY: Multi-Health Systems Inc. Corder, G. W., & Foreman, D. I. (2009). Nonparametric statistics for non-statisticians. Hoboken, NJ: John Wiley & Sons, Inc. Cronbach, L. (1970). Essentials of psychological testing (3 rd  ed.). New York, NY: Harper & Row Publishers Inc.  124  Cronbach, L., & Gleser, G. (1965). Psychological tests and personnel decisions (2 nd  ed.). Urbana, IL: University of Illinois Press. Cunningham, C. E., McHolm, A., & Boyle, M. H. (2006). Social phobia, anxiety, oppositional behavior, social skills, and self-concept in children with specific selective mutism, generalized selective mutism, and community controls. European Child and Adolescent Psychiatry, 15(5), 245-255. doi: 10.1007/s00787-006-0529-4 Cunningham, C. E., McHolm, A., Boyle, M. H., & Patel, S. (2004). Behavioral and emotional adjustment, family functioning, academic performance, and social relationships in children with selective mutism. Journal of Child Psychology and Psychiatry, 45(8), 1363–1372. doi: 10.1111/j.1469-7610.2004.00327.x Dadds, M. R., James, R. C., Barrett, P. M., & Verhulst, F. C. (2004). Diagnostic issues. In T. H. Ollendick and J. S. March (Eds.), Phobic and anxiety disorders in children and adolescents: A clinician's guide to effective psychosocial and pharmacological interventions (pp. 3-33). New York, NY: Oxford University Press. DeVellis, R. F. (2003). Scale development: Theory and application (2 nd  ed.). Sage Publications, Newbury Park, CA. DeVellis, R. F. (1991). Scale development: Theory and application.  Sage Publications, Newbury Park, CA. DiBartolo, P. M., Albano, A. M., Barlow, D. H., & Heimberg, R. G. (1998). Cross-informant agreement in the assessment of social phobia in youth. Journal of Abnormal Child Psychology, 26(3), 213-220. doi: 10.1023/A:1022624318795  Dow, S. P., Sonies, B. C., Scheib, D., Moss, S. E., & Leonard, H. L. (1995). Practical guidelines for the assessment and treatment of selective mutism. Journal of the American Academy  125  of Child and Adolescent Psychiatry, 34(7), 836-846. doi:10.1097/00004583-199507000- 00006 Drewes, K. M., & Akin-Little, A. (2002). Children with selective mutism: seen but not heard. The School Psychologist, 56(2), 37-75. Retrieved from http://www.indiana.edu/~div16/publications/school_psychologist_spring_02%20.pdf Dummit, E. S., III, Klein, R., Tancer, N. K., Ashe, B., Martin, J., & Fairbanks, J. A. (1997). Systematic assessment of 50 children with selective mutism. Journal of the American Academy of Child and Adolescent Psychiatry, 36(5), 653–660. doi:10.1097/00004583- 199705000-00016 Edwards, S. L., Rapee, R. M., Kennedy, S. J., & Spence, S. H. (2010). The assessment of anxiety symptoms in preschool-aged children: The revised Preschool Anxiety Scale. Journal of Clinical Child and Adolescent Psychology, 39(3), 400-409. doi: 10.1080/15374411003691701 Egger, H. L., & Angold, A. (2006). Common emotional and behavioral disorders in preschool children: presentation, nosology, and epidemiology. Journal of Child Psychology and Psychiatry, 47 (3), 313-317. doi:10.1111/j.1469-7610.2006.01618.x Elizur, Y., & Perednik, R. (2003) Prevalence and description of selective mutism in immigrant and native families: a controlled study. Journal of the American Academy of Child & Adolescent Psychiatry, 42(12), 1451-1459. doi:10.1097/00004583-200312000-00012 Lucas, C. P., & Shaffer, D. (2010). Childhood disorders: Elimination disorders and childhood anxiety disorders. In M. B. First and A. Tasman (Eds.), Clinical guide to the diagnosis and treatment of mental disorders, second edition (pp. 87-98). Chinchester, West Sussex, UK: John Wiley & Sons.  126  Ford, M. A., Sladesczek, I. E., Carlson, J., & Kratochwill, T. R. (1998). Selective mutism: Phenomenological characteristics. School Psychology Quarterly, 13(3), 192–227. doi:10.1037/h0088982 Fung, D. S., Manassis, K., Kenny, A., & Fiskenbaum, L. (2002). Web-based CBT for selective mutism. Journal of the American Academy of Child and Adolescent Psychiatry, 41(2), 112–113. doi:10.1097/00004583-200202000-00003 Garcia, A., Freeman, J., Francis, G., Miller, L. M., & Leonard, H. L. (2004). Selective mutism. In: T. Ollendick (Ed.), Phobic and anxiety disorders in children and adolescents: A clinician’s guide to effective psychosocial and pharmacological interventions (pp. 433– 455). London, UK: Oxford University Press. Goodman, R. (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38(5), 581-586. doi: 10.1111/j.1469- 7610.1997.tb01545.x Grills, A. E., & Ollendick, T. H. (2003). Multiple informant agreement and the Anxiety Disorders Interview Schedule for Parents and Children. Journal of the American Academy of Child and Adolescent Psychiatry, 42(5), 30-40. doi:10.1097/00004583- 200301000-00008 Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate data analysis (5 th  ed.). Upper Saddle River, NJ: Prentice Hall. Hayden, T. L. (1980). Classification of elective mutism. Journal of the American Academy of Child and Adolescent Psychiatry, 19(1), 118-133. Doi: 10.1016/S0002-7138(09)60657-9 House, A. E. (2002). DSM-IV Diagnosis in the Schools. New York, NY: The Guilford Press.  127  Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P., et al. (1997). Schedule for affective disorders and schizophrenia for school-age children – Present and lifetime version (KSADS-PL): Initial reliability and validity data. Journal of the American Academy of Child and Adolescent Psychiatry, 36(7), 980–988. doi:10.1016/j.jpsychires.2008.10.003 Kehle, T. J., Madaus, M. R., Baratta, V. S., & Bray, M. A. (1998). Augmented self-modeling as a treatment for children with selective mutism. Journal of School Psychology, 36(3), 247- 260. doi:10.1016/S0022-4405(98)00013-2 Kolvin, I., & Fundudis, T. (1981). Elective mute children: Psychological development and background factors. Journal of child Psychology and Psychiatry, 22(3), 219-232. doi: 10.1111/j.1469-7610.1981.tb00548.x Kraemer, H. C., Measelle, J. R., Ablow, J. C., Essex, M.J., Boyce, W. T., & Kupfer, D. J. (2003). A new approach to multiple informants: Mixing and matching contexts and perspectives. American Journal of Psychiatry, 160(9), 1566-1577. Retrieved from http://ajp.psychiatryonline.org Kratochwill, T. R., & Stoiber, K. C. (2000). Empirically supported interventions in school psychology: Conceptual and practice issues–Part II. School Psychology Quarterly, 15(2), 233–253. doi: 10.1037/h0088786 Kristensen, H. (2000). Selective mutism and comorbidity with developmental disorder/delay, anxiety disorder, and elimination disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 39(2), 249–256. doi: 10.1097/00004583-200002000-00026  128  Kristensen, H. (2001). Multiple informants report of emotional and behavioral problems in a nation-wide sample of selective mute children and controls. European Child & Adolescent Psychiatry, 10(2), 135–142. doi: 10.1007/s007870170037 Kristensen, H., & Torgersen, S. (2001). MCMI-II personality traits and symptom traits in parents of children with selective mutism: A case-control study. Journal of Abnormal Psychology, 110(4), 648-652. doi:10.1037/0021-843X.110.4.648 Kronenberger, W. G., & Meyer, R. G. (2001). The child clinician's handbook, second edition. Boston, MA: Allyn & Bacon. Kumpulainen, K., Rasanen, E., Raaka, H., & Somppi, V. (1998). Selective mutism among second-graders in elementary school. European Child and Adolescent Psychiatry, 7(1), 24-29. doi: 10.1007/s007870050041 La Greca, A. M., & Stone, W. L. (1993). Social Anxiety Scale for Children – Revised: Factor structure and concurrent validity. Journal of Clinical Child Psychology, 22(1), 17-27. doi: 10.1207/s15374424jccp2201_2 Layne, A. E., Bernstein, G. A., & March, J. S. (2003). Teacher awareness of anxiety symptoms in children. Child Psychiatry and Human Development, 36(4), 383-392. doi: 10.1007/s10578-006-0009-6 Lung, A. Y., & Sue, S. (1997). Chinese American Children. In G.P. Johnson & J. Yamamoto (Eds.), Transcultural child development: Psychological assessment and treatment (pp. 208-236). New York, NY: John Wiley & Sons. Lyneham, H. J., & Rapee, R. M. (2005). Agreement between telephone and in-person delivery of a structured interview for anxiety disorders in children. American Academy of Child and Adolescent Psychiatry, 44(3), 274-282. doi:10.1097/00004583-200503000-00012  129  Lyneham, H. J., Street, A. K., Abbott, M. J., & Rapee, R. M. (2008). Psychometric properties of the School Anxiety Scale – Teacher Report (SAS-TR). Journal of Anxiety Disorders, 22(2), 292-300. doi:10.1016/j.janxdis.2007.02.001 MacLeod, R. J., McNamee, J. E., Boyle, M. H., Offord, D. R., & Friedrich, M. (1999). Identification of childhood psychiatric disorder by informant: Comparisons of clinic and community samples. Canadian Journal of Psychiatry, 44(2), 144-150. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10097834 Manassis, K., Fung, D., Tannock, R., Sloman, L., Fiskenbaum, L., & McInnes, A. (2003). Characterizing selective mutism: Is it more than social anxiety? Depression and Anxiety, 18(3), 153-161. doi: 10.1002/da.10125 Manassis, K., Tannock, R., Garland, J., Minde, K., McInnes, A., & Clark, S. (2007). The sounds of silence: Language, cognition, and anxiety in selective mutism. The American Academy of Child and Adolescent Psychiatry, 46(9), 1187-1195. doi:10.1097/CHI.0b013e318076b7ab March J. S., & Albano, A. M. (1998). New developments in assessing pediatric anxiety disorders. In T. Ollendick (Ed.), Advances in clinical child psychology (pp. 213-241). Washington, DC: American Psychological Press. March, J. S., Parker, J., Sullivan, K., Stallings, P., & Conners, C. (1997). The Multidimensional Anxiety Scale for Children (MASC): Factor structure, reliability and validity.  Journal of the American Academy of Child and Adolescent Psychiatry, 36(4), 554-565. doi:10.1097/00004583-199704000-00019 Martinez, Y. J., Manassis, K., Tannock, R., McInnes, A., & Clark, S. (under review). Psychometric properties of the Teacher Telephone Interview: Selective Mutism & Anxiety  130  in the School Setting (TTI-SM-DSM-IV) in clinical samples. Journal of Child Psychology and Psychiatry.  McInnes, A., Fung, D., Manassis, K., Fiksenbaum, L., & Tannock, R. (2004). Narrative skills in children with selective mutism: An exploratory study. American Journal of Speech- Language Pathology, 13(4), 304–315. doi:10.1044/1058-0360(2004/031) McInnes, A., & Manassis, K. (2005). Why silence is not golden: An integrated approach to selective mutism. Seminars in Speech and Language, 26(3), 201-210. doi: 10.1055/s- 2005-917125 Mesman, J., & Koot, H. M. (2000). Child-reported depression and anxiety in preadolescence: I. Associations with parent- and teacher-reported problems. Journal of the American Academy of Child and Adolescent Psychiatry, 39(11), 1371-1378. doi:10.1097/00004583- 200011000-00011 Messick, S. (1989). Meaning and values in test validation: The science and ethics of assessment. Educational Researcher, 18(5), 5-11. doi: 10.3102/0013189X018002005 Modaff, J. V., & Modaff, D. P. (2000). Technical notes on audio recording. Research on Language and Social Interaction, 33(1), 101-118. doi: 10.1207/S15327973RLSI3301_4 Muris, P., Mayer, B., Kramer Freher, N., Duncan, S., & van den Hout, A. (2010). Children’s internal attributions of anxiety-related physical symptoms: Age-related patterns and the role of cognitive development and anxiety sensitivity. Child Psychiatry and Human Development, 41(5), 535-548. doi: 10.1007/s10578-010-0186-1 National Research Council. (2004). Improved operational testing and evaluation. Washington, DC: The National Academies Press.  131  Nauta, M. H., Scholing, A., Rapee, R. M., Abbott, M., Spence, S. H., & Waters, A. (2004). A parent report measure of children's anxiety. Behaviour Research and Therapy. 42(7), 813-839. doi:10.1016/S0005-7967(03)00200-6 Nowakowski. M. E., Cunningham, C. C., McHolm, A. E., Evans, M., Edison, S., St. Pierre, J., et al. (2009). Language and academic abilities in children with selective mutism. Infant and Child Development, 18(3), 271-290. doi: 10.1002/icd.624 Omdal, H., & Galloway, D. (2008). Could selective mutism be re-conceptualised as a specific phobia of expressive speech? An exploratory post-hoc study. Child and Adolescent Mental Health, 13(2), 74-81. doi: 10.1111/j.1475-3588.2007.00454.x Pionek Stone, B., Kratochwill, T. R., Sladezcek, I., & Serlin, R. C. (2002). Treatment of selective mutism: A best-evidence synthesis. School Psychology Quarterly, 17(2), 168– 190. doi:10.1521/scpq.17.2.168.20857 Pumariega, A. J., Glover, S., Holzer, C. E., & Nguyen, H. (1998). Utilization of mental health services in a tri-ethnic sample of adolescents. Community Mental Health Journal, 34(2), 145-156. doi: 10.1023/A:1018788901831 Rapee, R. M., Barrett, P. M., Dadds, M. R., & Evans, L. (1994). Reliability of the DSM-III-R childhood anxiety disorders using structured interview: Interrater and parent-child agreement. Journal of the American Academy of Child and Adolescent Psychiatry, 33(7), 984-992. doi:10.1097/00004583-199409000-00008 Remschmidt, H., Mathias, P., Herpertz-Dahlmann, B., Hennighausen, K., & Gutenbrunner, C. (2001). A follow-up study of 45 patients with elective mutism. European Archives of Psychiatry and Clinical Neuroscience, 251(6), 284-296. doi: 10.1007/PL00007547  132  Richters, J. E., & Pellegrini, D. (1989). Depressed mothers’ judgments about their children: An examination of the depression-distortion hypothesis. Child Development, 60(5), 1068- 1075. Retrieved from http://www.jstor.org/action/showPublication?journalCode=childdevelopment Rohde, P., Lewinsohn, P. M., & Seeley, J. R. (1997). Comparability of telephone and face-to- face interviews in assessing axis I and II disorders. American Journal of Psychiatry, 154(11), 1593-1598. Retrieved from http://ajp.psychiatryonline.org/ Ruscio, A. M., Brown, T. A., Chiu, W. T., Sareen, J., Stein, M. B., & Kessler, R. C., (2008). Social fears and social phobia in the USA: Results from the National Comorbidity Survey Replication. Psychological Medicine, 38 (1), 15-28. doi: 10.1017/S0033291707001699. Schniering, C. A., Hudson, J. L., & Rapee, R. M. (2000). Issues in the assessment and diagnosis of anxiety disorders in children and adolescents. Clinical Psychology Review, 20(4), 453– 478. doi:10.1016/S0272-7358(99)00037-9 Shaffer, D., Fisher, P., Lucas, C. P., Dulcan, M. K., & Schwab-Stone, M.E. (2000). NIMH Diagnostic Interview Schedule for Children Version IV (NIMH DISC-IV): Description, differences from previous versions, and reliability of some common diagnoses. Journal of the American Academy of Child and Adolescent Psychiatry, 39(1), 28-38. doi: 10.1097/00004583-200001000-00014 Siegel, S. (1956). Nonparametric statistics for the behavioral sciences. New York, NY: McGraw-Hill Book Company. Sharp, W. G., Sherman, C., & Gross, A. M. (2007). Selective mutism and anxiety: A review of the current conceptualization of the disorder. Journal of Anxiety Disorders, 21(4), 568- 579. doi:10.1016/j.janxdis.2006.07.002  133  Silverman, W. K., & Albano, A. (2004). Anxiety disorder interview schedule ADIS for DSM-IV: Parent Interview Schedule. New York: Oxford University Press. Silverman, W. K., Saavedra, L. M., & Pina, A. A. (2001). Test-retest reliability of anxiety symptoms and diagnoses with the Anxiety Disorders Interview Schedule for DSM-IV: Child and Parent Versions. Journal of the American Academy of Child and Adolescent Psychiatry, 40(8), 937-944. doi:10.1097/00004583-200108000-00016 Smith, S. (2007). Making sense of multiple informants in child and adolescent psychopathology : A guide for clinicians. Journal of Psycoeducational Assessment, 25(2). 139-149. doi: 10.1177/0734282906296233 Spence, S. H. (1998). A measure of anxiety symptoms among children. Behavior Research and Therapy, 36(5), 545-566. doi:10.1016/S0005-7967(98)00034-5 Spence, S. H., Rapee, R., McDonald, C., & Ingram, M. (2001). The structure of anxiety symptoms among preschoolers. Behavior Research and Therapy, 39(11), 1293-1316. doi:10.1016/S0005-7967(00)00098-X Standart, S., & Le Couteur, A. (2003). The quiet child: a literature review of selective mutism. Child and Adolescent Mental Health, 8(4), 154-160. doi: 10.1111/1475-3588.00065 Statistics Canada website. (2003). Ethnic Diversity Survey: A Portrait of a multicultural society. Retrieved on April 27, 2011 from http://www.statcan.gc.ca/pub/89-593-x/89-593- x2003001-eng.pdfSummerfeldt, L. J., & Antony, M. M. (2004). Structured and semistructured diagnostic interviews. In M. M. Antony and D. H. Barlow (Eds.), The Handbook of Assessment and Treatment Planning for Psychology Disorders (pp. 3-37). New York, NY: Guilford Press.  134  Steinhausen, H. C., & Juzi, C. (1996). Elective mutism: An analysis of 100 cases. Journal of the American Academy of Child and Adolescent Psychiatry, 35(5), 606–614. doi: doi:10.1097/00004583-199605000-00015 Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4 th  ed.). Needham Heights, MA: Allyn & Bacon. Tannock, R., Fung, D. S., & Manassis, K. (2003). Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV). Unpublished instrument. Used with permission by authors. Toppelberg, C. O., Patton, T., Coggins, A., Lum, K., & Burger, C. (2005). Differential diagnosis of selective mutism in bilingual children. Journal of the American Academy of Child & Adolescent Psychiatry, 44(6), 592-595. doi:10.1097/01.chi.0000157549.87078.f8 Verhulst, F. C., & Akkerhuis, G. W. (2006). Agreement between parents’ and teachers’ ratings of behavioral/emotional problems of children aged 4-12. The Journal of Child Psychology and Psychiatry, 30(1), 123-136. doi: 10.1111/j.1469-7610.1989.tb00772.x Visu-Petra, L., Cheie, L., Benga, O., & Packiam Alloway, T. (2011). Effects of anxiety on memory storage and updating in young children. International Journal of Behavioral Development, 35(1), 38-47. doi: 10.1177/0165025410368945 Waddell, C. (2007). Improving the mental health of young children: A discussion paper prepared for the British Columbia Health Child Development Alliance. Retrieved from http://www.firstcallbc.org/pdfs/Communities/4-alliance.pdf Woo, B. S., Ng, T. P., Fung, D. S., Chan, Y. H., Lee, Y. P., et al. (2007). Emotional and behavioral problems in Singaporean children based on parent, teacher and child reports.  135  Singapore Medical Journal, 49(5), 1100-1106. Retrieved from http://smj.sma.org.sg/smjcurrent.html Wood, J. J., Piacentini, J. C., Bergman, R. L., McCracken, J., & Barrios, V. J. (2002). Concurrent validity of the anxiety disorders section of the Anxiety Disorders Interview Schedule for DSM-IV: Child and Parent Versions. Journal of Clinical Child and Adolescent Psychology, 31(3), 335-342. doi: 10.1207/S15374424JCCP3103_05 Yeganeh, R., Beidel, D. C., & Turner, S. M. (2006).  Selective mutism: More than social anxiety? Depression and Anxiety, 23(3), 117-123. doi: 10.1002/da.20139 Yeganeh, R., Beidel, D. C., Turner, S. M., Pina, A. A., & Siverman, W. K. (2003).  Clinical distinctions between selective mutism and social phobia: An investigation of childhood psychopathology. The American Academy of Child and Adolescent Psychiatry, 42(9), 1069-1075.  doi: 10.1097/01.CHI.0000070262.24125.23 Youngstrom, E., Loeber, R., & Stouthamer-Loeber,. M. (2000). Patterns and correlates of agreement between parent, teacher, and male adolescent ratings of externalizing and internalizing problems. Journal of Consulting and Clinical Psychology, 68(6), 1038- 1050. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed    136   Appendix A: Teacher Telephone Interview: Selective Mutism & Anxiety in the School Setting (TTI-SM-DSM-IV)   137    138    139    140    141    142    143    144    145    146    147    148   Appendix B: Revised Teacher Telephone Interview: Selective Mutism in the School Setting, Paper-pencil Version (TTI-SM-R)   149    150     151     152  Appendix C: School Anxiety Scale – Teacher Report (SAS-TR)  153   Appendix D: Selective Mutism Questionnaire (SMQ)     154    155   Appendix E: Preschool Anxiety Scale (PAS)   156  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0072062/manifest

Comment

Related Items