A PSYCHOMETRIC EVALUATION OF TWO MEASURES OF EXPRESSED IN CAREGIVERS OF CHILDREN WITH MOOD DISORDERS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the

Graduate School of The Ohio State University

By

Nicole Klaus, M.A.

* * * * *

The Ohio State University 2006

Dissertation Committee:

Dr. Mary A. Fristad, Advisor Approved by

Dr. Michael Vasey

Dr. Steven J. Beck ______Graduate Advisor in Psychology

ABSTRACT

Expressed emotion (EE) refers to criticism, , and emotional overinvolvement (EOI) displayed by family members toward a patient. EE has been associated with higher rates of relapse and poor outcome in adults with a range of psychiatric disorders. In children, caregiver EE is related to presence and course of multiple disorders. EE measures developed for adult populations have been used in studies of children. However, questions have arisen regarding their appropriateness in such applications.

The present study examined reliability, stability, and validity of two measures of

EE, the Five Minute Speech Sample (FMSS) and Expressed Emotion Adjective Checklist

(EEAC), in caregivers of children with mood disorders. During the FMSS, a relative is asked to speak freely about the patient. Audiotaped speech samples are later coded by qualified raters. The EEAC is a self-report measure listing adjectives that are rated according to their frequency of expression by the patient and relative toward each other.

Both EE measures were completed by 180 mothers and 106 fathers of children with mood disorders as part of a study examining efficacy of family psychoeducational treatments. Data were also collected on constructs theoretically related and unrelated to

EE including reports of the child’s behavior, mood, and symptoms. A subset of parents completed EE measures again six and twelve months later.

ii FMSS interrater reliability was lower in this sample than in previous studies.

Stability was greater at one-year than at six-month follow-up, perhaps reflecting seasonal variations in child mood symptoms. Little evidence was found for validity of the EOI scale. Criticism was concurrently related to child, but not parent, variables. Mothers’ criticism predicted child mood and anxiety symptoms one year later.

EEAC scores remained stable over time. Child subscale scores were most strongly related to other child variables. Parent subscale scores were more strongly related to parent variables. Mothers’ EEAC scores predicted child mood symptoms and fathers’

EEAC scores predicted child behavior problems one year later. Further, mothers’ EEAC scores moderated the effects of psychoeducational treatment. The EEAC appears to be an inexpensive and convenient measure of EE, with validity comparable to or exceeding the

FMSS.

iii ACKNOWLEDGMENTS

I wish to express my deepest appreciation to my advisor, Dr. Mary Fristad, for the

intellectual support and guidance which made this dissertation possible. Her influences in

shaping my clinical and research skills will remain with me for a lifetime. I would also like to thank my committee members who provided valuable suggestions and feedback.

I would like to think the families who participated in this research study. Their

willingness to open their lives for this research has contributed directly to this study and

has provided me with unique and rewarding clinical experiences and inspiration.

I am grateful to the dedicated research staff of the MFPG project who have been a

to work with. I also appreciate the efforts of many undergraduate research

assistants who devoted considerable time to transcribing speech samples and preparing

data for entry.

I have benefited greatly from the support and encouragement of family and

friends throughout the preparation of this dissertation. My husband, Martin Klaus, has

given me unconditional and support, and this is dedicated to him. I also wish to

thank Tena and Rod Wasinger, Crystal Wasinger, Cory Wasinger, Ruth Dysart, and

Amanda Howard for their words of encouragement and comic relief over the past year.

Finally, these data were collected with financial support from a grant awarded to

Dr. Fristad from the National Institute of Mental Health.

iv VITA

June 26, 1978…………………………….. Born- Ransom, Kansas

2000……………………………………… B.A., Psychology and Human Development, University of Kansas

2004……………………………………… M.A., Psychology The Ohio State University

2000-2005……………………………...... Graduate Teaching and Research Associate, The Ohio State University

2005-2006………………………………..Pre-doctoral Psychology Intern West Virginia University School of Medicine

PUBLICATIONS

Klaus, N. & Fristad. M. A. (2005). Family psychoeducation: An adjunctive intervention for children with . Directions in Psychiatry, 25, 217-229.

FIELDS OF STUDY

Major Field: Psychology

v TABLE OF CONTENTS

Page Abstract……………………………………………………………………………...... ii Acknowledgments………………………………………………………………………. iv Vita………………………………………………………………………………………. v List of Tables………………………………………………………………………...... ix List of Figures………………………………………………………………………….... xi

Chapters:

1. Introduction………………………………………………………………………. 1

Development of EE concept.…………………………………………….. 1 Adult studies of EE………….…………………………………………… 4 …………………………………………………..... 5 Mood disorders………………………………………………...… 6 Eating disorders………………………………………………….. 8 Anxiety disorders………………………………………………… 9 Borderline personality disorder……………………………….… 11 Alcoholism…………………………………………………….... 11 Diabetes……………………………………………………….… 11 Summary………………………………………………………... 12 EE in children………………………………………………...………… 12 Association of EE with child psychopathology………….……... 14 Longitudinal studies…………………………………….………. 16 Contribution of parent characteristics to EE status…….……….. 19 Summary……………………………………………….……….. 20 Theoretical models and treatment implications……………………….... 20 Measurement issues in child research…………………………………... 22 Psychometrics of the FMSS…………………………………………….. 24 Interrater reliability...…………………………………………… 26 Concurrent validity…………………………………………...… 27 Predictive validity………………………………………………. 28 Test-retest reliability……………………………………………. 28 Convergent validity…………………………………………….. 29 Factor structure…………………………………………………. 32 Psychometrics of the EEAC……………………………………………. 33 Concurrent validity……………………………………………... 33 Internal consistency…………………………………………….. 34

vi Convergent validity…………………………………………….. 35 Parent-child concordance………………………………………. 36 Purpose of present study and hypotheses………………………………. 36 Concurrent validity…………………………………………...… 38 Test-retest reliability…………………………………………… 38 Convergent validity…………………………………………….. 39 Discriminant validity…………………………………………… 39 Predictive validity………………………………………………. 39

2. Methods…………………………………………………………………………. 41

Participants and procedure……………………………………………… 41 Measures………………………………………………………………... 43 Demographics………………………………………………..…. 43 Five Minute Speech Sample (FMSS)…………………………... 44 Expressed Emotion Adjective Checklist (EEAC)………………. 44 Children’s Interview for Psychiatric Syndromes, Child and Parent versions (ChIPS and P-ChIPS)………………….. 44 Children’s Rating Scale, Revised (CDRS-R)…….... 45 Mania Rating Scale………………………………...…………… 45 Mood Severity Index (MSI)…………………………………….. 46 Kaufman Brief Intelligence Test (K-BIT)…………………….... 46 Home and Community Social Behavior Scales (HCSBS)……… 46 Treatment Beliefs Questionnaire- Parent Version (TBQ-P)….… 47 Social Support Scale for Children (SSS)……………………….. 47 Teacher Report Form (TRF)……………………………………. 48

3. Results…………………………………………………………………………... 49

Descriptive statistics……………………………………………………. 50 EEAC…………………………………………………………… 50 FMSS…………………………………………………………… 51 FMSS interrater reliability……………………………………………… 51 Concurrent validity……………………………………………………... 52 Categorical data………………………………………………… 52 Continuous data……………………………………………….... 52 Test-retest reliability……………………………………………………. 54 EEAC…………………………………………………………… 54 FMSS………………………………………………………….... 55 Convergent validity……………………………………………………... 56

vii EEAC…………………………………………………………… 56 FMSS…………………………………………………………… 56 Discriminant validity…………………………………………………… 57 EEAC………………………………………………………….... 59 FMSS…………………………………………………………….60 Predictive validity………………………………………………………..61 Predictions of symptoms one year later………………………….61 Treatment-related changes in EE predicting mood improvement……………………………………………. 64 Additional analyses…………………………………………………..…. 65

4. Discussion………………………………………………………………………. 67

FMSS………………………………………………………………….... 67 Reliability………………………………………………………. 67 Stability…………………………………………………………. 68 Validity…………………………………………………………. 69 Summary and conclusions……………………………………… 75 EEAC………………………………………………………………….…77 Internal consistency…………………………………………….. 77 Stability………………………………………………………..... 77 Validity…………………………………………………………. 77 Relationship to FMSS…………………………………………... 80 Summary and conclusions…………………………………….... 81 Limitations and directions for future research.…………………………. 83

References………………………………………………………………………………. 84

Appendices

A: Tables………………………………………………………………………... 96 B: Figures…………………………………………………………………..….. 123 C: Questionnaires……………………………………………………………….126

viii LIST OF TABLES

Table Page

1 EEAC Descriptives …………….………………………………………………. 97

2 FMSS Descriptives ...……………………………………………………...…… 98

3 FMSS CRIT Predictions of EEAC Scores …………... ……………………….. 99

4 EEAC Means by FMSS CRIT Status ………………………………………… 100

5 Correlations between EEAC and FMSS Continuous Scales…………………...101

6 EEAC Test-retest Reliability………………………………………………….. 102

7 FMSS Continuous Scales Test-retest Reliability……………………………… 103

8 FMSS Categorical Scales Test-retest Reliability ……………...…………..….. 104

9 EEAC Convergent Validity Correlations for Mothers ……………………..…. 105

10 EEAC Convergent Validity Correlations for Fathers ...………………………. 106

11 FMSS Convergent Validity Correlations for Mothers ………………………... 107

12 FMSS Convergent Validity Correlations for Fathers…………………………. 108

13 FMSS EOI Predictions of Cost/Benefit Treatment Beliefs for Mothers ……... 109

14 FMSS CRIT Predictions of HCSBS Scores for Mothers ………………….…. 110

15 EEAC Discriminant Validity Correlations for Mothers………………………. 111

16 EEAC Discriminant Validity Correlations for Fathers ….……………………. 112

17 FMSS Discriminant Validity Correlations for Mothers ………………………..113

18 FMSS Discriminant Validity Correlations for Fathers…………………………114

ix 19 Hierarchical Regression Predicting P-ChIPS Mood Symptoms from EEAC Child Positive Adjective Ratings at One-year Follow-up for Mothers ………………………………………………………………………...115

20 Hierarchical Regression Predicting P-ChIPS Mood Symptoms from EEAC Child Negative Adjective Ratings at One-year Follow-up for Mothers ………………………………………………………………………...116

21 Hierarchical Regression Predicting P-ChIPS Behavior Problems from EEAC Child Negative Adjective Ratings at One-year Follow-up for Fathers ……………………………………………………………………..…...117

22 Hierarchical Regression Predicting P-ChIPS Mood Symptoms from FMSS Critical Comments at One-year Follow-up for Mothers………………...118

23 Hierarchical Regression Predicting P-ChIPS Anxiety Symptoms from FMSS Critical Comments at One-year Follow-up for Mothers …………..…...119

24 FMSS CRIT Predictions of P-ChIPS Mood Symptoms at One-year Follow-up for Mothers………………………………………………………….120

25 Hierarchical Regression Predicting P-ChIPS Six-month MSI Score from Treatment X EEAC Child Negative Adjective Interaction for Mothers…….…121

26 Hierarchical Regression Predicting P-ChIPS Six-month MSI Score from Treatment X EEAC Parent Negative Adjective Interaction for Mothers...….…122

x LIST OF FIGURES

Figure Page

1 Interaction between Treatment Status and EEAC Child Negative Adjective Ratings Predicting Post-treatment Mood Symptoms for Mothers………….… 124

2 Interaction between Treatment Status and EEAC Parent Negative Adjective Ratings Predicting Post-treatment Mood Symptoms for Mothers………….… 125

xi

CHAPTER 1

INTRODUCTION

While the cause of many serious mental illnesses, such as schizophrenia and

major mood disorders, is now understood to be influenced by biological factors, the

course of these disorders is shaped by environmental influences. The influence of the family environment may be of particular importance in children who rely on parents to meet basic physical, emotional, and social needs (Asarnow, Goldstein, Tompson, &

Guthrie, 1993). The impact of parenting and family interactions have been shown to be important in the course of many childhood diagnoses, such as conduct disorder, depression, and anxiety disorders (Chiariello & Orvaschel, 1995; Reid, Patterson, &

Snyder, 2002; Siqueland, Kendall, & Steinberg, 1996). Within the past decade the concept of expressed emotion (EE) has been adapted as a measure of family emotional climate in the child psychopathology literature.

Development of EE Concept

The concept of EE originated in the schizophrenia research with the work of

Brown and colleagues (Brown, Birley, & Wing, 1972; Brown, Carstairs, & Topping,

1958; Brown, Monck, Carstairs, & Wing, 1962). They conducted a series of studies

examining the impact of social experiences on social adjustment and relapse in

psychiatric patients. Patients with schizophrenia who lived with parents, a spouse, or in a

1

hostel were more likely to relapse within the year following discharge than those living

with siblings or in lodgings (Brown et al., 1958). They hypothesized that emotional characteristics of the family environment were important contributors to relapse and developed measures to reliably and validly assess various aspects of family relationships

(Brown et al., 1962; Brown & Rutter, 1966; Rutter & Brown, 1966). They measured rater-observed as well as frequency of emotive remarks during an extensive assessment procedure involving individual interviews with one or two key relatives in addition to an observed family interaction involving the patient and his or her relatives.

The behavior of the patient toward relatives during these interactions was not found to predict outcome beyond what could be predicted based only on the relative’s behavior toward the patient, therefore, research focused on the of key relatives (Brown et al., 1962).

The initial conceptualization of EE included criticism, hostility, dissatisfaction, emotional over-involvement (EOI), and warmth (Brown & Rutter, 1966; Rutter &

Brown, 1966). Criticism was measured as a frequency count of the number of comments the family member made indicating , disapproval, or dislike of the patient’s behavior based on tone of voice and statement content. Hostility was rated when the relative rejected the patient as a person rather than just the behavior or when multiple critical comments were generated spontaneously. Dissatisfaction was a rating of general dissatisfaction with various aspects of family life, which may or may not have been critical or hostile. EOI was a rating of unusual concern for the patient or reported self- sacrificing behavior for the patient. Warmth was a global rating of the warmth toward the patient demonstrated by positive comments and tone of voice. This assessment

2

procedure, which became known as the Camberwell Family Interview (CFI), required five to six hours to administer plus additional scoring time (Brown et al., 1972; Van

Humbeeck, Van Audenhove, De Hert, Pieters, & Storms, 2002). For analyses, families were categorized as high EE (high in criticism, hostility, or EOI) or low EE (low in criticism, hostility, and EOI). High EE was found to strongly and independently predict relapse in patients with schizophrenia, beyond previous levels of impairment, with 58% of patients with high EE relatives relapsing within 9 months compared to 16% of those with low EE relatives. Further, less time spent with a high EE relative was shown to have a protective effect (Brown et al., 1972; Brown et al., 1962).

Research indicated that criticism was most predictive of long-term outcome; however EOI and hostility were also seen as important conceptual components of EE.

Dissatisfaction was only related to relapse if it was critical. Dissatisfaction which was not accompanied by negative emotion did not contribute to the prediction of relapse. Warmth was found to have a complex relationship with the other components of EE. It was positively correlated with EOI and negatively associated with criticism. Warmth without

EOI or criticism was associated with better outcome. Because of its complex associations, warmth was not included in the final index of EE. Therefore, EE became a measure of strongly expressed negative emotion consisting of criticism, hostility, and

EOI (Brown et al., 1972). Emotions expressed during the CFI were assumed to reflect a relative’s propensity to react toward the patient with criticism, hostility, or EOI; however other factors may be necessary to precipitate such reactions. The predictive utility of EE was seen as evidence that the emotions expressed during the CFI were a reflection of the relative’s interaction style with the patient (Brown et al., 1972; Vaughn & Leff, 1976a).

3

Adult Studies of EE

The lengthy interview procedure developed by Brown and colleagues was

shortened by Vaughn and Leff (1976b) to an administration time of approximately 1½

hours plus three to four hours for scoring (Van Humbeeck et al., 2002). Due to overlap

with criticism, hostility is often not examined as a separate component of EE (Vaughn &

Leff, 1976a). While the CFI remains the gold standard of EE assessment, several other

instruments, including interviews and questionnaires have been developed to further

reduce assessment time and facilitate research (Van Humbeeck et al., 2002). In the Five

Minute Speech Sample (FMSS) a relative is asked to speak uninterrupted about the

patient and his or her relationship with the patient. The tape recorded speech sample is

coded for levels of criticism and EOI (Magaña et al., 1986). The FMSS underestimates the CFI EE score in 20-30% of cases and has less predictive power than the CFI (Van

Humbeeck et al., 2002). Despite these limitations, the relatively brief amount of time

required to administer and score the FMSS has made it a commonly used tool to assess

EE (e.g., Asarnow et al., 1993; Bachmann et al., 2002; Baker, Heller, & Henker, 2000;

Hinrichsen & Pollack, 1997; Peris & Baker, 2000; Peris & Hinshaw, 2003; Stubbe,

Zahner, Goldstein, & Leckman, 1993). A variety of questionnaires that assess EE from

the relative’s or the patient’s perspective have been developed, but are less commonly

used (e.g., Friedmann & Goldstein, 1994; Gavazzi, McKenry, Jacobson, Julian, &

Lohman, 2000; Hooley & Teasdale, 1989; Van Humbeeck et al., 2002). One

questionnaire, the Expressed Emotion Adjective Checklist (EEAC), will be more

carefully examined in this study (Friedmann & Goldstein, 1993, 1994).

4

Adult studies of EE have been conducted around the world, including studies in

England, Germany, Israel, Japan, the Netherlands, and the United States (e.g., Brown et

al., 1972; Fichter, Glynn, Weyerer, Liberman, & Frick, 1997; Marom, Munitz, Jones,

Weizman, & Hermesh, 2002; Miklowitz, Goldstein, Nuechterlein, Snyder, & Mintz,

1988; Shimodera et al., 2002; Van Furth et al., 1996). Similar effects of EE have been

found across nationalities. Premorbid adjustment, severity of symptoms, and chronicity

of illness have generally not been found to influence EE status in relatives (Bachmann et al., 2002; Heikkila et al., 2002; Vaughn, 1989).

Schizophrenia. Several studies have found EE to be a robust predictor of relapse

in schizophrenia. A meta-analysis by Butzlaff and Hooley (1998) found a weighted mean

effect size of .30 from 27 studies using the shortened version of the CFI, which is a small

effect size according to Cohen’s (1988) guidelines. Similar results have been found using

the FMSS to measure EE (Marom et al., 2002). While a few studies have failed to

replicate these findings (e.g., Parker, Johnston, & Hayward, 1988; Tompson et al., 1995),

Butzlaff and Hooley (1998) calculated a fail safe N which indicated that 1246 null

findings would be required to make the findings from their meta-analysis non-significant.

Examining the components of EE in a retrospective study, Van Os and colleagues (2001)

found that high FMSS criticism was associated with higher rates of psychotic relapse, while high FMSS EOI was related to increased family involvement in treatment.

Most studies of EE in relatives of patients with schizophrenia have assessed one key relative, usually a mother or other female relative. Miklowitz and colleagues (1984), however, administered the CFI to both parents and found significantly higher rates of EE

5

in mothers than in fathers. Specifically, a high level of criticism was found 3.5 times

more often in mothers and EOI was found seven times more often in mothers than

fathers.

EE rated by the CFI has been associated with observed affective style during

family interactions. High scores on criticism were specifically associated with critical

statements and high EOI scores were specifically associated with intrusive statements to

the patient during an interaction task (Miklowitz et al., 1984). Low EE, as measured by

the FMSS, has been associated with adequate and a neutral-to-positive

family atmosphere (Hahlweg et al., 1989). In high EE families, both patients and relatives

were found to contribute to the initiation and maintenance of negative escalation patterns,

indicating that patient and relative behavior interact to produce a more emotionally

charged environment (Hahlweg et al., 1989).

Mood disorders. Based on six studies of EE in mood disorders, including both

unipolar and bipolar disorders, Butzlaff and Hooley (1998) found a weighted mean effect size ranging from .39 to .45, depending on the cutoff for scoring critical comments, which is a small to medium effect size according to Cohen’s (1988) guidelines. In the first study to explore the effects of EE in depression, Vaughn and Leff (1976a) sought to replicate Brown and colleagues’ (1972) earlier findings and determine whether the effects of EE on relapse were specific to schizophrenia. While a cutoff score of six critical comments for the schizophrenia group replicated the Brown and colleagues (1972) findings, this threshold did not discriminate relapse status in patients with depression.

However, a threshold of two critical comments was found to provide the best separation of relapse rate, with 67% of patients with high EE relatives relapsing within nine months

6

compared to 22% with low EE relatives. The authors concluded that patients with depression were more vulnerable to the effects of criticism and relapsed at a lower level

of criticism. They also noted that the hostility and EOI components of EE did not add

anything to the prediction of relapse, and were therefore not used in assigning EE status.

Subsequent studies of unipolar depression using the CFI have generally adopted a cutoff

score of two criticisms for assigning high EE status (Hayhurst, Cooper, Paykel, Vearnals,

& Ramana, 1997; Hooley, Orley, & Teasdale, 1986).

Further research on EE in unipolar depression has been equivocal (Bachmann et

al., 2002; Wearden, Tarrier, Barrowclough, Zastowny, & Rahill, 2000). Relatively few

studies have been done with patients with depression and some have not found EE to be

predictive of outcome in this population (Hayhurst et al., 1997). The effects of EE may

depend on the patient’s relationship to the caregiver (i.e., parent, sibling, child, or spouse), with EE from adult children caring for a parent being more predictive of poor outcome than EE from a spouse (Hinrichsen & Pollack, 1997). Spouses rated high on EE have been found to be more critical, disagreeable, and less accepting in their interactions with a patient with depression (Hooley, 1986). Observations of the patient’s behavior found the patient to have more neutral nonverbal behavior and less self-disclosure with a high EE spouse, rather than more negative behavior as was found in schizophrenia samples (Hahlweg et al., 1989; Hooley, 1986).

Two studies focusing specifically on bipolar disorder have found that high EE is

predictive of relapse in this population (Miklowitz et al., 1988; Priebe, Wildgrube, &

Muller-Oerlinghausen, 1989). While both studies used the CFI, the criteria for

determining EE status in these studies was different, with Miklowitz and colleagues

7

(1988) using a cutoff of six critical comments for high EE and Priebe and colleagues

(1989) using a cutoff of two critical comments. Miklowitz and colleagues (1988) found

that observed affective style (i.e., critical, -inducing, or intrusive statements) during a

family interaction was not related to EE status in a population of patients of with bipolar

disorder, however EE and affective style made independent as well as interactive

contributions of the prediction of relapse. That is, the imperfect predictive power of one

variable could be dramatically improved by considering the other variable. When a

relative had high EE or a negative affective style, 94% of patients relapsed within nine

months, compared to 17% of patients whose relatives had neither risk factor (Miklowitz et al., 1988).

Eating disorders. Based on three studies examining EE in eating disorders, which

included anorexia nervosa, bulimia nervosa, and obesity studies, the weighted mean

effect size was .51, a medium effect size (Butzlaff & Hooley, 1998; Cohen, 1988).

Results were particularly strong in the two studies investigating obesity management

(Wearden, Tarrier, Barrowclough et al., 2000). High EE in husbands (defined by three or more critical comments on the CFI) predicted relapse in women maintaining weight loss

(Fischmann-Havstad & Marston, 1984). No husbands were high in EOI in this study.

High EE was associated with 72% relapse compared to 10% relapse in patients with low

EE husbands (Fischmann-Havstad & Marston, 1984). Using a cutoff of 6 critical comments on the CFI, Flanagan and Wagner (1991) found that EE in a key relative predicted weight loss in obese patients, with 17% of patients with a high EE relative compared to 83% of those with a low EE relative losing weight on a diet within five months.

8

Studies of anorexia nervosa and bulimia nervosa have included adolescent and young adult patients (Blair, Freeman, & Cull, 1995; Le Grange, Eisler, Dare, & Hodes,

1992; Szmukler, Eisler, Russell, & Dare, 1985; Van Furth et al., 1996). Criticism and

EOI in parents have both been associated with poorer outcome in eating disorders, although studies have used different assessment criteria (Hedlund, Fichter, Quadflieg, &

Brandl, 2003; Le Grange et al., 1992; Van Furth et al., 1996). Le Grange and colleagues

(1992) used the traditional high versus low EE split, adopting a cutoff of six critical comments on the CFI. Van Furth (1996), on the other hand, treated critical comments and

EOI from the CFI as continuous variables. Blair and colleagues (1995) found that EOI was associated cross-sectionally with severity of anorexia nervosa and cystic fibrosis.

EE has also been associated with treatment dropout in patients

(Szmukler et al., 1985). Szmukler and colleagues (1985) investigated EE using the CFI in each parent and, similar to the findings in schizophrenia research, mothers were found to display higher rates of EE than fathers (Miklowitz et al., 1984). Therefore Szmukler and colleagues (1985) adopted different scoring criteria for each parent, using a cutoff of six critical comments for mothers and three critical comments for fathers. EOI did not add to the predictive power of EE and was therefore not included in interaction analyses. When mothers were high in EE, families were more likely to drop out of family treatment

(Szmukler et al., 1985).

Anxiety disorders. High EE, as measured by the CFI, has been associated with less improvement in PTSD patients (Tarrier, Sommerfield, & Pilgrim, 1999). Criticism and hostility were specifically related to outcome, accounting for approximately 20% of the variance in outcome scores (Wearden, Tarrier, & Davies, 2000).

9

Constructs related to EE, such as , criticism, and negative attributions

concerning patient’s illness, have been associated with poorer outcome in patients with

obsessive-compulsive disorder (OCD) (Steketee & Van Noppen, 2003). High EE has been associated with OCD relapse in one study (Emmelkamp, Kloek, & Blaauw, 1992).

It has also been shown that EE (measured by the Family Attitude Scale questionnaire) could be decreased by a family involvement component added to traditional exposure and response prevention treatment for OCD and that patients receiving the family treatment showed a greater symptom reduction at follow-up (Grunes, Neziroglu, & McKay, 2001).

While EOI is conceptually related to excessive accommodation of symptoms, which has been associated with poorer outcome in OCD, further research is needed to clarify how

EOI, as well as the other components of EE are associated with OCD treatment response and outcome (Steketee & Van Noppen, 2003).

In a combined sample of patients with OCD and agoraphobia, an examination of the specific components of EE measured by the CFI showed that EOI and hostility were associated with treatment dropout (Chambless & Steketee, 1999). Hostility was associated with poorer response to treatment, while criticism was associated with better response to treatment (Chambless & Steketee, 1999). The unexpected finding that critical comments predicts better treatment response was also found by Peter and Hand (1988) in patients with disorder. These authors rated husband-wife dyads as high or low EE based on CFI interviews with both patient and spouse. They hypothesized that criticism reflected general dissatisfaction which increased the patient’s to change. Further research will be important to clarify the influence of family criticism in anxiety disorders

(Wearden, Tarrier, Barrowclough et al., 2000).

10

Borderline personality disorder. Criticism and hostility have not been found to

predict one year outcome in patients with borderline personality disorder, however higher

levels of EOI were related to better outcome and less rehospitalization (Wearden, Tarrier,

Barrowclough et al., 2000). In cross sectional research, more knowledge about borderline personality disorder was associated with higher levels of CFI rated hostility toward patients (Hoffman, Buteau, Hooley, Fruzzetti, & Bruce, 2003). Further research will be necessary to clarify the role of EE in the course of borderline personality disorder

(Hoffman et al., 2003; Wearden, Tarrier, Barrowclough et al., 2000).

Alcoholism. High EE in a spouse or significant other has been associated with relapse and shorter time to relapse in patients undergoing treatment for alcohol abuse

(Fichter et al., 1997; O'Farrell, Hooley, Fals-Stewart, & Cutter, 1998). Critical comments were specifically related to increased risk of relapse, while warmth and EOI were protective against relapse (Fichter et al., 1997; O'Farrell et al., 1998). The number of critical comments on the CFI required for assignment of high EE status was different in these studies, with Fichter and colleagues (1997) establishing four critical comments as

the cutoff and O’Farrell (1998) establishing six critical comments as the cutoff.

Diabetes. Research on the effects of EE on management of diabetes in adults has

yielded contradictory results (Wearden, Tarrier, Barrowclough et al., 2000). While one

study found critical EE to be associated with poorer glucose control independent of

treatment compliance in patients involved in an intensive treatment, another study found no relationship between EE and glucose control (Wearden, Tarrier, Barrowclough et al.,

2000; Wearden, Tarrier, & Davies, 2000).

11

Summary. EE has been associated with poor outcome and relapse in a variety of adult diagnoses around the world. The criticism component has most consistently been associated with poor outcome, with the notable exception being anxiety research, where

criticism was related to better response to treatment, possibly due to the patient’s

increased desire to change when dissatisfaction and criticism was present in the family

(Chambless & Steketee, 1999; Peter & Hand, 1988). EOI has less consistently been

associated with relapse. In some disorders EOI has been associated with better outcome

(Fichter et al., 1997; Wearden, Tarrier, Barrowclough et al., 2000). Some studies reported

that EOI occurred infrequently and had little predictive power and was therefore excluded

from analyses (Fischmann-Havstad & Marston, 1984; Szmukler et al., 1985; Vaughn &

Leff, 1976a).

The measurement of EE has been inconsistent across studies. The number of

critical comments required for high EE status has not been established for specific

diagnoses and cutoff scores have frequently been determined based on predictive utility

in a specific sample. There is no clear consensus regarding appropriate cutoff scores other

than the traditional cutoff of six critical comments in schizophrenia research. While

Brown and colleagues (1962) focused assessment on one key relative, usually female,

some subsequent studies have included two relatives or have assigned a composite family

rating of EE. A few studies have also included the patient in EE assessment. The effect of

including multiple relatives or the patient in assessment remains unclear.

EE in Children

More recently, EE has been applied to child psychopathology research. The

FMSS has been the most commonly used assessment technique in caregivers of children,

12

although the scoring criteria have been modified in some studies to take developmental

issues into consideration. Questions in the child literature have arisen about the

appropriateness of applying adult measures of overprotection to child samples and whether it has the same meaning in childhood. Parental overprotective behavior that is inappropriate when directed toward an adult child may be developmentally appropriate when directed at a young child (McCarty, Lau, Valeri, & Weisz, 2004; Wamboldt,

O'Connor, Wamboldt, Gavin, & Klinnert, 2000). Further, overprotective parenting of

adult children may have harmful effects because it is clearly developmentally

inappropriate and the distinction between highly involved, supportive parenting and

overinvolvement which is excessive and harmful is less clear in reference to children

(Wamboldt et al., 2000).

Several studies have developed modified FMSS scoring criteria for child samples.

For example, St. Jonn-Seed and Weiss (2002) made several adaptations to the FMSS for

use in mothers of infants. They videotaped parents responding to five to ten minutes of open-ended questions, which were rated high or low on negative EE (hostility and criticism), positive EE (positive remarks and warmth), and overinvolved EE. Caspi and colleagues (2004) coded four variables from the FMSS in mothers of preschool children: a frequency count of positive comments; a frequency count of negative comments; a 0-5 rating of overall negativity based on the amount of negativity and whether it was directed toward the child as a person or the child’s behaviors; and a 0-5 rating of overall warmth based on tone of voice, spontaneity, and toward the child. Baker and colleagues

(2000) used the traditional FMSS EE scoring criteria and also developed expanded codes which assessed positive (warmth, enjoyment of parenting, encouragment, and

13

positive tone) and ( about what to do and concern or worry about child’s

behavior). Wamboldt and colleagues (2000) completed analyses with the traditional

FMSS criteria and also with two revised EOI categories, which separated the emotional display and overprotective EOI ratings from positive statements about the child. One study adapted the FMSS into a three minute speech sample for use with children to assess the child’s expressed emotion toward each parent (Marshall, Longwell, Goldstein, &

Swanson, 1990).

Some authors have included borderline EE ratings as a separate category in

analyses or included them with high EE (e.g., Jacobsen, Hibbs, & Ziegenhain, 2000;

McGuire & Earls, 1994; Nelson, Hammen, Brennan, & Ullman, 2003; Stubbe et al.,

1993). The scoring criteria for the FMSS allow borderline scores to be included in the

high category in situations when a relative may be reluctant to express strong attitudes of

criticism or overinvolvement and this practice is specifically encouraged for parents of

young children (Jacobsen et al., 2000).

Association of EE with child psychopathology. Cross sectional studies in both

community and clinic-referred samples have explored the relationship between parental

EE and various problems in children. High EE has been consistently related to child

psychopathology and general problems with adjustment (Asarnow, Tompson, Woo, &

Cantwell, 2001; Hibbs et al., 1991; Jacobsen et al., 2000; McCleary & Sanford, 2002;

Peris & Baker, 2000; Peris & Hinshaw, 2003; Schwartz, Dorer, Beardslee, Lavori, &

Keller, 1990; Stubbe et al., 1993; Vostanis, Nicholls, & Harrington, 1994). Additionally,

EE has been related to symptom management of diabetes (Liakopoulou et al., 2001;

Stevenson, Sensky, & Petty, 1991).

14

High EE has been related to a broad range of childhood disorders, including

attention-deficit/hyperactivity disorder (ADHD), separation anxiety, and disruptive

behavior disorders (Peris & Hinshaw, 2003; Stubbe et al., 1993). High maternal EE has

been associated with a threefold increase in risk for depressive disorders, substance

abuse, and conduct disorders beyond the increased risk associated with parental psychopathology (Schwartz et al., 1990). The association between high EE and ADHD was the same for both the inattentive and combined types and held when the effects of comorbid aggressive behavior were controlled (Peris & Hinshaw, 2003). Hibbs and colleagues (1991) found significantly higher rates of high EE in parents of children with disruptive behavior disorders and OCD (88% and 82% respectively) than in a control

group without psychopathology (41%). High EE has also been related to childhood

disorganized attachment and poorer social functioning in adolescence, both of which

place a child or adolescent at higher risk for problems with adjustment (Jacobsen et al.,

2000; McCleary & Sanford, 2002).

Several studies have addressed the issue of whether the components of EE are

associated with type of diagnosis in children. Mothers of children with depression were

rated high on critical EE more often than mothers of children with ADHD or a

community control group (49% compared to 29% and 23% respectively) (Asarnow et al.,

2001). There is significant evidence that disruptive behavior disorders are specifically

related to the criticism scale of EE (McCarty & Weisz, 2002; Peris & Baker, 2000;

Stubbe et al., 1993; Vostanis et al., 1994). EOI, on the other hand, has been specifically associated with anxiety disorders, such as separation anxiety and overanxious disorder, in children (Stubbe et al., 1993). Considering specific variables that comprise EOI, McCarty

15

and Weisz (2002) found that maternal overprotection and crying during the FMSS were

both related to higher levels of child problems, while the number of positive remarks was

related to lower levels of child externalizing behavior.

Similar to the adult research on diabetes, there is contradictory evidence regarding

the effects of parental EE on child glucose control. Stevenson and colleagues (1991) found that EOI in either parent was related to better glucose control. However, the opposite result was found by Liakopoulou and colleagues (2001) who reported that EOI was associated with poorer glucose control in children. The excessive detail item used in scoring EOI was specifically related to poorer glucose control (Liakopoulou et al., 2001).

Longitudinal studies. Given the predictive value of EE in adult studies,

longitudinal studies have examined whether parental EE predicts relapse and course of

disorders in childhood, including mood disorders, disruptive behavior disorders, ADHD,

psychosis, and diabetes.

Asarnow and colleagues (1993) found that maternal EE predicted one year

outcome in children who had been hospitalized for depression beyond the effects of

demographic variables, diagnosis type, treatment, comorbidity, and chronicity. Children

recovered significantly more often when returning to low EE homes. McCleary and

Stanford (2002) found no effect of parental overall EE, critical EE, or EOI EE on one

year persistence of major depressive disorder in their total sample of children with

adolescents. However, there was an interaction with ADHD comorbidity such that EE

only predicted mood outcome in adolescents who did not have comorbid ADHD. The

reason for this interaction is not clear and should be interpreted with caution because the

multiple analyses conducted in this study increased the possibility of type I error.

16

In a long-term naturalistic follow-up of children with bipolar disorder, a measure of maternal warmth was found to predict rates of relapse after recovery (Geller et al.,

2002; Geller, Tillman, Craney, & Bolhofner, 2004). While no specific measure of EE was included in this study, parent and child measures of maternal and paternal warmth and hostility were available. Of the many variables assessed, including baseline psychosocial functioning, medications received, and psychotherapy, low maternal warmth was the only variable found to predict relapse. This finding held at both two-year and four-year follow- up. Children with low maternal warmth were about four times more likely to relapse after recovery than children with high maternal warmth (Geller et al., 2002; Geller et al.,

2004). Similarly, In a sample of adolescents with bipolar disorder Miklowitz (2005) found that initial parental EE status predicts level of depressive symptoms nine months later, with adolescents of low EE parents having a lower symptom severity than those with high EE parents.

There is also evidence that parental EE status may affect response to treatment in adolescents with bipolar disorder. Miklowitz (2005) compared a comprehensive treatment package including family-focused treatment plus pharmacotherapy visits and medication to an enhanced care group, which also received pharmacotherapy visits and medication plus three sessions of psychoeducation and crisis management as needed. In families where parents were initially rated low in EE status, adolescents improved at a similar rate with either treatment. Among families with high parental EE, however, those assigned to family-focused treatment showed improvement similar to the low EE group, but those with the less comprehensive treatment package did not show as much improvement, with a trend toward a significant interaction effect.

17

In a community sample, maternal EE when children were in preschool predicted

third grade ADHD beyond the effects of preschool levels of behavior problems and

maternal stress (Peris & Baker, 2000). The criticism scale accounted for this finding and

explained an additional 8% of the variance in ADHD symptom scores beyond preschool

behavior problems and maternal stress.

In a sample of 565 pairs of monozygotic twins, maternal EE when the child was

five-years-old predicted antisocial behavior problems at age seven after controlling for

the effects of behavior problems at age five (Caspi et al., 2004). Within twin pairs,

children who received more negativity at age five had more externalizing behavior at age

seven. A unique strength of this study was the use of twin comparison. This methodology

helped rule out the possibility that genetic differences between children or families led to

both antisocial behavior in children and greater expressed emotion from parents. The use

of parent and teacher reports of antisocial behavior also ruled out the possibility that

higher levels of behavior problems were due to maternal reporting bias. In another community sample, Baker and colleagues (2000), however, found that maternal EE when

the child was in preschool did not predict child behavior problems in first grade but

maternal stress was a significant predictor.

Adolescents with psychotic disorders were found to relapse more often when

returning to a home with high EE (Jarbin, Grawe, & Hansson, 2000). This relationship

was only significant when borderline EE scores were included in high EE. Two years

after hospital discharge, all of the patients from high EE homes had relapsed, compared

to half of those from low EE homes.

18

Using a version of the CFI, Worall-Davies, Owens, Holland, and Haigh (2002) studied the effects of EE from each parent in glucose control in children with diabetes.

They found that mothers were rated high on EE more often than fathers (55% and 33% respectively). While no effect was found for any component of maternal EE on glucose control, a significant effect was found for fathers’ level of hostility. Fathers’ hostility was related both retrospectively and prospectively to poorer glucose control in children.

Contribution of parent characteristics to EE status. Hibbs and colleagues (1991) found that that parental psychiatric diagnosis was related to high EE status for both mothers and fathers. Several studies have shown maternal depression to be associated with high EE (Baker et al., 2000; Bolton et al., 2003; McCarty & Weisz, 2002; Nelson et al., 2003). However, some studies have failed to replicate this finding (McCleary &

Sanford, 2002; Sisson, 2005). Nelson and colleagues (2003) found that maternal depression and critical EE had independent associations with adolescent externalizing problems. Studies using both the FMSS and CFI have found support for the role of EE as a partial mediator in the relationship between maternal depression and child externalizing behavior (Bolton et al., 2003; McCarty & Weisz, 2002).

In the same sample of children with mood disorders used in the present study,

Sisson (2005) found that high FMSS status in both mothers and fathers was related to lower global functioning in the child, but was not associated with parental psychopathology. When the EEAC was used as the measure of EE, however, maternal psychopathology, Axis II symptoms, and mood symptoms were all related to self- reported negativity toward the child. In fathers, Axis II symptoms and lower knowledge about mood disorders was associated with more self-reported negativity toward the child.

19

Additionally, mothers’ reports of the child’s negativity toward her on the EEAC were

related to the child’s global functioning and severity of mood symptoms (Sisson, 2005).

Summary. In child samples, cross sectional research has shown that high EE is associated with the presence of a variety of behavioral and emotional problems (Schwartz

et al., 1990; Stubbe et al., 1993). Longitudinal research has provided some evidence that

high EE predicts worse symptoms at follow-up, controlling for initial symptom level

(Asarnow et al., 1993; Caspi et al., 2004; Jarbin et al., 2000). There is evidence that

maternal depression is related to EE and that EE partially mediates the association

between maternal depression and child externalizing behaviors (Baker et al., 2000;

Bolton et al., 2003; McCarty & Weisz, 2002; Nelson et al., 2003). There are several

conflicting findings, however, and more research is needed to clarify how parental EE is

related to childhood psychiatric disorders and health conditions.

Theoretical Models and Treatment Implications

Much of the research on EE in various patient populations has been conducted

without a clear theory indicating why EE should be related to the disorder under

investigation, how the different components of EE might affect the patient, or what the

appropriate threshold for assigning EE status should be (Wearden, Tarrier, Barrowclough

et al., 2000). While EE’s predictive utility in adults is established, the precise

mechanisms through which EE operates to increase relapse risk is not clear, and has been

the suggested focus of future research in adult patients (Butzlaff & Hooley, 1998).

Patients with schizophrenia have been found to have heightened physiological in

the presence of a high EE relative, indicating that the home environment has a direct

physiological effect on the patient (Wearden, Tarrier, Barrowclough et al., 2000).

20

Bidirectional effects are likely, with worsening symptoms increasing EE and high EE

increasing patient stress and exacerbating symptoms. It is possible that EE operates

differently in different diagnoses. For example, there is evidence that patients with

depression may be more sensitive to the effects of criticism than patients with

schizophrenia and that criticism may have a beneficial effect in some anxiety disorders

(Chambless & Steketee, 1999; Peter & Hand, 1988; Vaughn & Leff, 1976a). EOI may be

quite stressful for patients with schizophrenia, but may have a positive effect in promoting treatment adherence and protecting against relapse in patients with diabetes or alcoholism (Wearden, Tarrier, Barrowclough et al., 2000).

How EE is developed, its association with other variables, and its susceptibility to change have important treatment implications (Vaughn & Leff, 1976a). Research has begun to explore characteristics of relatives that predict EE status, such as psychopathology and cognitions (Wearden, Tarrier, Barrowclough et al., 2000). As described above, there is evidence in the child literature that parental psychopathology is associated with EE status (Hibbs et al., 1991; Sisson, 2005). High EE relatives have also been found to make certain types of attributions about the patient and the disorder.

Relatives who score high on criticism or hostility have consistently been found to attribute more control over symptoms to the patient than those who display low levels of criticism (Barrowclough & Hooley, 2003). Critical relatives also tend to attribute patient problems to internal, stable, and personal factors (Barrowclough & Hooley, 2003). Most attribution studies have been conducted in adult samples, however a recent study of children with behavior problems found that mothers high on critical EE attributed the child’s problems to factors internal, personal, and controllable to the child (Bolton et al.,

21

2003). While critical EE mothers tended to hold the child responsible for the problems,

mothers high in EOI tended to attribute control to themselves for the child’s problems.

CFI rated warmth was related to fewer controllability attributions (Bolton et al., 2003).

Based on these findings, family treatments that involve education and cognitive- behavioral techniques to decrease negative attributions, thereby decreasing EE, may help

reduce relapse (Barrowclough & Hooley, 2003). Family psychoeducational treatments for

schizophrenia, mood disorders, and eating disorders have been found to reduce EE and

improve patient outcome (Fristad, Gavazzi, & Mackinaw-Koons, 2003; Miklowitz et al.,

2000; Tarrier et al., 1988; Uehara, Kawashima, Goto, Tasaki, & Someya, 2001; Vaughn,

1989). There is also evidence to suggest that initial EE level may predict response to

treatment, with high EE status parents needing more comprehensive family treatment to

achieve the same child symptom improvement that families with low EE parents can

achieve with a less intense treatment (Miklowitz, 2005).

Measurement Issues in Child Research

Common concerns in child EE research involve the measurement of EE.

Measurement instruments, such as the FMSS and CFI, which have been developed to

study the emotional relationship between an adult patient and a relative, have been used

in research involving children and their caretakers. There are concerns regarding whether

it is appropriate to apply the same criteria in adult and child research. There are also

questions of whether the individual components of EE have the same meanings,

determinants, and effects in children as they do in adults (Wamboldt et al., 2000).

Specific concerns have been raised regarding the FMSS EOI scale. Overinvolvement and

excessive praise may be problematic with adult patients, but developmental appropriate

22

with children (McCarty et al., 2004; Wamboldt et al., 2000). The level of involvement or

praise that is excessive in children is not established. Several studies have adapted the

FMSS for use in children, as described above, but there is no single adaptation that has

been widely adopted for use in child studies, making comparisons across studies difficult.

The borderline category of the FMSS has also been used differently across studies.

The FMSS focuses on two main components of EE, criticism and EOI. However,

studies have emphasized the importance of hostility (Lenior, Dingemans, Schene, Hart, &

Linszen, 2002; Worrall-Davies et al., 2002) and warmth (Geller et al., 2004; Vostanis &

Nicholls, 1995) in the assessment of EE. Hostility is often not assessed due to its overlap

with criticism, however, Lenior and colleagues (2002) found that a significant number of

relatives expressed hostility without criticism. Different levels of warmth have

specifically been associated with conduct and emotional disorders in children, leading

Vostanis and Nicholls (1995) to recommend including a rating of warmth in the FMSS.

The high versus low categorization of EE initiated by Brown and colleagues

(1972) has been maintained through much of the subsequent research, despite the loss of

information with the transformation of continuous variables into dichotomous ratings

(Bachmann et al., 2002). Some studies have developed continuous variables from the

scoring criteria for the CFI or FMSS (Caspi et al., 2004; Lenior, Dingemans, & Linszen,

1997; Van Furth et al., 1996; Vostanis et al., 1994).

Questions have also arisen concerning whether EE is a state or a trait. In Brown

and colleagues’ (1972) original conceptualization of EE, it was seen as a stable tendency to interact with the patient in a certain way, which may be activated by various factors.

EE has generally been treated as a trait in the literature (Wearden, Tarrier, Barrowclough

23

et al., 2000). Based on their findings, Van Os and colleagues (2001) hypothesized that criticism reflects a trait associated with poor prognosis and EOI reflects a state associated with the relative’s attempts to care for the patient. Due to the changing nature of child behavior and parent-child relationships over time, EE in parents of young children is not expected to be as stable as it is in relatives of adults (Vostanis & Nicholls, 1995).

Studies have traditionally assessed EE in mothers or another female relative

(Brown et al., 1972). More recent findings have shown that the rates of EE are different in mothers and fathers and may have different correlates, emphasizing the importance of examining EE in both parents (Asarnow et al., 2001; Miklowitz et al., 1984; Worrall-

Davies et al., 2002).

The present study examines the psychometric properties of two measures of EE, the FMSS and EEAC, in a sample of children with mood disorders and their caretakers.

This study will explore the validity of the EE scales, the usefulness of dichotomous compared to continuous ratings, the stability of EE, comparisons across mothers and fathers, and the relative validity of the two measures in this sample. First, the known psychometric properties of each measure will be outlined.

Psychometrics of the FMSS

The speech sample method, originally developed by Gottschalk and Gleser

(1969), was modified by Magaña and colleagues (1986) as a measure of EE. A relative is asked to speak uninterrupted for five minutes about the patient and his or her relationship with the patient. Scoring time is approximately 10-20 minutes (Magaña et al., 1986).

Similar to the CFI, the FMSS is scored based on statement content as well as tone and emotional expression (Magaña et al., 1986; Wamboldt et al., 2000). The speech sample is

24

rated on nine variables, including a frequency count of critical comments; a positive, neutral, or negative rating of the quality of the initial statement; a positive, neutral, or negative rating of the quality of the relationship; a high or low rating of dissatisfaction; a present, borderline, or absent rating of overprotective or self-sacrificing behavior; a present or absent rating of emotional display; a present or absent rating of excessive detail; a frequency count of statements of attitude; and a frequency count of positive remarks (Lenior et al., 1997; Magaña et al., 1986). It is then scored on two dimensions: criticism and EOI. The relative is scored high on the critical dimension if one or more critical comments are made, the relationship with the patient is described as negative, or the opening statement is negative. The relative is scored high on the EOI dimension if any of the following criteria are met: 1) self sacrificing or overprotective behavior is reported, 2) emotional display (i.e., crying) during the interview, 3) two or more of the following: excessive details about the past, expressions of strong positive for the patient, or five or more positive remarks (Magaña et al., 1986; Van Humbeeck et al.,

2002). High EE status is assigned when the relative scores high on either dimension. The low EE rating is further divided into pure low or borderline EE. The relative is rated as borderline critical if only dissatisfaction is present. Borderline EOI is scored if the relative has only one of the following: excessive details about the past, expressions of strong positive feelings for the patient, or five or more positive remarks (Magaña et al.,

1986; Shimodera et al., 2002).

The criticism component of the FMSS is similar to the CFI. The EOI component of the FMSS combines concepts the CFI rates separately as EOI, warmth, and positivity

(Magaña et al., 1986; Rutter & Brown, 1966; Wamboldt et al., 2000). Psychometric

25

properties of the FMSS have been examined across cultures (American, Dutch, Canadian,

German, Japanese, and American minority populations), diagnoses (schizophrenia, mood

disorders, behavior disorders, and asthma) and patient age groups (age three to adult)

(Leeb et al., 1991; Magaña et al., 1986; Malla, Kazarian, Barnes, & Cole, 1991; McGuire

& Earls, 1994; Shimodera et al., 2002; Shimodera et al., 1999; Van Furth, Van Strien,

Van Son, & Van Engeland, 1993; Wamboldt et al., 2000).

Interrater reliability. Interrater reliability in scoring the FMSS has been shown to

be adequate to excellent (Magaña et al., 1986; Wamboldt et al., 2000). In studies of adult

patients with schizophrenia, mood disorders, and eating disorders, kappas for the overall

high or low EE categorization range from .70 to .86 (Magaña et al., 1986; Shimodera et

al., 1999; Van Furth et al., 1993). Magaña and colleagues (1986) reported interrater reliability for the critical and EOI categories separately. Kappas for the items within the critical scale (initial statement, quality of relationship, and criticisms) ranged from .36 to

.94. Kappas for the EOI scale ranged from .56 to .85.

In adolescent community and asthmatic samples, kappas have ranged from .63 to

1.0 for the critical scale and .70 to 1.0 for the EOI scale (Nelson et al., 2003; Wamboldt et al., 2000). In samples of children with mood disorders and ADHD, kappas for the overall high or low EE categorization range from .61 to .80 (Asarnow et al., 1993; Peris

& Hinshaw, 2003). In samples of children with asthma and ADHD, kappas have been reported for the separate scales and range from .64 to .78 for the critical scale and .63 to

.76 EOI scale (Peris & Hinshaw, 2003; Wamboldt et al., 2000).

Concurrent validity. The association between the FMSS and the CFI has been

examined in relatives of adults with schizophrenia and mood disorders, as well as an

26

adolescent to young adult sample of patients with eating disorders (Leeb et al., 1991;

Magaña et al., 1986; Malla et al., 1991; Shimodera et al., 2002; Shimodera et al., 1999;

Van Furth et al., 1993). Overall agreement of 73-75 % with the CFI has been found (Leeb

et al., 1991; Magaña et al., 1986). Ratings of sensitivity, or the degree to which the FMSS

correctly classifies high EE, have ranged from 48% to 65% in schizophrenia samples

(Leeb et al., 1991; Magaña et al., 1986; Malla et al., 1991; Shimodera et al., 1999).

Sensitivity in a sample was found to be 67% (Shimodera et al., 2002). The specificity, or ability to correctly classify low EE, of the FMSS is considerably better, with most ratings in patients with schizophrenia ranging from 84% to 91% (Leeb et al.,

1991; Magaña et al., 1986; Malla et al., 1991). Shimodera and colleagues (1999), however, found a lower specificity rate in Japanese patients with schizophrenia of 65%.

Specificity was found to be 97% in a mood disorder sample (Shimodera et al., 2002).

Looking at specific subtypes, Malla and colleagues (1991) found that the critical dimensions were significantly related but failed to find a significant relationship between the EOI dimensions of the two measures. In an eating disorder sample, limited overlap was found between the FMSS and CFI with the FMSS criticism variables explaining 33% of the variance in CFI critical comments and the FMSS overprotection and emotional display items accounting for only 13% of the variance in CFI EOI ratings (Van Furth et al., 1993).

The FMSS has been found to underestimate CFI rated EE in 20-30% of cases

(Van Humbeeck et al., 2002). Shimodera and colleagues (2002) found that by including

27

borderline FMSS scores in the high categorization, sensitivity could be increased to

100%, with 90% specificity. However, this study included an unusually low number (8%)

of CFI rated high EE relatives, and these results require replication.

Predictive validity. The predictive validity of the FMSS has not been well

established (Van Humbeeck et al., 2002). In adult schizophrenia research, one recent

study found that the FMSS predicts outcome and relapse (Marom et al., 2002), but others

have not found evidence of predictive validity (Nugter, Dingemans, Van der Does,

Linszen, & Gersons, 1997; Tompson et al., 1995). The FMSS has been more commonly

used in longitudinal studies with children, sometimes with adaptations, and findings have

been mixed. While Asarnow and colleagues (1993) found FMSS EE scores to predict

outcome in childhood depression, McCleary and Stanford (2002) failed to replicate this

finding. Similarly, conflicting results have been found concerning the prediction of

behavior problems (Baker et al., 2000; Caspi et al., 2004). Studies have found FMSS

scores predict ADHD and psychotic relapse in children, but these results have not yet been replicated (Jarbin et al., 2000; Peris & Baker, 2000). There is some evidence that the

predictive validity improves when borderline EE scores are included in high EE (Jarbin et al., 2000).

Test-retest reliability. Five week test-retest reliability in a sample of 28 relatives

of adults with schizophrenia is .64 (Leeb et al., 1991). Using the traditional high versus low FMSS scoring criteria in a sample of 39 mothers of minority children ages three to

14, a 22 day test-retest reliability of .25 was found, which was not significant (McGuire

& Earls, 1994). However, when borderline scores were considered as a separate category rather than included in low EE, the test-retest reliability rose to .51 and when borderline

28

relatives were included in the high EE group, test-retest reliability increased to .69, both of which were significant correlations (McGuire & Earls, 1994). These findings indicate

that borderline EE may be important to include as a separate category in studies of

children and may be more closely related to high than low EE (McGuire & Earls, 1994).

Examining longer term stability, Peris and Baker (2000) found modest, but significant stability across two years in 48 mothers of children from preschool to first grade, with 74% maintaining low EE status and 70% maintaining high EE status. The correlations of the critical and EOI components, however, were not significant from preschool to first grade (Peris & Baker, 2000). Examining the stability of EE over nine years in 120 parents of young adult patients (age 15 to 26 at study entry) with schizophrenia, using either a high-low dichotomy or a quantitative rating of EE based on the FMSS, EE was found to decrease over the first 34 months following intervention then increase to pre-intervention levels in the following years (Lenior et al., 2002). Spanish and English versions of the FMSS have been shown to have similar stability (McGuire &

Earls, 1994).

Convergent validity. Several studies have provided support for the use of EE as a

measure of a relative’s attitudes and behaviors toward the patient in the home

environment. As described above, Hahlweg and colleagues (1989) found FMSS EE status

to be related to family interaction patterns in a sample of adult patients with

schizophrenia.

Several studies have examined the relationship between high EE status and family

interactions in samples of children and adolescents and their parents. In a sample of

minority children age 3-14, McGuire and Earls (1994) reported that maternal critical EE

29

was related to reported use of physical and verbally aggressive discipline methods, providing evidence that the behavior rated by the FMSS reflects some aspects of mothers’ general behavior toward the child. They also found that these types of discipline increased from low critical EE to borderline critical EE to high critical EE, further emphasizing the potential importance of the borderline category as distinct from low EE.

When borderline scores were included in high EE, maternal EE was associated with a disorganized attachment style at age six (Jacobsen et al., 2000).

Wamboldt and colleagues (2000) examined EE and family interactions in a sample of 129 children and adolescents age 6-18, most of whom were receiving treatment for asthma. This study provided further evidence of validity for the criticism scale. Parent ratings of high critical EE were associated with observations of parental negative affect in interactions with children, poorer parent and adolescent problem solving, and poorer parent and adolescent emotional attunement. High critical EE was also associated with poorer family functioning as reported by both parent and adolescent and with behavior

problems as reported on the Child Behavior Checklist and Youth Self Report, particularly

externalizing behavior. The validity of the EOI subscale and the overall EE rating

received less support. High EOI was associated with poorer observed adolescent

interpersonal boundaries and less parent reported affective involvement with the

adolescent. Any associations between overall ratings of EE and family interactions were generally due to associations with the criticism scale.

FMSS EOI ratings can be assigned either based on emotional display and

overprotective behavior or excessive praise, which may constitute distinct categories,

particularly for children (Wamboldt et al., 2000). The findings of Lenior and colleagues

30

(Lenior et al., 1997), discussed in more detail below, indicate that ratings of emotional

display and overprotection create one unidimensional scale, while ratings of excessive

praise are more closely associated with the criticism scale. Wamboldt and colleagues

(2000) created revised FMSS variables to separately analyze these two categories of EOI

and found that emotional displays and overprotective behavior were associated with

poorer observed parental and adolescent interpersonal boundaries, parent reported

affective responsiveness, and adolescent diagnosis of depression. When number of

positive remarks was analyzed separately, it was strongly and consistently related to

better family functioning and less behavior problems in children and adolescents,

consistent with the findings of McCarty and Weisz (2002) that number of positive

remarks was related to fewer externalizing problems in children.

McCarty and colleagues (2004) developed an observational system specifically

designed to capture behaviors reflecting EOI and criticism in parents of children,

including intrusive control, less fostering of independence, affective punishment, antagonism, negativity, and . While they found evidence for the validity of the

criticism scale, EOI was not associated with any of the observed behaviors in a sample of

252 clinic-referred children ages 7-17. Additionally, parents scoring borderline critical

EE more closely resembled low critical EE than high critical EE parents in their interactions with their children.

Several cross-sectional studies have explored the relationship between parental

EE and child psychopathology, including ADHD, anxiety, behavior problems,

depression, and substance abuse. Higher rates of psychopathology have been found in

children of high EE parents compared to low EE parents. Higher rates of EE have also

31

been found in clinical groups compared to nonclinical groups (Asarnow, Tompson,

Hamilton, Goldstein, & Guthrie, 1994; Asarnow et al., 2001; Baker et al., 2000; McCarty

& Weisz, 2002; Peris & Hinshaw, 2003; Stubbe et al., 1993). When the components of

the FMSS have been analyzed separately, criticism has been associated with disruptive

behavior and EOI has been specifically associated with anxiety (Stubbe et al., 1993).

Factor structure. Lenior and colleagues (1997) explored the associations among

the nine variables rated in the FMSS to examine the justification for a global EE index and the subscales of criticism, EOI, and hostility. Using Mokken scale analysis and

principal component analysis in a sample of adolescent and young adult patients with schizophrenia, they found strong evidence for a unidimensional scale of EE reflecting criticism and satisfaction which was applicable across both mothers and fathers. This scale contained the ratings of quality of initial statement, quality of the relationship, positive remarks, dissatisfaction, excessive detail, and critical remarks, which could be combined into a 0-6 scale reflecting how many of these items were rated present or less than positive. A second scale did not meet all of the criteria to be considered an internally

consistent scale across mothers and fathers and needs further study. However, this second

scale resembles the traditional definition of EOI, consisting of the variables emotional

display and overprotective/self-sacrificing behavior.

The creation of quantitative scales to measure EE has the advantage of assessing

variability better and potentially increasing the predictive power of the instrument. The

scales created by Lenior et al., however, have not been widely adopted as an alternative

to the dichotomous ratings developed by Magaña et al. (Van Humbeeck et al., 2002). One

32

study using these scales found the criticism scale to be associated with relapse in young patients with schizophrenia, while neither the EOI scale nor the dichotomized EE ratings showed such a relationship (Lenior et al., 2002).

Psychometrics of the EEAC

The EEAC is a self report developed by Friedmann and Goldstein (1993) to directly measure relatives’ perceptions of their own feelings toward a patient. Relatives are asked to rate their behavior toward the patient over the past three months on a 1

(never) to 8 (always) Likert scale. Relatives report their behavior in terms of ten positive adjectives (loving, good-natured, friendly, devoted, easy to get along with, cooperative, considerate, clear, accepting, and active) and ten negative adjectives (rude, mean, lazy, irritable, irresponsible, hostile, deceitful, contrary, bored, and angry). Relatives are further asked to rate the patient’s behavior toward the relative over the same time period using the same adjectives.

Concurrent validity. EEAC scores were compared to FMSS and CFI ratings in relatives of adults with schizophrenia (Friedmann & Goldstein, 1993). Negative adjective scores were associated with CFI ratings. High critical EE relatives rated both themselves and the patient as more negative on the EEAC compared to high EOI EE and low EE relatives. There was also a tendency for high EOI EE relatives to rate themselves as less negative on the EEAC compared to high critical EE and low EE relatives. The positive adjective scores were not related to CFI EE status.

Both negative and positive adjective scores on the EEAC were related to FMSS

EE status. High critical EE relatives rated both themselves and the patient as more negative, and themselves as less positive on the EEAC compared to high EOI EE and

33

low EE relatives. High EOI EE relatives rated the patient as more positive than low

EE relatives, who rated the patient more positively than high critical EE relatives.

In the Friedmann and Goldstein (1993) study, the CFI was administered at hospital admission. The FMSS and EEAC were both administered five to six weeks after discharge. The EEAC was found to be more closely related to the FMSS than the CFI, but these results are difficult to interpret due to the different administration times. The largest effect size was found when EEAC scores were compared to stable patterns of EE status across the FMSS and CFI. Because the EEAC assesses behavior over the past three months it may be a more sensitive measure of consistency of affective attitudes over time

(Friedmann & Goldstein, 1993).

Internal consistency. Friedmann and Goldstein (1993) calculated alpha

coefficients for the positive and negative adjective scales for the relative’s self-ratings

and patient’s reported behavior separately in a sample of relatives of adults with schizophrenia. Alpha coefficients were high for all four scales, ranging from .877 to .942.

Examining the relationships among the four scales, they found significant correlations

between relatives’ self ratings and patients’ reported behaviors (r = .64 for negative

adjectives and r = .71 for positive adjectives). Significant negative correlations were also

found between the positive and negative adjectives (r = -.73 for self-ratings and r = -.64

for patient ratings). These high correlations suggest that the relative’s responses reflect

general attitudes about the emotional quality of their relationship with the patient

(Friedmann & Goldstein, 1993).

Cerel and Fristad (1999b) used exploratory factor analysis to examine the factor

structure of the EEAC scales in a sample of 78 children and their parents. For the four

34

parent-reported scales, nearly all of the items loaded on a single factor. The rating for

lazy was the only item which did not load onto the self-rated negative adjective scale.

In relatives of both children and adults, the four EEAC scales appear to be

internally consistent. The significant associations among the scales indicate that the

scales are not measuring distinct behaviors, but rather reflect the general emotional

relationship between the relative and patient.

Convergent validity. One study has examined the relationship between EEAC

responses and family interactions in a sample of adults with schizophrenia (Friedmann &

Goldstein, 1994). Relatives who were observed to display consistently negative affective

behavior across two interactions rated themselves as more negative on the EEAC than

those who displayed a benign affective style. Those who were inconsistently negative in

their affective behavior did not rate themselves as more negative on the EEAC. Similarly,

patients who were observed to have a consistently negative coping style during a family

interaction were rated by the relative as more negative on the EEAC. Patients who were

inconsistently negative in their behavior were not rated by relatives as more negative on the EEAC. EEAC scores were better predictors of both patient and relative behavior during an interaction than either the CFI or the FMSS measures of EE. These results indicate that this self report of EE may be a better reflection of consistent emotional

interactions within a family (Friedmann & Goldstein, 1994).

Parent reported EEAC scores were found to discriminate children with mood

disorders from a nonclinical group of children and adolescents (Fristad, Schock, Gavazzi,

& Goldberg-Arnold, 1999; Schock, Gavazzi, Fristad, & Goldberg-Arnold, 2002).

Specifically, parents reported more negative and less positive child behavior in children

35

with mood disorders than the nonclinical group. Fathers of children with mood disorders

reported that they displayed more negative behavior toward their children and a similar trend was found for mothers. There was no difference between the clinical and nonclinical groups in the amount of positive behavior parents reported showing toward their children.

Parent-child concordance. Cerel and Fristad (1999a; 1999b) administered the

EEAC to children age nine to 23 and their parents. They found moderate agreement

between parents and children concerning child behavior and low agreement concerning parent behavior. They noted that some children had difficulty understanding the items, which may have contributed to the inconsistent reports.

Purpose of Present Study and Hypotheses

This study will address several issues related to the assessment of EE in child

samples by examining psychometric properties of the FMSS and EEAC in a sample of

children with mood disorders and their caregivers. This study will examine the

concurrent validity between the two measures as well as the convergent and discriminant

validity, stability, and predictive validity of each measure. All analyses will be conducted

separately for mothers and fathers to examine whether the validity of the measures in this

population is different depending on parent gender.

The EEAC will be analyzed in terms of four subscales: positive child to parent,

negative child to parent, positive parent to child, and negative parent to child. Several methods of scoring the FMSS have been used in the child literature. In order to thoroughly examine the various possible ways that EE can be conceptualized in children,

36

separate analyses will be conducted using five FMSS scales which have been utilized in past studies:

1) high, borderline, and low categories of criticism;

2) high, borderline, and low categories of EOI;

3) a modified dichotomous EOI scale created by Wamboldt and colleagues (2000)

in which relatives are scored high EOI if they report overprotective behavior or

cry during the interview;

4) a continuous score of positive remarks; and

5) a continuous score of critical comments.

These FMSS scales will capture criticism and EOI as well as a measure of positive emotion. Criticism, which has most consistently been shown to reflect family interaction and predict outcome, will be assessed using both the traditional categorical criteria and a continuous scale to determine which assessment strategy is most reliable and valid.

Because research has suggested that the positive components of the EOI scale may be conceptually different from the emotional and self-sacrificing components (McCarty &

Weisz, 2002; Wamboldt et al., 2000), the analyses will use the traditional criteria for scoring EOI as well as a purer measure of overinvolvement and a measure of positive feelings. The borderline categories of criticism and EOI will be examined separately to explore whether they constitute distinct categories or if they are more similar to high or low ratings.

Dichotomizing EE status based on a specified cutoff score on critical comments has been a common procedure in EE research, yet the assumptions that this procedure is based on have not been explicitly tested. The FMSS scoring procedures categorize one or

37

more critical comments as high CRIT, assuming that there is a difference between caregivers who express some versus no criticism, but there is no difference between those

who make one critical comment and those who make multiple comments. Additionally,

the assumption that five or more positive statements reflect emotional overinvolvement has not been tested in children. This study will explore various patterns in the relationship

that numbers of critical comments and positive remarks have with other variables,

including linear, quadratic, and cubic relationships, to more clearly examine the

assumptions of the FMSS categorical scoring system.

Concurrent validity. Concurrent validity will be assessed by comparing the two

measures. It is hypothesized that the negative EEAC scales will have a positive

relationship with FMSS criticism, measured categorically or continuously, and a negative

relationship with FMSS positive remarks. The positive EEAC scales are hypothesized to

have a positive relationship with FMSS positive remarks and a negative relationship with

FMSS criticism, measured categorically or continuously. As these instruments theoretically measure the same construct and previous research has found that they are related in adults, effect sizes are expected to be medium to large (Cohen, 1988;

Friedmann & Goldstein, 1993). The conceptual relationship of EOI to the EEAC scales is not clear, therefore no prediction is made and analyses of EOI will be exploratory.

Test-retest reliability. Six month and one year stability of each measure will be

examined. Based on previous child and adult studies, it is hypothesized that the FMSS

will show moderate stability at six months, which will decrease at 12 months as the

38

parent-child relationship changes with development. As an adult study has shown the

EEAC to be a sensitive measure of consistent patterns of EE, it is hypothesized that

EEAC will show moderate to strong stability.

Convergent validity. It is hypothesized that measures of EE will reflect aspects of

parenting behavior and beliefs. Specifically, parent and child reported parental neglect

and use of physical punishment are hypothesized to be associated with higher FMSS

criticism, higher EEAC negativity, and lower EEAC positivity. Based on findings

regarding parental attributions, parental beliefs about the child’s disorder and treatment

are hypothesized to be associated with EE, although specific predictions are not made

about how the scales will be related to these beliefs. It is predicted that child reported

support received from parents will be associated with lower FMSS criticism, more FMSS

positive remarks, lower EEAC negativity, and higher EEAC positivity.

Because EE has previously been associated with social functioning in adolescents

(McCleary & Sanford, 2002), it is hypothesized that both measures of EE will be related

to a measure of social competence in this child sample. It is unclear which components of

EE should be most closely associated with social competence, therefore specific

predictions are not made.

Discriminant validity. Factors to which EE should have no meaningful relationship include social support the child perceives from teachers and peers and the child’s academic and intellectual functioning. The association between these variables

and both EE measures will be examined to assess discriminant validity.

Predictive validity. Longitudinal studies have suggested that EE may predict

worse psychopathology at follow up. However, few child longitudinal studies have been

39

conducted and some have produced conflicting results. No child studies have examined the predictive validity of the EEAC. This study will expand on the existing literature by investigating whether FMSS and EEAC scores predict number of behavior, mood, and anxiety symptoms at one year follow up, controlling for initial symptom level.

The predictive validity of these two measures will be further examined by testing whether EE levels are decreased by a family psychoeducational treatment and whether changes in EE are related to an improvement in mood symptoms.

40

CHAPTER 2

METHOD

Participants and Procedure

Participants for this study took part in one of two larger studies on the efficacy of psychoeducational treatments for families of children with mood disorders. Twenty families were recruited for an Individual Family Psychoeducation (IFP) study and 165 were recruited for a Multi-Family Psychoeducational Group (MFPG) study. Families were referred to these studies through a variety of sources, including pediatricians, mental health providers, social workers, libraries, workshops, and media advertising. Families who called to inquire about participation were given a brief mood screening assessment over the phone. If a mood disorder seemed probable from the screening an initial assessment was scheduled. Inclusion criteria for the studies included children age 8-11 with a study diagnosis of mood disorder and IQ greater than 70. Both study treatments were considered adjunctive and participants continued to receive any other services or

treatments available in their community throughout the study.

For study assessments, families were asked to choose one caretaker to serve as the primary informant regarding the child’s symptoms and family history. The child also reported on his or her own symptoms. For the MFPG study, a secondary caregiver, if available, also participated in a shorter assessment of EE and caretaker psychopathology.

41

Each informant was interviewed separately. The initial primary informant assessment lasted approximately 3½ hours and included assessments of the child’s developmental and mood symptom history, current psychopathology, parental EE, parental lifetime and current psychopathology, and family history. Following the initial assessment, eligible

families were randomized into either immediate treatment or a one-year waitlist

condition. Regardless of condition, follow-up assessments were conducted at six, twelve,

and eighteen months. Follow-up assessments lasted approximately 1½ hours at six and

eighteen months and 3 hours at twelve months. They assessed current and worst child

mood symptoms since the last interview, parental EE, child psychopathology (twelve

month follow-up only), parental psychopathology (twelve month follow-up only), family

history (twelve month follow-up only), and updates on medications and services

received. Following each assessment, the principal investigator and another senior staff

member who were both masked regarding treatment status carefully reviewed a summary

of assessment data and assigned a specific mood diagnosis, ratings of service utilization,

and a global functioning rating. Discussions were held concerning any disagreements

until consensus was reached.

MFPG treatment consisted of eight 1½ hour group sessions. Parents and children

met together at the beginning and end of each group to review homework assignments

and share child group content. Parents and children met separately to cover new material.

Parent group content included information about symptoms, disorders, and medication;

school issues; treatment utilization; problem solving; communication; coping strategies; and managing mood symptoms (Fristad & Goldberg-Arnold, 2003). Child group content included information about symptoms, disorders, and medication; distinguishing

42

symptoms from self; mood and anger management; problem solving; and communication

(Goldberg-Arnold & Fristad, 2003). IFP adapted the MFPG content into a 16 session treatment for individual families. Eight parent and eight child sessions were alternated to cover material similar to MFPG with the addition of a healthy habits unit for parents and children (Fristad et al., 2003).

One extra subject was enrolled in MFPG and two families from MFPG switched to IFP, resulting in a total of 184 families who participated in the initial assessment across both studies. Six-month follow-up data were available from 152 families and one-year follow-up data were available from 107 families. Children were age 8-11 (M = 10.4) at the time of the initial assessment. Children were predominantly Caucasian (92%) and male (74%). Diagnoses of bipolar spectrum disorders (75%) were more common than depressive spectrum disorders (25%). Most children had comorbid behavior disorders

(97%) or anxiety disorders (68%). A variety of family structures were represented, including married biological parents (44%), step families (19%), adoptive parents (16%) single parents (2%), and other family types (19%). All family income levels were represented and the average income was $40,000 to $59,000 per year.

Measures

A subset of the data collected from families in these larger studies is used in the current study. Brief descriptions of each measure used are presented below.

Demographics. Demographic data were gathered from the primary informant at the initial interview, including birth dates of all informants, family structure, ethnicity, parental education and occupation, and family income.

43

The Five Minute Speech Sample (FMSS). As described above, the FMSS is a brief measure of EE in which a relative is asked to speak for five minutes about the patient and how the two of them get along (Magaña et al., 1986). Speech samples were tape recorded, transcribed, and sent to a laboratory at UCLA for scoring by raters who were masked to the child’s diagnosis and the family’s treatment status. FMSS data were collected from both primary and secondary informants at all assessment times.

Expressed Emotion Adjective Checklist (EEAC). As described above, the EEAC

is a self report measure of positive and negative emotion expressed by the patient and

relative toward each other (Friedmann & Goldstein, 1993). EEAC data were collected

from both primary and secondary informants at all assessment times.

Children’s Interview for Psychiatric Syndromes- Child and Parent Versions

(ChIPS and P-ChIPS). The ChIPS and P-ChIPS are structured clinical interviews which

assess DSM-IV symptoms in children and adolescents (Weller, Weller, Rooney, &

Fristad, 1999a, 1999b). The reliability and validity of these measures have been

established in both inpatient and outpatient populations. The ChIPS and P-ChIPS were

administered to the child and parent, respectively, at the initial interview and at one-year

follow-up. Symptoms experienced in the previous two weeks were assessed for all

disorders except mood symptoms, psychosis, and elimination disorders. At the initial

interview, mood symptoms were assessed in terms of the child’s worst lifetime episode.

At one-year follow-up, mood symptoms were assessed in terms of the worst episode

since the last assessment. Psychosis and elimination disorders in the past month were

44

assessed. Psychosocial stressors were assessed for the past two weeks and over the

child’s lifetime. In the present sample, interrater reliability was excellent for the parent

(weighted kappa = .784) and child (weighted kappa = .824) versions.

Children’s Depression Rating Scale, Revised (CDRS-R). The CDRS-R is a 17-

item clinician rated scale which assesses the severity of child depressive symptoms,

including dysphoric feelings, depressed affect, weeping, , difficulty having fun,

social withdrawal, appetite disturbance, sleep disturbance, hypoactivity, tempo of

language, excessive fatigue, excessive guilt, impaired schoolwork, physical complaints,

low self-esteem, morbid ideation, and suicidal ideation (Poznanski & Mokros, 1996).

Three items are rated on a 1-5 scale and 14 items are rated on a 1-7 scale, with higher

ratings reflecting greater symptom severity. The CDRS-R has been shown to be a

reliable, valid, and sensitive measure of depressive symptoms in both inpatient and

outpatient samples (Poznanski & Mokros, 1996). CDRS-R ratings obtained from the primary informant concerning child symptoms during the two weeks prior to each

assessment were used in this study. In the present sample, interrater agreement was

substantial (weighted kappa = .685).

Mania Rating Scale (MRS). The MRS is an 11-item clinician rated scale which

assesses the severity of manic symptoms, including elevated mood, irritability, thought

content, decreased need for sleep, language/ thought disorder, increased rate or amount of

speech, increased motor activity, increased sexual , disruptive/ aggressive

behavior, appearance, and insight (Young, Biggs, Ziegler, & Meyer, 1978). Seven items

are scored on a 0-4 scale and four items are scored on a 0-8 scale, with higher ratings

reflecting greater symptom severity. The reliability and validity for adults and children

45

have been shown to be acceptable (Fristad, Weller, & Weller, 1992; Fristad, Weller, &

Weller, 1995; Young et al., 1978). MRS ratings obtained from the primary informant

concerning child symptoms during the two weeks prior to each assessment were used in

this study. In this sample, interrater agreement was substantial (weighted kappa = .712).

Mood Severity Index (MSI). A mood severity index was created for this study to combine the severity scores from the CDRS-R and MRS. Adjustments were made to account for the different lower boundary of scores, the greater number of items on the

CDRS-R, and the duplicated irritability score on the two scales. Therefore, the MSI is calculated by dividing the irritability score on each scale in half, and applying the following formula: (CDRS-R score – 17 x 11/17) + MRS score.

Kaufman Brief Intelligence Test (K-BIT). The K-BIT is a brief (15 to 30 minute)

intellectual screening test which provides nonverbal, verbal, and composite scores

(Kaufman & Kaufman, 1990). Internal consistency and stability of the K-BIT are high

and the composite score has a correlation of .80 with the Wechsler Intelligence Scale for

Children- Revised (WISC-R) full scale IQ (Kaufman & Kaufman, 1990).

Home and Community Social Behavior Scales (HCSBS). The HCSBS is a parent

self-report measure of a child’s social competence and antisocial behavior (Merrell &

Caldarella, 1999). The HCSBS has been found to distinguish between children at risk for

social competence problems and children without problems (Lund & Merrell, 2001;

Merrell & Caldarella, 1999). It has also been shown to be associated with scores on other

commonly used, theoretically related scales (Merrell, Streeter, Boelter, Caldarella, &

Gentry, 2001).

46

The Treatment Beliefs Questionnaire- Parent Version (TBQ-P). The TBQ-P is a

36-item self-report scale which assesses parental beliefs about treatment (Davidson &

Fristad, 2004). Six subscales assess beliefs specific to the child, general beliefs about

medications, seriousness of the illness, costs/benefits of treatment, knowledge of

treatment, and relationships with health care providers. Items are rated on a scale of 1

(strongly disagree) to 5 (strongly agree). Scores on the TBQ-P have been shown to have

high internal consistency and six-month stability. Correlations between subscales are low

to moderate. Preliminary analysis also provided evidence for the validity of the TBQ-P

(Davidson & Fristad, 2004).

The Social Support Scale for Children (SSS). The SSS is a child self-report

measure of perceived social support from parents, teachers, classmates, and close friends

(Harter, 1985). The subscale tapping support from parents will be used to test the

convergent validity of the EE scales in this study and the remaining subscales will be

used to test the discriminant validity of the EE scales. Internal consistency is adequate for

all four subscales and there are some data suggesting that each scale is related to

measures of similar constructs. Factor analysis has shown that a four factor solution fit

best in a middle school sample, but a three factor solution was more appropriate in an

elementary school population, where the classmate and close friend scales combined to form one factor (Harter, 1985). As the sample for the present study consists mostly of elementary school age children, the two peer scales will be combined for analyses.

47

Teacher Report Form (TRF). The TRF is a self-report measure completed by school personnel and normed for children ages five to 18 (Achenbach & Rescorla, 2001).

The academic performance scale, which will be used in this study, assesses adaptive functioning in school. This scale has been shown to have excellent stability.

48

CHAPTER 3

RESULTS

All analyses were conducted separately for mothers and fathers to determine whether the FMSS and EEAC have different psychometric properties by gender and to avoid the dependence of data resulting from two informants reporting on the same child.

Following the procedure used by Sisson (2005), all female informants were combined and are referred to as mothers and all male informants were combined and are referred to as fathers. The majority of female informants were biological mothers (86%), followed by adoptive mothers (8%), grandmothers (2%), step-mothers (2%), and other (2%). The majority of male informants were biological fathers (62%), followed by step-fathers

(21%), adoptive fathers (9%), grandfathers (3%), and other relationships (5%).

Dividing participants by gender resulted in a total of 180 mothers and 106 fathers.

Mothers included 174 primary informants (97%) and fathers included 10 primary informants (9.4%). Eight fathers did not attend the initial assessment, but completed follow-up assessments. In 17 cases, two female informants reported on the same child and the secondary informant was deleted from analyses. Follow-up data were available from 149 mothers and 80 fathers at six-months. Follow-up data were available from 105 mothers and 53 fathers at twelve-months.

49

Effect sizes were examined in all possible analyses and were evaluated according

to the small, medium, and large criteria proposed by Cohen (1988). He described small effect sizes (r ≥ .10, R2 ≥ .02, and ω2 ≥ .02) as acceptable in new research, where

measures are not well developed or uncontrollable variability makes effects more

difficult to detect. Medium effect sizes (r ≥ .243, R2 ≥ .13, and ω2 ≥ .06) are “large

enough to be visible to the naked eye” and large effect sizes (r ≥ .371, R2 ≥ .50, and ω2 ≥

.15) are “grossly perceptible” (p. 26-27).

Descriptive Statistics

EEAC. At the initial assessment, the EEAC was completed by 176 mothers and

98 fathers. Across the four subscales, 25 scores (out of 704) were missing for mothers and four scores (out of 392) were missing for fathers due to individual items left blank.

Descriptive statistics are presented for mothers and fathers in Table 1. The possible range of scores for each subscale is 10 to 80. Ceiling and floor effects were not observed in the initial or follow-up data.

All four scales had high internal consistency. For mothers, Chronbach’s alpha ranged from .833 to .885 for the scales. For fathers the two child scales and the parent positive scale had similar alphas, ranging from .862 to .890. The parent negative adjective scale had slightly lower internal consistency, with an alpha of .751.

Strong negative correlations were found between positive and negative ratings for both the child (mothers: r = -.612, p <.01 and fathers: r = -.455, p <.01) and the parent

ratings (mothers: r = -.509, p <.01 and fathers: r = -.556, p <.01). Moderate to strong

50

positive correlations were found between the parent and child scales for both positive

(mothers: r = .377, p <.01 and fathers: r = .359, p <.01) and negative (mothers: r = .393,

p <.01 and fathers: r = .287, p <.01) adjective ratings.

FMSS. At the initial assessment, 155 FMSS scores were available for mothers and

69 were available for fathers. Missing data resulted from speech samples not yet scored

(39 cases), inaudible tape recordings (6 cases), and speech samples not recorded or missing (17 cases). Five scales were created from the FMSS data. Descriptive statistics

are presented in Table 2. High scores on the modified EOI scale occurred so infrequently

that meaningful analyses with this scale were not possible. Therefore, all further FMSS analyses were conducted with the remaining four scales only.

FMSS Interrater Reliabiliy

To date, a total of 42 speech samples (7.7%) have been coded by a second rater for reliability. Examining the subscales separately, the raters agreed on 74% of the high/borderline/low CRIT ratings (weighted kappa = .524) and 69% of the high/borderline/low EOI ratings (weighted kappa = .372). According to the criteria

proposed by Landis and Koch (1977), the kappa value for CRIT reflects moderate

agreement and the kappa value for EOI reflects fair agreement between raters. For the

continuous scales of critical comments and positive remarks, raters agreed in 52% and

36% of cases, respectively. The correlation between raters on critical comments was .77,

with an average disagreement of .09 comments (weighted kappa = .626). The correlation

51

between raters on positive remarks was .63, with a mean disagreement of 1.14 remarks

(weighted kappa = .372). These kappa values reflect substantial agreement for critical

comments and fair agreement for positive remarks.

Concurrent Validity

Data from the initial assessment were used to calculate concurrent validity

between the FMSS and EEAC. For the categorically measured FMSS scales, ANOVAs

were used to determine whether FMSS EE status is associated with higher or lower

scores on the EEAC. For continuously measured FMSS scales, correlations were used to

examine the association between FMSS and EEAC scores.

Categorical data. FMSS CRIT ANOVA results are presented in Table 3 for

mothers and fathers. EEAC means for each FMSS CRIT category are presented in Table

4 for mothers and fathers. For mothers, FMSS CRIT was associated only with the

emotions reportedly expressed by the child on the EEAC, including positive and negative

adjective ratings, with small (ω2 = .048) and moderate (ω2 = .071) effect sizes,

respectively. For fathers FMSS CRIT only predicted child positive adjectives, with a

moderate effect size (ω2 = .126). For both mothers and fathers, post-hoc Tukey tests

indicated that parents high on CRIT rated the child as less positive than low or borderline

CRIT parents, with no differences between low and borderline CRIT parents.

Additionally, mothers high on CRIT rated the child as more negative than low CRIT

parents, while borderline CRIT parents were not significantly different from the low or

high CRIT parents. EOI ratings were not significantly associated with EEAC scores for mothers or fathers.

52

Continuous data. Pearson correlation coefficients for the EEAC and the

continuously rated FMSS scales are presented in Table 5. For mothers, EEAC child

negative adjective ratings were found to be positively related to the FMSS number of

criticisms and negatively related to the number of positive remarks, with medium and

small effect sizes respectively. A small negative relationship was found between criticism and child positive adjective ratings.

For fathers, FMSS number of criticisms were related to both EEAC parent adjective scales, with medium effect sizes. Specifically, more critical comments were associated with higher self-ratings of negative adjectives and lower self-ratings of

positive adjectives. Additionally, the number of criticisms were associated with lower

ratings of child positive adjectives, with a large effect size. One male outlier was

identified who made a large number of critical comments in the FMSS, yet rated all child

adjectives on the EEAC, both positive and negative, low. Consistent with his speech

sample, he rated himself high on negative adjectives and low on positive adjectives.

When data from this one participant were removed, correlations between criticism and

child EEAC scales became -.484 (p < .01) and .256 (p < .05) for positive and negative

adjectives, respectively, which are large and medium effect sizes. Positive remarks were

not associated with EEAC scores for fathers.

Further analyses were performed to examine the possibility of nonlinear

relationships between the two continuous FMSS scales and the EEAC scales. Scatterplots

of the data were visually inspected and an orthogonal polynomial comparison procedure

was used to test the fit of quadratic and cubic models. A curvilinear model was only

53

found to describe the association between FMSS number of critical comments and the

EEAC child negative adjective scale for males. However, when the outlier described

above was removed from the data, a linear model fit best.

Test-retest Reliability

Ninety-four mothers and 56 fathers were assigned to the wait list condition,

allowing their data to be used for test-retest reliability analyses. The six-month and one-

year stability of the FMSS and EEAC were evaluated. Pearson correlation coefficients

were used to assess the relationship between initial and follow-up scores. To further

evaluate the stability of scores over time, paired samples t-tests were performed on the

continuous measures to test the hypothesis that there is no change in scores from the

initial assessment to follow-up. For categorical variables, percentages of caregivers who maintained the same rating, increased, or decreased were calculated.

EEAC. Seventy-two mothers and 38 fathers completed both the initial and six- month EEAC measures. Fifty-four mothers and 25 fathers completed both the initial and twelve-month EEAC measures. Correlations and t-values are presented in Table 6.

Correlations at six-month follow-up reflect large effect sizes for both mothers and

fathers. T-tests indicate that mothers tended to rate the child as more positive and less

negative at six-month follow-up than at the initial assessment. No other differences were found between initial ratings and six-month follow-up ratings for mothers or fathers.

Correlations at twelve-month follow-up reflect large effect sizes for all scales except

mothers’ self-ratings of negative emotions, which was a medium size effect. T-tests indicate that fathers tended to rate the child as less negative at twelve-month follow-up

54

than at the initial assessment. No other differences were found between initial ratings and twelve-month follow-up ratings for mothers or fathers.

FMSS. Data were available from 46 mothers and 20 fathers at both the initial and six-month follow-up assessments. Data were available from 30 mothers and 17 fathers at both the initial and twelve-month follow-up assessments.

Data for the continuous scales are presented in Table 7. At six-month follow-up none of the correlations were statistically significant with the available N. An examination of effect size indicates small to moderate relationships may exist between initial and six-month critical comments. At twelve-month follow-up, there was a large and statistically significant correlation for critical comments for both mothers and fathers.

There was a large and statistically significant correlation for positive remarks for mothers, but not fathers. T-tests did not reveal any significant differences between initial and follow-up scores.

Data for the categorical scales are presented in Table 8. For fathers, the twelve- month correlation for CRIT was significant, with a large effect size. A closer examination of the stability shows that most fathers (59%) had the same CRIT rating one year later and a substantial portion had higher ratings (35%). No other correlations for mothers or fathers reached statistical significance with the available N; however, an examination of the effect sizes indicates that CRIT ratings may have moderate stability over time.

When the stability of CRIT and EOI ratings were examined separately at each initial level (i.e., high, borderline, and low), high CRIT ratings appeared to have the greatest stability at 12-month follow-up, with 67% of mothers and 87% of fathers who

55

scored high CRIT at the initial assessment scoring high CRIT twelve months later. For mothers, high EOI appeared to have the lowest stability, with 20% remaining high at six- months and 0% remaining high at twelve-months. Due to the very small sample sizes, these observations require further investigation.

Convergent Validity

The convergent validity of the FMSS and EEAC were calculated using initial assessment data. The items assessing parental neglect and use of physical punishment on

ChIPS and P-ChIPS were summed so that higher scores reflect a greater number of items endorsed. Creating separate scales for items measuring physical punishment and neglect yielded two highly correlated scales (r = .764, p <.001). Therefore, the composite scale was used. Pearson correlation coefficients were used for continuous scales and ANOVAs were used for categorical scales to compare EE scores to reports of parental neglect and physical punishment, SSS reports of support received from parents, HCSBS reports of social competence and antisocial behavior, and parent reported treatment beliefs.

EEAC. Convergent validity correlations are presented in Tables 9 and 10 for mothers and fathers. Both positive and negative child adjective ratings were associated with ratings of the child’s social competence and antisocial behavior for both mothers and fathers, with medium to large effect sizes. Parents who rated the child more positive and less negative on the EEAC tended to rate the child as more socially competent with fewer antisocial behaviors on the HCSBS. When parents rated the child as more antisocial, they tended to rate themselves as more negative toward the child on the

EEAC, with small to medium effect sizes. Mothers, but not fathers, who rated the child as

56

more socially competent, rated themselves as more positive and less negative toward the child on the EEAC, with small effect sizes.

For mothers and fathers, several aspects of their treatment beliefs were associated with self-ratings on the EEAC. Specifically, parents’ higher self-ratings of positive adjectives were related to beliefs that the child’s illness is more serious, the benefits of treatment outweigh the costs, more knowledge about treatments, and better relationships with health care providers. For mothers only, higher self-ratings of negative adjectives were related to negative beliefs about treatment in general, beliefs that the child’s illness is less serious, and poorer relationships with health care providers. Fathers’ higher self ratings of negative adjectives were related only to less knowledge about treatments.

Effect sizes for all of these relationships were small, with the exception of the association between fathers’ positive self-ratings and knowledge about treatment, which was a medium effect size. For fathers only, EEAC ratings of the child’s behavior (both positive and negative adjectives) were related to beliefs about the seriousness of the child’s illness and the child’s need for treatment, with moderate effect sizes.

The EEAC was unrelated to parental neglect, use of physical punishment, and support that the child reports receiving from parents.

FMSS. Convergent validity correlations are presented in Tables 11 and 12 for mothers and fathers. For mothers, more critical comments were associated with higher ratings on the antisocial behavior scale and more positive remarks were related to higher ratings on the social competence scale of the HCSBS, both of which were moderate effect sizes. For fathers, an examination of the data indicated that the outlier described

57

earlier affected several of the relationships with other variables. Therefore correlations

presented in Table 12 do not include data from this one participant. Similar to the pattern

for mothers, fathers who made more critical comments tended to rate the child higher on the antisocial behavior scale of the HCSBS. There was also a trend for HCSBS social competence ratings to be associated with fewer critical comments and more positive remarks (p ≤ .06). Effect sizes were moderate for all of these relationships. Additionally,

more critical comments in fathers were associated with a belief that the costs of treatment

outweigh the benefits, with a moderate effect size. The number of critical comments and

positive remarks were not correlated with parental neglect, use of physical punishment,

support that the child reports receiving from parents, or most aspects of treatment beliefs.

Further analyses were performed to examine the possibility of nonlinear

relationships between the two continuous FMSS scales and the theoretically related

scales under investigation. Scatterplots of the data were visually inspected and an

orthogonal polynomial comparison procedure was used to test the fit of quadratic and

cubic models. These models were not found to provide a better explanation of the data.

ANOVA’s were performed to examine differences among categorically rated

levels of CRIT and EOI in the variables under investigation. For mothers, EOI was only

found to predict beliefs about the costs and benefits of treatment. Results from this

analysis are presented in Table 13. This effect was small in magnitude (ω2 = .027) and

post hoc Tukey tests indicated that mothers with borderline EOI believed that treatment

benefits exceeded costs more than mothers with high EOI. CRIT in mothers was only

found to predict HCSBS scores. Results from these analyses are found in Table 14. The

58

effect for social competence was small in magnitude (ω2 = .032) and post hoc Tukey

tests indicate that low CRIT mothers rated the child higher in social competence than

high CRIT mothers. The effect for antisocial behavior was medium in magnitude (ω2 =

.067) and post hoc Tukey tests indicate that high CRIT mothers rated the child higher on

antisocial behaviors than low CRIT mothers. No effects were found for CRIT or EOI in

fathers.

Discriminant Validity

The discriminant validity of the FMSS and EEAC were calculated using initial

assessment data. Pearson correlation coefficients were used for continuous scales and

ANOVAs were used for categorical scales to compare EE scores to K-BIT composite

scores, teacher reported academic functioning on the TRF, and child reported support

received from teachers and peers on the SSS.

EEAC. Correlations are presented in Tables 15 and 16 for mothers and fathers.

For mothers, all EEAC scales were associated with teacher reported academic functioning and effect sizes were in the moderate range. Higher ratings on positive adjective scales and lower ratings on negative adjective scales were related to better academic functioning. For fathers, only the lower ratings on the child negative adjective scale were significantly associated with better academic performance, also with a moderate effect size. An examination of the effect sizes of nonsignificant correlations in

fathers suggests that the child and parent positive adjective scales may also have small

associations with academic performance. The direction of the relationship between

59

fathers’ self-ratings of negative adjectives and academic performance was different than for mothers, however, with lower positive adjective ratings related to better academic performance. As this result was not statistically significant for fathers, further research is needed to determine whether differences exist between parents in this relationship.

For mothers, higher scores on the two EEAC negative adjective scales were associated with lower intelligence scores on the K-BIT, with small effect sizes. An examination of the effect sizes of nonsignificant correlations in fathers suggests that higher ratings of child negative adjectives may be related to lower intelligence scores, with a small effect size. Higher self ratings of positive adjectives and lower self ratings of negative adjectives may also be related to lower intelligence scores for fathers. Again, further research is needed to explore these nonsignificant relationships.

For mothers, higher scores on the two EEAC positive adjective scales were associated with child reports of more support from teachers, with small effect sizes.

Fathers’ self-ratings of positive adjectives also had a small, but nonsignificant, positive

association with support received from teachers.

FMSS. Correlations are presented in Tables 17 and 18 for mothers and fathers.

For mothers, more critical comments were associated with lower intelligence scores on

the K-BIT and less child reported support from peers, both with small effect sizes. For

fathers, more positive remarks were associated with higher intelligence scores, with a

moderate effect size. An examination of the effect sizes of nonsignificant correlations

suggests that mothers’ positive remarks may have a small positive association with

intelligence scores. For fathers, fewer critical comments and more positive remarks may

60

have a small association with better academic performance. In addition, more positive

remarks from fathers may have a small association with increased support the child

reports receiving from teachers and friends, and more critical comments may be related to

more support from peers.

For the categorical EOI and CRIT scales, ANOVA results indicated no significant group effects for any of the theoretically unrelated measures under investigation.

Predictive Validity

Predictions of symptoms one year later. The predictive validity of the FMSS and

EEAC were first analyzed in terms of their ability to predict behavior, mood, and anxiety

symptom levels at one-year follow-up, controlling for initial symptom levels. Wait-list

families were used for these analyses. Fifty-seven mothers and 29 fathers from wait-list

families had completed ChIPS and P-ChIPS at one-year follow-up and were included in

these analyses. A dimensional scale was constructed based on ChIPS and P-ChIPS data

using the BAMO method developed by Cerel and Fristad (2001), whereby a proportion of

symptoms endorsed for each disorder is calculated and the proportions are summed

across major categories of disorders. For continuous scales, hierarchical linear regression

was used to test the effects of initial EE on symptom levels one year later, with initial

symptom level entered into the first step of the model and initial EE ratings entered in the

second step of the model. For categorical EE scales, ANOVA was used to test for the

effects of initial EE group on one-year symptom level, with initial symptom level entered

as a covariate.

For mothers, none of the four EEAC scales significantly contributed to the

61

prediction of P-ChIPS behavior (all R2 ∆ ≤ .026) or anxiety (all R2 ∆ ≤ .008) disorders at

one-year follow-up beyond the effects of initial symptom levels. Mothers’ ratings of child adjectives on the EEAC, however, predicted P-ChIPS mood symptoms at one-year follow-up. The results of these regression models are presented in Tables 19 and 20.

Mothers who rated the child higher on positive adjectives had children with fewer mood symptoms one year later, with EEAC scores contributing to 15% of the variance in mood symptoms beyond initial levels of mood symptoms. Mothers who rated the child higher on negative adjectives had children with more mood symptoms one year later, with

EEAC scores contributing to 14% of the variance in mood symptoms beyond initial levels of mood symptoms. Both of these results reflect medium effect sizes. Mothers’ self-ratings of adjectives on the EEAC were not significantly related to P-ChIPS mood symptoms at one-year follow-up (all R2 ∆ ≤ .023). Further, mothers’ ratings on the four

EEAC scales did not significantly contribute to the prediction of ChIPS behavior (all R2

∆ ≤ .005), anxiety (all R2 ∆ ≤ .046), or mood (all R2 ∆ ≤ .019) disorders at one-year

follow-up beyond the effects of initial symptom levels.

For fathers, none of the four EEAC scales significantly contributed to the prediction of P-ChIPS anxiety (all R2 ∆ ≤ .079) or mood (all R2 ∆ ≤ .107) disorders at

one-year follow-up beyond the effects of initial symptom levels. Fathers’ ratings of child negative adjectives on the EEAC, however, predicted P-ChIPS behavior disorders at one- year follow-up. Results from this regression model are presented in Table 21. Fathers who rated the child higher on negative adjectives had children with more behavior problems one year later, with EEAC scores contributing to 17% of the variance in

62

behavior problems beyond initial levels of behavior problems. This finding has a

moderate effect size. Fathers’ other three EEAC scale scores did not significantly

contribute to the prediction of behavior problems at one year follow-up (R2 ∆ ≤ .060).

Further, fathers’ ratings on the four EEAC scales did not significantly contribute to the

prediction of ChIPS behavior (all R2 ∆ ≤ .013), anxiety (all R2 ∆ ≤ .012), or mood (all R2

∆ ≤ .010) disorders at one-year follow-up beyond the effects of initial symptom levels.

For mothers, FMSS number of critical comments made a significant contribution to the prediction of P-ChIPS mood and anxiety disorders. Results from these regression models are presented in Tables 22 and 23. More criticisms predicted higher levels of mood and anxiety symptoms at one-year follow-up, beyond the effects of initial symptom levels, accounting for 12% and 14%, respectively, of the variance in one-year symptom levels. Effect sizes for these results are small and medium. The number of critical comments did not significantly contribute to the prediction of behavior disorders (R2

∆ = .003). The number of positive remarks made by mothers did not significantly

contribute to the prediction of P-ChIPS behavior (R2 ∆ = .008), anxiety (R2 ∆ = .000), or

mood (R2 ∆ = .067) disorders. Neither the number of critical comments nor positive remarks made significant contributions to the prediction of one-year ChIPS symptoms

beyond initial symptom levels (all R2 ∆ ≤ .013). Additionally, scatterplots of the data

were visually inspected and an orthogonal polynomial comparison procedure was used to

test the fit of quadratic and cubic models. No evidence was found for nonlinear effects.

ANOVA results indicate that categorically rated CRIT was associated with P-ChIPS

mood symptoms beyond initial symptom levels for mothers. Results from this analysis

63

are presented in Table 24. When mothers were high on CRIT, 41% of children had more

mood symptoms at one-year follow-up than they did initially, compared to 17% with borderline and 13% with low CRIT mothers. Mothers’ levels of CRIT and EOI did not significantly predict any other P-ChIPS (all F ≤ 1.548, p ≥ .215) or ChIPS (all F ≤ 2.002,

p ≥ .146) disorders.

For fathers, FMSS number of critical comments made a significant contribution to

the prediction of P-ChIPS anxiety disorders (R2 ∆ = .112). A closer examination of the anxiety data, however, shows that the effect was accounted for by one father who made five critical comments (the highest in this subsample of 23 fathers). When data from this participant were removed, the change in R2 was reduced to .004, which was

nonsignificant. The number of critical comments did not significantly contribute to the

prediction of behavior (R2 ∆ = .068) or mood (R2 ∆ = .114) disorders. The number of positive remarks made by fathers did not significantly contribute to the prediction of P-

ChIPS behavior (R2 ∆ = .008), anxiety (R2 ∆ = .001), mood (R2 ∆ = .120), or other (R2

∆ = .060) disorders. Neither the number of critical comments nor positive remarks made significant contributions to the prediction of one-year ChIPS symptoms beyond initial symptom levels (all R2 ∆ ≤ .042). Additionally, scatterplots of the data were visually

inspected and an orthogonal polynomial comparison procedure was used to test the fit of

quadratic and cubic models. No evidence was found for nonlinear effects. ANOVAs

examining the contribution of categorically rated CRIT and EOI did not find any

significant effects beyond initial symptom level for P-ChIPS (all F ≤ 2.777, p ≥ .086) or

ChIPS (all F ≤ .691, p ≥ .513) disorders.

64

Treatment-related changes in EE predicting mood improvement. The predictive validity of the FMSS and EEAC were next tested by examining whether change in EE mediates symptom improvement following psychoeducational treatment. According to the guidelines suggested by Kraemer and colleagues (2002), to establish mediation, it must first be shown that changes in EE correlate with treatment status. Then it must be shown that changes in EE have a main or interactive effect on child symptom level, as measured by parent current MSI scores.

Change scores were calculated between initial and six-month values for all EE scales. Families were considered to have completed treatment if they attended at least

75% of treatment sessions (6 out of 8). Thirteen mothers and seven fathers assigned to immediate treatment did not complete treatment and were deleted from this analysis.

ANOVA’s were performed to determine whether treatment status (immediate treatment vs. wait-list) was associated with changes in EE. No significant results were found for mothers (all F ≤ 2.097, p ≥ .150) or fathers (all F ≤ 1.825, p ≥ .185). Because treatment was not found to impact EE scores, no further analyses were performed.

Additional Analyses

A recent study has suggested that parental EE status may moderate treatment response (Miklowitz, 2005). Additional analyses were performed to determine whether the EE scales under investigation may moderate the association between treatment status and mood improvement in this sample. Moderators are baseline characteristics which interact with treatment in predicting outcome (Kraemer et al., 2002). To test the moderational model, hierarchical regression was used. Initial mood symptom level,

65

treatment status (immediate vs. baseline), and initial EE status were entered in the first

step of the model. The treatment X initial EE interaction term was entered in the second step of the model. Families assigned to immediate treatment who did not complete treatment were deleted from this analysis. Standardized scores were used for all variables

and interaction terms to obtain appropriate regression coefficients and facilitate

interpretation (Aiken & West, 1991).

The interaction term was not significant for any of the FMSS scales for mothers

(all R2 ∆ ≤ .011) or fathers (all R2 ∆ ≤ .024). However, both of the EEAC negative

adjective scales interacted with treatment status for mothers. Results of these regression

models are presented in Tables 25 and 26 for adjectives expressed by the child and the parent. The child negative adjective interaction term accounted for an additional 3% of

the variance in mood scores and the parent negative adjective interaction term accounted for an additional 4% of the variance in mood scores, which are small effect sizes. Figures

1 and 2 show that there was a greater effect of treatment among parents who rated either

their child or themselves high on negative adjectives. The interaction terms for the

positive EEAC scales were not significant for mothers (both R2 ∆ ≤ .000) and the

interaction term was not significant for any of the EEAC scales in fathers (all R2

∆ ≤ .022).

66

CHAPTER 4

DISCUSSION

The goal of this study was to evaluate the psychometric properties of the FMSS and EEAC in caregivers of children with mood disorders and to address some of the unresolved issues regarding measurement of EE in child samples. In particular, this study sought to examine how the FMSS and EEAC relate both to each other and other constructs, considering the contributions of parent and child characteristics. The validity of the categorical scoring method of the FMSS and the role of borderline ratings were assessed. Additionally, differences between mothers and fathers were examined in all analyses.

FMSS

Reliability. A concern with the FMSS data in this sample is the relatively low interrater reliability. Agreement was higher for the categorical and continuous criticism scales, falling in the range of moderate to substantial agreement, than for the EOI and positive remarks scales, which had fair agreement between raters. Kappas were all below values obtained in other child and adult samples (Magaña et al., 1986; Nelson et al.,

2003; Peris & Hinshaw, 2003; Wamboldt et al., 2000).

One additional rater was used to calculate interrater reliability. It is possible that rater drift occurred with only one of the two raters, contributing to the lower kappa

67

values. Ratings from a third rater will need to be obtained to more accurately determine

whether speech samples can be reliably coded in this sample. The higher kappa values

obtained in previous studies, some of which were rated by the primary rater in the present

study, suggest that the FMSS can be reliably coded in a variety of samples (e.g., Asarnow

et al., 1993; Bullock, Bank, & Burraston, 2002; Peris & Hinshaw, 2003; Wamboldt et al.,

2000). Further analyses with the FMSS in the current sample must be interpreted

cautiously, particularly those involving the EOI and positive remarks scales, as failure to

find evidence for validity may reflect unreliable coding.

Stability. Results did not support the hypothesis that the FMSS would show moderate stability at six-months, which would decrease at twelve-months as the parent- child relationship changed. Unexpectedly, stability was higher at twelve-months than at six-months for all scales, except the EOI scale in fathers, which had low stability at both times. The twelve-month stability of criticism was particularly strong. Effect sizes were large for fathers when criticism was examined either categorically or continuously. In mothers, categorical criticism had a medium effect size and continuous criticism had a large effect size. At six-months, however, effect sizes for the stability of criticism were small to medium. A possible explanation for this finding is that there are seasonal patterns to FMSS criticism. Twelve-month follow-up assessments occur at the same time of year as the initial assessments, while six-month follow-up assessments occur during the opposite season. Using the same sample, Sisson (2005) found that higher mood severity was related to high EE status in parents. Additionally, seasonal patterns have been found in child and adolescent mood symptoms (Smith, 2005). The lower stability of

68

FMSS criticism may be a reflection of changes in the child’s mood symptoms during a

different season and the higher stability at one year may reflect a more similar mood severity in the child during the same season.

Overall, there was very little stability in EOI. Positive remarks had a large association with initial ratings only at twelve-month follow-up, and only for mothers.

This finding may again reflect seasonal changes in the child’s mood.

Previous studies in children have found modest stability in overall FMSS ratings over time, however, when the critical and EOI components have been examined separately, correlations across time were not significant (McGuire & Earls, 1994; Peris &

Baker, 2000). It is also important to note that previous studies of the stability FMSS EE status have included community samples, children with behavior problems, or adults with schizophrenia (Leeb et al., 1991; Lenior et al., 2002; McGuire & Earls, 1994; Peris &

Baker, 2000). Seasonal patterns have not been found in overall psychopathology and appear to be unique to mood disorders (Smith, 2005). As the FMSS has been used extensively in mood disorder research in children and adults (Asarnow et al., 1993;

Asarnow et al., 2001; Butzlaff & Hooley, 1998; McCleary & Sanford, 2002), the present results highlight the importance of examining the stability and other psychometric properties that may be specific to samples with mood disorders.

Validity. As evidence for validity, it was hypothesized that criticism, measured

both categorically and continuously, would be associated with parental neglect and use of

physical punishment and less child-reported support from parents. Positive remarks were

hypothesized to be related to more child-reported support from parents. Parental

treatment beliefs and child social competence were hypothesized to be related to EE,

69

although specific predictions were not made about the particular scales that would be involved in these relationships. It was hypothesized that the FMSS scales would not be associated with social support that the child perceives from teachers or peers or with the child’s academic and intellectual functioning. Additionally, the FMSS scales were hypothesized to play a role in predicting behavior, anxiety, and mood symptoms one year later beyond the effects of initial symptom level. Further, treatment related changes in EE were hypothesized to mediate improvement in mood symptoms with treatment.

These hypotheses received mixed support. The most robust findings involved the associations between parental criticism and social competence. The social competence measure used in this study, the HCSBS, included two scales: social competence and antisocial behavior. Both mothers and fathers who made more critical comments rated the child higher on antisocial behavior, with medium effect sizes. For fathers, there was also a medium-sized, but nonsignificant, trend for more critical comments to be related to lower scores on the social competence scale. For mothers only, the categorical criticism scale had a medium effect size in predicting antisocial behavior and a small effect size in predicting social competence. Mothers rated high on criticism were more likely to rate the child higher on antisocial behaviors and lower on social competence than mothers low on criticism. Mothers with borderline criticism ratings were not significantly different from those with high or low criticism in their HCSBS ratings. Additionally, more positive remarks were related to higher ratings of social competence, with medium effect sizes, although this effect did not reach significance for fathers.

The social competence results replicate and extend the results obtained by

McCleary and Sanford (2002) who found EE status to be related to social competence in

70

adolescents. The current findings indicate that this relationship holds using a different

measure of social competence and a child sample. While McCleary and Sanford (2002)

used the overall EE status in their analyses, the present findings suggest that the criticism

component is specifically related to poorer social competence. EOI was not related to social competence, but more positive remarks were associated with higher social competence.

The study hypothesis concerning parent treatment beliefs was partially supported.

Effects were only found for beliefs concerning the costs and benefits of treatment. In

fathers only, more critical comments had a medium association with beliefs that the costs

outweigh the benefits of treatment. This finding was in the predicted direction, with more

criticism being associated with more negative beliefs. The corresponding correlation in

mothers did not approach significance. Mothers rated high on EOI tended to believe that

costs outweighed the benefits of treatment more than those rated borderline EOI, with

EOI status accounting for approximately 3% of the variance in this belief. Unexpectedly,

low EOI did not differ significantly from borderline or high scores in this belief. This

may be because borderline EOI can be assigned based on a large number of positive

remarks (≥ 5), while high EOI ratings require additional strong feelings be expressed,

which may include overprotective behavior or crying during the interview. Parents who

express high EOI may be particularly overwhelmed by the child’s treatment. The positive

feelings comprising borderline EOI, on the other hand, may serve a protective function,

which translates into more positive beliefs about treatment. The lack of any strong

71

emotion in low EOI may not dispose the parent toward or protect against negative beliefs.

The variance accounted for by EOI was quite small, however, and further research will be needed to more carefully explore this relationship.

The expected associations between the FMSS and other parental treatment beliefs, child-reported support from parents, and parental neglect and use of physical punishment were not found. The constructs to which FMSS EE was not found to be related tended to reflect parent behavior and beliefs. The most robust association was found with social competence, a child behavior. These findings are consistent with the findings of Sisson

(2005) that the FMSS was associated with child mood symptom level and global functioning, but not to any measure of parental psychopathology in this sample. These findings are not consistent, however, with previous findings that critical EE is related to parent behaviors and beliefs, including use of verbally and physically aggressive discipline (McGuire & Earls, 1994), more displays of negative affect, poorer problem solving, less parent-child emotional attunement, and poorer family functioning

(Wamboldt et al., 2000).

Several small to medium sized relationships were found between the FMSS continuous scales and the theoretically unrelated measures. Higher intellectual functioning in the child was related to fewer critical comments from mothers and more positive remarks from fathers. For mothers only, more critical comments were associated with less support from peers. These findings again suggest that the FMSS is related to characteristics of the child. Nonsignificant effect sizes were also closely examined in this analysis to help rule out the possibility that low power supported discriminant validity.

These small, but nonsignificant, effects further suggest that the FMSS continuous scales

72

may measure child characteristics, including academic performance and support from teachers and peers. The small associations with social support may reflect the

associations that the FMSS was found to have with social competence.

Evidence for the predictive validity of FMSS criticism was found for mothers

only in predicting parent rated symptoms levels. Continuously measured criticism

accounted for 12% of the variance in mood symptoms and 14% of the variance in anxiety symptoms one year later after accounting for initial symptom levels. Considering categorically measured criticism, more than twice as many children with mothers rated high on criticism had more mood symptoms at one-year (41%) than children with mothers rated borderline or low on criticism (17% and 13%). Positive remarks and EOI were not found to significantly predict child symptom levels.

The findings regarding the utility of criticism in predicting mood symptom levels is consistent with previous research in children and adults (Asarnow et al., 1993; Butzlaff

& Hooley, 1998). While Geller and colleagues (2002; 2004) found evidence that maternal warmth affects the course of bipolar disorder in children, there were surprisingly no significant effects of positive remarks in this sample with a large portion of bipolar children.

No previous studies have examined the effects of EE on the course of anxiety disorders in children. The current findings help to extend adult research in EE and anxiety disorders to a child sample. Adult research has provided contradictory findings regarding the effects of criticism on anxiety disorders (Chambless & Steketee, 1999;

Peter & Hand, 1988; Steketee & Van Noppen, 2003; Wearden, Tarrier, Barrowclough et al., 2000). Two studies have found that criticism was related to better treatment response

73

in adults (Chambless & Steketee, 1999; Peter & Hand, 1988). The present study

examined the longitudinal course of anxiety symptoms while children received treatment

as usual in the community, finding that more criticism is associated with worsening

symptoms over time.

The lack of support for the hypothesis that the FMSS would predict behavior

disorders is inconsistent with previous child studies finding that criticism is associated with worsening ADHD and disruptive behavior symptoms (Caspi et al., 2004; Peris &

Baker, 2000). Both of the previous studies used community samples. It is possible that criticism has an effect on the levels of behavior problems seen in the community but not the more severe problems seen in clinical samples, such as the present one.

None of the predictive effects of criticism in mothers were replicated in fathers.

One explanation for the lack of effects is that symptom levels were reported by the primary informant, most of whom were mothers. It is also possible that the lower sample size for fathers lowered the power to detect similar effect sizes in fathers. An examination of effect sizes indicated that this may be the case. The number of critical comments and positive remarks in fathers accounted for 11% and 12% of the variance, respectively, in mood symptoms, which are small effect sizes and did not reach statistical significance.

The sizes of these effects however are similar to those seen in mothers and further

research will be needed with a larger sample of fathers to determine how paternal EE affects child mood symptoms.

74

Results did not support the hypothesis that changes in EE would mediate mood symptom improvement following treatment, as treatment was not found to lead to

changes in EE. Additional analyses indicate that the FMSS scales also do not moderate the effects of treatment.

Summary and conclusions. As with previous studies, the present study found

more evidence for the validity of the criticism scale than the EOI scale. High scores on

EOI were relatively uncommon (11% of mothers and 15% of fathers). No evidence was

found for the stability of predictive validity of EOI status. Regarding convergent validity,

EOI in mothers had a small relationship with beliefs about costs and benefits of

treatment, but with unexpected differences between high, borderline, and low scores. EOI

was not related to any of the other constructs to which it was compared. The general failure to find evidence for the validity of EOI is consistent with previous research

(McCarty et al., 2004; Wamboldt et al., 2000), however it is also possible that failure to

find significant effects in the present study may be due to the low interrater reliability for this scale. Attempts to examine a purer version of EOI consisting of emotional displays and overprotective behavior were not possible due to very low rates of occurrence.

The continuous positive remarks scale was related to the child’s social competence and intellectual functioning. Additionally, positive remarks were found to have seasonal patterns in stability for mothers only. Effect sizes suggest that positive remarks from fathers may predict lower levels of future mood symptoms, but more research is needed to support this result. Positive remarks were not related to any of the other constructs to which they were compared. Again, the general lack of evidence for validity may be due to low interrater reliability.

75

Both categorical and continuous criticism were found to have seasonal patterns in

stability, possibly due to an association with seasonally fluctuating mood symptoms in

the child. Criticism was found to be related to child characteristics, including social competence, intellectual functioning, and lack of peer support. Less support was found for relationships with parent characteristics; however there was evidence that criticism was related to fathers’ beliefs about the costs and benefits of treatment. Findings also indicated that criticism has validity in predicting the course of anxiety and mood treatment, particularly the continuous measure of criticism. No results were found for the categorical measure of criticism that were not also detected with the continuous count of critical comments. Several results, however, were found with continuous criticism that were not significant with the categorical scale. When the categorical scale did detect differences, the borderline rating was either not different from high or low ratings (for social competence ratings) or was more closely related to low ratings (in predicting one- year mood symptoms and EEAC ratings). No evidence was found to support the inclusion of borderline ratings with high ratings.

The continuous scales of criticism and positive remarks were examined for evidence of nonlinear relationships with other variables, which might support the practice of using cutoff scores in assigning categorical EE status. No evidence was found for nonlinear relationships with either scale.

Mothers and fathers were examined separately in all analyses to explore potential differences in psychometric properties depending on parent gender. While mothers and fathers had similar rates of low criticism, mothers had higher rates of high criticism ratings and fathers had higher rates of borderline criticism ratings. Similarly, fathers

76

made slightly fewer critical comments, on average, than mothers. Mothers and fathers had similar rates of EOI ratings and numbers of positive remarks. Many of the findings were consistent across mothers and fathers, including seasonal patterns in the stability of criticism, moderate associations between low social competence and criticism, and the effects of criticism in predicting mood symptom levels. Occasional differences were found between mothers and fathers, such as the effects of criticism in predicting anxiety symptoms and the associations of the criticism and EOI scales with beliefs about costs and benefits of treatment. Further research will be needed to explore the reasons for these

differences.

EEAC

Internal consistency. The four EEAC scales had high internal consistency in this

sample. The scales were also found to be related across child and self ratings and across

positive and negative adjective ratings. These results are similar to findings in adults and

indicate that the EEAC is measuring general attitudes about the parent-child relationship

(Friedmann & Goldstein, 1993).

Stability. Six-month EEAC ratings were significantly related to initial ratings for

all scales. There was a tendency, however, for mothers to rate the child more positive and

less negative at six-months. Compared to the six-month stability, ratings at twelve-

months were associated as closely or more closely with initial ratings. Fathers tended to

rate the child less negative at twelve-months than at the initial assessment. These findings

support the hypothesis that the EEAC is a largely stable measure of the parent-child

relationship over time. Slightly less stability was found in ratings of child adjectives than

in parent adjective self-ratings.

77

Validity. As evidence for validity, it was hypothesized that EEAC scales would be related to child social competence and parent beliefs about treatment. There was considerable support for these hypotheses in both mothers and fathers. The child adjective scales were specifically related to the social competence scales, with medium to large effect sizes. Parent adjective scales were also related to the social competence scales, although with smaller effects. The parent adjective scales were specifically related to several aspects of parent treatment beliefs, with small to medium effect sizes. For fathers only, child adjective ratings were associated with beliefs about the seriousness of the child’s illness and the child’s need for treatment. These findings suggest that parent and child adjective ratings are most strongly related to behavior of the individual who is being rated, which is consistent with the findings of Sisson (2005) that parent adjective ratings were associated with parent characteristics and symptom levels while child adjective ratings were associated with the child’s mood symptom severity and global functioning in this sample.

The hypotheses that higher scores on the negative EEAC scales and lower scores on the positive EEAC scales would be associated with more parental neglect and use of physical punishment and less child-reported support received from parents were not supported in this sample. It was further hypothesized that the EEAC scales would not be associated with social support that the child perceives from teachers or peers or with the child’s academic and intellectual functioning, which received mixed support. There were several indications that the child’s intellectual functioning and academic performance are related to the EEAC scales. For mothers, higher ratings on the negative scales were related to lower intellectual functioning and poorer academic performance and higher

78

ratings on the positive scales were related to better academic performance. Higher child

positive adjective ratings were also related to more child perceptions of support from

teachers. For fathers, only child negative adjective ratings were related to poorer

academic functioning. Nonsignificant effect sizes were also closely examined in this

analysis to help rule out the possibility that low power supported discriminant validity.

Examination of effect sizes suggest that fathers’ higher self-ratings of positive adjectives

may be related to lower intellectual functioning and poorer academic performance, which

is the opposite of the significant effects in mothers. Further research will be needed to determine whether these nonsignificant effects reflect a difference between mothers and

fathers.

To further establish validity, the EEAC scales were hypothesized to play a role in

predicting behavior, anxiety, and mood symptoms one year later beyond the effects of

initial symptom level. Evidence for the predictive validity of the EEAC indicates that

child adjective scales contribute to the prediction of parent rated symptom levels;

however differences emerged between mothers and fathers in the types of symptoms that

could be predicted by the EEAC. For mothers, lower child positive adjective ratings and

higher child negative adjective ratings predicted worse mood symptoms, accounting for

15% and 14%, respectively, of the variance in mood symptom levels one year later. For

fathers, higher ratings on child negative adjectives predicted more behavior problems one year later, accounting for 17% of the variance. It is possible that the lower sample size for

fathers lowered the power to detect additional effects that were similar in magnitude of

79

those found in mothers. An examination of effect sizes indicated that this may be the

case. Child positive adjective ratings in fathers accounted for 11% of the variance in mood symptoms, but did not reach significance.

Treatment related changes in EE were further hypothesized to mediate improvement in mood symptoms with treatment. This hypothesis was not supported; however additional analyses indicate that negative adjective ratings in mothers interact

with treatment effects. Specifically mothers’ high initial ratings of child and parent negative adjectives predicted greater mood symptom improvement following treatment.

This interaction made a small (3-4%) contribution to the variance in mood symptoms.

Relationship to FMSS. As hypothesized, FMSS criticism was significantly related

to the EEAC scales. Concurrent validity data show that FMSS high ratings of

categorically measured criticism were associated with EEAC lower ratings for child

positive adjectives in both parents and higher ratings for child negative adjectives in

mothers only. Higher numbers of critical comments on the FMSS were associated with

lower child positive adjective ratings for both parents, higher child negative ratings in

mothers only, and both parent scales for fathers only. The finding that FMSS criticism

was most consistently related to EEAC child adjective scales provides further evidence

that the FMSS is more closely related to child than parent characteristics.

Positive remarks were hypothesized to have a positive relationship with the

EEAC positive scales and a negative relationship with the EEAC negative scales. This

hypothesis was not supported. In addition, no association was found between EOI and the

80

EEAC scales. The low interrater reliability for the EOI and positive remarks scales of the

FMSS may account for the failure to find relationships with the EEAC for these two variables.

Summary and conclusions. The EEAC has not been widely used as a measure of

EE and there are currently no published studies using this measure with a child sample.

The current findings provide important and previously unavailable information regarding

the utility of this measure in child EE research. The EEAC was an internally consistent

and stable measure of general impressions of the parent-child relationship in this sample.

Convergent and divergent validity results were similar to those found for FMSS criticism.

Possible differences between mothers and fathers in the association between parent adjectives and child intellectual functioning and academic performance should be confirmed and examined in future studies. Overall, results for mothers and fathers were

very similar. The predictive validity of the EEAC had not previously been examined in adult or child samples (Van Humbeeck et al., 2002). For mothers and possibly for fathers, the child scales were found to be influential in predicting child mood symptoms.

Additionally, for fathers only, the child negative adjective ratings were found to predict

behavior problems.

The EEAC scales were found to be related to FMSS criticism, which has been

found to have the greatest validity of the FMSS scales in this and other studies. In

addition to showing patterns of validity similar to FMSS criticism, the EEAC appears to

have several advantages over the FMSS in this sample. As the FMSS requires significant

time and money to transcribe and rate and the EEAC demonstrates similar patterns of

validity to the FMSS, the ease of administering the EEAC makes it a good alternative

81

measure. There were also concerns in this sample about whether the FMSS can be rated

reliably, which further supports the use of a measure which does not require observer

ratings. While the FMSS appears to provide a better reflection of child characteristics in

this sample, the EEAC provides a measure of both child and parent characteristics.

Finally, the identification of mediators and moderators of treatment is important to better

understand and improve treatment effects (Kraemer et al., 2002). The moderator effects

found for the EEAC, but not for the FMSS, indicate that the more stable relationship

patterns represented by the EEAC may be more useful in understanding treatment effects

among children with mood disorders.

Limitations and Directions for Future Research

Several of the analyses in this study, particularly for fathers, had low sample sizes

when only a subset of the total data could be used (i.e. waitlist or follow-up data). The power to detect results was lowered in these instances and examination of effect sizes suggests that EE may have effects in the prediction of future symptom levels for fathers similar to those in mothers, which were not detectible in this study with the available

sample size. Further research, particularly with fathers, will be necessary to more clearly

understand how EE is associated with future symptom levels.

The sample used in this study includes children with mood disorders and their

families. As this is a group in which the psychometric properties of the FMSS have not

previously been examined, the present results provide an important extension of the

research. However these results may not be generalizable to samples of other clinical

disorders or to community samples.

82

This is the first psychometric examination of the EEAC in a child sample. The

EEAC appears to be a promising assessment tool which deserves replication of the current results and further research.

83

REFERENCES

Achenbach, T. M., & Rescorla, L. A. (2001). Manual for the ASEBA School-Age Forms and Profiles. Burlington, VT: University of Vermont, Research Center for Children, Youth, & Families.

Aiken, L. S., & West, S. G. (1991). Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks, CA: Sage Publications.

Asarnow, J. R., Goldstein, M. J., Tompson, M., & Guthrie, D. (1993). One-year outcomes of depressive disorders in child psychiatric in-patients: Evaluation of the prognostic power of a brief measure of expressed emotion. Journal of Child Psychology & Psychiatry, 34, 129-137.

Asarnow, J. R., Tompson, M., Hamilton, E. B., Goldstein, M. J., & Guthrie, D. (1994). Family expressed emotion, childhood-onset depression, and childhood-onset schizophrenia spectrum disorders: Is expressed emotion a nonspecific correlate of child psychopathology or a specific risk factor for depression? Journal of Abnormal Child Psychology, 22, 129-146.

Asarnow, J. R., Tompson, M., Woo, S., & Cantwell, D. P. (2001). Is expressed emotion a specific risk factor for depression or a nonspecific correlate of psychopathology? Journal of Abnormal Child Psychology, 29, 573-583.

Bachmann, S., Bottmer, C., Jacob, S., Kronmuller, K.-T., Backenstrass, M., Mundt, C., Renneberg, B., Fiedler, P., & Schroder, J. (2002). Expressed emotion in relatives of first-episode and chronic patients with schizophrenia and major depressive disorder--a comparison. Psychiatry Research, 112, 239-250.

Baker, B. L., Heller, T. L., & Henker, B. (2000). Expressed emotion, parenting stress, and adjustment in mothers of young children with behavior problems. Journal of Child Psychology & Psychiatry, 41, 907-915.

Barrowclough, C., & Hooley, J. M. (2003). Attributions and expressed emotion: A review. Clinical Psychology Review, 23, 849-880.

Blair, C., Freeman, C., & Cull, A. (1995). The families of anorexia nervosa and cystic fibrosis patients. Psychological Medicine, 25, 985-993.

84

Bolton, C., Calam, R., Barrowclough, C., Peters, S., Roberts, J., Wearden, A., & Morris, J. (2003). Expressed emotion, attributions and depression in mothers of children with problem behaviour. Journal of Child Psychology & Psychiatry, 44, 242-254.

Brown, G. W., Birley, J. L., & Wing, J. K. (1972). Influence of family life on the course of schizophrenic disorders: A replication. British Journal of Psychiatry, 121, 241- 258.

Brown, G. W., Carstairs, G. M., & Topping, G. (1958). Post-hospital adjustment of chronic mental patients. Lancet, ii, 685-689.

Brown, G. W., Monck, E. M., Carstairs, G. M., & Wing, J. K. (1962). Influence of family life on the course of schizophrenic illness. British Journal of Preventative and Social Medicine, 16, 55-68.

Brown, G. W., & Rutter, M. (1966). The measurement of family activities and relationships: A methodological study. Human Relations, 19, 241-263.

Bullock, B. M., Bank, L., & Burraston, B. (2002). Adult sibling expressed emotion and fellow sibling deviance: A new piece of the family process puzzle. Journal of Family Psychology, 16, 307-317.

Butzlaff, R. L., & Hooley, J. M. (1998). Expressed emotion and psychiatric relapse: A meta-analysis. Archives of General Psychiatry, 55, 547-552.

Caspi, A., Moffitt, T. E., Morgan, J., Rutter, M., Taylor, A., Arseneault, L., Tully, L., Jacobs, C., Kim-Cohen, J., & Polo-Tomas, M. (2004). Maternal expressed emotion predicts children's antisocial behavior problems: Using monozygotic-twin differences to identify environmental effects on behavioral development. Developmental Psychology, 40, 149-161.

Cerel, J., & Fristad, M. A. (1999a). A comparison of child/parent ratings and impact of psychopathology on a brief measure of expressed emotion. Unpublished manuscript.

Cerel, J., & Fristad, M. A. (1999b). Parent:child concordance & impact of mood on expressed emotion ratings. In M. A. Fristad (Chair), Issues in Assessing Expressed Emotion in Children and Adolescents. Symposium conducted at the 46th Annual Meeting of the American Academy of Child and Adolescent Psychiatry, Chicago, IL.

Cerel, J., & Fristad, M. A. (2001). Scaling structured interview data: A comparison of two methods. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 341-346.

85

Chambless, D. L., & Steketee, G. (1999). Expressed emotion and behavior therapy outcome: A prospective study with obsessive-compulsive and agoraphobic outpatients. Journal of Consulting & Clinical Psychology, 67, 658-665.

Chiariello, M. A., & Orvaschel, H. (1995). Patterns of parent-child communication: Relationship to depression. Clinical Psychology Review, 15, 395-407.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Earlbaum Associates.

Davidson, K. H., & Fristad, M. A. (2004). The Treatment Beliefs Questionnaire (TBQ): An instrument to assess parent and child beliefs about the child's illness and treatment needs. Manuscript submitted for publication.

Emmelkamp, P. M. G., Kloek, J., & Blaauw, E. (1992). Obsessive-compulsive disorders. In P. H. Wilson (Ed.), Principles and Practice of Relapse Prevention. (pp. 213- 234). New York: Guilford.

Fichter, M. M., Glynn, S. M., Weyerer, S., Liberman, R. P., & Frick, U. (1997). Family climate and expressed emotion in the course of alcoholism. Family Process, 36, 202-221.

Fischmann-Havstad, L., & Marston, A. R. (1984). Weight loss maintenance as an aspect of family emotion and process. British Journal of Clinical Psychology, 23, 265- 271.

Flanagan, D. A., & Wagner, H. L. (1991). Expressed emotion and panic in the prediction of diet treatment compliance. British Journal of Clinical Psychology, 30, 231-240.

Friedmann, M. S., & Goldstein, M. J. (1993). Relatives' awareness of their own expressed emotion as measured by a self-report adjective checklist. Family Process, 32, 459-471.

Friedmann, M. S., & Goldstein, M. J. (1994). Relatives' perceptions of their interactional behavior with a schizophrenic family member. Family Process, 33, 377-387.

Fristad, M. A., Gavazzi, S. M., & Mackinaw-Koons, B. (2003). Family psychoeducation: An adjunctive intervention for children with bipolar disorder. Biological Psychiatry, 53, 1000-1008.

Fristad, M. A., & Goldberg-Arnold, J. S. (2003). Family interventions for early- onset bipolar disorder. In B. Geller & M. P. DelBello (Eds.), Bipolar Disorders in Childhood and Early Adolescence. New York: Guilford.

86

Fristad, M. A., Schock, A., Gavazzi, S. M., & Goldberg-Arnold, J. S. (1999). Expressed emotion differences among clinical and nonclinical mothers and fathers. In M. A. Fristad (Chair), Issues in Assessing Expressed Emotion in Children and Adolescents. Symposium conducted at the 46th Annual Meeting of the American Academy of Child and Adolescent Psychiatry, Chicago, IL.

Fristad, M. A., Weller, E. B., & Weller, R. A. (1992). The Mania Rating Scale: Can it be used in children? A preliminary report. Journal of the American Academy of Child & Adolescent Psychiatry, 31, 252-257.

Fristad, M. A., Weller, R. A., & Weller, E. B. (1995). The Mania Rating Scale (MRS): Further reliability and validity studies with children. Annals of Clinical Psychiatry, 7, 127-132.

Gavazzi, S. M., McKenry, P. C., Jacobson, J. A., Julian, T. W., & Lohman, B. (2000). Modeling the effects of expressed emotion, psychiatric symptomology, and marital quality levels on male and female verbal aggression. Journal of Marriage & the Family, 62, 669-682.

Geller, B., Craney, J. L., Bolhofner, K., Nickelsburg, M. J., Williams, M., & Zimerman, B. (2002). Two-year prospective follow-up of children with a prepubertal and early adolescent bipolar disorder phenotype. American Journal of Psychiatry, 159, 927- 933.

Geller, B., Tillman, R., Craney, J. L., & Bolhofner, K. (2004). Four-year prospective outcome and natural history of mania in children with a prepubertal and early adolescent bipolar disorder phenotype. Archives of General Psychiatry, 61, 459-467.

Goldberg-Arnold, J. S., & Fristad, M. A. (2003). Psychotherapy for children with bipolar disorder. In B. Geller & M. P. DelBello (Eds.), Bipolar Disorder in Childhood and Early Adolescence. New York: Guilford.

Gottschalk, L. A., & Gleser, G. C. (1969). The measurement of psychological states through the content analysis of verbal behavior. Berkeley: University of California Press.

Grunes, M. S., Neziroglu, F., & McKay, D. (2001). Family involvement in the behavioral treatment of obsessive-compulsive disorder: A preliminary investigation. Behavior Therapy, 32, 803-820.

Hahlweg, K., Goldstein, M. J., Nuechterlein, K. H., Doane, J. A., Miklowitz, D. J., & Snyder, K. S. (1989). Expressed emotion and patient-relative interaction in families of recent onset schizophrenics. Journal of Consulting & Clinical Psychology, 57, 11-18.

87

Harter, S. (1985). Manual for the Social Support Scale for Children. Denver: University of Denver.

Hayhurst, H., Cooper, Z., Paykel, E. S., Vearnals, S., & Ramana, R. (1997). Expressed emotion and depression. British Journal of Psychiatry, 171, 439-443.

Hedlund, S., Fichter, M. M., Quadflieg, N., & Brandl, C. (2003). Expressed emotion, family environment, and parental bonding in bulimia nervosa: A 6-year investigation. Eating & Weight Disorders, 8, 26-35.

Heikkila, J., Karlsson, H., Taiminen, T., Lauerma, H., Ilonen, T., Leinonen, K.- M., Wallenius, E., Virtanen, H., Heinimaa, M., Koponen, S., Jalo, P., Kaljonen, A., & Salakangas, R. K. R. (2002). Expressed emotion is not associated with disorder severity in first-episode . Psychiatry Research, 111, 155-165.

Hibbs, E. D., Hamburger, S. D., Lenane, M., Rapoport, J. L., Kruesi, M. J., Keysor, C. S., & Goldstein, M. J. (1991). Determinants of expressed emotion in families of disturbed and normal children. Journal of Child Psychology & Psychiatry, 32, 757- 770.

Hinrichsen, G. A., & Pollack, S. (1997). Expressed emotion and the course of late-life depression. Journal of Abnormal Psychology, 106, 336-340.

Hoffman, P. D., Buteau, E., Hooley, J. M., Fruzzetti, A. E., & Bruce, M. L. (2003). Family members' knowledge about borderline personality disorder: Correspondence with their levels of depression, burden, distress, and expressed emotion. Family Process, 42, 469-478.

Hooley, J. M. (1986). Expressed emotion and depression: Interactions between patients and high- versus low-expressed-emotion spouses. Journal of Abnormal Psychology, 95, 237-246.

Hooley, J. M., Orley, J., & Teasdale, J. D. (1986). Levels of expressed emotion and relapse in depressed patients. British Journal of Psychiatry, 148, 642-647.

Hooley, J. M., & Teasdale, J. D. (1989). Predictors of relapse in unipolar depressives: Expressed emotion, marital distress, and perceived criticism. Journal of Abnormal Psychology, 98, 229-235.

Jacobsen, T., Hibbs, E., & Ziegenhain, U. (2000). Maternal expressed emotion related to attachment disorganization in early childhood: A preliminary report. Journal of Child Psychology & Psychiatry, 41, 899-906.

88

Jarbin, H., Grawe, R. W., & Hansson, K. (2000). Expressed emotion and prediction of relapse in adolescents with psychotic disorders. Nordic Journal of Psychiatry, 54, 201-205.

Kaufman, A. S., & Kaufman, N. L. (1990). Kaufman Brief Intelligence Test (K- BIT) Manual. Los Angeles: Western Psychological Services.

Kraemer, H. C., Wilson, G. T., Fairburn, C. G., & Agras, W. S. (2002). Mediators and moderators of treatment effects in randomized clinical trials. Archives of General Psychiatry, 59, 877-883.

Landis, J., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33, 159-174.

Le Grange, D., Eisler, I., Dare, C., & Hodes, M. (1992). Family criticism and self- starvation: A study of expressed emotion. Journal of Family Therapy, 14, 177-192.

Leeb, B., Hahlweg, K., Goldstein, M. J., Feinstein, E., Muller, U., Dose, M., & Magaña-Amato, A. (1991). Cross-national reliability, concurrent validity, and stability of a brief method for assessing expressed emotion. Psychiatry Research, 39, 25-31.

Lenior, M. E., Dingemans, P. M. A. J., & Linszen, D. H. (1997). A quantitative measure for expressed emotion. Psychiatry Research, 69, 53-65.

Lenior, M. E., Dingemans, P. M. A. J., Schene, A. H., Hart, A. A. M., & Linszen, D. H. (2002). The course of parental expressed emotion and psychotic episodes after family intervention in recent-onset schizophrenia. A longitudinal study. Schizophrenia Research, 57, 183-190.

Liakopoulou, M., Alifieraki, T., Katideniou, A., Peppa, M., Maniati, M., Tzikas, D., Hibbs, E. D., & Dacou-Voutetakis, C. (2001). Maternal expressed emotion and metabolic control of children and adolescents with diabetes mellitus. Psychotherapy & Psychosomatics, 70, 78-85.

Lund, J., & Merrell, K. W. (2001). Social and anitsocial behavior of children with learning and behavioral disorders: Construct validity of the Home and Community Social Behavior Scales. Journal of Psychoeducational Assessment, 19, 112-122.

Magaña, A. B., Goldstein, M. J., Karno, M., Miklowitz, D. J., Jenkins, J., & Falloon, I. R. H. (1986). A brief method for assessing expressed emotion in relatives of psychiatric patients. Psychiatry Research, 17, 203-212.

Malla, A. K., Kazarian, S. S., Barnes, S., & Cole, J. D. (1991). Validation of the Five Minute Speech Sample in measuring expressed emotion. Canadian Journal of Psychiatry, 36, 297-299.

89

Marom, S., Munitz, H., Jones, P. B., Weizman, A., & Hermesh, H. (2002). Familial expressed emotion: Outcome and course of Israeli patients with schizophrenia. Schizophrenia Bulletin, 28, 731-743.

Marshall, V. G., Longwell, L., Goldstein, M. J., & Swanson, J. M. (1990). Family factors associated with aggressive symptomatology in boys with attention deficit hyperactivity disorder: A research note. Journal of Child Psychology & Psychiatry, 31, 629-636.

McCarty, C. A., Lau, A. S., Valeri, S. M., & Weisz, J. R. (2004). Parent-child interactions in relation to critical and emotionally overinvolved expressed emotion (EE): Is EE a proxy for behavior? Journal of Abnormal Child Psychology, 32, 83-93.

McCarty, C. A., & Weisz, J. R. (2002). Correlates of expressed emotion in mothers of clinically-referred youth: An examination of the five-minute speech sample. Journal of Child Psychology & Psychiatry, 43, 759-768.

McCleary, L., & Sanford, M. (2002). Parental expressed emotion in depressed adolescents: Prediction of clinical course and relationship to comorbid disorders and social functioning. Journal of Child Psychology & Psychiatry, 43, 587-595.

McGuire, J. B., & Earls, F. (1994). The test retest stability of the Five Minute Speech Sample in parents of disadvantaged, minority children. Journal of Child Psychology & Psychiatry, 35, 971-979.

Merrell, K. W., & Caldarella, P. (1999). Social-behavioral assessment of at-risk early adolescent students: Psychometric characteristics and validity of a parent report form of the School Social Behavior Scales. Journal of Psychoeducational Assessment, 17, 36-49.

Merrell, K. W., Streeter, A. L., Boelter, E. W., Caldarella, P., & Gentry, A. (2001). Validity of the Home and Community Social Behavior Scales: Comparisons with five behavior-rating scales. Psychology in the Schools, 38, 313-325.

Miklowitz, D. J. (2005, April). New psychosocial interventions for juvenile-onset bipolar disorder. Paper presented at the NIMH Pediatric Bipolar Conference, Coral Gables, Florida.

Miklowitz, D. J., Goldstein, M. J., Falloon, I. R., & Doane, J. A. (1984). Interactional correlates of expressed emotion in the families of schizophrenics. British Journal of Psychiatry, 144, 482-487.

90

Miklowitz, D. J., Goldstein, M. J., Nuechterlein, K. H., Snyder, K. S., & Mintz, J. (1988). Family factors and the course of bipolar affective disorder. Archives of General Psychiatry, 45, 225-231.

Miklowitz, D. J., Simoneau, T. L., George, E. L., Richards, J. A., Kalbag, A., Sachs-Ericsson, N., & Suddath, R. (2000). Family-focused treatment of bipolar disorder: 1-year effects of a psychoeducational program in conjunction with pharmacotherapy. Biological Psychiatry, 48, 582-592.

Nelson, D. R., Hammen, C., Brennan, P. A., & Ullman, J. B. (2003). The impact of maternal depression on adolescent adjustment: The role of expressed emotion. Journal of Consulting & Clinical Psychology, 71, 935-944.

Nugter, A., Dingemans, P., Van der Does, J. W., Linszen, D., & Gersons, B. (1997). Family treatment, expressed emotion and relapse in recent onset schizophrenia. Psychiatry Research, 72, 23-31.

O'Farrell, T. J., Hooley, J., Fals-Stewart, W., & Cutter, H. S. G. (1998). Expressed emotion and relapse in alcoholic patients. Journal of Consulting & Clinical Psychology, 66, 744-752.

Parker, G., Johnston, P., & Hayward, L. (1988). Parental "expressed emotion" as a predictor of schizophrenic relapse. Archives of General Psychiatry, 45, 806-813.

Peris, T. S., & Baker, B. L. (2000). Applications of the expressed emotion construct to young children with externalizing behavior: Stability and prediction over time. Journal of Child Psychology & Psychiatry, 41, 457-462.

Peris, T. S., & Hinshaw, S. P. (2003). Family dynamics and preadolescent girls with ADHD: the relationship between expressed emotion, ADHD symptomatology, and comorbid disruptive behavior. Journal of Child Psychology & Psychiatry, 44, 1177-1190.

Peter, H., & Hand, I. (1988). Patterns of patient-spouse interaction in agoraphobics: Assessment by Camberwell Family Interview (CFI) and impact on outcome of self-exposure treatment. In I. Hand & H. U. Wittchen (Eds.), Panic and Phobias 2: Treatments and Variables Affecting Course and Outcome (pp. 240-251). Berlin: Springer-Verlag.

Poznanski, E. O., & Mokros, H. B. (1996). Children's Depression Rating Scale, Revised Manual. Los Angeles, CA: Western Psychological Services.

Priebe, S., Wildgrube, C., & Muller-Oerlinghausen, B. (1989). Lithium prophylaxis and expressed emotion. British Journal of Psychiatry, 154, 396-399.

91

Reid, J. B., Patterson, G. R., & Snyder, J. (2002). Antisocial behavior in children and adolescents: A developmental analysis and the Oregon model for intervention. Washington, D.C.: American Psychological Association.

Rutter, M., & Brown, G. W. (1966). The reliability and validity of measures of family life and relationships in families containing a psychiatric patient. Social Psychiatry, 1, 38-53.

Schock, A. M., Gavazzi, S. M., Fristad, M. A., & Goldberg-Arnold, J. S. (2002). The role of father participation in the treatment of childhood mood disorders. Family Relations: Interdisciplinary Journal of Applied Family Studies, 51, 230-237.

Schwartz, C. E., Dorer, D. J., Beardslee, W. R., Lavori, P. W., & Keller, M. D. (1990). Maternal expressed emotion and parental affective disorder: Risk for childhood depressive disorder, substance abuse, or conduct disorder. Journal of Psychiatric Research, 24, 231-250.

Shimodera, S., Mino, Y., Fujita, H., Izumoto, Y., Kamimura, N., & Inoue, S. (2002). Validity of a five-minute speech sample for the measurement of expressed emotion in the families of Japanese patients with mood disorders. Psychiatry Research, 112, 231-237.

Shimodera, S., Mino, Y., Inoue, S., Izumoto, Y., Kishi, Y., & Tanaka, S. (1999). Validity of Five-Minute Speech Sample measuring expressed emotion in the families of patients with schizophrenia in Japan. Comprehensive Psychiatry, 40, 372-376.

Siqueland, L., Kendall, P. C., & Steinberg, L. (1996). Anxiety in children: Perceived family environments and observed family interaction. Journal of Clinical Child Psychology, 25, 225-237.

Sisson, D. P. (2005). Expressed Emotion in Parents of Children with Early-Onset Mood Disorders. Unpublished Dissertation, The Ohio State University, Columbus.

Smith, K. D. (2005). Seasonal Changes in Mood and Behavior Among Children and Adolescents. Unpublished Dissertation, Ohio State University, Columbus.

St. Jonn-Seed, M., & Weiss, S. (2002). Maternal expressed emotion as a predictor of emotional and behavioral problems in low birth weight children. Issues in Mental Health Nursing, 23, 649-672.

Steketee, G., & Van Noppen, B. (2003). Family approaches to treatment for obsessive compulsive disorder. Journal of Family Psychotherapy, 14, 55-71.

92

Stevenson, K., Sensky, T., & Petty, R. (1991). Glycaemic control in adolescents with Type I diabetes and parental expressed emotion. Psychotherapy & Psychosomatics, 55, 170-175.

Stubbe, D. E., Zahner, G. E., Goldstein, M. J., & Leckman, J. F. (1993). Diagnostic specificity of a brief measure of expressed emotion: A community study of children. Journal of Child Psychology & Psychiatry, 34, 139-154.

Szmukler, G. I., Eisler, I., Russell, G. F., & Dare, C. (1985). Anorexia nervosa, parental "expressed emotion" and dropping out of treatment. British Journal of Psychiatry, 147, 265-271.

Tarrier, N., Barrowclough, C., Vaughn, C., Bamrah, J. S., Porceddu, K., Watts, S., & Freeman, H. (1988). The community management of schizophrenia: A controlled trial of a behavioural intervention with families to reduce relapse. British Journal of Psychiatry, 153, 532-542.

Tarrier, N., Sommerfield, C., & Pilgrim, H. (1999). Relatives' expressed emotion (EE) and PTSD treatment outcome. Psychological Medicine, 29, 801-811.

Tompson, M. C., Goldstein, M. J., Lebell, M. B., Mintz, L. I., Marder, S. R., & Mintz, J. (1995). Schizophrenic patients' perceptions of their relatives' attitudes. Psychiatry Research, 57, 155-167.

Uehara, T., Kawashima, Y., Goto, M., Tasaki, S., & Someya, T. (2001). Psychoeducation for the families of patients with eating disorders and changes in expressed emotion: A preliminary study. Comprehensive Psychiatry, 42, 132-138.

Van Furth, E. F., Van Strien, D. C., Martina, L. M. L., Van Son, M. J. M., Hendrickx, J. J. P., & Van Engeland, H. (1996). Expressed emotion and the prediction of outcome in adolescent eating disorders. International Journal of Eating Disorders, 20, 19- 31.

Van Furth, E. F., Van Strien, D. C., Van Son, M. J., & Van Engeland, H. (1993). The validity of the Five-Minute Speech Sample as an index of expressed emotion in parents of eating disorder patients. Journal of Child Psychology & Psychiatry, 34, 1253- 1260.

Van Humbeeck, G., Van Audenhove, C., De Hert, M., Pieters, G., & Storms, G. (2002). Expressed emotion: A review of assessment instruments. Clinical Psychology Review, 22, 321-341.

Van Os, J., Marcelis, M., Germeys, I., Graven, S., & Delespaul, P. (2001). High expressed emotion: Marker for a caring family? Comprehensive Psychiatry, 42, 504-507.

93

Vaughn, C. (1989). Expressed emotion in family relationships. Journal of Child Psychology & Psychiatry, 30, 13-22.

Vaughn, C., & Leff, J. (1976a). The influence of family and social factors on the course of psychiatric illness: A comparison of schizophrenic and depressed neurotic patients. British Journal of Psychiatry, 129, 125-137.

Vaughn, C., & Leff, J. (1976b). The measurement of expressed emotion in the families of psychiatric patients. British Journal of Social & Clinical Psychology, 15, 157- 165.

Vostanis, P., & Nicholls, J. (1995). Nine-month changes of maternal expressed emotion in conduct and emotional disorders of childhood: A follow-up study. Journal of Child Psychology & Psychiatry, 36, 833-846.

Vostanis, P., Nicholls, J., & Harrington, R. (1994). Maternal expressed emotion in conduct and emotional disorders of childhood. Journal of Child Psychology & Psychiatry, 35, 365-376.

Wamboldt, F. S., O'Connor, S. L., Wamboldt, M. Z., Gavin, L. A., & Klinnert, M. D. (2000). The Five Minute Speech Sample in children with asthma: Deconstructing the construct of expressed emotion. Journal of Child Psychology & Psychiatry, 41, 887-898.

Wearden, A. J., Tarrier, N., Barrowclough, C., Zastowny, T. R., & Rahill, A. A. (2000). A review of expressed emotion research in health care. Clinical Psychology Review, 20, 633-666.

Wearden, A. J., Tarrier, N., & Davies, R. (2000). Partners' expressed emotion and the control and management of Type 1 diabetes in adults. Journal of Psychosomatic Research, 49, 125-130.

Weller, E. B., Weller, R. A., Rooney, M. T., & Fristad, M. A. (1999a). Children's Interview for Psychiatric Syndromes (ChIPS). Washington, D.C.: American Psychiatric Press, Inc.

Weller, E. B., Weller, R. A., Rooney, M. T., & Fristad, M. A. (1999b). Children's Interview for Psychiatric Syndromes, Parent Version (P-ChIPS). Washington, D.C.: American Psychiatric Press, Inc.

Worrall-Davies, A., Owens, D., Holland, P., & Haigh, D. (2002). The effect of parental expressed emotion on glycaemic control in children with type 1 diabetes: Parental expressed emotion and glycaemic control in children. Journal of Psychosomatic Research, 52, 107-113.

94

Young, R. C., Biggs, J. T., Ziegler, V. E., & Meyer, D. A. (1978). A rating scale for mania: Reliability, validity and sensitivity. British Journal of Psychiatry, 133, 429- 435.

95

APPENDIX A

TABLES

96

C to P Positive C to P Negative P to C Positive P to C Negative Mothers Mean 45.76 48.03 60.48 27.02 SD 9.89 10.90 9.15 9.24 Minimum 27 18 30 11 Maximum 70 72 78 65

Fathers Mean 48.84 43.68 61.26 26.23 SD 10.30 10.96 8.78 7.63 Minimum 23 24 36 14 Maximum 68 67 78 44

Table 1: EEAC Descriptives

97

Mothers Fathers

FMSS CRIT Low 16% (N=25) 19% (N=13) Borderline 19% (N=30) 41% (N=28) High 65% (N=100) 41% (N=28)

FMSS EOI Low 39% (N=61) 51% (N=35) Borderline 50% (N=77) 35% (N=24) High 11% (N=17) 15% (N=10)

Modified EOI Low 95% (N=147) 96% (N=66) High 5% (N=8) 4% (N=3)

Positive remarks Mean 1.65 1.70 SD 1.95 1.90 Mode 0 0 Minimum 0 0 Maximum 9 8

Critical comments Mean 1.37 0.94 SD 1.61 1.50 Mode 0 0 Minimum 0 0 Maximum 7 7

Table 2: FMSS Descriptives

98

Dependent Variable df SS MS F

Mothers

C to P Positive 2 850.803 425.402 4.668* Error 143 13031.224 91.127 Total 145 13882.027

C to P Negative 2 1494.073 747.036 6.644** Error 144 16191.329 112.440 Total 146 17685.401

P to C Positive 2 132.955 66.478 .769 Error 146 12623.730 86.464 Total 148 12756.685

P to C Negative 2 117.145 58.573 .639 Error 144 13208.855 91.728 Total 146 13326.000

Fathers

C to P Positive 2 1045.608 522.804 5.764** Error 63 5714.513 90.707 Total 65 6760.121

C to P Negative 2 461.442 230.721 1.988 Error 65 7543.676 116.057 Total 67 8005.118

P to C Positive 2 196.230 98.115 1.365 Error 65 4672.050 71.878 Total 67 4868.279

P to C Negative 2 277.936 138.968 2.305 Error 65 3919.050 60.293 Total 67 4196.985

* p≤.05 **p≤.01

Table 3: FMSS CRIT Predictions of EEAC Scores

99

C to P C to P P to C P to C Positive Negative Positive Negative Mothers High CRIT 43.522 50.362 59.442 27.202 Borderline CRIT 48.379 46.276 61.828 27.931 Low CRIT 48.680 41.917 60.560 25.083

Fathers High CRIT 44.308 46.607 60.143 28.357 Borderline CRIT 52.556 40.815 63.074 23.926 Low CRIT 52.231 43.923 64.308 25.308

Table 4: EEAC Means by FMSS CRIT Status

100

Mothers Fathers EEAC Scales Criticism Positive Criticism Positive C to P Positive -.164* .162 -.525** .124 C to P Negative .248** -.214** .107 -.014 P to C Positive -.031 .141 -.242* .041 P to C Negative .118 .074 .263* -.078

* p≤.05 **p≤.01

Table 5: Correlations between EEAC and FMSS Continuous Scales

101

Six-Month Twelve-Month Mothers r t r t C to P Positive .470** -2.020* .508** -1.937 C to P Negative .566** 2.438* .588** 1.757 P to C Positive .611** -1.782 .630** -.772 P to C Negative .515** 1.208 .313* 1.067

Fathers C to P Positive .412* -1.273 .383 -.641 C to P Negative .468** 1.954 .733** 2.890** P to C Positive .769** -1.025 .764** -1.110 P to C Negative .558** -.162 .470* .572

* p≤.05 **p≤.01

Table 6: EEAC Test-retest Reliability

102

Six-Month Twelve-Month Mothers r t r t Critical Comments .251 .000 .432* .000 Positive Remarks .033 -.397 .407* .409

Fathers Critical Comments .219 .000 .665** .000 Positive Remarks .085 .535 .191 .651

* p≤.05 **p≤.01

Table 7: FMSS Continuous Scales Test-retest Reliability

103

Six-Month Twelve-Month Mothers r same inc. dec. r same inc. dec. CRIT .215 46% 24% 30% .300 57% 23% 20% EOI .050 44% 28% 28% .154 50% 20% 30%

Fathers CRIT .298 45% 20% 35% .747** 59% 35% 6% EOI .109 35% 10% 55% .008 35% 24% 41%

* p≤.05 **p≤.01

Table 8: FMSS Categorical Scales Test-retest Reliability

104

C to P C to P P to C P to C Positive Negative Positive Negative

Abuse/ Neglect (parent report) -.016 .008 -.045 .138 Abuse/ Neglect (child report) -.114 .058 -.067 -.059

SSS Support from Parents .140 .012 .089 -.062

HCSBS Social Competence .444** -.508** .207* -.205* HCSBS Antisocial Behavior -.351** .611** -.081 .214**

TBQ Specific Beliefs .037 .033 .060 -.046 TBQ General Beliefs -.034 -.036 .102 -.232** TBQ Seriousness -.040 .070 .164* -.195*

TBQ Costs and Benefits .004 .046 .236** -.085 TBQ Knowledge .140 -.074 .160* -.115 TBQ Relationships .150 -.085 .229** -.200** TBQ Total Score .016 .039 .181* -.239**

* p≤.05 **p≤.01

Table 9: EEAC Convergent Validity Correlations for Mothers

105

C to P C to P P to C P to C Positive Negative Positive Negative

Abuse/ Neglect (parent report) -.160 -.089 -.102 .057 Abuse/ Neglect (child report) -.035 -.012 -.051 .002

SSS Support from Parents .123 .015 .048 .020

HCSBS Social Competence .441** -.418** .090 -.046 HCSBS Antisocial -.333** .552** .005 .244*

TBQ Specific Beliefs -.246* .222* .011 -.033 TBQ General Beliefs -.032 .131 .157 -.073 TBQ Seriousness -.288** .286** .208* -.055

TBQ Costs and Benefits -.014 -.003 .235* .015 TBQ Knowledge -.042 .039 .347** -.236* TBQ Relationships -.007 .073 .270** -.071 TBQ Total Score -.269** .256* .202 -.079

* p≤.05 **p≤.01

Table 10: EEAC Convergent Validity Correlations for Fathers

106

Number of Number of Criticisms Positives

Abuse/ Neglect (parent report ) .005 .008 Abuse/ Neglect (child report ) .081 -.091

SSS Support from Parents .010 -.059

HCSBS Social Competence -.222* .327** HCSBS Antisocial .267** -.159

TBQ Specific Beliefs -.094 .039 TBQ General Beliefs -.052 .086 TBQ Seriousness -.022 -.086

TBQ Costs and Benefits -.037 .075 TBQ Knowledge -.035 -.097 TBQ Relationships -.154 .015 TBQ Total Score -.065 -.040

* p≤.05 **p≤.01

Table 11: FMSS Convergent Validity Correlations for Mothers

107

Number of Number of Criticisms Positives

Abuse/ Neglect (parent report ) .138 -.134 Abuse/ Neglect (child report ) .218 -.151

SSS Support from Parents -.011 .012

HCSBS Social Competence -.240 .243 HCSBS Antisocial .267* -.154

TBQ Specific Beliefs -.043 .121 TBQ General Beliefs -.162 .061 TBQ Seriousness .140 -.026

TBQ Costs and Benefits -.288* -.085 TBQ Knowledge .005 .032 TBQ Relationships -.160 .113 TBQ Total Score -.062 .106

* p≤.05 **p≤.01

Table 12: FMSS Convergent Validity Correlations for Fathers

108

Dependent Variable df SS MS F

Cost/ Benefit 2 4.483 2.241 3.126* Error 149 106.828 .717 Total 151 111.311

* p≤.05 **p≤.01

Table 13: FMSS EOI Predictions of Cost/ Benefit Treatment Beliefs for Mothers

109

Dependent Variable df SS MS F

Social Competence 2 1858.673 929.337 3.228* Error 131 37720.379 287.942 Total 133 39579.052

Antisocial Behavior 2 5495.538 2747.769 5.733** Error 129 61828.182 479.288 Total 131 67323.720

* p≤.05 **p≤.01

Table 14: FMSS CRIT Predictions of HCSBS Scores for Mothers

110

C to P C to P P to C P to C Positive Negative Positive Negative

K-BIT Composite .055 -.189* .040 -.205** TRF Academic Performance .250* -.348** .272* -.247*

SSS Support from Teachers .185* -.070 .120 -.068 SSS Support from Peers .050 -.008 .093 -.102

* p≤.05 **p≤.01

Table 15: EEAC Discriminant Validity Correlations for Mothers

111

C to P C to P P to C P to C Positive Negative Positive Negative

K-BIT Composite -.052 -.106 -.155 .128 TRF Academic Performance .121 -.282* -.134 .087

SSS Support from Teachers .008 .081 .124 -.013 SSS Support from Peers -.005 .068 .070 .033

* p≤.05 **p≤.01

Table 16: EEAC Discriminant Validity Correlations for Fathers

112

Number of Number of Criticisms Positives

K-BIT Composite -.196* .118 TRF Academic Performance .035 .078

SSS Support from Teachers -.092 .030 SSS Support from Peers -.182* .061

* p≤.05 **p≤.01

Table 17: FMSS Discriminant Validity Correlations for Mothers

113

Number of Number of Criticisms Positives

K-BIT Composite .031 .285* TRF Academic Performance -.187 .188

SSS Support from Teachers .049 .211 SSS Support from Peers .132 .230

* p≤.05 **p≤.01

Table 18: FMSS Discriminant Validity Correlations for Fathers

114

Variable B SE B β t R2 R2 ∆ Step 1 .002 .002 Constant 1.009 .256 3.937** Initial Mood Symptoms .062 .191 .044 .324 Step 2 .152 .150** Constant 1.908 .378 5.050** Initial Mood Symptoms .058 .177 .041 .326 EEAC C to P Positive -.019 .006 -.388 -3.067**

* p≤.05 **p≤.01

Table 19: Hierarchical Regression Predicting P-ChIPS Mood Symptoms from EEAC Child Positive Adjective Ratings at One-year Follow-up for Mothers

115

Variable B SE B β t R2 R2 ∆ Step 1 .001 .001 Constant 1.020 .260 3.919** Initial Mood Symptoms .041 .193 .030 .215 Step 2 .141 .140** Constant .185 .381 .484 Initial Mood Symptoms .068 .181 .050 .378 EEAC C to P Negative .017 .006 .374 2.851**

* p≤.05 **p≤.01

Table 20: Hierarchical Regression Predicting P-ChIPS Mood Symptoms from EEAC Child Negative Adjective Ratings at One-year Follow-up for Mothers

116

Variable B SE B β t R2 R2 ∆ Step 1 .241 .241** Constant .680 .243 2.799** Initial Behavior Problems .454 .155 .491 2.927** Step 2 .414 .173** Constant -.327 .424 -.771 Initial Behavior Problems .319 .147 .344 2.164* EEAC C to P Negative .026 .009 .441 2.769**

* p≤.05 **p≤.01

Table 21: Hierarchical Regression Predicting P-ChIPS Behavior Problems from EEAC Child Negative Adjective Ratings at One-year Follow-up for Fathers

117

Variable B SE B β t R2 R2 ∆ Step 1 .029 .029 Constant .741 .280 2.650* Initial mood symptoms .255 .208 .169 1.227 Step 2 .147 .118* Constant .505 .279 1.808 Initial mood symptoms .314 .198 .208 1.582 FMSS # Criticisms .137 .052 .346 2.629*

* p≤.05 **p≤.01

Table 22: Hierarchical regression predicting P-ChIPS Mood Symptoms from FMSS Critical Comments at One-year Follow-up for Mothers

118

Variable B SE B β t R2 R2 ∆ Step 1 .122 .122 Constant .996 .238 4.186** Initial anxiety symptoms .541 .203 .350 2.668** Step 2 .257 .135** Constant .522 .271 1.922 Initial anxiety symptoms .613 .190 .396 3.226** FMSS # Criticisms .351 .117 .370 3.009**

* p≤.05 **p≤.01

Table 23: Hierarchical Regression Predicting P-ChIPS Anxiety Symptoms from FMSS Critical Comments at One-year Follow-up for Mothers

119

Source of Variance df SS MS F Corrected Model 3 2.008 .669 2.890* Intercept 1 1.138 1.138 4.912* Initial Mood Symptoms 1 .438 .438 1.893 Initial CRIT 2 1.625 .813 3.509* Error 49 11.347 .232 Total 53 74.388 Corrected Total 52 13.355

* p≤.05 **p≤.01

Table 24: FMSS CRIT Predictions of P-ChIPS Mood Symptoms at One-year Follow-up for Mothers

120

Variable B SE B β t R2 R2 ∆ Step 1 .092 .090** Constant .027 .080 .331 Initial MSI .251 083 .262 3.014** Treatment Status .019 .081 .020 .238 EEAC C to P Negative .082 .082 .087 .995 Step 2 .120 .028* Constant .009 .080 .112 Initial MSI .229 .083 .239 2.763** Treatment Status .006 .080 .006 .073 EEAC C to P Negative .079 .081 .084 .973 Treatment X EEAC -.160 .077 -.169 -2.075*

* p≤.05 **p≤.01

Table 25: Hierarchical Regression Predicting Six-month MSI Score from Treatment X EEAC Child Negative Adjective Interaction for Mothers

121

Variable B SE B β t R2 R2 ∆ Step 1 .099 .099** Constant .038 .082 .464 Initial MSI .304 081 .309 3.752** Treatment Status -.006 .083 -.006 -.074 EEAC P to C Negative .032 .083 .032 .385 Step 2 .135 .036* Constant .003 .082 .032 Initial MSI .282 .080 .286 3.513** Treatment Status -.022 .082 -.022 -.265 EEAC P to C Negative -.002 .083 -.002 -.029 Treatment X EEAC -.201 .084 -.196 -2.389*

* p≤.05 **p≤.01

Table 26: Hierarchical Regression Predicting Six-month MSI Score from Treatment X EEAC Parent Negative Adjective Interaction for Mothers

122

APPENDIX B

FIGURES

123

2.5

2

1.5

1 Treatment Waitlist 0.5

0

from Baseline to Six-Months to -0.5 Baseline from Change in Child Mood Symptoms Symptoms Mood Child in Change -1 Low Child Neg High Child Neg

Figure 1: Interaction between Treatment Status and EEAC Child Negative Adjective Ratings Predicting Post-treatment Mood Symptoms for Mothers

124

2.5

2

1.5

1 Treatment Waitlist 0.5

0

from Baseline to Six-Months to -0.5 Baseline from Change in Child Mood Symptoms Symptoms Mood Child in Change -1 Low Parent Neg High Parent Neg

Figure 2: Interaction between Treatment Status and EEAC Parent Negative Adjective Ratings Predicting Post-treatment Mood Symptoms for Mothers

125

APPENDIX C

QUESTIONNAIRES AND RATING SCALES

126

The Expressed Emotion Adjective Checklist (EEAC)

Please use the following adjectives to describe YOUR SON/DAUGHTER’S BEHAVIOR OVER THE LAST THREE MONTHS as it was DIRECTED TOWARD YOU. Please completely fill in the circle that is the best answer.

NEVER ALWAYS 1. Accepting 1 2 3 4 5 6 7 8 ------2. Active 1 2 3 4 5 6 7 8 ------3. Angry 1 2 3 4 5 6 7 8 ------4. Bored 1 2 3 4 5 6 7 8 ------5. Clear 1 2 3 4 5 6 7 8 ------6. Considerable 1 2 3 4 5 6 7 8 ------7. Contrary 1 2 3 4 5 6 7 8 ------8. Cooperative 1 2 3 4 5 6 7 8 ------9. Deceitful 1 2 3 4 5 6 7 8 ------10. Devoted 1 2 3 4 5 6 7 8 ------11. Easy to get along 1 2 3 4 5 6 7 8 with ------12. Friendly 1 2 3 4 5 6 7 8 ------13. Good-natured 1 2 3 4 5 6 7 8 ------14. Hostile 1 2 3 4 5 6 7 8 ------

15. Irresponsible 1 2 3 4 5 6 7 8 ------16. Irritable 1 2 3 4 5 6 7 8 ------17. Lazy 1 2 3 4 5 6 7 8 ------

127

18. Loving 1 2 3 4 5 6 7 8 ------19. Mean 1 2 3 4 5 6 7 8 ------20. Rude 1 2 3 4 5 6 7 8

Now please use the same adjectives to describe YOUR OWN BEHAVIOR OVER THE LAST THREE MONTHS as it was DIRECTED TOWRD YOU SON/DAUGHTER. ** Please completely fill in the circle that is the best answer.

NEVER ALWAYS 21. Accepting 1 2 3 4 5 6 7 8 ------22. Active 1 2 3 4 5 6 7 8 ------23. Angry 1 2 3 4 5 6 7 8 ------24. Bored 1 2 3 4 5 6 7 8 ------25. Clear 1 2 3 4 5 6 7 8 ------26. Considerable 1 2 3 4 5 6 7 8 ------27. Contrary 1 2 3 4 5 6 7 8 ------28. Cooperative 1 2 3 4 5 6 7 8 ------29. Deceitful 1 2 3 4 5 6 7 8 ------30. Devoted 1 2 3 4 5 6 7 8 ------31. Easy to get along 1 2 3 4 5 6 7 8 with ------32. Friendly 1 2 3 4 5 6 7 8 ------33. Good-natured 1 2 3 4 5 6 7 8 ------34. Hostile 1 2 3 4 5 6 7 8 ------

35. Irresponsible 1 2 3 4 5 6 7 8 ------36. Irritable 1 2 3 4 5 6 7 8 ------37. Lazy 1 2 3 4 5 6 7 8

128

------38. Loving 1 2 3 4 5 6 7 8 ------39. Mean 1 2 3 4 5 6 7 8 ------40. Rude 1 2 3 4 5 6 7 8

129

Mania Rating Scale

Elevated Mood 0 Absent. 1 Mildly or possibly increased. 2 Definite subjective elevation; optimistic, self-confident; cheerful; appropriate to content. 3 Elevated, inappropriate to content; humorous. 4 Euphoric; inappropriate laughter; singing.

Irritability 0 Absent. 2 Subjectively increased. 4 Irritable at times during interview; recent episodes of anger or . 6 Frequently irritable during interview; short, curt, throughout. 8 Hostile, uncooperative; interview impossible.

Content 0 Normal. 2 Questionable plans, new interest. 4 Special project (s); hyper-religious. 6 Grandiose or paranoid ideas; ideas of reference. 8 Delusions; hallucinations

Sleep 0 Reports no decrease in sleep. 1 Sleeping less than normal amount by up to one hour. 2 Sleeping less than normal by more than one hour. 3 Reports decreased need for sleep. 4 Denies need for sleep.

Language-Thought Disorder 0 Absent. 1 Circumstantial; mild distractibility; quick thoughts. 2 Distractible; loses goal of thought; changes topics frequently; racing thoughts. 3 Flight of ideas; tangentiality; difficult to follow; rhyming; echolalia. 4 Incoherent; communication impossible.

Speech (Rate and Amount) 0 No increase. 2 Feels talkative. 4 Increased rate or amount at times, verbose at times. 6 Push; consistently increased rate and amount; difficult to interrupt. 8 Pressured; uninterruptible, continuous speech.

130

Increased Motor Activity 0 Absent. 1 Subjectively increased. 2 Animated; gestures increased 3 Excessive energy; hyperactive at times; restless (can’t be calmed). 4 Motor excitement; continuous hyperactivity.

Sexual Interest 0 Absent. 1 Mildly or possibly increased. 2 Definite subjective increase. 3 Spontaneous sexual content; elaborates on sexual matters; hypersexual by self- report. 4 Overt sexual acts (towards interviewer).

Disruptive-Aggressive Behavior 0 Absent, cooperative. 2 Sarcastic, loud at times, guarded. 4 Demanding, threatening. 6 Threatens interviewer; shouting; interview difficult. 8 Assaultive; destructive; interview impossible.

Appearance 0 Appropriate dress and grooming. 1 Minimally unkempt. 2 Poorly groomed; moderately disheveled; overdressed. 3 Disheveled; partly clothed; garish make-up. 4 Completely unkempt; decorated; bizarre garb.

Insight 0 Present; admits illness; agrees with need for treatment. 1 Possibly ill. 2 Admits behavior change but denies illness. 3 Admits possible behavior change. 4 Denies any behavior change.

131

Child Depression Rating Scale – Revised (CDRS-R)

Depressed Feelings 0 Unable to rate. 1 Occasional feelings of unhappiness, which quickly disappear. 2 3 Describes sustained periods of unhappiness that last 1 or more hours. May report shorter periods that occur several times a week. Unhappiness may be ascribed to everyday reality events but not to major life events. 4 5 Feels unhappy an entire day without a major precipitating cause. 6 7 Feels unhappy most of the time. Accompanied by psychic , e.g., “can’t stand it”.

Depressed Affect 0 Unable to rate. 1 Definitely not depressed. Facial expression and voice animated during interview. 2 Mild suppression of affect. Some loss of spontaneity. 3 Overall loss of spontaneity. Looks distinctly unhappy during parts of interview. May still be able to smile when discussing non-threatening areas. 4 5 Moderate restriction of affect throughout most of interview. Has longer and more frequent periods of looking distinctly unhappy. 6 7 Severe. Looks sad, withdrawn. Minimal verbal interaction throughout interview. Cries or may appear tearful

Weeping 0 No information, unable to rate. 1 Normal for age. 2 Suggestive statements that child cries, or feels like crying, more frequently than peers. 3 Child cries slightly more than peers. 4 5 Child cries or feels like crying frequently (several times a week). Admits to crying without knowing reasons why. 6 7 Cries nearly every day.

Irritability 0 Unable to rate. 1 Rare. 2 Occasional. 3 Several times a week for short period. 4 5 Several times a week for longer periods. 6 7 Constant. 132

Capacity to have fun 0 Unable to rate. 1 Interest and activities realistically appropriate for age, personality, and social environment. Shows no appreciable changes with present illness. Any feelings of transient. 2 3 Describe some activities realistically available several times a week but not on a daily basis. Shows interest but not . May express some episodes of boredom more than once a week. 4 5 Is easily bored. Complains of “nothing to do”. Participates in structured activities with a “going through the motions” attitude. 6 Shows no enthusiasm or real interest. Has difficulty naming activities. May express interest primarily in activities that are (realistically) unavailable on a daily or weekly basis. 7 Has no initiative to become involved in any activities. Primarily passive. Watches others play or watches television but shows little interest in program. Requires coaxing and/or pushing to get involved in activity.

Social Withdrawal/Peer Problems 0 Unable to rate. 1 Enjoys friendships with peers at school and at home. 2 3 May not actively seek out friendships but waits for others to initiate a relationship or may occasionally reject opportunities to play without a desirable alternative. 4 5 Frequently rejects opportunities for desirable interactions with others and/or sets up situations where rejection is inevitable. 6 7 Does not currently relate to other children. States he or she has “no friends” or actively rejects new or former friends.

Appetite or Eating Patterns 0 Unable to rate. 1 No problem or change in eating pattern. 2 3 Mild change from usual eating habits with onset of current behavioral problems. 4 5 Definitely anorexic. Is not hungry most of the time or has excessive food intake since onset of current behavioral problems or marked increase in appetite.

133

Sleep 0 Unable to rate 1 No (or occasional) difficulty (Goes to sleep within ½ hour or less) 2 3 Frequently has mild difficulty with sleep 4 5 Moderate difficulty with sleep nearly every night (If applicable, circle number indicating time of difficulty) 1. Initial 2. Middle 3. Early Morning Awakening

Hypoactivity 0 Unable to rate. 1 None. 2 3 Mild. Slow body movement 4 5 Moderate. Definite motor retardation. 6 7 Severe. Sits of lies in bed most of the time.

Tempo of Language 0 Unable to rate. 1 Normal. 2 Slow. 3 Slow; delays interview 4 5 Severe. Low; marked interference with interview.

Excessive Fatigue (consider age and activities of child) 0 Unable to rate. 1 No unusual complaints of “ tired” during the day. 2 3 Complaints of fatigue which seem somewhat excessive and not related to boredom 4 5 Daily complaints of feeling tired 6 7 Complaints of feeling tired most of the day. May voluntarily take long naps without feeling refreshed. Interfere with play activities.

134

Guilt 0 Unable to rate. 1 Does not express any undue feelings of guilt appears appropriate to precipitating event. 2 3 Exaggerates guilt and/or out of proportion to event described. 4 5 Feels guilty over things not under his or her control. Guilt is definitely pathological. 6 7 Severe delusions of guilt.

Schoolwork (consider change in performance and change in concentration) 0 Unable to rate. 1 Performance consistent with ability. 2 Minor interference with some subjects. 3 Decrease in school performance. 4 5 Major interference in most subjects. 6 7 No motivation to perform.

Physical Complaints (Complaints on a non-organic basis) 0 Unable to rate. 1 Occasional complaints. 2 3 Complaints appear mildly excessive. 4 5 Complains daily or some interference with the ability of the child to function. 6 7 Preoccupied with aches and ; interferes with play activities several times a week.

Self Esteem 0 Unable to rate. 1 Describes self in primarily positive terms. 2 3 Describes self with one important area where child feels deficit. 4 5 Describes self in preponderance of negative terms or gives bland answers to questions. 6 7 Refers to self in derogatory terms. Reports that other children refer to him/her frequently by using derogatory nicknames and child puts him/her self down.

135

Morbid Ideation 0 Unable to rate. 1 None. 2 3 Has some morbid thoughts, all of which relate to a real event but seem excessive 4 5 Preoccupied with morbid thoughts several times a week. Morbid thoughts extend beyond external reality 6 7 Preoccupied with death themes or morbid thoughts that are elaborate, extensive, bizarre and occur on a daily basis.

Suicidal Ideation 0 Unable to rate. 1 Understands the word “suicide” but does not apply the term to self. 2 Sharp denial of suicidal thoughts. 3 Has thoughts about suicide, usually when angry. 4 5 Has recurrent thoughts of suicide. If moderately depressed, strongly denies thinking about suicide. 6 7 Has made suicide attempt within the last month or is actively suicidal.

136

Home & Community Social Behavior Scales

Please fill in the bubble that corresponds with your beliefs about your child. Social Competence Never Sometimes Frequently

1. Cooperates with peers 1 2 3 4 5 2. Makes appropriate transitions between 1 2 3 4 5 different activities 3. Completes chores without being 1 2 3 4 5 reminded 4. Offers to help peers when needed 1 2 3 4 5 ------5. Participates effectively in family or 1 2 3 4 5 group activities 6. Understands problems and needs of 1 2 3 4 5 peers 7. Remains calm when problems arise 1 2 3 4 5 8. Listens to and carries out direction from 1 2 3 4 5 parents of supervisors ------9. Invites peers to participate in activities 1 2 3 4 5 10. Asks appropriately for clarification of 1 2 3 4 5 instructions. 11. Has skills or abilities that are admired 1 2 3 4 5 by peers 12. Is accepting of peers 1 2 3 4 5 ------13. Completes chores or other assigned 1 2 3 4 5 tasks independently 14. Completes chores or other assigned 1 2 3 4 5 tasks on time 15. Will give-in or compromise with peers 1 2 3 4 5 when appropriate 16. Follows family and community rules 1 2 3 4 5 ------

137

Never Sometimes Frequently 17. Behaves appropriately in a variety of 1 2 3 4 5 school settings 18. Asks for help in an appropriate manner 1 2 3 4 5 19. Interacts with a wide variety of peers 1 2 3 4 5 20. Produces work of acceptable quality for 1 2 3 4 5 his/her ability level ------21. Is good at initiating or joining 1 2 3 4 5 conversations with peers 22. Is sensitive to the feelings of others 1 2 3 4 5 23. Responds appropriately when corrected 1 2 3 4 5 by parents or supervisors 24. Controls temper when angry 1 2 3 4 5 ------25. Enters appropriately into ongoing 1 2 3 4 5 activities with peers 26. Has good leadership skills 1 2 3 4 5 27. Adjusts to different behavioral 1 2 3 4 5 expectations 28. Notices and compliments 1 2 3 4 5 accomplishments of others ------29. Is appropriately assertive when he/she 1 2 3 4 5 needs to be 30. Is sought out by peers to join in activities 1 2 3 4 5 31. Shows self-restraint 1 2 3 4 5 32. Is “looked up to” or respected by peers 1 2 3 4 5 ------Antisocial Behavior

1. Blames others for his/her problems 1 2 3 4 5 2. Takes things that are not his/hers 1 2 3 4 5 3. Is defiant to parents or supervisors 1 2 3 4 5 4. Cheats on schoolwork or in games 1 2 3 4 5 ------5. Gets into fights 1 2 3 4 5 6. Is dishonest; tells lies 1 2 3 4 5 7. Teases and makes fun of others 1 2 3 4 5 8. Is disrespectful or “sassy” 1 2 3 4 5 ------9. Is easily provoked; has a “short fuse” 1 2 3 4 5 10. Ignores parents or supervisors 1 2 3 4 5 11. Acts as if he/she is better than others 1 2 3 4 5 12. Destroys or damages other’s property 1 2 3 4 5 ------138

Never Sometimes Frequently 13. Will not share with others 1 2 3 4 5 14. Has temper outbursts or tantrums 1 2 3 4 5 15. Disregards feelings or needs of others 1 2 3 4 5 16. Is overly demanding of attention from 1 2 3 4 5 adults ------17. Threatens peers; is verbally aggressive 1 2 3 4 5 18. Swears or uses obscene language 1 2 3 4 5 19. Is physically aggressive 1 2 3 4 5 20. peers 1 2 3 4 5 ------21. Whines and complains 1 2 3 4 5 22. Argues and quarrels with peers 1 2 3 4 5 23. Is difficult to control 1 2 3 4 5 24. Bothers and annoys others 1 2 3 4 5 ------1. Gets in trouble at school or in the 1 2 3 4 5 community 2. Disrupts ongoing activities 1 2 3 4 5 3. Is boastful; brags 1 2 3 4 5 4. Is not dependable 1 2 3 4 5 ------5. Is cruel to other persons or to animals 1 2 3 4 5 6. Acts impulsively without thinking 1 2 3 4 5 7. Is easily irritated 1 2 3 4 5 8. Demands help from peers 1 2 3 4 5 ------Additional Information

Please use the following space to provide any additional information about this child that you believe would be useful for understanding his or her social behavior:

139

Treatment Beliefs Questionnaire – Parent Form (TBQ-P)

Directions: To what extent do you agree with the following statements on a scale of 1 (Strongly Disagree) to 5 (Strongly Agree). Fill in N/A if the item is not applicable to you.

N/A Strongly Disagree Neutral Agree Strongly Disagree Agree

9. For my child to be healthy, s/he must take medication. 0 1 2 3 4 5 10. For my child to be healthy, s/he needs therapy/counseling. 0 1 2 3 4 5 11. It worries me that my child takes medication. 0 1 2 3 4 5 12. Medicine is necessary to control my child’s mood/behavior. 0 1 2 3 4 5 13. Therapy/counseling is necessary to control my 0 1 2 3 4 5 child’s mood/behavior. 14. My child’s medicines are effective. 0 1 2 3 4 5 15. I worry about my child becoming too dependent on his/her medication. 0 1 2 3 4 5 16. I worry about side effects of my child’s medication 0 1 2 3 4 5 17. I worry about long-term negative side effects of my child taking medication. 0 1 2 3 4 5 18. I feel that part of my child’s personality might be lost by taking medication. 0 1 2 3 4 5 19. I would rather deal with my child’s mood/behavior problems than have him/her sedated by medication. 0 1 2 3 4 5 20. Without medicines, doctors would be less able to cure people. 0 1 2 3 4 5 21. Most medicines are addictive. 0 1 2 3 4 5 22. People who take medicines should stop their treatment every now and then. 0 1 2 3 4 5 23. Medicines only work if taken as prescribed. 0 1 2 3 4 5 24. Doctors use too many medications. 0 1 2 3 4 5 25. Most medicines are safe. 0 1 2 3 4 5 26. If doctors had more time with patients they would prescribe fewer medicines. 0 1 2 3 4 5 27. Herbal supplements are safer than prescribed medication. 0 1 2 3 4 5

140

N/A Strongly Disagree Neutral Agree Strongly Disagree Agree

28. My child needs medication to get along with others. 0 1 2 3 4 5 29. My child needs medication to help him/her function on a daily basis. 0 1 2 3 4 5 30. Without medication, my child would be okay. 0 1 2 3 4 5 31. My child will need medication for the rest of his/her life. 0 1 2 3 4 5 32. My child will need therapy/counseling for the rest of his/her life. 0 1 2 3 4 5 25. My child can control his/her mood/behavior without medications. 0 1 2 3 4 5 26. My child can control his/her mood/behavior 0 1 3 4 5 5 without therapy/counseling 27. My child’s illness is very serious. 0 1 2 3 4 5 28. If my child’s illness does not improve, s/he will have many future problems. 0 1 2 3 4 5 29. I can tolerate my child’s medicine-related side effects because the medicine is effective. 0 1 2 3 4 5 30. Even though my child’s medication is expensive, s/he needs it to be healthy. 0 1 2 3 4 5 31. Even though therapy sessions are expensive my child needs them to be healthy. 0 1 2 3 4 5 32. I understand why my child is taking medication. 0 1 2 3 4 5 33. I understand about the possible side effects of my child’s medication. 0 1 2 3 4 5 34. I my child’s psychiatrist. 0 1 2 3 4 5 35. I trust my child’s psychologist/therapist. 0 1 2 3 4 5 36. My child’s psychiatrist and therapist agree on treatment needs for my child. 0 1 2 3 4 5 37. My child’s doctor consults me about treatment decisions for my child. 0 1 2 3 4 5

141