I».
3.: .
Q w.

5:,
t .
.9:

rent-‘- . .
.3, E: , .V
b emWezwm.

Mm
u n e 4
.. wg‘mm .

p
a...

.ov.9. ;

11:? {Edy
.11}.

. I3...- J

..

 

, it". .
. av, .I. ..
.i . «magi
b...
53...
f:

 

 

 

‘ , .. 4
: . . . .
,1 v
r
:33
3.... 13..
z i

..s

Amman . Name. at.

.9 s 3.4. .1 . A

. 2.. e .23. 2. a...km.,naL..Wii.

.5 Pain-554‘ “.71.qu. « 3:93... .41 » 1|
\ H£51.. 3%: u... is.

.3. 52!... .. .5. x
>hF‘l:..l..:!....s AM \‘1
1. 13.»; 5,2. .l n
'46-‘55...) 5.. a
:1} :3

I;
l. S. ‘P n) r
fiz¥.271
0 .31.}: .

l5.»

 

“we. .. .2.” ., , , .. «#3: ..
‘ . . # ,u .

lllllllllllllllllllllllllllllllllllll’lllllllllllllllllllll

31293 0177182267

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the
dissertation entitled
THE IMPACT OF REACTANCE ON THE TREATMENT PROGRESS
OF INCARCERATED MALE ADOLESCENT OFFENDERS
presented by

Michele Ollie Poorman

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Psyghg 19 g):

 

[dam/A jam/c

M or ofessor

Date 5/ 6(1/ 9 9

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINE return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1M WMpfiS—p.“

 

THE IMPACT OF REACTANCE ON THE
TREATMENT PROGRESS OF
INCARCERATED MALE ADOLESCENT OFFENDERS
By

Michele Ollie Poorman

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology

1999

ABSTRACT

THE IMPACT OF REACTANCE ON THE TREATMENT PROGRESS OF
INCARCERATED MALE ADOLESCENT OFF ENDERS

By

Michele Ollie Poorman

Psychological reactance is an internal motivational force to regain lost
or threaten freedoms (Brehm, 1966) that has both individual difference (trait
reactance) and situation specific (state reactance) characteristics. Both the
theoretical and empirical literatures have evaluated the effect of reactance on
treatment progress for predominantly adult populations and have been
inconsistent in their conclusions regarding the differential impact of trait and
state reactance. In addition, the implications of increased levels of reactance
during adolescence on adolescent treatment outcomes need to be clarified.
Accordingly, this study investigated the nature of the relationship between
trait and state reactance and whether both aspects of reactance interact in
producing change in adolescents’ psychological and behavioral problems.
The participants were 29 males, 13 to 18 years of age, in a psychiatric
residential treatment program for adjudicated adolescent boys. Trait
reactance was assessed using the Therapeutic Reactance Scale (Dowd et al.,

1991) and state reactance was assessed using a measure of satisfaction with

the institution and perceived need for treatment. Multiple outcome measures
were completed by both adolescent residents and program staff. As
expected, trait reactance was more stable than state reactance over time
(Hypothesis 1), and trait and state reactance were moderately correlated,
although the level of trait reactance restricted the range of state reactance
(Hypothesis 2). A posteriori identification of an “unstable” state reactance
group required modification of Hypothesis 3. Nevertheless, for some
outcome variables, findings were as expected with state reactance
moderating the effect of trait reactance on treatment progress (Hypothesis 3).
Support for this hypothesis varied by type of pathology (internalizing versus
externalizing) and informant (self versus staff). The results have

implications for filtUI'C research and clinical practice.

For Willoughby
for enriching my life in more ways

than I could ever imagine

iv

ACKNOWLEDGEMENTS

This work would not have been possible without the assistance and
support of many people. First, I would like to thank my dissertation chair,
Susan Frank, for her valuable contributions and guidance. I would also like
to thank the other members of my committee, Antonio Nunez, Gary Stollak,
Ellen Strommen, and Daniel Field, for their constructive feedback.

This project would not have been possible without the cooperation of
the staff and residents at Rivendell Youth Center. I would especially like to
thank Diana Yount for her trust, support, and friendship. I am also grateful
to the residents who surprised me on a daily basis with their candor,
sensitivity, and generosity.

I would also like to thank my family and friends for their support. In
particular, I would like to thank my mother for her continuous love and
encouragement throughout graduate school, my father for never doubting
that I would finish, my sister for moving closer and for babysitting in a
pinch, my brother for laughs and phone calls, and Laurie Van Egeren for her
ideas and sense of perspective. Finally, I would like to thank my husband,
Jerry Urquhart, for being my best friend and biggest supporter. His patience,

flexibility, love, and belief in me have meant more than I can say.

TABLE OF CONTENTS

LIST OF TABLES ............................................................ ix
LIST OF FIGURES ........................................................... xi
INTRODUCTION ............................................................ l
Reactance ...................................................................... 4
State versus Trait Components of Reactance ........................ 5
Personality Characteristics Associated with Reactance ............. 6
Sex Differences in Reactance .......................................... 7
Reactance and Outcome ...................................................... 7
Theoretical Literature ................................................... 8
Research Literature ...................................................... 9
Trait Reactance ........................................................ 16
State Reactance ........................................................ 18
Trait and State Reactance ............................................. 19
Adolescent versus Adult Reactance ........................................ 20
Developmental Considerations ......................................... 21
Referral Source ........................................................... 22
Research on Adolescent Reactance and Personality 22
Characteristics .......................................................
Research on Adolescent Reactance and Outcome ................... 23
Juvenile Offenders, Outcome, and Reactance ........................ 24
The Proposed Study .......................................................... 26
Hypothesis 1 .............................................................. 27
Hypothesis 2 .............................................................. 28
Hypothesis 3 .............................................................. 29
Specific Predictions ...................................................... 31
METHOD ...................................................................... 31
Participants ..................................................................... 31
The Program ................................................................... 36
Measures ....................................................................... 38
Adolescent Reactance ................................................... 38
Trait Reactance ........................................................ 38
State Reactance ........................................................ 39
Functional Impairment ................................................... 41
Staff Report ............................................................ 41

vi

Adolescent Report ..................................................... 42

Behavior Problems ....................................................... 43
Staff Report ............................................................ 43
Adolescent Report ..................................................... 45
Staff Behavioral Observations ....................................... 46

Procedures ....... - .............................................................. 49
RESULTS ..................................................................... 54

Hypothesis 1: Comparing the Stability of Trait and State Reactance. 55
Hypothesis 2: Evaluating the Relationship Between Trait and State 58

Reactance ...............................................................
Classification of Trait and State Reactance Groups ............... 60
Hypothesis 3: Investigating State Reactance as a Moderator 64

Between Trait Reactance and Treatment Progress ...............
Collapsing the Reactance Groups ...................................... 64
Revising the Specific Predictions for hypothesis 3 .................. 65
Testing For Confounding Variables ................................... 66
Reducing the Outcome Variables ..................................... 68

Self and Staff Report .................................................. 68

Behavioral Observations ............................................. 69
Reactance and Changes in Outcome Variables ...................... 74

Intemalizing Problems ................................................ 79

Extemalizing Problems ............................................... 84

Total Problems ......................................................... 87
Treatment Group as a Between Subjects Variable .................. 90

DISCUSSION ................................................................. 97
Findings for Hypothesis 1 ................................................... 98
Findings for Hypothesis 2 ................................................... 99
Findings for Hypothesis 3 ................................................... 100
Self versus Staff Report ................................................. 101
Intemalizing Versus Extemalizing Problems ........................ 102
The Moderating Effect of State and Trait Reactance on 104

Treatment Progress ...................................................
Direct Effects of Trait and State Reactance on Treatment 107

Progress ...............................................................
Impact of Trait and State Reactance on Change for the Three 109

Treatment Groups ...................................................

Limitations and Future Directions .......................................... 110
Conclusions .................................................................... 1 12

vii

Appendix A: Trait Reactance ............................................... l 16

Appendix B: State Reactance ............................................... 119
Appendix C: Achenbach-Staff Report .................................... 121
Appendix D: Achenbach-Self Report ..................................... 124
Appendix E: CBRF .......................................................... 127
Appendix F: lnforrned Assent Agreement ............................... 129
Appendix G: Informed Consent Agreement ............................. 132
Appendix H: Self Report of Intemalizing Problems ..................... 135
Appendix I: Staff Report of Intemalizing Problems ..................... 138
Appendix J: Staff Report of Extemalizing Problems .................... 140
Appendix K: Self Report of Total Problems .............................. 142
LITERATURE CITED ...................................................... 145

viii

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

Table 9:

LIST OF TABLES

Percentages of Residents with DSM-IV Axis I Diagnoses... 32

Percent of Residents Taking Each Type of Medication at
Admission (Time 1) and 3 Months Post Admission (Time

2) ..................................................................... 33
Distribution of Crime Convictions Among the Six Offense

Categories .......................................................... 34
Timeline for Data Collection .................................... 53

Ranges, Means, and Standard Deviations for Trait and
State Reactance .................................................... 54

Intercorrelations Among Trait and State Reactance and the
Means and Standard Deviations for Trait and State
Reactance ........................................................... 59

Intercorrelations Among State Reactance and Trait
Reactance ........................................................... 60

Crosstabulation for Trait and State Reactance ................. 63

Crosstabulation for Trait and State Reactance Using
Revised Classifications ........................................... 64

Table 10: Demographic Variable Tested for Differences Among the

State and Trait Reactance Groups ............................... 67

Table 11: Interrater Agreement at Time 1 for Achenbach and

F ISCA Variable ................................................... 70

Table 12: Interrater Agreement at Time 2 for Achebach Variables... 71

Table 13: Interrater Agreement at Time 2 for F ISCA Variables ...... 72

ix

Table 14:

Table 15:

Table 16:

Table 17:

Significant Differences Between Groups for CBRF
Variables ........................................................... 73

Number of Residents Without Pathology at Time 1... 74

F -tests Associated with the Trait Reactance Group x State
Reactance Group x Change ANOVA’s Using Self Report
of Change .......................................................... 76

F-tests Associated with the Trait Reactance Group x State
Reactance Group x Change ANOVA’S Using Staff Report

of Change .......................................................... 77

Figure 1:

Figure 2:

Figure 3:

Figure 4:

Figure 5:

Figure 6:

Figure 7:

Figure 8:

Figure 9:

LIST OF FIGURES

Trait Reactance at Time 1, 2, and 3 and State Reactance at
Time 1, 2, 3, and 4 for All Participants in the Study......... 56

Distributions of the Standard Deviations for Trait and
State Reactance ................................................... 57

State Reactance at Time 1, 2, 3, and 4 for Low, Moderate,
High, and Unstable State Reactance Groups .................. 62

Interactions Between State Reactance Group and Change
Within the High Trait Reactance Group for Self Report of
Intemalizing Problems ........................................... 80

Interactions Between State Reactance Group and Change
Within the Moderate Trait Reactance Group for Self
Report of Intemalizing Problems .............................. 81

Interaction Between State Reactance Group and Change
Within the High and Moderate Trait Reactance Groups
for Staff Report of Somatic Problems ......................... 83

Interaction Between Trait Reactance and Changes in Self
Report of Aggressive Problems and Attention Problems... 85

Interaction Between Trait Reactance and Changes in Staff
Report of Impairment in Undercontrolled Behavior. . . . . 86

Interaction Between Trait Reactance and Changes in Self
Report of Total Problems ....................................... 88

Figure 10: Interaction Between State Reactance and Changes in

Self Report of Total Impairment .............................. 98

Figure 11: Interaction Between Group, State Reactance, and

Changes in Self Report of Intemalizing Problems .......... 91

xi

Figure 12: Interaction Between Group, State Reactance, and
Changes in Self Report of Anxiety/Depression Problems. 92

Figure 13: Interaction Between Group, State Reactance, and
Changes in Self Report of Total Problems ................... 93

Figure 142 Interaction Between Group, State Reactance, and
Changes in Self Report of Thought Problems ............... 94

xii

INTRODUCTION

Both theoretical and empirical literatures have evaluated the effect of
resistance on treatment process and outcome. However, the concept of
resistance is often broadly defined or used to describe many different
patterns of behavior. "Reactance" is one dimension of resistance defined by
Brehm (1966) as a motivational force to restore lost or threatened freedoms.
Like resistance, reactance can be both situation-specific and a characteristic
of personality. Reactance levels can be placed along a continuum from
abnormally low to abnormally high. Low reactant individuals tend to
follow directions without resistance, tolerate others making decisions for
them, respond in a non-defensive manner, accept interpretation, and often
seek direction from others. However, abnormally low reactant individuals
have been more negatively characterized as helpless (Dowd and Sanders,
1994). In contrast, high, especially abnormally high, reactant individuals
have intense needs to maintain autonomy, are resistant to external
influences, often refuse interpretation, and appear dominant.

Empirical studies of reactance theory have predominantly focused on
adult populations and evaluated the relationship between reactance and

other personality dimensions. It was not until recently that research has

1

focused on the impact of reactance on the effectiveness of interventions.
However, most of this research has focused on the effect of reactance on
treatment outcome for college students from non-clinical populations (i.e.,
procrastinators, athletes, smokers, individuals with sleep problems or test
anxiety). There has been less focus on how reactance affects the treatment
of clinical populations.

Both researchers and clinicians have been inconsistent in their
conclusions about the impact of reactance on treatment. Some researchers
have found that high reactant individuals show poor response to treatment
(Dowd, Hughs, Brockbank, and Halpain, 1988; Morgan, 1986; as cited in
Dowd, Milne, & Wise, 1991; Mulry, Fleming, and Gottschalk, 1994),
whereas others report that reactance does not affect treatment outcome
(Calvert, Beutler, & Crago, 1988; Hunsley, 1993; Swoboda, Dowd, and
Wise, 1990). Alternatively, theorists have argued that while reactance is
commonly seen as a negative, complicating factor in therapy, reactance can
be used in service of therapeutic change (Tennen, Press, Rohrbaugh, and
White, 1981).

While the adult literature does not point to any clear conclusions, at
least it has attempted to grapple with the impact of reactance on treatment

effectiveness. Almost no research has been done on adolescent reactance

2

and treatment outcome. This is consistent with other areas of adolescent .
outcome research. Outcome and process research in child and adolescent
psychotherapy has lagged behind research in adult psychotherapy
(Digiuseppe, Linscott, & Jilton, 1996). Given that reactance is thought to
be especially high in adolescence as well as old age (Tennen et al., 1981),
studying the impact of reactance on treatment effectiveness appears to be
particularly relevant for adolescents.

Frank, J ackson-Walker, Marks, Van Egeren, Loop, and Olson (1998)
have shown that reactance does affect outcome for middle adolescent
inpatients. However, they found a curvilinear relationship between
reactance and outcome whereby high and low scores on a reactance measure
predicted different outcomes than moderate reactance. Patterns of
reactance-change relationships also varied as a function of adolescents‘ sex.
Frank et al.'s (1998) study highlights the importance of broadening the
conceptualization of the relationship between reactance and outcome to
include not only direct linear effect models but also more complex models
such as curvilinear models and moderated models.

The present study is designed to further the work by F rank and her
colleagues and to explore in more detail the nature of the relationship

between adolescent reactance and treatment effectiveness. In order to

3

understand the complexity of this relationship, it is necessary to review the
literature on reactance in general, adult reactance and treatment outcome,

and potential distinctions between adolescent and adult reactance.

Reactance

State versus Trait Components of Reactance

Brehm (1966) originally conceived reactance as a situational variable
(state reactance) with little emphasis on individual differences (trait
reactance). In more recent years, researchers and theorists generally agree
that the disinclination to be influenced by others (i.e., reactance) has both
state and trait properties (Beutler, Sandowicz, Fisher, and Albanese, 1996).
However, theoretical definitions of reactance differ in the degree to which
trait and state factors are implied. For example, Dowd et a1. (1991)
hypothesized that psychological reactance is in part an individual difference
variable that is relatively stable over time and across situation. However,
based on their research on the trait reactance measure, they concluded that
reactance is only partly a trait-like variable. Dowd and Walbrown (1993)
concluded that reactance is partly situational and partly due to individual
differences. Similarly, Beutler argued that reactance potential is an internal

motivational state that can be increased by characteristics of both the

4

individual and the situation.

PersonalitL Characteristics Associated with Reactance

Several researchers have studied the personality characteristics
associated with trait reactance. Generally, they found that high reactance
was correlated with aspects of personality that have a negative connotation.
For example, Beutler et a1. (1996) concluded that reactant individuals are
unusually angry, have strong needs for control, are impulsive, and are
avoidant of relationships with others. Dowd et al. (1994) found that high
reactant college students had personality styles characterized by a lack of
interest in making a good impression and independence. In addition, high
reactant male students were found to be intolerant of other's beliefs and
dominant. Joubert (1990) demonstrated that reactance positively correlated
with fear of failure and negatively correlated with self-esteem for females.
For both males and females, reactance positively correlated with loneliness
and negatively correlated with happiness. Hong and Faedda (1996) reported
that reactance was positively correlated with depression and trait anger and
slightly negatively correlated with religiosity and life satisfaction. Note
however that Hong and Giannakopoulos (1994) found no relationship

between reactance and life satisfaction in Australian adults.

5

While Dowd and Wallbrown (1993) also linked negative attributes to
reactance, they alluded to potentially positive qualities of high reactant
individuals. They found high reactant students to be defensive, aggressive,
dominant, autonomous, quick to take offense, non-affiliative and non-
supportive. But they speculated that high reactant people also would make
effective leaders, have confidence in their decisions, and would stick with
their goals. Hong and Faedda (1996) and Hellman and McMillin (1997)
reported that at least one subscale of Hong and Page’s (1989) measure of
reactance was positively correlated with self-esteem. These findings
demonstrate the inconsistencies in the literature about the personality

correlates of reactance and suggest that the relationship is complex.

Sex Differences in Reactance

There are inconsistencies in the literature about the relationship
between reactance and gender. Hong (1990; Hong & Page, 1989) studied
college undergraduates and found no sex differences in reactance. In
contrast, Dowd, Wallbrown, Sanders, and Yesenosky (1994), Courchaine,
Loucka, and Dowd (1995), and J oubert (1990) found that male college
students were significantly more reactant than female college students. It is

noteworthy that the research by Hong used a different measure of reactance

6

and was conducted using Australian subjects (as opposed to American). It
is possible that sex role socialization differs significantly in these countries

and that may account for some of the difference in the findings.

Reactance and Outcome

A significant body of research exists on the relative effectiveness of
different therapeutic techniques in producing positive treatment outcomes.
However, this research has not adequately addressed why different people
respond differently to the same intervention. A relatively small number of
psychotherapy researchers have attempted to identify client characteristics
that correlate with positive treatment outcome. Beutler and Crago (1991)
concluded that client characteristics accounted for the greatest source of
variance in client outcome. In addition, Beutler and Clarkin (1990)
identified reactance as a client characteristic that appeared promising for

predicting differential responsiveness to different types of psychotherapy.

Theoretical Literature
Most theorists appear to agree that, in general, a negative relationship
exists between reactance and prognosis. For example, Dowd and his

colleagues (1990; 1991; 1994) argued that low reactant individuals would
7

be especially compliant in the treatment process and should be expected to
improve more rapidly than high reactant clients. They hypothesized that
high reactant clients would engage in oppositional behavior and that
therapeutic progress would be slower and ultimate outcome poorer. More
specifically, they suggested that the most difficult clients were high reactant
clients with ego-syntonic symptoms whereas clients with low reactance or
ego-dystonic symptoms were less difficult.

In contrast, Kirmayer (1990) expressed concerns about extremely low
reactant individuals. He warned that the relative absence of reactance might
be considered a helpless stance and that a shift to higher reactance reflects
an improved prognosis. Similarly, Tennen et al. (1981) argue that reactance
can be used therapeutically to accomplish treatment goals. They
hypothesize that client’s reactance can the channeled by the therapist to

produce successful treatment outcomes.

Research Literature

Researchers investigated the relationship between reactance and
treatment outcome in order to validate the theoretical predictions. This
research can be divided into research on trait reactance, state reactance, and

both trait and state reactance. Most of the research on reactance and

8

outcome was done on college students from non-clinical samples. The
generalizability of this research to actual clients in treatment is not clear.
However, a review of the smaller number of studies with clinical
populations shows a similar pattern of results, with not all studies showing a
negative effect, but none demonstrating a positive effect of reactance on
treatment effectiveness.

Trait Reactance. Of the research with trait reactance reviewed below,

 

four studies found a negative relationship between trait reactance and
treatment process or outcome; three studies found that trait reactance did not
affect treatment outcomes, and seven found a more complex relationship
between reactance and treatment outcome.

Courchaine et a1. (1995) studied the impact of trait reactance,
counselor type, and counselor style of interpretation on positive feelings
about the counselor. They measured trait reactance using both the
Therapeutic Reactance Scale (Dowd et al., 1991) and the Merz Reactance
Questionnaire (Merz, 1983). They found that high trait reactant students
rated counselors as less positive, less expert, and less trustworthy than low
trait reactant students. They concluded that client personality variables
potentially had a greater effect on outcome than therapeutic techniques and

suggested that trait reactance was a prime individual difference variable

9

impacting treatment process and outcome.

Dowd, Hughes, Brockbank, Halpain, Seibel, and Seibel (1988)
studied the impact of trait reactance on the treatment of procrastination and
test anxiety. They predicted that trait reactance (measured by the
Therapeutic Reactance Scale; Dowd et al., 1991) would moderate the
relationship between treatment approach (i.e., restraining, refrarning, and
non-paradoxical interventions) and outcome. Contrary to predictions, none
of the treatment conditions were differentially effective for high and low
trait reactant students. However, they found that high trait reactant subjects
experienced less satisfaction with performance and more anxiety after
treatment than low trait reactant students.

Most empirical studies on the impact of trait reactance on treatment
outcome evaluated individual therapy approaches. However, Oliver,
Mattson, and Moore (1993) evaluated the effectiveness of a marital
enrichment program for seventeen couples who volunteered for the study.
Results indicated that only females demonstrated improvements in outcome
measures (i.e., marital satisfaction, dyadic consensus, cohesion, and
affection). However, high trait reactance (TRS; Dowd et al., 1991) was
negatively associated with improvement. They concluded that trait

reactance acted as an antagonist in regard to females' improvement in

10

perception of marital adjustment. While males‘ trait reactance had no
impact on treatment outcome (in general, they did not show improvement so
no relationship was possible), their trait reactance was found to increase
over the time of treatment. The researchers concluded that trait reactance,
while an important factor in determining the effectiveness of marital
interventions, had a differential impact on males and females in a marital
relationship.

One study evaluating a clinical sample found that reactance had a
negative impact on outcome. Morgan (1986; as cited in Dowd et al., 1991)
found that high trait reactant clients had higher no-show rates and were in
treatment longer than low trait reactant clients.

Three studies found that reactance had no significant effects on
treatment outcome. One study of a non-clinical population found that trait
reactance (as measured by the Therapeutic Reactance Scale; Dowd et al.,
1991) did not have an impact on perceived treatment acceptability (Hunsley,
1993). In a clinical population, Swoboda, Dowd, and Wise (1990) studied
the effect of different treatment approaches on depressed adults. Their
results indicated that trait reactance (TRS; Dowd et al., 1991) had no effect
on treatment outcome. They hypothesized that the lack of significant

findings may be due to the more severe client problems in their study.

11

Similarly, Calvert, Beutler, and Crago (1988) tested the assumption
that matching psychotherapy approach to theory relevant patient
characteristics (i.e., trait reactance) would result in improved treatment
outcome. They measured trait reactance using a subscale of the
Fundamental Interpersonal Relations Orientation Scale, Behavior (FIRO—B;
Schutz, 1959; as cited in Calvert et al., 1988). The subscale assessed desire
for interpersonal control and the extent to which an individual allows his or
her behavior to be controlled or influenced by others. Contrary to
prediction, matching patient trait reactance to therapist approach did not
affect outcome.

More complex relationships between trait reactance and outcome
were investigated by several researchers. Dowd, Trutt, and Watkins (1992)
hypothesized that individual differences in students' trait reactance (TRS;
Dowd etal., 1991) would interact with their responses to tentative and
absolute interpretations in predicting therapist ratings. Results
demonstrated that high trait reactant students more positively rated
therapists that used absolute (versus tentative) interpretation while low trait
reactant students showed the opposite correlation. This finding was in the
opposite direction from what was predicted but was similar to findings by

Courchaine et al. (1995).
12

Kelly and Nauta (1997) studied whether trait reactance moderated the
relationship between therapeutic approach and thought suppression. They
conducted two studies of college students in introductory psychology
classes and measured reactance using Dowd’s (1991) trait reactance scale.
They concluded that reactance moderated the effects of thought control
strategies. In particular, high reactant students in the thought suppression
group reported more negative outcomes (e.g., feeling out of control, feeling
disturbed by their thoughts) than high reactant students in the thought
expression group. The opposite finding was reported for low reactant
students with students in the expression group reporting more negative
outcomes than students in the suppression group.

Horvath and Goheen (1990) studied adults with sleep problems. The
results indicated that more- and less- trait reactant individuals (as measured
by the TRS; Dowd et al., 1991) benefited similarly from different types of
treatment (i.e., paradoxical and behavioral interventions). However, trait
reactance appeared to moderate the relationship between type of treatment
and improvement after treatment. For example, for clients exposed to
paradoxical interventions, more trait reactant clients continued to improve
beyond active treatment, whereas less trait reactant clients tended to lose

some of the initial gains. In contrast, clients exposed to behavioral

13

treatment showed the opposite results.

Carter and Kelly (1997) found similar results when studying anxiety
in athletes. They studied differential impact of traditional and paradoxical
imagery interventions aimed at anxiety reduction on athletes with high or
low trait reactance ( TRS; Dowd et al., 1991). They hypothesized that high
reactant athletes would show less anxiety and better performance with
paradoxical imagery while low reactant athletes would show better
outcomes with the traditional confidence building imagery. They found
partial support for their predictions. In particular, low reactant athletes
reported more anxiety and less confidence with paradoxical imagery.
However, they found no differences in levels of anxiety for the high reactant
athletes in the different intervention groups. The opposite finding was
found when predicting performance. The low reactant group did not show
any differences in performance in the different intervention groups.
However, the high reactant group performed best with no intervention (i.e.,
the control group). They concluded that trait reactance moderates the effect
of treatment strategy on athletes’ anxiety and performance, although the
direction of their findings was difficult to interpret.

Two clinical studies evaluated the impact of trait reactance as a

moderator of relationships between treatment characteristics and treatment

14

outcomes. Graybar, Antonuccio, Boutilier, and Varble (1989) found that
trait reactance (TRS; Dowd et al., 1991) moderated the relationship between
smoking cessation and doctor tone or amount of advice. For high trait
reactant patients, a low amount of negatively toned advice was most
effective in reducing smoking. In contrast, for low trait reactant patients a
high amount of either positively or negatively toned advice produced the
best outcomes.

Beutler, Mohr, Grawe, Engle, and McDonald (1991) suggested that
therapists working with high trait reactant individuals should avoid direct
confrontation and interpretation and use non-directive approaches that
negotiate solutions, give choices, and/or empower individuals. Their own
research supported these predictions. They measured trait reactance using
nine instruments that described characteristics of highly reactant individuals
(e.g., dominance, sensitivity to interpersonal control, and competitiveness).
They found that high trait reactance was negatively correlated with
improvement at outcome when directive treatments were used and
positively correlated with improvement at outcome when non-directive
treatments were used.

The research does not present a clear picture regarding the impact of

trait reactance on outcome. While half of the studies found either no

15

relationship or a negative relationship between reactance and outcome, half
of the reviewed studies indicated a more complex relationship. In
particular, treatment approach moderated the relationship between trait
reactance and outcome.

State Reactance. With a non-clinical population, Shoham-Salomon,
Avner, and Neeman (1989) investigated the impact of state reactance as a
moderator of the relationship between treatment approach and outcome.
State reactance was experimentally manipulated by placing participants in
different experimental conditions. For example, in the high state reactant
condition participants were told they could choose the treatment method and
then were assigned to the non-preferred method without explanation. They
found that high state reactant students showed more improvement in
paradoxical interventions than in behavioral interventions. The opposite
relationship was found for low state reactant students. They concluded that
state reactance affected the relationship between treatment approach and
outcome.

Using clinical populations, two studies investigated the impact of
resistance on treatment outcome. Their operationalized definition of
resistance was analogous to state reactance. Chamberlain, Patterson, Reid,

Kavanagh, & F orgatch (1984) investigated the extent to which resistance
16

affected treatment outcome. They measured resistance using a coding
system designed to study client resistant behavior during therapy sessions.
They rated client's responses to therapist's direction as resistant if the client
interrupted, was unwilling to cooperate, confronted, changed the subject, or
refused to respond. Results showed that a reduction in resistance/state
reactance from initial to termination sessions was positively correlated with
treatment outcome.

Stoolmiller, Duncan, Bank, and Patterson (1993) studied patterns of
change in client resistance during parent training therapy for mothers of
conduct problem children. Resistance to therapy was coded at admission,
rnid-therapy, and termination using the same coding system described above
(Chamberlain et al., 1984). They found that families with accelerating and
chronically high levels of resistance (i.e., state reactance) were likely to fail
in treatment. In addition, they found families with low levels of resistance
(i.e., state reactance) throughout treatment also appeared unlikely to benefit
from parent training. This finding supports Kirmayer's (1990) warnings
about the negative impact of extremely low reactance. These results
demonstrate a curvilinear relationship between reactance and outcome with
high and low reactance having a negative impact on outcome.

While there were fewer studies on the relationship between state

17

reactance and outcome than trait reactance and outcome, the findings
demonstrated a similar inconsistent pattern. In particular, one study found a
negative relationship, one found a curvilinear relationship, and one found a
moderated relationship between state reactance and outcome.

State and Trait Reactance. Only one study, using a non-clinical
population, evaluated the impact of both state and trait reactance on
outcome. Mulry et al. (1994) studied the effects of both state and trait
reactance on treatment approaches for procrastination. They assessed trait
reactance using the Therapeutic Reactance Scale (Dowd et al., 1991). State
reactance was experimentally manipulated similar to the aforementioned
study by Shoham-Salomon et a1. (1989). They found that high state reactant
students showed less improvement than low state reactant students. In
addition, they concluded that state reactance generalized across situations.
They manipulated state reactance using relevant or non-relevant situations.
In the relevant high state reactant condition, state reactance was
manipulated by assigning participates to the non-preferred choice without
explanation. In the non-relevant high state reactant condition, participants
attempted to solve impossible visual discrimination puzzles. Both relevant
and non-relevant manipulated state reactance produced the same impact on

outcome. Unlike state reactance, trait reactance did not impact treatment

18

outcome. They concluded that state reactance was more influential than
trait reactance in predicting treatment outcome.

The experimental manipulation used in this study may create an
artificial situation with conclusions partially lacking external validity.
Consequently, the differential impact of state and trait reactance should be
further investigated in clinical settings. In addition, since clinical settings
do not require the experimentally manipulation of state reactance, the
relationship between state and trait reactance more easily can be

investigated.

Adolescent versus Adult Reactance

All the aforementioned literature reported findings for adults. In
contrast, this research will focus on adolescents. Several studies indicate
that adolescents experience higher rates of reactance and consequently, may
show different personality profiles and responses to treatment. In particular,
Frank et al. (1998) suggested that adolescence is an especially reactant
phase of development and reported that adolescent scores on the adolescent
version of the TRS (Dowd et al., 1991) were higher and more varied than
those found with adults (Dowd et al., 1991). However, they were unable to

make generalizations to normal adolescents because they used an inpatient

l9

sample. Hong, Giannakopoulos, Laing, and Williams (1994) studied 18-40
year old normal adults and found levels of reactance decreased with age.
However, while their study included 18 to 22 year olds (often referred to as
late adolescence), they did not study early or middle adolescents.

Finally, Taylor, Adelman, and Kaser-Boyd (1985) investigated
adolescents' treatment reluctance (state reactance). They conceptualized
reluctance as adolescents' reports of refusal to participate, expressions of
ambivalence, complaints of coercion, avoidance, and dropping out of
therapy. They found that 79% of the adolescents in their study manifested
some form of reluctance (state reactance). The reasons for the high level of
reactance found among adolescents may be better understood by evaluating

adolescent development and how adolescents enter treatment.

Developmental Considerations

Dowd and Seibel (1990) viewed reactance as a vehicle for creating
and maintaining autonomy. Given that the developmental task of
adolescence is to establish autonomy, it would be expected that reactance
would vary with development. Adolescents may characteristically be
resistant to advice and direction from others and appear hostile,

argumentative, and oppositional to change because a primary focus of

20

adolescence is the establishment of independence (Diguiseppe et al., 1996).
Similarly, adolescents may be especially resistant when their autonomy is
threatened and they may attempt to gain control by being negative and

oppositional (Bow, 1988).

Referral Source

Most adolescents and children are brought to treatment by third party
referrals (i.e., parents, teachers, or courts). Third party referrals reduce
commitment to treatment and may produce reactance (Brehm, 1972).
Chamberlain et al. (1984) demonstrated that families who were agency
referred tended to have higher levels of resistance than those who were self-
referred. Similarly, Prandoni and Wall (1990) reported that court-mandated
evaluation or treatment aroused reactance. According to Diguiseppe et al.
(1996), the field of psychotherapy has always had difficulty with court-
mandated referrals and clients who attend therapy at the insistence of others.
They concluded that child and adolescent psychotherapy differs
significantly from adult psychotherapy because the client is not self-

referred.

21

Research on Adolescent Reactance and Personality Characteristics

Because adolescence is a developmental stage characterized by high
reactance, it is important to investigate whether reactance among
individuals correlates meaningfully with different personality profiles and
predicts differential responses to treatment. Frank et al. (1998) studied
adolescents and found similar relationships between reactance and
personality as found in the adult literature. In particular, reactance was
associated with counter-conventional, antisocial, narcissistic, non-
affiliative, and distrustful personality traits. As predicted, they found no sex
differences and no relationship between reactance and psychological

distress.

Research on Adolescent Reactance and Outcome

Frank et a1. (1998) also found that trait reactance affected hospital
length of stay and outcome at 3 and 6 months post discharge for adolescent
inpatients. They measured trait reactance using a modified version of the
Dowd et a1. (TRS; 1991) measure. Whereas higher trait reactance was
associated with longer length of stays, the relationships between trait
reactance and outcome were curvilinear. In particular, high and low trait

reactance predicted different outcomes (changes in functional impairment)

22

than moderate trait reactance for middle adolescents. However, results
differed for boys and girls. They found that both high and low trait
reactance were associated with a reduction in aggressive and depressive
behaviors for boys and a reduction in alcohol and substance abuse for both
boys and girls. Alternatively, moderate trait reactance was associated with
an increase or no change in these problem areas. In contrast, moderate trait
reactant girls showed more reduction in aggressive and depressive behaviors

than high and low trait reactant girls.

J_uvenile OffendersLOutcome, and Reactance

Juvenile offenders are a group of adolescents that are thought to be
particularly resistant to treatment. Dowd and Seibel (1990) observed that
antisocial clients are notoriously difficult to engage in treatment.
Incarcerated juvenile offenders are thought to be highly reactant because
limitations on personal autonomy are an inevitable fact of life for
incarcerated adolescents and the involuntary nature of the placement is
likely to increase reactance (Martin & Osgood, 1987). In addition,
reactance and antisocial traits in adolescence appear to go hand in hand
(Frank et al., 1998).

The majority of the studies on the effectiveness of treatment for

23

adolescent offenders have focused on the treatment program and/or the
treatment approach (Adams & Vetter, 1982; Hollin, 1993; Nir & Cutler,
1973; Velasquez & Lyle, 1985) and have produced contradictory results.
Kazdin (1987) reported that published studies on treatment of antisocial
adolescents are generally pessimistic regarding success rates. In contrast,
Basta and Davidson ( 1988) reviewed the literature on the treatment of
adjudicated juvenile offenders and concluded that the overall findings were
positive. Similarly, Hollin (1993) reviewed the literature on treatment
approaches used with juvenile offenders and concluded that there was
reason for optimism in the field of offender treatment. He described client
resistance as a significant banier to treatment success that must be more
clearly understood. Investigating the impact of offender reactance on the
effectiveness of treatment may help explain the inconsistencies in the

research findings.

The Proposed Study
The proposed study will investigate the relationship between
reactance and treatment progress for incarcerated male juvenile offenders.
The ambiguity regarding the differential effect of state and trait

characteristics of reactance suggests the need for research aimed at

24

clarifying the impact of both aspects of reactance. The only research which
simultaneously evaluated both aspects of reactance investigated non-clinical
populations and experimentally manipulated state reactance. Consequently,
it is important to investigate the impact of both types of reactance on
treatment outcome with clients in clinical settings.

For the present study, trait reactance will be assessed using a self-
report questionnaire developed by Dowd et a1. (1991) and modified for
adolescents by Frank et al. (1993). Of the three trait reactance measures
found in the literature (Dowd et al., 1991; Hong & Page, 1989; Merz, 1983),
Dowd’s measure has been most consistently used and is the only measure
validated on adolescent populations. The measure was based on Brehm’s
(1966) definition of psychological reactance.

Measuring state reactance or situation specific reactance is
significantly more difficult. Several measures have been developed to
observe reactance as it varies during treatment (Chamberlain, Davis,

F orgatch, Frey, Patterson, Ray, Rothchild, Trombley, 1986; Kavanagh,
Gabrielson, & Chamberlain, 1982; Mahalik, 1994). However, all of these
require ratings of micro behaviors within the treatment situation. This
micro level analysis is too intrusive and potentially non-representative when

studying reactance in residential treatment programs. Consequently, a

25

different method of assessing state reactance was identified.

Taylor et al. (1985) explored the components of adolescents'
situational reactance during treatment. The adolescents explained their
reactance as stemming from three sources: a negative attribute of therapy,
lack of need for treatment/denial of problems, and lack of choice in the
decision to receive treatment. Their findings indicated that state reactance
may be assessed by asking adolescents about their satisfaction with the
treatment program and their beliefs about whether they needed treatment.
Two measures by Osgood, Gruber, Archer, and Newcomb (1985) were used
to measure these constructs (Satisfaction with the Institution and Perceived

Need for Treatment).

Hypothesis 1

Given that personality is thought to originate within the individual
and be consistent across situations (Burger, 1990), it is expected that trait
reactance will be stable over time because it measures a personality
characteristic. In contrast, since the environment is constantly changing and
individuals experience each situations differently, state reactance will

change over time because it is by definition a situation-specific variable.

26

Hypothesis 2

Given that both state and trait reactance are dimensions of a larger
construct, it is expected that these variables will be moderately positively
correlated even though trait reactance will set limits on the range of state
reactance. Specifically, trait and state reactance may function analogously
to genotypes and phenotypes. Trait reactance is like an individual's
genotype. Each individual has a level of reactance that s/he is likely to
experience. This is the reactance potential that s/he brings to every
situation. State reactance is the “phenotype” or actual reactance the
individual experiences in each situation. Accordingly, high reactant
individuals bring that potential to all situations but may respond in less or
more reactant ways depending on the characteristics of the situation.
However, on the average across situations, this individual will be more
reactant than another individual with low trait reactance and the possibility
that he or she will demonstrate relatively low reactance (as opposed to high
or moderate) levels of reactance is low. Consequently, on average trait

reactance and state reactance should be moderately correlated.

27

Hypothesis 3

The present study will investigate the differential effect of state and
trait reactance on outcome. It is expected that state reactance will moderate
the relationship between trait reactance and treatment progress. This
relationship would explain inconsistencies in the literature. For example, it
can explain why under certain conditions high trait reactant individuals
would show the best treatment outcomes and under other conditions show
the worst treatment outcomes. If high trait reactant individuals accept the
treatment goals, perceive themselves as needing change, and are satisfied
with the treatment program, i.e., "state" reactance is relatively low, they
should show relatively good outcomes. In addition, high trait reactant
individuals who are invested in treatment (i.e., lower state reactance) are
more likely to be actively involved in the treatment process, internalize the
ideas of the therapy, and maintain positive change over time. They are more
motivated to accomplish the short and long term goals of therapy because
they have "made them their own." In essence, high trait reactant individuals
with relatively lower state reactance should demonstrate the positive aspects
of high reactance introduced by Dowd and Wallbrown (1993).

In contrast, high trait reactance individuals with high state reactance

will often refirse to actively participate in treatment and will resist the

28

changes others are trying to make in them. These individuals will not feel
they have a reason to change and will not view the treatment program as
able to help them. Accordingly, these individuals should show little to no
treatment progress.

An individual who has low trait reactance is likely to have low state
reactance. However, if they perceive the treatment program as not being
able to help them or as having negative effects on them, their state reactance
may be relatively higher. Individuals with low trait reactance who believe
they need help in order change, are satisfied with the treatment program,
and accept the goals of treatment (i.e., low state reactance) are likely to
show more treatment progress than those individuals with low trait

reactance and relatively higher state reactance.

Specific Predictions
Based on the aforementioned hypotheses, the following specific
predictions were made:
1. Trait reactance will be relatively stable over time whereas state
reactance will change over time.
2. Trait and state reactance will be moderately correlated and trait

reactance group classifications will be related to state reactance

29

group classifications. In particular, low trait reactant subjects are
more likely to exhibit low or moderate (but not high) state reactance
whereas high trait reactant subjects are more likely to exhibit high or
moderate (but not low) state reactance.

. State reactance will moderate the effect of trait reactance on
treatment effectiveness. In particular, if state reactance is lower than
trait reactance, positive changes in psychological and behavioral
problems will be greater than when state reactance is higher or equal
to trait reactance.

a) When trait reactance is high and state reactance is moderate
or low, positive changes in psychological and behavioral
problems will be better than when trait reactance and state
reactance are both high.

b) When trait reactance is moderate and state reactance is
relatively low, positive changes in psychological and
behavioral problems will be better than when trait reactance
is moderate and state reactance is high.

0) When trait reactance is low and state reactance is also low,
treatment progress will be better than when trait reactance is

low and state reactance is moderate or high.

30

METHOD

Participants

Participants were 29 males who were being treated at a psychiatric
residential treatment program for adjudicated adolescent boys. They ranged
in age from 13 to 18 with 80% of the sample 15 years old or older. 53% of
the residents were Caucasian, 37% were African American, 3% were Latino,
and 7% were Native American. 17 % were in junior high, 70% were in high
school, and 3% had completed high school. 73% of the residents reported
being in special education. The average IQ was 83 (range =54-114;
estimated from IQ data from 75% of the residents).

All of the participants were admitted to the treatment program within
the same 3 month period of time upon orders from their local probate judge.
All residents were wards of the court and identified for placement in this
particular program because they had psychiatric diagnoses that warranted
treatment as well as incarceration. 67% of the residents had four or more
DSM-IV Axis I diagnoses, 27% had three diagnoses, and 6% had two
diagnoses. Table 1 presents the distribution of diagnostic categories based
on the admitting psychiatrist diagnosis: 33% of the residents had at least

one DSM-IV Axis II diagnosis with 80% of these residents being diagnosed

31

as mentally retarded (IQ’s for 75% of those with data ranged from 54 to 79).
On Axis IV, all but one of the residents showed severe social and
environmental problems. The average global assessment of current

functioning indicated serious impairment (mean = 45.52; Axis V).

Table 1

Percentages of Residents with DSM-IV Axis I Diagnoses

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DSM-IV Diagnosis Percentage

Major Depressive Disorder 13
Dsthymia 80
Bipolar Disorder 7

Post Traumatic Stress Disorder 33
Anxiety Disorders 10
Obsessive Compulsive Disorder 3

Attention Deficit Hyperactivity Disorder 60
Psychosis 13
Conduct Disorder 67
Oppositional Defiant Disorder 37
Intermittent Explosive Disorder 23
Impulse Control Disorder 53
Substance Abuse 20
Other (enuresis & articulation disorder) 7

 

 

32

 

Due to their psychiatric problems, 77% of the residents were on
medications at the time of admission and 87% were on medication 3 months
after admission. The number of medication categories ranged from 0 to 5.
Table 2 reports the percentage of residents on each type of medication at

admission and 3 months post admission.

Table 2

Percent of Residents Taking Each Type of Medication at Admission (Time
1) and 3 Months Post Admission (Time 2)

 

 

 

 

 

 

 

 

 

 

 

 

Medication Category T_im;1_ £1113;
Antidepressants 33 57
Anti-anxiety 7 17
Mood Stabilizer 37 47
Anti-psychotic 3O 33
Stimulant 20 3O
Cogentin 13 23
Sleep Medication 10 20
Other Psychiatric 7 7
Physical Problems 20 17

 

 

 

In addition to psychiatric problems, all of the residents had criminal

histories. The number of convictions ranged from 1 to 10 with mode being
33

 

three crimes. The types of crimes committed can be divided into six
categories describing the nature of the crime (Table 3). The most common
status offenses include truancy, running away, and incorrigibility.

Probation violation was the only crime in its category. The most common
misdemeanors were driving without a license, carrying a concealed weapon,
and escaping a correctional program. The most common drug offenses
include possession of marijuana. The most common property offenses were
petty larceny, attempted arson, grand larceny, and breaking and entering.
Finally, the most common person offenses include aggravated assault,

assault and battery, and rape.

Table 3

Distribution of Crime Convictions Among the Six Offense Categories

 

 

 

 

 

 

 

 

 

 

Offense Category Number of Convictions
Status Offense 22
Probation Violations 9
Misdemeanors l 3
Drug Offenses 5
Property Crimes 35
Person Crimes 29

 

 

34

 

Only two of the residents came from intact families. Approximately
half of the residents came from divorced families and one quarter had
parents who were never married. The remaining five residents reported that
their father had died. None of the residents had mothers who had died.
Half of the resident did not know enough about their fathers to report their
fathers’ education. Of the residents that knew their father’s education, half
had fathers who had not completed high school and only one had a father
who graduated from college. In contrast, all but three residents knew their
mother’s educational level. Mothers of only four of the residents had not
completed high school and five mothers completed college.

Prior to initial incarceration, 83% of the residents lived with either or
both parents. Of the 5 residents not living with their parents, one lived with
relatives, one lived with a family friend, and three lived in foster care.
However, prior to placement at this facility, only one resident lived with his
family. The others either were placed in foster care (3%), residential
treatment (3%), or juvenile detention (90%). Three months after admission,
90% of the residents had had contact with their family since admission
either by phone, letters, or visitation and 73% of the residents families

participated in family therapy.

35

The Program

At time of admission, all residents were informed they would be in
the program for approximately one year. It is expected that the residents
would be returned to their local communities after they were discharged
from the program. Funding for the program was provided by the Family
Independence Agency.

At time of admission, residents were randomly assigned to one of
three treatment groups depending on the space available. The residents
completed all of their treatment and activities with their group and never
interacted with members of other groups. Each group was essentially a
replication of the others, with programming and treatment protocols the
same for all treatment groups. The three groups of 10 residents each had a
group supervisor. The program was based on a positive peer culture model.
The fundamental assumption was that peer influence could be manipulated
in the direction of acceptance of treatment goals and positive change. The
program was designed to elicit pro-social norms in the groups by giving
residents autonomy to make certain choices. This was expected to result in
a greater internalization of the group's goals.

All groups participated in the same daily schedule. Residents

attended school taught by special education teachers for six hours each day.

36

The remainder of their time was divided among activities of daily living
(i.e., eating, showering, and sleeping) and recreational, psychoeducational,
and behavior modification groups. All residents were required to attend all
groups. The only reason they were excused from group was if they were
seriously ill or they were in seclusion/restraints for unsafe behavior.
Recreational activities included quiet time and gross motor activities in the
gym or outside. Psychoeducational groups focused on thinking errors (e. g.,
lack of empathy, blaming others, criminal excitement, lying, etc.), victim
awareness, social skills training, life skills training, parenting skills, current
events, and human sexuality. In addition, each day all residents wrote in
reflection logs that staff read and provided feedback on each night.
Behavior management groups focused on group norms, anger
management, substance abuse, behavior control, and the level system
(described below). All residents earned points on an hourly basis
throughout every day. They earned points for avoiding thinking errors,
following unit rules, participating in activities, interacting with others
appropriately, and following safety guidelines. All residents started at
orientation level and progressed to higher levels as they earned points. Each
level (i.e., orientation, starter, leader, and champion) was associated with

basic privileges with higher levels receiving more privileges. All residents

37

could lose a level for negative behavior.

Measures
Three constructs germane to this study were measured: reactance,

functional impairment, and behavior problems.

Adolescent Reactance

Trait Reactance. The 28-item Therapeutic Reactance Scale-
Adolescent Version (TRS-A) (Frank, Van Egeren, and Poorman, 1993
version of Dowd et al., 1991) was used to measure adolescent offenders trait
reactance (Appendix A). Respondents rated each item on a 4-point scale,
indicating whether they strongly disagree, disagree, agree, or strongly agree.
Trait reactance scores were calculated by averaging the responses on all 28
items. Frank et al. (1993) revised Dowd et al.'s (1991) Therapeutic
Reactance Scale (TRS) to make it more appropriate for adolescents. An
initial examination of the items on the TRS indicated that the content and
vocabulary of approximately half of the items were already appropriate for
adolescents. The other half easily could be modified without sacrificing the
original meaning (e.g., "I am not very tolerant of others' attempts to

persuade me" on the TRS appears as "It irritates me when other people try to

38

change my mind" on the TRS-A.) (See Dowd et al. (1991) for information
on reliability and validity of the TRS.) Frank et al. (1998) provided
evidence of reliability and validity for the TRS-A. They reported internal
consistency coefficients of .72 and .79. While they found that mean scores
were higher and more varied on the TRS-A than reported for college
students and other adult populations on the TRS, the relationships between
trait reactance and personality traits were as predicted and similar to
findings using the TRS. In particular, trait reactance was associated with
counter-conventional, antisocial, narcissistic, non-affiliative, and distrustful
personality traits. As predicted, they found no relationship between trait
reactance and psychological distress. The internal consistency coefficients
for this sample were .85, .89, and .77 at the three administrations.

State Reactance. State reactance was measured by two scales:
Perceived Need for Treatment (Osgood, Gruber, Archer, & Newcomb,
1985) and Satisfaction with the Institution (Osgood et al., 1985) (Appendix
B). Perceived Need for Treatment is a 4 item scale assessing the extent to
which residents believe that they need treatment in order to stay out of
trouble (e.g., "Do you think you need help from someone so you can change
and stay out of trouble?"). Satisfaction with the Institution is a 4 item scale

assessing the extent to which residents believe that the institution is helping

39

them with their problems (e. g., "On the whole, has this place been good for
you so far?"). Two items on this scale were reworded so that they could be
answered using the same scale as the remaining items (e.g., "Do you think
your life will be better or worse because you have spent time here?" was
changed to "Do you think your life will be better because you have spent
time here?"). Residents rated each item on a 4-point scale, indicating
definitely no, probably no, probably yes, and definitely yes. State reactance
scores were calculated by averaging the responses on all 8 items. Osgood
(etal., 1985; and Martin & Osgood, 1987) reported internal consistency
coefficients of .69, and .69 and .72 and .76, respectively. Osgood and his
colleagues (et al., 1985; Gold, Mattlin, & Osgood, 1989; and Martin &
Osgood, 1987) demonstrated construct validity. They reported that state
reactance negatively correlated with offenders' experiences of autonomy in
the program. In particular, offenders who felt they had a great deal of
autonomy reported a positive attitude about the program and acknowledged
their need for treatment (i.e., low state reactance). In addition, they found
that state reactance was negatively correlated to ties with staff and
positively correlated with antisocial values and experiences of anxiety and
depression. Internal consistency coefficients for the entire scale for this

sample were .75, .84, .88, and .89 at each administration.

40

Functional Impairment

 

Staff Report. Program staffs report of resident’s functional
impairment was measured using the Functional Impairment Scale for
Children and Adolescents (F ISCA, Frank & Paul, 1995a) at both 1.5 and 3
months post admission. The F ISCA is a l83-item questionnaire measuring
child and adolescent impairment in 8 different domains of functioning (e. g.,
school, home, community, thinking, undercontrolled behavoir, moods, self-
harm, and substance abuse). Program staff reported on each resident's
functioning in 4 of the domains. Four domains were eliminated because
they were not relevant to residents who had been incarcerated (i.e.,
substance abuse, school, home, and community). F ISCA items use a true-
false, likert, or multiple-choice format and are geared to readers with a 6th
grade education or higher. Staff typically completed the questionnaire in
15-20 minutes.

Scoring criteria, keyed to specific items on the questionnaire, defined
impairment as mild, moderate, and severe. Impairment levels within each
area are defined by several different criteria, although meeting any one
criterion within a level will satisfy the requirements for that particular level.
Scores for each impairment area reflect the highest level of impairment met

by the respondent. Impairment criteria for scoring the F ISCA are based on

41

modified and abbreviated version of the Child and Adolescent Functional
Assessment Scale (CAFAS, Hodges, 1994; Hodges & Gust, 1995).

Validity studies for the FISCA provide support for the measure's
construct, discriminant, and predictive validity. Impairment on the F ISCA
correlated with other measures of child impairment (Child Global
Assessment Scale, Shaffer, Gould, Brasic, Ambrosini, Fisher, Bird,
Aluwahlia, 1983; Columbia Impairment Scale, Bird, Shaffer, Fisher, &
Gould, 1993), discriminated between inpatients and outpatients, and
predicted length of hospital stay and hospital recidivism (Frank, Paul,
Marks, & VanEgeren, under review; Paul, 1997).

Adolescent Report. Resident’s self report of their functioning at
admission and 3 months post admission was measured using the Youth Self

Report Form of the Functional Impairment Scale for Children and

 

Adolescents (YSR F ISCA, Frank & Paul, 1995b). The YSR F ISCA uses

 

the same items as the adult report form although the measure was reworded
to simplify the language. The scale is scored using the same criteria and
domains as the adult form. At the time of admission, the adolescents
answered questions about their functioning during the previous 3 months in
all 8 domains. At 3 months following admission, they reported on their

functioning for the previous 3 months in 4 domains only (i.e., thinking,

42

undercontrolled behavior, moods, and self-harm) because the other 4 of the
domains were not relevant to functioning during incarceration (i.e., home,
school, community, and substance abuse).

A pilot study (cf Frank et al., 1998) comparing FISCA scores from
parents with the adolescent's scores on the FISCA-YSR demonstrated
relatively good agreement. Correlations from 114 adolescent/parent pairs

ranged from .43 to .66.

Behavior Problems

Staff Report. Program staff used the Child Behavior Checklist

 

(CBCL; Achenbach, 1991a) to rate the resident’s behavior problems at 1.5
months and 3 months following admission to the facility. (Appendix C).
The CBCL is a parent-completed rating scale for children ages 4 to 18. It
contains 1 18 items describing behavior problems and 20 items describing
social competence. Only the behavior problem items were used in this
study. The CBCL asks respondents to describe the child now and within the
past 6 month by rating each of the behavior problems as not true, somewhat
or sometimes true, or very true or often true. In order to make the time
frame of this measure consistent with the one used with the F ISCA,

respondents were asked to rate the resident's behavior now and for the past

43

1.5 months at 1.5 and 3 months after admission. While the CBCL was
designed to be completed by parents, a full-time program staff member from
the day and afternoon shift completed the CBCL based on their interactions
with and observations of the residents.

The CBCL is comprised of 8 syndromes: Withdrawal, Somatic
Complaints, Anxious/Depressed, Attention Problems, Thought Problems,
Social Problems, Delinquent Behavior, and Aggressive Behavior. Items
from the syndromes are combined to form to broad scales for Intemalizing
(Withdrawal, Somatic Complaints, and Anxious/depressed) and
Extemalizing problems (Aggression and Delinquency). In addition, all
reported behavior problems are combined to form a Total Behavior Problem
score.

The CBCL manual describes extensive evidence for the reliability of
the CBCL. Reliability data was provided separately for children ages 4 to
l 1 and 12-18. Internal reliability of the behavior problem scales using
Cronbach's alpha ranged from .62 (Thought Problems; boys 4-11) to .92
(Aggressive Behavior; entire sample of girls and boys). Alphas for both the
Extemalizing and Intemalizing scales were .93 and .89, respectively. The
test-retest reliability of the behavior problem scales over a one-week

interval was .72.

44

The current (1991) version of the CBCL is based on a well-validated
earlier version (Achenbach & Edelbrock, 1983). Validity studies
demonstrate convergent and discriminant validity. The CBCL has been
shown to classify effectively the different behavioral patterns of juvenile
offenders (Brannon & Williams, 1986). In addition, the CBCL
discriminates between referred and non-referred children (Achenbach,
1991a)

Adolescent Rpport. Residents used the Youth Self Report (YSR;
Achenbach, 1991b) to rate their problem behaviors at admission and 3
month later (Appendix D). The YSR is a self-report measure for
adolescents ages 11-18. The questionnaire contains 103 specific behavior
problem items and 17 social competence items. Only the behavior problem
items were used in this study. The YSR was developed as a self-report
version of the CBCL, and the two measures have 89 items in common. The
YSR behavior problem section has the same scoring and factor structure as
the CBCL. The YSR asks respondents to describe themselves now and
within the past 6 month by rating each of the behavior problems as not true,
somewhat or sometimes true, or very true or often true. In order to make the
time frame of this measure consistent with the F ISCA-YSR, at admission

adolescents were asked to rate their behavior over the past 3 months and for

45

the 3 months period since admission.

The YSR manual provided sufficient evidence of reliability. Internal
reliability of the behavior problem scales using Cronbach's alpha ranged
from .59 (Withdrawn; boys and girls) to .90 (Anxious/depressed; girls).
Alphas for both the Extemalizing and Intemalizing scales were .89. The
test-retest reliability of the behavior problem scales over a one-week
interval was .72.

Like the CBCL, the current version of the YSR is based on a well-
validated earlier version (Achenbach & Edelbrock, 1987). Validity studies
based on the new version demonstrated convergent and discriminant
validity. The YSR discriminated children referred for treatment from non-
referred children (Achenbach, 1991b). In addition, the YSR scales
moderately correlated with similar scales on the CBCL and the Teacher's
Report Form (Stanger & Lewis, 1993).

Staff Behavioral Observations. Staff used the Child Behavior Rating
Form-Revised (CBRF-R; Van Egeren, 1996) to rate resident’s behavior
problems. (Appendix E). The CBRF-R is a 27-item checklist that asks
raters to check off all problem behaviors observed and then rate the
behaviors using a 3-point scale (l=mi1d problem, 2=moderate problem,

3=severe problem). The CBRF-R is rated by team supervisors as part of
46

their weekly progress notes and summarizes behaviors for the preceding
week.

The CBRF-R is comprised of five problem behavior scales (19 items)
and three clinically relevant scales (8 items). The five problem behavior
scales consist of Oppositional (4 items), Attention Problems (3 items),
Overactivity (3 items), Withdrawal/depression (6 items), and Anxiety (3
items). The three clinical scales included to assess clinically relevant
content areas are Aggression (3 items), Self-harm (2 items), and Thought
Problems (3 items). In addition, three scales can be combined to form an
externalizing factor (Oppositionalism, Attention Problems, and
Overactivity).

The CBRF-R is a modified version of the Child Behavior Rating
Form (Edelbrock, 1985) which was a 65 item measure of child behavior
problems and positive behavior. The CBRF-R was revised in order to limit
redundancy, clarify meaning, and add clinically relevant information. In
addition, the directions and format were revised to accelerate the rating
process (e.g., the CBRF-R is rated on a 3 point scale rather than the 4 point
scale use for the original CBRF).

Van Egeren (1996) provided evidence for the reliability of the CBRF-

R using daily ratings over a period of weeks. Internal reliability of the five

47

behavior problem scales using Cronbach's alpha ranged from .67 (Anxiety)
to .90 (Oppositionalism). Alpha for the Extemalizing factor was .89. The
test-retest reliability of the behavior problem scales over a two-week
interval ranged from .11 (Anxiety) to .75 (Overactivity). Interrater
reliability was examined using trained student pairs and student/program
staff pairs. Interrater reliability for student rater pairs ranged from .41 to .84
(except for anxiety for which one set of raters observed none of the targeted
behaviors, so reliability could not be estimated). Pairs consisting of one
student rater and one program staff member resulted in lower coefficient
(.30 to .61, except for attention problems which averaged .16). Internal
reliability for this study ranged from .12 (anxiety) to .94 (oppositionalism)
for the behavior problems scales and .00 (self harm) to .59 (aggression) for
the clinical scales.

Validity studies by Van Egeren (1996) demonstrated concurrent and
predictive validity. She found that the scales related as expected to each
other and to other measures of child behavior problems (e.g., CBCL and
YSR). In addition, her data supported a predicted positive relationship

between the CBRF-R and the patient length of hospital stay.

48

Procedure

At the time of admission, the residents completed self-report
measures assessing trait reactance, functional impairment, and behavior
problems as part of the program's routine assessment procedures.
Information obtained from these questionnaires is used for diagnosis and
treatment planning. (See Table 4 for detailed timeline.)

At approximately six weeks after admission and prior to gathering
information beyond the initial intake assessment, adolescents were asked to
give informed assent for participation in the study (Appendix F). They were
given an assent agreement that outlines the questionnaires they would be
asked to complete, the purpose of the study, and the procedures ensuring
confidentiality. They were informed that they did not have to participate in
the study and told that they could stop participation at any time during the
study. They were assured that a decision not to participate in the study
would not affect the treatment they received in the program. They were
given the opportunity to receive feedback about their progress and told that
they could be informed about the results of the study if they were interested.
Twenty-nine of the thirty residents signed the informed assent agreement.
Only one resident declined to participate. While the one nonparticipant had

a higher IQ than the average participant and reported on the Achenbach
49

fewer problems prior to admission, he did not differ in regard to age or trait
reactance.

Case workers of adolescents assenting to participate were contacted
to give informed consent. (All adolescents are wards of the court and a
caseworker is appointed as their legal guardian.) Caseworkers were
contacted by phone and informed about the study. If they gave informed
consent, a consent form was faxed to them and they faxed back a signed
consent agreement. (Appendix G). In seven cases, the caseworker
requested that the resident’s parents be contacted to be informed about the
study and to give pennission for the resident to participate. All caseworkers
and parents contacted gave permission for the resident to participate in the
study.

All residents who signed the assent agreement and whose caseworker
signed the informed consent participated in the remainder of the study.
Demographic information was gathered from the residents' medical record.
Residents completed all questionnaires in their school classrooms. If
residents had difficulty reading or concentrating, they were given the option
of having the questionnaires read aloud to them individually or in small
groups. Eight of the 29 residents requested that the questionnaires be read

to them. These residents were diagnosed as either mentally retarded or

50

learning disabled.

Approximately 7 weeks (1.5 months) after admission, the residents
completed measures of both trait and state reactance. Every additional two
weeks, residents completed the state reactance measure (i.e., at 9, l 1, and 13
weeks post admission). Before completing the state reactance measure,
residents were instructed to report on how they were feeling at the time of
testing and informed that they should not try to remember what they
answered in the past. Three months after admission, residents again
completed questionnaires assessing trait reactance, fiinctional impairment,
and behavior problems.

One full-time program staff member from each shift (day and
afternoon) also completed questionnaires assessing functional impairment
and behavior problems for each resident 1.5 and 3 months after the
residents' admission to the program. The staff members were paid their
normal hourly rate to complete these questionnaires after work hours. In
addition, each team supervisor completed weekly summaries of residents’
behaviors as part of the standard program progress notes. Each team
supervisor used the CBRF-R to rate the occurrence of behavior problems
that occurred during the previous week. These ratings were used as another

measure of residents' behavior during the last 1.5 months of the study.

51

Members of the research team transformed information from the
residents' charts, residents' questionnaires, and staffs’ questionnaires into
numbers that were entered into computers along with code numbers to
identify the resident. By necessity, the research staff were initially aware of
the residents' names. However, all research staff signed and were bound by
the Center's existing confidentiality agreement. Only the computer data
files were removed from the center. Assessment protocols and charts were
never removed from the center grounds by any person for the purpose of
research. All information collected on residents was entered into the

computer and identified by a two digit code.

52

Table 4

Timeline for Data Collection

Time Resident Measures

Admission Behavior for past 3 months:

(0 Weeks) Functional Impairment

Behavior Problems

Trait Reactance

Informed Assent

7 Weeks Trait Reactance

(1'5 Months) State Reactance

8 Weeks
9 Weeks State Reactance
10 Weeks
11 Weeks State Reactance

12 Weeks

13 Weeks Behavior for past 3 months:

(3 Months) Functional Impairment

Behavior Problems
Trait Reactance

State Reactance

 

53

Staff Measures

Informed Consent from F IA
Behavior for past 1.5 months:
Functional Impairment
Behavior Problems
Child Behavior Rating Form
Child Behavior Rating Form
Child Behavior Rating Form
Child Behavior Rating Form

Child Behavior Rating Form

Behavior for past 1.5 months:
Functional Impairment
Behavior Problems

Child Behavior Rating Form

RESULTS

Table 5 shows the ranges, means, and standard deviations for trait and
state reactance at all time points. The scores for trait reactance at all three
time points were slightly higher but were within one standard deviation of
scores reported for a non-incarcerated inpatient adolescents sample (Frank

etaL,1998)

Table 5

Ragges, Means, and Standard Deviations for Trait and State Reactance

 

Trait Reactance

 

 

 

 

 

 

 

 

 

 

 

 

 

Range Mean SD
Time 1 1.86-3.43 2.62 .39
Time 2 1.82-3.79 2.64 .45
Time 3 1.82-3.25 2.56 .33

1 State Reactance

Time 1 1.25-3.75 2.15 .63
Time 2 125-363 2.28 .67
Time 3 LOO-4.00 2.30 .78
Time 4 1.00-3.63 2.02 .73

 

 

 

 

54

 

Hypothesis 1: Comparing the Stability of Trait and State Reactance

Several analyses indicated the greater stability of trait reactance
compared to state reactance. In Figure l pictorial plots of each resident’s
reactance overtime indicated that trait reactance was relatively stable over
three points in time, with most individual plots approaching straight lines
with slopes of zero. In contrast, state reactance was relatively unstable over
four points in time, with many individuals deviating Considerably from their
own mean. Relatedly, mean standard deviations for the two types of
reactance differed significantly (t(28)=-4.9, p<.001). As can be seen in
Figure 2, the range of standard deviations for State reactance (.06 to 1.16)
overlapped with, but exceeded the range of standard deviations for trait
reactance (.02 to .78). The great majority (90%) of standard deviations for
trait reactance were below .26, whereas for state reactance the majority
(59%) exceeded .26.

Repeated measures ANOVA’s assessed whether changes in trait and
state reactance were statistically significant. The F-test for the within
subjects factor for trait reactance (measured at three time points) was not
significant (F =1 .34, df(2,52), p=.27). However, the F-test for the four time
points of state reactance resulted in a statistical trend (F=2.26, df (3,84),

p=.09) with significant differences between Times 2 and 3 (p<.03) and
55

 

[4* l-I

 

 

 

 

 

 

 

 

 

 

4
U \ .
3 ’\\ o_____,_ H
0 :~:- ~\ -- _..——- l“—
g - 4’3:_:.~<
0 “Wk-'— ,-
8 2 _, w
m __
.i-D
8
H
I—
1 _
0 r r
l 2 3
Time
4
3 ..

 

State Reactance
N

 

 

 

 

 

Time
Figure 1

Trait Reactance at Time 1. 2, and 3 and State Reactance at Time 1.2. 3. a_r_1_(_1
4 for All Participants in the Study

56

14

 

12*

d
O

Frequency
e 9°

 

 

 

1Y2Y3‘415‘c‘7‘a‘9
Standard Deviation (:10)

Figure 2

1O

11

12

 

— Trait Reactance
— State Reactance

 

 

 

Distributions of the Standard Deviations for Trait and State Reactance

57

 

Times 3 and 4 (p<.02). As can be seen in Table 6, intercorrelations among
the trait reactance scores at each time point (range=.73 to .81; mean = .77)
generally are larger and do not overlap with the distribution of
intercorrelations among state reactance at each time point (range-+242 to .67;
mean = .55).

In sum, these data supported Hypothesis 1. The plots of changes in
trait and state reactance, the significantly narrower distribution of standard
deviations of trait reactance as compared to state reactance, and the
generally higher correlations among measures of trait reactance as compared
to state reactance over time indicated the greater stability of trait reactance

as compared to state reactance.

Hypothesis 2: Evaluating the Relationship Between Trait and State
Reactance
Hypothesis 2 predicting that trait reactance and state reactance would
be moderately correlated and that trait reactance group classifications would
be related to state reactance group classifications also was supported by the
data. As can be seen in Table 7, half of the correlations assessed at different
time points were statistically significant, with the mean scores for trait and

state reactance (averaged over time) correlating significantly (r=.45, p<.05).

58

Table 6

Intercorrelations Among Trait and State Reactance and the Means and

Standard Deviations for Trait and State Reactance

 

 

 

 

 

 

 

 

 

 

 

 

Trait Time 1 Time 2 Time 3

Reactance

Time 1 1.00

Time 2 .81C 1.00

Time 3 .73C .76c 1.00

Trait .92C .95c .88c

State Time 1 Time 2 Time 3 Time 4
Reactance

Time 1 1.00

Time 2 .50b 1.00

Time 3 .422’ .67c 1.00

Time 4 .443 .59c .67c 1.00
State .70C .84c .87c .84°

 

 

 

 

 

Note. Trait = Mean of Trait Reactance at Time 1, 2, and 3; State = Mean of

State Reactance at Time 1, 2, 3 and 4.

a: p<.05, b = p<.01, ° = p<.OOl.

59

 

Table 7 -

Intercorrelations Among State Reactance and Trait Reactance

 

 

 

 

 

 

 

Trait Time 1 Trait Time 2 Trait Time 3 Trait
State Time 1 .32 .51b .40a .48b
State Time 2 .43a .19 .31 .30
State Time 3 .46a .28 .52b .428
State Time 4 .31 .14 .40a .511)
State .45a .33 .51b .453

 

 

 

 

 

Note. Trait = Mean of Trait Reactance at Time 1, 2, and 3.
State = Mean of State Reactance at Time 1, 2, 3 and 4.
a = p<.05, b = p<.01, ° = p<.001.

Classification of Trait and State Reactance Groups

Different patterns of change and stability were used to classify
residents into trait and state reactance groups. All residents whose scores
for state reactance remained within one standard deviation of the sample
mean were placed in the Moderate State Reactance Group (SRG; p=10). If
state reactance was below the sample mean at all time points, and one
standard deviation below the sample mean for at least one time point, the

resident was placed in the Low SRG (3:7). Alternatively, if state reactance

6O

 

was above the sample mean at all time points and at least one standard
deviation above the mean for at least one time point, the resident was placed
in the High SRG (3:5). The remaining residents were placed in an Unstable
SRG which was not anticipated a priori, but fit the data for a significant
minority of participants (13:7). In this group, state reactance at one or more
time points was at least one standard deviation above or below the sample
mean and at one or more time points shifted to the opposite side of the
mean. Figure 3 shows the plots of the four state reactance groups.

Residents were divided into trait reactance groups (TRG) using the
same rules as state reactance. Five residents were classified in the Low
TRG, fifteen in Moderate TRG, and eight in the High TRG. Only one
resident met criteria for the Unstable TRG (compared to seven in the
unstable state reactance group).

Table 8 shows the crosstabulation of the TRG’s and SRG’s. As
expected trait reactance restricted the range of state reactance. Low trait
reactant residents fell either in the Low or Moderate SRG’S with none in the
High SRG. The Moderate trait reactant residents were found in all four
SRG’S. With the exception of one resident in the Moderate SRG, all High
trait reactant residents were either in the High or Unstable SRG’S. The one

Unstable trait reactant resident was also in the Unstable SRG suggesting

61

Low State Reactance Moderate State Reactance

 

 

   

 

 

 

 

 

 

82.5 J 82.5
§15 $1.5
20.5 E05 -
£305 130.5
“-1.5 ~ "31.5 4
-2 5 . r t -2.5 t t .
1 2 3 4 1 2 3 4
Time Time
High State Reactance Unstable State Reactance

 

 

 

 

 

 

 

 

 

Time

Figure 3

State Reactance at Time 1, 2, 3, and 4 for Low, Moderate. High, and
Unstable State Reactance Groups

62

that this resident’s responses may have been random on both measures and

hence of questionable validity. A significant Chi-square (X2(9)=18.31,

p<.03) indicated that the crosstabulation between trait and state reactance

would not be expected by chance.

Table 8

Crosstabulation for Trait and State Reactance

 

 

 

 

 

 

 

 

Trait State Reactance Groups (SRG)

Reactance

Groups Low Moderate High Unstable Total
(TRG)

Low 3 2 5
Moderate 4 7 l 3 15
High 1 4 3 8
Unstable 1 1
Total 7 10 5 7 29

 

 

 

 

 

 

63

 

Hypothesis 3: Investigating State Reactance as a Moderator Between

Trait Reactance and Treatment Progmss

Collapsing the Reactance Groups

Crosstabulations of the TRG’s and SRG’S resulted in six empty cells

(Table 8). Accordingly, the one subject exhibiting unstable trait and

unstable state reactance was excluded from further analyses. In addition,

the Low and Moderate SRG’s were collapsed because the distinction

between low and moderate state reactance was not essential for testing the

hypotheses and preliminary analyses indicated that the two groups had

similar patterns of change. The resulting design is a three TRG’s (Low,

Moderate, and High) by three SRG’s (Low/moderate, High, and Unstable)

matrix with 28 subjects and two empty cells (see Table 9).

Table 9

Crosstabulation for Trait and State Reactance Using Revised Classifications

 

 

 

 

 

 

 

Trait State Reactance Groups

Reactance

Groups Low/Mod. High Unstable Total
Low 5 5
Moderate 1 1 1 3 15
High 1 4 3 8
Total 1 7 5 6 28

 

 

 

 

 

64

 

 

Revising the Specific Predictions for Hypothesis 3

Due to the observed (as opposed to expected) distribution of residents
in the trait and state reactance groups, it was not possible to fully test the
specific predictions for Hypothesis 3 as originally stated. In particular, no
predictions could be made about the Low TRG because all of these residents
were classified as one and only one SRG (i.e., Low/moderate). In addition,
no predictions could be made about the combinationof moderate trait and
high state reactance or of high trait and low/moderate state reactance
because in each case only one resident could be classified as such.

Finally, the nature of the Unstable SRG needed to be considered
before specific predictions could be revised. The distribution of state
reactance across TRG’s implied that unstable state reactance is an
intermediary between moderate and high state reactance. In particular, the
High TRG was more likely to display unstable rather than moderate state
reactance suggesting that unstable state reactance is a higher level of
reactance than moderate state reactance. In contrast, the Moderate TRG was
more likely to display unstable state reactance than high state reactance
suggesting that Unstable state reactance is a lower level of reactance than
high state reactance. Unstable state reactance may represent a reduction in

expected state reactance for the High TRG and an increase in expected state

65

reactance for the Moderate TRG. If this interpretation of the data is
accurate, the unstable state reactance should have different implications for
treatment progress for the Moderate and High TRG’s.

Accordingly, specific predictions regarding the moderating effects of
state reactance on the relationship between trait reactance and treatment
progress were restated as follows:

1. When trait reactance is high and state reactance is unstable
(relatively lower), treatment progress will be better than when trait
reactance is high and statereactance is also high.

2. When trait reactance is moderate and state reactance is also
moderate or lower (i.e., low/moderate), treatment progress will be
better than when trait reactance is moderate and state reactance is

unstable (relatively higher).

TestingFor Confounding Variables

One-way ANOVA’s on a total of 29 demographic variables (see
Table 10) were performed separately for trait and state reactance groups in
order to identify potential confounds. No variables differed significantly
among the trait reactance groups and only three variables differed

significantly among state reactance groups. In particular, the Unstable SRG

66

Table 10

Demographic Variable Tested for Differences Among the State and Trait

Reactance Groups

 

Trait Reactance Group

 

 

Age
Minority Status
1Q
Father Education
Mother Education
Medication Categories:
Time 1
Time 2
Number of Crimes Committed
Crime Catpgories:
Status Offenses
Probation Violation
Misdemeanors
Drug Offenses
Property Crimes
Person Crimes

 

Number of Axis I Diagnosis
Axis I Diagnostic Categories:

Major Depressive Disorder
Dysthymia

Bipolar Disorder

PTSD

Anxiety

Obsessive-Compulsive Disorder
ADHD

Psychosis

Conduct Disorder

Oppositional Defiant Disorder
Intermittent Explosive Disorder
Impulse Control Disorder
Substance Abuse

Axis V (current)

 

Note. Variables in boldface differ significantly among state reactance
groups. No variables show significance for trait reactance.

67

 

was taking more different types of medications (M=3.71, SD=1.17) and
comrrritted more status offenses (M=1.17, SD=.41) than the Low/moderate
SRG (M=l .35, SD=1.37 and M=1.06, SD=.24 respectively; F(2,25)=3.71
and F (2,25)=3.89 respectively, p<.05). In addition, the High SRG had more
drug offenses (M=l .60, SD=.55) than Low/moderate (M=1.06, SD=.24) or
Unstable (M=1.17, SD=.41) SRG’s (F(2,25)=4.76, p<.05). Separate Chi-
Square analyses for TRG’s and SRG’s investigating possible differences in

race or treatment group were all insignificant.

Reducing the Outcome Variables

Self and Staff Report. Eleven Achenbach and five FISCA scales
were used to assess treatment progress. However, because a large number
of residents had no initial impairment in thinking or self-harm on the
FISCA, these variables were not considered per se in subsequent analyses
but were included in calculations of a total impairment score. The three
remaining F ISCA variables include impairment in undercontrolled behavior
(an aspect of Achenbach’s externalizing problems), impairment in moods
(an aspect of Achenbach’s internalizing problems), and total impairment (an

aspect of Achenbach’s total problems).

68

Correlations between self and staff reports for the final set of outcome
variables were used to assess interrater agreement (see Tables 11, 12 and
13). At Time 1 (1.5 months post admission), two staff raters reported on
each resident’s behavior and at Time 2 (3 months post admission) the
resident himself as well as the two staff raters reported on the resident’s
behavior. Intercorrelations between staff reports at Time 1 were statistically
significant for total problems on both the Achenbach and the F ISCA and for
externalizing, social, thought, and attention problems on the Achenbach.

No internalizing problems showed significant intercorrelations between
staff reports on either the Achenbach or the F ISCA. At Time 2,
intercorrelations between both staff raters were significant for all outcome
variables. Intercorrelations between self report and staff report typically
were not significant. The FISCA showed better interrater agreement
between self and staff report than the Achenbach with agreement
coefficients significant for both impairment in moods and total impairment.

Behavioral Observations. The CBRF scales were dropped as
outcome variables because of a lack of internal consistency. None of the
clinical scales (self-harm, aggression, thought problems) nor the anxiety
problem behavior scale had acceptable alpha’s (alpha’s <.51). In addition,

ratings of the remaining behavior problem scales (externalizing,

69

Table 1 l

Interrater Agreementat Time 1 for Achenbach and F ISCA Variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Achenbach Variables
Rater 1/ Mean (SD) T-Test
Rater 2 Rater 1 Rater 2 df(28)
Intemalizing Problems .32 17.66 18.90 -.91
(7.43) (4.62)
Withdrawal .36 7.45 7.86 -.5 l
(4.61) (2.29)
Somatic Problems .33 3.24 4.72 -2.66a
(2.84) (2.31)
Anxiety/Depression .15 8.21 7.62 -.68
(4.44) (2.26)
Extemalizing Problems .79C 15.93 20.76 -4.00c
(9.99) (9.84)
Delinquency .56b 4.21 5.34 -245:
(2.88) (2.33)
Aggression .81C 1 1.72 15.41 -4.16c
(7.64) (7.83)
Social Problems .65° 3.97 5.24 -284"
(2.99) (2.73)
Thought Problems .56" 1.55 1.45 .27
(2.43) (1.74)
Attention Problems .59° 6.31 7.31 -1.52
(3.97) (3.85)
Total .62“ 51.83 61.07 -2995
(20.42) (17.3%
FISCA Variables
Moods .36 2.21 2.83 -3.70c
’ (.94) (.63L
Undercontrolled .15 2.17 2.48 -l .51
(1.00) (.639
Total .55b 48.28 61.38 -422“
(18.72) (15.97)

 

 

 

 

 

Note. a = p<.05, b = p<.01, ° = p<.001.

 

Table 12

Interrater Agreement at Time 2 for Achenbach Variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Self/ Self/ Rl/ Mean F-Test
R1 R2 R2 Self Raterl Rater2 df(2,27)
IntemalizingPr. .25 .30 .60° 21.83y 19.102 15.14“ 6.881)
(15.57) (8.38) (3.73)
Withdrawal .22 .36 .68C 531" 8.38” 6.692 7.88b
(3.93) (3.33) (1.67)
Somatic Pr. .50b .17 .43a 5.79"y 2.76" 3.03y 6.84b
(5.09) (2.53) Q64)
Anx/depression .12 .25 .473 11.59y 9.41Z 6.41” 1181"
(8.67) (4.95) (2.15)
Extemalizing Pr. .06 .20 .61° 19.59 18.48 19.03 .14
(10.44) (7.88) (5.75)
Delinquency -.09 .1 l .453 6.34" 4.17"Z 4.97Z 3.91a
(3.95) (1.95) (1.80)
Aggression .17 .30 .64c 13.24 14.31 14.07 .18
(7.93) (6.75) (4.39)
Social PT. .20 .41: .70° 4.79 4.48 4.24 .57
(4.05) (2.91) (1.83)
Thought Pr. .35 .20 .62“ 4.72xy 1.31" 1.28y 11.98c
_ (3.93) (1.73) (2.05)
AttentionPr. .41a .36 .48b 7.14 6.59 5.38 3.26
(4.65) (3.41) (2.50)
Total .23 .31 .67c 70.86y 55.862 50.07” 5.50b
(43.46) (19.85) (12.13)

 

Note. Anx = Anxiety, Pr. = Problems, R l = Rater l, R 2 = Rater 2.

x = self report and rater 1 are significantly different, y = self report and rater
2 are significantly different, 2 = rater l and rater 2 are significantly different.

a = p<.05, b = p<.01, ° = p<.001.

71

 

Table 13

Interrater Agreement at Time 2 for F ISCA Variables

 

 

 

 

 

 

 

 

 

 

 

 

Self/ Self/ Raterl/ Mean (SD) F -Test
Rater l Rater 2 Rater 2 Self Rater 1 Rater 2 df(2,27)
Mood .3921 .68c .60C 2.34 2.41 2.14 2.04
(.94) (.87) (.99)
Ucontrol .29 .24 .73C 2.17 2.38 2.14 2.86
(.93) (.68) (.83)
Total .48b .43a .77: 57.93y 54.147- 47.24yz 4.1721
(35.76) (17.43) (21.53)

 

 

Npte. Ucontrol = Impairment in undercontrolled behavior, Mood
Impairment in mood, Total = Total impairment in thinking, undercontrolled
behavior, moods, and self harm.
x = self report and rater 1 are significantly different, y = self report and rater

2 are significantly different, 7‘ = rater l and rater 2 are significantly different.
d = df(1,28). “ = p<.05, b = p<.01, C = p<.001.

oppositional, attention problems, overactivity, and withdrawal/depression)

differed by treatment group (see Table 14). Because each team was rated by

a different rater, rater idiosyncrasies on the instrument could account for

these differences. This seems likely because there were few group

differences on the Achenbach or the F ISCA (5 of 84 analyses). Unreliable

reporting by the group leaders is also likely because the program was

relatively new and group leaders had not been formally trained to use the

rating scale.

72

 

Table 14

Significant Differences Between Groups for CBRF Vgriables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mean (SD) F -Test, df

Group 1 I Group 2 I Group 3
Extemalizing
Time 1 .93 (.43)" 1.90 (.14)"2 .56 (.56)" 5.70" df(2,8)
Time 2 3.18 (1.32)y 3.32 (.90)" 1.48 (1.67)yz 3.89" df(2,19)
Time 4 .87 (.40)" 1.68 (86)" .61 (.90)2 10.24°df(2,21)
Time 6 .84 (.32)" 1.34 (48)” .40 (.30)2 10.12°df(2,18)
Time 7 .57 (.37)x 1.59 (.61)"2 .5 (.44)2 10.56°df(2,17)
Oppositional
Time 2 1.88 (.89)y 2.45 (.74)2 .57 (.80)” 8.49" df (2,19)
Time 3 1.39 (.80)y 1.14 (1.06)2 .31 (.43)yz 4.39a df(2,20)
Time 4 1.36 (.73)"y 2.50 (.63)"" .47 (.61)” l6.94°df(2,21)
Time 5 1.78 (.83)y 1.72 (1.06)2 .61 (.90)yz 3.86" df(2,23)
Time 6 1.68 (.70)y 2.06 (.48)" .50 (.79)yz 10.25°df(2,18)
Time 7 1.21 (.70)" 2.64 (66)”- .85 (.55)" 15.38°df(2,17)
Attention Pr.
Time 1 .33 (.47)" 1.83 (24)“ .40 (.37)" 11.19b df(2,8)
Overgctivity
Time 1 .42 (.50) .83 (.24)" .00 (.00)" 5.33" df(2,8)
Time 4 .33 L551 .94 (1.04)2 .07 (.15)2 3.62" df(2,21)
With/Deg
Time 2 .75 (.45) 1.20 (.68)" .29 (.46)2 4.79" df(2,l9)
Time 3 .74 (.53) .93 (.56)2 .28 (.34)2 4.05a df(2,20)
Time 4 .78 (.61)y 1.03 (.41)2 .24 (.33)yz 5.61a df(2,21)
Time 6 .60 (.46)" 1.27 (.75)"2 .14 (.22)2 7.60bdf(2,18)

 

 

 

Note. Pr. =Problems, With/Dep=Withdrawal/depression.
"= group 1 and group 2 are significantly different, y = group 1 and group 3
are significantly different, 2 = group 2 and group 3 are significantly

different.

"= p<.05, b = p<.01, ° = p<.001.

73

 

Reactance and Changes in Outcome Variables

Analyses of data for each of the remaining outcome variables
excluded residents who had a score of “0” at Time 1. These residents by
definition were unable to Show progress because of lack of initial pathology.
Table 15 shows the number of residents excluded from further analyses for

each outcome variable.

Table 15

Number of Residents Without Pathology at Time 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Achenbach Variables -
Self Report Staff Report
Intemalizing 0 0
Withdrawal l 0
Somatic Problems 3 l
Anxiety/Depression 1 0
Extemalizing 0 0
Delinquency 0 0
Aggression 0 0
Social Problems 2 0
Thought Problems 3 9
Attention Problems 2 0
Total Problems 0 0
FISCA Variables
Moods 3 0
Undercontrol Behavior 1 0
Total Impairment 1 0

 

 

 

74

 

At Time 1, scores on the outcome variables did not differ
significantly among the three TRG’S. Repeated measures ANOVA’S were
used to assess whether changes in outcomes over the thirteen weeks of the
study period varied as a function of trait and state reactance (Hypothesis 3).
TRG (3 levels) and SRG (3 levels) defined the between group factors and
Time (of assessment, 1 or 2) defined the two levels of the within subjects or
change factor. For self reports, Time 1 refers to admission with behavior
reported for the 3 months prior to admission. For staff report, Time 1 refers
to 1.5 months after admission with behavior reported from the time of
admission. For both staff and self reports, Time 2 outcome measures were
completed at 3 months post admission and described behavior since the last
report (i.e., admission for self report and 1.5 months post admission for staff
report). Analyses using staff reports also included Rater (1 or 2) as a within
subjects factor.

Table 16 shows the results of analyses for each outcome variable
based on self report. Table 17 shows the results for each variable based on
staff report. Although Rater x Change interactions were significant for
seven of the fourteen outcome variables, none of the Rater x TRG x Change
interactions and none of the Rater x SRG x Change interactions were

statistically significant. These effects are not presented in the table.

75

Table 16

F-tests Associated with the Trait Reactance Group x State Reactance Group
x Change ANOVA’S Using Self Report of Change

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Achenbach Variables
Outcome Change (C) Trait (T x C State (S)xC Tx S xC
Variables F-Test df F-Test df F-Test df F-Test df
Intemalizing 9.55b 1,21 17.98“ 2,21 9.03“ 2,21 7.16" 2,21
Withdrawal .58 1,20 6.00"" 2,20 1.21 2,20 2.63" 2,20
SomaticPr 5.95" 1,18 1.40 2,18 .76 2,18 1.13 2,18
Anx/Dep 7.50" 1,20 41.30“ 2,20 22.59“ 2,20 15.03“ 2,20
Extemalizing 3.11" 1,21 2.45 2,21 .34 2,21 .14 2,21
Delinquency 7.42a 1,21 1.03 2,21 .45 2,21 .89 2,21
Aggression .16 1,21 3.48" 2,21 .51 2,21 1.51 2,21
SocialPr 1.92 1,19 2.05 2,19 .53 2,19 1.39 2,19
ThoughtPr .28 1,18 1.66 2,18 .73 2,18 1.75 2,18
Attention Pr .01 1,19 3.77“ 2,19 1.25 2,19 1.00 2,19
Total .38 1,21 10.65“ 2,21 3.22" 2,21 3.13" 2,21

FISCA Variables
Mood .01 1,18 1.43 2,18 1.29 2,18 2.63t 2,18
Undercontrol 3.08" 1,20 1.41 2,20 .16 2,20 .57 2,20
Total .67 1,20 2.03 2,20 3.67" 2,20 .31 2,20

 

Note. Anx/Dep = Anxiety and depression, Pr = Problems, Undercontrol =

Impairment in undercontrolled behavior, Mood = Impairment in moods,

Total = Impairment in thinking, undercontrolled behavior, moods, and self

harm.

t= <.10, " = p<.05, ‘“ = p<.01, “ = p<.OOl.

76

 

Table 17

F-tests Associated with the Trait Reactance Group x State Reactance Group
x Change ANOVA’S Using Staff Report of Change

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Achenbach Variables
Outcome Change (C) Trait (T) x C State (S) x C T x S x C
Variables F -Test df F-Test df F -Test df F -Test df
Intemalizing .72 1,21 .11 2,21 .85 2,21 2.97" 2,21
Withdrawal .32 1,21 .27‘ 2,21 .70 2,21 .41 2,21
Somatic Pr 2.37 1,20 1.66 2,20 .42 2,20 4.02a 2,20
Anx/Dep 2.00 1,21 .64 2,21 1.38 2,21 1.42 2,21
Extemalizing .02 1,21 .23 2,21 1.48 2,21 .67 2,21
Delinquency .60 1,21 .03 2,21 1.76 2,21 .25 2,21
Aggression .22 1,21 .34 2,21 1.21 2,21 .83 2,21
Social Pr .08 1,21 2.92t 2,21 1.83 2,21 1.96 2,21
Thought Pr .81 1,12 1.45 2,12 .53 2,12 .25 2,12
Attention Pr .20 1,21 .11 2,21 .11 2,21 .89 2,21
Total .00 1,21 .02 2,21 1.78 2,21 2.35 2,21
FISCA Variables
Mood .43 1,21 .46 2,21 .15 2,21 .1 17 2,21
Undercontrol 1.20 1,21 4.13" 2,21 .37 2,21 2.60t 2,21
Total 1.03 1,21 1.51 2,21 .92 2,21 .33 2,21

 

Note. Anx/Dep = Anxiety and depression, Pr = Problems, Undercontrol =
Impairment in undercontrolled behavior, Mood = Impairment in moods,
Total = Impairment in thinking, undercontrolled behavior, moods, and self

harm.

t= <.10, " = p<.05, " = p<.01, “ = p<.001.

77

 

AS can be seen in Table 16, there are two significant main effects for
self report of change (somatic problems and delinquency) and two statistical
trends (externalizing problems and undercontrolled behavior) that are not
accompanied by significant 2-way or 3-way interactions. The main effect
for self report of somatic problems indicates that residents developed more
somatic problems over time (means Tl= 5.24 and T2=6.12). In contrast, the
main effect for delinquency indicates that the residents showed a significant
decrease in delinquent behaviors over time (means T1=10.11 and T2=6.32).
In addition, statistical trends for self report of externalizing problems
(means Tl=24.32 and T2=19.36) and undercontrolled behavior (means
Tl=2.78 and T2=2.15) suggested improvement. Notably, there were no
significant main effects for staff reported change.

Interaction results associated with Hypothesis 3 will be discussed
separately for internalizing, externalizing, and total problems and within
each of these, for self and staff report. All significant interactions are
plotted using z-scores to facilitate interpretation across variables.
Significant change (p<.05) and statistical trends for change (p<.10) are
represented by straight lines, nonsignificant change by dashed lines, and
non-interpretable changes (based on n’s of one) in dotted lines. Because

Hypothesis 3 predicts that trait and state reactance interact in predicting

78

changes overtime, the relevant effects testing this hypothesis involve 3-way
interactions (TRG x SRG x Change).

Intemalizing Problems. Analyses of the self report data for
internalizing problems resulted in two Significant TRG x SRG x Change
interactions (total internalizing problems and anxiety/depression problems)
and two statistical trends (withdrawal problems and impairment in moods).
All four 3-way interactions using self reports resulted in similar patterns
within each of the TRG’S and these patterns generally supported Hypothesis
3. Figures 4 and 5 Show the two significant interactions (graphs of the two
trends are in Appendix H). Within the High TRG (Figure 4), the Unstable
SRG showed significant improvement for anxiety/depression problems
(t(2)=4.73, p<.05; means=l 7.67 at T1 and 9.33 at T2) and a similar non-
significant pattern of improvement for internalizing problems (means=28.33
at T1 and 19.00 at T2). In contrast, the High SRG demonstrated significant
Egan—ye change over time for both internalizing and anxiety/depression
problems (t(3)=-25.98, p<.001; means=21.50 at T1 and 29.00 at T2 and
t(3)=-4.33, p<.05; means=12.25 at T1 and 15.50 at T2, respectively).

Within the Moderate TRG (Figure 5), the Low/moderate SRG showed
a statistical trend for improvement for anxiety/depression problems

(t(10)=1.96, p<.10; means=9.63 at T1 and 8.36 at T2) and a similar
79

High Trait

 

 

15‘ --I - Low/mod. State
1‘ (n=1)
‘- _ + High State (n=4)

 

" 4" .. Unstable State (n=3)

 

 

 

Self Report of Intemalizing
Problems
:3

 

 

 

Time

High Trait

 

 

- - I - - Low/mod. State

_ (n=1)
0'5 ‘ >< +High State (n=4)
0 a

-05 . -—O— Unstable State (n=3)

 

 

 

 

Self Report of Anxiety/depression

 

 

Time

Figure 4

Interactions Between State Reactance Group and Change Within the Iflgh
Trait Reactance Group for Self Report of Intemalizingjroblems

8O

Moderate Trait

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.5
u -
.5 2
a 1.5 ~ "
e 1 — , ,-
g E 0-5 ‘ ¢// - l- Low/mod State (n=ll)
:cé 0‘ " '-*--l-Iighstate(n=1)
l--—————I
E 3: ’0'5 ‘ I" —o——Unstablc State (n=3)
a. ,1 -
0
g -1.5 ~
r3 .2 -
-2.5 r
1 2
Time
Moderate Trait
2.5
E 2 ~ ‘
§ 1.5 — v'
a .9
e 1 ~
13‘ 0.5 —I—Low/rnod State (n=ll)
E 04 ~~1t--Highstate(n=1)
if 0 5 —O———Unstable State (n=3)
6 " .
g -1 - ..
a; -1.5 - ‘
-2.5
1 2
Time
Figure 5

Interactions Between State Reactance Group and Change Within the
Moderate Trait Reactance Group for Self Report of Intemalizing Problems

81

non-significant pattern of improvement for internalizing problems
(means=16.72 at T1 and 15.91 at T2). In contrast, the Unstable SRG
demonstrated significant negative change over time for internalizing
problems and a statistical trend for anxiety/depression problems (t(2)=-4.95,
p<.05; means=24.67 at T1 and 39.33 at T2 and t(2)=-2.77, p<.10;
means=10.67 at T1 and 18.67 at T2, respectively). Notably, the Low trait
(and Low/moderate state) reactant group showed significant improvement
over time for anxiety/depression (t(3)=3.29, p<.05; means=14.00 at T1 and
5.00 at T2) and a statistical trend in the same direction for internalizing
problems (t(4)=2.31, p<.10; means=17.20 at T1 and 7.00 at T2).

Staff report data provided more mixed support for Hypothesis 3. The
one significant TRG x SRG x Change interaction for staff report of
internalizing problems (somatic problems; Figure 6) and the one statistical
trend (internalizing problems; Appendix I) Showed the same pattern.
Contrary to Hypothesis 3, within the High TRG, the High SRG showed
improvement (t(3)=4.90, p<.05, means=3.37 at T1 and 2.38 at T2) whereas
the Unstable SRG demonstrated no change over time (means=3.67 at T1
and 3.33 at T2). Alternatively, and more consistent with Hypothesis 3,
within the Moderate TRG, the Low/moderate SRG showed improvement

(t(10)=3.91, p<.01, means=4.64 at T1 and 2.55 at T2) whereas the Unstable
82

High Trait

I"
or

 

.5
01”

d
I _ l

 

.0
0t

. - - C . Low/mod. State (n=1)‘
---"'t'"':‘ "' .. +High State (n=4) i

I. ' ' ~ 4" ~ Unstable State (n=3) J

 

 

 

I
. p
-‘> 0'! o

-1.5 4

Staff Report of Somatic Problems

 

 

-2.5 r

 

Time

Moderate Trait

 

 

0-5 ‘ +1.0w/mod State (n=11)
0. .- -.A --*--Highstate(n=l)

-0.5 ~ , . - ~ ‘ ' ' +Unstable State (n=3)

 

 

 

Staff Report of Somatic Problems

 

 

 

Time

Figure 6

Interaction Between State Reactance Grogp and Charge Within the High
_a_n_d Moderate Trait Reactance Groups for Staff Report of Somatic Problems

83

SRG only demonstrated a statistical trend for change over time (t(2)=3.46,
p<.10, means=5.00 at T1 and 4.00 at T2). Notably, the Low trait (and
Low/moderate state) reactant group Showed no change over time.

For the most part, the 2-way interactions (TRG x Change or SRG x
Change) for internalizing problems were insignificant or else accompanied
by a statistically significant 3-way interaction. The one exception, a TRG x
Change interaction for self report of change in withdrawal problems, was
accompanied by a statistical trend for the 3-way interaction (see Appendix
H). Hence, the 2-way interaction is not discussed any further.

Extemalizing Problems. There were no Significant TRG x SRG x
Change interactions for externalizing problems using either self or staff
report. However, there were several Significant findings for TRG x Change
interactions. The most interesting were the two significant interactions for
self reported change (Figure 7). The Low TRG showed a statistical trend
for improvement in self reported aggression problems (T(4)=2.48, p<.10,
means=12.00 at T1 and 5.40 at T2) and a similar non-significant pattern for
self reported attention problems (means=7.25 at T1 and 3.25 at T2). In
contrast, Moderate and High TRG’S showed no significant change. The one
significant TRG x Change interaction for staff report of externalizing

problems (i.e., impairment in undercontrolled behavior, Figure 8) differed in

84

 

 

1 i + Low Trait (n=5)

- I - Moderate Trait

""""" (n=15)
-0.5 4 \ .. .‘ .. High Trait (n=8)
-1 4

 

 

 

Self Report of Aggression
o
I

 

 

 

Time

 

 

1 ‘ — o- Low Trait (n=4)

- I - Moderate Trait
‘ ~ (n=14)

'95 ‘ “-- -—-4- High Trait (n=8)

 

 

 

Self Report of Attention Problems
0
p
II
I i
i
l
I
I
|
i
1
I
I

 

 

 

Time
Figure 7

Interaction Between Trait Reactance and Changes in Self Report of
Agggssive Problems and Attention Problems

85

 

 

 

 

 

 

 

 

_ 3 2-
“'3
.5 3g 1.51
‘2 8 1—
=1 3", ._ _ — . =
f: '2 3 0'5 7 “ ******** A § Low Trait (n 5)
"ED-E 04 l-———_ ,v’. -l-ModerateTrait(n=15)
9.58 -05- ,-""--I - . _
8'2 ' o" ‘ "it- H1ghTra1t(n—8)
at; -1~
Er}; 4.5-
mg _24
’2.5 1
1 2
Time
Figure8

Interaction Between Trait Reactance and Changes in Stgfifl' Report of
Impairment in Undercontrolled Bemvior

86

pattern from that just described for self report of change in externalizing
problems. However, in the case of staff reports of undercontrolled behavior,
none of the changes within each of the TRG’S were statistically significant.
Notably, the 3-way interaction for staff reports of undercontrolled behavior
Showed a statistical trend (see Appendix J for graph). There were no
significant SRG x Change interactions for self or staff report of
externalizing problems.

Total Problems. In addition to the findings for internalizing and
externalizing problems, there were two significant findings for self report
(but not staff report) of overall problems. Figure 9 illustrates the significant
TRG x Change interaction for total problems on the Achenbach with the
Low TRG showing improvement in total problems (T(4)=2.45, p<.10,
means=62.80 at Tland 27.00 at T2) and the Moderate and High TRG’S
showing no change. Notably, the 3-way interaction for this variable showed
a statistical trend and was consistent with Hypothesis 3 (see Appendix K for
graph). Finally, there was one significant SRG x Change interaction for self
report of total impairment on the F ISCA. Figure 10 shows that the Low
SRG showed significant improvement in impairment over time (T(15)=3.32,
p<.01, means=75.63 at T1 and 45.63 at T2) whereas High and Unstable

SRG’S Showed no change.
87

 

 

0.5 ~ A- ~~~~~~~~~~~ 1. + Low Trait (n=5)
- I - Moderate Trait (n=15)

 

 

 

-0.5 - \‘ " "' High Trait (n=8)
-1 1

-1.5 ‘1

Self Report of Total Problems
0
l
l
l
l

l

i

i

l
I
h

 

 

 

Time

Figure 9

Interaction Between Trait Reactance and Changes in Self Report of Total
Problems

88

 

 

 

 

 

 

 

 

 

s 2*
g 1.5 '1
3. 1 - +Low/mod
E 0.5 - State (n=l6)
2,! 0 g - i - High State
I" =5)
"5 -0 5 4 (n
t: ' r: 4" .... Unstable State
E- -1 ‘ (n=6)
g -1.5 -
t2 -2 4
-2.5
1 2
Time

Figure 10

Interaction Between State Reactance and Charges in Self Rgport of Total
Impairment

89

Treatment Group as a Between Sulficts Variable
As noted earlier, residents were randomly assigned to one of three

treatment groups and completed all of their treatment and activities with
their group. Although programming was essentially the same for all
treatment groups, residents never interacted with members of other groups.
Due to the small sample size, it was not possible to investigate the impact of
treatment group by running a 3 x 3 x 3 ANOVA with Group, TRG, and SRG
as between subjects variables. However, two separate 3 x 3 ANOVA’S were
run in order to assess for any Group x TRG x Change or Group x SRG x
Change interactions. No Group x TRG x Change interactions were
significant. However, four Group x SRG x Change interactions were
observed for the self report (but not the staff report) of changes in
Achenbach variables, including internalizing problems (F(3,20)=5.50,
p<.01), anxiety/depression problems (F (3,19)=8.60, p<.001), thought
problems (F(3,17)=3.66, p<.05), and total problems (F(3,20)=4. l 3, p<.05).

Figures 11, 12, 13, and 14 show the patterns of SRG x Change
interactions within each of the three treatment groups. The pattern for
Group 1 was similar across all outcome variables. The Unstable SRG
showed significant improvement for internalizing problems (t(1)=29.00,

p<.05, means=3l.00 at T1 and 16.50 at T2). Also, a statistical trend for
90

Group 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

» 2.5
5 2«
- z 1.5-
g E 1 ‘ k
a 0 5 " — —- =
g u -0 ‘k a Low/Mod. State (11 8)
i g .05 4 ”‘3 —~a—-—- Unstable State (n=2)
2 7'5 -1 .
'3 E -1.5-
3 -2 i
.5 -2.5
1 2
Time
Groupz
-» 2.5
8 2-
'5 E 1.5 ~ A
:5 0;; ”Me-r“ —-I-Low/Mod.State(n=6)
32° or 5;" """""" ' -a-— HighState(n=2)
:5 “0'5 ‘ " -------- I n. - -Unstab1e State (n=1)
2 '3 -1 .
‘3 E 4.54
3 -2 -
.5 -2.5
1 2
Time
Group3
a 2.5
3 24
‘5 a 1.5 .
1: E 1 ‘ — l — Low/Mod. State (n=3)
2' 00 0.54 | .
a 0 4 +ngh State (n=3)
5 s -05- r” ————————
, 'i ;1 ‘ " +Unstable State (n=3)
(3 E 4.54
a .24
.5 -2.5 .
1 2
Time
Figure 11

Interaction Between Group, State Reactance, and Changes in Self Report of
Intemalizing Problems

91

Group I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.5
= 2 .
«5 'g 1.5 ‘
t 2 O; a..-
g g- ’0 , ._x_ _. — I — Low/Mod. State (n=8)
g >. -0.5 - \‘fi .4... Unstable State (n=2)
’6 -1 -
5 E 4.5-
< -2
-2.5
1 2
Time
Group 2
2.5
3 2'
‘5 - “15‘ ,A
E- g- 05 : I ...... :fl- (3,:- ’2. I - '. "' LOW/MOd. State (11:6)
‘25 .0151: t‘:: _____ I --A--HighStatc(n=2)
i; ;1; -“ ~-I~~Unsrablesmte(n=l)
-1.5 -
< -2 <
-2.5
1 2
Time
Group 3
2.5
fl 2 '1
'5- '§ 1.? -
E. g 0.5 ._ ‘ ~ / - l - Low/Mod. State (n=2)
aé 0g: .7— ~~e~~ ‘ +HighState(n=3)
3 g‘ ;1 , 7' +Unstab1e State (n=3)
4.5 J
5 -2
-2.5
1 2
Time
Figure 12

Interaction Between Group. State Reactance, and Changes in Self Report of
Anxiety/Depression Problems

92

Group 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.5
1 2
“ 1.5 1
t2 ,, 1 fl
'5 E F N .s
g .2 “3‘ H 'r- “-2 ... ..... * _ +1.0w/Mod. State (n=8)
g- $- -0.5 - 1 - ‘B - Unstable State (n=2)
a 9n -1 .
h -1.5 —
t3 -2 ~
-2.5
1 2
Time
Group 2
2.5
11 2 -
:3 1.5 «
‘5 E 0;: ”,4 —I—LowlModState(n=6)
£73 '0. E'Wfixif ..... g «as-— High' State(n=2)
E '0.5 ‘ 2 ——————— l . . . v =
a at -1 1 I Unstable State (11 1)
a: -1.5 -
t3 -2-
-2.5
1 2
Time
Group 3
2.

 

 

5

2

.5 -

:15: / - l- Iflw/Mod State(n=3)
0

5

 

 

 

 

Self Report of Total
Problems

 

 

o — 4 .
< t- ******* — i - High State (n=3)
.03 .— ———————— -I +Unstable State (n=3)
4.5 «
.2 .
-2.5
1 2
Time

Figure 13

Interaction Between Group, State Reactance, and Changes in Self Report of
Total Problems

93

Group 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

*- 2.5
Ea 1?: “"~--......_~_.
0.5 ,2 0.3 ._ ______ ‘“ “‘3 - -l - Low/Mod. State (n=7fl
E- ?- -0.5 d _. ~— +8 - Unstable State (n=2) 1
o a. -1 J
II! -1.5 ~
’5 -2 l
m '2.5 r
1 2
Time
Group 2
'- 2.5
'5. 2 4
g 1.5 i i
i " 1 J .9”?
._ g 0.5 a I,“ - I - Low/Mod. State (n=5)
1?; .5 _og- r’f’ffih _. --A- HighState(n=2)
§. 3. ;1 2 e" --I - -Unstab1eState(n=1)
K 45 j
% -2
m '2.5 r
1 2
Time
Group 3
o- 2.5
'5. 2 «
3 1.5-
E g 0.; +Low/Mod. State (n=2)
“9% 424 ---A--HighState(n=3)
é. a. ;1 j - a - Unstable State (n=3)
:1 -1.5 «
.2 a
'3 -2.5
1 2
Time
Figure 14

Interafiction Between Group, State Reactance; and Changes in Self Report of
Thoght Problems

94

anxiety/depression problems (t(l)=10.00, p<.10, means=18.00 at T1 and
8.00 at T2), and a similar non-significant pattern of change for thought
problems and total problems (means=9.50 at T1 and 16.50 at T2,
means=101.50 at T1 and 65.50 at T2, respectively) suggested improvement
for the Unstable SRG. In contrast, the Low/moderate SRG within Group 1
did not change over time. None of the residents in Group 1 exhibited High
state reactance.

Changes in the outcome variables in Group 2 generally were
insignificant. However, Group 3 showed different patterns of change
depending on the outcome variable with internalizing, anxiety/depression,
and total problems showing a different pattern of change than thought
problems. In general, for the first three variables, changes for the High SRG
and Unstable SRG were negative, whereas there was essentially no change
among the Low/moderate SRG. In particular, the High SRG Showed
Significant negative change for internalizing problems (t(2)=-22.00, p<.01,
means=18.33 at T1 and 26.67 at T2), a statistical trend for
anxiety/depression problems (t(2)=-4.0, p<.10, means=11.67 at T1 and
14.33 at T2) and a similar non-Significant pattern for total problems
(means=77.67 at T1 and 80.67 at T2). Similarly, the Unstable SRG showed

significant negative change for internalizing and total problems (T(2)=-4.95,

95

p<.05, means=24.67 at T1 and 39.33 at T2 and t(2)=-8.36, p<.05,
means=94.67 at T1 and 125.33 at T2, respectively) and a statistical trend for
anxiety/depression problems (t(2)=-2.77, p<. 10, means=10.67 at T1 and
18.67 at T2). In contrast, the Low/moderate SRG within Group 3 showed
no change over time for internalizing, anxiety/depression, or total problems.
Alternately, for thought problems (Figure 14), the High and Unstable SRG’S
did not change whereas the Low/moderate SRG showed improvement

(t(1)=13.00, p<.05, means=9.50 at T1 and 3.00 at T2).

96

DISCUSSION

Although these results were relatively complex, in general, the

following findings emerged:

l. Hypothesis 1 was supported by the results. Trait reactance was
relatively stable over time whereas state reactance evidenced more
change over time.

2. Hypothesis 2 also was supported by the results. Trait and state
reactance were moderately correlated and trait reactance appeared
to restrict the range of state reactance.

3. Results for Hypothesis 3 were mixed and included the following
patterns:

a. Change in general, as well as interactions between reactance
and change in particular, emerged for the most part when
the data were based on self, rather than staff, reports of
outcomes.

b. The direction of change varied as a function of type of
pathology with residents in general showing more positive
change for externalizing problems than internalizing

problems.

97

c. State reactance moderated the relationship between trait
reactance and change for internalizing problems but not
externalizing problems.

(1. Trait reactance, and to a much lesser degree state reactance,
were directly related to change in externalizing problems
and total problems but not internalizing problems, with only
low (and not moderate or high) reactant residents Showing
positive change.

e. Implications of state reactance for change was more affected
by context than the implications of trait reactance for
change.

The discussion considers each of these general findings.

Finding for Hypothesis 1
The results provided overwhelming support for the greater stability of
trait reactance as compared to state reactance. As a personality variable,
trait reactance was expected to be more stable whereas state reactance was
expected to be less stable because by definition it is situationally driven. In
fact, state reactance was even less stable than anticipated. The residents’

environment in the facility was much more controlled than a “real world”

98

environment with the residents interacting with the same people each day
and following the same schedule each day. As a result, state reactance
might be expected to be more stable in a residential environment in which
freedom is restricted than in an environment that allows more freedom to
choose. Accordingly, it was assumed that at each time point state reactance
would vary at most between two adjacent levels; for example, high state
reactant individuals would demonstrate state reactance that varied between
high and moderate levels. It was not expected that some residents could not
be classified as low, moderate, or high because their state reactance scores

spanned the range of all three classifications.

Findings for Hypothesis 2

Hypothesis 2 was also supported by the data. As predicted, state and
trait reactance were moderately correlated with trait reactance to some
extent, but not entirely, constraining the individual variation in state
reactance. In particular, it was hypothesized that low trait reactant subjects
mostly would exhibit low to moderate state reactance whereas high trait
reactant subjects would be more likely to exhibit high to moderate state
reactance. The data demonstrated that trait reactance did in fact restrict the

range of observed state reactance. Each resident appeared to have a

99

characteristic level of reactance (trait reactance) but characteristics of the
environment to some extent affected the actual level of reactance
experienced in each Situation (state reactance). Nevertheless, low trait
reactant residents never reported high state reactance and high trait reactant
residents never reported low state reactance. It may not be possible for high
trait reactant residents to experience low state reactance for any length of
time due to their predisposition not to be influenced by others. Instead, the
closest approximation to low reactance for this group was that state
reactance was highly variable fluctuating between high, moderate, and

occasionally low (i.e., unstable state reactance).

Findings for Hypothesis 3

Support for Hypothesis 3 was mixed. Hypothesis 3 predicted state
reactance would moderate the relationship between trait reactance and
treatment progress. In particular, it was expected that treatment progress
would be better if state reactance was lower than trait reactance and not as
good if state reactance was higher than trait reactance. However, the
emergence of the unstable state group required some modification of the
original hypotheses. The distribution of state reactance across trait

reactance groups implied that unstable state reactance was an intermediary

100

between the moderate and high state reactance groups. Unstable state
reactance appeared to be relatively “low” for the high trait reactance group
and relatively “high” for the moderate trait reactance group. Hence, it was
expected that high trait reactant residents with unstable state reactance
would Show more treatment progress than high trait reactant residents with
high state reactance. In contrast, it was expected that moderate trait reactant
residents with unstable state reactance would Show less treatment progress
than moderate trait reactant residents with either low or moderate (i.e.,

low/moderate) state reactance.

Self versus Staff Report

For residents as a whole, the self report data showed significant
change on half of the outcome variables, whereas the staff report data
Showed no significant (overall) change. This discrepancy can be better
understood when differing methodologies and perspectives associated with
self and staff report are taken into account. Firstly, the time span for
observing changes differed for self and staff reports. Changes in self report
data reflect changes over a three month period whereas changes in staff
report data reflect changes over a one and a half month period. Whereas

self report data were collected at admission and three months post

101

admission, gaff report data were collected at 1.5 months post admission and
at three months post admission. The discrepancy in reporting intervals was
necessary in order to allow staff raters a reasonable period of time to
observe the resident’s behavior before reporting on it. Secondly, self report
data reflect changes between no treatment and three months post treatment
whereas staff report data reflect changes between 1.5 and 3 months post
treatment. It would be expected that there would be a bigger difference
between behavior prior to treatment and 3 months into treatment than
between 1.5 months into treatment and 3 months into treatment. Thirdly,
staff can not readily observe internal changes in the residents and must rely
on the residents’ disclosing information regarding changes in cognitions and
feelings. Finally, staff reports may be biased by residents’ reputations,

which can be relatively stable despite change in behaviors.

Intemalizing Versus Extemalizing Problems

These data showed that the overall direction of change varied as a
function of type of pathology. Similarly, Swenson and Kennedy (1995)
studied treatment outcomes for adolescent offenders and found different
predictors of treatment outcome for adolescent offenders with externalizing

behavior problems as compared to internalizing behavior problems. In the

102

present study, residents reported a reduction in externalizing problems and
an increase in certain internalizing problems over time. For example,
delinquency and undercontrolled behavior declined during treatment. This
is not surprising given the high degree of structure and supervision in the
treatment facility that limit opportunities for acting out and delinquency.
However, whether the changes in externalizing behavior will be maintained
after discharge remains unknown.

Although the pattern of change for several internalizing problems was
complicated by reactance, self reported somatic problems increased over
time across all reactance groups. There are two possible explanations for
the increase in this component of internalizing pathology: one Situational-
Specific and one more theoretical. AS part of the program, the residents
were able to fill out daily forms detailing physical concerns. After
completing the form, the residents met with the unit nurse to discuss these
health concerns. The nurse was very nurturing and these contacts were
probably reinforcing. In addition, because of the increased access to
medical care and increased medical education, the residents may have
become more aware of their physical problems since being placed in the
facility.

A more theoretical explanation focuses on manipulation of distress in

103

order to facilitate change. Beutler (1988, 1990, & 1991) argues that in order
for treatment to be effective there needs to be an optimal level of
motivational distress. Since externalizing problems are not experienced as
distressing, in order to increase motivation for change, it may be necessary
to initially increase acute distress, a defining aspect of internalizing
problems (Phares & Copas, 1990). Similarly, Swenson and Kennedy (1995)
found that adolescent offenders with externalizing problems showed more

favorable treatment outcomes when they reported anxiety.

The Moderating Effect of State and Trait Reactance on Treatment Proggss
The expectation that state reactance would moderate the relationship
between trait reactance and treatment progress found support when progress
was defined as a decrease in internalizing problems but not externalizing
problems. It is possible that internalizing problems may be more
susceptible to influence by internal motivations such as reactance whereas
externalizing problems may be more controlled by the external environment.
However, it is not clear whether improvements in externalizing problems
may by linked to internal motivations as well when adolescents are not
living in such a controlled environment. For example, when the residents

are living in the community and are responsible for controlling their own

104

behavior, control of externalizing behavior may be more influenced by
internal motivations. In fact, in the absence of environmental controls, it
may be necessary to Shape internal motivations such as state reactance in
order to produce positive change.

Patterns of change associated with Trait reactance group x State
reactance group x Change interactions for self reports of internalizing
problems were consistent with Hypothesis 3. In particular, within the high
trait reactance group, the unstable state reactance group showed a greater
decrease in internalizing problems than the high state reactance group; and
within the moderate trait reactance group, the low/moderate state reactance
group Showed more treatment progress than the unstable state reactance
group. Accordingly, treatment progress in the unstable state reactance
group varied as a function of trait reactance group, with change in opposite
directions for the high and moderate state reactance groups. Assuming, as
suggested, that unstable state reactance is a relative reduction in reactance
for high trait reactant residents and an increase in reactance for moderate
trait reactant residents, these findings support the hypothesis that the
unstable state reactance has different implications for moderate and high
levels of trait reactance.

The 3-way interaction found for staff report of somatic problems

105

partially supported Hypothesis 3. In particular, the within group patterns of
change for the moderate trait reactance group were Similar to those for the
self report data and as such, consistent with the hypothesis. In contrast, the
pattern for the high trait reactance group was inconsistent with Hypothesis
3, predicting that unstable state reactance group would Show more treatment
progress than the high state reactance group. Rather, the high state
reactance group showed more treatment progress than the unstable state
reactance group. This discrepancy may be related to the nature of the
outcome variable (i.e., somatic problems). Self reported somatic problems
increased over time. AS discussed earlier, this increase may have been due
to the residents seeking out comfort from the staff. High trait reactant
residents would be less likely to overtly seek help and comfort because they
value autonomy and do not like to rely on others for help (Dowd &
Wallbrown, 1993, Dowd et al., 1994). Consequently, these high trait
reactant residents might appear to staff to have fewer somatic problems
since help seeking behavior may be the only way in which somatic problems
would be identifiable to the staff. In contrast to staff report, self report data
would more accurately reflect changes residents’ private experiences of in
somatic problems. Alternatively, it is possible that the increase in somatic

problems was due to an informal norm maintaining by peer pressure for

106

residents to express distress in the form of somatic problems. Since the high
trait and state reactant residents would be more likely to resist pressures to
conform, this group may have been the only group to show treatment

progress for these types of problems.

Direct Effects of Trait and State Reactance on Treatment Progress

Although Hypothesis 3 predicted that state reactance would moderate
the effects of trait reactance on treatment progress, several direct
relationships between trait reactance and change emerged in the data.
Consistent with past research (Courchaine et al., 1985; Dowd. et al., 1988;
Oliver et al., 1993), the low trait reactance group showed improvement in
self report of internalizing problems, aggressive problems, attention
problems, and total problems. In addition, even when the data identified
significant effects of state reactance within the moderate and high trait
reactance groups, the low trait group, all of whom consistently reported only
low or moderate state reactance, showed relative improvement in
internalizing problems.

The one significant direct association between state reactance and
change showed that the low/moderate state reactance group was the only

group to show significant change in self reports of overall impairment. This

107

finding is consistent with previous work by Chamberlain et al. (1984).
However, the larger number of direct effects for trait reactance as opposed
to state reactance and the magnitude of these effects suggest that trait
reactance may be more influential in predicting treatment progress than state
reactance. This is inconsistent with Mulry et al. (1994) who studied the
effects of both state and trait reactance on treatment approaches for a non-
clinical population. Based on their findings, these researchers concluded
that state reactance was more influential than trait reactance in predicting
treatment outcome. This discrepancy may be due to the manner in which
state reactance was measured. In particular, the study by Mulry artificially
manipulated state reactance and produced two extremely different levels of
state reactance (i.e., high or low). High state reactant participants were told
they could choose the treatment method and were then assigned to the non-
preferred method without explanation. The level of state reactance
produced by being purposely misled is likely to be different in intensity and
quality from the state reactance experienced in a treatment setting designed
to help residents. This type of extreme fi'ustration may have been
responsible for the differential impact of state reactance as opposed to trait
reactance. In addition, the artificial manipulation of state reactance may

have minimized the impact of trait reactance in predicting the range of state

108

reactance.

Impact of Trait and State Reactance on Change for the Three Treatment
9%

Additional analyses evaluated the impact of trait and state reactance
on change as a function of treatment group. The results showed a clear
pattern with four significant Group x State x Change interactions and no
significant Group x Trait x Change interactions, suggesting that state
reactance is more affected by social context than trait reactance. In
particular, these interactions suggest that different levels of state reactance
may be more beneficial in different social settings. In one treatment group,
the unstable group was the only group to show significant improvement. In
another treatment group, the low/moderate group Showed improvement
whereas the high and unstable groups became more dysfunctional over time.
Since it is not possible to test the 4-way interaction due to sample size, it is
not possible to determine if the differential impact of state reactance is
moderated by trait reactance. As discussed in previous findings, unstable

state reactance had different effects given moderate or high trait reactance.

109

Limitations and Future Directions

While the present study expanded the conceptualization of reactance
by Simultaneously investigating the differential impact of trait and state
reactance on treatment progress in the “real world”, generalizability is
limited by several methodological problems. Most importantly, a larger
sample is needed to bolster statistical power and ensure generalizability.
Basta and Davidson (1988) noted that when studying the effectiveness of
treatment for juvenile offenders, this is an especially salient concern.

In addition, the generalizibility of the present findings is limited to
male middle adolescent offenders. Future research should investigate
whether these findings can be replicated with females, individuals at other
stages of developmental (i.e., children, preadolescents, early adolescents,
and adults) and other clinical populations (i.e., outpatients, psychiatric
inpatients, and non incarcerated offenders). This is particularly important in
light of Frank et al.’S (1998) work that studied psychiatric inpatients and
found that trait reactance had different implications on treatment outcomes
for male and female participants and participants in early and middle
adolescence.

Although this study assessed treatment progress using multiple

measures completed by different informants, only one measure of each type

110

of reactance was used. Future research may want to include additional
measures of trait reactance (i.e., Hong & Page, 1989; Merz, 1983) and
investigate different components of trait reactance. For example, future
research should evaluate whether Dowd et al.’s (1991) verbal and
behavioral subscales of trait reactance and Hong and Page’s (1989) freedom
of choice, conformity, reactance to advise, and behavioral freedom
subscales of trait reactance have a differential impact treatment progress. In
addition, future research should further validate the current measure of state
reactance.

While this study attempted to assess treatment progress through
behavioral observations, these data were not sufficiently reliable to be
included in the analyses. This study demonstrates the risks of relying on
untrained treatment providers as direct observers. When including
behavioral observations, future researches should either train treatment
providers or use trained unbiased observers.

Since this research demonstrated that both state and trait reactance
impact treatment progress, future research should evaluate which
components of the treatment account for observed changes. In particular,
the relationship between different treatment approaches and different levels

of reactance needs to be assessed for adolescents. Similarly, which

111

treatment approaches are associated with a reduction in adolescent’s state
reactance needs to be investigated.

Finally, future research Should expand the window of change to
include longer periods for change and post treatment follow up. Ideally, the
goal of treatment effectiveness studies is to present data on the long-term
effectiveness of treatment. In particular, investigations of offender
populations are interested in the prevention of antisocial acts in the long
term (Basta & Davidson, 198 8). Conceivably, high reactant individuals
may be more likely than low reactant individuals to maintain improvements
because they resist the negative influence of others. This is particularly
important for adolescents, Since most adolescents must confront family
members and peers who may resistent any changes that could alter the

family or peer group homeostasis.

Conclusions
In general, the results provided support for the first two hypotheses
and mixed support for the third. In particular, trait reactance was more
stable than state reactance over time (Hypothesis 1) and trait and state
reactance were moderately correlated, with the level of trait reactance

restricting the range of state reactance (Hypothesis 2). Also for certain

112

outcome variables, State reactance moderated the effect of trait reactance on
treatment progress (Hypothesis 3). The last hypothesis only found support
when evaluating change in internalizing problems (as opposed to
externalizing problems). In addition, only self reports (as opposed to staff
reports) of change demonstrated the expected effects of state reactance as a
moderator between trait reactance and treatment progress.

The emergence of the unstable state reactance group was an
unexpected but theoretically important discovery that may also prove
clinically significant. In particular, unstable state reactance appeared to be
an intermediate level between moderate and high state reactance.

Moreover, the implications of unstable state reactance varied as a function
of trait reactance. In particular, the unstable state reactance group was more
likely to change in a positive direction when accompanied by high trait as
opposed to moderate trait reactance.

There are several implications of the current findings for clinicians
working with adolescent offenders. First, the results indicated that, on
average, lower reactant (state or trait) adolescents are more likely to Show
treatment progress than higher reactant adolescents. Accordingly, clinicians
Should develop treatment plans that focus on manipulating adolescent

reactance and use treatment approaches proven to lower state reactance

113

levels, especially for high reactant adolescents. For example, Beutler (et a1,
1991) suggested using non-confrontational or paradoxical approaches with
high reactant adult clients. In addition, while it appears possible for a
clinician to impact the level of state reactance, the potential range of state
reactance is constrained by the adolescent’s level of trait reactance. In
particular, for high trait reactant adolescents, clinicians Should recognize
that low state reactance may not be feasible and instead strive for more
variable state reactance that at some points is low (i.e., unstable state
reactance). Finally, effectively shaping the client’s state reactance may
facilitate treatment progress for internalizing problems but not necessarily
externalizing problems. Since externalizing behavior appears to be more
directly controlled by the external environment, clinicians working with the
adolescent with externalizing problems should consider involving the
family and implementing structural changes in the environment that limit

the opportunity for acting out.

114

APPENDICES

115

< Appendix A

Trait Reactance

116

Personal Attitudes Inventory
Instructions: Please answer each question by circling your answer.

SD=Strongly Disagree D=Disagree A=Agree SA=Strongly Agree

 

1. If I disagree with my teachers, I let them know. SD D A SA

 

2. I get annoyed at adults whotryto tell me what to SD D A SA
do.

 

3. Ioften find I have to question adults' decisions. SD D A SA

 

4. I enjoy seeing someone else do something that I SD D A SA
know we are not supposed to do.

 

5. It's very important to me to be free to do what I SD D A SA
want.

 

6. I often keep an argument going because I just SD D A SA
don't want to give in.

 

7. In discussions, I am easily persuaded by others. SD D A SA

 

8. Nothing turns me on as much as a good SD D A SA
argument.

 

9. When I have a job to do, I like it better when no SD D A SA
one tells me how to do it.

 

10. If I am told what to do, I often do the opposite. SD D A SA

 

11. I am sometimes afraid to disagree with others. SD D A SA

 

 

12. It really bothers me when police officers tell SD D A SA
people what to do.

 

117

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

13. I don't get upset about changing my plans when SD D A SA
someone in the group wants to do something else.

14. I don't mind other people telling me what to do. SD D A SA

15. I enjoy debating with other people. SD D A SA

16. If someone asks a favor of me, I wonder what SD D A SA
they're really after.

17. It irritates me when other people try to change my SD D A SA
mind.

18. I often follow other's suggestions. SD D A SA

19. I have very strong opinions about things. SD D A SA

20. It's important to me to have power over others. SD D SA

21. I'm very willing to listen to other people's advice SD D SA
when it comes to solving problems.

22. I enjoy "Showing up" other people who think they SD D A SA
are right.

23. I'm the kind of person who likes to compete more SD D A SA
than cooperate.

24. I don't mind doing something for someone even SD D A SA
when I don't know why I'm doing it.

25. I usually go along with others' advice. SD D A SA

26. I feel it is better to stand up for what I believe SD D A SA
than to be silent.

27. I am very stubborn and set in my ways. SD SA

28. When I have to work with other people, it's very SD A SA

important to me to get along well with them.

 

118

 

Appendix B

State Reactance

119

TIME: 1 2 3 4

Name:

Instructions: Please answer each question by circling your answer.

l=Definitely N o

 

Program Inventory

=Probably No

MRII:

Date:

3=Probably Yes

 

 

4=Definitely Yes

 

1.

Do you agree with the changes the staff is trying
to make in you?

1

2

3

4

 

Suppose you had a friend who was committing
some burglaries and car thefts. He was caught
once and put on probation and then caught again.
Do you think it would help this friend to be sent
here?

 

Do you think some of the staff here can help you
find out why you get into trouble and help you
change?

 

Do you think you need help from someone so
you can change and stay out of trouble?

 

On the whole, has this place been good for you
so far?

 

When you think about the time you have spent
here, do you think it was happy?

 

Do you think your life will be better because
you have spent time here?

 

 

Has being here been a bad experience for you?

 

120

 

Appendix C

Achenbach-Staff Report

121

0

0°

0

Below is a list oi items that describe children and youth. For each item that describes yoor child now or within the past 6 months. please Circle
the 2 ii the item is very true or often true oi your child. Circle the 1 ii the item is somewhat or sometime true oi your child. it the item is not
true oi y0ur child. circle the 0. Please answer all items as well as yoo can. even it some do not seem to apply to yoor child.

1 2
1 2

2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2

2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2
1 2

2
1 2
1 2
1 2

0 = Not True (as tar as you know)

1.
2.

9'

.V

10.

11.
12.

13.
14.

15.
16.

17.
18.

19.
20.

21.

22.

23.
24.
25.

27.
28.

29.

30.

Please Print
1 3 Somewhat or Sometimes True

Acts too y0ung for his/her age
Allergy (describe):

 

 

Argues a lot
Asthma

Behaves like opposite sex
Bowel movements outside toilet

Bragging. boasting
Can't concentrate. can't pay attention ior long

Can't get hislher mind oii certain thoughts;
obsessions (describe):

 

 

Can't sit still, restless. or hyperactive

Clings to adults or too dependent
Complains oi loneliness

Coniused or seems to be in a log
Cries a let

Cruel to animals
Cruelty. bullying. or meanness to others

Day-dreams or gets lost in his/her thoughts
Deliberately harms seli or attempts suicide

Demands a lot of attention
Destroys hislher own things

Destroys things belonging to his/her family
or others
Disobedient at home

Disobedient at school
Doesn't eat well

Doesn't get along with other kids
Doesn't seem to loot guilty aiter misbehaving

Easily jealous
Eats or drinks things that are not iood —
don't include sweets (describe):

 

Fears certain animals. situations. 0r places.
other than school (describe):

 

Fears going to scnoot

 

0

0° O0 00 CO

1

‘ddd

d“‘

2

N

N” N

NM

”NM”

”NM”

31.

40.

41.

42.

44.

45.

46.

47.

48.
49.

51.

2 a Very True or Otten True

Fears helshe might think or do something
bad

Feels helshe has to be periect
Feels or complains that no one loves himlher

Feels others are Out to get himlher
Feels worthless or interior

Gets hurt a lot. accident-prone
Gets in many tights

Gets teased a lot
Hangs around with others who get in trouble

Hears sounds or voices that aren't there
(describe):

 

 

impulsive or acts without thinking

Would rather be alone than with others
Lying or cheating

Bites iingernails
Nervous. highstrung. or tense

Nervous movements or twitching (describe):

 

 

Nightmares

Not liked by other kids
Constipated. doesn't move bowels

Too ieariul or anxious
Feels dizzy

Feels too guilty
Overeating

Overtired
Overweight

Physical problems without known medical

cause:

.Achesorpains(notstomachorheadeches)

. Headaches

Nausea. ieels Sick

. Problems with eyes (not it corrected by gas”)
(describe):

. Rashes or other skin problems
Stomachaches or cramps

. Vomrting. throwing up

. Other (describe):

apar-

 

30,0

 

 

 

’40! 1

122

Please see other side

0 a Not True (as far as you know)

Please Print
1 2 Somewhat or Sometimes True

2 a Very True or Often True

a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o 1 2 57. Physically attacks people 0 1 2 84. Strange behavior (describe):
0 1 2 58. Picks nose. skin. or other parts oi body
(describe):
0 1 2 85. Strange ideas (describe):
0 2 59 Plays with own sex parts in public
0 1 2 60 Play: with own :6)! parts too much a 1 2 86. Stubborn. sullen. or irritable
0 1 2 61. Poor school work 0 1 2 87. Sudden changes in mood or feelings
O 1 2 62 Poorly coordinated or clumsy o 1 2 88, Sulks a lot
0 1 2 63 Prefers being with older kids a 1 2 89. Suspicious
o 1 2 64 Prefers being with younger kids a 1 2 90. Swearing or obscene language
0 1 2 65 Refuses to talk 0 1 2 91. Talks about killing self
0 1 2 66 Repeats certain acts over and over: 0 1 2 92. Talks or walks in sleep (describe):
compulsions (describe):
0 1 2 93. Talks too much
0 1 2 67 Runs away from home 0 1 2 94. Teases a lot
0 1 2 68 Screams a lot
0 1 2 95. Temper tantrums or hot temper
0 2 69. Secretive. keeps things to self 0 1 2 96. Thinks about sex too much
0 1 2 70 Sees things that aren't there (describe):
0 1 2 97. Threatens people
0 1 2 98. Thumb-sucking
o 1 2 99. Too concerned with neatness or cleanliness
O 1 2 100. Trouble sleeping (describe):
0 2 71 Self-conscious or easily embarrassed
O 1 2 72 Sets fires
0 1 2 73 Sexual problems (describe): 0 1 2 101. Truancy. skips school
0 1 2 102. Underactive. slow moving. or lacks energy
0 1 2 103. Unhappy. sad. or depressed
0 1 2 104. Unusually loud
0 1 2 74 Showing off or clowning
o 1 2 105. Uses alcohol or drugs for nonmedlcal
o 2 75 Shy or timid “”9““ ‘°°’°"°°"
o 1 2 76 Sleeps less than most kids 0 ‘ 2 106 Vandalism
O 1 2 77 Sleeps more than most kids during day 0 1 2 107 Wets self during the day
and/or night (describe): a 1 2 108. Wets the bed
. 0 1 2 109. Whining
O 1 2 78 Smears or plays with bowel movements 0 ‘ 2 110 Wishes to be of opposite sex
° ' 2 79 59"“ p'°°'°""°”c”°°" o 1 2 111. Withdrawn. doesn't get involved with others
0 1 2 112. Worries
° 1 2 8° Stares blankly 113. Please write in any problems your child has
that e n t l' t d :
0 2 81. Steals at home w re 0 '3 e above
0 1 2 82 Steals outsrde the home 0 1 2
O 1 2 83 Stores up things helshe doesn't need 0 1 2
(describe):
0 1 2

 

 

 

 

PLEASE BE SURE YOU HAVE ANSWERED ALL ITEMS.

DAG. a

123

UNDERLINE ANY YOU ARE CONCERNEO ABOUT.

Appendix D

Achenbach-Self Report

124

Below is a list of items that describe kids. For each item that
very true or often true of you. Circle the t if the item is some

3
describes you now or within the past’months. please circle the 2 it the item is
what or sometlmes true of you. If the item is not true of you. circle the a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Please Print 0 a Not True 1 - Somewhat or Sometimes True 2 a Very True or Often True
0 1 2 1. i set too young for my age 0 1 2 4o. i hear saunds or voices. that other people
0 1 2 2. l have an allerdy (describe): mm" "9"" there (“3m“):
0 ‘ 2 3 l argue a lot a 1 2 41. I act without stopping to think
0 ‘ 2 ‘ ‘ have asthma 0 1 2 42. I would rather be alone than with others
0 1 2 5. l act like the opposite sex 2 : : :3 : :eo'mihieizgemails '
o 1 2 e. i like animals ' '
0 ‘ 2 7‘ ' brag O 1 2 45. I am nervous or tense
. o 1 2 46. Parts of My body twitch or
o 1 2 a l have trouble concentrating
0' paying attention make nervous movements (describe):
0 1 2 9. l can‘t get my mind off certain thoughts
(describefi
0 1 2 47. l have nightmares
0 1 2 48. lam not llked'by other kids
0 1 2 10. i have trouble sitting still ° 1 2 ‘9' :51"ng ”2;" """9' m"
0 1 2 11. I'm too dependent on adults 0 ‘ 2 50 i am we fearful “ anxious
o 1 2 12. I feel lonely 0 1 2 51. I feel diay
o 1 2 13. i feel confused or in a log 0 ‘ 2 52; I feel we gum,
° ' 2 "‘ ' c" ' '°' 0 1 2 53. I eat too much
0 1 2 15. I am pretty honest 0 ‘ 2 5‘ I t”. overtired
O 1 2 16. i am mean to others 0 1 2 55' i am overweight
O 1 2 17. l daydream a lot ' .
56. Physical problems without blown medal
o 1 2 18 l deliberately try to hurt or kill myself m:
0 l 2 ‘9- ' "7 ‘° 9“ ' '°‘ °' "mm“ o 1 2 a. map-hernotmermi
o 1 2 20. l destroy my own things 0 1 2 b. Headaches
o 1 2 21. l destroy things belonging to others 0 ‘ 2 c. m ieefsick
° 1 2 22 'dimmp‘m‘” o 1 2 d ProblemwilheyeunetloonseledbyM)
0 1 2 23. i disobey at school (m);
0 1 2 24. I don't eat as well as I should
0 1 2 25. I don‘t get along with other kids
0 1 2 26. i don‘t feel guilty after doing
something I shouldn't
0 1 2 27‘ ' am 10.10118 of others 0 1 2 e Rashes or other skin problems
0 1 2 23. i am willing to help others 0 ' 2 " s‘°""_ a, “ch“ °' mm“
when they need help 0 1 2 9- Vomiting, 1W1” 0P
o 1 2 29. I am afraid of certain animals. situations. 0 1 2 it. Other (describe):
or places. other than school
(describe):
0 1 2 57. l physically attack people
0 1 2 58. lpiekmyskinorotherpartsofmybody
o 1 2 30. I am afraid of going to school (describe):
a 1 2 31. i am afraid i might think or
do something bad
0 1 2 32. I feel that l have to be perfect
0 1 2 33. i feel that no one loves me
0 1 2 34. i feel that others are Out to get me 0 1 2 59' i can be pretty friendly
0 1 2 35. i feel worthless or interior 0 1 2 80. , like ,0 "if new things
0 1 2 36. l accidentally get hurt a lot 0 , 2 61. My school work is m
o ‘ 2 37- ' 9“ "‘ "“"Y "9"" o 1 2 62. I am poorly coordinated or clumsy
0 ‘ 2 38‘ ' 9" "‘3’“ a '°‘ 0 1 2 83. I would rather be with older
0 ' 2 39 I hang a'OUna W11" kids WM 9.1 1" "00°19 “'05 than W11" kids my m ac.
not 1 Please see other Side

125

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OsNotTlue 1=SomewhatorSometlmesThie 2=VeryTrueorOften1hie
0 1 2 64. I would rather be with younger 0 1 2 85. I have thoughts that other people would
kids than with kids my own age think are strange (describe):
0 1 2 85. I refuse to talk
0 1 2 66. 1 repeat certain acts over and over
(descnbe):
0 1. 2 86. I am stubborn
0 1 2 67- I run away Irfor" home ' o 1 2 87. My moods or feelings change suddenly
o ‘ 2 68' ' scream ' '°' 0 1 2 88. I enjoy being with other people
0 1 2 69. I am secretive or keep things to myself 0 1 2 89. ‘ am suspicious
0 1 2 70. I see things that other people think aren't
there (describe): 0 1 2 90. I swear or use dirty language
0 1 2 91. I think about killing myself
0 1 2 92. I like to make others laugh
0 1 2 93. i talk too much
u 1. 2 94. i tease others a lot
0 1 2 71. i am self-conscious or easily embarrassed 0 1 y 2 g. l have a hot temper
° ‘ 2 77* '””"” . o 1 2 96. Ithink aboutsextoomuch
: 1 i :3 : 32w? 3.3:? my hands 0 1 2 97. I threaten to hurt people
0 1 2 75 I am W o 1 2 m. I like to help others
0 g 2 76. [simmtm most kids 0 1 2 99. lamtoo concernedaboutbelng
o 1 2 77. I sleep more than most kids during day "'3‘ °' clean
endlor night (describe): 0 1 2 1m. l have trouble sleeping (describe):
0 1 2 78. l have a good imagination
o , 2 79. , m . Speech problem (describe):__ 0 1 2 101. (out classes or skip school
0 1 2 102. i don’t have much energy
a 1 2 10:3. 1 am unhappy. sad. or depressed
o 1 2 104. I am louder than other kids
0 1 2 105. I use alcohol or drugs tor nonmedlcal
purposes (describe;
0 1 2 so. I stand up for my rights
0 1 2 81. I steal at home
0 1 2 82 I steal from places other than home
0 1 2 83. I store up things I don't need (describe):
0 1 2 1G. I try to be lair to others
0 1 2 107. I sing, a good joke
0 1 2 84. i do things other people think are strange 0 1 2 108' I like to take life easy
(describe): 0 1 2 109. i try to help other people when I can
0 1 2 110. I wish I were oi the opposite seat
0 1 2 111. I keep trom getting involved with others
0 1 2 112 I worry a lot

 

 

 

Please write down anything else that describes your feelings. behavior, or interests

 

M21

126

PLEASE BE SURE YOU HAVE ANSWERED ALL ITEMS

Appendix E

CBRF

127

Child Behavior Rating Form

Date: Reeldent'e Name:

Behavior problem: 1) Check all behavror problems that occurred during this week. 2) Circle 1. 2 or 3 tor the Meg items only.

1 - Mild problem 2 :- Iloderete problem 3 - Severe problem

__ 1. Apathetic or unmotivated ............ 1 2 3 _ 15. Restless. high energy level .......... 1 2
_ 2. Argued ......................... 1 2 3 _ 16. Saw/heard things that others can't . . . . 1 2
_ 3. Deliant. challenged adult authority . . . . 1 2 3 _ 17. Secretive. kept things to sell ......... 1 2
_ 4. Destroyed property ................ 1 2 3 __ 18. Sexually inappropriate .............. 1 2

5. Dilliculty concentrating ............. 1 2 3 __ 19. Short attention span ............... 1 2
_ 6. Disoriented. outoloontact Mh reality. 1 2 3 _20. Stubborn. hadtodothlngsownway... 1 2
_ 7. Easily distracted .................. 1 2 3 _ 21. Sulked. was silent and moody ........ 1 2
__ 8. Fidgeled ......................... 1 2 3 _ 22. Talked back to stall ................ 1 2
_ 9. Harmed sell ...................... 1 2 3 __ 23. Threatened to harm sell ............. 1 2
_ 10. Isolated sell lrom others ............ 1 2 3 _ 24. Too learlul or anxious .............. 1 2
_11.Nervousortense .................. 1 2 3 __25.Unl‘nppyorsad ................... 1 2
___12. Obsessions/convulsions ........... 1 2 3 __26. Wlthdrawn. uninvolved with others . . .. 1 2
_ 13. Overly excited. exrberant ........... 1 2 3 _ 27. Worrying ......................... 1 2
_ 14. Physically attacked others .......... 1 2 3

Ability to function as pert or the group (participation in groups and GA‘s. acceptance oi treatment goals. desire to change):

0000090999990

 

 

 

 

Positive Chengee:

 

 

 

 

Medical Probieme or Side Effects of Medications:

 

 

 

 

128

Appendix F

Informed Assent Agreement

129

Informed Assent Agreement

At Rivendell we are always trying to make the services we provide
better. Our goal is to help all people we work with deal with their emotional
and behavioral difficulties so that they will be able to lead happier lives.
Residents' opinions are important in checking how well we are doing in
meeting this goal. We are interested in whether the residents are doing
better in the program. In order to check this we are doing a study to see if
things get better and what helps or hurts things getting better. Ms. Poorman
will be doing this study as a part of her dissertation.

We will be asking you to answer questionnaires like the ones you
answered when you got to Rivendell. You will do the questionnaires during
school time. If you have trouble reading, someone will be read the
questionnaires to you. The questionnaires should not take more than an
hour to do in total and you will not need to do them all at one time. You
will be asked to fill out information every other week for the next several
weeks. We will ask about your feelings about the program, personal
attitudes about being influenced by others, behavior towards self and others,
moods and emotions, quality of thinking, and school problems. Other
information will be gathered from your medical record such as your age,
who you lived with before coming to Rivendell, the medications you are
taking, your answers to the questionnaire you completed when you got here,
and information from weekly progress notes completed by staff. In
addition, staff will be asked to complete a similar questionnaire about your
behavior.

In order to protect your confidentiality, the information you provide
will be changed into numbers and entered into a computer data file. Your
information in the computer will be identified by a code number. Only the
computer data files will be removed from Rivendell and no information with
your name on it will leave the facility.

Please read this form carefully. We very much hope that you will
agree to participate. Studies like this one help us provide you with the
better services.

By signing this form, you are agreeing to participate in the study.
You understand:

130

1. If you decide not to participate in any or all parts of this study, your
choice will in no way affect the services you will receive or how you will
be treated in the program.

2. You can decide to stop at any time.

3. If you have any questions or concerns about the study or would like to
get information about the results of the study, you can talk with Ms.
Poorman.

4. After you have finished the study, you can receive information about
your progress. You can meet with Ms. Poorman to go over the
questionnaires you completed and discuss areas in which you changed
and did not change.

5. Your caseworker will be contacted to give permission for you to
participate in the study since you are still a minor.

6. If you are upset by any of the questions asked by the study, you can talk
with Ms. Poorman or the staff at Rivendell.

 

 

Resident's Signature Date

131

Appendix G

Informed Consent Agreement

132

Informed Consent Agreement

At Rivendell we are always working to improve the services we
provide. Our goal is to help all people we work with cope with their
emotional and behavioral difficulties so that they will be able to lead more
fulfilling and productive lives. Residents are our most important source of
information in evaluating how well we are meeting this goal. We obviously
are interested in whether the residents are making progress in the program.
In order to assess this we are conducting a study to evaluate residents'
treatment progress and look at variables that affect their progress. The
study is being conducted by Ms. Poorman as a part of her dissertation.

We will be asking residents to complete questionnaires similar to the
ones they completed when they got to Rivendell. They will complete the
questionnaires during school time. If they have difficulty reading, someone
will be available to read the questionnaires to them. The questionnaires
should not take more than an hour to complete in total and they will not
need to do them all at one time. They will be asked to fill out information
every other week for the next several weeks. We will ask them to report on
their feelings about the treatment program, personal attitudes about being
influenced by others, behavior towards self and others, moods and
emotions, quality of thinking, and school problems. Other information will
be gathered from their medical record such as demographic information,
intake assessment information, and information from weekly progress notes
completed by staff. In addition, two full time staff will be asked to
complete a questionnaire about the resident's behavior.

In order to protect their confidentiality, the information they provide
will be transformed into numbers and entered into a computer data file.
Their information in the computer will be identified by a code number.
Only the computer data files will be removed from Rivendell and no
information with their name on it will leave the facility.

The residents have been asked to give assent to participation in the
study. You were only contacted if the resident agreed to participate. The
residents were given an assent agreement similar to this but in language
easier for them to understand. They were informed of the following:

133

1. If they decide not to participate in any or all parts of this study, their
decision will in no way affect the services they will receive or how they
are treated in the program.

2. They can decide to stop participation atany time.

3. If they have any questions or concerns about the study or would like to
receive information about the results of the study, they can contact Ms.
Poorman or ask the staff to contact her.

4. After they have completed the study, they can receive information about
their progress. They can meet with Ms. Poorman to go over the
questionnaires they completed and discuss areas in which they changed
and did not change.

5. If they are upset by any of the questions asked by the study, they can talk
with Ms. Poorman or the staff at Rivendell.

By signing this agreement, you are agreeing to allow the resident to
participate in the study.

 

 

 

Resident's Name Caseworker's Signature Date

134

Appendix H

Self Report of Intemalizing Problems

135

Self Report of Withdrawal

Self Report of Withdrawal

Self Report of Withdrawal

Low Trait

 

N

op-s

bémouahmh

l l 1 l

. .1.
N
4‘

 

-I- Low/mod. State (n=4)J

 

 

 

t'v
or

Time

Moderate Trait

 

 

 

 

 

 

— I - Low/mod State(n=ll)
"at - ~Highstate(n=l)
—+— Unstable Slate (n=3)

 

 

 

 

Time

High Trait

 

2.5

1.5 ~
1.
0.5 ~

-0.5 4
-1 .
-1.5 .
-2 1
-2.5

 

 

 

Time

136

 

. . u - ~Low/mod. State (n=l)
_‘~m@&mmfl)
— -e -- Unstable State (n=3)

 

 

 

Low Trait

=4fl

 

 

{— I — Low/mod. State (It

 

 

 

 

2

EthunE— no team: zom

- 4 u

5251505

1

up

too: 5

u

1

a
5 9..
Jr

5.
9..

Time

Moderate Trait

 

— I - Low/mod State (n=10)
- - I - -I-Iighstate(n=l)
.. e - Unstable State (n=3)

 

 

 

 

 

 

 

52515054545
2 1 0 .w 4 9..
coo—Z :_
E2519:— ue toned .=om

Time

High Trait

 

 

 

I Low/mod State (n=l)
«— I -- High State (n=4)
-- e -- Unstable State (n=3)

 

 

 

 

2 al 0 0.. .
e8: 5
ago—Ema...— ._o «.893- =om

 

Time

137

Appendix I

Staff Report of Intemalizing Problems

138

Low Trait

- I — Low/mod. State (n=5) ]

 

 

 

-____—————.

 

 

 

5....r.+.a.

2 1 o a 4 9..

253?... wag—«E8..—
.o 5...... new

Time

Moderate Trait

 

 

 

 

 

 

 

 

 

a w
m n m
a F e
s. t a...
Mm..
/ w m
m w U
. .
+...
n .
h...
a. 2 m
. m
H
e
m
T
sp51ama+aas
2 1 0 mo 4 9..
2:039..— “£3153:—
... :33. :5

 

G Low/mod State (n=l)
-— I -- High State (n=4)
'- -O -- Unstable State (n=3)

 

 

 

 

..--l
,....

££Li:-_-_.

fl.”

 

 

 

« W A ‘1 d d u 41
525150545
2 1 0 0. 4|

.2 —l

5.
2
. .

253°...— mafia—«Eu...—
..o tone: cam

Time

139

Appendix J

Staff Report of Extemalizing Problems

140

I
l

I - Low/mod. State (n=5) '

 

 

 

Low Trait

 

 

 

cesium
60:93:83.5 =_
.55.:38— no tone: 28m

Time

Moderate Trait

 

- I — Low/mod State(n=ll)
' - I - -High state (n=l)
-~ I -- Unstable State (n=3)

 

 

 

 

 

 

 

.8320:
3:22.333: .5
.coEafinE— he tone: Ezm

Time

High Trait

 

 

 

I Low/mod State (11:1)
—- I —- High State (n=4)
- 4 ~ Unstable State (n=3)

 

 

:Lu‘vv“““—‘
--4~~~~e
“I
2

Time

I.
Q:
1

 

 

 

.3330:
3:22.353: 5
2.25.2.8— we team 9.5

141

Appendix K

Self Report of Total Problems

142

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

143

Time

 

 

 

 

1U ) )
._._. 4. .4 _._. m...
( ()n
m an ma(
m w. ... mom
S 8mm .55
.e m1...
m was ms
1m. mmU mU
_ _ . . .
a a.- lie
_ _ w .
.fl
1 n .u
0' T m
I e
m a 2 m a. a 2 T 4.
w . m . _ .ona I
.m . u ._ m :
. e o .._ e z
. .m M .._ 1 m _
. _.. 2
. _.. 2
52515054545 saswsmsrs25 5251505452
21044.... 2.1.04.4... 2104...
3.039..— mEoEeam 2.5—so...—

_s.... s :25. .em 3.; .5 :25. .3 3.... .o :25. .em

LITERATURE CITED

144

LITERATURE CITED

Achenbach, T. (1991a). Manual for the Child Behavior Checklist/ 4-
18 and 1991 Profile. Burlington, VT: University of Vermont, Department
of Psychiatry.

Achenbach, T. (1991b). Manual for the Youth Self Report and 1991
Profile. Burlington, VT: University of Vermont, Department of Psychiatry.

Achenbach, T., & Edelbrock, C. (1983). Manual for the Child
Behavior Checklist and Revised Child Behavior Profile. Burlington, VT:
University of Vermont, Department of Psychiatry.

Achenbach, T., & Edelbrock, C. (1987). Manual for the Youth Self
Report and Profile. Burlington, VT: University of Vermont, Department of
Psychiatry.

 

Adams, R., & Vetter, H. (1982). Social structure and psychodrama
outcome: A ten-year follow-up. Journal of Offender Counseling. Services,
and Rehabilitation. 6, 111-1 19.

Basta, J ., & Davidson, W. (1988). Treatment of juvenile offenders:
Study outcomes since 1980. Behavioral Sciences and the Law. 6, 355-384.

Beutler, L., Clarkin, J. (1990). Systematic Treatment Selection:
Toward Targeted Therapeutic Interventions. New York: Brunner/Mazel.

Beutler, L., & Crago, M. (1991). Psychotherapy Research: An
International Review of Programmatic Studies. Washington, DC:
American Psychological Association.

Beutler, L., Mohr, D., Grawe, K., Engle, D., & McDonald, R. (1991).
Looking for differential treatment effects: Cross-cultural predictors of
differential psychotherapy efficacy. Journal of Psychotherapy Integration,
_1_, 121-141.

Beutler, L., Sandowicz, M., Fisher, D., & Albanese, A. (1996).
Resistance in Psychotherapy: What conclusions are supported by research.
In Session: Psychotherapy in Practiceg, 77-86.

145

Bird, H., Shaffer, D., Fisher, P., & Gould, M. (1993). The Columbia
Impairment Scale (CIS): Pilot findings on a measure of global impairment
for children and adolescents. International Journal of Methods in
Psychiatric Research, 3, 167-176.

Bow, J. (1988). Treating resistant children. Child and Adolescent
Social Work, 5, 3-15.

Brannon, J ., & Williams, D. (1986). The effectiveness of the Child
Behavior Checklist in identifying the behavioral patterns and security
requirements of juvenile offenders. International Journal of Offender
Therapy and Comparative Criminology, 30, 195-201.

Brehm, J. (1966). A Theory of Psycholggical Reactance. New York:
Academy Press.

Brehm, J. (1972). Responses to loss of freedom: A theory of
psychological reactance. Morristown, NJ: General Learning Press.

Brehm, S., & Brehm, J. (1981). Psychological Reactance: A Theory
of Freedom and Control. New York: Academic Press.

Burger, J. (1990). Personality (3rd ed.). Belmont, CA: Wadsworth
Publishing Company, Inc.

Calvert, S., Beutler, L., & Crago, M. (1988). Psychotherapy outcome
as a function of therapist-patient matching of selected variables. Journal of
Social and Clinical Psychology, 6, 104-117.

Carter, J ., & Kelly, A. (1997). Using traditional and paradoxical
imagery interventions with reactant intramural athletes. Sport Psychologist.
fl, 175-189.

Chamberlain, R., Davis, J ., Forgatch, M., Frey, J ., Patterson, 6., Ray,
J ., Rothchild, A., & Trombley, J. (1986). The Therapy Process Code: A
Multidimensional System for Observing Therapist and Client Interaction. .
Unpublished instrument, Oregon Social Learning Center, Eugene, OR.

Chamberlain, R., Patterson, 6., Reid, J ., Kavanagh, K., and Forgatch,
M. (1984). Observation of client resistance. Behavior Therapy. 15, 144-
155.

146

Courchaine, K., Loucka, P., & Dowd, E. (1995). Effects of
counselor interpretations and client reactance on social influence and the
working alliance. Counseling Psycholggy Ouarteiy, 8, 123-138.

DiGuiseppe, R., Linscott, J ., & Jilton, R. (1996). Developing the
therapeutic alliance in child-adolescent psychotherapy. Applied and
Preventive Psychology, 5, 85-100.

Dowd, E., Hughes, 8., Brockbank, L., & Halpain, D. (1988).
Compliance-based and defiance-based intervention strategies and
psychological reactance in the treatment of free and unfree behavior.
Journal of Counseling Psychology, 35, 370-376.

Dowd, E., Milne, C., & Wise, S. (1991). The Therapeutic Reactance
Scale: A measure of psychological reactance. Journal of Counseligg and
Development, 69, 541-545.

Dowd, E. & Sanders, D. (1994). Resistance, reactance, and the
difficult client. Special Issue: Perspectives on working with difficult
clients. Canadian Journal of Counseling, 28, 13-24.

Dowd, E. & Seibel, C. (1990). A cognitive theory of resistance and
reactance: Implications for treatment. Journal of Mental Health
Counselinggl 2, 458-469.

Dowd, E., Trutt, S., & Watkins, C. (1992). Interpretation style and
reactance in counselor's social influence. Psychological Reports. 70, 247-
254.

Dowd, E. & Wallbrown, F. (1993). Motivational components of
client reactance. Journal of Counselitpggnd Developmem, 71, 533-538.

Dowd, E., Wallbrown, F ., Sanders, D., & Yesenosky, J. (1994).
Psychological reactance and its relationship to normal personality variables.
@gnitive Therapy and Resgrch, 18, 601-612.

Edelbrock, C. (1985). Child Behavior Rating Form.
Psychopharmacolggy Bulletin, 21, 835-837.

147

Frank, S., Jackson-Walker, 8., Marks, M., Van Egeren, L., Loop, K.,
& Olson, K. (1998). From the laboratory to the hospital, adults to
adolescents, and disorders to personality: The case of psychological
reactance. Journal of Clinical Psychology, 54, 361-381.

Frank, S., & Paul, J. (1995a). Functional Impairment Scale for
Children and Adolescents (F ISCA). Department of Psychology, Michigan
State University.

Frank, S., & Paul, J. (1995b). Functional Impairment Scale for
Children and Adolescents-Youth Self Report Form (FISCA-YSR).
Department of Psychology, Michigan State University.

Frank, 8., Paul, J ., Marks, & Van Egeren, L. (Under review).
Validation of the Functional Impairment Scale for Children and Adolescents
(FISCA). Manuscript submitted to firmal of the American Academy of
Child and Adolescent Psychiatry. .

Frank, S., Van Egeren, L., & Poorman, M. (1993). Therapeutic
Reactance Scale-Adolescent Version (TRS-A). Unpublished measure.
Department of Psychology, Michigan State University.

Gold, M., Mattlin, J ., & Osgood, D. (1989). Background
characteristics and responses to treatment of two types of institutionalized
delinquent boys. Criminal Justice and Behavior, 16, 5-33.

Graybar, S., Antonuccio, D., Boutilier, L, & Varble, D. (1989).
Psychological reactance as a factor affecting patient compliance to
physician advise. Scandinavian Journal of Behavior Therapy, 18, 43-51.

Hellman, C., & McMillin, W. (1997). The relationship between
psychological reactance and self-esteem. Journal of Social Psychology,
Q7, 135-138.

Hodges, K. (1994). Child and Adolescent Functional Assessment
Scale (CAFAS). Department of Psychology, Eastern Michigan University.

Hodges, K., & Gust, J. (1995). Measures of impairment for children
and adolescents. Journal of Mental Health AdministrationLZZ, 403-413.

Hollin, C. (1993). Advances in the psychological treatment of
delinquent behavior. Criminal Behaviour and Mental Health, 3, 142-157.

148

Hong, S. (1990). Effects of sex and church attendance on
psychological reactance. Psychological Reports. 66, 494.

Hong, S., & F aedda, S. (1996). Refinement of the Hong
Psychological Reactance Scale. Educational and Psycholgcal
Measurement, 56, 173-182.

Hong, S. & Giannakopoulos, E. (1994). The relationship of
satisfaction with life to personality characteristics. Journal of Psychology,
l_2_8_, 547-558.

Hong, S., Giannakopoulos, E., Laing, D., & Williams, N. (1994).
Psychological reactance: Effects of age and gender. Journal of Social
Psychology, 134, 223-228.

Hong, S., & Page, S. (1989). A psychological reactance scale:
Deve10pment, factor structure and reliability. Psychological Reports, 64,
1323-1326.

Horvath, A. & Goheen, M. (1990). Factors mediating the success of
defiance- and compliance- based interventions. Journal of Counseling
Psychology, 37, 363-371. ‘

Hunsley, J. (1993). Treatment acceptability of symptom prescription
techniques. Journal of Counseling Psychology. 40, 139-143.

Joubert, C. (1990). Relationship among self-esteem, psychological
reactance, and other personality variables. Psychological Reports, 66, 1147-
1151.

Kavanagh, K., Gabrielson, P., & Chamberlain, P. (1982). Manual for
Codinngient Resistance. Unpublished instrument, Oregon Social
Learning Center, Eugene, OR.

Kazdin, A. (1987). Treatment of antisocial behavior in children:
Current status and future directions. Psychological Bulletin, 102, 187-203.

Kelly, A., & Nauta, M. (1997). Reactance and thought suppression.
Personality and Sociial Psychology Bulletin, 23, 1123-1132.

149

Kirmayer, L. (1990). Resistance, reactance, and reluctance to
change: A cognitive attributional approach to strategic interventions.
Journal of Cognitive Psychotherapy, 4, 83-104.

Mahalik, J. (1994). Development of the Client Resistance Scale.
Journal of Counseling ngchology, 41, 58-68.

Martin, F., & Osgood, D. (1987). Autonomy as a source of pro-
social influence among incarcerated adolescents. Journal of Applied Social
Psychology, 17, 97-107.

Merz, J. (1983). Fragebogan zur messung der psychologischen
reactanz. Diagnostica, 29, 75-82.

Mulry, G., Fleming, R., & Gottschalk, A. (1994). Psychological
reactance and brief treatment of academic procrastination. Journal of
College Student Psychotherapy. 9, 41-56.

Nir, Y., & Cutler, R. (1973). The therapeutic utilization of the
juvenile court. American Journal of Psychiatry, 130, 1112-1117.

Oliver, R., Mattson, D., & Moore, J. (1993). Psychological reactance
in marital enrichment training. TCA Journal. 22, 3-10.

Osgood, D., Gruber, E., Archer, M., & Newcomb, T. (1985).
Autonomy for inmates counterculture or cooperation? Criminal Justice and
Behavior, 12, 71-89.

Paul, J. (1997). Validation of the Functional Impairment Scale for
Childretflnd Adolescents (FISCAL Unpublished master’s thesis, Michigan
State University, East Lansing.

Phares, V., & Copas, B. (1990). Adolescents’ subjective distress over
their emotional/behavioral problems. Journal of Consulting and Clinical
Psychology. 58, 596-603.

Prandoni, J. & Wall, S. (1990). Effects of mental health and
reactance-reducing information on offenders' compliance with court-
mandated psychiatric evaluations. Professional Psychology: Research and
Practice, 21, 204-209.

150

Shaffer, D., Gould, M., Brasic, Ambrosini, Fisher, P., Bird, H., &
Aluwahlia. (1983). A Children's Global Assessment Scale (C-GAS).
Archives of General Psychiatry, 40, 1228-1231.

Shoham-Salomon, V., Avner, R., & Neeman, R. (1989). You're
changed if you do and changed if you don't: Mechanisms underlying
paradoxical interventions. Journal of Consulting and Clinical Psychology,
51, 590-598.

Stranger, C., & Lewis, M. (1993). Agreement among parents,
teachers, and children on internalizing and externalizing behavior problems.
Journal of Clinical Child Psychology, 22, 107-115.

Stoolmiller, M., Duncan, T., Bank, L., & Patterson, G. (1993). Some
problems and solutions in the study of change: Significant patterns in client
resistance. Journal of Consulting and Clinical Psychology, 61, 920-928.

Swenson, C., & Kennedy, W. (1995). Perceived control and
treatment outcome with chronic adolescent offenders. Adolescence, 30,
565-578.

Swoboda, J ., Dowd, E., & Wise, S. (1990). Reframing and
restraining directives in the treatment of clinical depression. Journal of
Counseling Psychology, 37, 254-260.

Taylor, L., Adelman, H., & Kaser-Boyd, N. (1985). Exploring
minor's reluctance and dissatisfaction with psychotherapy. Professional
Psychology: Research and Practice, 16, 418-425.

Tennen, H., Press, 8., Rohrbaugh, M., & White, L. (1981).
Reactance theory and therapeutic paradox: A compliance-defiance model.
Psychotherapy: Theory, Research. and Practice, 18, 14-22.

Velasquez, J ., & Lyle, C. (1985). Day versus residential treatment
for juvenile offenders: The impact of program evaluation. Child Welfare,
6g, 145-156.

Van Egeren, L. (1996). A Model of Inpatient Child and Adolescent
Behaviors: Validation of the Child Behavior Ratirggjom-Revised.
Unpublished master's thesis, Michigan State University, Lansing.

151

"I11111111111111111111115