w

4 Wm?“

 

in; .
. .2}
.194“:

. 7v .
i. 384 .
burr»?
.. x4 .

mm
.M

NH"?
lit.

-33.» . . ,

Anna? In

. isleifi‘.
. .2:

 

 

33.7 1

, i .w.

p.531: .I .5; ,xrwlriii:

ME

:004
sydfixéa?

 

- innnuv

L:~‘og;~§fihy “A t .
starlight-n 3&3 9

University l

W“

 

 

 

This is to certify that the
dissertation entitled

OBSERVABLE CLIENT RESISTANCE AND EARLY
PSYCHOTHERAPY DROPOUT

presented by

JONATHAN N. WELLER

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Psychology_

 

[U mg LSAT/€57

Major Professor’s Signature
irl «1/0;

Date

MSU Is an Affinnatlve Action/Equal Opportunity lnsfiMlon

 

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c/CIRC/DateDuepGS-pJS

 

OBSERVABLE CLIENT RESISTANCE AND
EARLY PSYCHOTHERAPY DROPOUT
By

Jonathan N. Weller

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
‘ for the degree of
DOCTOR OF PHILOSOPHY
Department of Psychology

2003

 

lllls study sou:

 

1993 meta-analysis in

psythothcrapy dropoul
scssions. Psychotherur
variety of problems f0
form of resistance. WI
dcspilc having sought
WM by theoretic

in "fitment and drOpt

PaniCipants ol

ABSTRACT
OBSERVABLE CLIENT RESISTANCE AND
EARLY PSYCHOTHERAPY DROPOUT
By
Jonathan N. Weller

This study sought to establish a link between patient resistance and dropout. A
1993 meta-analysis investigating psychotherapy dropout reported an overall
psychotherapy dropout rate of 47% with most attrition occurring within the first couple of
sessions. Psychotherapy dropouts represent a significant public health concern and pose a
variety of problems for psychotherapists. Patient dropout can be considered an extreme
form of resistance. When a patient is unwilling to engage in the therapeutic process
despite having sought treatment, it may be the result of intense resistance. While this is
supported by theoretical writings, the relationship between resistance experienced early
in treatment and dropout has not been established empirically.

Participants of this study were forty clients (twenty dropouts and twenty persisters
matched demographically) of the MSU Psychological Clinic. A ten-minute sample from
the first and third sessions of participants’ treatment were coded by two independent
raters using the Client Resistance Scale (CR8) and the Client Resisance Code (CRC) to
evaluate the level of resistance early in treatment. It was hypothesized that dropouts
would exhibit more resistance early in treatment, resistance would increase between
sessions with dropouts outpacing persisters, and a discriminant function capable of

predicting early dropout could be developed using resistance data.

As predicted. d
{48.2% vs. 34.4%) as .

significant differences.

 

The anticipated
resistance. though stall.
. |

submit, Opposmg Re

0fSmith between scss

 

 

Discriminant fu
”mbmhip from the s
therapist, third sag
CR3 Opposing Recoil.-
associated With group

overall classification a

As predicted, dropouts tended to have a higher proportion of resistant responses
(48.2% vs. 34.4%) as measured by the CRC. CRS data failed to find statistically
significant differences, though dropouts expressed greater intensity of resistance.

The anticipated increasing levels of resistance were exhibited on both measures of
resistance, though statistical analyses failed to reach levels of significance. One CRS
subscale, Opposing Recollection of Material, produced a statistically significant pattern
of growth between sessions F (l, 38) = 4.25, p<.05.

Discriminant function analysis produced a model that predicted group
membership from the scores of four variables (CRC resistance proportion, CRS Opposing
the Therapist, third session CRS Opposing Expression of Painful Affect, and third session
CRS Opposing Recollection of Material). This discriminant function was significantly
associated with group membership (X2(4) = 11.37, p < .05). The discriminant function an

overall classification accuracy of 77.5%.

To Raini and Lincoln

iv

ACKNOWLEDGMENTS

I would like to express my gratitude to those that have assisted and supported me
throughout this research project. Serving as the chair of my dissertation committee, Dr.
Norman Abeles was an excellent source of feedback, support and encouragement. His
help and timely feedback greatly facilitated the completion of this dissertation. I would
also like to thank Dr. Gershen Kaufman, Dr. Bertram Karon, and Dr. M. Janice Gutfreund
for serving on my committee and for their prior influences on my education and
development as a clinical psychologist.

I would also like to extend my gratitude to Ken Murray and Patrick Perkins for
their assistance in coding the numerous hours of psychotherapy samples that made this
study possible. I would further like to acknowledge the support, encouragement and
informal assistance of many friends and colleagues, including David Cordry, Quinn
Bastian, and Ted Moore amongst many others. Finally, I must also recognize the

consistent support and inspiration provided by my wife, children, and family members.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................... viii
LIST OF FIGURES ........................................................................... ix
INTRODUCTION ............................................................................. 1
LITERATURE REVIEW
Resistance: Theoretical Background ................................................. 4
Empirical Research on Resistance ................................................... 7
Psychotherapy Dropout ................................................................ 10
RATIONALE .................................................................................... l7
HYPOTHESES .................................................................................. 19
METHOD
Data ...................................................................................... 21
Participants .............................................................................. 21
Therapists ....................................................................... 23
Therapy Dyads .................................................................. 24
Tape Recorded Therapy Samples ..................................................... 24
Measures ................................................................................. 25
Raters ..................................................................................... 28
Procedures ................................................................................ 29
RESULTS
Demographic comparisons between groups ......................................... 32
Interrater Reliabilities .................................................................. 32
Overview of Resistance Variables .................................................... 33
Hypothesis I .............................................................................. 35
Hypothesis II ............................................................................. 36
Hypothesis III ............................................................................ 38
Hypothesis IV ............................................................................ 39
DISCUSSION
Hypothesis I ......................................... 43
Hypothesis II ............................................................................. 47
Hypothesis III ............................................................................ 53
Hypothesis IV ........................................................................... 58
Practical Implications of the Findings ................................................ 61
Limitations of the Present Study ...................................................... 64

vi

DISCUSSION (continued)

Directions for Future Research ......................................................... 68
TABLES ........................................................................................... 71
FIGURES .......................................................................................... 83
APPENDIX A- Client Resistance Scale ....................................................... 9O
APPENDD( B- Client Resistance Code Categories .......................................... 93
APPENDIX C- Symptom Checklist 90—Revised ............................................ 94
LIST OF REFERENCES ......................................................................... 96

vii

LIST OF TABLES

 

Table 1. Sample Demographic Descriptive Statistics and ANOVA’s .................. 72
Table 2. Sample Demographic Descriptive Statistics: Crosstabulations ............... 73
and Chi-square estimates.

Table 3. Sample Means, Standard Deviations, and Interrater Reliabilities:

Client Resistance Scale (CRS) .............................................................. 74
Table 4. Client Resistance Scale: Subscale Intercorrelation Matrix .................... 75

Table 5. Sample Means, Standard Deviations, and Interrater Reliabilities

Client Resistance Code (CRC) .............................................................. 76
Table 6. Concurrent Validity Correlations: Client Resistance Scale (CR8)

and Client Resistance Code (CRC) .......................................................... 77
Table 7. Group Means, Standard Deviations, and t-test for Client Resistance

Code Resistant Response Proportions ....................................................... 78
Table 8. Group Means, Standard Deviations, and t-test for Client Resistance

Scale Summed Scores. ........................................................................ 79
Table 9. Canonical Discriminant Function Coefficients ................................. 80
Table 10. Discriminant Classification Results. .......................................... 81

 

Table 11. Pooled Within-Groups Correlations Between Variables
and Canonical Values .......................................................................... 82

viii

LIST OF FIGURES

Figu_re 1. CR8 Subscale Means: Dropout vs. Persister Group (Session 1) .................. 84
Frgm 2. CR8 Subscale Means: Dropout vs. Persister Group (Session 3) .................. 85

Figure 3. Change in CRS Total Means Between Sessions 1 and 3:
Dropout vs. Persister Group ....................................................................... 86

FrguLe 4. Change in CRC Resistant Response Proportion Means Between
Sessions 1 and 3: Dropout vs. Persister Group .................................................. 87

Figure 5. Change in CRS Opposing Recollection of Material Means Between
Sessions 1 and 3: Dropout vs. Persister Group ................................................. 88

ix

INTRODUCTION

Historically psychotherapy research has been dominated by studies of process and
outcome with clients who complete treatment or who remain in therapy long enough to
provide sufficient data. Somewhat less inquiry and investigation has been spent
examining premature termination of psychotherapy (or psychotherapy dropout), despite
this being a widespread and fundamental obstacle to the successful and effective delivery
of mental health services. A recent meta-analysis of 125 studies on psychotherapy
dropout by Wierzbicki and Pekarik (1993) reported an overall mean dropout rate of
approximately 47%. Most psychotherapy studies of process and outcome, typically
report dropout rates between 30% and 60% with a majority of the attrition occurring
within the first couple of sessions. This is particularly troubling considering only 25% of
the people in the US. with a diagnosable mental disorder seek treatment (Regier,
Narrow, Rae, Manderscheid. Locke, & Goodwin, 1993). Not only do psychotherapy
dropouts represent a significant public health concern, but they also pose clinical,
financial, and morale problems for mental health professionals. Without a better
understanding of the underlying dynamics of such enormous attrition rates, clinical
psychologists have little power to make informed decisions with regard to how to
respond to or prevent them.

Until very recently, most studies on psychotherapy dropout have focused on the
easily obtainable demographic characteristics of clients and therapists. However, despite
the ease of securing such data, the research findings have been equivocal and sparsely

informative, lending little insight into how to decrease the frequency of this problem.

 

Irrcreascd risk for drU

of education, and low

 

 

findings have been re
recommend the inves‘
dropout. Despite this i
client and therapist in

suggested for further 1'
interpersonal relations

(e.g., level of warmth.

 

“tannent factors (e. g,.

While only a f.
Wnflity variables .
1995; Walters, Solom
more have been condi
the”reuse alliance d,

found Significant I'Bla

Increased risk for dropout has been found to be associated with minority race, low level
of education, and low Socio-Economic Status (SES). However, effects sizes for such
findings have been relatively small. This prompted Wierzbicki and Pekarik (1993) to
recommend the investigation of more complex psychological variables in relation to
dropout. Despite this suggestion, a minimal number of studies investigating potential
client and therapist factors related to dropout have ensued. Among the variables
suggested for further investigation are client factors (e.g., personality traits, health of
interpersonal relationships, expectations, and level of psychopathology), therapist factors
(e.g., level of warmth, empathic failures, level of training, conceptualization accuracy),
treatment factors (e. g., theoretical model, directivity), and interactions between these.

While only a few studies have looked directly at the relationship between client
personality variables and premature termination (Hilsenroth, Hanler, Toman, & Padawer,
1995; Walters, Solomon, & Walden, 1982; Budman, Demby, Soldz, & Merry, 1996),
more have been conducted that investigated the impact of personality factors on
therapeutic alliance development (Mallinckrodt, 1993). Recently, investigators have also i
found significant relationships between the pattern of a patient’s pre-therapy
interpersonal relationships and the therapeutic alliance established in treatment (Piper,
Azim, Joyce, McCallum, Nixon, & Sega], 1991; Luborsky, McLellan, Woody, O’Brien,
& Auerbach, 1985). This is important and may be especially helpful in the pursuit of
variables related to therapy dropout, as a negative therapeutic alliance {is likely associated
with attrition (Mallinckrodt, 1993).

The lack of a positive therapeutic alliance could be associated with a variety of

variables ranging from lack of clinician skill, poor client engagement, pathological

personality traits to client resistance. One interesting and theoretically-based way to
understand patient dropout would be to consider it an extreme form of resistance
(Beutler, Rocco, Moleiro, & Talebi, 2001). The fact that a patient is no longer willing to
engage in the therapeutic process, despite their having just begun treatment, may be an
indication of intense resistance. And yet, the relationship between level of resistance
experienced early in treatment and patient attrition has yet to be firmly established. This

study sets out to establish a link between patient resistance and dropout.

LITERATURE REVIEW
Resistance: Theoretical Backgr_qund

Freud (1920) emphasized the importance of resistance. For example in his
“Introductory Lectures on Psychoanalysis,” he stated, “ when we undertake to restore a
patient to health, to relieve him of the symptoms of his illness, he meets us with a violent
and tenacious resistance, which persists throughout the whole length of treatment.
(p.354)” Freud’s contemporaries have since expanded and elaborated the construct of
resistance. According to Greenson (1967), resistance refers to all forces within the client
which are in opposition to the process and procedures of psychoanalysis. This view has
also been extended to include many other forms of psychotherapy (e.g., cognitive-
behavioral, gestalt, and rational emotive therapy) with an emphasis being placed on the
client’s defense of the “status quo” of their condition. Patients may employ any variety of
psychological defenses to resist treatment. These resistances can be both consciously
and/or unconsciously determined and driven.

Interestingly, despite its being a central issue in arguably all forms of
psychotherapy, client resistance has received little empirical attention concerning its
relation to therapy process, outcome, or attrition. Many theorists have indicated that
certain transference reactions can lead to a specific form of resistance referred to in the
literature as “flight into health,” which causes the client to exit therapy prematurely with
relatively little therapeutic gain. (Greenson, 1967) Here the client is denying the reath of
their internal state and seeking to avoid the pain of approaching their affects, experiences,

and treatment issues. It is not uncommon for flight into health to occur early in

treatment, while ambivalence toward therapy, the therapist and change are typically
intense.

In Strean’s (1985) text on resistance, he proposed that it has been theoretically
understood for decades that clients enter therapy with a great deal of ambivalence.
Despite their unhappiness, life dissatisfaction, or painful internal state the idea of seeking
therapy as a solution is still often frightening and aversive. However, while much
theoretical attention has been given to this phenomenon it has been infrequently
examined empirically, due in great part to difficulties in operationalizing and measuring
ambivalence and resistance (Strean, 1985).

Resistances are founded on many fears, negative fantasies, and cultural
influences. These include (particularly in the US. and other Western cultures) the
cultural value placed on independence and self-sufficiency; fears of being infantilized,
placed in a submissive position to the therapist, or treated like a child; and negative
fantasies of being exploited, seduced, or rejected by the therapist to name only a few.
(Strean, 1985)

As resistance is naturally activated by thinking of or beginning therapy, it should
not be surprising, to those that understand and recognize it, that clients are quick to delay,
postpone, or even drop out of treatment. Perhaps then a certain portion of clients will be
destined to leave treatment prematurely. However, armed with an understanding of this
dynamic and how to effectively respond could potentially aid clinicians from all
theoretical orientations in helping clients or prospective clients remain in treatment long
enough for it to benefit them. (Strean, 1985). Patton et al. (1997) highlighted the point

that resistance is both an obstacle and an opportunity in therapy. They emphasize that if it

is not recognized and highlighted by the therapist and then understood and worked
through by the client, less therapeutic change is likely to occur. Perhaps in the worst case
scenario, clients’ resistance and a lack of understanding of its function by the therapist
can lead to the client leaving treatment prematurely. Berri gan and Garfield’s (1981)
study of the relationship between missed therapy appointments and premature
termination bears this out. They suggest that clients missing therapy sessions may be
manifesting their initial disillusionment with the process of treatment (it can also easily
be interpreted as resistance), but noted that little attention was given to the missed
sessions by the therapists treating the clients who eventually dropped out. The authors go
on to suggest that these therapists ought to have at very least explored the absences with
their clients as a first step toward increasing the probability of their continuing in
treatment.

Although there are many forms of resistance that may be inaccessible to an
observer, the measures that will be employed in the current study have been shown to
reliably measure episodic (transient/limited interruptions) and tactical (immediate,
objectively observable behaviors) resistances (Kavanagh, et al., 1982; Schuller, et al.,
1991; Mahalik, 1994). Although episodic and tactical resistance do not reflect the deepest
levels of client resistance, it has been suggested that successful handling of resistance at
this level is important, as it helps clients-feel safe and understood and provides a model

for dealing with more difficult issues (Schuller et al., 1991).

Empirical Research on Resistance

In Hill, Corbett, Kanitz, Rios, Lightsey, and Gomez’ (1992) development of a
pantheoretical measure of client behavior they found that resistance or defensive verbal
responses, as measured by the Client Resistance Code (Kavanagh, et al., 1982),
accounted for 11% of their sample’s behavior during the middle stages of psychotherapy.
However, they noted difficulty in obtaining high levels of reliability while coding
resistance. This was likely due to the most common and difficult challenge presented to
researchers of resistance, that of defining and operationalizing the construct.
Additionally, the prior research done in this area has frequently failed to recognize the
multidimensional nature of resistance and rarely formulated the measurement of the
construct based on a comprehensive theory. Further evidence suggests that measuring
client resistance on a moment-to-moment or speaking turn basis lends itself to greater
reliability and less “observer bias” than global (e.g., entire session) ratings
(Chamberlain, Patterson, Reid, et al., 1984).

Chamberlain, et al. (1984) studied resistance exhibited by participants in a
behaviorally oriented, parent training intervention. Using an observational system they
developed to measure client resistance, the Client Resistance Code (CRC), they found
statistically significant relationships between the levels of resistance exhibited and the
stage of therapy, the completion of therapy, and whether clients were self or agency
referred. Their findings were supportive of the proposition that resistance first increases,

peaks, and then diminishes over the course of treatment, for clients who complete

treatment. They also reported clients who dropped out of treatment exhibited a
significantly higher frequency of resistant responses than those who completed the
treatment did. In relation to this finding, they also reported that agency-referred clients
were more resistant and more likely to dropout of treatment than self-referred clients.

Schuller, Crits—Christoph, and Connolly (1991) in developing their Resistance
Scale, found four principle dimensions or subtypes of resistance: Abrupt/Shifting,
Flat/Halting, Oppositional, and Vague/Doubting. They also found that a majority of the
variance in resistance was related to within session fluctuations and client differences.
Surprisingly, they found little support for the influence of therapist intervention
differences. These findings were suggestive of the Resistance Scale tapping stable
characteristics of the client, what might be referred to as character resistances, as well as
in session resistance fluctuations.

Of particular interest for the purposes of the current study, was the development
of the Client Resistance Scale (CRS) by Mahalik (1994). He took Greenson’s definition
and operationalization of resistance and expanded it to make it inclusive of nonanalytic
approaches and the motivational aspect of the concept (i .e., the avoidance of painful
affect), which fits well with the aims of this study. The formulation of resistance used to
develop the CRS and which will be applied in the current study was:

“With its motivation being the avoidance of painful affect. resistance opposes the client’s
recollection or the therapeutic material, as well as, the therapist’s efforts, change, and insight”

From this basic formulation, Mahalik elaborated on the multi-dimensional nature of
resistance using Greenson’s (1967) work as a starting point. These five dimensions,

which will be of particular interest in the present study are described below:

Opposing Expression of Painful Affect: This form of resistance has as its primary aim
the avoidance of experiencing or expressing any unpleasant emotion, most commonly
depression, anxiety, and aggression. It is most typically exhibited when a client
communicates their experience verbally, but with an absence of affect, in spite of the fact
that what they are sharing would be expected to be highly emotionally charged.

Opposing Recollection of Material: As the client’s primary task of psychotherapy is
openly discussing their experiences and feelings as freely as possible, all activity and
motivation to do otherwise could be broadly labeled resistance. It has been said that
resistance acts against speaking and results in the failure to share relevant information. As
such, the most obvious and transparent expressions of this kind of resistance include
when the client states they don’t feel like talking, they remain silent, or says, “I don’t
know.” In other cases, the client may talk about superficial or insignificant events, use
nondescript language, or summarize events in broad outlines lacking details all in the
service of hiding from the pain that a more in depth or descriptive account may
precipitate.

Opposing the Therapist: This dimension of resistance involves the degree of client
compliance with therapist instruction, direction, or ground rules. The uncooperativeness
exhibited can take the form of contrary responses to the therapist’s request or
intervention, non-acceptance of the therapist’s influence, or a negative interpersonal
dynamic. This form of resistance is thought to originate out of self-protection on the part
of the client, where they may experience the therapist as an attacker or intruder. Their
uncooperative behavior is thus a means of fending off further intrusion. Resistance of
this kind is often, but not always, a product of therapist errors (e.g., poorly timed
interventions, moving in a direction that does not meet the current needs of the client,
etc.).

Opposing Change: Here a client’s opposition to change disrupts the therapeutic process
and can be viewed as their wish to maintain “status quo,” despite the recognition that it is

maladaptive, limiting, and painful. Fear, the search for security, and secondary gain keep
the client stuck in their familiar, maladaptive patterns.

Opposing Insight: This dimension of resistance acts to oppose client self-knowledge,
particularly the ability to make connections between their experiences, thoughts, feelings,
and behaviors. This allows the client to “not know,” which helps them avoid painful
feelings that may accompany insight.

Mahalik (1994) effectively applied his scales to the analysis of the process in six classic
and well-documented sessions of therapy between expert therapists (Beck , Ellis,
Meichenbaum, Perls, Rogers, and Strupp) and two clients (Gloria and Richard). The

scale was found to be sensitive to differing therapeutic approaches (therapist responses),

clients, therapists, and within session shifts. The utility of a measure able capable of
capturing different subtypes of client resistant behavior will hopefully be of value in the
current study. These dimensions of resistance are readily applied to factors previously
found to be predictive of dropout, as a review of past attrition research will bear out.
Frnally, in their study of process dimensions within the context of psychodynarnic
counseling, Patton, Kivlighan, and Multon (1997) found that during the course of
counseling there was a linearly decreasing growth pattern of client resistance. They also
found that the mid-treatment (at on average the eighth session) level of client resistance
was related to client outcome, with less resistant clients showing greater levels of change
following treatment. Finally, they were able to lend further evidence of the low-high-low

curve of resistance being related to positive outcomes.

Psychotherapy Dropput

As stated previously, psychotherapy dropout has most commonly been studied
from the standpoint of easily obtainable demographic information. However, following
the admonitions of researchers interested in preventing this phenomenon, there have been
an increasing number of studies conducted that have taken more complex variables into
account (e.g., a process or client-therapist interaction approach). Less common are
studies that take into account variables (i.e., resistance, transference, attachment) that can
be derived from or grounded in well-established psychological theory. Occasionally, the
demographically oriented studies have included resistance relevant tidbits of information
(e.g., client-therapist differences), but this data has not been elaborated on or interpreted
from such a theoretical perspective. Additionally, much research indicates that the early

phase of therapy is crucial with regards to client continuation, with dropout rates

10

lessening significantly when clients remain beyond four sessions (Reis & Brown, 1999).
Thus, the following review of dropout studies attempts to summarize a broad base of
studies in an effort to provide a large pool of evidence from which theoretically and
empirically grounded hypotheses can be drawn linking early psychotherapy dropout to
client resistance.

In Hilsenroth, et al.’s (1995) study of Rorschach and MMPI—Z correlates of early
psychotherapy termination, they found that patients who terminated prematurely (before
the eighth session) from dynamically oriented psychotherapy, had fewer texture
responses, more cooperative movement responses, and fewer aggressive movement
responses, compared with patients staying in treatment. This suggests that individuals
who dropout of treatment prematurely are somewhat less psychologically disturbed, less
aggressive, more capable of establishing cooperative relationships, and have less need for
close interpersonal contact, than those patients remaining in treatment. In viewing these
results from the perspective of resistance, it follows that clients who fit this particular
profile may find the process of psychotherapy particularly intrusive and thus, “oppose the
therapist” in an effort to ward off the intrusion. Further, such clients may be fairly
satisfied with the status quo of their situation, since their life is not marked by extreme
pathology, overt conflict, or relationship difficulties. Thus, by dropping out of treatment
they “oppose change,” leaving everything as it was before they entered treatment.

Tian and Kazmierczak (1997) found that cognitive-behavioral ”group therapy
clients’ level of participation within therapy sessions was significantly related to early
dropout, while demographic and pathology factors were not. Dropouts in this study

tended to be withdrawn and less able to freely express themselves within the group. This

11

could be indicative of any number of the Greenson (1967) categories of resistance, most
obviously “opposing recollection of material” and “opposing insight.”

Beckham (1992) in looking for predictors of dropout at a medical school
outpatient mental health clinic, found that a negative impression of therapist by the
patient was the singular variable that reliably predicted attrition. Beckham further
suggested that dropouts may have a sensitivity to whether or not their therapist’s
personality and style will meet their needs. This could be another indication of how
resistance can take the forrrr of opposing the therapist. Missed and cancelled
appointments early in treatment were also negatively prognostic for staying in treatment.
In some cases particularly in those of the “no show,” these incidents may represent the
most extreme form of avoidance of painful affect. In essence, the client by not attending
their session, completely avoids the possibility of confronting painful emotions and
experiences.

In an intriguing descriptive study of thirty independent practice patients who
dropped out prematurely, Levinson, McMurray, Podell, and Weiner (1978) found that
reactive factors were common (e.g., fear of loss of defenses, fear of dependence, fear of
aggression, transference) as well as intrinsic factors (e.g., self-defeating behavior,
negativism, suspiciousness). Reactive factors were found with 87% of the patients that
dropped out, while intrinsic factors were influential 57% of the time. These results seem
to be consistent with Stone’s (1973) two dimensions of resistance, with the reactive
factors being similar to the tactical resistance (situational) dimension and intrinsic factors
being similar to what Stone and others have labeled characterological resistance. Tactical

resistances are those that are readily observed in outward behaviors. They serve

12

immediate “tactical” functions such as avoiding painful affects, shifting the focus of
discussion to something less threatening, and keeping the therapist at a distance.
Characterological resistances on the other hand are much less overt and/or situationally
bound. These are long term patterns of behavior that again serve a defensive/protective
function for the client, but that are much more generalized across situations and time. In
other words, a much more static and global approach to the world.

Cartwright, Lloyd, and Wicklund (1980) attempted to develop a rapid method for
identifying patients not likely to stay in psychotherapy. Their goal was to develop a
screen that was easy to administer and score and that could then be used to direct a subset
of clients into a program designed to increase psychological mindedness and motivation
prior to treatment. They found that they could predict with moderate reliability those
clients likely to drop out using a cutoff score on the Counseling Readiness Scale in
combination with intaker ratings on two scales, “difficulty of remaining in treatment” and
“patient’s ability to generate inner life data in interview.” The latter item is of particular
relevance to the present studies aims in that low ratings on the patients’ ability to
generate “clinical material” from their inner experience was predictive of dropout. This
item seems to coincide with another of Mahalik’s (1994) dimensions of resistance,
specifically the opposition to the recollection of material. Again, part of what is being
reported here may be an observable expression of resistance.

Dubrin and Zastowny (1988) found similar results to Levinson, et al. (1978),
noting a connection between attrition and client character structure and personality

characteristics. These authors strongly suggested that researchers turn their attention

13

toward the specifics of the therapy encounter, so called “rrricroanalytic process”
variables, which could better capture client-therapist interactions and relatedness.

In a study of ninety-one adult therapy clients at a university training hospital,
Kolb, Beutler, Davis, Crago, and Shanfield (1985) found that premature terminators rated
their therapists as having fewer facilitative relationship skills. They also found that the
therapists of the dropouts rated their clients as having been less involved in the
therapeutic process. Finally, the investigators also reported that those clients, who
remained in treatment made significantly greater positive changes in comparison to
dropouts, based on therapist ratings. Ironically, the dropout clients themselves rated their
improvement as being relatively high, despite their negative perceptions of the therapist.
In viewing these result from the perspective of resistance theory, it seems feasible that
these clients may have been engaging in classic defensive behavior with the aim of
resisting change and the perceived intrusion of the therapist. Thus, they exit therapy and
endorse having somehow benefited in spite of dissatisfaction with the therapist and a lack
of engagement in the therapeutic process, a somewhat suspicious outcome that seems to
fit well with the concept of “flight into health.”

In their study of therapeutic alliance and interpersonal behavior variables,
Samstag, Batchelder, Muran, Sufran, and Winston (1998) found a few factors that
distinguished between their dropout and good outcome groups (comprised of 25 and 28
subjects, respectively). Therapeutic alliance ratings by both the client and therapist from
early in treatment were able to discriminate between dropout and good outcome clients,
with scores being significantly worse for the dropouts. Further, therapist ratings of client

hostility and session smoothness and client ratings of therapist friendliness and session

14

depth were also able to successfully discriminate between the dropout and good outcome
groups. Patients seen as most hostile paired with therapists viewed as unfriendly were
typical of the dropout scenario, as was a lack of session depth and smoothness. These
variables also appear to have good potential for relating to resistance. Schuller, et al.’s
(1991) measure of resistance was found to have four factors descriptive of differing
manifestations of client resistance, Abrupt/Shifting, Oppositional, Flat/Halting, and
Vague/Doubting. It would seem that the lack of therapeutic alliance development and
hostility observed by Samstag, et al. would be indicative of potentially oppositional
behavior. Additionally, the variables related to session depth and smoothness could easily
be correlated with the abrupt-shifting and/or flat/halting resistant behaviors.

Brogan, Prochaska, and Prochaska (1999) used measures based on the
transtheoretical model of change to predict termination/continuation status of sixty clients
from a community mental health center, university counseling center and a doctoral
training clinic. They found that when compared to client who either terminated therapy
appropriately or continued treatment, premature terminators were more oriented toward
changing their environment rather than themselves. They look to change the
environment because they do not view themselves as having a problem. Thus, they are
unable to face the necessary therapeutic issues, which may inevitably lead to their
dropping out. This lack of desire to change oneself is not unlike the previously
mentioned Opposing Change dimension of resistance established by Mahalik (1994) in
his Client Resistance Scale.

Reis and Brown (1999) conducted a very thorough review of 30 years worth of

psychotherapy dropout research. They concluded that divergent expectations for each

15

other and the therapy process between client and therapist, a weak working alliance, and
client dissatisfaction all increase the likelihood of unilateral termination. These factors
seem to at least have the potential to be related to resistance, especially the dimensions of
opposing the therapist and opposing change.

While the previously summarized group of studies varies greatly with regard to
the independent variables of interest and treatment setting, each of them has at least one
variable that could be conceptualized in terms of client resistance. Thus, there is previous
empirical evidence, from which theoretically sound hypotheses can be derived. Further,
because there has been conflicting evidence with regard to the level of psychopathology
present with psychotherapy dropouts, I have made this an additional variable of interest.
Little has been written suggesting whether there may be differences in the level of
resistance present dependent upon the level of pathology individuals are experiencing.

Therefore, there are multiple reasons to include a measurement of psychopathology.

l6

RATIONALE

Its important to remember that how one views and interprets client behavior may
be very different depending on whether one looks at manifest behavior on a molecular
level versus when attempting to understand the same behavior from the informed
perspective of an inclusive and broad-based theory. Drawing on theory helps the
clinician draw meaningful links between otherwise isolated behaviors. This is well
illustrated in VanDenberg and VanDenberg’s (1992) exploration of a single case of
premature termination from three different theoretical perspectives.

Some authors have also criticized studies that operationalize premature
termination based on a cutoff of number of sessions attended. However, while this may
not be an all inclusive way to capture every premature terminator, it is likely that many, if
not a majority of the clients qualify as dropouts based on this criteria, especially when
viewed from a dynamic/analytical theoretical perspective. Again, without a guiding
theoretical perspective, it may be easy to accept that a client leaves treatment early
because “they got what they need ” in a couple sessions, when in actuality it may be a
manifestation of “flight into health” or an extreme expression of resistance. Indeed, there
is fairly strong empirical evidence to suggest that there is a minimal amount of therapy by
which clients show improvement. In a meta-analysis of outcome studies by Howard,
Kopta, Krause, and Orlinsky (1986), they found that with a sample of 2,400 patients,
eight sessions was the point by which 50% of them began showing improvement. Many
other studies have found that early dropout is especially associated with lack of
improvement, worse outcome, and in some cases even an increase in symptomology

(Reis & Brown, 1999). Chamberlain, et al. (1984) defined dropout as any clients who

17

completed a minimum of two, but not more than four, sessions of treatment. They then
made successful hypotheses based on the presumption that clients who leave therapy
prematurely are more resistant to treatment than those who stay. Thus, in the case of the
present study where resistance (and its varied expressions) will be used to attempt to
predict early dropout, the use of a minimum number of sessions attended prior to
termination to define “prematurity” is appropriate and has some precedence. Simply
stated, the broadest application of resistance will be used, meaning anything that leads to
a client entering therapy only to exit soon thereafter could theoretically be termed
resistance. As so defined, there may be other forms of premature termination or other
critical periods of client resistance that cannot be addressed by this study. However, the
benefit of making an empirically-based link between early dropout and resistance theory
outweighs these limitations.

Thus, this study aims to respond to the suggestions of reviewers such as
Mennicke, Lent, and Burgoyne (1988), Wierzbicki and Pekarik (1993), and others that
have indicated the need for more complex, theory-based models of therapy process and
attrition be examined by applying a broad-based theory of resistance, similar to that
employed by Mahalik (1994) and Schuller et al. (1991) in their development of resistance
measures, to attempt to differentiate between those clients who dropout of therapy early

versus those who continue to the point of mutual termination.

18

 

HYPOTHESES
Hyp_othesis 1:

Psychotherapy clients prematurely terminating during the early phase of
treatment without discussing it with their therapist (Dropouts) will not differ
significantly from clients who continue in treatment beyond twenty sessions
(Persisters) with regard to their self-reported level of psychopathology at intake as
measured by the SOL-90R global severity index (GSI).

HMesis 2:

Psychotherapy dropouts will differ significantly from Persisters on measures
of quantity, intensity and quality of resistance they exhibit during their first and
third sessions of psychotherapy. The differences in estimated frequency (quantity) of
resistant responses as a proportion of overall coded responses will be measured by the -
Client Resistance Code (CRC) (Kavanagh, et al., 1984). It is hypothesized that Dropouts
will tend to respond with a higher proportion of resistant responses than Persisters. The
overall intensity and quality of the two groups’ resistance will be measured by the
Mahalik (1994)Client Resistance Scale (CRS). It is hypothesized that Dropouts will
exhibit higher overall levels of resistance and that they will specifically score higher on
the Opposing Change, Opposing Recollection of Material, and Opposing the Therapist

scales.

l9

Hypothesis 3:
Dropouts and Persisters will both exhibit a growing level of resistance

between the first and third sessions, with Dropouts beginning at a higher level and
outpacing the Persisters in amount of increase in resistance between sessions.
Persisters will exhibit a slower level of resistance growth between sessions, as

measured by the Client Resistance Scale (Mahalik, 1994) and Client Resistance

Code (Kavanagh, et al., 1984).

Hypothesis 4:

By evaluating levels of resistance in two groups of patients (Dropouts vs. Persisters)
with multiple, multidimensional measures (CRS and CRC) and then entering these
variables into a discriminant function, a model that predicts group membership can

be achieved that will consistently exceed the expected chance rate of prediction for

two groups (50%).

20

METHOD

_D_a_t§

The present study utilized data, which had been collected for research purposes at
the Michigan State University Psychological Clinic (The MSU Psychotherapy Research
Project). The clinic is a training and research facility and is part of the Michigan State
University Department of Psychology. Doctoral-level clinical psychology graduate
students, who are under the supervision of PhD. clinical psychologists, staff and provide
services at the clinic. The clinic is a non-profit, fee for service setting which provides
low-cost outpatient psychotherapy to residents of the Mid-Michigan region (greater
Lansing, MI) and to MSU students who have been referred from the university’s
counseling center for long-tenn psychotherapy, a service that is generally unavailable
there. The clientele served typically presents with a wide variety of psychopathology
including, but not limited to moderate to severe depression and anxiety, eating disorders,
anger control difficulties, and interpersonal relationship difficulties. The data was
originally collected with thepr of gathering relatively unintrusive information on
the clinic’s clientele and their experiences in psychotherapy, which could be utilized in
later research endeavors.
Particimts

The participants for the current study were selected from the pool of clients who
had previously volunteered to take part in the MSU Psychotherapy Research Project.
During initial intake interview at the MSU Psychological Clinic, these participants were
asked to take part in the ongoing therapy research project. They were informed that their

choice to participate or not would not affect the services they were receiving. A small

21

incentive was offered. If they chose to participate and completed the necessary pre- and
post-therapy forms (consent form, SCL-90R, Interpersonal Checklist, and Dream
Questionnaire), they were given a refund equal to ten percent of their fees paid for
therapy up to a maximum of eighty dollars. Historically, over ninety-five percent of
incoming clients at the MSU Psychological Clinic have agreed to participate.

The participants of this study were 40 clients (20 early treatment dropouts and 20
persisters matched by demographic information) of the Michigan State University
Psychological Clinic. This study’s “Dropout” group was created from participating
clients who completed the consent form and initial battery of questionnaires and returned
them at their first session of psychotherapy only to later drop out of treatment prior to
their eighth therapy session (m=5.55 sessions attended; SD=2.06). Considerable efforts
were made to ensure that individuals terminating treatment due to a medical condition,
hospitalization, unexpected financial hardship, relocation, or who left to seek treatment
from an alternative mental health facility or professional were not included in the dropout
group. These efforts included listening for mention of such information in audio taped
sessions, reviewing research follow-up questionnaires, and looking for any available data
or information contained in the research database (e.g., notes to research coordinator
from therapists).

After identifying twenty participants for the dropout group, each of the members
of that group were matched with a participating client who persisted in treatment beyond
a minimum of twenty sessions of therapy (m=24.9 sessions attended; SD=3.55), thus
creating an equal sized “Persister” comparison group. An extensive effort was made to

match these groups as closely as possible based on demographic information (sex, age,

22

SES, racial group, etc.). These efforts were generally successful in regard to sex, age,
and SES data, but were less successful in matching on racial group due to limited
minority-group representation in the overall sample.

The sample of 40 clients consisted of 28 females and 12 males ranging in age
from 20 to 57 years old with an overall mean age of 31 years. This was a predominantly
Caucasian sample (37 of 40 participants) with minorities underrepresented (only one '
African-American and two Asian-American participants total). Reported yearly income
ranged from $0 to $42,000. Marital status of clients broke down as follows: seven
married; twelve never married; eleven divorced; nine lived with a partner; and one was
widowed. The participants’ education levels ranged from nine to twenty years with two
years of post-secondary education being the modal level.

Therapists

All the participating clients were treated by graduate student clinicians with
experience ranging from one to four years of formal training in psychological assessment
and intervention. The therapists were beginning practicum students, advanced practicum
students, and students who were serving half-time clinical assistantships within the clinic,
many of whom had completed a Master’s degree. Despite the clinicians in this study
being made up of doctoral candidates in training, their level of training and education is
fairly comparable to that found in community mental health centers and other agencies
that provide similar therapy services. Specific demographic data and experience levels for

these clinicians was not available as it was not included in the data set.

23

Therapy Dyads

Not only were subjects matched by demographic data, but efforts were made to
match subjects also based on their having the same type of therapy dyad pairings (e.g.,
male therapist-male client, female therapist-female client, etc.). This was successfully
achieved for all subjects. Their were eighteen clients who were paired with a male
therapist and twenty-two clients paired with a female therapist. The percentages for the
four types of dyads broke down as follows: Four male therapist-male client dyads (10%):
14 female therapist-female client dyads (35%); 14 male therapist-female client dyads
(35%); and eight female therapist-male client, dyads (20%). Again, fifty percent of each

of these dyad types belonged to the dropout and persisting client comparison groups.

Tag Recorded Therapy Samples

All clients (and therapists) who participated in the therapy research project agreed
to have periodic therapy sessions audio taped The student clinicians involved in treating
the participants aided in data collection by turning in tape recordings from the first and
third sessions of therapy (and beyond for those clients who remained in treatment). These
recordings were assigned a numeric identification number to provide anonymity to
participants and protect the clients’ confidentiality. They were then compiled into a
library of therapy tapes used for research. The therapy samples for the present study were
collected between January of 1994 and June of 1998.

Following the identification of and creation of an equal sized comparison groups
of early dropout and persisting participants, a ten-minute sample from both the first and

third sessions were duplicated from each of the participants’ therapy tapes. In order to

24

avoid unintended effects due to. the stage of the session sampled from entering into data
collection and analysis, the ten-minute sample of session material was randomly selected
from either the first, middle or last twenty nrinutes of the session. Again, matched
subjects had similar stages of their therapy sessions sampled.

In order to ensure uniformity of the samples and to provide the coders with the
context of the initial client material presented in the recording, samples always began
with a clinician speaking turn. Samples were ended as closely to ten minutes as possible
without destroying the codeablity of the final client speaking turn. The samples were
digitized and attempts were made to improve the sound quality for optimal audibility
during coding. These individual samples were assigned a specific sample number to
help identify the resulting data and allow it to be assigned to the appropriate subjects
during data entry.

The sequence of the samples was randomized and recorded to compact disc (CD)
to create a “master” digital recording that allowed the raters to be blind to which group
the session material came from, as well as, whether the material was from the first or
third session. The same master recordings were employed by both trained coders to

produce the resistance data for statistical analysis.

Measures

SCL 90-R: The Symptom Checklist 90 Revised. Derogatis’ (1983) Symptom
Checklist 90-Revised (SCL-90) was used to measure participants’ number of symptoms,
level of self-reported intensity of symptoms and global psychopathology and at the time

of their entering therapy. It is a 90 item self-administered questionnaire composed of

25

nine subscales measuring nine symptom dimensions: somatization, obsessive-
compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety,
paranoid ideation, and psychoticism. Participants are instructed to read each of the items,
which ask to what degree are they distressed by particular symptoms (e. g., headaches,
nervousness, shakiness, etc.) and then to rate each item on a Likert type scale that ranges
from 0, not at all, to 4, extremely. Means are then computed for each of the nine
subscales. A Global Severity Index (681) is computed as the sum of all item responses
divided by 90. According to Derogatis (1983), the 681 is the best single indication of the
current level of pathology.

Internal consistencies for the SCL-90 subscales were reported to range between a
low of .77 for psychoticism to a high of .90 for depression. The test-retest reliability for
the SCL-90 was reported as ranging from .78 for hostility to .90 for phobic anxiety
(Derogatis, Rickels & Rock, 1976). Convergent validity has been supported (Derogatis,
et al., 1976) by successfully comparing the dimensions scores of the SCL-90 with the

scale scores of the MMPI.

CRS: The Client Resistance Scale. Mahalik’s (1994) Client Resistance Scale
(CRS) was used to measure participants’ level of resistance at two points early in
treatment (first and third therapy sessions). The CRS is a process measure of five
dimensions associated with client resistance. Trained judges use this rating system to
judge the level of “observable” resistance present across the five dimensions during a
therapeutic interaction between a client and therapist dyad. The CRS’s 5 subscales are
based on R.R. Greenson’s (1967) formulation of client resistance. These scales are

Opposing Expression of Painful Affect, Opposing Recollection of Material, Opposing

26

Therapist, Opposing Change, and Opposing Insight. Judges rate the client portion of
recorded (audio— or videotaped) interactions on each subscale according to a 7-point
Likert scale with higher ratings reflecting the presence of greater amounts of those
variables. These ratings are then summed to produce a global resistance scale score.
Reliabilities for each of the five resistance subscales using two judges ranged
between .92 for Opposing Recollection of Material to .71 for Opposing Insight.
Reliability increased across all subscales, as three and four judges were employed.
Interconelations between subscales ranged from .31 to .62, indicating that the CRS is
unidimensional. However, there were meaningful differences in level of resistance found
based on the assessment of different clients, which suggests an examination of the
subscales has qualitative value and supports the view that resistance, as a construct, is

multidimensional.

Client Resistance Code (CRC). Kavanagh, Gabrielson, and Chamberlain’s (1982)
Client Resistance Code was used to measure the frequency of participants’ resistant
responses during segments of their first and third sessions of psychotherapy. The CRC is
a seven-category coding system that was designed to measure resistance during single
therapy sessions. The system has five categories that are used for coding resistant client
behavior (e.g., Interruptions/talking over, Not tracking, and Negative attitude) and two
for coding cooperative behaviors. For the purpose of the current study only the five
categories of resistant behaviors will be used and as in the case of Chamberlain, et al.’s
(1984) study these five categories will be collapsed into a single global scale of

resistance. This scale measures the total number of resistant behaviors observed during a

27

given therapy session, which is then divided by the total number of minutes coded to
produce a rate per minute score. Rather than just counting the number of resistant
responses and then dividing them by the same ten nrinute intervals, this study will tally

the number of resistant responses and then look at what the proportion of responses

during that coded segment were resistant (CRC resistant responses + Total client
speaking turns). Inter-rater reliability was reported to be r=.98, p<..001, for the
resistance scale. The successful application of this coding system to corroborate
theoretically based assumptions about resistance during differing stages of treatment,
resistance and dropout, and resistance levels in self vs. agency referred clients provides

support for its construct validity (Chamberlain, et al., 1984).

Ram

The primary research investigator trained two raters to employ the CRS and CRC
using clinical material, prior to their employing these measures with the actual research
therapy samples. Both raters were male graduate students in clinical psychology who had
completed their second year of graduate level coursework and a year of clinical
practicum. Initially the raters were given readings to familiarize them with the
background theory from which the resistance scales were developed. Following this, the
raters met with the primary researcher and were given basic instruction with regard to the
standardized procedures for employing each of the resistance measures. A number of
additional samples of therapy sessions were obtained from the therapy tape library and
used for a series of rating practice sessions. These practice sessions were held until

adequate(r>=.80) inter-rater reliabilities were achieved. Post-training inter-rater reliability

28

for each rater considered singly with a Pearson’s Product Moment Correlation achieved
r1=.82, p<.05 levels. The aggregate or effective reliability that considers both raters
simultaneously by using the Spearman-Brown formula achieved rk=.90, p<.05 levels.
Both raters coded all of the selected therapy samples (80 ten-nrinute samples) using both
the CRC and CRS. Coding was completed over the course of approximately six weeks.

Raters averaged spending three minutes per minute of therapy sample to code the data.

Procedures

Upon agreeing to take part in the psychotherapy research project, participants
completed a consent form and the Symptom Checklist-90 Revised (SCL-90R). These
forms had to be completed prior to their first therapy session. The forms were returned to
the student clinician by which they were treated at the commencement of treatment. The
student clinician in turn passed these forms and designated session tape recordings to a
graduate student who independently managed the data collection for the ongoing project.
This study did not use recently collected data, rather all the data was archived at
minimum three years prior to its initiation.

The archived data did have to be checked for completeness and sound quality.
Following the training of raters and initial reliability checks, the pre-therapy SCL-90
questionnaires were compiled and scored for the two groups and entered into a computer
database.

Coding of the ten-minute segments of both the first and third therapy sessions
commenced shortly after training sessions were completed. Coders were blind to whether

the segments were from first or third sessions and client group assignment (dropout vs.

29

persister), only identifying the segments by an identification number. They were
instructed not to discuss their ratings with each other and were assigned a different
sequence of the segments to follow. The raters coded the therapy samples independent
from each other following the directions outlined in the Manual for the Client Resistance
Spale (Mahalik, 1994) and as described by the authors of the Client Resisance Code
(CRC) in the Manual for Coding Client Resistance (Kavanagh, Gabrielson, &
Chamberlain, 1982). The coders rated each therapy segment with both measures
simultaneously, though this typically required multiple reviews of the segments. The
coders rated each speaking turn independently, but these individual speaking turn ratings
were later summarized into segment averages for each of the CRS scales. An accounting
of how many speaking turns per segment was maintained which allowed for the
proportion of CRC resistant responses to be calculated for each segment.

A mid-coding reliability check was conducted using non-research therapy session
samples to ensure that inter-rater reliability was being maintained. Reliability showed
some decline with the Pearson correlation slipping to r1=.72, p<.05 and the Spearman-
Brown correlation falling to rk=.83, p<.05. However, with the aggregate correlation
(Spearrnan-Brown) still above the .80 level, this was still considered acceptable to the
primary researcher. Further practice coding and discussion was conducted in addition to
the reliability check to “recalibrate” the reliability of the coders with the hopes of
improving the reliability of the remaining ratings.

Following the completion of all session sample coding, the sample ratings were
reorganized back into appropriate group (dropout and persister). The data was then

entered into an SPSS spreadsheet for statistical data analysis. A combination of chi-

30

squared analysis, correlations, one-tailed t-tests, one-way analyses of variance, 2x2 rrrixed
(within-between) subjects analyses of variance, and discriminant function analyses were

applied to the data set to evaluate the accuracy of a priori hypotheses.

31

RESULTS

Demographic comparisons between grpups

The presence of possible confounds was investigated prior to performing the
pre-planned data analyses. A combination of one-way analysis of variance (ANOVA)
and Chi-squared analysis was used to check if there were significant differences
between the Dropout and Persister groups on any demographic variables that may have
impact on the other analyses, as prior research has shown such variables to have weak,
but statistically significant relationships to dropout . As stated previously, subjects were
paired specifically by sex and and the dyad pairing (e.g., male client-male therapist,
etc.). therefore there was no need to measure differences in these categories.
Statistically significant differences were not found between groups on other potential
cenfounding demographic variables (See Tables 1 and 2). Therefore, the data analysis

proceeded as planned.

Interrater Reliabilities

The interrater reliabilities for the Client Resistance Scale (CRS) and Client
Resistance Code (CRC) ratings are shown in Tables 3 and 5, respectively. Pearson
product-moment correlations give the measure of reliability for each of the individual
judges independently, these ranged from .51 (Opposing Change) to .76 (Opposing
Recollection of Material) for the CRS subscales with an overall correlation of r=.71 for
the subscales collapsed into a single measure of resistance. The CRC had Pearson
correlations of .78 and .85, for the resistance and cooperative scales respectively. These

correlations were a bit lower than hoped for by the investigator. To improve the

32

reliability of these ratings, the ratings for each scale were averaged to produce a single
score for each session-segment rated. By treating the data in this way, it allows the
investigator to amplify reliability data through the application of the Spearman-Brown

formula for estimating “effective” or aggregate reliabilities:

R = number of ratcrs*(average of individual correlations)

1 + (numer of raters - 1)* average of individual correlations

(Rosenthal & Rosnow, 1991). This procedure is essentially the same as that used for
determining split-half reliabilities, only in this case the two raters are treated as if they
were the halves. By looking at the interrater effective reliabilities, the CRS subscale
reliabilities improved to a low of .68 (Opposing Change) to a high of .86 (Opposing the
Recollection of Material) and an overall correlation of .83 for the subscales collapsed
into a single measure. These reliabilities were significantly lower than those reported by
Mahalik, the CRS author, who had achieved subscale reliabilities that ranged from .71
to .92. However, as the overall correlation exceeded .80 this was viewed as acceptable
by the primary investigator. Effective reliability conversions improved the correlations
on the two CRC scales to .88 and .91, for the Resistance and Cooperative response

scales respectively, which is an acceptable rate by most research standards.

Overview of Resistance Variables

Subcale intercorrelations were statistically significant between all possible
combinations of CR8 subscales (correlations ranged from m .33 to .45), with the
exception of Opposing the Expression of Painful Affect scale and Opposing the
Therapist which were found to have little relationship (:07, non-significant at p<.05).

This pattern of intercorrelation supports Mahalik’s (1994) conclusion that the CRS is

33

generally a unidimensional scale, though the subscales are worthy of consideration
individually to add to qualitative understanding of the different facets of resistance. It
was quite discrepant for the Opposing Expression of Painful Affect and Opposing the
Therapist scales to be uncorrelated using this sample’s ratings as Mahalik had found
this combination of subscales to be the most related of all subscale combinations
(r=.62).

The two scales of the CRC were found to have no relationship (r=.08,
nonsignificant at p<.05) land are therefore viewed as independent variables. This was as
expected due to the mutually exclusive nature of the two scales and this allowed the
investigator to focus analysis on the resistant response scale alone.

Concurrent validity between the two measures of resistance was calculated using
Pearson product-moment correlations between the overall and subscale scores of the
CRS and the resistant response scale of the CRC. Table 6 summarizes this analysis. The
correlation between the CRS and CRC was statistically significant (r=.46, p<.05)
suggesting that these two measures are moderately related. Intercorrelations between
the CRC and the CRS subscales were statistically significant for four out of five of the
subscales, ranging from .29 (Opposing Experiencing Painful Affect) to .53 (Opposing
the Therapist), only the Opposing Change subscale was not correlated. This is
consistent with the view that these two measures are measuring different parts of the

same thing (frequency/quantity (CRC) vs. intensity/quality of resistance (CRS)).

34

Hyppthesis I
Psychotherapy clients prematurely terminating during the early phase of

treatment without discussing it with their therapist (Dropouts) will not differ
significantly from clients who continue in treatment beyond twenty sessions
(Persisters) with regard to their self-reported level of psychopathology at intake as
measured by the SCL-90R global severity index (GSI).

In order to evaluate possible differences between the dropout and persister
groups in initial level of psychopathology, a one-tailed, independent samples t-test was
conducted comparing group means on the SCL-90R global severity index. In addition
to the global severity index, additional t-tests were also conducted to explore possible
differences on the positive symptom index and the nine subscales of the SCL-90.

Results of the statistical analysis confirmed the hypothesized lack of differences
between dropouts and persisters on presenting level of psychopathology. Global
Severity Index scores for the two groups did not significantly differ, t (38) = —0.282,
p=. 78. Dropouts actually had a slightly lower average GSI score (m=l.16; SD=.67)
than the persister group (m=l.23; SD=.78), but again these differences were unable to
achieve statistical significance. Further analysis of SCL-90R data was unable to
produce any positive results. Neither the positive symptom index nor any of the nine
SCL-90R subscales produced significant mean differences between the two groups.
Point-biserial conelations between the groups and SCL-90R scales further failed to
establish any relationships.

As the study’s focus was on resistance levels, correlations between the SCL-90R

global severity scale and CRC and CRS scales were also evaluated. First session

35

resistance scores on both the CRC and CRS (including all subscales) did not produce
any significant Pearson correlations. Out of the analysis of third session variables, only
levels of Opposing Insight correlated with the SCL-90R global severity index (r=.29,

p<.05, one-tailed).

H esis II

Psychotherapy dropouts will differ significantly from Persisters on
measures of quantity, intensity and quality of resistance they exhibit during their
first and third sessions of psychotherapy.

Independent samples t-tests were conducted in order to evaluate differences
between dropout and persister groups on the Client Resitance Code resistant response
proportions, the total Client Resistance Scale scores, and the CRS subscales. T-test
results for the CRC resistance response proportion scores for both the first and third
sessions approached statistical significance, t(38) = 1.83, p=.075 and t (38)=l.98,
p=.056, respectively. Confirming hypothesis two, when the overall proportion for both
sessions combined was evaluated the t-test was found to be statistically significant, t
(38) =2.70, p = .01. As predicted, participants in the dropout group tended to have a
higher proportion of resistant responses (46.26 % vs.33.45 % in the first session and
51.54% vs. 37.39% in the third session). The overall percentage of resistant responses
for both sessions were 48.2% (SD=18) for dropouts and 34.4% (SD=14) for persisters.
The strength of this relationship between group and CRC resistant response proportion

as indexed by 1] was .40 indicating a moderate effect size. The effect size as estimated

36

by the d statistic was .86 for CRC resistant response proportion. See Table 7 for
summary of CRC t-test data.

Contrary to predictions, the effect for group (dropout vs. persister) did not
achieve significance for overall CRS scores, I (38) = .387, p > .05. On average
participants in the dropout group demonstrated slightly higher scores on the CRS but
this difference was not significant. The strength of the relationship between group and
CRS scores as indexed by d was .12 indicating a small effect size. See Table 8 for
summary of CRS t-test data.

Further analyses of group differences on the CRS subscales failed to reveal any
statistically significant results. Figures 1 and 2 provide graphic depiction of mean
comparisons for both first and third session ratings on the CRS subscales. The lack of
positive results is likely influenced by small sample size. Three non-significant results
are of note due to effect size estimates that are suggestive of a relationship between
these variables and group membership. First session Opposing the Therapist scores
produced near significant results, I (38) = 1.78, p=.08 with effect size estimates of 7)
=27 and d = .56. Third session Opposing Recollection of Material and Opposing
Change also approached statistically significant mean differences, r (38) = 1.43, p=.16
and t (38) = 1.52, p=.14, respectively. Effect size estimates for these two scales were :7
=22 and d = .45 and 17:24 and d = .46, respectively. Seven out of the ten subscale
measurements (first and third sessions for each of the five subscales), produced mean
differences in the predicted direction (dropout group higher score). Three subscales
interestingly produced mean differences that were in the opposite direction of the

hypothesis, first session Opposing Recollection of Material, third session Opposing

37

Expression of Painful Affect and third session Opposing Change. Only the third session

Opposing Change differences approached statistical significance.

HMesis III

Dropouts and Persisters will both exhibit a growing level of resistance
between the first and third sessions, with Dropouts beginning at a higher level and
outpacing the Persisters in amount of increase in resistance between sessions.
Persisters will exhibit a slower level of resistance growth between sessions, as
measured by the Client Resistance Scale (Mahalik, 1994) and Client Resistance
Code (Kavanagh, et al., 1984).

In order to evaluate differences in the change in resistance over time between
dropout and persister groups, 2x2 within subjects ANOVAs were conducted. Means
and standard deviations for CRC and CRS data in sessions 1 and 3 are summarized in
Tables 7 and 8 . Graphical depiction of the between session growth on the two
resistance measures is shown in Figures 3 and 4.

Contrary to the hypothesis, dropout and persister groups demonstrated fairly
similar levels of CRS measured resistance at session one and did not demonstrate
significant differences in patterns of resistance growth (Group X Session) F (1,38) =
.039, p > .05. The witlrin-subjects main effect for session supported the hypothesis that
there would be increasing levels of resistance, though this analysis failed to reach
statistical significance F (1,38) = 2.166, p =.15. This is likely due to insufficient sample
size and there being a relatively small effect, eta was estimated to be .23. Observed

power estimate was beta equals .30 for this main effect analysis.

38

Additional analysis of CRS subscales only produced one statistically significant
main effect for Opposing Recollection of Material (See Figure 5) F (1, 38) = 4.25,
p<.05. There was a medium effect size of eta=.32 and power was estimated to be beta
equals .52. This same variable produced an interaction (Group X Session) that
approached significance F (1, 38) = 2.05, p=.16. Effect size of the interaction was
estimated to be eta equals .23 representing a relatively small effect and power was very
low for this analysis at beta equals .29.

Consistent with the hypothesis, dropout and persister groups demonstrated
differences that approached statistical significance on CRC resistant proportion scores
at both first and third sessions. The within subjects main effect for session demonstrated
the anticipated increasing levels of resistance, though this analysis failed to reach
statistical significance F (1,38) = 1.05, p 2.31. This is likely due to insufficient sample
size and there being a relatively small effect, eta was estimated to be .16. Observed
power estimate was beta equals .17 for this main effect analysis. Dropout and persister
groups did not demonstrate significant differences in patterns of resistance growth

(Group X Session) F (1,38) = .022, p > .05.

Hymthesis IV

By evaluating levels of resistance in two groups of patients (Dropouts vs.
Persisters) with multiple, multidimensional measures (CRS and CRC) and then
entering these variables into a discriminant function, 'a model that predicts group
membership can be achieved that will consistently exceed the expected chance rate

of prediction for two groups (50%).

39

A step-wise discriminant function analysis was conducted to develop a
mathematical equation capable of classifying participants into either the dropout or
persister groups using the resistance variable data. The group variable degrees of
fieedom value for this two-group design is one; so one discriminant function was
calculated. The analysis produced a model that predicted group membership from the
scores of four variables (overall CRC resistant response proportion across both sessions,
overall CRS Opposing the Therapist across both sessions, third session CRS Opposing
Expression of Painful Affect, and third session CRS Opposing Recollection of
Material). The single discrinrinant function was significantly associated with group
membership [X2(4) = 11.37 , p < .05] and the canonical correlation coefficient related to
this function was r = .52. Standardized discriminant function coefficients for the four
variables included in the analyses following the step-wise procedure are presented in
Table 9.

The classification accuracy of the discriminant function is shown in Table 10. The
discriminant function achieved sensitivity, or a true positive rate for dropouts of 70%
and a specificity, or true negative rate for persisters of 85%, with an overall
classification accuracy of 77.5%. Furthermore, the use of the discrirrrinant function
results in a decrease in error rate of 27.5% compared to decisions of group membership
based on sample size alone. Cross-validation was conducted for all the cases in the
analysis. In cross validation, each case was classified by the functions derived from all
cases other than that case, the cross-validation yielded an overall accuracy of 72.5%.

The reduction in accuracy from the original discriminant function analysis was due to

40

misclassification of two persister group participants, results for classification of
dropouts remained the same (70%).

Each predictor variable increased the overall classification rate as it was added
to the function. The combined CRC resistance response scale was the most influential
predictive variable, as when it was used alone, it was able to successfully classify
62.5% of the subjects into their correct group. Upon adding the third session CRS
Opposing Recollection of Material variable to the discriminant function, the overall
classification rate rose to 65%. Followed by a 5 % increase to 70% with the addition of
third session CRS Opposing Expression of Painful Affect and finally, a 7.5% increase
when combined first and third session CRS Opposing the Therapist was added. When
all of the variables are entered together, the contribution of each variable becomes more
apparent. In Table 11 pooled within-groups correlations between discriminating
variables and standardized canonical discrirrrinant functions are displayed. This matrix
provides another way to study the usefulness of each variable in the discriminant
function when all variables are entered simultaneously.

The intercorrelations between the variables of interest in this study revealed
significant correlations ranging from .07 to .53. The intercorrelations and redundancy
of many of the variables entered into the model likely contributed to the final
discriminant function using only four variables out of a possible 18. Although the
variables account for different facets of resistance, different periods of resistance
expression (1’t or 3rd session), and differing ways of applying these measures (combined .
versus individual sessions measures) the resultant overlap in variance accounts for the

limited discriminating power of some of the variables during the step-wise procedure.

41

DISCUSSION

This study set out to establish further empirical evidence of a link between
observable forms of client resistance and premature termination from psychotherapy.
This was pursued using two previously created measures of client resistance, the Client
Resistance Code (CRC) and the Client Resistance Scale (CRS). The CRC was chosen for
its ease of application and previous success at providing evidence of the link between
resistance and dropout. It functioned as a measure of the quantity or proportion of
resistant responses given by the research participants. The CRS, on the other hand was
selected due to its theoretical relevance to Greenson’s (1967) writings and for its
demonstrated broad applicability to samples from a variety of theoretical orientations.
The CRS was viewed as a measure of the quality and intensity of resistance exhibited by
the subjects. It provided more specific information regarding how the participants
manifested their resistance to treatment. It was hoped that by applying these instruments
in tandem to samples of psychotherapy sessions of both dropOuts and a comparable group
of patients who remained in treatment, new insights into the relationship between
resistance and dropout would emerge and a model capable of predicting group
membership would be developed. What follows is an explanation and an integration of

the reported results with the extant literature on resistance and dropout.

42

Hyppthesis I

Hypothesis I was included to address the influence of psychopathology levels on
dropout. The hypothesis in this case was that there would be no difference in initial levels
of psychopathology between the dropout and persister groups. This hypothesis was tested
to rule out potential confounding variables that might interfere with the ability to assess
the influence of resistance on dropout. Had there been differential levels of pathology
between the groups, level of initial psychopathology would have been covaried in the
statistical analysis implemented in Hypotheses II and III and entered into the stepwise
discriminant function analysis in Hypothesis IV.

Hypothesis I was confirmed with there being little difference in initial levels of
pathology between groups as measured by the SCL-90 Global Symptom Index. In fact,
additional statistical analyses were unable to produce any significant differences between
the dropout and persister groups when looking at SCL-90 subscales or the Positive
Symptom Index, lending additional strength to the hypothesized lack of difference in
initial pathology.

There have been many studies with conflicting outcomes that address the
influence of pathology on premature termination. Strongly supporting the present study’s
findings, a sizable study of 815 outpatients at a federally funded community mental
health center by Stahler and Eisenman (1987) also found no difference between dropouts
(patients attending less than three sessions) versus non-dropouts on initial self-reported
levels of symptomology as measured by the SCL-90. However, they did find that patients

differed on diagnoses with the non-dropout group containing more psychotic patients .

43

They also found that therapist ratings of the patients’ level of functioning at intake were
higher for the dropout group, suggesting they were higher functioning than the non-
dropout group. While this study’s results appear to closely approximate those of the
present study, especially regarding SCL-90 scores, it must be noted that the dropout
patients in the present study would have been considered non-dropouts using Stahler and
Eisenman’s dropout criteria, as all of the present study’s subjects attended a minimum of
three sessions of therapy.

In Smith, Koenigsberg, Yeomans, Clarkin, and Selzer’s (1995) study of
psychodynarnic psychotherapy for borderline personality disorder with thirty-six patients,
dropout versus continuation in treatment was studied using demographic, prior treatment,
and clinical variables (including the SCL-90R Global Severity Index). SCL-90R GSI
scores failed to contribute successfully to the prediction of group membership (dropout
vs. continuer). The only clinical variable that contributed was a measure of hostility
(Buss-Durkee Hostility Inventory). These findings appear consistent with those of the
present study and in this case, the type of psychotherapy being employed by the therapist,
as well as the length of treatment received by dropouts closely approximates those found
in the present sample.

Some additional studies also reporting corroboratory findings include Martin,
McNair, and Hight (1988) who studied 148 student clients at a university counseling
center. Their study specifically found that State-Trait Anxiety Inventory (STAI) scores
did not differ between early prematurely terminating (EFT) clients (those unilaterally
discontinuing treatment after one or two sessions) and non-EFT clients. Additionally, Oei

and Kazmierczak (1997) using a variety of intake assessment measures were not able to

predict group membership between dropout and completer clients (n=131) from
cognitive-behavioral group therapy for mood disorders. Further Masher-Ashley (1994)
found no significant differences between groups of elderly psychotherapy patients who
dropped out versus those completing treatment with regard to diagnosis, severity of
symptoms at intake, and level of discomfort caused by the reported symptoms.

The present study’s results were contrary to those of four dropout studies
reviewed that considered the influence of initial levels of pathology on premature
termination (Richmond, 1992; McCallum, Piper, & Joyce, 1992;1-Iilsenroth, et al., 1995;
and Hatchett, Han, & Cooker, 2002). Richmond (1992) conducted a dropout study with
624 individual adult therapy clients at a nonprofit outpatient mental health clinic.
Advanced doctoral students with a primarily psychodynamic theoretical orientation
staffed this clinic. The results of this study found evidence of differences between therapy
completers and dropouts on a variety of initial symptomatic measures including: DSM
III-R Axis II diagnoses, grandiosity, and guilt. This study’s findings are of note due to its
large sample size and the similarity to the patient p0pulation and setting to the present
study. As noted in the literature review, Hilsenroth et al. (1995) found that dropouts were
less disturbed, more capable of establishing cooperative relationships, and had less need
for interpersonal contact. McCallum, Piper and Joyce’s (1992) study of dropout from
short-term group therapy successfully employed SCL-90 scores as one of three predictors
that differentiated dropouts (n=17) from remainers (n=3 8) at an overall seventy-two
percent success rate.

While there appears to be more studies that support a lack of differences in initial

pathology between therapy dropouts and persisters, the previous review illustrates well

45

the problems that arise from the different approaches employed by researchers in defining
dropout and premature termination. As there is such a lack of consistency in the
supporting research and there are weaknesses and limits to the current study, the present
results should be interpreted with caution and cannot be viewed as unequivocal. The
reviewed literature would also suggest that factors such as the type of treatment setting,
the kind of intervention employed (e.g., individual vs. group therapy), and therapist
factors (e.g., theoretical orientation, training level) may have interactive or moderating
influences on the eventual persistence or premature termination of a patient. Improving
definitions of dropout and collecting follow up data directly from those patients who
dropout would likely clarify this and improve the quality of future research.

Something to be considered regarding the present study’s negative findings is that
despite a lack of differences in initial psychopathology endorsed, we know nothing
regarding the differential impact of the psychotherapy (3-8 sessions) received by these
patients on their pathology. This issue was addressed by Pekarik (1992) in an outcome
study evaluating differences between 94 dropout and completing outpatients (47 child
«and 47 adult cases) at a public mental health clinic in a mid-sized Midwestern city. One
focus of Pekarik’s outcome study was that early (1-2 sessions) and late dropouts (3 or
more sessions) differed in their differential response to therapy, with the early dropouts
continuing to exhibit a greater lack of adjustment than patients from the later dropout
group. Little difference in response to treatment as measured by self-reported pathology
could be detected between the late dropout group and those that completed treatment. The
present study’s dropout group was more similar to Pekarik’s late dropout group and the

lack of differences in initial pathology are consistent with his study’s’ findings. Had post-

46

dropout follow up measures of pathology been collected on the present study’s dropout
group, there may have been similar lack of differences in response to treatment. This
leads to speculation that the further a patient proceeds in the treatment process prior to
dropping out, the more likely it may be that they are leaving treatment due to actual
symptomatic relief rather than a conscious or unconscious resistance to treatment.
However, caution regarding such speculation should still be taken, as self-reported level
of pathology could also be an outward manifestation of a patient’s denial and/or resistant
“flight into health.” Clearly this is an aspect of psychotherapy dropout that warrants
continued research attention, but that will likely continue to pose a challenge to
investigators with regard to deriving clear and interpretable results.

For the purposes of the present study, the fact that there were not differences in
initial pathology endorsed by members of the two groups was considered fortuitous, as it
provided a natural control for a potentially confounding variable. Since this study was
employing a quasi-experimental design with a convenience sample, where random
assignment and a control group could not be employed, being able to rule out at least one
potentially influential external source of variance on the variable of interest (observable

resistance) was very helpful.

Hymthesis II
This hypothesis tested the primary contention of the study that there would be

differences in the quality, quantity, and level of resistance observed between dropout and
persister subjects. It was contended that dropouts would exhibit a higher proportion of

resistant responses as measured by the Client Resistance Code (CRC), with greater

47

   

intensity and predictable patterns of subscale elevations on the Client Resistance Scale
(CRS). Statistical analysis confirmed that the dropouts in this sample responded to their
therapists with a higher proportion of resistant responses than persisters. This is
consistent with the results of Chamberlain, et al.’s (1984) study in which the CRC was
originally applied to adults participating in parent-training. Chamberlain found that
significantly more “high resistant” participants dropped out of the parent training
program than “low resistant” participants (54% vs. 14%, respectively). The present study
demonstrated the utility of the CRC to other forms of treatment beyond didactic parent-
training.

Statistical analysis of the Client Resistance Scale (CRS) and its subscales was met 7
with less success than that of the CRC. Two problems existed that likely undermined the
ability for this study to confirm the hypotheses related to differences on the CRS
measures. First, the complexity of the CRS is much higher than that of the CRC. It
required the research coders to simultaneously assess five coexisting subcategories of
resistance for every client speaking turn versus only two mutually exclusive categories
with the CRC. The difficulty of this task contributed to increased variance and less than
ideal reliability rates on some of the CRS subscales (see Table 3). This combined with a
relatively small sample size made it difficult to attain statistical significance when
making comparisons between groups on the CRS subscales. Those subscales where
results seemed to approach significance were very likely the victim of small sample size
and inadequate power for the size of effects exhibited. There is still value in evaluating
the trends found in the results of the present data analysis and comparing it with existing

research results, as meta-analyses or future prospective studies may find such information

48

of value. However, interpretation and generalization of such findings should be done
tentatively with an emphasis on encouraging further confirmatory research that will
overcome the aforementioned shortcomings of this study.

While the CRS subscale scores did not differentiate between dropouts and
persister groups, they provide some interesting insight into the quality of the observed
resistant behaviors amongst this sample of psychotherapy clients. Overall, there appears
to be a somewhat consistent pattern to the resistance exhibited in early sessions of
psychotherapy (See Figures 1 and 2). Avoiding expression of painful affect seems to be
the most intensely experienced form of resistance for all patients, followed by resistance
to change and difficulty making connections between the past and present (opposing
insight). Most patients at least initially are able to verbalize personally relevant material
and follow the lead of their therapist, though they may not elaborate or pursue greater
depth of discussion.

Opposing expression of painful affect was the most strongly manifest subtype of
resistance amongst this sample and there was the least amount of difference between the
dropout and persister groups on this measure. This is very consistent with the well-
accepted theoretical view of resistance as an inherent, unconscious striving to avoid
thoughts and feelings that cause discomfort (Beutler, Rocco, Moleiro, & Talebi, 2001).
The subjects’ scores on the Opposing Change and Opposing Insight subscales were not
far behind opposition to expression of painful affect in level of intensity exhibited. These
three subscales were moderately interconelated (ranging from .38 to .45). The elevation
on the Opposing Insight scale could represent an extension of the client’s avoidance of

uncomfortable thoughts and feelings (Opposing Expression of Painful Affect). Making

49

connections between present experiences and feelings to historical events in one’s life
require first, identifying and recognizing current (possibly painful) emotions. The insight
process then intensifies those emotions and experiences by requiring the patient to look at
unresolved conflicts, past traumas, etc. that contribute to these in a way that is likely to
enlist and engage the patient’s psychic defenses. However, variations in patients’ level of
psychological mindedness and/or lack of experience or development of the ability to
make such connections early in treatment may artificially inflate their scores on this
subscale.

The elevated initial scores for both groups on the CRS Opposing Change subscale
can be interpreted as representing the well-established theory that Strean (1985) details in
his text on resistance. He asserts that all clients enter therapy with a great deal of
ambivalence despite their acknowledged difficulties and presenting problems. The idea of
seeking therapy as a solution is still often frightening and aversive. Interestingly, in the
current study resistance to change was almost unchanged between sessions one and three
for the dropout group, while it increased between sessions for the persister group. The
difference between dropouts and persisters on the third session Opposing Change scale
approached statistical significance. This is contrary to findings that are suggestive of
opposition to change having an influence on dropout in psychotherapy. Brogan,
Prochaska, and Prochaska’s (1999) study successfully used the transtheoretical model of
change to develop a discriminant function that correctly classified 92% of the sixty
patients in their study into two groups (premature and appropriate terminators).

Premature terminators exhibited a “precontemplation profile,” which could be considered

50

similar to the CRS subscale of opposing change, as it is indicative of a lack of recognition
for the need to change and tends to hinder movement away from the status quo.

CRS subscales that were less intensely expressed were the Opposing Recollection
of Material and Opposing the Therapist forms of resistance. However, when looking at
group differences these two scales tended to be more differentially expressed and
between session growth and predictive power of these scales seemed to be higher than the
other three subscales, as will be discussed later.

The first session Opposing the Therapist scale scores came closest to achieving
statistical significance, with dropouts exhibiting higher levels of this type of resistance
than persisters. This is consistent with other studies that address the development of
therapeutic alliance and/or client reactions to the therapist. Kolb et al. (1985) reported
that in their sample of 91 outpatients, those who dropped out were less involved in and
less changed by their treatment and tended to view their therapist as less supportive and
facilitative. In looking at initial session process variables within a sample of 32
outpatients, Bottari and Rappaport (1983) found that patient perceptions of the therapists’
style of relating and affect, related to the number of therapy sessions attended and benefit
derived from treatment. Beckham (1992) also reported negative client impressions of the
therapist are predictive of early dropout. Martin et al. (1988) reported that at least a
subset of their early premature terminators left due to negative feelings toward their
therapist (e.g., not liking them, feeling they were close-minded, etc.) These past studies
as well as the literature on the importance of the therapeutic alliance lend support to the

“nonsignificant” results of the current study regarding direct opposition/resistance to the

51

therapist and would further imply the need for further research as has been suggested by
prominent figures in this field of study (Beutler, et al., 2001).

The Opposing Recollection of Material and Opposing the Therapist subscales
were also moderately correlated (r=.40). It makes intuitive sense that a client’s resistance
to their therapist will influence their openness and willingness to produce personally
relevant material. Beutler et al. (2001) in reviewing previous resistance research reported
evidence that directive and nondirective therapeutic interventions were differentially
responded to by patients with differing levels of trait and state resistance. He indicated
that more resistant patients responded better to nondirective interventions and vice versa.
Differences on third session scores on the Resisting Recollection of Material subscale
approached statistical significance, with dropouts having higher levels than pesisters. It
may be that the mediating effects of different therapist styles are coming into play in this
situation, as they seem especially relevant to the measure of opposing recollection of
material. This alludes to the importance of pursuing such interactional data in future
research, as will be discussed later.

Again, data analysis failed to find statistically significant differences between the
dropout and persister groups on the CRS summed scores for both first and third sessions.
Analysis of the individual subscales also failed to find statistically significant results,
however, in a few cases the results approached significance. In attempting to understand
the pattern of results on the qualitative CRS measures of resistance a couple ideas emerge
for consideration. First, it may be that patients who have become engaged in the work of
psychotherapy will exhibit qualitatively similar resistant behaviors to those exhibited by

dropouts, but less frequently and in the process of actually approaching and dealing with

52

the psychological issues that brought them into treatment. Whereas the dropouts employ
these defenses more readily and frequently, with the goal being to maintain the status quo
by keeping the therapy from moving forward. The persisting patients are in essence
taking the previously mentioned “opportunity” of working through their resistance, while
the dropouts (consciously or unconsciously) are attempting to frustrate and “challenge”
their therapist and the process of psychotherapy. Second, it seems that the scales
(Opposing the Therapist and Opposing Recollection of Material) where greater
differential expression existed between dropouts and persisters may tap into factors that
have been both theoretically and empirically supported to be especially relevant to the
success of the early phase of psychotherapy (e. g., development of therapeutic alliance,
expectations for treatment, match between therapist style and client presentation, etc.).
Overall, the present study’s findings seem fairly congruent with Beutler et al.’s (2001)
report that most studies that investigated the prognostic value of patient resistance found

that it had a negative impact on treatment outcome.

Hyppflesis 1]]

The purpose of this hypothesis was to address within-subjects changes in level of
resistance between sessions, as well as, to analyze possible differential growth rates
between the dropout and persister groups. The Client Resistance Code (CRC) measured
growth in frequency of resistant responses, while the Client Resistance Scale measured
changes in intensity and quality of resistance between sessions. Contrary to expectations,
data analysis revealed very little difference in resistance growth rates between the

dropout and persister groups (See Figures 3 and 4). As expected drapouts did exhibit

53

higher initial levels of resistance and reached higher levels by session‘three on both
measures. Within-subjects resistance growth approached statistical significance, with
average levels of resistance on both measures increasing noticeably. This trend of
progressively increasing resistance levels during the early phase of treatment is consistent .
with both theoretical and empirical expectations.

Reviewed theoretical and practical writings on resistance (Greenson, 1967;
Teyber, 2000; Gabbard, 1990; Benjamin, 2003) suggest that intensification of resistance
is an expected and typical pattern of patients during the early phase of treatrrrent. From a
psychodynamic perspective, patients entering treatment progressively intensify their
resistance against approaching emotionally laden issues by employing a variety of
intrapsychic defense mechanisms. This intensification continues as the therapist moves
the patient toward “working through” the resistance. The origins of the resistance (and
transference) are explored and often addressed with repetitious interpretation during the
“working through” phase, until eventually insights and understanding are eventually
integrated into the patient’s conscious awareness. At this point, the level of resistance
will subside.

Chamberlain et al.’s (1984) study found this pattern in their initial application of
the CRC to parent-training therapy with adults with child management problems. They
reported significant within-subjects changes in overall frequency of resistant responses
between early treatment (average of first two sessions’) scores and mid-treatment
(average of seventh and eighth sessions’) scores. They further observed a progressive
decrease in resistance from mid-treatment to late treatrnent/tennination. Patton, et al.

(1997) also reported a low-high-low pattern of resistance, as measured by Schuller, et

54

al.’s Resistance Scale, amongst clients described as having better outcomes. Further, they
reported that clients with lower levels of rnidtreatment resistance showed greater change
from counseling.

In both of the previously cited studies, measurements of change in resistance level
were assessed further apart (on average more than five sessions apart) than those made in
the present study (two sessions apart). The added time and further progression of
treatment in these studies allowed for additional manifestation of resistance growth, thus
increasing their probability of reaching statistically significant levels of change. When
comparing the current study’s average per session change on the CRC with those
reported by Chamberlin et al. (1984), the slope of resistance growth was actually steeper
in the present study. Small sample size (relatively low statistical power) and a relatively
brief pre-lpost-measurement interval in the present study likely explains the lack of
statistically significant results. However, the overall trend observed amongst this sample
is consistent with the previous research. Not only was there an increase in frequency of
resistance, there was also increase in intensity of the resistance exhibited.

In pursuing additional understanding of the pattern of resistance growth, further
data analysis looking at the CRS subscales was conducted. This was to provide additional
information on the quality of the changes in resistance growth. Of the five subscales, only
the Opposing Recollection of Material scale produced statistically significant results.
Growth of this particular subset of resistant behavior was significant between sessions.
Initially the persister group exhibited slightly higher levels of resisting recollection of
material. However, persisters scored sinrilarly on this scale in both the first and third

sessions. Whereas in the third session, dropouts experienced a fairly dramatic increase

55

that outpaced the persister group. Dropouts thus ended up with higher scores on this scale
by the third session as mentioned in discussion of Hypothesis 11 (See Figure 5 for graphic
illustration). The substantial increase in resisting recollection of material by dropouts
represents a shift away from discussing self-focused material to focusing on others or
verbally shutting down altogether.

Supporting the present findings, Piper et al.’s ( 1999) study of 44 patients (22
dropouts and 22 matched completers) in time-limited individual psychotherapy found that
in the last session prior to dropping out, the patients resisted their therapists’ attempts to
focus on painful affect, opposed the therapist directly through arguing or becoming silent,
and resisted “dynamic exploration.” Oei and Kazmierczak’s (1997 ) study of dropout
from group cognitive behavioral therapy found that dropouts participated much less than
those who completed treatment. This lack of participation also seems similar to resisting
recollection of material, as it appeared to involve the patients engaging in less
verbalization and production of clinical material.

There would appear to be a fairly simple explanation for the relationship between
patients’ resisting recollection of material and dropout. From the time of Freud’s original
development of psychoanalysis and it’s being labeled the “talking cure” until present day,
free association and/or discussion of clinical material has been viewed as the central
mechanism of psychotherapy. Therefore, it would follow that those who refuse to engage
in such behavior or who do it in a less than therapeutic way (e. g., remaining focused on
external problems and/or refusing to take personal responsibility), may find participation
in treatment less than helpful and eventually dropout. By behaving in such a manner

these patients are undermining the curative process. Without a change in patient

56

behaviors, the therapist will not have the opportunity to address any other forms of
defense or resistance. This counterproductive stance is analogous to a person taking their
car to a mechanic for engine problems, but refusing to open the hood or put the car up on
a hydraulic lift.

There is a major difference between someone being initially guarded and slow to
pursue depth of clinical material and someone who either refuses to talk, responds
minimally with short utterances, or someone who turns attention away from themselves.
Over time the initially guarded person may begin approaching more threatening material,
while the person intensely resisting production of appropriate therapeutic material will
find little benefit or reinforcement for remaining in treatment. There are ways therapists
can either facilitate communication (e.g., active listening) or aid the pursuit of clinical
material (e. g., asking open ended questions). However, the patient is able to exert their
greatest degree of control over therapy by deciding how much or little they are going to
talk and which topics they are willing to pursue. Relevant to these issues, Beutler et al.’s
(2002) chapter on resistance related the social psychology theory of “psychological
reactance” to the resistant patient’s behavior in treatment. He reports that this theory
suggests the patient perceives a threat to their “legitimate freedom,” motivating them to
restore this thwarted freedom through oppositional behavior, noncompliance, and
rigidity. This facet of resistance as well as the relevance of this body of psychological

theory is deserving of further research and exploration.

57

Hymthesis IV

Building upon the findings of the previous hypotheses, Hypothesis IV’s purpose
was to attempt to develop an discriminant function using the CRC and CR8 resistance
variables that could then be used to predict group membership. This hypothesis was
further attempting to validate empirically the predictive utility of resistance theory, as it
relates to dropout from psychotherapy. Specifically, the notion that dropout may
represent an extreme form of resistance or “flight into health.” By identifying specific
variables that could be useful to clinicians hoping to identify patients at risk for early
therapy dropout, this hypothesis was also attempting to make a practical contribution to
the dropout literature.

The prior t-test and AN OVA comparisons of the dropout and persister groups
were able to speak to differences in quantity and quality of resistance exhibited in the
sample of clients used. However, these forms of statistical analysis were not able to
address the predictive utility of the two measures. Thus, this was the added benefit of
conducting a discriminant function analysis.

The prior hypotheses’ data analyses were used to inform the researcher in
establishing which variables to enter into the discriminant function. Following multiple
iterations of entering the resistance variables in a step-wise fashion into equations aimed
at predicting group membership (dropout vs. persister), a model emerged that employed
four variables successfully to predict group membership at an overall rate of 77.5%. The
four predictor variables were: 1) the combined first and third session CRC resistant
response proportion; 2) third session CRS Opposing Recollection of Material; 3) third

session CRS Opposing Expression of Painful Affect; and 4) combined first and third

58

session CRS Opposing the Therapist. Three out of the four variables had a positive
relationship to dropout, while third session CRS Opposing Expression of Painful Affect
had a negative relationship with dropout. These findings suggest that not only the
quantity, but the specific quality of resistance exhibited influences dropout and that some
forms of resistance may aid rather than hinder the therapeutic process.

The success of these variables to predict dropout versus continuation status was
not only statistically significant, but also has potential practical significance. The ability
to potentially identify growing resistance prior to its influencing a client to dropout of
therapy can help guide clinical interventions in a way that will potentially reduce overall
dropout rates. The results of the discriminant function analysis suggest the development
of clinical guidelines that address both the overall amount and types of resistant behavior
engaged in by the client will be most helpful. Specifically, at the outset of psychotherapy
therapists should emphasize the development of a collaborative working alliance,
wherein discussion of real life, personally relevant material is facilitated, and where the
expression of painful affect can be contained and supported in a somewhat controlled
fashion. As the early stage of psychotherapy proceeds, adjustments (informed by
relevant theory and empirical research) should be made on the part of the therapist if and
when, 1) the client begins to shut down and/or moves away from discussing, self-focused
clinical material; 2) the client refuses to follow the therapists direction, becomes
tangential, or actively blocks the therapist’s direction; or 3) the patient has difficulty
modulating and/or containing their expression of painful affect. By the predictive model
drawing on multiple dimensions of resistance (quantity, quality, intensity), it is able to

more specifically guide and direct clinicians to relevant practical, research and theoretical

59

literature for tailoring their approach in the clinical setting. More specific discussion of
thi 3 literature will be addressed later. ,

The present predictive model was better at correctly classifying persisters than
dropouts (85% vs. 70%, respectively). In looking at the impact of the associated enor
rates. misclassification of persisters as dropouts has less potential for negative
consequences than that of misclassifying dropouts as persisters. It is difficult to think of
what harm could be done should specific interventions aimed at reducing resistance be
overapplied early in treatment with patients who would persist in treatment anyway.
However, in pursuing clinical efficiency and congruence between intervention and client
presentation, such overapplication of technique or specialized intervention should be
avoided. Misclassification of dropouts into the persisting group was done 30% of the time
using this model. The obvious result of not detecting “at-risk” resistant clients is that they
end up dropping out of treatment. While this is the undesirable result that the current
study set out to address, it is unlikely that interventions, no matter how well developed
and tested, will ever fully prevent dropout. Even the best clinicians will make strategic
technical errors and in some cases the patient dropping out may truly believe they have
gotten their maximal benefit from treatment. Prior research that has focused on
dWIOgraphic variables in predicting dropout has failed to provide the mental health
PmVidcr with clinically relevant and changeable factors to target. Thus, in this case a
70% success rate of classifying dropouts using resistance variables still moves us toward
the goal of realistically reducing the rate of premature termination, by at least identifying

relevant theoretically-based process and client variables that are amenable to change

through clinical intervention.

practical Implications of the Findings

Despite the mixed results and sometimes lack of statistical significance this study
pt-¢)¢iuced, there appears to be support of previous findings in similar research that
suggests high client resistance levels will negatively impact treatment outcomes (Bender,
Mo] Ciro, & Talebi, 2002). This study was also able to develop a predictive model of
dropout using resistance variables. As such, treatment will be more effective, if therapists
can either avoid stimulating excessive levels of resistance within their patients or
effectively address the resistance in ways that temper and/or suppress the effects of the
resistance expressed (Beutler, Clarkin, Bongar, 2000). In order to be most clinically
effective, it is necessary for therapists to recognize, identify, and respond to different
types and styles of resistance.

The Client Resistance Scale (CRS) and Client Resistance Code (CRC) both
prOVide organized and systematic approaches to identifying and tracking both episodic
and tactical forms of resistance, allowing a clinician or researcher to assess the quantity
and quality of resistance being manifest. Though regular use of these measures would
likely prove cumbersome in clinical practice, clinicians (particularly trainees/students)
familiarizing themselves with them will likely find themselves more attune to resistance
in their patients aiding their awareness and ability to respond therapeutically. Beutler et
31. (2001) made recommendations for responding to state (episodic/tactical) resistance
ineluding, acknowledging and reflecting the patient’s concerns, directly discussing the
therapeutic relationship, and making modifications/adjustments to therapeutic goals.

These recommendations are aimed at returning a sense of control to the patient thus

61

defusing the effects of resistance.

Previous research (Beutler, et al., 2002; Piper, et al., 1999) suggests that it is also
important for clinicians to make adjustments to their therapeutic approach upon
evaluating their patient’s level of resistance in order to avoid impasses in treatment
and/or dropout. This is particularly true when dealing with patients who present with high
levels of trait or character resistance. Evidence suggests that such highly resistant patients
respond better to less directive interventions, while less resistant patients respond well to
higher levels of directiveness. In dealing with resistant patients, Beutler and Harwood
(2000) specifically recommend employing a treatment approach that de-emphasizes
therapist authority and that bolster patient self-direction. Therapists should do more
listening and avoid giving too many instructions. Rigid homework assignments are to be
avoided and assignments that require minimal overt behavior» are suggested to reduce
opportunities for oppositional behavior. Thus, self-directed assignments and reading
might replace more highly structured and specific tasks. It has further been suggested
that paradoxical interventions (e. g., discouraging rapid change, symptom prescriptions,
encouraging the exaggeration of symptoms) may have some benefit in dealing with some
resistant patients. Here the patient’s tendency toward violating directives and
oppositional impulses are enlisted to aid their movement toward change and/or symptom
improvement.

In his book on interpersonal processes in psychotherapy, Teyber (2000) provides
a number of suggestions for responding to resistance at differing stages of treatment. He
recommends inviting a new patient to express all of their concerns regarding entering

. treatment. He then encourages the therapist to explore these concerns in a nonjudgrnental

62

 

way that leaves the patient in control and responsible for making the decision regarding
continuation of treatment. As therapy proceeds, he further encourages therapists to
empathize with the patient’s resistant stance, “honoring (the origins of) their resistance”
and then giving them permission to discuss their reticence and doubts about entering
treatment openly and directly. Teyber also suggests educating the patient in a way that
will aid their having realistic expectations and help them understand the ambivalence
they are experiencing (for further details see Chapter 3 of Teyber’s (2000) text).
Similarly, Reis and Brown’s (1999) research also suggested that preparing patients prior
to therapy by providing them with information and addressing their expectations tenkd
to reduce dropout rates. There is additional empirical evidence for following Teyber’s
recommendations as Patton et al. (1997) found that therapist that adhered to
psychoanalytic techniques aimed at highlighting and exploring client resistance were
more successful at reducing the overall level of resistance exhibited across sessions. This
reduction in resistance was associated with better patient outcomes in treatment. As such,
it may be to the advantage of therapists of all theoretical orientations to read and
familiarize themselves with psychoanalytic theory as it relates to resistance (e.g.,
Greenson (1967)) and then incorporate these nondirective, interpretive approaches aimed
at ‘working through’ with their more resistant patients.

In an era of managed care and emphasis on short-term treatment, it is important
that training programs and currently practicing professionals not ignore resistance and the
theory and writing that informs this topic. In fact, the effective management of patient
resistance (with its empirically proven ability to negatively impact treatment) may be as

relevant as ever, particularly in light of the emphasis being placed on treatment efficacy

63

and efficiency. If we can continue to predict the majority of premature terminations and
develop appropriate interventions and responses, we will then have the potential to
reduce the impact of dropout on public health, as well as, reducing the related financial

and morale issues posed by dropout to mental health professionals.

Limitations of the Present Study

The design of the present study had some significant challenges, weaknesses and
limitations that ought to be considered. The archival nature of this study presented some
specific challenges in gathering data. One of the most difficult challenges was
establishing a large enough sample of demographically matched clients who had
complete SCL-90R data and whose therapy sessions were tape recorded with sufficient
audio quality as to render codeable samples. Despite extensive measures being utilized to
enhance sound quality, a small percentage of the therapy samples remained difficult to
hear clearly. In these cases, coders were encouraged to code the portions they could hear
adequately and/or attempt to make single ratings based on the entire sample rather than
individual speaking turns. Again, this was done with less than five percent of the
samples. However, such deviation from standard procedure may have contributed to
decreased reliability and accuracy in these cases. A related weakness is the fact that the
present study employed audiotaped session samples for its observations of client resistant
behavior. As this is a “single channel” approach to assessing qualitative aspects of
communication exchanges between therapists and clients, it is less than ideal. Social
psychologists have suggested that a large proportion of communication is nonverbal and

therefore, a sizable portion of potentially codeable resistant responses are being lost due

to the researchers having access only to the verbiage and tonation of communications
between the clients and therapists in this study.

The use of archival data established limits to the sample size that introduced a
number of additional weaknesses into the present study. When using a “convenience
sample,” researchers are not able to establish and maintain, as many controls as ideally
would exist in a pure experimental design. Subjects were assigned to naturally occurring
groups in this study. There was no random assignment or control group. Instead, a group
of treatment dropouts were matched with a similar group of persisting clients. Even if this
had been a prospective study, resistant traits/behaviors cannot be randomly assigned to
subjects. Also, there was limited data available on the therapists that treated these
subjects and other than their sex none of this data was taken into consideration. As such,
many unintended sources of systematic error and variance may have entered into the data
analysis. However, there seems to be a fair amount of consistency between this study’s
findings and previous research on the effects of resistance on psychotherapy dropout and
outcome.

An additional sampling issue was the impact that available data had on the
establishment of the criteria for dropout status. This study’s intent was to address early
dropout. It is debatable whether or not the present sample truly represents a group of
“early” dropouts. All of the subjects in the dropout group unilaterally discontinued
treatment and generally would meet most definitions of premature termination and/or
dropout. However, due to the desire to improve statistical power by achieving a minimum
sample size of forty subjects, the number of sessions attended prior to dropping out

exceeded five sessions amongst half of the dropout group. Previous research studies (e.g.,

65

Pekarik,1992) have provided some evidence of there being significant differences on a
number of factors between subjects that dropout early (less than three to five sessions)
versus those that drapout later. Stahler and Eisenman (1987) also suggested that-a few
sessions of therapy may benefit a subset of patients that are either in acute crisis or who
have better ego resources that allow them to benefit from the briefer exposure to
treatment. Some patients may also come into therapy never intending to pursue long-term
treatment. Therefore, closer evaluation of such factors may improve a study’s ability to
pinpoint differences between patients.

This issue notwithstanding, it was important to achieve a reasonable samplesize,
as without sufficient statistical power even moderate to large effects may be missed due
to Type 11 error. Small samples are also more vulnerable to the effects of outliers,
measurement error/reliability, and differences in variance between groups. Thus, the
4 compromise in number of sessions attended by dropouts was considered worthwhile in
the service of improving statistical power. Some additional exploratory data analysis
where the dropout group was divided into two groups (less than or greater than five
sessions attended) suggested that future research should pursue expanding sample size
and delineating between the stage at which dropouts terminate. In discussing their
findings, Martin, et al. (1988) also suggested that the assumption of uniformity or
homogeneity amongst premature terminators may not hold and that clients who dropout
of psychotherapy may be quite a diverse papulation with different motivations
influencing their decisions to leave treatment.

An additional area of weakness of this study was the use of only two therapy

sample coders and the resulting diminished reliability levels obtained. Client Resistance

Code (CRC) interrater reliabilities were high enough as to not consider them a significant
liability, while Client Resistance Scale (CRS) reliabilities were lower than desired,
particularly on the subscales. Mahalik (1994) had recommended the use of three or more
judges when using the CRS, as this was found to significantly enhance reliability across
the subscales. Unfortunately, due to the limited resources of the primary researcher, only
two judges were used in the present study. There were also significant time constraints
that led to less time being spent on training the judges than may have been optimal. More
mid-coding reliability checks would have been helpful as well, but again limited
availability on the part of the judges made this unfeasible. The reliabilities on the CRS
Opposing the Therapist and Opposing Change subscales are low enough that there is
concern regarding measurement error.

When interpreting and attempting to generalize the present study’s findings, one
must consider a number of potential confounds and limitations. A limitation to the
generalizability of the present study was its clinical setting, a training clinic that was
staffed by graduate student clinicians. Additionally, psychodynamic psychotherapy was
the most common form of therapy offered, but not the only kind offered. Future research
should attempt to address these limits of external validity by employing carefully
controlled combinations of therapists and settings (e. g., licensed psychologists in a
variety of treatment settings including college counseling centers, community mental
health centers, and outpatient hospital clinics). It would also be advantageous to be more
systematic about the inclusion of therapies, either restricting the research to one form of
therapy or being more inclusive and tracking the impact of type of therapy as an

additional variable of interest. The present sample was also somewhat homogenous with

67

 

regard to ethnic/racial representation, as only three minorities were included. Thus,
generalizations to minority psychotherapy clients must be made very cautiously.

A final concern is the fact that some therapists could have more than one client in
the study. This factor could be seen as increasing the possibility of chance findings. As
such, the successful t-test results and discriminant function need to be replicated and

cross-validated in future studies.

Directions for Future Research

As is frequently the casein psychological research, the present study’s findings
created as many or more questions than it answered. The present study’s limitations also
prompt the need for further validation and replication. An initial recommendation would
be to conduct similar research with a more diverse sample and employing a variety of
therapist with broader variance of training level and theoretical orientations.

Building upon the results and discussion of this study’s first hypothesis that
addressed initial pathology levels in the sample, research looking into differential patterns
of symptom response between dropouts and persisters would be highly recommended. Of
great interest would be how levels of psychopathology in dropouts change as they
approach the point where they exit treatment. Arguments have been made speculating
both increase and decrease in disturbance. It would be additionally important to not only
rely on patient self-reported symptomology, but to also incorporate therapist ratings of
pathology and dysfunction. This would help address concerns over the manifestation of

“flight into health” by highly resistant patients.

68

 

This study’s application of the Client Resistance Scale (CRS) (Mahalik, 1994)
and the Client Resistance Code (CRC) (Kavanagh, et al., 1982) to individual
psychotherapy samples collected in a natural setting was something that has not been
done but a relatively few times in the present psychological literature. Therefore, much
more research with larger and more representative samples is needed in order to
strengthen the external validity and generalizability of these measures. The positive
results from the present study combined with those reported by the authors of these two
measures, suggests these measures hold promise for broader application. Another issue
related to the use of the CRS and CRC is that these measures tend to emphasize the more
observable and transient forms or “tactical” subtypes of resistance. Further exploration of
unconscious and/or character resistances using measures such as Schuller et al.’s (1991)
Resistance Scale or additionally developed measures could provide additional empirical
insights and understanding of the effects of resistance on the treatment process and
premature termination. Also, future research that further elaborates the patterns and
qualities of resistance exhibited over a greater period of treatment would help guide the
development of more precise interventions for responding most appropriately.

The current study was limited by its strictly addressing client pathology and
resistance, as it related to dropout without looking at therapist and interactive variables
suggested by theorists and other studies that may also contribute to attrition (Bender, et
al. .2002; Bischoff & Tracey, 1995). It would be very helpful to exarrrine such therapist -
variables in tandem with client variables and then enter this data into a discriminant

function equation simultaneously. In theory such a multivariable, multimethod approach

69

 

should increase positive predictive power. Future research should incorporate measures
of both client and therapist behaviors to address this issue.

As specific recommendations for addressing resistance have been suggested both
in this study and theoretical writings (see Practical Implications section), it would be of
great value to empirically validate the efficacy and utility of these recommended
responses in reducing client resistance and dropout. Further development and testing of
additional new interventions that address the interaction between resistance and dropout,
perhaps, drawing further on social psychology and other subfields of psychology outside
of psychoanalytic and psychodynanric theory would also be welcome in expanding
breadth of understanding within this area of research. or additional interest, especially
within in the context of today’s biologically dominated mental health field, would be the
impact of psychotropic medications on resistance levels and whether this impact would
facilitate or hinder progress in psychotherapy.

Finally, arguably the most practical and useful results from the present study,
were those from the discrirrrinant function analysis that were successful at developing a
predictive model of dropout using resistance variables. In order for these results to be
generalizable and applied with any degree of confidence, this model must be tested by
applying the present study’s discriminant function to data from a new sample. Such

cross-validation is viewed as an essential step for furthering this area of research.

70

 

TABLES

71

l
u.

Table 1. Sample Demographic Descriptive Statistics and ANOVA’s.

E Mega
AG_E

Dropout 20 31.30
Persister 20 31.26
Total 40 31.28
EDUCATION (years)
Dropout 20 - 14.20

Persister 20 15.55

Total 40 ' 14.88
INCOME Q E year)

Dropout 20 10995.00
Persister 20 15610.53

Total 40 13243.59

S_D £110 2

11.06 .00 (1) .99 (ns)
6.41

8.98

2.24 387(1) .056 (ns)
2.09

2.24

6560.45 2.573 .117 (ns)
10973.24

9166.21

72

Table 2. Sample Demographic Descriptive Statistics:
Crosstabulations and Chi-square estimates.

OCCUPATIONAL LEVEL X GROUP ASSIGNMENT

Persister Totals
ve
business owner

worker
worker

0
0
1
2
6
1
5
3

CO‘Nr—mMNt—t

otals 18*

N
C

 

Pearson x2= 12.793, df=7, p=.077 (ns)
* 2 missing cases

MARITAL STATUS X GROUP ASSIGNMENT

married
with

vorced
1dowed
otals 20

 

Pearson x2 = 4.846, dr=4, p=.303 (ns)
*1 rrrissing case

73

Table 3. Sample Means, Standard Deviations, and Interrater Reliabilities:
Client Resistance Scale (CRS).

Client Resistance Code

EA RM I Q I Summed Scale

Means 5.19 3.97 3.82 4.75 4.79 90.14
Standard Deviations .75 .91 .86 .66 .96 7.76 ’
Pearson Product-

Moment Correlations .72 .76 .52 .51 .68 .71
Swan-Brown

Effective Reliability .83 .86 .68 .68 .81 .83

Mahalik’s refitted
Effective Reliabilities .88 .92 .84 .81 .71

Using two raters

 

All correlations significant at p<.05 (two-tailed)

EA: Opposing Expression of Painful Affect, RM=Opposing Recollection of Material, T=Opposing
Therapist, C=Opposing Change, and l=Opposing Insight

74

 

Table 4. Client Resistance Scale: Subscale Intercorrelation Matrix

 

Subscale 1 2 3 4 5
1. Expression of Affect - .344 .066 .448 .391
2. Recollection of Material - .407 .323 .434
3. Therapist - .332 .422
4. Change - .410
5. Insight -

 

All intercorrelations were significant at p<.01 (two-tailed).

75

Table 5. Sample Means, Standard Deviations, and Interrater Reliabilities:

Client Resistance Code (CRC)

Speaking
Turns/p_er session

Means 9.5
Standard Deviations 4.42
Pearson Product-
Moment Correlations -
Swan-Brown
Effective Reliability -

Client Resistance Code
Resistant % Coograti ve %

41.3 58.7

17.4 17.4

.78 .84

.88 .92

 

All correlations significant at p<.05 (two—tailed)
Intercorrelation between scales was r=.08 and non-significant

76

1 —_.¢

Table 6. Concurrent Validity Correlations:
Client Resistance Scale (CRS) and Client Resistance Code (CRC)

Client Resistance Code
1_3_A_ _R_M_ I Q I Summed Scale

Client Resistance Code 29* 39* 53* .09 .45* .46*
Resistance scale

 

EA= Opposing Expression of Painful Affect, RM=Opposing Recollection of Material, T=Opposing
Therapist, C=Opposing Change, and l=Opposing Insight

‘Correlation significant at p<.05 (two—tailed)

77

Table 7. Group Means, Standard Deviations, and t-test for Client Resistance Code
Resistant Resmnse Propprtions.

CRC Resistant
Resp_onse
Promrtion 1%)
Session 1 Dromut
Persister
Session 3 Dromut
Persister
Sessions Dropput
Combined
Persister

Mean

46.26

33.45

51.54

37.39

48.21

34.41

23.48

20.61

26.82

17.37

18.03

14.05

78

t p(2-tailed) 11

1.83 .075 .28
1.98 .055 .30
' 2.69 .010 .40

Table 8. Group Means, Standard Deviations, and t-test for Client Resistance Scale

Summed Scores.
CRS Summed Score
SessiOn 1 Dromut
Persister
Session 3 Dromut
Persister
Sessions Dromut
Combined
Persister

Mean

44.56

43.85 .

46.05

45.81

90.62

89.66

4.71

5.98

6.32

4.32

8.51

7.12

79

.419

.144

.387

.88

.70

.13

..05

.12

Table 9. Canonical Discriminant Function Coefficients

 

 

 

 

Standardized Canonical
Discriminant Function Coefficients
CRS: Opposing .943
Recollection of Material -

3" Session

ICRC-Resistant Response .958

Proportion 1“ and 3“1
sessions combined

 

 

CRS: Opposing the -.689
Therapist—l“ and 3"1

sessions combined

CRS:Opposing -.746

Expression of Painful
Affect—3'“ Session

 

 

 

 

8O

Table 10. Discriminant Classification Results

 

 

 

 

 

 

 

 

 

 

 

 

Predicted Group Total
Membership
GROUP Dropout Persister
Count dropout l4 6 20
persister 3 17 20
‘/o dropout 70 30 100
persister 15 85 100

 

81

 

 

Table 11. Pooled Within-Groups Correlations Between Variables and Canonical Values

 

 

 

 

 

 

 

Structure Matrix
Discriminant Function
CRC-R%1“+ 3“I .718
Recollection Material 3rd .382
Therapist 1st+ 3rd .301
Expression Affect 3rd -.213

 

 

82

FIGURES

83

 

Frgu_re l. CRS Subscale Means: Dromut vs. Persister Group (Session 1)

12.0

10.0

IDropout

.0
6 uPerslster

4.0

2.0

0.0

 

 

Scores reported represent the sum of both raters average rating for each scale. All t-tests failed to find
significant differences between dropout and persister groups across all subscales. .EA= Opposing
Expression of Painful Affect, RM=Opposing Recollection of Material, T=Opposing Therapist, C=Opposing
Change, and l=Opposing Insight

 

Egg 2. CRS Subscale Means: Drop_out vs. Persister Group (Session 3)

12.0

10.0

8.0

6.0 ~ 'D'W
uPersister

4.0

0.0

 

 

Scores reported represent the sum ofboth raters average rating for each scale. All t—tests failed to find
significant differences between dropout and persister groups across all subscales. EA= Opposing
Expression of Painful Affect, RM=Opposing Recollection of Material, T=Opposing Therapist, C=Opposing
Change. and l=Opposing Insight

85

 

Fr .Chan inCRSTtalM sBe eenScssionsland:
D_romut vs. Persisg Group

CR8 Total Score

 

Session 1 Session 3

86

 

Egg 4. Change in CRC Resistant Resppnse Prom‘on Means
Betwgn Sessigns l and 3: Dropput vs. Persister m

cncssnum-m
8828886:$88$2$88

 

Session 1 Session 3

87

 

Fr . Chan e in CRS sin Recollection of Material Means
gtween Sessions 1 and 3: Dromut vs. Persister Group

a a 8

V

CB? Oppoolng MW of Him-l

as

 

88

 

APPENDICES

89

APPENDIX A "‘

Client Resistance Scale
(Mahalik, 1994)

Opposing Expression of Painful Affect

Raters examine to what extent the emotions HURT—DEPRESSION (i.e., themes of
actual or perceived loss, feeling down, disappointed, or down on self), ANGER (i.e.,
themes of wanting to strike back or stand up to protect), and/or ANXIETY (i.e., themes
of concern or uneasiness over anticipated loss) are present.

1 = Reporting experiencing painful feelings (current or past) by using emotional
words with (a) uncontrollable or overwhelming sobbing (almost or cannot speak
through the tears), (b) shouting tirade, or (c) where anxiety is the dominant theme
and material causes panic.

2 = Reporting experiencing painful feelings (current or past) by using emotional
words with (a) lots of tears (talking while weeping), (b) shouting (short burst),
and/or (c) anxiety is the dominant theme of the client's material and client has
difficulty speaking because of anxiety.

3 = Reporting experiencing painful feelings (current or past) by using emotional
words with (a) a welling of tears or tears (talks through the tears), (b) raising
voice, and/or (c) where anxiety is the dominant theme of client's material.

4 = Reporting experiencing painful feelings (current or past) by using emotional
words (e.g., "I feel said," "I feel angry," "I feel scared" that indicate he or she is or
was feeling hurt/angry/anxious).

S = Indirect indications of hurt (e.g., discusses struggling through the day,
withdrawing, sounds hurt, depressed, or down without directly using those
words), anger (e. g., uses insults-"he was a jerk"-or profanity, sounds angry or
worried without directly using those words, uses sarcasm and/or has an edge to
his or her voice, expresses frustration, uses punctuated speech), or anxiety (e.g.,
lists worries or concerns, sounds anxious or worried without directly using those
words, has pressured speech, is starting to hear stress, is starting to hear
concern/anxiety, client material has themes of anxiety, fidgeting, restlessness).

6 = Intellectualizing, translating painful emotions into cognitions (e.g., "although
I may have had hurt/angry/anxious feelings after that happened, they were due to
the inevitable let down following my heightened expectations"); laughter that
covers painful affect.

7 = Other work than painful feelings, informational responses, not dealing with
subject of negative feelings, laughter, matter of factness to expression,
conversational.

 

Opposing Recollection of Material

1 = Client is the clear focus of the material and the material is a reporting of real
life events for the client. That is, client gives specific occurrences in daily life
with examples. These examples are made up of identifiable people, specific
actions, and descriptions of specific events in a verbatim fashion.

2: Client rs the clear focus of the material and majority of material rs detailed.
Examples are used, but they are not an accounting of real life events.

3 = Client is the clear focus of the material and describes material in both vague
and detailed terms. The client may use an example (or examples), but it is not an
accounting of real life events.

4 = Client is the clear focus of the material, but client material is not a report of
real life events. The client recollects material in a vague, general way, void of
details, and without examples. There must be some material present.

5 = Material is half self-focused and half other-focused, or client talks about self
entirely in the third person.

6 = Client material is mostly or totally other focused. Other client responses that
are minimal or simply a closed response to counselor's closed question (i .e., "yes"
or "no," or gives a long-worded yes or no response, or responds to a closed
question).

7 = Client does not provide any material. Client says, "I don't know," says "I don't
remember," is silent, or provides material that is completely tangential so that the
material is totally nonresponsive (e.g., word salad, responding with weather).

Opposing Therapist

1= Follows therapist' s direction and deepens the material significantly (i ..,e as
client talks material becomes much more emotionally meaningful/impactful to the
client).

2 = Follows therapist's direction and elaborates a great deal (i.e., gives a great deal
more information), and/or client deepens material somewhat (i.e., as client talks
material becomes more emotionally meaningful/impactful to the client).

3 = Follows therapist's direction and elaborates material (i.e., gives some more
information) but does not deepen material.

4 = Follows therapist's direction but does not elaborate. This occurs most often
when the client simply gives the minimal amount of material that answers a
question or replies no further than the therapist's lead.

5 = Client responds to the therapist's direction by following direction for half of
the client material, then going in a direction different from the therapist for the
other half of the client material.

6 = Tangential to therapist‘s direction (e.g., acknowledging topic, then proceeding
in totally different direction).

7 = Actively blocking therapist's direction.

91

Opposing Change

1 = Appears to be very uncomfortable with the status quo and expresses desire to
change.

2 = Appears to be uncomfortable or dissatisfied with the status quo and expresses
desire to change.

3 = Appears to be slightly uncomfortable or slightly dissatisfied with the status
quo of life circumstances and expresses desire to change.

4 = Appears to be uncomfortable/very uncomfortable or dissatisfied/very
dissatisfied with status quo but does not directly express a desire to change.

5 = Appears to be slightly uncomfortable or slightly dissatisfied with status quo
but does not directly express a desire to change.

6 = Appears to be comfortable or satisfied with the status quo of life
circumstances and is characterized by neutral statements or by those not reflecting
satisfaction or dissatisfaction with the status quo; gives informational responses.
7 = Expresses opposition to change (e.g., "It's not my problem, he's got to be the
one to change," "I like myself just fine the way I am," or "I don't want to be here,
I was sent here").

Opposing Insight

1 = Verbalizing understanding and/or making connections between experiences,
feelings, thoughts, and behaviors. In addition, the client provides older historical
references to illustrate the understanding.

2 = Verbalizing understanding and/or making connections between experiences,
feelings, droughts, and behaviors. In addition, the client introduces recent
historical references.

3 = Verbalizing understanding and/or making connections between experiences,
feelings, thoughts, behaviors.

4 = Agreeing with counselor's insights; may add new material but does not add
new connections.

5 = Expresses willingness or desire to understand. This may be implied by
phrases such as "and I don't know why!" The client must be pursuing
understanding not just complaining.

6 = Client does not verbalize any self-understanding in material. May be reporting
material in matter-of-fact way or discussing material in such a way that does not
indicate attempt to gain insight.

7 = In response to counselor or material brought up by client, the client actively
opposes self-understanding (e.g., "I don't even want to think about it"; "I don't
have a problem"; "The problem is out there, not in me").

92

 

Resistant muses
l. Interrupt/talkover

. 2. Negative attitude

3. Challenge/confront

4. Own agenda

5. Not tracking

Coomrative rcsmnses

6. Nonresistant

7. Facilitative

APPENDIX B

Client Resistance Code Categories

(Kavanagh, et al., 1984)

Coded only when the client is obviously cutting the
therapist off or talking over the therapist.

Responses indicating unwillingness/inability to cooperate
with therapist’s suggestions (e.g., blaming others,
statements of hopelessness, defeat, disagreement)
Responses challenging the therapist’s qualifications and/or
experience; responses that indicate that the therapist
doesn’t know what s/he is doing.

Bringing up new topics/concerns to avoid discussing or to
block the issue(s) that the therapist was on.

Inattention, not responding, disqualifying a previous
statement.

All responses that are neutral, cooperative, or following the
directions set by the therapist.

Short utterances indicating attention or agreement.

93

 

APPENDIX C ""

Symptom Checklist 90—Revised
Derogatis, et al. (1983)

Instructions: Below is a list of problems and complaints that people sometimes have. Please read
each one carefully. After you have done so, please rate the item according to HOW MUCH
DISCOMFORT THAT PROBLEM HAS CAUSED YOU DURING THE PAST WEEK
INCLUDING TODAY, according to the following scale: O-NOT AT ALL; l-A LITTLE BIT; 2-
MODERATELY; 3-QUITE A BIT; 4-EXTREMELY. Use only one number to rate each item.

Headaches

Nervousness or shakiness inside

Repeated unpleasant thoughts that won’t leave your mind
Faintness or dizziness

Loss of sexual interest or pleasure

Feeling critical of others

The idea that someone else can control your thoughts
Feeling others are to blame for most of your troubles
Trouble remembering things

10. Worried about sloppiness or carelessness

11. Feeling easily annoyed or irritated

12. Pains in heart or chest

13. Feeling afraid in open spaces or on the streets

14. Feeling low in energy or slowed down

15. Thoughts of ending your life

16. Hearing voices that other people do not hear

l7. Trembling

18. Feeling that most people cannot be trusted

19. Poor appetite

20. Crying easily

21. Feeling shy or uneasy with the opposite sex

22. Feelings of being trapped or caught

23. Suddenly scared for no reason

24. Temper outbursts that you could not control

25. Feeling afraid to go out of your house alone

26. Blaming yourself for things

27. Pains in lower back

28. Feeling blocked in getting things done

29. Feeling lonely

30. Feeling blue

31. Worrying too much about things

32. Feeling no interest in things

33. Feeling fearful

34. Your feelings being easily hurt

35. Other people being aware of your private thoughts
36. Feeling others do not understand you or are unsympathetic
37. Feeling that people are unfriendly or dislike you

38. Having to do things very slowly to insure correctness
39. Heart pounding or racing

)0?°>’.°‘.V‘PPN:"

94

40.
41.
42.
43.
. Trouble falling asleep
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
. Awaking in the early morning
65.
. Sleep that is restless or disturbed
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
. Having thoughts about sex that bother you a lot
85.
86.
87.
88.
89.
. The idea that something is wrong with your mind

Nausea or upset stomach
Feeling inferior to others
Soreness of your muscles
Feeling that you are watched or talked about by others

Having to check and double check what you do

Difliculty making decisions

Feeling afraid to travel on buses, subways or trains

Trouble getting your breath

Hot or cold spells

Having to avoid certain things, places, or activities because they frighten you
Your mind going blank

Numbness or tingling in parts of your body

A lump in your throat

Feeling hopeless about the future

Trouble concentrating

Feeling weak in parts of your body

Feeling tense or keyed up

Heavy feelings in your arms or legs

Thoughts of death or dying

Overeating

Feeling uneasy when people are watching or talking about you
Having thoughts that are not your own

Having urges to beat, injure, or harm someone

Having to repeat the same actions such as touching, counting, or washing

Having urges to break or smash things

Having ideas or beliefs that others do no share

Feeling very self-conscious with others

Feeling uneasy in crowds, such as shopping or at a movie
Feeling everything is an effort

Spells of terror or panic

Feeling uncomfortable about eating or drinking in public
Getting into frequent arguments

Feeling nervous when you are left alone

Others not giving you proper credit for your achievements
Feeling lonely even when you are with people

Feeling so restless you couldn’t sit still

Feelings of worthlessness

The feeling that something bad is going to happen to you
Shouting or throwing things

Feeling afraid you will faint in public

Feeling that people will take advantage of you if you let them

The idea that you should be punished for your sins
Thoughts and images of a fiightening nature

The idea that something serious is wrong with your body
Never feeling close to another person

Feelings of guilt

95

 

LIST OF REFERENCES

 

LIST OF REFERENCES ._

Beckham, E. (1992). Predicting patient dropout in psychotherapy. Psychotherapy,
22, 177-182.

Benjamin, LS. (2003). Integpgrsonal reConstructive therapy: Promoting change in
nonresmnders. New York : The Guilford Press.

Berrigan, L.P., & Garfield, S.L. (1981). Relationship of missed psychotherapy
appointments to premature termination and social class. British Journal of Clinical
Psychology, 20, 239-242.

Bender, L.E., Clarkin, J .P.,‘ & Bongar, B. (2000). Guidelines for the systematic
treatment of the depressed patient. New York: Oxford University Press.

Beutler, L.E., & Harwood, MT. (2000). Prescriptive therapy: A practical gg'de
to systematic treatment selection. New York: Oxford University Press.

Beutler, L.E., Rocco, F., Moleiro, C.M., & Talebi, H. (2002). Resistance. In J.C.

Norcross (Ed), Psychotherapy Relationshifi that work: Therist contributions and
resmnsiveness to patients. (pp. 129-143). London: Oxford University Press.

Beutler, L.E., Rocco, F., Moleiro, C.M., & Talebi, H. (2001). Resistance.
Psychotherapy, 38(4), 431-436.

Bischoff, M.M. & Tracey, T.J.G. (1995). Client resistance as predicted by
therapist behavior: A study of sequential dependence. Journal of Counseling Psychology,
QM), 487 -495.

Bottari, M.A. & Rappaport, H. (1983). The relationship of patient an therapist-
reported experiences of the initial session to outcome: An initial investigation.

Psychotherapy: Theogy, Research, and Practice, 20(3), 355-358.

Brogan, M.M., Prochaska, J.O., & Prochaska, J.M. (1999). Predicting termination
and continuation status in psychotherapy using the transtheoretical model.
Psychotherapy, 36(2), 105-113.

 

Cartwright, R., Lloyd, 3., & Wicklund, J. (1980). Identifying early dropouts from
psychotherapy. Ps chothera : Theo Research and Practice 17 3 263-267.

Chamberlain, P., Patterson, 6., Reid, 1., Kavanagh, K., & Forgatch, M. (1984).
Observation of client resistance. Behavior Therapy, ,1_§, 144-155.

97

Costa, P.T., & McCrae, RR. (1992). Revised NEO Personality Inventory (EEO
PI-R and NBC Five-Factor Invento O- : Professional manual. Odessa, FL:

Psychological Assessment Resources.

Derogatis, L. (1983). SCL-90-R administration, scoring, and medures manual-
I_I; Towson: Clinical Psychometric Research.

Derogatis, L., Rickels, K. & Rock, A. (1976). The SCL-90-R and the MMPI: A
step in the validation of a new self-report scale. British Journal of Psychiagy, m, 280-
289.

Dubrin, J. & Zastowny, T. (1988). Predicting early attrition from psychotherapy:
An analysis of a large private-practice cohort. Psychothegpy,25(3), 393408.

Freud, S. (1920). Introductog Lectures on Psychoanalysis. New York: Liveright
Publishing Corporation.

Gabbard, 6.0. (1990). Psychodmamic Psychiag in Clinical Practice.
Washington, DC : American Psychiatric Press, Inc.

Greenson, R. (1967). The Technique and Practice of Psychoanalfiis, Volume 1.
New York: International Universities Press, Inc.

Hatchett, G.T., Han, K., & Cooker, RC. (2002). Predicting premature
termination from counseling using the Butcher Treatment Planning Inventory.
Assessment 9(2), 156-163.

 

Hill, C.E., Corbett, M.M., Kanitz, B., Rios, P., Lightsey, R., & Gomez, M. (1992).
Client behavior in counseling and therapy sessions: Development of a pantheoretical
measure. Journal of Counseling Psychology, Q, 539-549.

Hjordis, B., & Gunnar, E. (1989). Characteristics of drop-outs from a long-term

behavioral treatment program for obesity. International Journal of Eating Disorders, 8,
363-368.

Howard, K.L., Kopta, 8., Krause, M., Orlinsky, D. (1986). The dose effect
relationship in psychotherapy. American Psychologist, 41, 159-164.

Kavanagh, K., Gabrielson, P., & Chamberlain, P. (1982). Manual for Coding
Client Resistance. Unpublished instrument, Eugene, Oregon.

Kolb, D.L., Beutler, L.E., Davis, C.S., Crago, M., & Shanfield, SB. (1985).
Patient and therapy process variables relating to dropout and change in psychotherapy.

Psychotherapy, 22(4), 702-710.

98

 

Levinson, P., McMurray, L., Podell, P., & Weiner, H. (1978). Causes for
premature interuption of psychotherapy by private practice patients. American Journal of
Psychiafl. .13—5. 826-829.

Mahalik, J .R. (1994). Development of the client resistance scale. Journal of
Counseling Psychology, 41(1), 58-68.

Mahalik, JR. (1994). Manual for the Client Resistance Scale. Unpublished
Instrument, Boston College, Boston, Massachusetts.

Mallinckrodt, B. (1993). Session impact, working alliance, and treatment
outcome in brief counseling. Journal of Counseling Psychology, 40, 25-32.

Martin, G., McNair, D, & Hight, W. (1988). Contributing factors to early
premature termination at a college counseling center. Journal of Counseling and
Development, 66, 233-236.

McCallum, M., Piper, W.E., & Joyce, AS. (1992). Dropping out from short-term
group therapy. Psychotherapy, 29(2), 206-214.

Mennick S., Lent, R., & Burgoyne, K. (1988). Premature termination from

university counseling centers: A review. Journal of Counseling and Development, 66,
458-465.

Mosher-Ashley, RM. (1994). Therapy termination and persistence patterns of
elderly clients in a community mental health center. The Gerontologist, 34(2), 180-189.

Mueller, M., & Pekarik, G. (2000). Treatment duration prediction: Client
accuracy and its relationship to dropout, outcome, and satisfaction. Psychotherapy,
3_7(2), 117-123.

Norman, G.R., & Streiner, D.L. (1999). FE Statistics, Second Edition.
Hamilton, Ontario: B.C. Decker, Inc.

Oei, T. & Kazmierczak, T. (1997). Factors associated with dropout in group
cognitive behaviour therapy for mood disorders. Behavioral Research and Theory,
3_5_(l 1), 1025-1030.

Ogles, B.M., Lambert, M.J., & Fields, S.A. (2002). Essentials of Outcome
Assessment. New York, New York: John Wiley & Sons, Inc.

Patton, M.J., Kivlighan, B.M., & Multon, K.D. (1997). The Missouri
psychoanalytic counseling research project relation of changes in counseling process to
client outcomes. Journal of Counseling Psychology, 43, 189-208.

Pekarik, G. (1992). Posttreatrnent adjustment of clients who drop out early vs.
late in treatment. Journal of Clinical Psychology, 48(3), 379-387.

Piper, W., Azim, H., Joyce, A., McCallum, M., Nixon, G., & Sega], P. (1991).
Quality of object relations versus interpersonal functioning as predictors of therapeutic
alliance and psychotherapy outcome. Journal of Nervous and Mental Disease 1E 432-
438.

 

Piper, W., Joyce, A., Azim, H., Rosie, J. (1994). Patient characteristics and
success in day treatment. Journal of Nervous and Mental Disease 1%, 381-386.

 

Piper, W., Joyce, A., Rosie, J ., Ogrodniczuk, J., McCallum, M., & O’Kelly, J.
(1999). Prediction of dropping out in time-limited. interpretive individual psychotherapy.
Psychotherapy, 36(2), 114-122.

‘ Regier, D, Narrow, W., Rae, D., Manderscheid, R., Locke, B. Goodwin, F.
(1993). The de facto U.S. mental and addictive disorders service system: Epidemiologic
Catchment Area prospective l-year prevalence rates of disorders and services. Archives

of General Psychiat_ry, fl, 85-94.

Reis, B.F., & Brown, LG. (1999). Reducing psychotherapy dropouts:
Maximizing perspective convergence in the psychotherapy dyad. Psychothergpy, 36,
123-136. .

Richmond, R. (1992). Discriminating variables among psychotherapy dropouts
from a psychological training clinic. Professional Psychology: Research and Practice,
23(2), 123-130.

Rosenthal, R., & Rosnow, R.L. (1991). Essentials of Behavioral Research:
Methods and Data Analysis, Second Edition. New York, New York: McGraw-Hill, Inc.

Samstag, L., Batchelder, S., Muran, J .C., Safran, J.D., & Winston, A. (1998).
Early identification of treatment failures in short-term psychotherapy. An assessment of
therapeutic alliance and interpersonal behavior. Journal of Psychotherapy: Research and

Practice, 7(2), 126-143.

Schuller, R., Grits-Christoph, P. & Connolly, MB. (1991). The resistance scale:
background and psychometric properties. Psychoanalm'c Psychology, 8(2), 195-211.

Seeman, M.V. (1974). Patients who abandon psychotherapy. Why and when.
Archives of General Psychiapg, 30, 486-491.

Smith, T.E., Koenigsberg, H.W., Yeomans, F.E., Clarkin, J .F., & Selzer, M.A.
(1995). Predictors of dropout in psychodynamic psychotherapy of borderline personality

disorder. Journal of Psychotherapy Practice and Research, 4, 205-213.

100'

Stahler, G.J. & Eisenman, R. (1987). Psychotherapy dr0pouts: Do they have poor
psychological adjustment? Bulletin of the Psychonomic Society, 25(3), 198-200.

Strean, HS. (1985). Resolving Resistances in Psychotherapy. New York: John
Wiley& Sons, Inc.

Sue, 8., McKinney, H., & Allen, D. (1976). Predictors of the duration of therapy
for clients in the community mental health system. Community Mental Health Journal,
12, 365-375.

Tabachnick, B.G. & Fidel], LS. (2001). Using Multivariate Statistics, 4"I ed.
Needham Heights, Massachusetts : Allyn and Bacon.

Teyber, E. (2000). Intemrsonal mess in psychothegpy: A relational approach
(4th ed.). Chicago, Illinois : Wadsworth Publishing.

Tracey, T]. (1986). Interactional correlates of premature termination. Journal of
Consulting and Clinical Psychology, _6_, 784-788.

VanDenberg, T.F., & VanDeberg, E]. (1992). Premature termination in the

midst of psychotherapy: Three psychoanalytic perspectives. Psychothegpy, 22(2), 183-
190.

Wierzbicki, M., & Pekarik, G. (1993). A meta-analysis of psychotherapy
dropout. Professional Psychology: Research and Practice, 24, 190-195.

Zois, C. (1986). Handling resistance in depressed patients. International Journal of
Intensive Short-term Dmamic Psychotherapy, 1, 17-30.

101

    

  

   

      

\l‘xijiijjl

504

nihiihtill'u