PROCESS AND OUTCOME IN COGNITIVE THERAPY

In recent decades, the efficacy of psychotherapy and the process by which psychotherapy works
have been increasingly scrutinized. Outcome studies have become more specific with regard to
the disorder treated and the types of treatments used. With the manualization of cognitive
therapy for depression (Beck, Rush, Shaw, & Emery, 1979), it has become one of the most
widely researched of the psychotherapies. This chapter will describe the current status of process
and outcome research in cognitive therapy. There will first be a brief review of the current state
of outcome research for psychotherapy in general followed by a more detailed examination of
outcome research in cognitive therapy. We will then examine the process research pertaining to
cognitive therapy. Reviews of outcome studies will focus most heavily on clinically diagnosed
disorders with less emphasis on analog studies.

Several reviews have found psychotherapy in general to be more helpful than no treatment
(Lubarsky, Singer, & Lubarsky, 1975; Smith, Glass, and Miller, 1980). However, there has
generally been a lack of hard evidence that one treatment is superior to another (Lubarsky et al.,
1975), especially over the long term (Frank, 1979). In reviewing the literature, Frank ( 1979)
observed that there may be a short -term superiority of behavior therapy for phobias and
cognitive therapy for the relief of depression. In a meta-analytic review, Smith et al., ( 1980)
surveyed 475 studies and found that the average outcome score for all treated groups was .85
standard deviations more improved than the mean of all control groups. The effect size of
cognitive therapies was considerably higher than any of the other 17 types of therapy included in
the meta-analysis (Smith et al., 1980, p. 89).

In a meta-analytic review of the efficacy of cognitive behavior therapies (CBT), Miller and
Berman (1983) analyzed 48 studies in which depression, anxiety, and psychophysiological
disorders had been treated. Statistically, the average patient treated with CBT had an
outcome .83 standard deviations more improved than the average untreated patient. CBT patients
did not have significantly better outcomes than patients treated with systematic desensitization,
but the lack of statistical significance may have been due to the small number of comparisons (13
studies) and to the fact that the patient disorders in the studies surveyed included anxiety as well
as depression. A significant difference favoring CBT over other therapies was found only on
self-report measures. When effectiveness by diagnosis was examined, the effect size of CBT
compared to no-treatment controls was numerically larger for depression than for treatment of
anxiety and psychophysiological disorders.

Outcome studies of cognitive therapy have varied in the type of cognitive therapy tested (Beck's
cognitive therapy of depression, Meichenbaum's stress innoculation training, Ellis's rational-
emotive therapy), their target populations, extent of therapist training, and general
methodological rigor. Most have used depressed patients, and these studies will be given the
greatest amount of attention in this chapter.

OUTCOME STUDIES OF COGNITIVE THERAPY OF DEPRESSION

Only a little over a decade ago, Shaw (1977), Taylor and Marshall (1977), and Rush, Beck,
Kovacs, and Hollon (1977) provided the first empirical evidence that cognitive therapy was
effective in treating depression. One methodological problem of these early studies was their
lack of clearly defined diagnostic categories. More recent studies have tended to use inclusion
criteria based on diagnoses by the Feighner criteria (Feighner et al., 1972), the Research
Diagnostic Criteria (Spitzer, Endicott, & Robins, 1978) or the DSM-III (American Psychiatric
Association, 1980). Generally, the required diagnosis has been major depression.

Another trend starting with Rush et al. ( 1977) has been to test cognitive therapy within medical
settings and to compare it with medication as a reference condition. The advantage of comparing
it with tricyclic antidepressants, especially imipramine and amitriptyline, has been that they are
well accepted within the medical community as effective treatments for depression. Findings of
comparable or ~uperior efficacy therefore command attention from a wide array of researchers
and mental h~;dth professionals. Rush et al. (1977) compared cognitive therapy with imipramine
and found the outcome with cognitive therapy to be superior at termination.

Several studies followed Rush et al. (1977) in this vein and found cognitive therapy either to be
more effective than tricyclic medication (Blackburn, Bishop, Glen, Whalley, & Christie, 1981,
with general practice outpatients; Teasdale, Fennel, Hibbert, & Amies, 1983) or equally as
effective (Beck, Hollon, Young, Bedrosian, & Budenz, 1985; Blackburn et al., 1981 with
psychiatric outpatients; Elkin, Shea, Watkins, & Collins, 1986; Evans et al., 1985; Murphy,
Simons, Wetzel, & Lustman, 1984). Although, on the whole, these results have been very
supportive of the effectiveness of cognitive therapy for depression, medication groups do not
provide a totally adequate substitute for control groups. There is, of course, a good reason for the
use of medication groups because the use of control groups with severely depressed patients
raises serious ethical problems. The NIMH Treatment of Depression Collaborative Research
Program (TDCRP; Elkin, Parloff, Hadley, & Autrey, 1985) attempted to deal with this problem
by using a pill placebo plus clinical management (minimal supportive therapy) condition. In the
seven studies where control groups have been used, each has found cognitive therapy to be more
effective (Comas-Diaz, 1981; Larcombe & Wilson, 1984; Reynolds & Coats, 1986; Ross &
Scott, 1985; Shaw, 1977; Taylor & Marshall, 1977; Wilson, Goldin, & Charbonneau-Powis,
1983). Reporting on the TDCRP results, Elkin et al. (1986) found numerical but not statistically
significant superiority for cognitive therapy over the pillplacebo treatment at termination.

Behavioral treatments for depression typically have included assertiveness training and/or
training subjects to increase participation in pleasant events. Most direct comparisons of
cognitive therapy with these forms of behavior therapies have failed to find a difference in
efficacy (Comas-Diaz, 1981; Gallagher & Thompson, 1982; McNamara & Horan, 1986; Rehm,
Kaslow, & Rabin, 1987; Taylor & Marshall, 1977; Wilson et al., 1983; Zeiss, Lewinsohn, &
Munoz, 1979). Only Shaw (1977) has found cognitive therapy to be superior to behavior therapy.
Reynolds and Coats ( 1986) made the first comparison between cognitive therapy and systematic
desensitization for depression. Again they failed to find any significant difference in efficacy.
Two studies (de Jong, Treiber, & Henrich, 1986; Taylor & Marshall, 1977) have reported that a
combined cognitive behavioral treatment was more effective than a purely cognitive treatment.
However, Taylor and Marshall also found it to be superior to a purely behavioral intervention.

The NIMH Treatment of Depression Collaborative Research Program (Elkin et al., 1985)
remedied many of the weaknesses of earlier CBT studies (Williams, 1984 ). Cognitive therapy
was compared with imipramine, interpersonal psychotherapy (IPT; Klerman, Weissman,
Rounsaville, & Chevron, 1984) and pill placebo plus clinical management. Patients receiving a
placebo had time to talk with their pharmacotherapist about how they were feeling and to receive
support (Imber et al., 1986). At termination, there were no statistically significant differences
between CBT and the other treatments. CBT had an outcome on depressive symptoms slightly
better than placebo plus supportive therapy and slightly worse than either IPT or imipramine.
CBT patients were actually worse symptomatically than placebo patients at the 4- and 8-week
evaluations (Watkins, Leber, Imber, & Collins, 1986). This, of course, raises the question of why
cognitive therapy was not more effective. This may perhaps be answered as process research on
the TDCRP data archives is pursued.

Two groups of investigators have studied the effectiveness of cognitive therapy for geriatric
depression. Thus far, it has been found to be of approximately equal efficacy with
psychodynamic group therapy (Steuer et at., 1984) and behavioral and relational/insight
therapies (Gallagher & Thompson, 1982). Steuer et at. found it to be somewhat superior to the
psychodynamic therapy in improving scores on cognitive items on the BDI, and Gallagher and
Thompson found it to be superior to the relational/insight therapy in its effect on depressive
symptoms at 1-year follow-up.

Two studies have compared the effectiveness of group versus individually administered CBT.
Rush and Watkins ( 1981), in a pilot study contrasted individual cognitive therapy, group
cognitive therapy, and individual cognitive therapy plus antidepressant medication. They found
preliminary evidence that group CBT may not be as effective as individually administered CBT.
There was no control group to compare with. Ross and Scott (1985) also compared group and
individually administered CBT; however, they found no differences in

effectiveness.

Most studies have used patients free of complications such as major physical illnesses or alcohol
abuse. The treatment of depression, however, must often proceed in the context of just such
complicating factors, and demonstrating efficacy for a treatment eventually means testing its
utility with a wide variety of patients. Larcombe and Wilson (1984) found CBT to be superior to
a waiting list in a sample of depressed patients with multiple sclerosis. Tan and Bruini (1986)
found no superior effectiveness for group CBT over group supportive therapy or a wait-list
control for treating depression in epileptics. However, it should be noted that depression was not
the only focus of treatment in the Tan and Bruini study and that the average initial BDI score for
the CBT group was less than 12. Only one study could be found that specifically addressed the
issue of treating depression in alcoholics. Turner and Wehl (1984) found individual cognitive-
behavioral therapy to be significantly more effective for treating depression in inpatient
alcoholics than standard inpatient treatment.

The previously mentioned results regard efficacy of cognitive behavioral therapy at the time of
termination from treatment. An equally important question regards whether patients treated by
CBT are more immune to relapse than patients treated by medication or other psychotherapies.
Cognitive theory predicts that individuals treated with cognitive therapy will be less likely to
hold to dysfunctional assumptions and will possess more skills for coping with stress and
depression than persons treated with pharmacotherapy or supportive therapy. For that reason,
cognitive-therapy subjects would be predicted to have less depression during a follow-up period.
Kovacs, Rush, Beck, and Hollon (1981) reported that patients treated with cognitive therapy in
Rush et al. (1977) had lower levels of depression at 1-year follow-up than patients treated by
pharmacotherapy. However, there was no difference in rates of relapse between the two groups.
In the study of depressed elderly by Gallagher and Thompson (1982), patients treated either by
cognitive or behavioral therapy maintained gains 1-year posttreatment better than patients treated
by relational/insight therapy. Blackburn, Eunson, and Bishop (1986) reported a 2-year
naturalistic follow-up of the earlier Blackburn et al. (1981) study. Patients treated by cognitive
therapy or a combination of cognitive therapy and pharmacotherapy were less likely to relapse
than patients treated by pharmacotherapy alone. There was no significant difference between
groups, however, in average levels of symptomatology at the 6-month follow-up. The same
pattern also held in the results of Simons, Murphy, Levine, and Wetzel (1986). Patients who
received cognitive therapy, either alone or in combination with pharmacotherapy, were less
likely to relapse than those persons receiving pharmacotherapy alone. Finally, Evans et al.
(1985) compared follow-up results for CBT, CBT plus imipramine, imipramine alone without
maintenance dosages, and imipramine with maintenance dosages. The treatment consisting of
imipramine alone without maintenance resulted in higher relapse rates than the other three
treatments.

In summary, the effectiveness of cognitive therapy for depression is supported by a number of

studies. However, the evidence that it is more effective than other active treatments during the
acute treatment phase is not conclusive. There is promising evidence that it is superior to
pharmacotherapy and nonbehavioral psychotherapies in preventing relapse.

ANXIETY DISORDERS
Much of the literature on cognitive interventions with anxiety disorders has used subclinical
conditions or small numbers of subjects. For that reason, only selected studies will be reviewed
here. Treatment of phobias-especially agoraphobia-has generally received the most attention.
Emmelkamp and his colleagues have conducted a series of studies comparing cognitive
interventions and in vivo exposure. Emmelkamp, Kuipers, and Eggeraat (1978) compared
cognitive restructuring and prolonged in vivo exposure for agoraphobia and found improvement
in the exposure condition to be clearly superior. The cognitive intervention consisted of: (a)
learning to relabel situations and to replace irrational with rational thoughts; (b) explanation and
discussion of Ellis's irrational beliefs; and (c) using coping self-statements. It apparently did not
involve systematic questioning by therapists of the evidence and logic behind fear-producing
thoughts, and it is not clear from the published reports that the therapists had any training in
cognitive therapy. Therefore, it is not too surprising that the combined condition was not more
effective than behavior therapy alone. In a further comparison of cognitive restructuring, in vivo
exposure, and combined cognitive behavioral treatment (Emmelkamp and Mersch, 1982), the
behavioral and combined treatment conditions produced greater improvement at termination.
However, cognitive therapy patients improved during follow-up, and the behavioral patients
relapsed so that the two groups were equivalent at follow-up, and neither group was doing as
well as the combined treatment group. Emmelkamp, Brilman, Kuiper, and Mersch (1986)
compared exposure in vivo, self-instructional training, and RET, and again found the behavioral
treatment superior to the cognitive treatments in reducing agoraphobic behavior. Biran and
Wilson (1981) compared guided exposure with cognitive restructuring for simple phobias
(heights, elevators, darkness) and found that exposure produced superior effects on most
measures. Finally, Williams and Rappoport ( 1983) have also found no added improvement in
agoraphobics when cognitive coping procedures were added to in vivo exposure.

Other researchers have found more promlSlng results for cogmtive techniques. Mavissakalian,
Michelson, Greenwald, Kornblith, and Greenwald (1983) used both selfinstructional training and
paradoxical intention for treating agoraphobia. Paradoxical intention treatment emphasized how
anxious thoughts led to the feared anxiety in a self-fulfilling way and instructed subjects to
reverse their thoughts regarding their fears. Both treatments were effective, with paradoxical
intention being somewhat superior at posttreatment. In a later comparison of paradoxical
intention, muscle relaxation, and graduated exposure for agoraphobia, all three treatments
produced significant improvement, although the behavioral treatments produced effects more
rapidly (Michelson, Mavissakalian, & Marchione, 1985). Emmelkamp, Vander Helm, van
Zanten, and Plochg (1980) compared the effectiveness of gradual exposure in vivo against in vivo
exposure plus self-instructional training for 15 obsessive compulsive patients. The self-
instructional procedures were similar to those of Meichenbaum's stress inoculation therapy. The
addition of self-instructional training did not bring about greater improvement. This kind of
design, however, does not yield much information about the efficacy of cognitive therapy.
Studies with combination treatments for depression have often failed to find an advantage of
combined over individual treatments, and the same may prove to hold true for anxiety disorders
as well.

No studies using a purely cognitive treatment for panic disorder have been reported as yet.
However, Clark, Salkovskis, and Chalkley (1985) tested a cognitive behavioral treatment on 18
patients with panic attacks. Therapy was based on the theory that some persons react to stress by
hyperventilating, causing a drop in the partial pressure of carbon dioxide and thus resulting in
physiological symptoms. Apprehensive thoughts concerning these sensations increase the
hyperventilation and the symptoms. Treatment focused on educating patients about this cycle,
and patients were led to reattribute their panic attacks to the hyperventilation rather than to a
serious illness, such as epilepsy or heart disease. Subjects were also trained in slow breathing
techniques. Frequency of panic attacks showed a sharp and significant decrease in response to
this treatment package during the first 2 weeks of therapy. Further sessions employed a wider
variety of standard cognitive and behavioral techniques. Treatment gains were maintained at the
6-month and 2-year follow-ups. A second study (Salkovskis, Jones, & Clark, 1986) replicated
the results of the first study. It also demonstrated that the partial pressure of carbon dioxide of
panic subjects was significantly below normal before treatment began and rose to normal levels
by the end of treatment.

In a meta-analysis comparing cognitive therapy and desensitization for a variety of clinical and
subclinical anxiety conditions (Berman, Miller, & Massman, 1985), results were found to be
strongly dependent on the theoretical allegiance of the researchers. Cognitive and behavioral
researchers tended to find in favor of their preferred treatments. The meta-analysis found no
significant differences in outcome between the two techniques, although there was a small
nonsignificant difference favoring cognitive therapy.

At the current time, there is not enough evidence to judge the value of cognitive therapy for
anxiety disorders. Too much of the evidence that does exist is based on subclinical populations
or on studies by behavioral researchers using selected cognitive interventions. Moreover,
attempts to maintain a distinction between cognitive and behavioral procedures and thus protect
internal validity may have actually caused problems with internal validity. It is not possible to
fully separate cognitive and behavioral techniques. For example, cognitive therapists often
encourage patients to discover negative thoughts by participating in the feared situations, noting
the automatic thoughts, and testing them out.

Because there are several types of possible cognitive interventions for anxiety, it would be most
fruitful for these to be compared against each other before more comparisons are done between
cognitive therapy and other therapies. At the current time, there is no single cognitive therapy of
anxiety that is generally accepted, as there is with depression. Beck, Emery, and Greenberg
(1985) have written extensively on cognitive techniques for use with anxiety. This will hopefully
lead to greater delineation and comparison of specific cognitive interventions. Once it is
determined which cognitive interventions are most effective for anxiety, comparison with other
treatments will be more meaningful.

PSYCHOSOMATIC DISORDERS

Cognitive behavior therapy has been applied to the treatment of several psychosomatic disorders,
including tension and migraine headaches, duodenal ulcers, and myofascial pain. In addition, it
has been used with Type A behavior. Holroyd, Andrasik, and Westbrook (1977) treated tension
headaches with '~stress-coping training'' that consisted of an extensive set of cognitively oriented
therapeutic procedures based upon Beck (1976), Goldfried, Decenteceo, and Weinberg (1974),
and Meichenbaum (1974). The patients receiving ''stress-coping training" showed a marked
reduction in headache activity that was maintained at follow-up, whereas biofeedback patients
showed more modest improvement and a waiting-list control group showed no improvement. All
subjects receiving stress-coping training reported decreases in headache activity at posttreatment,
with improvement ranging from 43% to 100%. Lake, Rainey, and Papsdorf (1979) included
rational-emotive therapy in a comparative study of different biofeedback procedures in the
management of migraine headache. Digital temperature biofeedback alone and feedback
combined with three 30-minute RET sessions were found to be no more effective than control
conditions. However, three 30-minute sessions would not seem to be an adequate test of the
potential effectiveness of RET with migraine headache.

Stenn, Mothersill, and Brooke (1979) successfully applied biofeedback and cognitivebehavioral
modification in treating patients with myofascial pain dysfunction syndrome (MPDS), also
known as temporomandibular joint pain (TMJ). Patients treated had a diagnosis of TMJ for at
least 1 year with complete failure of previous conservative treatment. Eight treatment sessions
were provided to all subjects. The first half hour consisted of relaxation training (half of the
patients received biofeedback-assisted relaxation training) and the second half hour of one or
more modules of CBT, such as examining self-statements during and prior to the occurrences of
pain, developing coping skills such as assertiveness, and the use of rational-emotive therapy and
stress-inoculation training. All subjects showed significant reduction in subjective-rated pain and
in MPDS signs and symptoms, as rated by an independent physician blind to the treatments and
treatment assignments. Because all subjects received a combination of relaxation training and
CBT, it is not clear how much of the change may have been brought about by cognitive-behavior
modification.

Duodenal ulcer patients were treated by Brooks and Richardson ( 1980) in an eight -session
cognitive-behavioral protocol. The first four sessions focused on anxiety management training,
which included cognitive restructuring and relaxation training. The next four sessions consisted
of assertiveness training, including both cognitive and behavioral rehearsal of appropriate
expression of negative emotions and correction of erroneous beliefs about asserting oneself.
Treatment procedures were followed by a reduction of symptom severity, fewer days of pain,
and a reduction in consumption of antacid medication. During the 3-year follow-up only one of
the nine treatment patients had a recurrence of duodenal ulcer. This was a statistically significant
difference from the eight control patients, five of whom required surgery and/or were victims of
ulcer recurrence.

Jenni and Wollersheim (1979) compared the efficacy of stress management training, cognitive
therapy, and no treatment for altering Type A personality characteristics. Stress management
training consisted of visual rehearsal in which anxiety-arousing scenes were followed by
adaptive responses (e.g., relaxation), whereas cognitive therapy consisted of cognitive
restructuring (Ellis, 1973). Cognitive therapy was more effective than both stress management
training and no treatment in reducing self-perceived levels of Type A behavior, but neither
treatment reduced subjects' cholesterol levels nor blood pressure. Cognitive-behavior therapy
would appear to be quite promising with some stress-related psychosomatic disorders, especially
if one primarily considers those studies meeting basic methodological standards (treatments are
not confounded, adequate control groups and amounts of treatment). Studies meeting these
criteria are those by Holroyd et al. (1977) on tension headache and by Brooks and Richardson
(1980) on duodenal ulcer.

IMPULSE CONTROL DISORDERS

A cogmt1ve behavioral treatment of adults with impulse control disorders has been described by
Watkins (1977, 1983) and Boyer, Beckham, and Buck (1987). Boyer eta/. reported treatment
results for 154 persons for either (l) shoplifting, (2) rape or anger discontrol, (3) child
molestation, or (4) exhibitionism. Only 13% of treatment completers were rearrested within 2
years of the termination, whereas 37% of treatment dropouts had been rearrested by that time.
However, there was no control group to compare treatment results against. A specific goal for
sexual offenders was to increase age-appropriate sexual behavior. Among sexual offenders there
was a significant change away from the categories of ''nonmarried, nondating," and "married,
without sexual relations" to the categories of "nonmarried, dating'' and ''married with sexual
relations''. Predictors of rearrest were dropping out of treatment, being treated for pedophilia,
high levels of denial, high overall MMPI elevations, and not paying for treatment (either refusing
to pay or having treatment paid for by the state). Novaco (1975, 1977a,b) has also reported
success with a cognitive-behavioral approach for anger control with adults. Most studies of
impulsive behavior, however, have been with children or adolescents. For instance, group anger-
control training has been reported to be effective with institutionalized adolescents (Feindler,
Ecton, Kingsley, & Dubey, 1986). As a result of treatment, there was an increase in self-control
and appropriate verbalizations, a decrease in hostile verbalizations during conflict situations, and
lower rates of fines and restrictions during treatment and follow-up. In another study of 76
aggressive boys (Lochman, Burch, Curry, & Lampron, 1984), two cognitively oriented
treatments (anger coping and anger coping plus goal setting) were compared with two control
treatments (goal setting minimal treatment and no treatment). Both anger-coping interventions
improved aggressive and disruptive behavior to a greater degree than the two control treatments.
The cognitivebehavioral treatments appeared to have their greatest impact with those boys who
initially had the poorest problem-solving skills and the greatest amounts of disruptive behavior
(Lochman, Lampron, Burch, & Curry, 1985).

Cognitive and relaxation treatments for anger were compared by Hazaleus and Deffenbacher
(1986). Both resulted in significant anger reduction. Although some minor differences favored
the cognitive intervention posttreatment, the differences disappeared by the 4-week follow-up.
Compared to the control condition, both treatments led to less general and state anger, fewer
physical symptoms of anger, lower daily ratings of anger, and better coping with verbal
antagonism in response to provocations. A !-year follow-up also showed a continued significant
reduction in general anger for both treatment groups. However, cognitive coping skills for anger
reduction appeared to generalize and assist subjects to reduce general anxiety, whereas training
in relaxation coping skills surprisingly did not produce this effect. Overall, the results for the
cognitive condition were said to have replicated those in the literature, but those for the
relaxation condition were somewhat stronger than previously found (e.g., Novaco, 1975;
Schlichter & Horan, 1981).

Combination treatments have also been examined. Kendall and Braswell ( 1982) randomly
assigned 27 children with low self-control to either a behavioral condition, a combined
cognitive-behavioral condition, or an attention control condition. Both active treatments were
associated with improvements on blind ratings of hyperactivity, but only the combined
cognitive- behavioral condition led to improvements in teacher ratings of self-control. The author
viewed cognitive training as adding significantly to treatment effectiveness, but the behavioral
component was viewed as essential. In a review of treatment studies of anger and impulse
control, Kendall (1982) concluded that interventions that combine cognitive and behavioral
training are superior to cognitive or behavioral training alone. There have been a number of other
reports of successful use of self-control training programs involving cognitive behavioral therapy
with hyperactive children (e.g., Barkley, Copeland, & Sivage, 1980; Kendall & Zupin, 1981;
Neilans & Israel, 1981). Not all investigators, however, have found cognitive training to be
effective with hyperactive children. Brown, Wynne, and Medenis (1985), in a comparison of
cognitive therapy with methylphenidate found that only those children in the medication and the
medication-plus-cognitive-therapy conditions demonstrated improvement in attentional
deployment and behavioral ratings. Furthermore, the data indicated that the combined
medication and cognitive-therapy condition was no more effective than that condition involving
medication alone. This lack of any additive effect of cognitive therapy when combined with a
stimulant has been repo~d elsewhere (e.g., Abikoff & Gittelman, 1985; Brown et al., 1986).

SUBSTANCE ABUSE

Although the area of substance abuse has only begun to be examined by cognitive researchers,
two encouraging studies have appeared. Luborsky, McLellan, Woody, O'Brien, and Auerbach
(1985) compared cognitive psychotherapy, supportive expressive psychotherapy, and drug
counseling for male drug abuse clients. Patients in all three conditions improved. However, on
the whole, patients receiving one of the two forms of psychotherapies had better outcomes than
patients receiving drug counseling. Some differences in outcome between supportive-expressive
therapy and CBT were found, in that supportive-expressive patients did better in psychological
functioning and employment and CBT patients did particularly well avoiding legal problems.
Oei and Jackson (1984) found encouraging results for CBT with problem drinkers on the amount
of their alcohol consumption and on other other measures, such as neuroticism scores. Therapist
behaviors were systematically varied, and one of the two types of cognitive treatment
interventions resulted in substantial decreases in drinking and neuroticism scores (for further
discussion, see the section Process Research-Experimental Manipulation). However, there was
no control group, making overall interpretation of the efficacy of their treatments difficult.

EATING DISORDERS

Several studies suggest that cognitive therapy can be an effective treatment for eating disorders.
Connors, Johnson, and Stuckey (1984) used a multiple baseline approach to test a
psychoeducational intervention for bulimic behavior with 20 women subjects. Cognitive
restructuring was only one portion of the intervention. There was a 70% reduction in binge/
purge episodes following twelve 2-hour treatment sessions. Freeman, Sinclair, Turnbull, and
Annandale (1985) have reported pilot data comparing behavioral, cognitive behavioral, and
group interventions with bulimia. A control group was also included. The CBT treatment
involved monitoring and challenging automatic thoughts, changing maladaptive cognitions
regarding weight, improving self-esteem, and increasing assertiveness. Although only initial
results were available, they suggested that there were clinically significant reductions in binge
frequency and eating attitudes for all three therapies compared to the control group. Ordman and
Kirschenbaum (1985) compared a course of cognitive-behavioral therapy lasting an average of
15 weeks with a three-session minimal educational intervention for treating bulimia. Full-
intervention clients were taught to identify unrealistic and self-defeating cognitions regarding
their eating. In addition, they received exposure with response prevention and were taught to
self-monitor hinging and other food-related behavior. Subjects treated with the cognitive
behavioral intervention reduced their frequency of hinging/vomiting, improved their
psychological adjustment, and changed their food-related attitudes to a statistically greater extent
than minimal intervention clients.

In an obesity study, Collins, Rothblum, and Wilson (1986) treated subjects with cognitive
therapy, behavioral therapy, cognitive-behavioral therapy, or a nutrition-exercise (control)
regimen. Cognitive treatment focused on thoughts regarding the role of food in subjects' lives. A
wide variety of cognitive techniques were used. Behavioral techniques focused on stimulus
control of eating cues, self-monitoring of calorie intake, and exercise. Although all treatments
resulted in increased weight reduction compared to the control intervention, the result was
statistically significant only for subjects receiving the behavioral and combined
cognitivebehavioral treatments.
THE SEARCH FOR SPECIFIC EFFECTS

The positive outcome for cognitive therapy in many of the previously mentioned studies does not
establish the claim that it is cognitive change that is bringing about symptom change. It is quite
possible that processes held in common with other therapies (e.g., therapeutic alliance,
establishment of hope) may be the central catalytic agents. A finding that cognitive therapy
produces specific effects (and not just greater effects) compared to other therapies would
strengthen the probability that it possesses unique and effective therapy elements. Such a result
was found by Rush, Beck, Kovacs, Weissenburger, and Hollon (1982) who reported differential
effects of cognitive therapy and imipramine on hopelessness and self-concept in depression.
Compared with imipramine, cognitive therapy resulted in significantly greater reductions in
hopelessness and more generalized gains in self-concept. McNamara and Horan ( 1986) found
evidence for greater change on cognitive measures for cognitive therapy subjects but not for
greater change on behavioral measures (such as the Pleasant Events Schedule) for behavior-
therapy subjects.

Apart from these two studies, however, specific effects generally have not been found for
cognitive therapy. For instance, Simons, Garfield, and Murphy (1984) found that not only were
cognitive therapy and pharmacotherapy equally effective in bringing about remission of clinical
depression but that there was no difference in the rate of improvement on any of three cognitive
measures: the Cognitive Response Test (Watkins & Rush, 1983), the Dysfunctional Attitude
Scale (Weisman & Beck, 1978), or the Automatic Thoughts Questionnaire (Hollon & Kendall,
1980). Zeiss et al. (1979) compared a type of cognitive therapy with two different behavioral
interventions (assertiveness training and increasing pleasant events) with moderately depressed
patients. The three treatments produced equivalent symptomatic results. In addition, there were
no differences on three respective target-dependent (cognitive and behavioral) measures. In a
similar type of comparison, Rehm's self-control therapy was modified to be addressed toward
cognitive change, behavioral change, or both (Rehm et al., 1987). The treatments not only
produced similar symptomatic improvement but also equivalent improvement on cognitive and
behavioral target measures. Imber, Pilkonis, Sotsky, and Elkin (1986) failed to find any
difference in impact on the cognitive items of the Hamilton Rating Scale for Depression
(helplessness, pessimism, worthlessness) between cognitive therapy and imipramine. In addition,
Imber et al. examined the TDCRP data on other measures salient to cognitive processes,
including the Dysfunctional Attitude Scale and cognitive items on the Beck Depression
Inventory. Only the need for social approval factor of the DAS was found to respond
preferentially to CBT over Interpersonal Psychotherapy and imipramine. Silverman, Silverman,
and Eardley (1984) treated depressed patients using a combination of pharmacotherapy and
emotional support. The Dysfunctional Attitude Scale was used as a measure of depressive
attitudes. Patients improved greatly on DAS scores and at posttreatment were essentially at the
same average score as the normal subjects in the original validation sample for the DAS
(Weissman & Beck, 1978). Similarly, Hamilton and Abramson (1983) found that depressed
patients treated on an inpatient psychiatric unit were indistinguishable at discharge from control
subjects on DAS and Attributional Style Questionnaire (Peterson et al., 1982) scores. Neither of
these studies treated patients using cognitive therapy, and yet patients attained scores on
cognitive target measures in the normal range at posttreatment. This suggests that cognitive
therapy is not likely to be much more effective than standard psychiatric treatment in improving
dysfunctional attitudes or attributional styles during the course of a hospital stay.

In summary, the evidence for specific effects for cognitive therapy is very weak at the current
time. If specific effects do occur in cognitive therapy for depression, they are likely to be limited
in scope, and they have generally escaped being measured by the research instruments that have
been used thus far.

PROCESS RESEARCH

Until recently, outcome studies have usually been limited to demonstrating that a particular
treatment intervention (often a treatment "package") is followed at some distant point in time by
symptom alleviation and other changes. As a result, there generally has been little evidence that
specific components within a treatment are related to equally specific proximal or distal changes.
One role of process research is to reveal the most effective components of psychotherapy and the
causal relationship between those ingredients and client change. The identification of processes
in psychotherapy and the demonstration of their relationship to change is the ultimate goal of the
scientific study of psychotherapy. Kiesler ( 1981) has argued that scientific outcome research
requires process analysis of both therapist and patient interview behaviors. Specification of
exactly what these processes are has in the past often centered on broad concepts such as
"combating demoralization" (Frank, 1974) and "active participation of the patient" (Gomes-
Schwartz, 1978), or on relationship variables, for example, "warmth, genuineness, and empathy"
(Truax & Carkuff, 1967) and "a helping relationship with a therapist" (Luborsky eta!., 1975). In
addition, process studies have often focused on measurement of processes within therapy without
attempting to relate these processes to outcome, or they have used outcome criteria that were
poorly specified or overly subjective (therapist or patient self-reports).

SIGNIFICANCE OF DEMONSTRATED OUTCOME

A fruitful way for process psychotherapy research to proceed would be to examine the processes
in a therapy whose efficacy has been well demonstrated. Cognitive therapy would thus seem to
be an excellent choice among psychotherapies for relating process to outcome. Greenberg
( 1981) has commented that one of the most encouraging features of the cognitive behavioral
approach is its combination of a focus on internal mediating events with its rigorous attempts at
measuring and evaluating their change .... This is a domain in which research holds promise of
making a major contribution to the practice of psychotherapy by illuminating some of the
specific mechanisms of client change in therapy. (p. 30)
There are two basic approaches that may be taken toward process analysis. One approach, the
experimental design, involves manipulating key elements of a therapy (i.e., therapist or client
behaviors) and examining proximal and distal effects. Effects studied may be either symptomatic
change or subsequent patient behaviors in therapy. A second approach allows the therapeutic
process to unfold naturally and entails observation and measurement of key processes. Each of
these approaches possesses advantages and disadvantages.

THE NATURALISTIC APPROACH

The main advantage of the naturalistic approach is that artificial manipulations by the
experimenter do not obscure the interaction between patient and therapist. Experimental designs
generally impose restraints on therapist behavior that can conceal any reciprocal, unfolding
interactions. A naturalistic design is likely to be more suitable than an experimental design, for
example, to study the effects of patient noncompliance and resistance on therapist actions and
subsequent patient actions in reaction to therapist behavior. Patient noncompliance may be
followed by a variety of therapist reactions. Experimental manipulations of therapist behavior
would interfere with observing how therapist behavior is typically affected by noncompliance
and, in tum, how any such resulting therapist behavior then affects the client.

Attributional Process Analysis

Naturalistic designs are useful for studying small numbers of patients intensively over long
periods of time. Peterson, Luborsky, and Seligman ( 1983) analyzed tapes of an extended
analysis of a man with chronic depression. Using the revised learned helplessness model
(Abramson et al., 1978), it was predicted that global, stable, and internal attributions for bad
events would lead the patient to be more depressed. Tapes of therapy sessions were monitored at
places where there was an improvement or worsening of mood, and attributions of the patient
before and after each mood shift were rated. Peterson et al. (1982) discovered that mood
worsening tended to follow global, stable, and internal attributions by the patient.

Path Analysis

Another way that naturalistic designs may be informative about the process of cognitive therapy
is through a path analysis of symptom and behavior change. For example, it might be
hypothesized that in cognitive therapy, cognitive change (C; e.g., improvement in self-esteem
and hopefulness) precedes behavioral change (B; e.g., assertiveness, participation in social
events), and that both have direct effects on mood (M). A path analysis model could be
constructed and tested:

C B M
Of course another model could be constructed to te~t the opposite hypothesis: that it is
behavioral change that precedes cognitive change. More elaborate models could be constructed
that would include vegetative symptoms of depression, social support, environmental stressors,
and so on.

Task Analysis

Rice and Greenberg (1984) have proposed that psychotherapy research focus on patterns of
change within therapy. This requires analysis of what constitutes a particular type of pattern and
of ''markers'' that indicate the beginning of such a pattern. Such a method is interested in
predictable patterns of therapist and patient behavior once such a marker occurs. Safran (1985)
has suggested that one type of marker in cognitive therapy is a negative comment by a patient
with no awareness of the possibility that the thought is not realistic (patient is fully immersed in
the thought). A second type of marker occurs when a patient has a negative thought but knows
intellectually that it is not likely to be true (divided awareness). He suggests that different
therapist interventions are required depending on the type of marker.

Other Naturalistic Studies

Persons and Burns (1985) studied one session for each of 17 depressed and anxious patients in
cognitive therapy. As part of normal cognitive therapy procedure, patients were asked to describe
their feelings about an upsetting event and to rate the intensity of their feelings. The degree of
belief in automatic thoughts related to the events was also recorded. After the therapist and
patient worked on formulating rational responses, intensity of feelings and degree of belief in
automatic thoughts were again rated. Improvement in automatic thoughts and relationship with
the therapist (rated after each session) were both statistically related to session outcome. Two
other factors-presence of a personality disorder and high degree of belief in automatic thoughts at
the beginning of a session-were negative predictors of mood change. Although the study was
limited by small sample size and heterogeneity of diagnoses, it is one of the first to seriously
tackle the foremost issue of process research in cognitive therapy-whether cognitive change leads
to mood change. Unfortunately, its correlational nature does not rule out the possibility that it is
actually mood change that is causing the cognitive change or that a third factor is causing both.
Another study by Safran et al. (1987) also addresses the relationship of cognitive change and
mood change within session.

Cognitive change, as rated by the patient and the therapist, did not predict mood shift in the
session, but it did predict degree of problem resolution between that session and the next.
Luborsky et al. (1985) studied three types of treatments for male drug abuse patients: cognitive-
behavioral therapy, supportive-expressive therapy, and drug counseling. Therapist ability to
develop a helping alliance as measured by the Helping Alliance Questionnaire was highly
predictive of outcome. In addition, the degree to which sessions embodied CBT techniques
correlated positively and significantly with improved legal status, employment status, and
psychological status. The presence of supportive-expressive qualities in therapist behavior also
correlated well with outcome in all three treatment groups. Purity of the CBT intervention (i.e.,
being free of non-CBT techniques) was significantly and positively related to better outcomes,
suggesting that greater therapist adherence to the CBT manual led to better outcomes. However,
it was noted by Luborsky et al. that patients' responsiveness to their therapist may have enabled
the therapist to adhere to their intended technique. In fact, there is some evidence along these
lines that individual differences do affect therapist adherence.

Vallis and Shaw (1987) found that patient difficulty was "strongly and inversely related to
therapist competency. '' Ratings of higher patient difficulty were associated with lower ratings of
therapist competence. This relationship held both across cases and within cases. That is, not only
did patient difficulty averaged across sessions covary with therapist competency averaged across
sessions, but covariation was also found within series of sessions of therapist-patient dyads.

The importance of individual differences among patients in determining outcome has also been
noted by Fennell and Teasdale (1987). Patients who improved rapidly in response to CBT for
depression (labeled steeps for the gradient of change) maintained their gains and were less
depressed at termination and followup than initially slow responders (slights). Steeps and slights
were found to differ on several measures. Steeps were more likely to positively endorse the
cognitive conceptualization of depression. They also received a greater positive benefit from the
first homework assignment. Finally, they scored higher on a measure of depression about being
depressed.

EXPERIMENTAL MANIPULATION

As stated before, one of the primary problems with the naturalistic approach is interpreting
direction of causality. In the experimental approach, therapist behavior is protected to some
degree from the influence of patient behavior, and, therefore, interpreting directionality of effect
is Jess difficult. The experimental approach is also well suited to study low frequency or novel
therapist behaviors, which would not be likely to occur in naturalistic studies. One experimental
approach to process research involves dismantling a therapy into its component parts, as has
been done by Rehm et al. ( 1981) with self-control therapy. Dismantling involves a systematic
variation of treatment components across groups of patients. Rehm studied four versions of self-
control therapy: self-monitoring only, self-monitoring plus selfevaluation, self-monitoring plus
self-reinforcement, and a combination of all three components.

The self-evaluation component was similar to cognitive therapy in that it taught patients to set
realistic goals and make accurate attributions for success and failure. The results were very
surprising. All treatment conditions did better than the waiting-list controls. Differences among
the three treatment groups were minor and tended to occur on some outcome measures but not
others. In a later dismantling study (Komblith, Rehm, O'Hara, & Lamparski, 1983), the time of
treatment was lengthened from 7 to 12 weeks. Four treatments were compared: self-control
principles without homework; self-monitoring plus self-evaluation; comprehensive self-control
therapy; and a psychodynamic therapy group as a control. Again, no significant differences were
found.

In a somewhat similar design, Jarrett and Nelson (1987) varied the sequencing of three main
cognitive therapy components for depression: self-monitoring, logical analysis, and hypothesis
testing. Each group received self-monitoring first, but half of the subjects were taught logical
analysis before hypothesis testing, whereas the other half received hypothesis testing before
logical analysis. By conducting assessments of depressive symptoms, automatic thoughts,
pleasant events, and interpersonal functioning before and after each of the three components, it
was possible to determine whether hypothesis testing or logical analysis was more efficacious.
Self-monitoring alone was associated with reductions only in transient mood as measured by the
Depression Adjective Checklist (DACL). Both logical analysis and hypothesis testing were
associated with significant improvements on all measures. The interaction of time X sequence
was not significant for any of the dependent measures, suggesting that hypothesis testing and
logical analysis were approximately equal in effectiveness. On some measures, exposure to all
components combined was associated with greater improvement than exposure to only logical
analysis or hypothesis testing. However, for other measures, the combination (or more treatment)
equaled the effect of logical analysis or hypothesis testing alone.

Zettle (1987) compared the efficacy of three CBT components of treating depression: distancing
of thoughts, rational restructuring, and behavioral homework. Subjects receiving training in both
distancing and rational restructuring improved more than subjects trained in rational
restructuring only. In addition, subjects given behavioral assignments improved more than
subjects who were not.

Two studies have compared the effects of thought exploration and thought challenge.
Straatmeyer and Watkins (1974) and Teasdale and Fennell (1982) had one group of subjects to
explore and also to challenge their negative thoughts and another group to only identify thoughts
without challenging them. In both studies, the thought-challenging condition led to greater
immediate improvement in mood than thought exploration. These studies and the one by Jarrett
and Nelson provide very suggestive evidence that thought monitoring and exploration of
negative thoughts alone are ineffective in treating depression.

Still another way of comparing component procedures of CBT has been reported by Amkoff
( 1986). Students with test anxiety received either training in coping self-statements or in
restructuring of irrational beliefs. Whereas the coping group showed more improvement
compared to the control group, there was no significant difference in effects between the two
cognitive treatments. Moreover, there were also no posttreatment differences between the two
conditions on the Irrational Beliefs Test (Jones, 1969) that would logically be expected to change
more in the cognitive restructuring treatment.
Oei and Jackson (1984) tested two different styles of cognitive therapy with problem drinkers.
Both treatments used a combined cognitive restructuring and social skills treatment. In one
group, future-oriented statements and positive self-statements were reinforced. In addition,
therapists in that group used self-disclosure to demonstrate how they had dealt with their own
maladaptive attitudes. In the second group, therapists avoided self-disclosure and giving
opinions; they did not reinforce positive self-statements, healthy attitudes, or futureoriented
statements; and they did not challenge negative self-statements. There were equal amounts of
positive and aversive reinforcement given to each group, but it was given randomly to the second
group. Changes in cognitions were found to precede or to parallel behavioral changes in patients.
Subjects in the first group had more positive and future-oriented statements and fewer negative
statements in therapy sessions. The first group also improved more on behavioral ratings of
social skill and made greater decreases in alcohol consumption.

Moreover, the first group maintained their decrease in drinking during a 6-month follow-up
period, whereas the second group returned to pretreatment levels.

IMPORTANT VARIABLES TO BE ScRUTINIZED IN COGNITIVE PROCESS RESEARCH

Patient Difficulty

Cognitive therapists have recognized that there is wide variability in patient difficulty (Fennell &
Teasdale, 1982). There is little empirical evidence as to the best methods of effectively
overcoming noncompliance, although there have been some clinical suggestions on this topic
(Bums, Adams, & Anastopoulos, 1985). This is a field that would seem to hold considerable
promise for process researchers.

Presence of a personality disorder would be one variable likely to predict noncompliance

(Persons & Bums, 1985), although clinical experience suggests that dependent personalities and
some other types may actually be more compliant. Vallis and Shaw ( 1987) investigated the
types of patient characteristics that were related to observer ratings of patient difficulty. Negative
patient attitude toward the therapist was the strongest predictor of patient difficulty. Other
variables that were significantly associated with patient difficulty were greater patient
depression, lower patient activity levels, and less willingness for self-exploration. Simons,
Lustman, Wetzel, and Murphy (1986) found that patients scoring high on learned resourcefulness
as measured by the Self Control Schedule (Rosenbaum, 1980) were more responsive to cognitive
therapy for depression, and patients scoring low on the SCS responded preferentially to a
tricyclic antidepressant. Rehm et al. (1987) also found high scores on the Self-Control Schedule
to be predictive of success for all three treatments in their study: cognitively targeted,
behaviorally targeted, and combined target self-control therapy. Strength of dysfunctional
assumptions may also have an effect on treatment outcome. Keller ( 1983) found that depressed
patients who had high Dysfunctional Attitude Scale scores responded less well to cognitive
therapy than those with lower scores. It is unclear how strong dysfunctional attitudes and low
levels of learned resourcefulness may affect outcome-whether they affect patient behavior in
therapy or inhibit symptom improvement in some other ways.

Therapist Behavior-Purity and Competence of Cognitive Technique

The research of Luborsky et al. (1985) suggests that "pure" CBT interventions are more effective
than interventions that blend other approaches. It would seem logical that introduction of non-
CBT techniques might confuse patients and inhibit assimilation and consolidation of CBT coping
skills. However, their research left unclear whether pure interventions were indeed more
effective, or whether therapists were simply better able to remain true to "pure" cognitive
interventions when patients were improving.

Competence is distinguished from "purity of intervention" in that it involves skillful use of all
relevant aspects of cognitive therapy. For example, psychoanalytic process researchers have
begun to abandon gross measures such as frequency of interpretations and have begun to
investigate the "suitability" of interpretation on outcome (Silberschotz, Fretter, & Curtis, 1986).
In the same way, therapist application of cognitive techniques may be skillful or unskilled in the
way that it selects which negative thoughts and depressive behaviors to work on. Therapist
interventions may lead toward or away from resolution of the patient's underlying cognitive
schema. Little is known about relationship of competence in CBT in this sense and treatment
outcome.

Therapist and Patient Relationship

Because cognitive therapy emphasizes developing new ways of thinking, the relationship
between therapist and patient is not always thought of as an important ingredient in CBT. Yet it
is clearly spelled out as a necessary ingredient of cognitive therapy in Beck eta!. (1979). Persons
and Bums (1985) found relationship to relate to mood change during a single session, and
Lubarsky et a!. ( 1985) found patient perception of the therapist's ability to form a helping
relationship to correlate with outcome across all three treatments they studied, including CBT.

Further work is needed to determine whether relationship provides a truly independent

contribution to outcome, as suggested in the previously mentioned two studies, or whether it only
potentiates the assimilation of cognitive techniques.

Process Research in Cognitive Therapy-Conclusion

It is too early to draw any conclusions from the available process literature. Most of the
methodology is quite new. Researchers are only beginning to test the primary premise of
cognitive therapy-that it is cognitive change that causes symptomatic improvement. The next
decade is likely to see an increasing amount of research focused on this issue, which is of the
highest importance. With regard to predictors of outcome in cognitive therapy, the following
constructs are promising candidates: purity and competence of CBT technique, quality of the
therapist-patient relationship, cognitive style of the patient, and ''patient difficulty,'' a broad
concept that will undoubtedly prove to be very complex and heterogeneous.

INSTRUMENTATION AVAILABLE FOR PROCESS AND OuTCOME RESEARCH

Several instruments are now available for outcome and process research in cognitive therapy.
With regard to measurement of patient variables, the Beck Depression Inventory and the
Hamilton Rating Scale for Depression have traditionally been the measures of choice among the
multitude available to researchers for measuring depression. Because the BDI has items closely
identified with the cognitive triad, a subset of BDI items can also be used to measure it
specifically. The Hopelessness Scale (Beck, Weissman, Lester, & Trexler, 1974) specifically
measures one aspect of the triad-view of the future. An instrument specifically designed to
measure the cognitive triad-the Cognitive Triad Inventory (Beckham, Leber, Watkins, Boyer, &
Cook, 1986)-is designed to measure the entire triad and to be sensitive to changes in it during a
session or across the course of treatment. Other instruments measure patients' cognitive styles,
for example, the Cognitive Bias Questionnaire (Krantz & Hammen, 1979); the Attributional
Style Questionnaire (Peterson et al., 1982); and the Self-Control Schedule (Rosenbaum, 1980).
Other scales measure amounts of depressive cognitions (Auto-matic Thoughts Questionnaire;
Hollon & Kendall, 1980), and tendency to use cognitive coping strategies (Coping Strategies
Scales; Beckham & Adams, 1984). For a review of these and other instruments see Krantz and
Hammen (1985) and Goldberg and Shaw (Chapter 3 in this book).

There does not appear to be a satisfactory means of measuring underlying core beliefs
(dysfunctional assumptions). Two instruments have been used for this purpose, the Irrational
Beliefs Test (Jones, 1969) and the Dysfunctional Attitude Scale (Weissman & Beck, 1978).
However, although the DAS has been shown to have some value in predicting depressive relapse
above and beyond BDI scores (Simons et al., 1986), other studies suggest that the DAS simply
reflects the presence or absence of depression. Two groups have suggested more idiographic
approaches to assessment of core beliefs (Beckham, Boyer, Cook, Leber, & Watkins, 1984;
Safran, Vallis, Segal, & Shaw, 1986), but no data on empirical validity of such methods have yet
been presented.

The Cognitive Therapy Scale (Young & Beck, 1980) has been the measure primarily used thus
far for scoring therapist behavior in cognitive therapy. It consists of two major sections, one
measuring general interpersonal factors and a second measuring quality of application of
cognitive techniques. Several studies using the CTS suggest that it can provide reliable, valid
ratings of therapist competency when trained raters are used (Dobson, Shaw, & Vallis, 1985;
Hollon et al., 1981; Vallis, Shaw, & Dobson, 1986; Young, Shaw, Beck, & Budenz, 1981).
lntraclass reliability coefficients have ranged from .54 to . 96. Although the CTS was designed to
assess competency, other measures have been designed to assess the degree to which therapist
behaviors are consistent with the cognitive approach (DeRubeis, Hollon, Evans, & Bemis, 1982;
Lubarsky, Woody, McLellan, O'Brien, & Rosenzweig, 1982).