Psychological Bulletin
1990, Vol. 108, No. 1,30-49
Copyright 1990 by the American Psychological Association, Inc.
0033-2909/90/$00.75
Psychotherapy for the Treatment of Depression:
A Comprehensive Review of Controlled Outcome Research
Leslie A. Robinson, Jeffrey S. Herman, and Robert A. Neimeyer
Memphis State University
Previous quantitative reviews of research on the efficacy of psychotherapy for depression have included only a subset of the available research or limited their focus to a single outcome measure.
The present review offers a more comprehensive quantitative integration of this literature. Using
studies that compared psychotherapy with either no treatment or another form of treatment, this
article assesses (a) the overall effectiveness of psychotherapy for depressed clients, (b) its effectiveness
relative to pharmacotherapy, and (c) the clinical significance of treatment outcomes. Findings from
the review confirm that depressed clients benefit substantially from psychotherapy, and these gains
appear comparable to those observed with pharmacotherapy. Initial analysis suggested some differences in the efficacy of various types of treatment; however, once the influence of investigator allegiance was removed, there remained no evidence for the relative superiority of any 1 approach. In
view of these results, the focus of future research should be less on differentiating among psychotherapies for depression than on identifying the factors responsible for improvement.
Depression is a prevalent clinical disorder with high economic and emotional costs. Epidemiological research has indicated that 10%-20% of the population experience a major depressive episode at some point in their lifetime (Boyd & Weissman, 1981), with the incidence highest during the adult years
when family and career responsibilities may be most adversely
affected (Weissman & Myers, 1978). Although the remission
rate for depressive disorders is relatively high (Beck, 1967, chap.
3), a substantial portion of those afflicted remain chronically
depressed (Weissman & Klerman, 1977), and those who do improve are at an increased risk for further episodes (Belsher &
Costello, 1988;Kessler, 1978; Klerman, 1978).
Until recently, depression was treated almost exclusively with
medication, traditional insight-oriented therapy, or a combination of the two. However, the 1970s witnessed the development
of a number of new therapeutic approaches, each of which pos-
its a different etiological model for depression. For example, advocates of behavioral approaches treat depression as a consequence of a low rate of response-contingent positive reinforcement. The object of this therapy, then, is to increase
reinforcement, either by encouraging participation in pleasant
activities (Lewinsohn, 1974) or by building the assertion skills
necessary to elicit social rewards (LaPointe & Rimm, 1980;
Sanchez, Lewinsohn, & Larson, 1980). A second treatment approach, cognitive therapy, is derived from Beck's (1967) view of
depression as an affective response to negative beliefs. Modifying these unproductive beliefs is the primary focus of cognitive
therapy. In addition, a number of treatment packages have been
developed that explicitly integrate elements from both cognitive
and behavioral models. Examples include Lewinsohn's Coping
with Depression course (Lewinsohn, Steinmetz, Antonuccio, &
Teri, 1985) and Rehm's self-control therapy (Fuchs & Rehm,
1977).
With the development of these new therapies has come a dramatic increase in outcome research on the efficacy of treatments for depression (Brown & Lewinsohn, 1984; Elkin, Parloff, Hadley, & Autry, 1985;McLean, 1981). Two major evaluation strategies have been used: (a) comparisons between treated
clients and wait-list controls and (b) comparisons between clients receiving different types of psychotherapy. The results of
studies using wait-list comparisons have generally indicated
that a number of treatment approaches are effective. However,
as Kazdin (1981) has pointed out, the focus in recent depression research has been not so much on comparisons with untreated controls as on comparisons between different types of
therapy, and here the results have been remarkably inconsistent.
For example, Shaw (1977) found that cognitive therapy was
more effective than behavioral treatment, whereas others (e.g.,
Hodgson, 1981; LaPointe & Rimm, 1980) have reached the opposite conclusion. Similarly mixed results have been obtained
when cognitive or behavioral methods have been compared
This research was supported by a Centers of Excellence grant
awarded to the Department of Psychology at Memphis State University
by the state of Tennessee. The research was also aided by a Faculty Research Grant awarded to Jeffrey S. Herman by Memphis State University. While the research was being conducted, Leslie A. Robinson was
supported by a Van Vleet Memorial Fellowship.
Portions of this work were presented at the annual meeting of the
Society for Psychotherapy Research, Santa Fe, New Mexico, June 1988.
The research was also discussed as part of the symposium Psychotherapy for Specific Disorders: Depression, Agoraphobia, and Anorexia Nervosa (Jeffrey S. Herman, Chair), conducted at the 96th annual meeting
of the American Psychological Association, Atlanta, Georgia, August
1988.
We would like to thank Marna Barrett, Crystal Blyler, Diane Camp,
and Chris Quails for their assistance in locating and coding the research
studies included in the review.
Correspondence concerning this article should be addressed to
Jeffrey S. Herman, Department of Psychology, Memphis State University, Memphis, Tennessee 38152.
30
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
with more traditional approaches (e.g., Fleming & Thornton,
1980;LaPointe&Rimm, 1980; Thompson & Gallagher, 1984).
To further complicate matters, few studies have obtained consistent results across all outcome measures or assessment
points.
In attempting to account for these inconsistencies, researchers have advanced two primary hypotheses. First, it has been
suggested that genuine differences in efficacy do exist but have
been obscured by variations across studies in factors such as
treatment procedures, client selection, and therapist training.
According to this view, consistent differences in therapy outcome might emerge in large-scale, tightly controlled research
programs. Such is the philosophy underlying the Treatment of
Depression Collaborative Research Program (Elkin et al.,
1985), a multisite project initiated by the National Institute of
Mental Health to evaluate the efficacy of cognitive-behavioral,
interpersonal, and pharmacological treatments for depression.
In contrast, others have suggested that there are no significant
differences in the effects of the various therapies, partly because
of considerable overlap in their treatment methods. Both Kazdin (1981) and McLean (1982) have pointed out that there may
well be no treatment techniques that are used only within a
particular therapy. Instead,
what researchers and clinicians presently h a v e . . . is broad agreement on the characteristics of clinical depression . . . ; divergent
theory to account for the hypothesized mechanisms responsible for
the etiology, maintenance, and reversal of depression; and strikingly similar treatment procedures deriving from these diverse theories. (McLean, 1982, p. 22)
If the methods and effects of different forms of psychotherapy
are similar, why then do some studies reveal reliable differences
in treatment efficacy, whereas others do not? One possible explanation involves the impact of the researcher's allegiance. In
previous reviews of psychotherapy research (Herman, Miller, &
Massman, 1985; Smith, Glass, & Miller, 1980, chap. 5), the results of comparisons between therapies have been found to vary
according to the theoretical preference of the investigator. Although it seems reasonable to expect that a similar effect might
be operative in the depression literature, this possibility has not
yet been explored.
The typical response to the inconsistency in the existing literature has been a call for further research, in the hope that better
designed studies might reveal some client or treatment format
variables that could account for the contradictory findings of
previous research. Additional studies would undoubtedly contribute to the literature, but it appears unlikely that they can
provide any final answers to the central problems in depression
research. As Fiske (1983) has pointed out, although "the single
study may stimulate or irritate in a healthy fashion, only the
distillations from the entire body of research in an area have
lasting effects" (p. 65).
This type of broad integration of the depression treatment
literature has been attempted in a number of recent narrative
reviews. However, as of yet no clear and consistent conclusions
have been reached. For example, some commentators have indicated that the research supports the efficacy of both cognitive
and behavioral approaches to the treatment of depression (e.g.,
Blaney, 1981; Emmelkamp, 1986; McLean, 1982; Rehm &
31
Kornblith, 1979; Whitehead, 1979), whereas others have concluded that behavioral techniques have not yet received adequate empirical support (e.g., DeRubeis & Hollon, 1981; Hollon, 1981; Kovacs, 1980). In a few instances, it has been suggested that treatments incorporating both cognitive and
behavioral components may be more effective than either approach alone (e.g., Blaney, 1981; Rehm & Kornblith, 1979;
Whitehead, 1979). Other reviewers have simply concluded that
at this point, there is no clear treatment of choice (e.g., Emmelkamp, 1986; McLean, 1982; Rush, 1982). Furthermore, although differences in client populations and therapy formats
have often been cited as contributing to inconsistent results
across studies, few reviewers have systematically examined research relating these variables to outcome. Most often, they
have simply noted that there are at present no clear prognostic
or prescriptive indicators and have called for further research
in these areas (e.g., Blaney, 1981; DeRubeis & Hollon, 1981;
Kovacs, 1980; McLean, 1981; Rush, 1982; Whitehead, 1979).
Given the large number of studies on the treatment of depression and the complexity of their results, other reviewers have
turned to quantitative techniques for summarizing the research. The advantage of these quantitative, or "meta-analytic,"
procedures (e.g., see Glass, McGaw, & Smith, 1981) is that they
provide a powerful method for identifying trends that might
otherwise be overlooked. In addition, these techniques permit
the systematic assessment of factors that vary across individual
studies.
Although several previous quantitative reviews of the psychotherapy literature (Dush, Hirt, & Schroeder, 1983; Shapiro &
Shapiro, 1982; Smith etal., 1980) have reported separate analyses of depressed samples, in each case these analyses were limited to a small sample of the research on psychotherapy treatments for depression. Three other reviews that focused specifically on depression have suffered from a similar limitation. In
two of these reviews (Quality Assurance Project, 1983; Steinbrueck, Maxwell, & Howard, 1983), the primary purpose was
to estimate the relative benefits of pharmacotherapy and psychotherapy for depression; the third review (Conte, Plutchik,
Wild, & Karasu, 1986) used a "box-score" summary technique
to evaluate treatments combining pharmacotherapy with psychotherapy. Given their focus on drug treatments, however,
these reviews included relatively few studies that assessed the
efficacy of psychotherapy alone. Thus, many substantive issues
concerning psychotherapy for depression could not be addressed.
In contrast, two recent reviews (Dobson, 1989; Nietzel, Russell, Hemmings, & Gretter, 1987) have included substantially
larger portions of the literature on psychotherapy for depression. However, even these reviews were less than comprehensive,
because they were both limited to studies that used the Beck
Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, &
Erbaugh, 1961) as an outcome measure, and only the BDI was
analyzed. Moreover, the two reviews reached markedly different
conclusions. Nietzel et al. (1987) found that individual treatment was more effective than group treatment, but there were
no reliable differences in the efficacy of cognitive, behavioral,
and other forms of therapy. In contrast, Dobson (1989) concluded that Beck's cognitive therapy was more effective than
other therapeutic approaches. Perhaps one reason for these di-
32
L. ROBINSON, J. BERMAN, AND R. NEIMEYER
vergent findings is that neither review adjusted for the potential
influence of investigator allegiance in the studies they examined.
The purpose of this review was to provide a more complete
quantitative summary of the controlled research evidence on
psychotherapy for depression. In the first of the following analyses, we examined the efficacy of psychotherapy for depression
by using evidence from a substantially larger number of empirical studies than have been included in any previous quantitative
review of this literature. In a second set of analyses, we used the
available research to assess the comparative efficacy of psychotherapy and the leading alternative treatment for depression,
pharmacotherapy. Finally, we evaluated the clinical significance
of psychotherapy by comparing treated clients with nondepressed individuals. Thus, our analyses allowed us to assess the
effectiveness of psychotherapy for depression, its benefits relative to pharmacotherapy, and how much former clients resemble those who are not depressed.
Analysis 1: Effectiveness of Psychotherapy
In this first set of analyses, we investigated not only the overall
effectiveness of psychotherapy but also the relative effectiveness
of different forms of treatment. In addition, we examined a
number of other substantive issues in research on the treatment
of depression, including the role of investigator allegiance, the
impact of group and individual therapy formats, the importance of diagnostic screening procedures, and the influence of
other variables such as therapist training, length of treatment,
and client characteristics.
Method
Studies. The first analysis was based on a total of 58 studies of psychotherapy for the treatment of depression. (See Appendix A for a list of
the references for these studies.) The studies were identified through a
search of the volumes of Psychological Abstracts (1976-1986), references of published reviews and outcome studies, and an issue-by-issue
examination of the 1985 and 1986 volumes of relevant journals.1 Of the
58 studies included in this analysis, 47 were not included in the recent
Dobson (1989) review and 40 were not covered in the analyses by Nietzeletal.(1987).
We used several criteria in selecting studies for inclusion in the review.
First, to assess the effects of therapy on depressive disorders (rather than
depressive moods), the analysis was restricted to studies using samples
identified as primarily suffering from depression. Thus, studies that described clients in more general terms (e.g., neurotic) or in terms of another specific diagnostic category (e.g., alcoholic) were excluded, even
when the researcher reported that the clients were also depressed. Studies using inpatient samples and those that focused on children or adolescents were also omitted, because treatment methods used with these
groups often differ from those that are the focus of this review.
Second, we included a study in the review only if it contained a comparison between treatment and no treatment or between different types
or formats of therapy. Thus, we omitted case histories and studies using
simple pre-post designs.
Third, because our primary interest was in the effects of psychotherapy, we excluded research on treatments that did not have a prominent
verbal component. Thus, treatments such as exercise and bibliotherapy
were not considered. In addition, we omitted the very few studies of
family and marital therapy because their interactional focus differed
substantially from the more individualized goals of the treatments included in the reviewed studies.
The studies selected for review investigated a variety of psychotherapeutic methods. Distinctions between the different types of therapy were
often difficult to draw, in part because of overlap in the treatment techniques. However, a careful inspection of the method sections of these
studies indicated that most therapies could be classified into one of four
categories: (a) cognitive, (b) behavioral, (c) cognitive-behavioral, and
(d) general verbal therapy. The first category, cognitive therapy, included
those treatments that focused primarily on the evaluation and modification of cognitive patterns. For example, treatments that involved attributional retraining or challenging irrational beliefs were classified as
cognitive. However, therapies that simply directed clients to substitute
positive thoughts or images for negative ones were excluded from this
category. Such treatments differ from cognitive therapy as it is usually
practiced in that no evaluation of existing cognitions is undertaken by
the client (Ledwidge, 1978; also see Miller & Berman, 1983). The behavioral therapy category included treatments designed to decrease depression by changing behavioral patterns (e.g., by increasing assertive
behavior or participation in pleasant activities). Therapies that included
both cognitive and behavioral components were classified as cognitivebehavioral treatments. The final category, general verbal therapy, comprised treatments such as psychodynamic therapy, client-centered approaches, and other forms of interpersonal therapy such as that outlined
by Klerman, Weissman, Rounsaville, and Chevron (1984). The commonality among these treatments is that each places relatively greater
emphasis on insight than on the acquisition of a set of specific skills.
Although the general verbal category was broad, too few studies were
available to evaluate the effectiveness of the specific therapies within this
group.
Treatment outcomes were assessed in the studies by a variety of instruments. Many of the instruments were designed specifically to assess
depressive symptomatology, but some were more general or evaluated
other areas of functioning. We classified the following scales as specific
measures of depression: the BDI (Beck et al., 1961), the Zung Self-Rating Depression Scale (Zung, 1965), the Depression Adjective Check List
(Lubin, 1965), the Hamilton Rating Scale for Depression (Hamilton,
1960), the Depression Scale of the Minnesota Multiphasic Personality
Inventory (Hathaway & McKinley, 1967), the D-30 (Dempsey, 1964),
and the Center for Epidemiologic Studies Depression Scale (Radloff,
1977).
Descriptive characteristics of the 58 studies reviewed are presented
in Table 1. As the table reveals, the typical client was a middle-aged
woman who was experiencing moderate depression as measured by the
BDI. The therapy was usually brief, and treatment sessions occurred
approximately once a week. In the 34 studies that included a follow-up
assessment, this follow-up was conducted an average of 13 weeks
(range = 2-52) after treatment termination.
In 28 (48%) of the studies, clients were solicited from the community
through media announcements. Another 14 investigations (24%) relied
on students solicited in a university setting, and 9 studies (16%) used
traditionally referred outpatients. In the remaining 7 studies (12%), either the referral source was not reported or both solicited and traditionally referred clients were included.
Although all studies focused on the treatment of depression, some
screened clients more rigorously than did others. In 20 investigations
(35%), clients were required to meet formal diagnostic criteria for a
depressive disorder in order to be included. Most often, the Research
1
The issue-by-issue search was conducted for the following journals:
Archives of General Psychiatry, Behavior Modification, Behavior Therapy, Cognitive Therapy and Research, Journal of Clinical Psychology,
Journal of Consulting and Clinical Psychology, Journal of Counseling
Psychology, and Psychotherapy: Theory, Research and Practice.
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
Table 1
Characteristics of Psychotherapy Studies
Study characteristic
M
Range
No. of clients
Clients per group
Percentage attrition (posttreatment)
Percentage female clients
Client age (years)
Initial Beck Depression Inventory score
No. of therapists
Weeks of treatment
No. of sessions
40.4
14.3
11.0
79.6
39.4
22.7
4.0
6.9
8.7
9-155
4-47
0-65
50-100
19-71
12-30
1-18
1-36
1-46
Note. Each mean is based on at least 42 of the 58 studies.
Diagnostic Criteria (Spitzer, Endicott, & Robins, 1978) were used, but
other investigators used the Feighner criteria (Feighner et al., 1972) or
the Diagnostic and Statistical Manual of Mental Disorders (American
Psychiatric Association, 1980). The 38 other studies (65%) used less
stringent selection criteria, such as scores on self-report measures of
depression.
Estimating treatment effects. Each outcome measure reported in a
study was expressed in terms of Cohen's (1977) d, a standardized measure of effect size. Cohen's d is denned as
(D
where m, and m2 represent group means and s is the pooled withingroups standard deviation. For comparisons between controls and
treated clients, the control mean was subtracted from the treatment
mean. Thus, an effect size of 0.5 indicated that the mean of the treated
group was one half of a standard deviation larger than the mean of the
control group. In direct comparisons between different types or formats
of treatment, the difference in means was calculated in a consistent
manner. For example, we always calculated effect sizes for comparisons
between cognitive and behavioral therapy by subtracting the behavioral
group mean from the cognitive mean. For some outcome measures,
higher numbers indicated greater improvement, whereas for other measures lower numbers represented greater improvement. The sign of each
effect size was therefore adjusted so that positive values indicated that
clients in the first group improved more and negative values indicated
that clients in the second group improved more.
We calculated effect sizes for all of the measures within a study, with
a few exceptions. Process measures, such as client ratings of therapist
likability, were not coded, because these measures are not a direct indication of treatment efficacy. Measures of paralinguistic characteristics
(e.g., the latency of client speech) and observations of body movement
(e.g., counts of head movements) were also excluded, because as Jacobson (1981) has pointed out, the validity of these data in the assessment
of outcome has not been demonstrated. For similar reasons, the few
measures of physiological characteristics, such as heart rate, were omitted. Measures administered midway through treatment were not coded.
Also, when multiple follow-up assessments were conducted, only measures taken at the longest follow-up point were used to compute followup effect sizes.
Most effect sizes were calculated from means and standard deviations
reported in the studies. When this information was not available, we
used other statistics to estimate the effect size. The various methods for
computing such effect sizes are reported in previous work (e.g., Miller
&Berman, 1983; also see Glass etal., 1981,chap. 5).
Occasionally, researchers described a measure but did not report
findings for it or reported results only as nonsignificant. In these situations, an exact effect size could not be calculated. However, excluding
33
these measures would have inflated the estimate of overall effect size,
because investigators are more likely to provide adequate information
on measures that reveal reliable group differences. Thus, when findings
were not reported or were described simply as nonsignificant, we conservatively estimated the effect size to be zero.
Hedges (1982) has identified a small-sample bias in the estimate of
effect size. Although this bias is of practical concern only when sample
sizes are quite small (i.e., fewer than 20 subjects), we applied Hedges'
(1982, Formula 4) correction for the bias to all effect sizes reported in
our analyses.
Preliminary analyses. Differences between groups were usually assessed by more than one outcome measure (M = 6.0, range = 1-25).
Moreover, many studies reported results for more than one treatment
comparison (M = 2.1, range = 1-6). Other reviewers have treated the
effect sizes derived from individual outcome measures as separate observations in their analyses (e.g., see Smith et al., 1980, chap. 4). However,
this procedure arbitrarily weights studies according to the number of
outcome measures and treatment comparisons reported. Even worse,
multiple effect sizes derived from the same study may not represent statistically independent observations. If effect sizes within a study are not
independent, then using the effect size as the unit of analysis can seriously underestimate error variance and inflate tests of statistical significance (e.g., see Glass et al., 1981, chap. 6).
To assess the issue of nonindependence in our data, we first examined
whether the variation in effect sizes derived from different treatment
comparisons was greater than the variation of effect sizes within a single
treatment comparison. Our focus in this analysis was on the most frequent type of comparison, that between treated clients and untreated
controls. We conducted an analysis of variance in which the unit of
analysis was the individual effect size (N = 354) and the independent
variable was the treatment comparison (N = 78). This analysis revealed
that the variability among effect sizes drawn from different treatment
comparisons was indeed far greater than the variability of effect sizes
within a treatment comparison, intraclass R = .66, F\ll, 276) = 9.68,
p<.001. 2
Given this problem of nonindependence, we averaged all effect sizes
derived from the same treatment comparison. For example, if a treatment comparison yielded five different effect sizes, the mean of these
measures was calculated for use in overall analyses. However, separate
means for different types of outcome measures were also retained. For
instance, we calculated the mean of effect sizes derived from self-report
outcome measures and the mean of those based on observer ratings.
These additional means could then be used in analyses in which we
wished to assess treatment effects for particular types of outcome measures.
Many of the studies reported the results of more than one treatment
comparison. As a further check on nonindependence, we therefore examined whether the variability among treatment comparisons drawn
from different studies was greater than the variability of multiple treatment comparisons within the same study. As before, our focus was on
the most frequent type of comparison, that between treated clients and
untreated controls. We conducted an analysis of variance in which the
unit of analysis was the mean effect size for a comparison (N = 78) and
the independent variable was the study (N = 37). This analysis indicated
that the variability among comparisons from different studies was sub-
2
Because the number of effect sizes varied for different treatment
comparisons, we computed the intraclass correlation by using the procedure for unequal class membership outlined in Haggard (1958, chap.
2). This method of computing the intraclass correlation was also used in
the later analysis assessing the independence of treatment comparisons,
because the number of treatment comparisons varied for different studies.
34
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
Table 2
Efficacy of Psychotherapy at Posttreatment and Follow- Up
Effect size
Assessment
AT of
studies
M
SD
Posttreatment
Follow-up
37
9
0.73*
0.68*
0.69
0.68
Note. Means and standard deviations are based on weighted least
squares analyses in which effect sizes were weighted by sample size.
*p<.05.
stantially larger than the variation of comparisons within the same
study, intraclass R = .76, F(36,41) = 7.67, p < .001.
Because of this second level of nonindependence, we combined the
results of multiple treatment comparisons from the same study. For example, if a study yielded three different comparisons of treatment to no
treatment, we averaged the results of these comparisons for our overall
analyses. However, we also retained separate means for specific types of
treatment. For example, we calculated the mean effect size within a
study for all comparisons involving cognitive therapy or all comparisons
involving behavioral therapy. These additional means could then be
used in analyses comparing the effectiveness of particular types or forms
of treatment.
The goal of averaging multiple effect sizes within a treatment comparison and of averaging multiple comparisons within a study was to ensure
that estimates of error variance were always based on independent observations. In most cases, this procedure meant that no matter how
many effect sizes were derived from a study, the study provided only one
observation in each of our statistical analyses. However, there were two
types of analysis in which a study was represented by more than one
score. The first and simpler case occurred when a group of studies provided an effect size for each of two categories being compared. For example, some of the studies provided an effect size at both posttreatment
and follow-up, and we wanted to use these studies to assess whether
posttreatment and follow-up effects differed. In such cases, the dependency among pairs of scores was accommodated in a straightforward
fashion by means of a paired t test.
A more complex situation arose, however, when some studies in an
analysis provided an effect size for only one of the categories being compared, but other studies offered effect sizes for more than one of the
categories. For instance, most studies examined either an individual
treatment or a group treatment, but a small number of studies examined both treatment formats. In the overall analysis comparing the
effectiveness of individual and group therapies, we permitted studies
with information about both treatment formats to provide one observation to each treatment category. Thus, a study might contribute more
than one score to the analysis but would provide no more than one
observation to each treatment category. In this type of analysis, the estimate of variability within groups would still be based on independent
observations. However, given the clustering of effect sizes drawn from
the same study, the estimate of variability between groups would be
smaller than that expected if all the observations were from separate
studies (see Kenny & Judd, 1986, Equation 1). Therefore, these hybrid
analyses represent conservative inferential tests.
The studies under review varied widely in terms of the number of
clients in their samples. Analysis revealed that studies with fewer clients
yielded larger overall treatment effects than studies with more clients,
r(35) = —.32, p = .06. The most likely explanation for this pattern is
that it reflects a bias in the publication of research findings. That is,
journal editors may be more willing to publish studies that report statistically significant differences, and in studies with small samples only
large treatment effects are statistically significant.
One might also expect that the variability in the findings of studies
would decrease as the sample sizes of the studies increased. To evaluate
this issue, we conducted a regression analysis in which we used the sample size of the study to predict the overall treatment effect. As expected,
the squared residuals or errors from this regression analysis decreased as
the sample size of the study increased, r(35) = -.35, p = .03. Therefore,
studies with larger samples not only provided more conservative estimates of treatment effects but also offered more reliable estimates.
Given this difference in the reliability of studies with different sample
sizes, we conducted the following analyses using a weighted least squares
procedure (e.g., see Chatterjee & Price, 1977, chap. 5) in which the
results of each study were weighted by the number of clients in that
study. Thus, the means, standard deviations, and correlations reported
in the following sections are all weighted estimates.
Results
Table 2 presents the effect sizes for comparisons between psychotherapy and no treatment. As can be seen, posttreatment
outcomes of depressed clients receiving psychotherapy were almost three fourths of a standard deviation better than the outcomes of individuals not receiving therapy. These posttreatment results were quite similar to those obtained with followup measures of outcome, f(44) = 0.20, p = .8. In fact, among the
nine studies reporting both posttreatment and follow-up data,
effect sizes derived from posttreatment measures (M = 0.74,
SD = 0.61) were almost identical to effect sizes derived from
the same measures at follow-up (M = 0.70, SD = 0.67), paired
t(S) = 0.79, p = .5. Moreover, the posttreatment effect size for a
study was highly predictive of the effect size for that study at
follow-up, r(l) = .98, p < .001. Given this similarity between
posttreatment and follow-up findings, the results presented in
the remainder of the review are based only on posttreatment
data.
The no-treatment control groups in the preceding analyses
included not only wait-list controls but also placebo treatments,
which provided attention or used placebo drugs. Analysis indicated that effect sizes based on comparisons to wait-list controls
were larger than those based on comparisons to placebo controls, t(36) = 2.24, p = .03. As can be seen in Table 3, effect sizes
for comparisons to wait-list controls differed reliably from zero,
but effect sizes for comparisons to placebo controls did not. In
view of these findings, we chose not to combine the two types
of control groups in the remaining analyses. Instead, effect sizes
involving comparisons to untreated individuals are based on
the more frequently occurring type of control, the wait-list
group.
Table 3
Efficacy of Psychotherapy Based on Comparisons to Wait-List
Controls and Placebo Controls
Effect size
Control group
AT of
studies
SD
0.69
29
0.84*
Waitlist
0.52
Placebo
9
0.28
Note. Means and standard deviations are based on weighted least
squares analyses in which effect sizes were weighted by sample size.
*p<.05.
35
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
Type of treatment. Table 4 presents the effects of four specific
types of psychotherapy: (a) cognitive, (b) behavioral, (c) cognitive-behavioral, and (d) general verbal. For each type of therapy,
treated clients improved more than did wait-list controls. The
effect sizes for some therapies were larger than for others, but
this variation was not reliable, F(3,35) = 0.92, p = .4. However,
it is difficult to judge the relative efficacy of the treatments from
this evidence, because the results were drawn from different sets
of studies. For instance, studies evaluating general verbal therapy might have used clients that were less amenable to treatment than the client samples used in studies of other types of
therapy. Thus, differences in effect size may be due to variations
across studies in background variables such as sample characteristics rather than to differences in therapeutic efficacy.
Investigations that directly compare two or more types of
treatment provide a better assessment of relative efficacy, because background variables are held relatively constant across
the therapies. Thus, any differences between the treatment
groups can be attributed more clearly to differences in the therapies themselves. Table 5 presents the results of studies that directly compared two or more therapies. Effect sizes were coded
so that positive numbers indicate that the first therapy in the
comparison was more effective, whereas negative numbers indicate that the second therapy was more effective. As the effect
sizes in the table reveal, outcomes for cognitive therapy did not
appear to differ reliably from those of either a strictly behavioral
approach or a cognitive-behavioral intervention. However, cognitive-behavioral treatments did appear to produce more improvement than behavioral methods used in isolation. Furthermore, general verbal therapy seemed less effective than all three
of the alternative therapies with which it was compared.
Although these results suggest that some forms of therapy
may be more successful in alleviating depression than others,
they do not take into account the possible impact of the researcher's allegiance. Previous research has indicated that when
one therapy is compared with another, the preference of the researcher can have a substantial impact on outcome (Berman et
al., 1985; Smith et al., 1980, chap. 5). To determine whether a
similar effect is operative in the depression literature, two independent raters judged investigator allegiance on a 5-point scale
for each comparison between treatments. These ratings were
based on comments provided in the introductory section of
each study. For example, some researchers outlined specific
hypotheses concerning the relative efficacy of two treatments
under comparison. Even when explicit hypotheses were not re-
Table 4
Efficacy of Different Types of Psychotherapy
Effect size
Therapy
# of
studies
M
SD
Cognitive
Behavioral
Cognitive-behavioral
General verbal
8
12
13
6
0.96*
1.02*
0.85*
0.49*
0.64
0.79
0.74
0.35
Note. Means and standard deviations are based on weighted least
squares analyses in which effect sizes were weighted by sample size.
*/><.05.
Table 5
Direct Comparisons Between Different Types of Psychotherapy
Effect size"
Not
Comparison
studies
M
SD
Estimate if no
allegiance11
Cognitive vs. behavioral
Cognitive vs. cognitivebehavioral
Behavioral vs. cognitivebehavioral
Cognitive vs. general
verbal
Behavioral vs. general
verbal
Cognitive-behavioral vs.
general verbal
12
0.12
0.33
0.12(0.09)
4
-0.03
0.24
-0.03(0.12)
8
-0.24*
0.20
-0.16(0.10)
7
0.47*
0.30
-0.15(0.20)
14
0.27*
0.33
0.15(0.13)
8
0.37*
0.38
0.09(0.27)
Note. Means, standard deviations, and standard errors are based on
weighted least squares analyses in which effect sizes were weighted by
sample size.
* Positive numbers indicate that the first therapy in the comparison was
more effective; negative numbers indicate that the second therapy in the
comparison was more effective.
b
Standard error of the estimated effect size is in parentheses.
*/><.05.
ported, allegiance could often be rated on the basis of subtler
cues. For example, one type of treatment might be described
in considerable detail, whereas the other was only briefly and
incompletely discussed. Alternatively, the introductory comments may have presented a balanced picture of the two therapies under comparison. This evenhanded treatment often occurred when researchers of differing theoretical persuasions collaborated. In such cases, the researchers were rated as having
no preference.3
Examination of the allegiance ratings suggested that investigators frequently had substantial theoretical preferences for
particular forms of therapy. Furthermore, these ratings of allegiance were highly correlated with the results of direct comparisons between treatments, r(2B) - .58, p < .001. Thus, therapies
that were preferred by the investigator tended to achieve better
results than the less favored therapies with which they were
compared.
Although this finding might indicate that researcher allegiance had a sizable impact on a study's outcome, the opposite
could have also been true. That is, study outcome could have
influenced ratings of allegiance, because these ratings were
based on introductory comments that were often prepared after
a study was completed. To clarify the nature of the relation between allegiance and outcome, we conducted a second analysis
that was restricted to comparisons in which the researchers' allegiance could be verified on the basis of their comments in
some previous publication. In this smaller set of 13 studies, preferred therapies once again tended to produce more improvement than their less favored counterparts, r( 11) = .51, p = .08.
3
The two raters were able to judge allegiance on the basis of the introductory comments with a high degree of reliability. In fact, the intraclass correlation for the reliability of the mean of the two raters was
.95.
36
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
Therefore, it seems unlikely that the relation between outcome
and allegiance is solely attributable to the tendency of researchers to write introductions tailored to their results.
What would be the relative efficacy of the different therapies
in the absence of an allegiance effect? To answer this question,
we used the regression equation for the relation between effect
size and allegiance rating to estimate the effect size that would
have occurred if the investigators had no preference for either
of the therapies being compared. The estimated mean effect size
and its standard error are presented for each comparison in the
final two columns of Table 5. It should be noted that the effect
sizes for comparisons between cognitive and behavioral therapy
and for those between cognitive and cognitive-behavioral approaches required no correction, because for these comparisons the investigators indicated no preference. However, for the
four remaining comparisons in which there were substantial
theoretical preferences, the effect sizes representing differences
in treatment efficacy were substantially reduced. In fact, once
the influence of investigator allegiance was removed, there remained no evidence for the relative superiority of any one type
of therapy.
Some might wonder whether the lack of differences in treatment outcome is simply a reflection of incorrect treatment implementation. When studies do not verify that competing therapies were administered as prescribed, similarity in outcome
may indicate either that the therapies were equally effective or
that they were not implemented properly. In recognition of this
issue, a number of investigators have used videotapes, audiotapes, or observers to ensure that the therapies were delivered
in an appropriate manner. If the similarity in outcome across
therapies is simply a reflection of poor treatment implementation, then one might expect the use of these monitoring procedures to increase the differences observed between treatments.
However, in direct comparisons between treatments, our analyses revealed no reliable difference in the absolute magnitude of
effect sizes from 14 studies that used these monitoring procedures (M = 0.27, SD = 0.15) and 16 studies that did not (M =
0.38, SD = 0.41),/(28)= 1.04,p=.3.
Another method of attempting to standardize treatment delivery involves the use of therapy manuals. In some of the studies, therapists conducted treatment according to detailed manuals; in other studies, no such formalized therapy program was
used. When treatments were directly compared with one another, however, the absolute magnitude of effect sizes from 11
studies that used formal manuals (M = 0.28, SD = 0.30) did
not differ reliably from the absolute magnitude of effect sizes
from 14 studies in which no manuals were used (M = 0.34,
SD = 0.18), f(23) = 0.55, p = .6. Similar results were observed
when we examined studies comparing treated groups with waitlist controls. The effect sizes of 14 studies using manual-driven
therapies (M = 0.82, SD = 0.64) did not differ systematically
from the effect sizes of 17 studies for which no manual was developed (M = 0.84, SD = 0.74), t(29) = 0.07, p = .9. Although
the use of treatment manuals has increased in recent years,
these data provide no indication that their use either increases
therapeutic efficacy or allows for a finer differentiation of the
relative effectiveness of treatments.
The length of treatment might also affect the outcome of psychotherapy. To address this issue, we correlated the results of
Table 6
Efficacy of Individual and Group Psychotherapy
Effect size
Not
Format
studies
M
SD
Individual therapy
Group therapy
16
15
0.83*
0.84*
0.77
0.60
Note. Means and standard deviations are based on weighted least
squares analyses in which effect sizes were weighted by sample size.
*p<.05.
comparisons between treated groups and wait-list controls with
the length of treatment. The analyses revealed no reliable relation between effect size and either the number of weeks of treatment, r(26) = —.09, p = .6, or the total number of sessions,
r(27) = -. 11, p = .6. Thus, there is no indication from the studies that the benefits of psychotherapy increase systematically as
the length of treatment increases.
Treatment format. In some investigations clients were
treated individually, whereas in others clients were seen in
groups. To determine the possible influence of these different
formats on outcome, we analyzed studies comparing individual
therapy with a wait-list control and studies comparing group
therapy with a wait-list control separately. As indicated in Table
6, effect sizes for the group and individual methods were quite
similar, t(29) = 0.03, p > .9. Further data relevant to this point
were provided by five investigations that directly compared individual and group approaches. From this smaller set of studies,
effect sizes were calculated so that a positive number indicated
that individual therapy was more effective and a negative number indicated that group therapy was superior. Although individual therapy tended to produce better results than group therapy in these five studies, the effect sizes representing the difference between the two methods (M = 0.31, SD = 0.35) did not
differ reliably from zero, f(4) = 1.95, p = . 1.
In studies comparing group therapy with a wait-list control
group, the number of clients per therapy group ranged from 3
to 12 (M = 7). Some might suspect that smaller therapy groups
would be more effective than larger groups. However, we detected no consistent relation when we correlated group size with
the outcome of therapy, r( 12) = -. 13, p = .1. Thus, small treatment groups did not appear to be any more effective than therapy groups with more clients.
Type of outcome. A wide range of outcome measures was
used in these studies. Some of the instruments were designed
specifically to assess changes in depressive symptomatology.
Others measured constructs that were less directly related to
depression. To determine whether the effects of therapy varied
across these different domains of outcome, effect sizes representing comparisons of treated clients with wait-list controls
were calculated separately for measures of depression and measures of other constructs. These results are presented in the top
half of Table 7. For both types of outcome, there was a reliable
effect of therapy, with treated clients improving more than waitr
list controls, but the difference between the effect sizes for the
different measures was not reliable, t(41) = 1.37, p = .2. However, in the 20 studies that included both types of measures,
37
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
Table 7
Efficacy of Psychotherapy for Different
of Outcome Measures
Types
Effect size
Measure
Not
M
SD
characteristic
studies
Focus
Depression
29
0.93*
0.76
0.64*
0.63
20
Other
Source
Self-report
29
0.85*
0.73
7
0.81*
0.77
Observer
Note. Means and standard deviations are based on weighted least
squares analyses in which effect sizes were weighted by sample size.
*p<.05.
there was a tendency for effect sizes to be larger for measures
designed to assess depression (M = 0.85, SD = 0.79) than for
other instruments (M = 0.64, SD = 0.63), paired t( 19) = 1.81,
p = .09.
Estimates of the effects of therapy might also vary with the
perspective of the evaluator. In the studies under review, two
sources of outcome were commonly used: client self-report and
reports from independent observers. As indicated in the bottom
half of Table 7, the effect of therapy was substantial from both
perspectives and quite similar for clients and observers, f(34) =
0.13, p = .9. In seven studies that included both sources of outcome, effect sizes from observer measures (M = 0.81, SD =
0.77) were somewhat larger than those for self-report instruments (M = 0.70, SD = 0.96), but the difference was once again
not reliable, paired f(6) = 0.56, p - .6.
Sample characteristics. The client samples used in these
studies varied on a number of dimensions. For example, some
studies used female clients exclusively, whereas others included
both sexes. The mean age of the client sample also varied across
studies. It may be that the effects of psychotherapy on depression differ depending on these client characteristics. However,
analysis indicated that these demographic variables had no appreciable relation with treatment outcome. In comparisons between treated groups and wait-list controls, effect sizes did not
vary reliably as a function of the percentage of female clients,
r(24) = -.19, p = .3, or the age of the sample, r(19) = -.23,
p = .3.
Initial severity of depression is another factor that might
moderate the efficacy of treatment. One indicator of initial severity is the pretreatment score of clients on the BDI. When we
examined comparisons of treated groups with wait-list controls, however, we did not find a systematic relation between
effect size and initial scores on the BDI, r(2Q) = .25, p = .3.
Thus, there was no convincing evidence that clients scoring
higher on this measure responded differently to treatment than
did those who initially reported less depression.
Another indicator of the severity of depression is the diagnostic status of the sample. To determine the impact of this variable, we divided studies into two groups: those that used formal
diagnostic criteria in the screening process and those that used
less rigorous client selection procedures. Effect sizes for comparisons of treated clients with wait-list controls are reported
for both types of studies in the top half of Table 8. As this table
reveals, the effects of therapy were virtually identical for formally diagnosed samples and for those that were not formally
diagnosed, f(27) = 0.08, p > .9.
A final sample characteristic that might affect study outcome
is the referral source of the clients. In most of the studies, clients
were obtained through media announcements. Only rarely
were traditional outpatient referral procedures used. To assess
the influence of referral source, we compared the results of
studies using outpatients with those of studies using student or
community volunteers. Effect sizes for comparisons of treated
clients with wait-list controls are presented for the three groups
of studies in the bottom half of Table 8. As the values suggest,
there was no reliable relation between effect size and referral
source, 7^2,23) = 0.34, p = .1. Thus, treatment effects in studies using solicited clients did not differ from those observed in
studies using traditionally referred outpatients.
Other characteristics of the studies. In addition to using
different client samples, the reviewed studies also varied on a
number of other characteristics. One such factor that can have
a major impact on the validity of a study is the degree to which
attrition rates differ for treatment and control groups. When the
proportion of dropouts is unequal, the therapy manipulation
becomes confounded with these differences in attrition. Thus,
a treatment may appear ineffective simply because it has lost
more of its successful cases. If this were the case, then one would
expect studies with greater attrition in treatment groups than
in control groups to show smaller treatment effects. To investigate this possibility, we created a differential attrition variable
for each study, which was defined as the percentage of attrition
from treatment groups minus the percentage of attrition from
wait-list control conditions. When we correlated the treatment
effect reported in a study with this differential attrition variable,
no reliable relation emerged, r(20) = —. 13, p = .6.
The level of training of therapists is another factor that might
affect the outcome of a study. If student therapists are less skilled
than professionals, studies using students might provide an inadequate test of treatment benefits. However, our results argue
against this possibility. In fact, using comparisons of treated clients with wait-list controls, we found that the 19 studies in
which graduate students served as therapists yielded larger
effect sizes (M = 1.05, SD = 0.73) than those obtained from the
Table 8
Efficacy of Psychotherapy for Different Client Samples
Effect size
AT of
studies
Sample characteristic
SD
M
Formal diagnosis
Yes
0.86*
8
0.85
No
21
0.84*
0.64
Referral source
Solicited from community
0.64
13
0.75*
Solicited students
11
0.95*
0.82
Outpatients
0.74
2
1.13
Note. Means and standard deviations are based on weighted least
squares analyses in which effect sizes were weighted by sample size.
*/?<.05.
38
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
2 studies in which all therapists were fully trained professionals
(M = 0.65, SD = 0.05), although this numerical difference was
not statistically significant, t(\9) = 0.70, p = .5.
Analysis 2: Comparisons With Pharmacotherapy
As our analyses indicate, psychotherapy appears effective in
helping depressed clients. A further question, though, is the
effectiveness of psychotherapy relative to the leading alternative
treatment, pharmacotherapy. Several previous quantitative reviews have suggested that psychotherapy compares favorably
with drug therapy for the treatment of depression (e.g., Quality
Assurance Project, 1983; Smith et al., 1980; Steinbrueck et al.,
1983). However, in each of these reviews the efficacy of psychotherapy and pharmacotherapy was estimated from different sets
of studies. Thus, differences in variables such as client characteristics may account for the apparent variation in treatment
effects. In addition, the benefits of combining psychotherapy
with pharmacotherapy have received scant attention. Although
several reviewers using narrative or box-score techniques have
suggested that the combination approach may be more effective
than either treatment alone (e.g., Conte et al., 1986; Weissman,
1979), the evidence has not been subjected to rigorous inferential analysis.
Fortunately, a number of studies have recently been published in which both psychotherapy and pharmacotherapy or
the combination of the two have been evaluated for the treatment of depression. The following analyses provide a quantitative summary of this research.
Method
To locate appropriate studies, we conducted computerized searches
of both medical and psychological data bases and examined the references of published reviews and outcome studies in this area. In addition,
recent volumes of relevant journals were searched on an issue-by-issue
basis.4 We selected studies containing at least one of the following comparisons: (a) psychotherapy versus pharmacotherapy, (b) a psychotherapy-pharmacotherapy combination versus psychotherapy alone, and
(c) a combination treatment versus pharmacotherapy alone. Comparisons with "treatment as usual" were excluded when the treatment-asusual condition could not be clearly classified as psychotherapy, pharmacotherapy, or the combination of the two. In addition, when a psychotherapy-pharmacotherapy combination was evaluated, we required
that it be compared with one of its components. For example, a comparison between cognitive therapy and the combination of psychodynamic
therapy and imipramine would have been excluded, because the difference between the two conditions reflects not only the addition of pharmacotherapy but also a difference in the type of psychotherapy. Other
inclusion criteria paralleled those outlined earlier for the larger psychotherapy review. Thus, we included only studies using adult outpatients
suffering from unipolar depression, and once again the focus was on
treatments with individual or group formats.
Using these criteria we were able to locate 15 studies that provided
the relevant comparisons between psychotherapy, pharmacotherapy, or
the combination of the two.5 (Appendix B presents a list of the references for the studies.) Many of these studies were reported in more than
one publication. In fact, on average more than two reports were published from each research project. Thus, the actual number of separate
studies in this area is far smaller than might be expected on the basis of
the number of publications.
The client samples used in these studies were similar to those of the
psychotherapy studies reviewed earlier in terms of age (M =41.1 years),
gender (M = 68.5% female), and initial BDI score (M = 25.5). In all
except one of the studies, clients were required to meet formal diagnostic criteria for depression. Only three of these studies (21%) reported
that clients were solicited through media announcements. Five studies
(36%) used traditionally referred outpatients, and the remaining seven
studies (47%) either included both solicited and traditionally referred
clients or did not report the source of their subjects.
A variety of psychotherapeutic approaches were evaluated in this set
of studies. Cognitive-behavioral therapy was used in eight investigations
(53%), and behavioral treatments were tested in three studies (20%).
One study (7%) examined purely cognitive methods, whereas four studies (27%) assessed a general verbal therapy. In one study, the type of
therapy was not reported.
In 12 of the studies (80%), tricyclic antidepressants constituted the
pharmacotherapy under investigation, with amitriptyline most often
administered. One study (7%) examined a tetracyclic, and another 2
(13%) evaluated benzodiazepines. A final study allowed prescribing
physicians to use any of a variety of psychoactive drugs. Mean dosages
of these medications were generally within accepted therapeutic ranges,
although usually at the lower end. Treatment lasted on average 12 weeks,
allowing in most cases an adequate trial of the medication.
For each study, effect sizes representing differences in treatment outcome were calculated by using the procedures outlined for the earlier
psychotherapy analyses. Once again, multiple outcomes within a treatment comparison and multiple treatment comparisons within a study
were averaged to ensure that estimates of error were based on independent observations. As before, a weighted least squares procedure was
used to compensate for differences in the sample sizes of the studies.
Results
The top half of Table 9 presents the results of comparisons
between psychotherapy, pharmacotherapy, and the combination of both. As can be seen, psychotherapy appeared more
effective than pharmacotherapy in the treatment of depression.
However, the outcomes of a combination approach did not
differ systematically from the outcomes of either treatment
alone. Thus, the benefits of psychotherapy and pharmacotherapy do not appear to be additive.
Could the superiority of psychotherapy over pharmacotherapy be an artifact of researcher allegiance? To investigate this
possibility, we had two independent raters judge researcher allegiance by using the procedures outlined in the preceding psychotherapy review.6 We then used these allegiance ratings to
predict the effect size that would have occurred if the research-
4
The issue-by-issue search was conducted for the following journals:
American Journal of Psychiatry, Archives of General Psychiatry, Behavior Therapy, Behaviour Research and Therapy, British Journal of Psychiatry, Cognitive Therapy and Research, Journal of Consulting and
Clinical Psychology, and Psychopharmacology Bulletin.
5
One additional study (Daneman, 1961), which provided a comparison between a psychotherapy-pharmacotherapy combination and pharmacotherapy alone, was eliminated from our analyses because the effect
size derived from the study was more than nine standard deviations
above the mean of other studies providing this same comparison. We
also conducted our analyses with this study included. Although inclusion of the study altered effect size values, it did not change any of the
conclusions.
6
The intraclass correlation for the reliability of the mean of the two
raters was .85.
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
Table 9
Relative Efficacy of Psychotherapy, Pharmacotherapy,
and the Combination of Both
Effect size' Estimate if
no
studies M SD allegiance"
Af of
Comparison
All pharmacological treatments
Psychotherapy vs. drug therapy
8 0.13* 0.12 0.07(0.04)
Combination vs. psychotherapy
12 0.01 0.25 -0.01(0.08)
Combination vs. drug therapy
5 0.17 0.24 -0.05(0.21)
Tricyclic antidepressants only
7 0.12*0.13 0.07(0.04)
Psychotherapy vs. tricyclics
Combination vs. psychotherapy
9 0.02 0.25 -0.05(0.08)
Combination vs. tricyclics
4 0.15 0.27 -0.05(0.26)
Note. Means, standard deviations, and standard errors are based on
weighted least squares analyses in which effect sizes were weighted by
sample size.
' Positive numbers indicate that thefirsttherapy in the comparison was
more effective; negative numbers indicate that the second therapy in the
comparison
was more effective.
b
Standard error of the estimated effect size is in parentheses.
*p<.05.
ers had no preferences for particular treatments. The results of
this analysis are presented in the last two columns of Table 9.
As can be seen, once the preferences of the researchers were
taken into account, there was no reliable difference between
psychotherapy and pharmacotherapy.
In some of the studies clients received tricyclic antidepressants, whereas in others less established drug treatments were
evaluated. Some might argue that pharmacotherapy would appear more effective if only studies using tricyclics were considered. However, as can be seen in the bottom section of Table 9,
when we restricted our analyses to such studies the same pattern
of results emerged. Psychotherapy initially appeared to be superior to treatment with tricyclics, but when the effect of allegiance was removed the difference was not reliable. Moreover,
the combination of tricyclics and psychotherapy was no more
effective than either treatment alone.
Analysis 3: Clinical Significance of
Psychotherapy Effects
Does psychotherapy cure depression? Although our results
indicate that clients treated with psychotherapy function substantially better than untreated controls and as well as clients
treated with drugs, it is unclear to what extent former clients
resemble their nondepressed peers. Treated clients may function much better than controls but remain more depressed than
healthy individuals with no history of depression.
To evaluate this issue, we followed an approach suggested by
Jacobson, Follette, and Revenstorf (1984) and implemented in
the Nietzel et al. (1987) meta-analysis of the depression treatment literature. This method involves comparing treated clients with norms derived from nondisturbed peers. To serve as
the basis for this comparison, an outcome measure must have
been widely used in both normative studies and treatment studies. As did Nietzel et al. (1987), we found that the BDI met this
criterion admirably.
39
Method
We were able to identify a total of 39 studies reporting normative data
on the BDI. (See Appendix C for a list of the references for the studies.)
Of these studies, 20 had been previously identified by Nietzel et al.
(1987) in their search of the 1978 through September 1985 issues of
six psychology journals.7 We located another 19 normative studies by
extending Nietzel et al.'s journal search from September 1985 through
November 1987 and by reviewing bibliographies of articles on the assessment of depression. These normative studies included samples that
were similar in terms of their age and gender to the clients treated in the
psychotherapy studies. The average normative sample consisted of 64%
female subjects, and the average age of those in the sample was 30 years.
In 28 of the normative studies, the BDI was administered to a group
of subjects unscreened for mental health difficulties. Examples include
investigations that randomly sampled community residents and those
using data from large groups of university students. In 12 studies, norms
were reported for individuals who had been screened on some measure
of mental health. An example is a study that obtained data from subjects who did not meet Research Diagnostic Criteria for depression.
Not surprisingly, BDI scores for nondistressed samples were lower than
those derived from samples of the general population, /(38) = 2.14, p =
.04. For this reason, we calculated norms separately for nondistressed
and general population samples.
Results
Table 10 presents BDI scores for both types of norms along
with pretreatment and posttreatment scores derived from 22
studies that compared clients who received psychotherapy with
no-treatment controls. As can be seen, both treated clients and
untreated controls were moderately depressed at pretest. After
treatment, clients receiving psychotherapy were functioning at
the lower end of the mild range of depression, whereas controls
scored at the upper end of the mildly depressed range. Despite
this improvement relative to untreated controls, clients who
had received psychotherapy remained more depressed at the
end of therapy than subjects in the general population studies,
;(23.7) = 4.95, p < .001.8 A similar picture emerged when the
more stringent norms for the nondistressed samples were used.
Once again, treated clients were more depressed at the end of
therapy than subjects in the normative studies, f(31.4) = 6.23,
p < .001. Thus, although psychotherapy brings about a reduction in depression, clients who have received therapy can still
7
Two studies included in the Nietzel et al. (1987) normative group
(Kilpatrick-Tabak & Roth, 1978; Oliver & Burkham, 1979) were excluded from our analyses because they reported means but no standard
deviations for the BDI. Another six studies (Atkinson & Rickel, 1984;
Cutrona, 1984; Fiore, Becker, & Coppel, 1983; Jacob, Turner, Szekely,
& Eidelman, 1983; Nielsen & Williams, 1980; O'Hara, Rehm, & Campbell, 1982) were excluded because subjects in these studies were experiencing situational stresses such as medical illness.
8
As the standard deviations in Table 10 suggest, the variance of posttreatment BDI scores for treated clients among the psychotherapy studies was considerably larger than the variance of such scores among either
the general population studies, F(21,27) = 12.11, p< .01, or the studies
of nondepressed samples, F(2l, 11) = 4.73, p< .05. In assessing differences between the means of these groups of studies, we therefore used
the separate-variance approximation of a t value and Satterthwaite's
procedure for adjusting degrees of freedom (see Winer, 1971, Section
1.9).
40
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
Table 10
Beck Depression Inventory (BDI) Scores for
Normative Samples, Clients Treated With
Psychotherapy, and Untreated Controls
BDI"
Group
Normative samples
General population
Nondistressed
Treated clients
Pretreatment
Posttreatment
Untreated controls
Pretreatment
Posttreatment
Not
studies
M
SD
28
12
7.0
4.9
1.3
2.0
22
22
21.8
11.8
4.7
4.4
22
22
20.7
18.1
5.0
5.2
Note. Means and standard deviations are based on weighted least
squares analyses in which BDI scores were weighted by sample size.
" BDI scores from 10 to 20 indicate mild depression; from 20 to 30,
moderate depression; and over 30, severe depression (Kendall, Hollon,
Beck, Hammen, & Ingram, 1987).
be distinguished from healthy individuals who have not sought
treatment.
Jacobson et al. (1984) have suggested that one way to judge
the impact of treatment is to express client functioning relative
to the mean and standard deviation of a population of healthy
normal individuals. To generate such estimates, we pooled the
standard deviations of BDI scores reported in studies of the general population and those reported in studies of nondepressed
samples, and we used the pooled values as estimates of the dispersion of individual scores in each normative population. We
then expressed the BDI scores of clients treated with psychotherapy in terms of these normative standard deviations. The
results, presented in Figure 1, indicate that before treatment
clients reported substantially more depression on the BDI than
did normative subjects. After receiving psychotherapy, however,
clients had moved to within one standard deviation of the mean
of the general population and within one and a half standard
deviations of the mean of nondistressed samples.
treatments (e.g., Bowers &Clum, 1988; Casey &Berman, 1985;
Dush et al., 1983; Miller & Berman, 1983; Shapiro & Shapiro,
1983). However, these previous reviews were not restricted to
depressed samples. It may be that depression is particularly responsive to common curative factors occurring in both psychotherapy and placebo treatments. If so, then the specific procedures that define the type of treatment may be less important
in alleviating depression than previously recognized.
Despite their improvement, clients treated with psychotherapy remain distinguishable from healthy controls. When we
compared the BDI scores of treated clients with norms derived
from the general population and from nondistressed samples,
treated clients were found to be more depressed at the end of
therapy than both normative groups. However, the magnitude
of improvement over the course of therapy appeared impressive. On average, clients were functioning within one standard
deviation of the general population after treatment, compared
with a pretreatment difference of more than two standard deviations. Such a change clearly represents substantial improvement.
The research evidence demonstrates, furthermore, that the
benefits of psychotherapy for depression are not short-lived. In
those studies that included a follow-up assessment, improvement at posttreatment was quite similar to that observed at a
later follow-up. Not only does this finding emphasize the enduring nature of the changes that occur during treatment, but it
(Posttreatment - |i+OJto)
•o
I
ao
Q.
~co
(Pretreatment - u + 2.4<T)
I
Beck Depression Inventory
Discussion
As our analyses demonstrate, clinical research has firmly established the efficacy of psychological interventions for depression. In studies comparing psychotherapy with no treatment,
there was a positive effect of therapy that was both statistically
reliable and substantial in magnitude. Despite the relatively
brief duration of treatment provided in most of the reviewed
studies, the average client was functioning at treatment termination approximately three fourths of a standard deviation better than untreated controls.
It remains unclear, though, which aspects of psychotherapeutic treatment were responsible for producing this improvement.
When the effects of psychotherapy were compared with those
of placebo treatments, no reliable differences emerged. Such results are somewhat surprising, because previous quantitative reviews of the psychotherapy outcome literature have generally
shown that the effects of therapy are larger than those of placebo
(Posttreatment - \a + \
^Pretreatment - n + 3.4q)
I
Beck Depression Inventory
Figure 1. Typical pretreatment and posttreatment scores on the Beck
Depression Inventory relative to the distribution of individual scores
observed in the general population and in nondistressed samples. (Frequency distributions of general population and nondistressed samples
are hypothetical normal approximations.)
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
argues against the widely held assumption that follow-up assessment is essential because of unusually high relapse rates for depression. As in the more general review by Nicholson and Herman (1983), our results suggest that follow-up findings often
add little to information obtained at the end of therapy. Thus,
rather than including a costly follow-up assessment, psychotherapy researchers might justifiably choose to invest resources in
other aspects of their design. For example, researchers could
focus on obtaining larger samples of clients and therapists,
thereby increasing both the power and generalizability of their
analyses.
Although all forms of psychotherapy were more effective than
no treatment, our initial analyses indicated that there might be
some variation in the efficacy of different types of therapy. Such
a difference has also been reported in a recent review by Dobson
(1989), who concluded that cognitive therapy for depression
was superior to other approaches. However, after more detailed
analyses, we discovered that differences in the efficacy of different treatments may be an artifact of the theoretical allegiances
of the researchers conducting these studies. When a particular
type of therapy was preferred by an investigator, it tended to
produce more favorable results than the treatment with which
it was being compared. This pattern is similar to findings reported by Herman et al. (1985) and Smith et al. (1980, chap. 5),
who also found that the outcome of a psychotherapy study varied according to the allegiance of the researcher. Furthermore,
the effect is consistent with Ro'senthaTs (1969, 1976) wellknown research demonstrating the influence of experimenter
expectations. To take this influence into account, we used regression analysis to predict the outcomes that would have occurred under conditions of no allegiance. This predictive analysis indicated that if all of the researchers had been neutral, there
would have been no reliable differences in the effectiveness of
the various types of therapy.
The allegiance of the investigator also appeared to play a role
in comparisons between psychological and pharmacological
treatments for depression. Although psychotherapy initially appeared more effective than drugs, this difference was not reliable once we adjusted for the influence of investigator allegiance. Our analyses also indicated that combinations of psychotherapy with pharmacotherapy were not systematically
more effective than either of the treatments alone. Furthermore,
the efficacy of psychotherapy and pharmacotherapy remained
comparable even when we restricted the analysis to studies using tricyclic antidepressants, the best established of the pharmacological interventions for depression.
Advocates of drug therapy might argue that the effectiveness
of pharmacotherapy was underestimated in these studies for a
variety of reasons. For example, the clinical response to pharmacotherapy might have been better had higher levels of the
drugs been prescribed. In addition, some have suggested that
the efficacy of pharmacological treatment depends partly on client characteristics such as chronicity and the presence of endogenous symptoms (e.g., Becker & Schuckit, 1978; Weissman,
1981). Thus, given different dosages or client samples, pharmacotherapy could have appeared more effective. Alternatively,
advocates of particular psychological or pharmacological approaches may argue that our classification system was overly
broad, thereby obscuring important differences in the efficacy
41
of specific forms of treatment. Although such arguments have
merit, the onus would appear to be on the proponents of such
hypotheses to offer convincing empirical evidence in support of
them.
The relative efficacy of group and individual therapy has only
recently become an issue in depression research, perhaps because group therapy for depression has traditionally been considered a contradiction in terms. It has been feared that the special needs of depressed clients could not be met in a group setting and that depressed clients might act as a burden on a group
because of their withdrawal, pessimism, and self-absorption.
However, our analyses indicated that both group and individual
treatment formats produced more improvement than no treatment, and the effects of the two approaches were comparable.
Moreover, there was no evidence that treatment groups with
many members were less effective than smaller therapy groups.
Although this lack of difference in the efficacy of individual and
group treatments is consistent with the findings of several previous reviews (Miller & Herman, 1983; Shapiro & Shapiro, 1982;
Smith et al., 1980, chap. 5), it conflicts with Nietzel et al.'s
(1987) and Dush et al.'s (1983) evidence for the superiority of
individual therapy. The difference between our results and those
of Nietzel et al. are particularly striking, because their review
also focused specifically on the treatment of depression. However, it should be noted that our reviews differ not only in terms
of the studies included but also in the outcome measures analyzed.
Although variations in sample characteristics such as age,
sex, and initial symptom severity have often been offered as explanations for inconsistencies in the results of different studies,
we found no evidence that any of these variables were systematically related to outcome. Furthermore, studies that included
only clients meeting formal diagnostic criteria for depression
generated results that were virtually identical to those observed
in studies using less rigorous client selection procedures. The
reliance on self-report measures for client selection has often
been criticized on the grounds that these measures were not
developed as diagnostic tools. However, our data suggest that, at
least for the problem of depression, treatment findings based
on self-identified clients will mirror the results obtained from
formally diagnosed samples.
Self-report instruments have been viewed as suspect not only
when used for client selection but also when used as an outcome
measure. The underlying assumption is that clients may overestimate the benefits of treatment, whereas measures obtained
from observers will exhibit less bias. However, in a recent metaanalysis, Lambert, Hatch, Kingston, and Edwards (1986) found
greater indications of change on an interviewer measure of depression than on two self-report scales. In our analysis of a
broad range of self-report and observer measures, we found that
treatment effects derived from self-report instruments were
comparable in size to those obtained from observer measures.
As in the Lambert et al. (1986) review, there was no evidence
that measures based on the self-report of the client yielded
overly positive estimates of treatment efficacy.
Our analysis did suggest that the effects of therapy differed
according to the content of the outcome measures. Instruments
specifically designed to assess depression tended to produce
larger effects than measures assessing other constructs. Such
42
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
findings might suggest that the psychotherapies used in these
studies are more effective for depression than for other types
of complaints. However, it must be remembered that we only
included studies with clients whose primary problem was depression. Therefore, an equally plausible explanation for these
results is that change is more likely to occur for the problems
or symptoms that initially prompt a client to enter treatment.
In our review, we included studies that varied considerably in
terms of the quality of their research design. Critics of quantitative review methods (e.g., Wilson, 1985; Wilson & Rachman,
1983) have objected to this practice, arguing that the quality of
the research should be taken into account in the selection of
studies. Like other quantitative reviewers (e.g., Glass et al.,
1981, chap. 2), we consider the impact of study quality to be an
empirical question. In fact, our analyses generally failed to detect differences as a function of research design characteristics.
For instance, findings from studies with differential rates of attrition did not differ reliably from the findings of studies in
which dropout rates were equivalent across treatment groups.
Moreover, consistent with the findings of Herman and Norton
(1985), investigations that used fully trained professionals as
therapists did not report greater treatment benefits than studies
that relied on graduate students to administer therapy. Furthermore, studies that used treatment manuals or that monitored
therapist behavior during treatment yielded findings similar to
those of studies that did not include these procedures for ensuring correct treatment delivery.
Of the various study characteristics examined, only the number of clients per study related to treatment outcome. As our
preliminary analysis of the psychotherapy data indicated, studies with fewer clients yielded larger effects than studies with
many clients. The most likely explanation for this relation is
that it reflects a publication bias. That is, journals are more
likely to publish studies that achieve statistically significant
differences between groups, and small studies need larger effects
to achieve statistical significance. One implication of this finding is that literature reviews of published research (whether they
are narrative summaries or quantitative reviews such as our
own) may yield overly generous estimates of treatment effects.
Because of the way in which we conducted our analyses, however, such a publication bias was probably minimized in this
review. For example, our statistical analyses gave less weight to
studies with smaller samples, and it is these small-sample studies that are most likely to inflate estimates of treatment effects.
In addition, our procedures for estimating treatment effects
were often conservative. Thus, when results were reported simply as nonsignificant, we estimated that effect size to be zero.
Whether or not these conservative estimation procedures fully
offset the influence of a publication bias, our review can at least
be viewed as a conservative summary of the published evidence
on the efficacy of psychotherapy for depression.
The findings in this review are consistent with more general
quantitative reviews of the psychotherapy outcome literature
(e.g., Smith et al., 1980). As with these other reviews, we found
few differences in the efficacy of various therapeutic methods.
However, unlike these general analyses, ours provided an evaluation of comparative, treatment efficacy within a single diagnostic category. Thus, the similarity in the effects of various therapies cannot be attributed to their application to different noso-
logical groups. Even when the client sample was restricted to
the diagnosis of depression, different types of psychotherapy
yielded equivalent benefits. Furthermore, the similarity between our findings and those of general reviews including other
diagnostic groups raises the provocative possibility that diagnosis has much less impact on treatment response than previously
assumed. Psychotherapy may produce similar benefits not only
across different types of therapy but also across different types
of clients. Thus, extensive efforts to establish the diagnostic purity of samples for psychotherapy research may be unnecessary.
The evidence from the review, coupled with that of previous
quantitative reviews, indicates that additional comparisons between competing therapies for depression are not likely to prove
informative. Comparative studies are useful primarily in establishing new approaches as viable therapeutic options rather
than in elucidating the mechanisms through which they effect
change. As Kazdin (1981) has noted in a commentary on depression outcome research: "Relatively little is known about
major treatment contenders, including cognitive therapy and
behavior therapy, and the comparative studies do not shed light
on whether these individual treatments operate in many of the
ways proposed on conceptual grounds" (p. 319). Indeed, the
key question now is not whether psychotherapy works for the
treatment of depression but rather how these therapies produce
their benefits.
In a recent analysis of the cognitive theory of depression, Hollon, DeRubeis, and Evans (1987) pointed out that the similarity
in the effects of differing therapies does not necessarily disprove
the causal significance of cognitive changes as a mediator of recovery from depression. The possibility remains that all therapies are effective because all of them activate the cognitive
changes that are the specific target of cognitive therapy. Although this argument could be true, it must be recognized that
a comparable position could be advanced with equal plausibility by proponents of behavior therapy. That is, therapies may
promote recovery by producing behavioral changes, whether by
accident or by design. In this sense, cognitive or behavioral
changes may constitute nonspecific effects that occur naturally
over the course of therapy regardless of whether they are the
identified goal of treatment. Alternatively, factors such as clients' expectations of improvement, their acceptance of the therapeutic rationale, or the quality of the therapeutic relationship
may be the central mechanisms through which therapeutic
change occurs (e.g., see Frank, 1982; Zeiss, Lewinsohn, & Munoz, 1979).
Surprisingly little research has been directed toward evaluating these hypothesized mediators of change. Curative factors
common to all therapies have generally been mentioned only as
post hoc explanations in comparative studies finding nonsignificant differences in treatment outcomes. If researchers are to
progress in their understanding of how psychotherapy benefits
clients, these common factors may need to become a more central focus of future research efforts.
References
American Psychiatric Association. (1980). Diagnostic and Statistical
Manual of Mental Disorders (3rd ed.). Washington, DC: Author.
Atkinson, A. K., & Rickel, A. U. (1984). Postpartum depression in primiparous parents. Journal of Abnormal Psychology, 93, 115-119.
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
Beck, A. T. (1967). Depression: Clinical, experimental, and therapeutic
aspects. New \brk: Harper & Row.
Beck, A. X, Ward,. C. H., Mendelson, M., Mock, J., & Erbaugh, J.
(1961). An inventory for measuring depression. Archives of General
Psychiatry, 4, 561-571.
Becker, J., & Schuckit, M. A. (1978). The comparative efficacy of cognitive therapy and pharmacotherapy in the treatment of depressions.
Cognitive Therapy and Research, 2, 193-197.
Belsher, G., & Costello, C. G. (1988). Relapse after recovery from unipolar depression: A critical review. Psychological Bulletin, 104, 8496.
Berman, J. S., Miller, R. C., & Massman, P. J. (1985). Cognitive therapy
versus systematic desensitization: Is one treatment superior? Psychological Bulletin, 97, 451-461.
Berman, J. S., & Norton, N. C. (1985). Does professional training make
a therapist more effective? Psychological Bulletin, 98, 401-407.
Blaney, P. H. (1981). The effectiveness of cognitive and behavioral therapies. In L. P. Rehm (Ed.), Behavior therapy for depression: Present
status and future directions (pp. 1-32). New \brk: Academic Press.
Bowers, T. G., & Clum, G. A. (1988). Relative contribution of specific
and nonspecific treatment effects: Meta-analysis of placebo-controlled behavior therapy research. Psychological Bulletin, 103, 315323.
Boyd, J. H., & Weissman, M. M. (1981). Epidemiology of affective disorders: A reexamination and future directions. Archives of General
Psychiatry, 38, 1039-1046.
Brown, R. A., & Lewinsohn, P. M. (1984). A psychoeducational approach to the treatment of depression: Comparison of group, individual, and minimal contact procedures. Journal of Consulting and Clinical Psychology, 52, 774-783.
Casey, R. J., & Berman, J. S. (1985). The outcome of psychotherapy
with children. Psychological Bulletin, 98, 388-400.
Chatterjee, S., & Price, B. (1977). Regression analysis by example. New
York: Wiley.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences
(rev. ed.). New York: Academic Press.
Conte, H. R., Plutchik, R., Wild, K. V., & Karasu, T. B. (1986). Combined psychotherapy and pharmacotherapy for depression. Archives
of General Psychiatry, 43, 471-479.
Cutrona, C. E. (1984). Social support and stress in the transition to
parenthood. Journal of Abnormal Psychology, 93, 378-390.
Daneman, E. A. (1961). Imipramine in office management of depressive reactions (a double blind clinical study). Diseases of the Nervous
System, 22, 213-217.
Dempsey, P. (1964). A unidimensional depression scale for the MMPI.
Journal of Consulting Psychology, 28, 364-370.
DeRubeis, R. J., & Hollon, S. D. (1981). Behavioral treatment of affective disorders. In L. Michelson, M. Hersen, & S. M. Turner (Eds.),
Future perspectives in behavior therapy (pp. 103-129). New \brk: Plenum Press.
Dobson, K. S. (1989). A meta-analysis of the efficacy of cognitive therapy for depression. Journal of Consulting and Clinical Psychology, 57,
414-419.
Dush, D. M., Hirt, M. L., & Schroeder, H. (1983). Self-statement modification with adults: A meta-analysis. Psychological Bulletin, 94,408422.
Elkin, I., Parloff, M. B., Hadley, S. W., & Autry, J. H. (1985). NIMH
Treatment of Depression Collaborative Research Program: Background and research plan. Archives of General Psychiatry, 42, 305316.
Emmelkamp, P. M. G. (1986). Behavior therapy with adults. In S. L.
Garfield & A. E. Bergin (Eds.), Handbook of psychotherapy and behavior change (3rd ed., pp. 385-442). New York: Wiley.
Feighner, J. P., Robins, E., Guze, S. B., Woodruff, R. A., Winokur, G., &
43
Munoz, R. (1972). Diagnostic criteria for use in psychiatric research.
Archives of General Psychiatry, 26, 57-63.
Fiore, J., Becker, J., & Coppel, D. B. (1983). Social network interactions: A buffer or stress? A merican Journal ofCommunity Psychology,
11, 423-439.
Fiske, D. W. (1983). The meta-analytic revolution in outcome research.
Journal of Consulting and Clinical Psychology, 51, 65-70.
Fleming, B. M., & Thornton, D. W. (1980). Coping skills training as
a component in the short-term treatment of depression. Journal of
Consulting and Clinical Psychology, 48, 652-654.
Frank, J. D. (1982). Therapeutic components shared by all psychotherapies. In J. H. Harvey & M. M. Parks (Eds.), Psychotherapy research
and behavior change (Master Lecture Series, Vol. 1, pp. 9-37). Washington, DC: American Psychological Association.
Fuchs, C. Z., & Rehm, L. P. (1977). A self-control behavior therapy
program for depression. Journal of Consulting and Clinical Psychology, 45, 206-215.
Glass, G. V., McGaw, B., & Smith, M. L. (1981). Meta-analysis in social
research. Beverly Hills, CA: Sage.
Haggard, E. A. (1958). Intraclass correlation and the analysis of variance. New \fork: Dryden Press.
Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery, and Psychiatry, 23, 56-62.
Hathaway, S. R., & McKinley, J. C. (1967). Minnesota Multiphasic Personality Inventory: Manual (rev. ed.). New \brk: Psychological Corporation.
Hedges, L. V. (1982). Estimation of effect size from a series of independent experiments. Psychological Bulletin, 92,490-499.
Hodgson, J. W. (1981). Cognitive versus behavioral-interpersonal approaches to the group treatment of depressed college students. Journal of Counseling Psychology, 28, 243-249.
Hollon, S. D. (1981). Comparisons and combinations with alternative
approaches. In L. P. Rehm (Ed.), Behavior therapy for depression:
Present status andfuture directions (pp. 33-71). New \brk: Academic
Press.
Hollon, S. D., DeRubeis, R. J., & Evans, M. D. (1987). Causal mediation of change in treatment for depression: Discriminating between
nonspecificity and noncausality. Psychological Bulletin, 102, 139149.
Jacob, R. G., Turner, S. M., Szekely, B. C., & Eidelman, B. H. (1983).
Predicting outcome of relaxation therapy in headaches: The role of
"depression." Behavior Therapy, 14, 457-465.
Jacobson, N. S. (1981). The assessment of overt behavior. In L. P. Rehm
(Ed.), Behavior therapy for depression: Present status and future directions (pp. 279-300). New York: Academic Press.
Jacobson, N. S., Follette, W. C., & Revenstorf, D. (1984). Psychotherapy
outcome research: Methods for reporting variability and evaluating
clinical significance. Behavior Therapy, 15, 336-352.
Kazdin, A. E. (1981). Outcome evaluation strategies. In L. P. Rehm
(Ed.), Behavior therapy for depression: Present status and future directions (pp. 317-336). New York: Academic Press.
Kendall, P. C., Hollon, S. D., Beck, A. T., Hammen, C. L., & Ingram,
R. E. (1987). Issues and recommendations regarding use of the Beck
Depression Inventory. Cognitive Therapy and Research, 11, 289299.
Kenny, D. A., & Judd, C. M. (1986). Consequences of violating the
independence assumption in analysis of variance. Psychological Bulletin, 99, 422-431.
Kessler, K. A. (1978). Tricyclic anti-depressants: Mode of action and
clinical use. In M. A. Lipton, A. DiMascio, & K. F. Killam (Eds.),
Psychopharmacology: A generation of progress (pp. 1289-1302). New
York: Raven Press.
Kilpatrick-Tabak, B., & Roth, S. (1978). An attempt to reverse perfor-
44
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
mance deficits associated with depression and experimentally induced helplessness. Journal of Abnormal Psychology, 87, 141-154.
Klerman, G. L. (1978). Long-term treatment of affective disorders. In
M. A. Lipton, A. DiMascio, & K. F. Killam (Eds.), Psychopharmacology: A generation of progress (pp. 1303-1311). New 'York: Raven
Press.
Klerman, G. L., Weissman, M. M., Rounsaville, B. J., & Chevron, E. S.
(1984). Interpersonal psychotherapy of depression. New \fork: Basic
Books.
Kovacs, M. (1980). The efficacy of cognitive and behavior therapies for
depression. American Journal of Psychiatry, 137, 1495-1501.
Lambert, M. J., Hatch, D. R., Kingston, M. D., & Edwards, B. C.
(1986). Zung, Beck, and Hamilton Rating Scales as measures of treatment outcome: A meta-analytic comparison. Journal of Consulting
and Clinical Psychology, 54, 54-59.
LaPointe, K. A., & Rimm, D. C. (1980). Cognitive, assertive, and insight-oriented group therapies in the treatment of reactive depression
in women. Psychotherapy: Theory, Research, and Practice, 17, 312321.
Ledwidge, B. (1978). Cognitive behavior modification: A step in the
wrong direction? Psychological Bulletin, 85, 353-375.
Lewinsohn, P. M. (1974). Clinical and theoretical aspects of depression.
In K. S. Calhoun, H. E. Adams, & K. M. Mitchell (Eds.), Innovative
treatment methods in psychopathology (pp. 63-120). New "York: Wiley.
Lewinsohn, P. M., Steinmetz, J. L., Antonuccio, D., & Teri, L. (1985).
Group therapy for depression: The Coping with Depression course.
International Journal of Mental Health, 13, 8-33.
Lubin, B. (1965). Adjective checklists for measurement of depression.
Archives of General Psychiatry, 12, 57-62.
McLean, P. D. (1981). Matching treatment to patient characteristics
in an outpatient setting. In L. P. Rehm (Ed.), Behavior therapy for
depression: Present status and future directions (pp. 197-207). New
York: Academic Press.
McLean, P. (1982). Behavioral therapy: Theory and research. In A. J.
Rush (Ed.), Short-term psychotherapies for depression: Behavioral,
interpersonal, cognitive, andpsychodynamic approaches (pp. 19-49).
New York: Guilford Press.
Miller, R. C., & Berman, J. S. (1983). The efficacy of cognitive behavior
therapies: A quantitative review of the research evidence. Psychological Bulletin, 94, 39-53.
Nicholson, R. A., & Berman, J. S. (1983). Is follow-up necessary in
evaluating psychotherapy? Psychological Bulletin, 93, 261-278.
Nielsen, A. C., & Williams, T. A. (1980). Depression in ambulatory
medical patients: Prevalence by self-report questionnaire and recognition by nonpsychiatric physicians. Archives of General Psychiatry,
37, 999-1004.
Nietzel, M. T, Russell, R. L., Hemmings, K. A., & Gretter, M. L.
(1987). Clinical significance of psychotherapy for unipolar depression: A meta-analytic approach to social comparison. Journal of Consulting and Clinical Psychology, 55, 156-161.
O'Hara, M. W., Rehm, L. P., & Campbell, S. B. (1982). Predicting depressive symptomatology: Cognitive-behavioral models and postpartum depression. Journal of Abnormal Psychology, 91, 457-461.
Oliver, J. M., & Burkham, R. (1979). Depression in university students:
Duration, relation to calendar time, prevalence, and demographic
correlates. Journal of Abnormal Psychology, 88, 667-670.
Quality Assurance Project. (1983). A treatment outline for depressive
disorders. Australian and New Zealand Journal of Psychiatry, 17,
129-146.
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale
for research in the general population. Applied Psychological Measurement, 7,385-401.
Rehm, L. P., & Kornblith, S. J. (1979). Behavior therapy for depression:
A review of recent developments. In M. Hersen, R. M. Eisler, & P. M.
Miller (Eds.), Progress in behavior modification (pp. 277-318). New
York: Academic Press.
Rosenthal, R. (1969). Interpersonal expectations: Effects of the experimenter's hypothesis. In R. Rosenthal & R. L. Rosnow (Eds.), Artifact
in behavioral research (pp. 181 -277). New York: Academic Press.
Rosenthal, R. (1976). Experimenter effects in behavioral research (enlarged ed.). New York: Irvington.
Rush, A. J. (1982). Diagnosing depressions. In A. J. Rush (Ed.), Shortterm psychotherapies for depression: Behavioral, interpersonal, cognitive, and psychodynamic approaches (pp. 1-18). New York: Guilford
Press.
Sanchez, V. C., Lewinsohn, P. M., & Larson, D. W. (1980). Assertion
training: Effectiveness in the treatment of depression. Journal of Clinical Psychology, 36, 526-529.
Shapiro, D. A., & Shapiro, D. (1982). Meta-analysis of comparative
therapy outcome studies: A replication and refinement. Psychological
Bulletin, 92, 581-604.
Shapiro, D. A., & Shapiro, D. (1983). Comparative therapy outcome
research: Methodological implications of meta-analysis. Journal of
Consulting and Clinical Psychology, 51, 42-53.
Shaw, B. F. (1977). Comparison of cognitive therapy and behavior therapy in the treatment of depression. Journal of Consulting and Clinical
Psychology, 45,543-551.
Smith, M. L., Glass, G. V., & Miller, T. I. (1980). The benefits of psychotherapy. Baltimore: Johns Hopkins University Press.
Spitzer, R. L., Endicott, J., & Robins, E. (1978). Research Diagnostic
Criteria: Rationale and reliability. Archives of General Psychiatry, 35,
773-782.
Steinbrueck, S. M., Maxwell, S. E., & Howard, G. S. (1983). A metaanalysis of psychotherapy and drug therapy in the treatment of unipolar depression with adults. Journal of Consulting and Clinical Psychology, 51, 856-863.
Thompson, L. W., & Gallagher, D. (1984). Efficacy of psychotherapy in
the treatment of late-life depression. Advances in Behaviour Research
and Therapy, 6, 127-139.
Weissman, M. M. (1979). The psychological treatment of depression:
Evidence for the efficacy of psychotherapy alone, in comparison with,
and in combination with pharmacotherapy. Archives of General Psychiatry, 36, 1261-1269.
Weissman, M. M. (1981). Antidepressants and psychotherapy in depression. Advances in Biological Psychiatry, 7, 230-239.
Weissman, M. M., & Klerman, G. L. (1977). The chronic depressive
in the community: Unrecognized and poorly treated. Comprehensive
Psychiatry, /S, 523-531.
Weissman, M. M., & Myers, J. K. (1978). Affective disorders in a US urban
community: The use of Research Diagnostic Criteria in an epidemiological survey. Archives of General Psychiatry, 35, 1304-1311.
Whitehead, A. (1979). Psychological treatment of depression: A review.
Behaviour Research and Therapy, 17, 495-509.
Wilson, G. T. (1985). Limitations of meta-analysis in the evaluation of the
effects of psychological therapy. Clinical Psychology Review, 5, 35-47.
Wilson, G. T, & Rachman, S. J. (1983). Meta-analysis and the evaluation of psychotherapy outcome: Limitations and liabilities. Journal
of Consulting and Clinical Psychology, 51, 54-64.
Winer, B. J. (1971). Statistical principles in experimental design (2nd
ed.). New York: McGraw-Hill.
Zeiss, A. M., Lewinsohn, P. M., & Munoz, R. F. (1979). Nonspecific
improvement effects in depression using interpersonal skills training,
pleasant activity schedules, or cognitive training. Journal of Consulting and Clinical Psychology, 47, 427-439.
Zung, W. W. K. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70.
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
45
Appendix A
Sources for the Psychotherapy Studies Included in the Review
Information about a study was sometimes obtained from more than
one published report. In those cases, the other reports providing information about the study are listed in parentheses. Also indicated in parentheses are those cases in which a report provided information about
more than one study.
Barrera, M. (1979). An evaluation of a brief group therapy for depression. Journal of Consulting and Clinical Psychology, 47, 413-415.
Beach, S. R. H., & O'Leary, K. D. (1986). The treatment of depression
occurring in the context of marital discord. Behavior Therapy, 17,
43-49.
Beck, J. T, & Strong, S. R. (1982). Stimulating therapeutic change with
interpretations: A comparison of positive and negative connotation.
Journal of Counseling Psychology, 29, 551-559.
Bellack, A. S., Hersen, M., & Himmelhoch, J. (1981). Social skills training compared with pharmacotherapy and psychotherapy in the treatment of unipolar depression. American Journal of Psychiatry, 138,
1562-1567. (Information about this study was also obtained from
Bellack etal., 1983a, 1983b; Hersen etal., 1984.)
Bellack, A. S., Hersen, M., & Himmelhoch, J. M. (1983a). A comparison of social-skills training, pharmacotherapy, and psychotherapy for
depression. Behaviour Research and Therapy, 21, 101-107. (Information about this study was also obtained from Bellack et al., 1981;
1983b; Hersen etal., 1984.)
Bellack, A. S., Hersen, M., & Himmelhoch, J. M. (1983b). Social skills
training for unipolar depression. Psychotherapy in Private Practice,
1, 9-13. (Information about this study was also obtained from Bellack etal., 1981,1983a; Hersen etal., 1984.)
Beutler, L. E., Scogin, F, Kirkish, P., Schretlin, D., Corbishley, A.,
Hamblin, D., Meredith, K., Potter, R., Bamford, C. R., & Levenson,
A. I. (1987). Group cognitive therapy and alprazolam in the treatment of depression in older adults. Journal of Consulting and Clinical
Psychology, 55, 550-556.
Brown, R. A., & Lewinsohn, P. M. (1984). A psychoeducational approach to the treatment of depression: Comparison of group, individual, and minimal contact procedures. Journal of Consulting and Clinical Psychology, 52, 774-783.
Catanese, R. A., Rosenthal, T. L., & Kelley, J. E. (1979). Strange bedfellows: Reward, punishment, and impersonal distraction strategies in
treating dysphoria. Cognitive Therapy and Research, 3, 299-305.
Comas-Diaz, L. (1981). Effects of cognitive and behavioral group treatment on the depressive symptomatology of Puerto Rican women.
Journal of Consulting and Clinical Psychology, 49, 627-632.
Conoley, C. W., & Garber, R. A. (1985). Effects of refraining and selfcontrol directives on loneliness, depression, and controllability. Journal of Counseling Psychology, 32, 139-142.
Davis, G. R., Armstrong, H. E., Donovan, D. M., & Temkin, N. R.
(1984). Cognitive-behavioral treatment of depressed affect among epileptics: Preliminary findings. Journal of Clinical Psychology, 40,
930-935.
DiMascio, A., Weissman, M. M., Prusoff, B. A., Neu, C., Zwilling, M.,
& Klerman, G. L. (1979). Differential symptom reduction by drugs
and psychotherapy in acute depression. Archives of General Psychiatry, 36, 1450-1456. (Information about this study was also obtained
from Herceg-Baron et al., 1979; Prusoff et al., 1980; Weissman et
al., 1981; Weissman, Prusoff, & DiMascio, 1979; Weissman, Prusoff,
DiMascio et al., 1979; Zuckerman etal., 1980.)
Feldman, D. A., Strong, S. R., & Danser, D. B. (1982). A comparison
of paradoxical and nonparadoxical interpretations and directives.
Journal of Counseling Psychology, 29, 572-579.
Fleming, B. M., & Thornton, D. W. (1980). Coping skills training as
a component in the short-term treatment of depression. Journal of
Consulting and Clinical Psychology, 48, 652-654.
Fry, P. S. (1983). Structured and unstructured reminiscence training
and depression among the elderly. Clinical Gerontologist, 1, 15-37.
Fry, P. S. (1984). Cognitive training and cognitive-behavioral variables
in the treatment of depression in the elderly. Clinical Gerontologist,
3, 25-45.
Fuchs, C. Z., & Rehm, L. P. (1977). A self-control behavior therapy
program for depression. Journal of Consulting and Clinical Psychology, 45, 206-215.
Gallagher, D. (1981). Behavioral group therapy with elderly depressives:
An experimental study. In D. Upper & S. Ross (Eds.), Behavioral
group therapy, 1981 (pp. 187-224). Champaign, IL: Research Press.
Gallagher, D. E., & Thompson, L. W. (1982). Treatment of major depressive disorder in older adult outpatients with brief psychotherapies. Psychotherapy: Theory, Research and Practice. 19,482-490.
Gardner, P., & Oei, T. P. S. (1981). Depression and self-esteem: An investigation that used behavioral and cognitive approaches to the treatment of clinically depressed clients. Journal of Clinical Psychology,
37, 128-135.
Gold, S. R., Jarvinen, P. J., & Teague, R. G. (1982). Imagery elaboration
and clarity in modifying college students' depression. Journal of Clinical Psychology, 38, 312-314.
Graff, R. W., Whitehead, G. I., & LeCompte, M. (1986). Group treatment with divorced women using cognitive-behavioral and supportive-insight methods. Journal of Counseling Psychology, 33, 276-281.
Hammen, C. L., & Glass, D. R. (1975). Depression, activity, and evaluation of reinforcement. Journal of Abnormal Psychology, 84,118-721.
(This report provided information about two different experiments.
Each of these experiments was treated as a separate study in the analyses.)
Hayman, P. M., & Cope, C. S. (1980). Effects of assertion training on
depression. Journal of Clinical Psychology, 36, 534-543.
Herceg-Baron, R. L., Prusoff, B. A., Weissman, M. M., DiMascio, A.,
Neu, C., & Klerman, G. L. (1979). Pharmacotherapy and psychotherapy in acutely depressed patients: A study of attrition patterns in a
clinical trial. Comprehensive Psychiatry. 20, 315-325. (Information
about this study was also obtained from DiMascio et al., 1979; Prusoff et al., 1980; Weissman et al., 1981; Weissman, Prusoff, & DiMascio, 1979; Weissman, Prusoff, DiMascio et al., 1979; Zuckerman et
al., 1980.)
Hersen, M., Bellack, A. S., Himmelhoch, J. M., & Thase, M. E. (1984).
Effects of social skill training, amitriptyline, and psychotherapy in
unipolar depressed women. Behavior Therapy, 15, 21-40. (Information about this study was also obtained from Bellack et al., 1981,
1983a, 1983b.)
Hodgson, J. W. (1981). Cognitive versus behavioral-interpersonal approaches to the group treatment of depressed college students. Journal of Counseling Psychology, 28, 243-249.
Jarvinen, P. J., & Gold, S. R. (1981). Imagery as an aid in reducing
depression. Journal of Clinical Psychology, 37, 523-529.
Kelly, F. D., Dowd, E. T, & Duffey, D. K. (1983). A comparison of
cognitive and behavioural intervention strategies in the treatment of
depression. British Journal of Cognitive Psychotherapy, 1, 51-58.
Klein, M. H., Greist, J. H., Gurman, A. S., Neimeyer, R. A., Lesser,
D. P., Bushnell, N. J., & Smith, R. E. (1985). A comparative outcome
study of group psychotherapy vs. exercise treatments for depression.
International Journal of Mental Health, 13, 148-177.
Kornblith, S. J., Rehm, L. P., O'Hara, M. W, & Lamparski, D. M.
(1983). The contribution of self-reinforcement training and behav-
46
L. ROBINSON, J. HERMAN, AND R. NEIMEYER
ioral assignments to the efficacy of self-control therapy for depression.
Cognitive Therapy and Research, 7, 499-527.
LaPointe, K, A., & Rimm, D. C. (1980). Cognitive, assertive, and insightoriented group therapies in the treatment of reactive depression in
women. Psychotherapy: Theory, Research and Practice, 17,312-321.
Larcombe, N. A., & Wilson, P. H. (1984). An evaluation of cognitivebehaviour therapy for depression in patients with multiple sclerosis.
British Journal of Psychiatry, 145, 366-371.
McLean, P. D., & Hakstian, A. R. (1979). Clinical depression: Comparative efficacy of outpatient treatments. Journal of Consulting and
Clinical Psychology, 47, 818-836.
McNamara, K., & Horan, J. J. (1986). Experimental construct validity
in the evaluation of cognitive and behavioral treatments for depression. Journal of Counseling Psychology, 33, 23-30.
Neimeyer, R. A., Heath, A. E., & Strauss, J. (1985). Personal reconstruction during group cognitive therapy for depression. In F. R. Epting &
A. W. Landneld (Eds.), Anticipating personal construct theory (pp.
180-197). Lincoln: University of Nebraska Press.
Nezu, A. M. (1986). Efficacy of a social problem-solving therapy approach for unipolar depression. Journal of Consulting and Clinical
Psychology, 54, 196-202.
Padfield, M. (1976). The comparative effects of two counseling approaches
on the intensity of depression among rural women of low socioeconomic
status. Journal of'Counseling Psychology, 23, 209-214.
Pecheur, D. R., & Edwards, K. J. (1984). A comparison of secular and
religious versions of cognitive therapy with depressed Christian college students. Journal of Psychology and Theology, 12, 45-54.
Propst, L. R. (1980). The comparative efficacy of religious and nonreligious imagery for the treatment of mild depression in religious individuals. Cognitive Therapy and Research, 4, 167-178.
Prusoff, B. A., Weissman, M. M., Klerman, G. L., & Rounsaville, B. J.
(1980). Research diagnostic criteria subtypes of depression: Their
role as predictors of differential response to psychotherapy and drug
treatment. Archives of General Psychiatry, 3 7, 796-801. (Information
about this study was also obtained from DiMascio et al., 1979; Herceg-Baron et al., 1979; Weissman et al., 1981; Weissman, Prusoff, &
DiMascio, 1979; Weissman, Prusoff, DiMascio et al., 1979; Zuckermanetal., 1980.)
Rehm, L. P., Fuchs, C. Z., Roth, D. M., Kornblith, S. J., & Romano,
J. M. (1979). A comparison of self-control and assertion skills treatments of depression. Behavior Therapy, 10, 429-442.
Rehm, L. P., Kaslow, N. J., & Rabin, A. S. (1987). Cognitive and behavioral targets in a self-control therapy program for depression. Journal
of Consulting and Clinical Psychology, 55, 60-67.
Rehm, L. P., Kornblith, S. J., O'Hara, M. W., Lamparski, D. M., Romano, J. M., & Volkin, J. I. (1981). An evaluation of major components in a self-control therapy program for depression. Behavior
Modification, 5, 459-489.
Ross, M., & Scott, Mf. (1985). An evaluation of the effectiveness of individual and group cognitive therapy in the treatment of depressed patients in an inner city health centre. Journal of the Royal College of
General Practitioners, 35, 239-242.
Rude, S. S. (1986). Relative benefits of assertion or cognitive self-control
treatment for depression as a function of proficiency in each domain.
Journal of Consulting and Clinical Psychology, 54, 390-394.
Rush, A. J., & Watkins, J. T. (1981). Group versus individual cognitive therapy: A pilot study. Cognitive Therapy and Research, 5,
95-103.
Sallis, J. F, Lichstein, K. L., Clarkson, A. D., Stalgaitis, S., & Campbell,
M. (1983). Anxiety and depression management for the elderly. International Journal of Behavioral Geriatrics, 1, 3-12.
Sanchez, V. C., Lewinsohn, P. M., & Larson, D. W. (1980). Assertion
training: Effectiveness in the treatment of depression. Journal of Clinical Psychology, 36, 526-529.
Schmidt, M. M., & Miller, W. R. (1983). Amount of therapist contact
and outcome in a multidimensional depression treatment program.
Acta Psychiatrica Scandinavica, 67, 319-332.
Shaw, B. F. (1977). Comparison of cognitive therapy and behavior therapy in the treatment of depression. Journal of Consulting and Clinical
Psychology, 45, 543-551.
Shipley, C. R., & Fazio, A. F. (1973). Pilot study of a treatment for
psychological depression. Journal of Abnormal Psychology, 82, 372376. (This report provided information about two different experiments. Each of these experiments was treated as a separate study in
the analyses.)
Steuer, J. L., Mintz, J., Hammen, C. L., Hill, M. A., Jarvik, L. F, McCarley, T, Motoike, P., & Rosen, R. (1984). Cognitive-behavioral and
psychodynamic group psychotherapy in treatment of geriatric depression. Journal of Consulting and Clinical Psychology, 52, 180-189.
Taylor, F. G., & Marshall, W. L. (1977). Experimental analysis of a cognitive-behavioral therapy for depression. Cognitive Therapy and Research, 1, 59-72.
Teri, L., & Lewinsohn, P. M. (1986). Individual and group treatment of
unipolar depression: Comparison of treatment outcome and identification of predictors of successful treatment outcome. Behavior
Therapy, 77,215-228.
Thompson, L. W, & Gallagher, D. (1984). Efficacy of psychotherapy in
the treatment of late-life depression. Advances in Behaviour Research
and Therapy, 6, 127-139.
Turner, R. W, Ward, M. F, & Turner, D. J. (1979). Behavioral treatment
for depression: An evaluation of therapeutic components. Journal of
Clinical Psychology, 35, 166-175.
Weissman, M. M., Klerman, G. L., Prusoff, B. A., Sholomskas, D., &
Padian, N. (1981). Depressed outpatients: Results one year after
treatment with drugs and/or interpersonal psychotherapy. Archives of
General Psychiatry, 38, 51-55. (Information about this study was also
obtained from DiMascio et al., 1979; Herceg-Baron et al., 1979; Prusoff et al., 1980; Weissman, Prusoff, & DiMascio, 1979; Weissman,
Prusoff, DiMascio et al., 1979; Zuckerman et al., 1980.)
Weissman, M. M., Prusoff, B. A., & DiMascio, A. (1979). Research
directions on comparisons of drugs and psychotherapy in depression.
Psychopharmacology Bulletin, 15, 19-21. (Information about this
study was also obtained from DiMascio et al., 1979; Herceg-Baron et
al., 1979; Prusoffetal., 1980; Weissman etal., 1981; Weissman, Prusoff, DiMascio et al., 1979; Zuckerman etal., 1980.)
Weissman, M. M., Prusoff, B. A., DiMascio, A., Neu, C., Goklaney, M.,
& Klerman, G. L. (1979). The efficacy of drugs and psychotherapy
in the treatment of acute depressive episodes. American Journal of
Psychiatry, 136, 555-558. (Information about this study was also obtained from DiMascio et al., 1979; Herceg-Baron et al., 1979; Prusoff
et al., 1980; Weissman et al., 1981; Weissman, Prusoff, & DiMascio,
1979; Zuckerman etal., 1980.)
Wilson, P. H. (1982). Combined pharmacological and behavioural treatment of depression. Behaviour Research and Therapy, 20,173-184.
Wilson, P. H., Goldin, J. C., & Charbonneau-Powis, M. (1983). Comparative efficacy of behavioral and cognitive treatments of depression.
Cognitive Therapy and Research, 7, 111-124.
Zeiss, A. M., Lewinsohn, P. M., & Munoz, R. F. (1979). Nonspecific
improvement effects in depression using interpersonal skills training,
pleasant activity schedules, or cognitive training. Journal of Consulting and Clinical Psychology, 47, 427-439.
Zuckerman, D. M., Prusoff, B. A., Weissman, M. M., & Padian, N. S.
(1980). Personality as a predictor of psychotherapy and pharmacotherapy outcome for depressed outpatients. Journal of Consulting and
Clinical Psychology, 48, 730-735. (Information about this study was
also obtained from DiMascio et al., 1979; Herceg-Baron et al., 1979;
Prusoff et al., 1980; Weissman et al., 1981; Weissman, Prusoff, & DiMascio, 1979; Weissman, Prusoff, DiMascio etal., 1979.)
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
47
Appendix B
Sources for the Pharmacotherapy Studies Included in the Review
Information about a study was sometimes obtained from more than
one published report. In those cases, the other reports providing information about the study are listed in parentheses.
Beck, A. T, Hollon, S. D., Young, J. E., Bedrosian, R. C, & Budenz, D.
(1985). Treatment of depression with cognitive therapy and amitriptyline. Archives of 'General Psychiatry, 42, 142-148.
Bellack, A. S., Hersen, M., & Himmelhoch, J. (1981). Social skills training compared with pharmacotherapy and psychotherapy in the treatment of unipolar depression. American Journal of Psychiatry, 138,
1562-1567. (Information about this study was also obtained from
Bellack etal., 1983a, 1983b; Hersen etal., 1984.)
Bellack, A. S., Hersen, M., & Himmelhoch, J. M. (1983a). A comparison of social-skills training, pharmacotherapy, and psychotherapy for
depression. Behaviour Research and Therapy, 21, 101-107. (Information about this study was also obtained from Bellack et al., 1981,
1983b; Hersen etal., 1984.)
Bellack, A. S., Hersen, M., & Himmelhoch, J. M. (1983b). Social skills
training for unipolar depression. Psychotherapy in Private Practice,
1, 9-13. (Information about this study was also obtained from Bellack etal., 1981,1983a; Hersen etal., 1984.)
Beutler, L. E., Scogin, F, Kirkish, P., Schretlin, D., Corbishley, A.,
Hamblin, D., Meredith, K., Potter, R., Bamford, C. R., & Levenson,
A. I. (1987). Group cognitive therapy and alprazolam in the treatment of depression in older adults. Journal of Consulting and Clinical
Psychology, 55, 550-556.
Blackburn, I. M., & Bishop, S. (1981). Is there an alternative to drugs
in the treatment of depressed ambulatory patients? Behavioural Psychotherapy, 9, 96-104. (Information about this study was also obtained from Blackburn & Bishop, 1983; Blackburn et al., 1981,
1986.)
Blackburn, I. M., & Bishop, S. (1983). Changes in cognition with pharmacotherapy and cognitive therapy. British Journal of Psychiatry,
143, 609-617. (Information about this study was also obtained from
Blackburn* Bishop, 1981; Blackburn etal., 1981, 1986.)
Blackburn, I. M., Bishop, S., Glen, A. I. M., Whalley, L. J., & Christie,
J. E. (1981). The efficacy of cognitive therapy in depression: A treatment trial using cognitive therapy and pharmacotherapy, each alone
and in combination. British Journal of Psychiatry, 139,181-189. (Information about this study was also obtained from Blackburn &
Bishop, 1981,1983; Blackburn etal., 1986.)
Blackburn, I. M., Eunson, K. M., & Bishop, S. (1986). A two-year naturalistic follow-up of depressed patients treated with cognitive therapy,
pharmacotherapy and a combination of both. Journal of Affective
Disorders, 10, 67-75. (Information about this study was also obtained from Blackburn & Bishop, 1981, 1983; Blackburn et al.,
1981.)
Covi, L., & Lipman, R. S. (1987). Cognitive behavioral group psychotherapy combined with imipramine in major depression. Psychopharmacology Bulletin, 23, 173-176.
DiMascio, A., Klerman, G. L., Weissman, M. M., Prusoff, B. A., Neu,
C., & Moore, P. (1979). A control group for psychotherapy research
in acute depression: One solution to ethical and methodologic issues.
Journal of Psychiatric Research, 15, 189-197. (Information about
this study was also obtained from DiMascio, Weissman et al., 1979;
Herceg-Baron et al., 1979; Prusoff et al., 1980; Rounsaville et al.,
1981; Weissman et al., 1981; Weissman, Prusoff, & DiMascio, 1979;
Weissman, Prusoff, DiMascio et al., 1979; Zuckerman et al., 1980.)
DiMascio, A., Weissman, M. M., Prusoff, B. A., Neu, C., Zwilling, M.,
& Klerman, G. L. (1979). Differential symptom reduction by drugs
and psychotherapy in acute depression. Archives of General Psychia-
try, 36, 1450-1456. (Information about this study was also obtained
from DiMascio, Klerman etal., 1979; Herceg-Baron etal., 1979; Prusoff et al., 1980; Rounsaville et al., 1981; Weissman et al., 1981;
Weissman, Prusoff, & DiMascio, 1979; Weissman, Prusoff, DiMascio
etal., 1979; Zuckerman etal., 1980.)
Herceg-Baron, R. L., Prusoff, B. A., Weissman, M. M., DiMascio, A.,
Neu, C., & Klerman, G. L. (1979). Pharmacotherapy and psychotherapy in acutely depressed patients: A study of attrition patterns in a
clinical trial. Comprehensive Psychiatry, 20, 315-325. (Information
about this study was also obtained from DiMascio, Klerman et al.,
1979; DiMascio, Weissman et al., 1979; Prusoff et al., 1980; Rounsaville etal., 1981;Weissmanetal., 1981; Weissman, Prusoff, & DiMascio, 1979; Weissman, Prusoff, DiMascio et al., 1979; Zuckerman et
al., 1980.)
Hersen, M., Bellack, A. S., Himmelhoch, J. M., & Thase, M. E. (1984).
Effects of social skill training, amitriptyline, and psychotherapy in
unipolar depressed women. Behavior Therapy, 15, 21-40. (Information about this study was also obtained from Bellack et al., 1981,
1983a, 1983b.)
Kovacs, M., Rush, A. J., Beck, A. T, & Hollon, S. D. (1981). Depressed
outpatients treated with cognitive therapy or pharmacotherapy: A
one-year follow-up. Archives of General Psychiatry, 38, 33-39. (Information about this study was also obtained from Rush et al., 1977,
1978,1982.)
McLean, P. D., & Hakstian, A. R. (1979). Clinical depression: Comparative efficacy of outpatient treatments. Journal of Consulting and
Clinical Psychology, 47, 818-836.
Murphy, G. E., Simons, A. D., Wetzel, R. D., & Lustman, P. J. (1984).
Cognitive therapy and pharmacotherapy: Singly and together in the
treatment of depression. Archives of General Psychiatry, 41, 33-41.
(Information about this study was also obtained from Simons, Garfield, & Murphy, 1984; Simons, Levine et al., 1984; Simons et al.,
1985,1986.)
Prusoff, B. A., Weissman, M. M., Klerman, G. L., & Rounsaville, B. J.
(1980). Research diagnostic criteria subtypes of depression: Their
role as predictors of differential response to psychotherapy and drug
treatment. A rchives of General Psychiatry, 37, 796-801. (Information
about this study was also obtained from DiMascio, Klerman, et al.,
1979; DiMascio, Weissman et al., 1979; Herceg-Baron et al., 1979;
Rounsaville et al., 1981; Weissman et al., 1981; Weissman, Prusoff,
& DiMascio, 1979; Weissman, Prusoff, DiMascio et al., 1979; Zuckerman etal., 1980.)
Roth, D., Bielski, R., Jones, M., Parker, W, & Osborn, G. (1982). A
comparison of self-control therapy and combined self-control therapy
and antidepressant medication in the treatment of depression. Behavior Therapy, 13, 133-144.
Rounsaville, B. J., Klerman, G. L., & Weissman, M. M. (1981). Do
psychotherapy and pharmacotherapy for depression conflict? Archives of General Psychiatry, 38,24-29. (Information about this study
was also obtained from DiMascio, Klerman et al., 1979; DiMascio,
Weissman etal., 1979; Herceg-Baron etal., 1979; Prusoff etal., 1980;
Weissman et al., 1981; Weissman, Prusoff, & DiMascio, 1979; Weissman, Prusoff, DiMascio et al., 1979; Zuckerman et al., 1980.)
Rush, A. J., Beck, A. T., Kovacs, M., & Hollon, S. (1977). Comparative
efficacy of cognitive therapy and pharmacotherapy in the treatment
of depressed outpatients. Cognitive Therapy and Research, 1, 17-37.
(Information about this study was also obtained from Kovacs et al.,
1981; Rush etal., 1978, 1982.)
Rush, A. J., Beck, A. T, Kovacs, M., Weissenburger, J., & Hollon, S. D.
(1982). Comparison of the effects of cognitive therapy and pharmaco-
48
L. ROBINSON, J. BERMAN, AND R. NEIMEYER
therapy on hopelessness and self-concept. American Journal of Psychiatry, 139, 862-866. (Information about this study was also obtained from Kovacs et al., 1981; Rush et al., 1977, 1978.)
Rush, A. J., Hollon, S. D., Beck, A. T., & Kovacs, M. (1978). Depression: Must pharmacotherapy fail for cognitive therapy to succeed?
Cognitive Therapy and Research, 2, 199-206. (Information about
this study was also obtained from Kovacs et al., 1981; Rush et al.,
1977,1982.)
Rush, A. J., & Watkins, J. T. (1981). Group versus individual cognitive
therapy: A pilot study. Cognitive Therapy and Research, 5, 95-103.
Schneider, L. S., Sloane, R. B., Staples, F. R., & Bender, M. (1986).
Pretreatment orthostatic hypotension as a predictor of response to
nortriptyline in geriatric depression. Journal of Clinical Psychopharmacology, 6, 172-176.
Simons, A. D., Garfield, S. L., & Murphy, G. E. (1984). The process
of change in cognitive therapy and pharmacotherapy for depression:
Changes in mood and cognition. Archives of General Psychiatry, 41,
45-51. (Information about this study was also obtained from Murphy
etal., 1984; Simons, Levineetal., 1984; Simons etal., 1985,1986.)
Simons, A. D., Levine, J. L., Lustman, P. J., & Murphy, G. E. (1984).
Patient attrition in a comparative outcome study of depression: A
follow-up report. Journal of Affective Disorders, 6, 163-173. (Information about this study was also obtained from Murphy et al., 1984;
Simons, Garfield, & Murphy, 1984; Simons etal., 1985,1986.)
Simons, A. D., Lustman, P. J., Wetzel, R. D., & Murphy, G. E. (1985).
Predicting response to cognitive therapy of depression: The role of
learned resourcefulness. Cognitive Therapy and Research, 9, 79-89.
(Information about this study was also obtained from Murphy et al.,
1984; Simons, Garfield, & Murphy, 1984; Simons, Levine et al.,
1984; Simons etal., 1986.)
Simons, A. D., Murphy, G. E., Levine, J. L., & Wetzel, R. D. (1986).
Cognitive therapy and pharmacotherapy for depression: Sustained
improvement over one year. Archives of General Psychiatry, 43, 4348. (Information about this study was also obtained from Murphy et
al., 1984; Simons, Garfield, & Murphy, 1984; Simons, Levine et al.,
1984; Simons etal., 1985.)
Udelman, D. L., & Udelman, H. D. (1985). A preliminary report on
anti-depressant therapy and its effects on hope and immunity. Social
Science and Medicine, 20, 1069-1072.
Weissman, M. M., Klerman, G. L., Prusoff, B. A., Sholomskas, D., &
Padian, N. (1981). Depressed outpatients: Results one year after
treatment with drugs and/or interpersonal psychotherapy. Archives of
General Psychiatry, 38,51-55. (Information about this study was also
obtained from DiMascio, Klerman et al., 1979; DiMascio, Weissman
et al., 1979; Herceg-Baron et al., 1979; Prusoff et al., 1980; Rounsaville et al., 1981; Weissman, Prusoff, & DiMascio, 1979; Weissman,
Prusoff, DiMascio etal., 1979; Zuckerman etal., 1980.)
Weissman, M. M., Prusoff, B. A., & DiMascio, A. (1979). Research
directions on comparisons of drugs and psychotherapy in depression.
Psychopharmacology Bulletin, 15, 19-21. (Information about this
study was also obtained from DiMascio, Klerman et al., 1979; DiMascio, Weissman et al., 1979; Herceg-Baron et al., 1979; Prusoff et
al., 1980; Rounsaville etal., 1981; Weissman etal., 1981; Weissman,
Prusoff, DiMascio etal., 1979; Zuckerman etal., 1980.)
Weissman, M. M., Prusoff, B. A., DiMascio, A., Neu, C, Goklaney, M.,
& Klerman, G. L. (1979). The efficacy of drugs and psychotherapy
in the treatment of acute depressive episodes. American Journal of
Psychiatry, 136, 555-558. (Information about this study was also obtained from DiMascio, Klerman et al., 1979; DiMascio, Weissman et
al., 1979; Herceg-Baron etal., 1979; Prusoff etal., 1980; Rounsaville
et al., 1981; Weissman et al., 1981; Weissman, Prusoff, & DiMascio,
1979; Zuckerman etal., 1980.)
Weissman, M. M., Prusoff, B. A., Kleber, H. D., Sholomskas, A. J., &
Rounsaville, B. J. (1985). Alprazolam (Xanax) in the treatment of
major depression. In G. D. Burrows, T. R. Norman, & L. Dennerstein (Eds.), Clinical and pharmacological studies in psychiatric
disorders (pp. 52-58). London: John Libbey.
Wilson, P. H. (1982). Combined pharmacological and behavioural
treatment of depression. Behaviour Research and Therapy, 20, 173184.
Zuckerman, D. M., Prusoff, B. A., Weissman, M. M., & Padian, N. S.
(1980). Personality as a predictor of psychotherapy and pharmacotherapy outcome for depressed outpatients. Journal of Consult ing and
Clinical Psychology, 48, 730-735. (Information about this study was
also obtained from DiMascio, Klerman et al., 1979; DiMascio,
Weissman etal., 1979; Herceg-Baron etal., 1979; Prusoff etal., 1980;
Rounsaville et al., 1981; Weissman et al., 1981; Weissman, Prusoff,
& DiMascio, 1979; Weissman, Prusoff, DiMascio et al., 1979.)
Appendix C
Sources for the Normative Studies Included in the Review
Baum, A., Gatchel, R. J., & Schaeffer, M. A. (1983). Emotional, behavioral, and physiological effects of chronic stress at Three Mile Island.
Journal of Consulting and Clinical Psychology, 51, 565-572.
Baumgart, E. P., & Oliver, J. M. (1981). Sex-ratio and gender differences
in depression in an unselected adult population. Journal of Clinical
Psychology, 37, 570-574.
Beckneld, D. F. (1985). Interpersonal competence among college men
hypothesized to be at risk for schizophrenia. Journal of Abnormal
Psychology, 94, 397-404.
Bond, C. R., & McMahon, R. J. (1984). Relationships between marital
distress and child behavior problems, maternal personal adjustment,
maternal personality, and maternal parenting behavior. Journal of Abnormal Psychology, 93, 348-351.
Bumberry, W, Oliver, J. M., & McClure, J. N. (1978). Validation of the
Beck Depression Inventory in a university population using psychiatric estimate as the criterion. Journal of Consulting and Clinical Psychology, 46, 150-155.
Clark, D. A. (1986). Cognitive-affective interaction: A test of the "specificity" and "generality" hypotheses. Cognitive Therapy and Research,
10, 607-623.
Cohen, L. H., McGowan, J., Fooskas, S., & Rose, S. (1984). Positive life
events and social support and the relationship between life stress and
psychological disorder. American Journal of Community Psychology,
12, 567-587.
Dobson, K. S., & Breiter, H. J. (1983). Cognitive assessment of depression: Reliability and validity of three measures. Journal of Abnormal
Psychology, 92, 107-109.
Dobson, K. S., & Shaw, B. F. (1986). Cognitive assessment with
major depressive disorders. Cognitive Therapy and Research, 10,
13-29.
Ellis, E. M., Atkeson, B. M., & Calhoun, K. S. (1981). An assessment
of long-term reaction to rape. Journal of Abnormal Psychology, 90,
263-266.
Funabiki, D., Bologna, N. C., Pepping, M., & FitzGerald, K. C. (1980).
PSYCHOTHERAPY FOR THE TREATMENT OF DEPRESSION
Revisiting sex differences in the expression of depression. Journal of
Abnormal Psychology, 89, 194-202.
Gallagher, D., Nies, G., & Thompson, L. W. (1982). Reliability of the
Beck Depression Inventory with older adults. Journal of Consulting
and Clinical Psychology, 50, 152-153.
Gotlib, I. H. (1984). Depression and general psychopathology in university students. Journal of Abnormal Psychology, 93, 19-30.
Hammen, C., & Mayol, A. (1982). Depression and cognitive characteristics of stressful life-event types. Journal of Abnormal Psychology,
91, 165-174.
Hammen, C. L., & Padesky, C. A. (1977). Sex differences in the expression of depressive responses on the Beck Depression Inventory. Journal of Abnormal Psychology, 86, 609-614.
Hewitt, P. I., & Dyck, D. G. (1986). Perfectionism, stress, and vulnerability to depression. Cognitive Therapy and Research, 10, 137-142.
Hollon, S. D., Kendall, P. C., & Lumry, A. (1986). Specificity of depressotypic cognitions in clinical depression. Journal of Abnormal Psychology, 95, 52-59.
Johnson, M. H., Magaro, P. A., & Stern, S. L. (1986). Use of the SADSC as a diagnostic and symptom severity measure. Journal of Consulting and Clinical Psychology, 54, 546-551.
Katzman, M. A., & Wolchik, S. A. (1984). Bulimia and binge eating in
college women: A comparison of personality and behavioral characteristics. Journal of Consulting and Clinical Psychology, 52,423-428.
Linden, W., Paulhus, D. L., & Dobson, K. S. (1986). Effects of response
styles on the report of psychological and somatic distress. Journal of
Consulting and Clinical Psychology, 54, 309-313.
Mogg, K., Mathews, A., & Weinman, J. (1987). Memory bias in clinical
anxiety. Journal of Abnormal Psychology, 96, 94-98.
Monroe, S. M., Imhoff, D. F., Wise, B. D., & Harris, J. E. (1983). Prediction of psychological symptoms under high-risk psychosocial circumstances: Life events, social support, and symptom specificity. Journal
of Abnormal Psychology, 92, 338-350.
Nezu, A. M., Nezu, C. M., & Nezu, V. A. (1986). Depression, general
distress, and causal attributions among university students. Journal
of Abnormal Psychology, 95, 184-186.
Nezu, A. M., & Ronan, G. F. (1985). Life stress, current problems,
problem solving, and depressive symptoms: An integrative model.
Journal of Consulting and Clinical Psychology, 53, 693-697.
Norton, G. R., Dorward, J., & Cox, B. J. (1986). Factors associated with
panic attacks in nonclinical subjects. Behavior Therapy, 17, 239-252.
Olinger, L. J., Kuiper, N. A., & Shaw, B. F. (1987). Dysfunctional attitudes and stressful life events: An interactive model of depression.
Cognitive Therapy and Research, 11, 25-40.
Oliver, J. M., & Simmons, M. E. (1985). Affective disorders and depres-
49
sion as measured by the diagnostic interview schedule and the Beck
Depression Inventory in an unselected adult population. Journal of
Clinical Psychology, 41,469-477.
Pagel, M. D., Becker, J., & Coppel, D. B. (1985). Loss of control, selfblame, and depression: An investigation of spouse caregivers of Alzheimer's disease patients. Journal of Abnormal Psychology, 94, 169182.
Pasahow, R. J. (1980). The relation between an attributional dimension
and learned helplessness. Journal of Abnormal Psychology, 89, 358367.
Peterson, C. (1979). Uncontrollability and self-blame in depression: Investigation of the paradox in a college population. Journal of Abnormal Psychology, 88, 620-624.
Peterson, C. (1983). Clouds and silver linings: Depressive symptoms and
causal attributions about ostensibly "good" and "bad" events. Cognitive Therapy and Research, 7, 575-578.
Propst, L. R. (1980). The comparative efficacy of religious and nonreligious imagery for the treatment of mild depression in religious individuals. Cognitive Therapy and Research, 4, 167-178.
Rizley, R. (1978). Depression and distortion in the attribution of causality. Journal of Abnormal Psychology, 87, 32-48.
Strauss, J., & Ryan, R. M. (1987). Autonomy disturbances in subtypes
of anorexia neivosa. Journal of Abnormal Psychology, 96, 254-258.
Tanaka-Matsumi, J., & Kameoka, V. A. (1986). Reliabilities and concurrent validities of popular self-report measures of depression, anxiety, and social desirability. Journal of Consulting and Clinical Psychology, 54, 328-333.
Williamson, D. A., Kelley, M. L., Davis, C. J., Ruggiero, L., & Blouin,
D. C. (1985). Psychopathology of eating disorders: A controlled comparison of bulimic, obese, and normal subjects. Journal of Consulting
and Clinical Psychology, 53, 161 -166.
Wise, E. H., & Barnes, D. R. (1986). The relationship among life events,
dysfunctional attitudes, and depression. Cognitive Therapy and Research, 10. 257-266.
Zautra, A. J., Guenther, R. T, & Chartier, G. M. (1985). Attributions
of real and hypothetical events: Their relation to self-esteem and depression. Journal of Abnormal Psychology, 94, 530-540.
Zimmerman, M. (1986). The stability of the Revised Beck Depression
Inventory in college students: Relationship with life events. Cognitive
Therapy and Research, 10, 37-43.
Received August 12,1988
Revision received June 27,1989
Accepted August 10,1989