Content uploaded by Nikolaos Kazantzis
Author content
All content in this area was uploaded by Nikolaos Kazantzis on Dec 08, 2017
Content may be subject to copyright.
Meta-Analysis of Homework Effects in Cognitive and
Behavioral Therapy: A Replication and Extension
Nikolaos Kazantzis, School of Psychological Science, La Trobe University, Australia
Craig Whittington, National Collaborating Centre for Mental Health and University College,
London
Frank Dattilio, Faculty of Psychiatry at Harvard Medical School and University of Pennsylvania
School of Medicine
Kazantzis, Deane, and Ronan (2000) estimated the effect
size (ES) for homework’s causal effects on outcome, but
did not (a) estimate ES for ‘‘control’’ therapy conditions,
(b) incorporate data from correlational studies, or (c) test
for outliers. The present analysis (46 studies,
N
= 1,072)
replicated and extended Kazantzis and colleagues’
review and obtained a pre–posttreatment ES of
d
= 0.63
for control conditions, and a larger
d
= 1.08 for therapy
conditions with homework. A pooled ES of
d
= 0.48
favoring homework was obtained when the analysis was
restricted to controlled studies contrasting the same
therapy. No evidence was found for outlier or publication
bias effects. Results supported the conclusion that
homework assignments enhance therapy outcomes.
Key words:
cognitive–behavioral therapy, homework
assignments, meta-analysis, outcome.
[Clin Psychol Sci
Prac 17: 144–156, 2010]
Beck’s cognitive therapy (Beck, Rush, Shaw, & Emery,
1979), frequently labeled as cognitive behavioral ther-
apy (CBT), requires patients to engage in homework
assignments as a vehicle for testing out ideas, putting
new skills into practice, and improving functioning in
the diversity of situations in which their problems
occur. The effective use of homework is integral to
these theorized mechanisms of therapeutic change (see
Kazantzis, Deane, Ronan, & L’Abate, 2005; Roth &
Pilling, 2007). Although between-session tasks are a
feature of many cognitive and behavioral therapies (cf.
Kazantzis, Reinecke, & Freeman, 2010), the definition
and empirical evaluation of homework has been more
extensively considered in CBT (Lambert, Harmon, &
Slade, 2007; Nelson, Castonguay, & Barwick, 2007).
Homework has been studied more than any other
therapeutic process in CBT (Persons, Davidson, &
Tompkins, 2000), and research has continued during
the past decade (e.g., Coon & Thompson, 2003;
Dunn, Morrison, & Bentall, 2002, 2006; Neimeyer,
Kazantzis, Kassler, Baker, & Fletcher, 2008; Tolin,
Frost, & Steketee, 2007; Westra, Dozois, & Marcus,
2007). Although studies have examined relevant
in-session processes, such as therapist behaviors in
planning or reviewing homework (Bryant, Simons, &
Thase, 1999; Detweiler-Bedell & Whisman, 2005;
Startup & Edmonds, 1994), research on homework
has mainly focused on its effects in relation to pre–
posttreatment symptom reduction.
Several reviews of the evidence for homework effects
in various CBTs have been published (e.g., Beutler
et al., 2004; Burns & Auerbach, 1992; Macaskill, 1996;
Openshaw, 1998; Shelton & Levy, 1981). However,
these reviews have taken the research studies as a
cohesive group, and not considered the different research
Address correspondence to Nikolaos Kazantzis, School of
Psychological Science, La Trobe University, Melbourne,
VIC, 3086, Australia. E-mail: n.kazantzis@latrobe.edu.au.
2010 American Psychological Association. Published by Wiley Periodicals, Inc., on behalf of the American Psychological Association.
All rights reserved. For permissions, please email: permissionsuk@wiley.com 144
questions under investigation. Correlational studies have
associated homework adherence with therapy outcome,
whereas experimental studies have manipulated home-
work using a ‘‘control’’ therapy condition without
homework versus a comparable therapy condition with
homework. For example, Coon and Thompson (2003)
describe a correlational analysis between homework
adherence and therapy outcome in CBT for depression
among older adults. Neimeyer and Feixas (1990) con-
ducted an experimental study of homework’s causal
effects in group CBT for depression (i.e., outpatients
with major depressive disorder were randomly assigned
to a control group without homework [n= 31] or an
experimental group receiving a standard thought record
assignment for homework [n= 32]). Clearly, these two
different methodologies have different aims, namely to
examine either (a) the correlational relationship between
homework adherence and symptom reduction or (b) the
causal effects of homework assignments on symptom
reduction (see Figure 1).
If a homework assignment is included in therapy,
then the extent to which it is effective depends on
whether the patient attempts the assignment (i.e., the
patient’s level of adherence). The different research
questions being examined in studies of homework
effects also require a different role for the assessment of
homework adherence. Figure 1 illustrates the different
role of adherence in studies of homework’s correla-
tional and causal effects. As shown, patient homework
adherence is a variable correlated with outcome in cor-
relational studies, whereas adherence is used as a check
on treatment integrity in studies of causal effects.
Therefore, there are several reasons to consider these
two streams of research examining relationships
between homework and outcome separately.
The research on correlational effects has consistently
shown positive adherence–outcome relationships with
effect size (ES) in the small to medium range (see review
in Lambert et al., 2007). However, the research examin-
ing causal effects has not always shown superiority of
therapies involving homework. Only one review by
Kazantzis et al. (2000), published in Clinical Psychology:
Science and Practice, has comprehensively reviewed and
quantitatively aggregated the data from studies of home-
work’s effects in therapy. Kazantzis and colleagues’
meta-analysis involved 27 studies examining homework
effects in CBT. The analysis demonstrated that there was
consistency in the ESs for correlational homework effects
(r= .22) and causal homework effects (r= .36). It was
later concluded, using Rosenthal and Rubin’s (1982)
binomial ES display to interpret the causal effects r= .36
value, that 68% of patients would improve in treatment
when receiving a therapy with homework, compared
with only 32% receiving a comparable therapy minus
homework (Kazantzis & Lampropoulos, 2002).
Although this meta-analysis was helpful and made a use-
ful contribution, questions have been raised regarding
the meta-analytic methodology (see Beutler et al., 2004;
Lambert et al., 2007). If the methods used in Kazantzis
and colleagues’ review were faulty, then there is a possi-
bility that the ES regarding homework’s causal effects
was artificially inflated.
One limitation was that Kazantzis et al. (2000) did
not estimate ES for control therapy conditions (i.e., no
homework) included in studies of homework’s causal
effects. As homework demands considerable time and
effort for both therapists and patients, it can be argued
that it is just as important to quantify the benefit that
patients receive from therapy that excludes homework.
It is plausible that therapy without homework can still
be effective. In fact, one study that did not find a
Figure 1. Two main research designs used to examine homework effects
in relation to treatment outcome.
HOMEWORK ASSIGNMENTS AND THERAPY OUTCOME •KAZANTZIS ET AL. 145
benefit for their experimental condition (i.e., therapy
with homework) concluded that their findings called
for a reconsideration of the inclusion of homework in
CBT (see Zettle & Hayes, 1987). Therefore, calculat-
ing the ES for control conditions would have strength-
ened the contribution of Kazantzis and colleagues’
review to this circumscribed area of research.
A second limitation was that Kazantzis et al. (2000)
conducted separate analyses for studies examining cor-
relational and causal homework effects, when some of
these data could have been combined. Studies of
homework’s correlational effects (Figure 1a) could have
been included along with data from the experimental
conditions from studies of homework’s causal effects
(Figure 1b). Such an analysis would have enabled a
comparison of control conditions in relation to all the
available pre–postsymptom change data for conditions
of therapy with homework, and may have yielded dif-
ferent findings.
A third limitation was that Kazantzis and colleagues’
(2000) analysis attempted to examine moderators of the
ES by dividing the sample of studies into subgroups
based on patient population (i.e., anxiety, depression,
inpatient, outpatient), homework type (exposure, relax-
ation, social skills, etc.), source of homework adher-
ence assessment (patient, therapist, objective measure),
and timing of homework adherence assessment (regular
intervals, posttreatment). Although this analysis was
interesting in theory, it was problematic in practice
because it assumed consistency in other aspects of the
research methodology, such as the use of therapy man-
uals, therapist training, monitoring of therapist adher-
ence to protocols, evaluation of therapist competence,
and so forth. As Lambert et al. (2007) outlined, many
of the studies had poorly specified control therapy con-
ditions and were actually different therapeutic interven-
tions to those used in experimental conditions. A more
methodologically rigorous and conservative approach
to that undertaken by Kazantzis et al. would have been
to estimate ES studies of homework’s causal effects in
studies where the type of therapy was held constant
between control and experimental conditions.
Finally, Beutler et al. (2004) took a selection of the
studies examined by Kazantzis et al. (2000) in the
causal effects analysis and obtained a smaller ES. Thus,
Beutler and colleagues’ review raised the possibility that
the ES estimates reported in Kazantzis et al. may have
been skewed by outliers. Tests of outliers were not
conducted in Kazantzis and colleagues’ report.
The aim of the present study was to conduct a
replication of Kazantzis and colleagues’ (2000) meta-
analytic review of homework’s causal effects on therapy
outcome, and extend the analysis in several important
ways. Specifically, the present study was designed to:
(a) use all available data for the evaluation of pre–post-
treatment change within control conditions (i.e., ther-
apy without homework) and within therapy conditions
involving homework; (b) compute a between-group
ES for controlled studies examining homework’s causal
effects, with a separate analysis for studies holding
constant the type of therapy between control and
experimental conditions; and finally, (c) test for outliers
in studies of homework’s causal effects. It was hypothe-
sized that conditions of therapy with homework would
demonstrate larger ESs than conditions of therapy
without homework.
1
METHOD
Sample
Studies examining homework in therapy were identi-
fied using the same search strategy outlined in
Kazantzis and colleagues’ (2000) meta-analysis. Specifi-
cally, studies were identified through (a) computer
search of the PsycINFO database during the period of
January 1980 through December 2007 using the key
terms behavioral practice, extratherapy, extratreatment, home
practice, homework, and self-help assignments; (b) a manual
search of the reference sections of previous reviews;
and (c) a manual search of the reference sections of
studies examining the relationship between homework
(or homework adherence) and therapy outcome. (Con-
sistent with Kazantzis and colleagues’ strategy, the pres-
ent review focused on studies published after 1980
because the vast majority of previously published
treatment outcome studies did not assess homework
adherence; see Shelton & Levy, 1981.) As a result of
the two searches described, a total of 953 studies were
initially selected and considered for inclusion.
Studies were included in the meta-analysis if they
met Kazantzis and colleagues’ (2000) criteria: (a) were
published in English; (b) examined the effects of
homework assignments on therapy outcome assessed at
CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE •V17 N2, JUNE 2010 146
pre- and posttreatment (i.e., using experimental or
quasi-experimental methodology); and (c) included
sufficient statistical data to calculate ES (excluding non-
parametric and Tobit analyses). A large number of the
953 studies were excluded as they mentioned that
homework was part of the treatment protocol, but they
did not examine homework effects in relation to
therapy outcome. In addition, many of the abstracts
represented nonempirical articles that were inappropri-
ate for inclusion (e.g., case studies, theoretical articles).
Of the studies identified in the search, 46 studies
met inclusion criteria (27 studies were included in
Kazantzis and colleagues’ meta-analysis).
Calculation of Effect Sizes
In the original meta-analysis (Kazantzis et al., 2000),
the coefficient rwas used as the ES index. Since then,
Cohen’s d(Cohen, 1988), also known as the standard-
ized mean difference (the difference between two
means divided by their pooled standard deviation), has
become a more widely used and understood index.
Therefore, the present review utilized d, computed
from means and standard deviations, or where no other
data were reported, from sample size and p-values, or
from appropriate within-or between-groups t-or
F-tests. When results were reported only as significant,
p= .05 was assumed, and when only as nonsignificant,
p= .50 was assumed. Because sample sizes were rela-
tively small, we adjusted dusing Hedges’s correction
(Hedges’s g; Hedges & Olkin, 1985). Mean ES and
associated standard error were used where a study con-
tained multiple ESs (Hunter & Schmidt, 1990).
Data Synthesis and Analytic Strategy
Comprehensive Meta-Analysis software, Version
2.2.040 (Borenstein, Hedges, Higgins, & Rothstein,
2005) was used to conduct the meta-analyses and test for
publication bias. Positive ds indicated that the outcome
favored therapy with homework relative to therapy
without homework, or that there was more improve-
ment from pre- to posttreatment. To examine statistical
heterogeneity in the meta-analyses, we used both the
chi-squared statistic (Q) and I
2
(Higgins & Thompson,
2002), as well as a visual inspection of the forest plots. I
2
was used because it described the proportion of total var-
iation in the study of ES that is due to heterogeneity as
opposed to sampling error (Higgins & Thompson,
2002), and was not affected by low statistical power.
(Low statistical power can cause problems interpreting
other measures of heterogeneity, including the chi-
squared test.) To help judge the extent of heterogeneity,
the following guideline was used: 25% = low,
50% = moderate, and 75% = high heterogeneity (Hig-
gins, Thompson, Deeks, & Altman, 2003). A fixed
effects meta-analysis model was employed to calculate
the pooled ES where heterogeneity was low; otherwise
the impact of studies with extreme results (outliers) by
using a sensitivity analysis was used to exclude these
studies. In the case of important unexplained heteroge-
neity (i.e., I
2
> 50%), a random effects model was used
to estimate the pooled ES (DerSimonian & Laird, 1986).
Outliers were determined by calculating the sample-
adjusted meta-analytic deviancy (SAMD) statistic, with a
scree plot used to establish a cutoff SAMD value (Huff-
cut & Arthur, 1995).
RESULTS
Characteristics of Studies
Forty-six studies were included in the meta-analysis,
comprising 14 evaluations of pre–posttreatment changes
in control therapy conditions (Figure 2) and 32 evalua-
tions of pre–posttreatment changes in therapy conditions
with homework (Figure 3). Twelve studies (14 evalua-
tions) were specifically designed to examine the causal
effects of homework assignments in therapy (Figure 4), a
subset (n= 9) that held constant the type of therapy
between control (i.e., no homework) and experimental
conditions (i.e., therapy with homework; see Figure 5).
The remainder (n= 5) did not hold constant the type of
therapy between control and experimental conditions
(e.g., relaxation versus CBT). Although the aforemen-
tioned searches were not deliberately restricted to studies
of CBT, all studies examined the role of homework
within the stated treatment approaches.
Major characteristics of the studies included in the
pre–posttreatment analysis are presented in Figures 2
and 3. Overall, the number of participants in these
studies ranged from 5 to 94, with a mean of 23 partici-
pants. Six studies examined homework effects in the
therapy for depression, 26 studies examined homework
in the therapy for anxiety, 1 study examined home-
work in the therapy for either depression or anxiety,
HOMEWORK ASSIGNMENTS AND THERAPY OUTCOME •KAZANTZIS ET AL. 147
Note. Hedges’s g = ES estimate d with Hedges’s correction. A = Anxiety sample. CBT = Cognitive behavior therapy. CT = Cognitive therapy. CI = Confidence interval. CM = Covert modeling.
D = Depression sample. MC = Minimal contact. Mixed = Distancing and/or cognitive restructuring. O = Other outpatient sample. OR = Overt rehearsal. PHZ = Phenelzine. PLB = Placebo.
PMR = Pro
g
ressive muscle relaxation. Po
p
. = Patient
p
o
p
ulation. PS = Problem solvin
g
. Relax. = Relaxation. SP = Su
pp
ortive
p
s
y
chothera
py
. TB = Thermal biofeedback. WL = Wait list.
Figure 2. Meta-analysis on pre–posttreatment effects for ‘‘control’’ therapy conditions without homework.
N
ote. Hedges’s g = ES estimate d with Hedges’s correction. A = Anxiety sample. AMT = Anxiety management training. BT Steps = Computer-guided self-help.
CBT = Cognitive behavior therapy. CBT (G) = CBT for GAD. CBT (P) = CBT for panic disorder. CBT (S) = CBT for social phobia. CT = Cognitive therapy.
CI = Confidence interval. CM = Covert modeling. CM + OR= Covert modeling plus overt rehearsal. D = Depression sample. Exposure = exposure for phobia.
Mixed = Distancing and/or cognitive restructuring. O = Other outpatient sample. PHZ = Phenelzine. PLB = Placebo. PMR = Progressive muscle relaxation.
Pop. = Patient population. PP = Programmed practice. Relax. = Relaxation. Self-exp. = Self-exposure for phobia. Self-relax. = Self-relaxation for phobia.
TB = Thermal biofeedback.
Figure 3. Meta-analysis on pre–posttreatment effects for therapy conditions with homework.
CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE •V17 N2, JUNE 2010 148
and the remaining 13 studies examined homework in
the therapy for a range of other patient problems.
Major characteristics of controlled studies of home-
work’s causal effects are presented in Figure 4. Overall,
the number of participants in these studies ranged from
12 to 64, with a mean of 34 participants. All studies
defined and measured outcome in therapy as symptom-
atic improvement. Two of these studies examined
homework for patients with depression, six for patients
with anxiety-related disorders, and the remaining
six studies examined homework for other patient
problems.
Pre–Posttreatment Effects of Control Therapy Conditions
There were 14 studies (N= 234) that evaluated
therapy conditions without homework in the analysis
(Figure 2). As there was evidence of moderate hetero-
geneity across these studies (I
2
= 64%, Q= 35.82,
p< .001), the random effects model was utilized in this
analysis. The pooled ES was d= 0.63, 95% confidence
interval (CI) = 0.39–0.87, p< .0001.
Pre–Posttreatment Effects of Therapy Conditions With Homework
There were 32 studies (N= 838) that evaluated
therapy conditions with homework in the analysis
N
ote. Hedges’s g = ES estimate d with Hedges’s correction. A = Anxiety sample. AMT = Anxiety management training. CBT = Cognitive behavior therapy.
CT = Cognitive therapy. CI = Confidence interval. CM = Covert modeling. CM + = Covert modeling plus overt rehearsal. D = Depression sample. MC = Minimal contact.
Mixed = Distancing and/or cognitive restructuring. O = Other outpatient sample. PHZ = Phenelzine. PLB = Placebo. PMR = Progressive muscle relaxation.
Po
p
. = Patient
p
o
p
ulation. PP = Pro
g
rammed
p
ractice. PS = Problem solvin
g
. Relax. = Relaxation. SP = Su
pp
ortive
p
s
y
chothera
py
. TB = Thermal biofeedback. WL = Wait list.
Figure 4. Meta-analysis of controlled studies directly contrasting therapy with and without homework.
N
ote. Hedges’s g = ES estimate d with Hedges’s correction. A = Anxiety sample. CBT = Cognitive behavior therapy. CT = Cognitive therapy.
CI = Confidence interval. CM = Covert modeling. D = Depression sample. Mixed = Distancing and/or cognitive restructuring. O = Other outpatient sample.
PHZ = Phenelzine. PLB = Placebo. PMR = Pro
g
ressive muscle relaxation. Pop. = Patient population. Relax. = Relaxation. TB = Thermal biofeedback.
Figure 5. Meta-analysis of studies that hold constant the type of therapy in ‘‘control’’ and experimental conditions.
HOMEWORK ASSIGNMENTS AND THERAPY OUTCOME •KAZANTZIS ET AL. 149
(Figure 3). There was again evidence that heterogene-
ity across studies was high (I
2
= 79%, Q= 147.64,
p< .0001), so the random effects model was utilized.
The mean ES was d= 1.08, 95% CI = 0.89–1.27,
p< .0001.
As noted in the introduction, the present study
intentionally combined data from correlational and
experimental research for the evaluation of pre–post-
treatment change for therapy conditions with home-
work. However, in order to ensure that estimation of
change in therapy conditions with homework was not
due to differences in study design, we excluded 18
studies that did not have a control group and recom-
puted the meta-analysis with the remaining 14 studies
(N= 243). As there was evidence of moderate hetero-
geneity across these studies (I
2
= 53%, Q= 27.83,
p< .01), the random effects model was again utilized.
The mean ES was very similar to the analysis involving
the entire group of studies, d= 1.04, 95% CI = 0.81–
1.28, p< .0001, thereby justifying the grouping of 32
studies in the evaluation of pre–posttreatment effects.
Direct Comparisons of Therapy Conditions With and Without
Homework
There were 14 evaluations (N= 477) that had directly
compared therapy conditions with and without home-
work (Figure 4). For example, Hoelscher, Lichstein, and
Rosenthal (1986) contrasted two groups receiving the
same therapy, and added contingency contracting as
homework in one group only. Similarly, Neimeyer and
Feixas (1990) contrasted two groups of cognitive ther-
apy, and added a thought record for homework in one
group only. There was no evidence of important heter-
ogeneity across these studies (I
2
= 37%, Q= 20.71,
p= .08), so the fixed effects model was utilized. The
mean ES was d= 0.53, 95% CI = 0.35–0.72,
p< .0001). As some studies did not hold constant the
type of therapy in control and experimental conditions,
we reanalyzed the data as a function of research design.
Studies That Held Constant Type of Therapy Between
Control and Experimental Conditions. There were nine
evaluations (N= 299) that held constant the type of
therapy used for the comparison of control and experi-
mental conditions. There was no evidence of important
heterogeneity across these studies (I
2
= 38%,
Q= 12.95, p= .11), so the fixed effects model was
utilized. The mean ES was d= 0.48, 95% CI = 0.25–
0.71, p< .0001. These results are displayed in the for-
est plot presented in Figure 5.
Studies That Did Not Hold Constant Type of Therapy
Between Control and Experimental Conditions. There were
five studies (N= 178) that had utilized different ther-
apy conditions when contrasting control and experi-
mental therapy conditions. There was no evidence of
heterogeneity across these studies (I
2
= 44%, Q= 7.19,
p= .13), so the fixed effects model was utilized. The
mean ES was d= 0.63, 95% CI = 0.33–0.93,
p< .0001.
Analysis of Outliers
There was heterogeneity in the ES estimates for the
studies in the meta-analysis. However, outliers were
only present in one of the analyses, and sensitivity tests
indicated that these did not influence the overall mean
ES. That is, three studies were identified as outliers in
the group of 32 studies evaluating pre–posttreatment
effects for therapy with homework (i.e., Al-Kubaisy
et al., 1992; Bryant et al., 1999; Woods, Chambless, &
Steketee, 2002). When these three studies were
excluded from the analysis, there was still evidence of
moderate heterogeneity across the remaining studies
(I
2
= 63%, Q= 74.99, p< .0001), and the random
effects model was utilized. The mean ES was d= 1.00,
95% CI = 0.84–1.16, p< .0001, and did not differ
substantially from the mean ES when the outliers
were included (i.e., d= 1.08, 95% CI = 0.89–1.26,
p< .0001).
Evaluation of Hypothesized Findings
The results supported the hypothesis that therapy con-
ditions with homework would demonstrate larger ESs
than therapy conditions without homework. The find-
ings that contribute to this support were (a) the smaller
ES obtained for pre–posttreatment effects of control
conditions of therapy without homework, in compari-
son with pre–posttreatment effects of therapy with
homework (d= 0.63 and d= 1.08, respectively); and
(b) the positive ES obtained in studies that held con-
stant the type of therapy for comparisons of control
and experimental conditions (d= 0.48).
CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE •V17 N2, JUNE 2010 150
Analysis for Publication Bias (Fail-Safe
N
)
An estimated ES has the potential to overestimate the
true effect under investigation simply because the esti-
mate is derived from published research. A key concern
of publication bias is that some nonsignificant studies
are missing from analysis and that these studies, if
included, would nullify the observed ES and conclu-
sions from the analysis. The present meta-analysis of
pre–posttreatment effects of therapies with homework
incorporated data from 32 studies, which yielded a z-
value of 22.35, corresponding two-tailed p< .0001,
and a fail-safe Nof 4,129. This means that 4,129
‘‘null’’ studies would need to be identified and
included in order for the combined two-tailed p> .05.
The meta-analysis of control therapy conditions incor-
porated data from 14 studies, which yielded a z-value
of 8.39, corresponding two-tailed p< .0001, and a
fail-safe Nof 243. The meta-analysis of control versus
experimental therapy conditions incorporated data from
13 studies, which yielded a z-value of 5.15, corre-
sponding two-tailed p< .0001, and a fail-safe Nof 77.
Thus, it is very unlikely that the findings reported in
the present review were due to publication bias.
DISCUSSION
Three questions were addressed in synthesizing the
results of the empirical studies over the past 27 years:
Do therapy conditions with and without homework
produce different pre–posttreatment effects if all the
available data are analyzed together? What is the magni-
tude of effect for the direct comparison of therapy with
and without homework in a more comprehensive body
of research? What is the influence of holding constant
the type of therapy in trials directly contrasting control
(i.e., no homework) with experimental conditions (i.e.,
therapy with homework). We answered these questions
in primary, secondary, and sensitivity analyses testing for
possible outliers.
The meta-analysis produced a pre–posttreatment
finding of d= 0.63 for control conditions and larger
d= 1.08 for therapy conditions with homework (a sta-
tistically significant difference; z-value = 2.87, one-
tailed p= .002). The ES for therapy conditions with
homework was somewhat larger than the d= 0.77 (con-
verted from r= .36) reported in the previous meta-anal-
ysis of the same literature (i.e., Kazantzis et al., 2000).
The present analysis also found that studies directly
contrasting therapy with and without homework pro-
duced a statistically significant mean ES of d= 0.48,
which represented a small-to-medium effect in favor of
therapy with homework. (A larger mean ES estimate
was obtained when the type of therapy was not held
constant between control and experimental conditions,
d= 0.64.) Using Rosenthal and Rubin’s (1982) bino-
mial ES display, the value of d= 0.48 could be taken
to suggest that 62% of patients would improve when
receiving therapy with homework, compared with only
38% receiving therapy without homework. It is notable
that the conclusion is the same whether homework
effects are estimated by directly contrasting therapies
with and without homework, or by calculating ES
from pre–postcomparisons. There is a clear benefit for
homework assignments over and above the effects of
an already effective therapy. Therefore, the present
findings provided good evidence to support the
hypothesis that homework makes clinically meaningful
contributions to CBT outcome. These data can also be
taken to support the central role accorded to home-
work assignments in Beckian cognitive therapy (see
Beck et al., 1979; Kazantzis, Deane et al., 2005).
There was significant heterogeneity across the studies
included in the pre–posttreatment homework effects
analysis. Some of this variability could be accounted for
by variations in samples and other features of design
methodology. For instance, a frequent problem of this
research literature is the lack of design sensitivity to
detect homework effects. A previous power analysis by
Kazantzis (2000) demonstrated that studies attempting
to detect homework’s causal effects have, on average,
only had a 58% chance of detecting large effects (as
defined by Cohen’s [1988] conventions). Similarly,
studies had a low chance of detecting medium and small
ESs (32% and 9%, respectively). However, there are
several important conceptual issues in this literature that
may account for this variability and warrant attention.
Conceptual Issues in Research on Homework Effects
In conclusion to Kazantzis and colleagues’ (2000)
meta-analysis, the following was noted:
We hope that this meta-analytic review has changed
the focus from the general question of whether
HOMEWORK ASSIGNMENTS AND THERAPY OUTCOME •KAZANTZIS ET AL. 151
homework compliance is related to therapy out-
come, to more specific questions such as which
types of homework assignments facilitate improve-
ment in therapy for which patient problems, and
which therapist behaviors can enhance the effects of
homework assignments in therapy. (p. 198)
A decade later, the present review demonstrates that
research efforts have not followed this recommenda-
tion. We argue that there are conceptual issues inher-
ent in the studies reviewed in the present analysis that
now demand the change in research direction sug-
gested by Kazantzis et al. (2000).
The first conceptual issue concerns the meaning of
therapy conditions that exclude homework assign-
ments. As homework has always been integral to the
facilitation of mechanisms of change in many behav-
ioral and cognitive therapies (e.g., Beck et al., 1979),
the utility of extracting such a core process for empiri-
cal investigation is questionable. For example, the study
by Neimeyer and Feixas (1990) used an experimental
design to contrast two conditions of group cognitive
therapy (Beck et al., 1979) that differed in the assign-
ment of thought records for homework. That is, the
control condition involved no thought records for
homework. Given the integral role of homework to
cognitive therapy, it is debatable whether the therapy
condition used as the control can be considered adher-
ent to the tenets of the system of psychotherapy
devised by Aaron T. Beck.
We encourage researchers to begin exploring the
specific processes involved in integrating homework
assignments into session, such as linking homework
to case conceptualization (Kuyken, Padesky, & Dud-
ley, 2009), and through the development of a sound
collaborative empiricism (Kazantzis, MacEwan, &
Dattilio, 2005). Such investigations could be initially
conducted in analog studies involving a ‘‘standard’’
homework assignment offered to two therapy groups
differing only in therapist behaviors in integrating
homework. These studies would lay the foundations
for prospective treatment outcome evaluations where
the integration of homework into sessions was specif-
ically examined.
A second conceptual issue in this research literature
lies in the assessment of homework adherence. As a
patient may complete a thought record on a daily basis
over the course of a week and not receive as much
benefit as another patient who completes a thought
record once, we argue that ‘‘simple’’ adherence, or
quantity of homework completion, is not the most
theoretically meaningful construct to be assessed.
Behavior and cognitive theories suggest that a range of
antecedents, consequences, and beliefs are important to
understanding what leads a person to engage in a
health behavior (Cameron & Leventhal, 2003; Kanfer
& Goldstein, 1991; Kazantzis & L’Abate, 2007). In
addition, a range of factors are important in under-
standing different levels of generalization and mainte-
nance of adaptive behaviors (Kazantzis & Daniel, 2008;
Thoresen & Mahoney, 1974). These theoretically
meaningful determinants of homework adherence have
generally not been assessed in the research to date
(Kazantzis, Deane, & Ronan, 2004; Lambert et al.,
2007; Primakoff, Epstein, & Covi, 1986).
A comparatively small group of studies have exam-
ined expectancies (Westra et al., 2007) and willingness
to engage (Burns & Nolen-Hoeksema, 1991; Neimeyer
et al., 2008), despite the research supporting various
cognitive determinants of patients’ decision to engage
in health behaviors (Berg, Raminani, Greer, Harwood,
& Safren, 2008; Dunn et al., 2006; Russell & Kazant-
zis, 2008). However, efforts to study the role of home-
work in therapy are becoming more sophisticated, with
observations of nonlinear relationships between adher-
ence and symptom change (Gaynor, Lawrence, &
Nelson-Gray, 2006; Westra et al., 2007), and efforts to
produce indexes of session-by-session homework
adherence (Yovel & Safren, 2007). Lambert et al.
(2007) reviewed these issues and encouraged research-
ers to adopt more theoretically meaningful measures
(e.g., Kazantzis et al., 2004).
A final conceptual issue concerns the distinction
between homework adherence and the generalization of
adaptive coping skills. In practice, a patient’s completion
of homework assignments signals the beginning of gen-
eralization, given that a skill has been transferred from
the session to the patient’s everyday life. As the patient
engages in independent practice of the task in different
contexts with or without in-session discussion, there is
evidence of further generalization of the skill. Ulti-
mately, a goal is to enable a patient to develop complex
CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE •V17 N2, JUNE 2010 152
reasoning about the adaptive skill (i.e., higher-level gen-
eralization). Taking behavioral activation as the example,
it is important to monitor a patient’s completion of plea-
sure and mastery activities, movement to continued
practice of activities, through an awareness of how to
regulate and maintain his or her mood state through
activity scheduling. In fact, there are data to support the
assertion that there has been conceptual confusion and
inadequate distinction between the assessment of home-
work adherence and more complex generalization.
Kornblith, Rehm, O’Hara, and Lamparski (1983)
assigned 49 depressed patients to four therapy conditions
and showed no advantage for those conditions involving
homework assignments. However, Kornblith et al.
found that patients in the control conditions decided to
practice therapeutic activities of their own accord, and
suggested that this patient-initiated between-session
activity explained the null results obtained in the study.
Patients are routinely expected to continue the practice
of skills learned early in CBT (e.g., behavioral activation
through activity scheduling) alongside the learning of
new skills (e.g., cognitive restructuring through thought
records and behavioral experiments; see Dobson &
Dobson, 2009). We suggest that future research would
benefit from delineating effects that are due to home-
work adherence from the effects that are due to more
complex generalization of adaptive skills. Such research
would require a clear definition of what constitutes ‘‘a
homework assignment’’ and what constitutes ‘‘general-
ization’’ in CBT.
LIMITATION
A limitation of this study was the relatively small num-
ber of studies involved in some analyses. For this rea-
son, it is premature to generalize the results of this
meta-analysis more broadly. These studies suggest spe-
cific effects for homework assignments, but given the
absence of important methodological details of these
studies, such as a clear treatment manual and assessment
of therapist adherence and competence, it was not pos-
sible to evaluate the quality of the studies as a modera-
tor of aggregated effects. Such close inspection of study
methodology has proven useful in recent meta-analyses
(cf. O
¨st, 2008) and may serve to raise awareness of the
importance of specifying these details in published
reports.
CONCLUSION
The results of this meta-analysis allow us to conclude
that, when all of the data are analyzed together, study
conditions that include homework as a main component
produce superior treatment effects to those that do not.
These results support the central thesis that therapy with
homework is better than therapy without homework, in
the context of the CBTs. Unfortunately, less can be said
about the effect of the quality and nature of homework
tasks that were assigned in these studies. Future research
is needed, particularly on the process of integrating
homework assignments into sessions, effects of the qual-
ity of homework assignments, skill acquisition, general-
ization, and the theoretically meaningful determinants of
homework adherence. Such research would contribute
to a better understanding of how homework produces its
effects, and would lead directly to empirically based
guidance for the integration of homework assignments
in CBT.
NOTE
1. No tests of moderators were considered appropriate in the
present meta-analysis because of the small number of studies
that had carried out a direct comparison of the same therapy
in attempting to demonstrate homework effects. The
decision was also based on the critique of Kazantzis and
colleagues’ (2000) meta-analysis by Lambert et al. (2007).
ACKNOWLEDGMENTS
The authors thank Dr. Keith S. Dobson, University of
Calgary, for comments on a previous version of this article.
The senior author also thanks Aaron T. Beck, Judith S. Beck,
David A. Clark, Frank P. Deane, Keith S. Dobson, Louis G.
Castonguay, Ian M. Evans, Lydia Fehm, Arthur Freeman,
Luciano L’Abate, Michael J. Lambert, Georgios Lampropoulos,
Paul L. Merrick, Rona Moss-Morris, Robert A. Neimeyer,
Frederick L. Newman, Nancy A. Pachana, Kevin R. Ronan,
Gregoris Simos, Mehmet Sungur, and the students in the
Cognitive Behavior Therapy Homework Project’s Research
Laboratory for their collaboration, inspiration, and helpful
discussions on the role of homework in psychotherapy.
REFERENCES
*Al-Kubaisy, T., Marks, I. M., Logsdail, S., Marks, M. P.,
Lovell, K., Sungur, M., et al. (1992). Role of exposure
homework in phobia reduction: A controlled study.
Behavior Therapy,23, 599–621.
HOMEWORK ASSIGNMENTS AND THERAPY OUTCOME •KAZANTZIS ET AL. 153
*Barlow, D. H., O’Brien, G. T., & Last, C. G. (1984). Couples
treatment of agoraphobia. Behavior Therapy,15, 41–58.
Beck, A. T., Rush, J. A., Shaw, B. F., & Emery, G. (1979).
Cognitive therapy of depression. New York: Guilford.
Berg, C., Raminani, S., Greer, J., Harwood, M., & Safren,
S. (2008). Patients’ perspectives on cognitive-behavioural
therapy for adherence and depression in HIV. Psychother-
apy Research,18, 271–280.
Beutler, L. E., Malik, M., Alimohamed, S., Harwood, T. M.,
Talebi, H., Noble, S., et al. (2004). Therapist effects. In
M. J. Lambert (Ed.), Bergin and Garfield’s handbook of psy-
chotherapy and behavior change (5th ed., pp. 227–306). New
York: Wiley.
*Blanchard, E. B., Nicholson, N. L., Radnitz, C. L., Steffek,
B. D., Appelbaum, K. A., & Dentinger, M. P. (1991a).
The role of home practice in thermal biofeedback. Journal
of Consulting and Clinical Psychology,59, 507–512.
*Blanchard, E. B., Nicholson, N. L., Taylor, A. E., Steffek,
B. D., Radnitz, C. L., & Appelbaum, K. A. (1991b). The
role of regular home practice in the relaxation treatment
of tension headache. Journal of Consulting and Clinical Psy-
chology,59, 467–470.
Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H.
(2005). Comprehensive meta-analysis (Version 2) [Com-
puter software]. Englewood, NJ: Biostat.
*Bryant, M. J., Simons, A. D., & Thase, M. E. (1999).
Therapist skill and patient variables in homework compli-
ance: Controlling an uncontrolled variable in cognitive
therapy outcome research. Cognitive Therapy and Research,
23, 381–399.
Burns, D. D., & Auerbach, A. H. (1992). Does homework
compliance enhance recovery from depression? Psychiatric
Annals,22, 464–469.
Burns, D. D., & Nolen-Hoeksema, S. (1991). Coping styles,
homework compliance, and the effectiveness of cognitive-
behavioural therapy. Journal of Consulting and Clinical
Psychology,59, 305–311.
Cameron, L. D., & Leventhal, H. (Eds.). (2003). The self-
regulation of health and illness behaviour. London: Routledge.
*Carroll, K. M., Nich, C., & Ball, S. A. (2005). Practice
makes progress? Homework assignments and outcome in
treatment of cocaine dependence. Journal of Consulting and
Clinical Psychology,73, 749–755.
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.
*Coon, D. W., & Thompson, L. W. (2003). The relationship
between homework compliance and treatment outcomes
among older adult outpatients with mild-to-moderate
depression. American Journal of Geriatric Psychiatry,11, 53–61.
DerSimonian, R., & Laird, N. (1986). Meta-analysis in clini-
cal trials. Controlled Clinical Trials,7, 177–188.
Detweiler-Bedell, J. B., & Whisman, M. A. (2005). A lesson
in assigning homework: Therapist, client, and task charac-
teristics in cognitive therapy for depression. Professional
Psychology: Research and Practice,36, 219–223.
Dobson, D., & Dobson, K. S. (2009). Evidence-based practice of
cognitive-behavioral therapy. New York: Guilford.
Dunn, H., Morrison, A. P., & Bentall, R. P. (2002). Patients’
experiences of homework tasks in cognitive behavioral
therapy for psychosis: A qualitative analysis. Clinical Psy-
chology & Psychotherapy,9, 361–369.
Dunn, H., Morrison, A. P., & Bentall, R. P. (2006). The
relationship between patient suitability, therapeutic alli-
ance, homework compliance and outcome in cognitive
therapy for psychosis. Clinical Psychology & Psychotherapy,
13, 145–152.
Gaynor, S. T., Lawrence, P. S., & Nelson-Gray, R. O.
(2006). Measuring homework compliance in cognitive-
behavioral therapy for adolescent depression: Review, pre-
liminary findings, and implications for theory and practice.
Behavior Modification,30, 647–672.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-
analysis. Orlando, FL: Academic.
Higgins, J. P. T., & Thompson, S. G. (2002). Quantifying
heterogeneity in a meta-analysis. Statistics in Medicine,21,
1539–1558.
Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman,
D. G. (2003). Measuring inconsistency in meta-analyses.
British Medical Journal,327, 557–560.
*Hoelscher, T. J., Lichstein, K. L., & Rosenthal, T. L.
(1986). Home relaxation practice in hypertension treat-
ment: Objective assessment and compliance induction.
Journal of Consulting and Clinical Psychology,54, 217–221.
Huffcut, A. I., & Arthur, W. (1995). Development of a new
outlier statistic for meta- analytic data. Journal of Applied
Psychology,80, 327–334.
Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analy-
sis: Correcting error and bias in research findings. Newbury
Park, CA: Sage.
*Jannoun, L., Munby, M., Catalan, J., & Gelder, M. (1980).
A home-based treatment program for agoraphobia: Repli-
cation and controlled evaluation. Behavior Therapy,11,
294–305.
*Jannoun, L., Oppenheimer, C., & Gelder, M. (1982). A
self-help treatment program for anxiety state patients.
Behavior Therapy,13, 103–111.
Kanfer, F. H., & Goldstein, A. P. (1991). Helping people
change: A textbook of methods. New York: Pergamon.
CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE •V17 N2, JUNE 2010 154
Kazantzis, N. (2000). Power to detect homework effects in
psychotherapy outcome research. Journal of Consulting and
Clinical Psychology,68, 166–170.
Kazantzis, N., & Daniel, J. (2008). Homework assignments.
In G. Simos (Ed.), Cognitive behaviour therapy: A guide for
the practicing clinician (2nd ed., pp. 165–186). New York:
Routledge.
Kazantzis, N., Deane, F. P., & Ronan, K. R. (2000). Home-
work assignments in cognitive and behavioral therapy: A
meta-analysis. Clinical Psychology: Science and Practice,7,
189–202.
Kazantzis, N., Deane, F. P., & Ronan, K. R. (2004). Assess-
ing compliance with homework assignments: Review and
recommendations for clinical practice. Journal of Clinical
Psychology,60, 627–641.
Kazantzis, N., Deane, F. P., Ronan, K. R., & L’Abate, L.
(Eds.). (2005). Using homework assignments in cognitive behav-
ioral therapy. New York: Routledge.
Kazantzis, N., & L’Abate, L. (2007). Introduction and his-
torical overview. In N. Kazantzis & L. L’Abate (Eds.),
Handbook of homework assignments in psychotherapy:
Research, practice, and prevention (pp. 1–15). New York:
Springer.
Kazantzis, N., & Lampropoulos, G. L. (2002b). Reflecting on
homework in psychotherapy: What can we conclude from
research and experience? Journal of Clinical Psychology,58,
577–585.
Kazantzis, N., MacEwan, J., & Dattilio, F. M. (2005). A
guiding model for practice. In N. Kazantzis, F. P. Deane,
K. R. Ronan, & L. L’Abate (Eds.), Using homework assign-
ments in cognitive behavior therapy (pp. 359–407). New
York: Routledge.
Kazantzis, N., Reinecke, M. A., & Freeman, A. (Eds.).
(2010). Cognitive and behavioral theories in clinical practice.
New York: Guilford.
*Kazdin, A. E., & Mascitelli, S. (1982). Covert and overt
rehearsal and homework practice in developing assertive-
ness. Journal of Consulting and Clinical Psychology,50, 250–
258.
*Kenwright, M., Marks, I., Graham, C., Frances, A., &
Mataix-Cols, D. (2005). Brief scheduled phone support
from a clinician to enhance computer-aided self-help for
obsessive compulsive disorder: Randomized controlled
trial. Journal of Clinical Psychology,61, 1499–1508.
Kornblith, S. J., Rehm, L. P., O’Hara, M. W., & Lamparski,
D. M. (1983). The contribution of self-reinforcement
training and behavioral assignments to the efficacy of self-
control therapy for depression. Cognitive Therapy and
Research,7, 499–528.
Kuyken, W., Padesky, C. A., & Dudley, R. (2009). Collabo-
rative case conceptualization. New York: Guilford.
Lambert, M. J., Harmon, S. C., & Slade, K. (2007). Direc-
tions for research on homework. In N. Kazantzis &
L. L’Abate (Eds.), Handbook of homework assignments in psy-
chotherapy: Research, practice, and prevention (pp. 407–423).
New York: Springer.
*Lampropoulos, G., & Rector, N. A. (2004). Homework
compliance in cognitive behavior therapy for anxiety disorders:
Predictors of outcome. Poster presented at the annual Harvey
Stancer Research Day, University of Toronto, Ontario,
Canada.
Macaskill, N. D. (1996). Improving clinical outcomes in
REBT ⁄CBT: The therapeutic uses of tape-recording.
Journal of Rational-Emotive and Cognitive-Behavior Therapy,
14, 199–207.
*Neimeyer, R. A., & Feixas, G. (1990). The role of homework
and skill acquisition in the outcome of group cognitive
therapy for depression. Behavior Therapy,21, 281–292.
Neimeyer, R. A., Kazantzis, N., Kassler, D. M., Baker,
K. D., & Fletcher, R. (2008). Group cognitive behavior
therapy for depression: Outcomes predicted by willingness
to engage in homework, compliance with homework, and
cognitive restructuring skill acquisition. Cognitive Behavior
Therapy,37, 199–215.
Nelson, D. L., Castonguay, L. G., & Barwick, F. (2007).
Directions for the integration of homework in psycho-
therapy practice: Conceptual and clinical consensus toward
the benefits and implementation of between sessions
activities. In N. Kazantzis & L. L’Abate (Eds.), Handbook
of homework assignments in psychotherapy: Research, practice,
and prevention (pp. 425–444). New York: Springer.
Openshaw, D. K. (1998). Increasing homework compliance:
The SEA method. Journal of Family Psychotherapy,9, 21–29.
O
¨st, L. G. (2008). Efficacy of the third wave of behavioral
therapies: A systematic review and meta-analysis. Behaviour
Research and Therapy,46, 296–321.
Persons, J. B., Davidson, J., & Tompkins, M. A. (2000).
Essential components of cognitive-behavior therapy for depression.
Washington, DC: American Psychological Association.
Primakoff, L., Epstein, N., & Covi, L. (1986). Homework
compliance: An uncontrolled variable in cognitive therapy
outcome research. Behavior Therapy,17, 433–446.
*Rees, C. S., McEvoy, P., & Nathan, P. (2005). Relationship
between homework completion and outcome in cognitive
behavior therapy. Cognitive Behaviour Therapy,34, 242–247.
Rosenthal, R., & Rubin, D. B. (1982). A simple, general
purpose display of magnitude of experimental effect.
Journal of Educational Psychology,74, 166–169.
HOMEWORK ASSIGNMENTS AND THERAPY OUTCOME •KAZANTZIS ET AL. 155
Roth, A. D., & Pilling, S. (2007). Competencies required to
deliver effective cognitive and behaviour therapy for people with
depression and with anxiety disorders. London: HMSO,
Department of Health.
Russell, J., & Kazantzis, N. (2008). Medication beliefs and
adherence to antidepressants in primary care. New Zealand
Medical Journal,121, 14–20.
*Schmidt, N. B., & Woolaway-Bickel, K. (2000). The effect
of treatment compliance on outcome in cognitive-
behavioral therapy for panic disorder: Quantity versus
quality. Journal of Consulting and Clinical Psychology,68,
13–18.
Shelton, J. L., & Levy, R. L. (1981). Behavioral assignments
and treatment compliance: A handbook of clinical strategies.
Champaign, IL: Research Press.
*Solyom, C., Solyom, L., LaPierre, Y., Pecknold, J., & Mor-
ton, L. (1981). Phenelzine and exposure in the treatment
of phobias. Biological Psychiatry,16, 239–247.
*Stanley, M. A., Beck, J. G., & Glassco, J. D. (1996). Treat-
ment of generalized anxiety in older adults: A preliminary
comparison of cognitive-behavioral and supportive
approaches. Behavior Therapy,27, 565–581.
*Stanley, M. A., Beck, J. G., Novy, D. M., Avrill, P. M., &
Swann, A. C. (2003). Cognitive behavioral treatment of
late-life generalized anxiety disorder. Journal of Consulting
and Clinical Psychology,71, 309–319.
Startup, M., & Edmonds, J. (1994). Compliance with home-
work assignments in cognitive-behavioral psychotherapy
for depression: Relation to outcome and methods of
enhancement. Cognitive Therapy and Research,18, 567–579.
Thoresen, C. E., & Mahoney, M. J. (1974). Behavioral
self-control. New York: Holt, Rinehart & Winston.
*Tolin, D. F., Frost, R. O., & Steketee, G. (2007). An open
trial of cognitive-behavioral therapy for compulsive hoard-
ing. Behaviour Research and Therapy,45, 1461–1470.
*Westra, H. A., & Dozois, D. J. A. (2006). Preparing clients
for cognitive behavioral therapy: A randomized pilot study
of motivational interviewing for anxiety. Cognitive Therapy
and Research,30, 481–498.
*Westra, H. A., Dozois, D. J. A., & Marcus, M. (2007).
Expectancy, homework compliance, and initial change in
cognitive behavioral therapy for anxiety. Journal of Consult-
ing and Clinical Psychology,75, 363–373.
*Wetherall, J. L., Hopko, D. R., Diefenbach, G. J., Averill,
P. M., Beck, J. G., Craske, M. G., et al. (2005). Cognitive
behavioral therapy for late-life generalized anxiety disorder:
Who gets better? Behavior Therapy,36, 147–156.
*Woods, C. M., Chambless, D. L., & Steketee, G. (2002).
Homework compliance and behavior therapy outcome for
panic with agoraphobia and obsessive compulsive disorder.
Cognitive Behaviour Therapy,31, 88–95.
*Woody, S. R., & Adessky, R. S. (2002). Therapeutic
alliance, group cohesion, and homework compliance
during cognitive-behavioral group treatment of social
phobia. Behavior Therapy,33, 5–27.
*Yovel, I., & Safren, S. A. (2007). Measuring homework
utility in psychotherapy: Cognitive behavioral therapy
for adult attention-deficit hyperactivity disorder as an
example. Cognitive Therapy and Research,31, 385–399.
*Zettle, R. D., & Hayes, S. C. (1987). Component and
process analysis of cognitive therapy. Psychological Reports,
61, 939–953.
*References marked with an asterisk were included in
the meta-analysis.
Received March 16, 2009; revised August 18, 2009; accepted
October 8, 2009.
CLINICAL PSYCHOLOGY: SCIENCE AND PRACTICE •V17 N2, JUNE 2010 156
A preview of this full-text is provided by American Psychological Association.
Content available from Clinical Psychology Science and Practice
This content is subject to copyright. Terms and conditions apply.