The efficacy of problem solving therapy in reducing mental and
physical health problems: A meta-analysis
John M. Malouff⁎, Einar B. Thorsteinsson, Nicola S. Schutte
School of Psychology, University of New England, Armidale, NSW, 2351, Australia
Received 13 July 2005; accepted 17 July 2005
This paper describes a meta-analysis of 31 studies that examined the efficacy of problem solving therapy (PST). The meta-
analysis, encompassing 2895 participants, showed that PST is significantly more effective than no treatment (d=1.37), treatment as
usual (d=0.54), and attention placebo (d=0.54), but not significantly more effective than other bona fide treatments offered as part
of a study (d=0.22). Significant moderators included whether the PST included problem-orientation training, whether homework
was assigned, and whether a developer of PST helped conduct the study.
© 2006 Elsevier Ltd. All rights reserved.
Keywords: Problem solving; Therapy; Treatment; Efficacy; Meta-analysis
Problem solving therapy (PST) developed out of a trend toward providing psychotherapy by teaching clients
psychosocial skills (D'Zurilla & Goldfried, 1971). PST involves teaching a client how to use a step-by-step process to
solve life problems. The usual process taught can be broken into two major parts: (a) applying a problem-solving
orientation to life and (b) using rational problem-solving skills. Applying a problem-solving orientation usually
involves appraising problems as challenges, thinking that the problems can be solved, and realizing that effective
problem solving tends to require time and systematic effort (Nezu, 2004). Rational problem-solving skills include:
(1) attempting to identify a problem when it occurs, (2) defining a problem, (3) attempting to understand the
problem, (4) setting goals related to the problem, (5) generating alternative solutions, (6) evaluating and choosing
the best alternatives, (7) implementing the chosen alternatives, and (8) evaluating the efficacy of the effort at
problem solving (D'Zurilla & Nezu, 1999). If the efforts to solve the problem fail, one may return to any step and
try again. PST typically involves oral and written presentation of the steps by the therapist, along with guided
practice, both in session and as home assignments, in solving real problems. Developers of PST recommend that
clients receive 8–16 sessions of 1.5 to 2 hr each (D'Zurilla & Nezu, 1999).
Over the past few decades dozens of articles have been published reporting evaluations of the efficacy of PSTwith a
wide variety of problems, such as deliberate self-harm, depression, and obesity. D'Zurilla and Nezu (1999) summarized
Clinical Psychology Review 27 (2007) 46–57
⁎Corresponding author. Tel.: +61 2 6773 3776.
E-mail address: firstname.lastname@example.org (J.M. Malouff).
0272-7358/$ - see front matter © 2006 Elsevier Ltd. All rights reserved.
analysis that included four studies of the efficacy of PST for preventing repetition of deliberate self-harm found only a
non-significant trend for the effects of PST (Hawton et al., 1998). A meta-analysis by Townsend et al. (2001)
evaluated efficacy of PST for hopelessness, depression, and problem resolution in deliberate self-harm clients. This
analysis included the same four studies as in the prior meta-analysis plus another two studies. The comparison treatment
and problem resolution than the comparison treatment.The two meta-analyses produced interesting results but withsome
important limitations: the analyses (a) reviewed only treatment for individuals who had engaged in intentional self-harm,
(b) included only a small number of studies; and (c) showed different results with different types of outcome variables.
The mixed findings of prior meta-analyses suggest that there may be moderators of the efficacy of PST.For instance,
in a very large and important meta-analysis, Wampold et al. (1997) found that various types of mental health treatment
had sizeable but equivalent effect sizes, so one might wonder whether the effect of PSTis moderated by the comparison
group, with PST more effective than no treatment but no more effective than another bona fide treatment.
A study by Nezu and Perri (1989) showed that, as a group, depressed individuals who received the full PST, including
problem-orientation training, experienced significantly greater decreases in depression than depressed individuals who
received PST without problem-orientation training. Nezu (2004) commented later that the lack of problem orientation
training as part of PSTwould therefore be worth examining as a moderator of effect size in a meta-analysis.
Burke, Arkowitz, and Menchola (2003) found that treatment studies of motivational interviewing had a greater effect
size if they were done by a developer of the treatment method. Hence, one might wonder whether efficacy studies of PST
would show larger effect sizes if done by a developer of PST. One might also think that more hours of PSTwould lead to
might also affect treatment outcomes. A meta-analysis of home assignments in psychotherapy in general concluded that
giving home assignments led to greater treatment effects (Kazantzis, Deane, & Ronan, 2000). Although no reason exists
for PST to have greater effects with one gender than another or with young people versus adults, such a difference is
possible and this information is common in research reports, so examination of these two variables seems warranted.
Another variable of similar appeal and availability of information is whether the research participants were individuals
seeking psychotherapy or recruited into psychotherapy for the study. The more important group for generalization would
seemtobethose whoare seekinghelpontheir own;veryfew peoplewilleverberecruitedintopsychotherapyaspartofa
resistant to substantial long-term improvement from any type of psychological treatment (Norris et al., 2004). It is also
outcomes with either format (McRoberts, Burlingame, and Hoag, 1998).
It is possible that subjective outcomes (e.g., self-report of depression) differ in effect size from more objective
outcomes (e.g., interviewer ratings), so evaluating type of outcome as a moderator, such as Leichsenring and Leibing
(2003) did in a meta-analysis, makes sense. Further, the effects of PST may fade out over time, as appears to happen
with some psychological treatments of some disorders (Spiegler & Guevremont, 2003), although psychotherapy
studies usually find no overall association between effect size and length of followup (Lambert & Bergin, 1994). Still,
checking for an association between length of followup and treatment effect size seems wise.
1. Purpose of the present meta-analysis
The present paper reports the first meta-analysis of the efficacy of PSTacross all types of mental or physical health
problems. The meta-analysis had as its objectives to evaluate (1) whether PST is more effective in helping individuals
overcome health problems than no treatment, treatment as usual, attention–placebo treatment, and another bona fide
treatment presented specifically as part of a study; (2) whether PST that includes problem-orientation training is more
effective than PSTwithout this element; (3) whether more hours of PSTare associated with greater effects; (4) whether
including home assignments in PST is associated with greater effects; (5) whether studies that involve one of the
developers of PSTshow greater effects; (6) whether individual and group PST differ in efficacy; (7) whether effect size
varies with gender; (8) whether effect size differs for children and adults; (9) whether the effect size varies with whether
the participants are identified prior to the study as having a clinical problem or are recruited from the public; (10)
47J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
Summary statistics for analysis (Nd=39)
AnalysisN Problem treatedDepression
Format Hours of
Alexopoulos et al.
Arean et al. (1993)
⁎Arean et al. (1993)
Azrin et al. (2001)
25Depression and executive
Dysthymia in adults
Dysthymia in adults
Yes NoYesNo Individual na NoAttention/placebo0.963 0.130 1.790
Waiting list or no Tx
BFTx family behav
Barrett et al. (2001)
⁎Barrett et al. (2001)
Biggam and Power
Black and Scherba
Carey et al. (1990)
BFTx Milan approach
Tx as usual
14Obesity NoNo YesNo Group18 YesBFTx behav 1.6330.4202.840
22Mental illness and
na No YesYesGroup6 NoTx as usual0.485
Catalan et al. (1991)
Chaney et al. (1978)
Coche and Flick
Coche et al. (1984)
Dowrick et al. (2000)
⁎Dowrick et al.
Epstein et al. (2000)
Lerner and Clum
Malouff et al. (1988)
⁎Malouff et al.
McLeavey et al.
et al. (2000)
Tx as usual
Tx as usual
Tx as usual
BFTx cog-behav education
Waiting list or no Tx
BFTx family behav
BFTx rational emotive
Waiting list or no Tx
226 Unipolar depression YesNo YesNo Individual4 NoBFTx fluvoxamine0.151
J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
et al. (1995)
et al. (1995)
et al. (1997)
Nezu and Perri
Nezu et al. (2003)
Perri et al. (2001)
⁎Perri et al. (2001)
Pfiffner et al. (1990)
56 Unipolar depression YesNo Yes YesIndividual4 NoBFTx amitriptyline 0.234
55Unipolar depression Yes NoYes Yes Individual4 No Attention/placebo0.8640.310 1.420
58Emotional disorders naNo YesYesIndividual3 NoTx as usual
Waiting list or no Tx
Distressed cancer patients
Waiting list or no Tx
Waiting list or no Tx
BFTx behav parent
Tx as usual
Sahler et al. (2002)92 Mother of child
na NoYes NoIndividual8 Yes0.400
Salkovski et al.
Schwartz et al. (1998)
20na No YesYesIndividual5 YesTx as usual1.3950.4002.390
341Relative of breast
naNo YesNo Individual2 YesAttention/placebo0.031
Spaccarelli et al.
et al. (1992)
van den Hout et al.
Williams et al. (2000)
⁎Williams et al.
NoGroup6NoBFTx behave 0.473
38 Child behaviorNoNo Yesb
NoGroup6 NoWaiting list or no Tx 1.050 0.3701.730
76Low back pain NoNoYes YesGroup15YesAttention/placebo0.6310.170 1.090
Dysthymia in elders
Dysthymia in elders
Note. PST Developer = problem solving therapy (PST) developer one of the researchers; Clinical problem = whether the participants had been identified before the study as having a clinical problem;
na = information not available; BFTx = bona fide treatment; Tx = treatment.
⁎Denotes a second analysis in a single study, this one comparing problem solving with a waiting list or attention–placebo. In order to eliminate any sample bias caused by using more than one analysis
(effect size) per sample per study (Lipsey & Wilson, 2001) these second analyses were excluded from any meta-analysis that did not include comparison type as a variable.
#Using problem solving therapy by GPs versus medication alone.
aIncluded also teens.
bParents to help child behavior.
J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
whether the effect size varies with the type of mental or physical health problem; (11) whether PST has greater effects
on self-report outcome measures than on objective outcome measures; and (12) whether the effects of PST diminish
with increased follow-up time.
2.1. Literature search
The PsycINFO and PubMed databases were searched using the term “problem solving.” Relevant articles from this
search were then used to obtain additional articles. We also evaluated every treatment study cited in D'Zurilla and Nezu
(1999) and two meta-analyses of PST for self-harm treatment (Hawton et al., 1998; Townsend et al., 2001).
2.2. Characteristics of included and excluded studies
To be included, a study had to (a) evaluate PST as a treatment for a mental or physical health problem, (b) use a
comparison condition, (c) include random assignment to condition; (d) either describe the problem solving steps that the
participantsweretrainedinorstate thatthePSTfollowedD'Zurilla andGoldfried(1971)orD'ZurillaandNezu(1982)or
some closely related guidelines; and (e) report statistics essential to meta-analyses, such as the number of participants and
study, we attempted to obtain the information from the authors, but in all these cases we were unsuccessful. The meta-
analysis included studies that compared PST in addition to another treatment versus just the other treatment.
Eight studies that were considered closely for inclusion in the meta-analysis were excluded because they did not
clearly use PST but rather described vaguely a treatment focused on problem solving (e.g., Gibbons, Butler, Urwin, &
Gibbons, 1978). Hence, their results may tell us nothing about problem solving therapy. Seven studies (e.g., Kazdin,
means-end thinking, consequential thinking, and takingthe perspectiveof others as recommended by Spivack, Platt and
Shure (1976) in their description of interpersonal cognitive PST. Nine studies were excluded because they included in
PST distinct, major therapy methods such as assertion training, relaxation training, communication training, systematic
reinforcement for improved behavior, and extensive information booklets dealing with a specific problem of interest,
these studies because it is not possible to determine what contribution, if any, problem solving therapy made to the
(2004), who completed a meta-analysis of a specific therapy and excluded studies that included the specific therapy as
partofa “combinationtreatment” (p.8).One studywasexcluded becauseithad nooutcome measure involvinga mental
or physical health problem (Intagliata, 1978). Three studies were excluded because they lacked adequate data for the
meta-analysis (DeVellis, Blalock, Hahn, DeVellis, & Hochbaum, 1988; Graves, Meyers, & Clark, 1988; Hussian &
2.3. Coding studies
The studies were coded for the following variables: (1) type of comparison group: no treatment, treatment as usual,
attention–placebotreatment,oranother bonafidetreatmentpresentedspecificallyaspartofastudy;(2) whetherthe study
reportindicatedthatthe PSTusedincludedproblem-orientationtraining; (3) hoursofPST; (4)whether homeassignments
were mentioned as part of the PST; (5) whether one of the developers of PST participated in the study; (6) whether
the PST was presented in individual and group format; (7) gender of the participants: male, female, or mixed; (8)
whether the participants were children or adults; (9) whether the participants were identified prior to the studyashaving
a clinical problem or were recruited from the public; (10) the type of mental or physical health problem; (11) whether the
outcome variables were self-report, objective, or a combination of self-report and objective; (12) length of the follow up;
(13) the number of participants in each condition who completed the study; and (14) the key data for outcome measures.
Treatmentas usual characterizes the treatment individuals ina healthcaretreatment program would have received ifthere
was no study. Attention/placebo characterizes any study with a drug placebo or a psychological intervention intended to
serve as an attention/placebo.
50J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
Some studies used two different groups of problem solving therapy, e.g., one administered by physicians and one
administered by nurses. In such cases we randomly chose one of the two PST groups when the study did not report data
for the two problem solving groups combined. In many cases a study included more than one comparison group for
instance a behavioral treatment group and a waiting list control group. We completed separate analyses of comparisons
of problem solving therapy with (a) a bona fide treatment offered specifically for the study, (b) treatment as usual,
(c) attention/placebo treatment, and (d) no treatment or waiting list control. When a study provided outcome results for
different periods after completion of treatment we used the longest followup period.
We included as dependent variables only measures of some mental or physical health condition such as depression or
Problem-Solving Inventory (Heppner & Peterson, 1982), because these are not measures of mental or physical health. In
this regard, we followed the model of Burke et al. (2003), who completed a meta-analysis of a therapy method called
motivational interviewing, and included only measures of “the main behavioral and health outcomes” (p. 845).
When two or more outcome variables were mathematically related (e.g., body mass index and weight), we chose the
variable that seemed to best represent the desired outcome (e.g., body mass index).
statisticssuchasF,orpercentages ofparticipantswho moved intothe normalrange (seee.g.,Hedges, 1981,1982; Lipsey
& Wilson, 2001; Wolf, 1986); (b) using w, inverse variance weighting (Lipsey & Wilson, 2001), to compute descriptive
and inferential statistics; (c) calculating anaverage effectsizefor studies withmultipleoutcome measures; (d) usingthe Q
statistic (Lipsey & Wilson, 2001) to perform homogeneity analyses; and (e) examining effect sizes for univariate outliers
using as a criterion z=3.67, p=.001 two-tail, and Normal Q–Q plots and Detrended Normal Q–Q plots, following the
recommendations of Tabachnick and Fidell (2001). No effect sizes were identified as potential outliers.
Thirty-one studies, producing 39 effect-size analyses, met all the criteria for the meta-analysis. Table 1 describes the
studies, which encompassed a total of 2895 participants.
Meta-analysis summary statistics employing a mixed effects model (method of moments random effects) analysis
d (CI−95%, CI+95%) SEzpHomogeneity analysisFail-safe Na
One analysis per studyb
0.54 (0.31, 0.77)
0.56 (0.36, 0.76)
Note. A significant Q value indicates that homogeneity should be rejected (i.e., effect sizes are heterogeneous).
aReports the number of studies with d=0.00 needed to reduce the mean d to the d criterion value (±0.10).
bExcludes starred studies in Table 1.
Moderator analysis for four treatment comparison conditions, mixed effects model (method of moments random effects) analysis (Nd=39)
Source Waiting list or no treatmentTreatment as usual Attention-placeboOther experimental treatment
Comparison treatment Qbetween(3)=16.54, pb.001
51J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
Overall, PST had a significant effect size, using the best comparison group in each study, with another bona fide
treatment the top choice, treatment as usual and attention–placebo treatment next, and then no-treatment (see Table 2).
The homogeneity analysis in Table 2 indicates that the meta-analysis effect sizes tended to be heterogeneous,
suggesting that random effects models should be employed for analyses (Lipsey & Wilson, 2001). Random effects
models produce larger confidence intervals than fixed effects models, leading to more conservative conclusions about
significant differences. We therefore used random effects models for all meta-analyses.
With regard to type of comparison group, Table 3 shows that PSTwas significantly more effective than being on a
waiting list, treatment as usual, and an attention/placebo comparison group. However, PSTwas not significantly more
effective than a bona fide comparison treatment, although there was a trend in that direction. No one bona fide therapy
type was included in enough studies to allow comparison with a single specific type of therapy.
In order to search for moderators of effect size, we examined a number of coded variables to determine whether they
were associated with effect size. Two variables, gender of participants and whether the participants were adults or not,
had insufficient variability to support a meaningful analysis. Two variables involved continuous data, and their results
were as follows. The effect size for number of hours of PSTwas r(25)=.23, p=.12 (one-tailed; a non-significant trend);
the effect size for months of followup was r(29)=−.09, p=.32 (one-tailed). See Table 4 for the analyses of
dichotomous variables and Table 5 for the analysis of type of assessment format, a tripartite variable.
Three variables were significantly associated with effect size: inclusion in PST of training in problem orientation,
the statement in the report that PST included home assignments, and the participation in the study of one of the
developers of PST (in all cases Arthur Nezu). There was no significant difference in effect size for the other variables
examined: (a) individual versus group/family PST, (b) whether the participants had been identified before the study as
having a clinical problem or were recruited from the public, (c) whether the problem involved some specific type of
Moderator analysis, mixed effects model (method of moments random effects) analysis
Qbetween df pd (CI−95%, Cl+95%), SEQwithin dfpd (CI−95%, CI+95%), SE Qwithin dfp
Not depression related
.942 0.50 (0.14, 0.87), 0.186
No PST developer as one of the researchers
b.001 0.34 (0.15, 0.53), 0.098
.332 0.63 (0.34, 0.92), 0.149
.846 0.51 (0.18, 0.84), 0.167
.004 0.30 (0.01, 0.59), 0.146
No orientation training
b.001 0.19 (−0.06, 0.45), 0.131 21.26
.340 0.49 (0.16, 0.81), 0.164 27.14
PST developer as one of the researchers
26 .224 2.03 (1.45, 2.61), 0.294 16.78
18 .002 0.39 (0.01, 0.77), 0.194 13.74
13 .093 0.56 (0.22, 0.89), 0.171 34.62
18 .866 1.02 (0.62, 1.42), 0.204 34.00
16 .169 1.00 (0.69, 1.31), 0.159 31.82
Depression related 0.011 9.0389 .001
Clinical participants 0.94139.68 11.248
Treatment format 0.041 20.11 15 .003
Homework 8.291 11.6211 b.001
Training in problem orientation 15.261 13 .003
aExcludes starred studies in Table 1.
Moderator analysis for assessment type, mixed effects model (method of moments random effects) analysis (Nd=31)
Comparison treatment Qbetween(2)=2.85, p=.241
52J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
disorder (only studies of depression-spectrum disorders were numerous enough to compare with other disorders),
or (d) assessment format (i.e., self-report, objective or mixed).
In order to assess the relative importance of the three significant moderators, whether home assignments given,
whether the PST included problem orientation training, and whether a PST developer was one of the researchers, we
completed a multiple regression of effects sizes, calculated as suggested by Lipsey and Wilson (2001, pp. 138–142),
with the three significant moderators entered. Table 6 shows that only whether a PST developer was one of the
researchers made a significant independent contribution to effect size, with the other two variables close to
significance. The three moderators together accounted for 48% of the variance in outcome.
To determine whether the three studies that lacked some essential statistical data needed for meta-analysis showed
the same pattern as the studies that entered the meta-analysis, we looked closely at the results of each. Together the
results of the studies (DeVellis et al., 1988; Graves et al., 1988; Hussian & Lawrence, 1981) showed no clear pattern.
The studies of DeVellis et al. and Hussian and Lawrence showed no significant effect for PST, while the study by
Graves et al. showed a significant effect.
The meta-analysis showed that across all the studies PST had a significant effect size. PST was significantly more
effective than no treatment, attention/placebo treatment, and treatment as usual, but not significantly more effective
than other bona fide treatments with which it has been compared. The results, across 31 studies (39 analyses) and 2895
participants, provide strong evidence that PST tends to be effective in treating mental or physical health problems.
Whether PSTis more effective than other bona fide treatments is not perfectly clear because there was a non-significant
trend in favor of PST. The present meta-analysis examined only 16 studies that used for comparison a bona fide
treatment presented as part of a treatment study. It could be that with more studies, the trend in favor of PST would
become a statistically significant difference.
Finding equivalency of bona fide treatments in psychological treatment outcome research, as in this meta-analysis,
is common enough to have its own fanciful name, the Dodo Bird effect (Wampold et al., 1997). Equivalency may be
common because different psychological treatments tend to have similar elements such as a healer, a ritual, and social
support (Frank, 1973) and tend to lead to increased self-efficacy (Bandura, 1997). The finding of equivalent efficacy for
PSTand other psychological interventions suggests that PST may serve as a useful comparison treatment in evaluating
whether new types of treatment are more effective than an existing treatment.
The meta-analysis also found three other factors that were associated with effect size: participation of one of the PST
developers in the study, a clear statement in the study report that homework was assigned, and training in problem
orientation. Studies conducted by a PST developer had significantly larger effect sizes than studies conducted by non-
developers. This finding is similar to meta-analytic findings regarding a type of therapy called motivational
interviewing (Burke et al., 2003). The present results could be interpreted in various ways such as that the treatment
was done better or the therapists or clients had greater hope for the treatment. The main implication of the PST
developer finding is that other therapists are unlikely to obtain such large effects with PST as those of one of the
developers of PST. One might wonder whether a researcher's being a strong believer in a specific treatment might also
tend to be associated with greater effect sizes, but that could not be tested within the meta-analysis, as the current
tradition is not for researchers to describe their personal beliefs about the various treatments they test against each other.
PST studies that included training in problem orientation had larger effect sizes than studies that did not. This
finding is consistent with the finding of Nezu and Perri (1989) using an experimental research method that depression
improved more when clients received problem orientation training in addition to training in the rational steps of
Multiple regression analysis, random intercept, fixed slopes model (method of moments random effects) analysis (Nd=31)
SourceB SE Bβzp
PST developer as one of the researchers
Training in problem orientation
Note. This model had a d=0.50, R2=.48, Q(3)=38.30, pb.001; residual Q(27)=41.24, p=.039.
53J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
problem solving. Although the moderator findings of the present meta-analysis do not show causation, they raise the
possibility that the value of problem orientation training as part of PSTextends to treatment of problems in general, as
suggested by Nezu (2004). However, the multiple regression analysis showed that this moderator added no significant
explained variance to that of assignment of homework and participation of a PST developer.
The finding that reporting the giving of home assignments was associated with larger effect sizes is consistent with
findings in studies of psychotherapy in general (Kazantzis et al., 2000). The main implication of this moderator finding
is that PST may be more effective if home assignments are given, as suggested by the developers of PST (D'Zurilla &
Nezu, 1999). It is possible though that some or all of the studies that did not mention giving home assignments actually
gave them. The meaning of the difference in effect sizes would then be murky. Also, the moderator findings do not
show causation; some other factors may have led to both home assignments and greater effect sizes. Finally, the home-
assignments variable did not add a significant amount of explained variance to that of the PST-developer variable and
assignment of homework.
Several other variables were found not to be significantly related to effect size: the number of hours of PST
(although there was a trend in favor of a higher number of hours being associated with greater effect size); individual
versus group PST; whether the participants had been identified before the study as having a clinical problem; whether
the problem involved a depression-spectrum disorder; whether the study used self-report, objective or both self-report
and objective outcome measures; and the length of the followup.
The finding that number of hours of therapy was not significantly associated with effect size is similar to findings
regarding other types of psychotherapy, which also have found no significant relationship between number of hours of
treatment and outcome (Bennett & Gibbons, 2000; Dobson, 1989; Feske & Chambless, 1995; Koss & Shiang, 1994).
The finding that group PST was not significantly less effective than individual PST is consistent with most research
findings on individual versus group psychotherapy (McRoberts, Burlingame, & Hoag, 1998). The finding is important
because group therapy can be much more cost effective. The finding that whether the participants were previously
identified as having a clinical problem or were recruited for the study was unassociated with effectsize suggests that the
source of participants does not affect efficacy of PST. The finding that depression-spectrum problems and other
problems had similar effect sizes shows that as of yet there is no evidence that PSTis more effective with some specific
class of problems than with others. Lack of a significant association between effect size and length of followup is
consistent with most research on psychotherapy in general (Lambert & Bergin, 1994). The finding is important because
it suggests no reason to expect a diminishing in PST effects over time.
Prior meta-analyses using some of the studies included in the present meta-analysis showed that PST had a non-
significantly greater effect than treatment as usual on suicide attempt repetition in self-harm clients (Hawton et al.,
1998) but did have a significantly greater effect than treatment as usual on self-reported depression, hopelessness, and
improvement in problems in self-harm clients (Townsend et al., 2001). These two meta-analyses included some studies
that we excluded for various reasons, such as that they examined a “PST” treatment involving both PSTand some other
treatment method. The findings of the present meta-analysis, which covers many more studies with a much broader
range of psychological problems, are consistent with the significant findings of the meta-analysis of Townsend et al.
(2001) and with the non-significant trend found in the smaller meta-analysis of Hawton et al. (1998).
Some cautions apply to interpreting the results. First, the meta-analysis included 31 studies, providing, as in most
meta-analyses, limited power to search for moderators. Second, only published studies were included in the meta-
analysis. Thus, the analysis may have a “publication bias” in that non-significant findings are less likely to be published
than significant findings. However, the fail-safe analysis presented in Table 2 indicates that 136 unpublished studies
would have to exist to negate the significant effect found in this meta-analysis for PST in general. Third, with many
moderator analyses completed, there is the risk of inflated Type I error, leading to spurious findings. However, our use
in analyses of the conservative random effects model reduces this risk to some extent. Fourth, evidence supporting the
moderators is not experimental. For instance, the different studies were not randomly assigned to have assigned
homework or not. Hence, association between whether homework was assigned and effects size may not be causal;
some third variable may have led to both the assignment of homework and the effect level. This same sort of limitation
applies to all the significant moderator findings. Hence, one would better think of significant moderator findings as
suggesting promising avenues for future research with experimental methods than proving some causal link.
The present meta-analysis excluded seven studies of problem solving therapy combined with training in means-end
thinking, consequential thinking, and taking the perspective of others as recommended by Spivack et al. (1976) in their
description of interpersonal cognitive PST. If a few more outcome studies of interpersonal cognitive PST are
54J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
completed, the collection in its entirety might be large enough for a useful meta-analysis, assuming that the reports
provide adequate statistical information.
The present meta-analysis excluded findings of changes in problem solving strategies used. Although such
manipulation-check findings may not be as important as findings regarding client-sought health outcomes,
manipulation-check findings can be important in judging whether PST achieves its intermediate goal of changing
problem-solving behavior. A future meta-analysis could examine this issue.
Future research on PST might also profitably explore what efficacious therapeutic elements it shares with other
types of psychotherapy and what unique efficacious elements it has. For instance it is not clear that PST leads to solving
more real-life problems over time (Foxx & Faw, 2000; Mynors-Wallis, 2002; Tisdale & Lawrence, 1986). Do clients
benefit merely from solving problems during therapy, from building their problem solving self-efficacy (Bandura,
1997), or from solving problems after therapy ends?
*Alexopoulos, G. S., Raue, P., & Arean, P. (2003). Problem-solving therapy versus supportative therapy in geriatric major depression with executive
dysfunction. American Journal of Geriatric Psychiatry 11, 46−52.
Allen, S. M., Shah, A. C., Nezu, A. M., Nezu, C. M., Ciambrone, D., Hogan, J., et al. (2002). A problem-solving approach to stress reduction among
younger women with breast carcinoma: A randomized controlled trial. Cancer 94, 3089−3100.
*Arean, P. A., Perri, M. G., Nezu, A. M., Schein, R. L., Christopher, F., & Joseph, T. X. (1993). Comparative effectiveness of social problem-solving
therapy as treatments for depression in older adults. Journal of Consulting and Clinical Psychology 61, 1003−1010.
*Azrin, N. H., Donohue, B., Teichner, G. A., Crum, T., Howell, J., & DeCato, L. A. (2001). A controlled evaluation and description of individual-
cognitive problem solving and family-behavior therapies in dually-diagnosed conduct-disorder and substance-dependant youth. Journal of Child
and Adolescent Substance Abuse 11, 1−46.
Bandura, A. (1997). Self-efficacy. The exercise of control. New York: Freeman.
*Barrett, J. E., Williams, J. W., Oxmen, T. E., Frank, E., Katon, W., Sullivan, M., et al. (2001). Treatment of dysthymia and minor depression in
primary care. Journal of Family Practice 50, 405−412.
Bennett, D. S., & Gibbons, T. A. (2000). Efficacy of child cognitive-behavioral interventions for antisocial behavior: A meta-analysis. Child and
Family Behavior Therapy 22, 1−15.
*Bennun, I. (1986). Evaluating family therapy: A comparison of the Milan and problem solving approaches. Journal of Family Therapy 8, 225−242.
*Biggam, F. H., & Power, K. G. (2002). A controlled, problem-solving, group based intervention with vulnerable incarcerated young offenders.
International Journal of Offender Therapy and Comparative Criminology 46, 678−698.
*Black, D. R., & Scherba, D. S. (1983). Contracting to problem solve versus contracting to practical behavioral weight loss skills. Behavior Therapy
Burke, B. L., Arkowitz, H., & Menchola, M. (2003). The efficacy of motivational interviewing: A meta-analysis of controlled clinical trials. Journal
of Consulting and Clinical Psychology 71, 843−861.
*Carey, M. P., Carey, K. B., & Meisler, A. W. (1990). Training mentally ill chemical abusers in social problem solving. Behavior Therapy 21,
*Catalan, J., Gath, D. H., Anastasiades, P., & Bond, S. A. (1991). Evaluation of a brief psychological treatment for emotional disorders in primary
care. Psychological Medicine 21, 1013−1018.
*Coche, E., Cooper, J. B., & Petermann, K. J. (1984). Differential outcomes of cognitive and interactional group therapies. Small Group Behavior 15,
*Coche, E., & Flick, A. (1975). Problem-solving training groups for hospitalized psychiatric patients. The Journal of Psychology 91, 19−29.
DeVellis, B. M., Blalock, S. J., Hahn, P. M., DeVellis, R. F., & Hochbaum, G. M. (1988). Evaluation of a problem-solving intervention for patients
with arthritis. Patient Education Counseling 11, 29−42.
Dobson, K. S. (1989). A meta-analysis of the efficacy of cognitive therapy for depression. Journal of Consulting and Clinical Psychology 57,
*Dowrick, C., Dunn, G., Ayuso-Mateos, J. L., Dalgard, O. S., Page, H., Lehtinen, V., et al. (2000). Problem solving treatment and group
psychoeducational for depression: Multicentre randomised controlled trial. British Journal of Psychiatry 321, 1450−1454.
D'Zurilla, T. J., & Goldfried, M. R. (1971). Problem solving and behavior modification. Journal of Abnormal Behavior 78, 107−126.
D'Zurilla, T. J., & Nezu, A. M. (1982). Social problem solving in adults. In P. C. Kendall (Ed.), Advances in cognitive-behavioral research and
therapy (pp. 202−274). New York: Academic.
D'Zurilla, T. J., & Nezu, A. M. (1999). Problem-solving therapy: A social competence approach to clinical intervention. (2nd ed.). New York.
*Epstein, L. H., Paluch, R. P., Gordy, C. C., Saelen, B. E., & Ernst, M. M. (2000). Problem solving in the treatment of child obesity. Journal of
Consulting and Clinical Psycology 68, 717−721.
Feske,U.,& Chambless, D.L.(1995). Cognitivebehavioral versusexposureonly treatmentfor socialphobia:Ameta-analysis. BehaviorTherapy26,
⁎References marked with an asterisk indicate studies included in the meta-analysis.
55J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
Foxx, R. M., & Faw, G. D. (2000). The pursuit of actual problem-solving behavior: An opportunity for behavior analysis. Behavior and Social Issues
Frank, J. (1973). Persuasion and healing: A comparative study of psychotherapy. Baltimore: Johns Hopkins University Press.
Gibbons, J. S., Butler, P., Urwin, P., & Gibbons, J. L. (1978). Evaluation of social work service for self-poisoning patients. British Journal of
Psychiatry 133, 111−118.
Graves, T., Meyers, A. W., & Clark, L. (1988). An evaluation of parental problem-solving training behavioral treatment of childhood obesity. Journal
of Consulting and Clinical Psycology 56, 246−250.
Hawton,K., Arensman, E., Townsend, E., Bremner, S., Feldman,E., Goldney, R., et al. (1998). Deliberate self harm: Systematic review of efficacy of
psychosocial and pharmacological treatments in preventing repetition. British Medical Journal 317, 441−447.
Hedges, L. V. (1981). Distribution theory for glass's estimator of effect size and related estimators. Journal of Educational Statistics 6, 107−128.
Hedges, L. V. (1982). Estimation of effect size from a series of independent experiments. Psychological Bulletin 92, 490−499.
Heppner, P. P., & Peterson, C. H. (1982). The development and implications of a personal problem-solving inventory. Journal of Counseling
Psychology 30, 537−545.
Hussian,R.A.,&Lawrence,S. L.(1981).Socialreinforcementof activityandproblem-solvingtraininginthetreatmentofdepressedinstitutionalized
elderly patients. Cognitive Therapy and Research 5, 57−69.
Intagliata,J. C. (1978). Increasing the interpersonal problem-solving skills of an alcoholic population. Journal of Consultingand Clinical Psychology
Johnson,S.M.,& Greenberg,L.S.(1985). Differentialeffectsofexperientialandproblem-solving interventionsinresolvingmaritalconflict.Journal
of Consulting and Clinical Psychology 53, 175−184.
Kazantzis, N., Deane, F. P., & Ronan, K. R. (2000). Homework assignments in cognitive and behavioral therapy: A meta-analysis. Clinical
Psychology: Science and Practice 7, 189−202.
Kazdin, A. E., Esveldt-Dawson, K., French, N. H., & Unis, A. S. (1987). Problem-solving skills training and relationship therapy in the treatment of
antisocial child behavior. Journal of Consulting and Clinical Psychology 55, 76−85.
Koss, M. P., & Shiang, J. (1994). Research on brief psychotherapy. In A. E. Bergin& S. L. Garfield (Eds.), Handbook of psychotherapy and behavior
change (pp. 664−700). 4th ed. New York: Wiley.
Lambert, M. J., & Bergin, A. E. (1994). The effectiveness of psychotherapy. In A. E. Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and
behavior change (pp. 143−189). 4th ed. New York: Wiley.
Leichsenring, F., & Leibing, E. (2003). The effectiveness of psychodynamic therapy and cognitive behavior therapy in the treatment of personality
disorders: A meta-analysis. American Journal of Psychiatry 160, 1223−1232.
*Lerner, M. S., & Clum, G. A. (1990). Treatment for suicide ideators: A problem-solving approach. Behavior Therapy 21, 403−411.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, USA: Sage Publications.
*Malouff, J. M., Lanyon, R. I., & Schutte, N. S. (1988). Effectiveness of a brief group RET treatment for divorce-related dysphoria. Journal of
Rational-Emotive and Cognitive-Behavior Therapy 3, 162−171.
*McLeavey, B. C., Daly, R. J., Ludgate, J. W., & Murray, C. M. (1994). Interpersonal problem-solving skills training in the treatment of self-
poisoning patients. Suicide and Life-Threatening Behavior 24, 382−394.
McRoberts, C., Burlingame, G. M., & Hoag, M. J. (1998). Comparative efficacy of group versus individual psychotherapy. Group Dynamics 2,
Moyer, C. A., Rounds, J., & Hannum, J. W. (2004). A meta-analysis of massage therapy research. Psychological Bulletin 130, 3−18.
Mynors-Wallis, L. M. (2002). Does problem-solving treatment work through resolving problems? Psychological Medicine 32, 1315−1319.
*Mynors-Wallis, L. M., Gath, D., Lloyd-Thomas, A. R., & Tomlinson, D. (1995). Randomised controlled trial comparing problem solving treatment
with amitriptyline and placebo for major depression in primary care. General Practice 310, 441−445.
*Mynors-Wallis, L. M., Davies, I., Gray, A., Barbour, F., & Gath, D. (1997). A randomised controlled trial and cost analysis of treatment for
emotional disorders given by community nurses in primary care. British Journal of Psychiatry 170, 113−119.
*Mynors-Wallis, L., Gath, D. H., Day, A., & Baker, F. (2000). Randomised controlled trial of problem solving treatment, antidepressant medication,
and combined treatment for major depression in primary care. British Medical Journal 320, 26−30.
*Nezu, A. M. (1986). Efficacy of a social problem-solving therapy approach for unipolar depression. Journal of Consulting and Clinical Psycology
Nezu, A. M. (2004). Problem solving and behavior therapy revisited. Behavior Therapy 35, 1−33.
*Nezu, A. M., & Perri, M. G. (1989). Social problem-solving therapy for unipolar depression: An initial dismantling investigation. Journal of
Consulting and Clinical Psycology 57, 408−413.
*Nezu, A. M., Nezu, C. M., McClure, K. S., Felgoise, S. H., & Houts, P. S. (2003). Project genis: Assessing the efficacy of problem-solving therapy
for distressed adult cancer patients. Journal of Consulting and Clinical Psycology 71, 1036−1048.
Norris, S. L., Zhang, X., Avenell, A., Gregg, E., Bowman, B., & Serdula, M. e. a. (2004). Long-term effectiveness of lifestyle and behavioral weight
loss intervention in adults with type 2 diabetes: A meta-analysis. American Journal of Medicine 117, 762−774.
*Perri, M. G., Nezu, A. M., McKelvey, W. F., Shermer, R. L., Renjilian, D. A., & Viegener, B. J. (2001). Relapse prevention training and problem-
solving therapy in long term management of obesity. Journal of Consulting and Clinical Psychology 69, 722−726.
*Pfiffner, L. J., Jouriles, E. N., Brown, M. M., Etscheidt, M. A., & Kelly, J. A. (1990). Effects of problem-solving therapy on outcomes of parent
training for single-parent families. Child and Family Behavior Therapy 12, 1−11.
*Sahler, O. J. Z., Varni, J. W., Fairclough, D. L., Butler, R. W., Noll, R. B., Dolgin, M. J., et al. (2002). Problem-solving skills training for mothers of
children with newly diagnosed cancer: A randomized trial. Journal of Developmental and Behavioral Pediatrics 23, 77−86.
*Salkovskis, P. M., Atha, C., & Storer, D. (1990). Cognitive-behavioural problem solving in the treatment of patients who repeatedly attempt suicide:
A controlled trial. British Journal of Psychiatry 157, 871−876.
J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57
*Schwartz, M. D., Lerman, C., Audrian, J., Cella, D., Rimer, B., Stefanek, M., et al. (1998). The impact of a brief problem-solving training Download full-text
intervention for relatives of recently diagnosed breast cancer patients. Annals of Behavioral Medicine 20, 7−12.
*Spaccarelli, S., Cotler, S., & Penman, D. (1992). Problem-solving skills training as a supplement to behavioral parent training. Cognitive Therapy
and Research 16, 1−18.
Spiegler, M. D., & Guevremont, D. C. (2003). Contemporary behavior therapy (4th ed.). Belmont, CA: Thomson Wadsworth.
Spivack, G., Platt, J. J., & Shure, M. B. (1976). The problem solving approach to adjustment. San Francisco: Jossey-Bass.
Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Needham Heights, MA: Allyn & Bacon.
Tisdale, D. A., & Lawrence, J. S. (1986). Interpersonal problem-solving competency: Review and critique of the literature. Clinical Psychology
Review 6, 337−356.
Townsend, E., Hawton, K., Altman, D. G., Arensman, E., Gunnel, D., Hazell, P., et al. (2001). The efficacy of problem-solving treatment after
deliberate self-harm: Meta-analysis of randomised controlled trials with respect to depression, hopelessness and improvement in problems.
Psychological Medicine 31, 979−988.
*van den Hout, J. H. C., Vlaeyen, J. W. S., Heuts, P. H. T. G., Zijlema, J. H. L., & Wijnen, J. A. G. (2003). Secondary prevention of work-related
disability in nonspecific low back pain: Does problem-solving therapy help? A randomized clinical trial. The Clinical Journal of Pain 19, 87−96.
Wampold, B. E., Mondi, G. W., Moody, M., Stich, F., Benson, K., & Ahn, H. (1997). A meta-analysis of outcome studies comparing bona fide
psychotherapies: Empirically, “All must have prizes”. Psychological Bulletin 122, 203−215.
*Williams, J. W., Barrett, J., Oxman, T., Frank, E., Katon, W., Sullivan, M., et al. (2000). Treatment of dysthymia and minor depression in primary
care. Journal of the American Medical Association 284, 1519−1526.
Wolf, F. M. (1986). Meta-analysis: Quantitative methods for research synthesis. London: Sage Publications.
57J.M. Malouff et al. / Clinical Psychology Review 27 (2007) 46–57