Depressive realism: A meta-analytic review
Michael T. Moore⁎,1, David M. Fresco
Kent State University, Kent, OH, USA
a b s t r a c ta r t i c l ei n f o
Received 27 January 2011
Revised 7 May 2012
Accepted 10 May 2012
Available online 22 May 2012
The current investigation represents the first meta-analysis of the depressive realism literature. A search of
this literature revealed 75 relevant studies representing 7305 participants from across the US and Canada,
as well as from England, Spain, and Israel. Results generally indicated a small overall depressive realism effect
(Cohen's d=−.07). Overall, however, both dysphoric/depressed individuals (d=.14) and nondysphoric/
nondepressed individuals evidenced a substantial positive bias (d=.29), with this bias being larger in non-
dysphoric/nondepressed individuals. Examination of potential moderator variables indicated that studies
lacking an objective standard of reality (d=−.15 versus −.03, for studies possessing such a standard)
and that utilize self-report measures to measure symptoms of depression (d=.16 versus −.04, for studies
which utilize structured interviews) were more likely to find depressive realism effects. Methodological
paradigm was also found to influence whether results consistent with depressive realism were found
(d's ranged from −.09 to .14).
© 2012 Elsevier Ltd. All rights reserved.
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.Beck's theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2. Depressive realism hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3. Boundaries and potential functions of depressive realism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3.1. Situational constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3.2.Individual constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4.Critique of the depressive realism literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5.The present study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.Search procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.Coding procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3. Statistical procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Major Depressive Disorder (MDD) is a prevalent and debilitating
(Kessler et al., 2003), MDD hadthehighest lifetime and 12-monthprev-
Depression affects over 13 million individuals per year in the United
larger than the costs of heart disease (Greenberg, Stiglin, Finkelstein, &
Berndt, 1993). Cognitive therapy of depression (Beck, Rush, Shaw, &
pression (e.g., Blackburn & Moorhead, 2001; DeRubeis & Crits-Cristoph,
1998). The theory underlying cognitive therapy posits that the de-
pressed individual is negatively biased in their perceptions, while the
Clinical Psychology Review 32 (2012) 496–509
648 Beacon St., 6th Fl., Boston, MA 02215, USA. Tel.: +1 617 353 9610; fax: +1 617 353
E-mail address: email@example.com (M.T. Moore).
1Now at the Center for Anxiety and Related Disorders, Boston University, Boston,
0272-7358/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
Contents lists available at SciVerse ScienceDirect
Clinical Psychology Review
primary goal in cognitive therapy is returning these individuals to a
more objective state (Beck et al., 1979). However, there is research
that has shown that the depressed individual may be better able to
make certain judgments thannondepressed individuals, a phenomenon
referred to as “depressive realism” (see Alloy & Abramson, 1988, for a
review). The literature, regarding how best to characterize the cogni-
tions of depressed individuals, is mixed in its support. This debate
calls into question how it is that cognitive therapy exerts its therapeu-
tic effect. If depressed individuals may be less biased in their ability to
process information than their nondepressed peers (the position of
depressive realism), then how does cognitive therapy work? A recent
review (Longmore & Worrell, 2007) of the literature which investigat-
ed mediators of cognitive–behavioral therapy critiqued the lack of
research demonstrating that cognitive change precedes symptom
change. In addition, the review highlighted research that demonstrat-
ed that symptom change in cognitive–behavioral therapy may either
precede cognitive change or occur in its absence. While research
has consistently demonstrated that cognitive therapy is an effective
treatment for depression, knowledge of how it results in therapeutic
change can result in refinements of the treatment. These refinements
can potentially make cognitive therapy more concentrated, cost-
effective, and hence, available to more of the millions of people who
suffer from this debilitating condition. While the current study repre-
it is important to understand more specifically how this literature
differs from the prevailing theory on the cognition of depressed
1.1. Beck's theory
Beck's (1967, 1987) theory, which formed the basis for cognitive
therapy, posits that depressed affect is heavily influenced by recur-
rent thoughts with negative content, or automatic thoughts. These
thoughts arise from deeply-held dysfunctional beliefs, or schemas,
relating to the self, world, and future (e.g., “If I fail, no one will love
me”). Beck identified that schemas and automatic thoughts, and the
depressed affect that results from them, tend to be self-perpetuating
as the depressed person both attends more to negative events in
their lives and interprets events that occur after the onset of the
depressed mood in light of their own dysfunctional cognitions.
Beck (1987) characterizes the cognition of depressed individuals as
“schema-driven” and depressed individuals themselves as possessing
“depressive cognitive distortions.” The thoughts of nondepressed
individuals, however, are characterized as “data-driven” and he de-
scribed nondepressed individuals as possessing “nondepressive
accuracy,” implying that depressed individuals' cognitions are sys-
tematically less informed by reality and, hence, more irrational. For
instance, a depressed person may experience a significant success,
but may minimize the importance of that event as due to chance
because they believe that they are a failure. One of the primary
goals of cognitive therapy for depression (Beck et al., 1979) is teach-
ing depressed individuals to analytically monitor their own negative
thoughts. This monitoring is done in service of both challenging
and replacing these “schema-driven” thoughts with more accurate
1.2. Depressive realism hypothesis
The “depressive realism hypothesis” (Alloy & Abramson, 1979)
presented an alternative view to both conventional clinical wisdom
and Beck's theory (1967, 1987) of the cognition of the depressed per-
son. Research supportive of depressive realism illustrated not only
that depressed individuals can make realistic inferences, but that
they could do so to a greater extent than nondepressed individuals
under certain circumstances. The first evidence for this phenomenon
came in the form of studies utilizing what is called the “judgment of
contingency task.” In this task, participants are asked to press a
button, which results in the illumination of a light a percentage of
the time that is predetermined by the experimenter. The dependent
variable is the participant-rated contingency between pressing the
button and the illumination of the light. As such, there are two factors
that the participant needs to attend to: the occurrence of the outcome
(i.e. light illumination) in the presence of the response (i.e. button
press) and the occurrence of the outcome in the absence of the
response. Higher positive contingencies result when the outcome
occurs at a higher rate in the presence of the response than in its
absence (i.e. button non-press). Negative contingencies are also
possible where the outcome is less likely to occur in the presence of
the response than in its absence (i.e. if pressing the button suppressed
the illumination of the light). Consistent with the depressive realism
effect, depressed individuals have been shown to more accurately
make these kinds of judgments than nondepressed individuals
(Alloy, Abramson, & Kossman, 1985; Alloy, Abramson, & Viscusi,
1981; Musson & Alloy, 1987; Vazquez, 1987). Nondepressed individ-
uals experienced what has been referred to as an “illusion of control,”
where they overestimated their degree of control over the outcome.
Depressed individuals experienced no such bias. In addition, these
results were replicated over a variety of differing predetermined
contingency conditions (Abramson, Alloy, & Rosoff, 1981; Alloy &
Abramson, 1979; Dobson & Pusch, 1995; Ford & Neale, 1985; Martin,
Abramson, & Alloy, 1984; Msetfi, Murphy, Simpson, & Kornbrot, 2005;
Presson & Benassi, 2003; Vazquez, 1987).
Despite the number of studies utilizing the judgment of contin-
gency task, not all of the research in support of depressive realism
has used this methodological paradigm. Other methodological para-
digms, referred to as self-evaluation of task performance (Gotlib,
1983; Lobitz & Post, 1979; Rozensky, Rehm, Pry, & Roth, 1977) and re-
call of feedback studies (DeMonbreun & Craighead, 1977; Dennard &
Hokanson, 1986; Nelson & Craighead, 1977) have also produced find-
ings compatible with depressive realism. Studies examining the self-
evaluation of task performance have participants engage in a task,
then rate their performance on that task without the benefit of feed-
back. The participants' self-performance is then compared to their ac-
tual performance to determine how accurately it was perceived. In
research examining the recall of feedback, ratings of the participants'
performance is given immediately after each subtask is completed,
and the participants are then asked to rate their aggregate level of
performance across the task as a whole. The participants' recall of
the feedback they received is compared to the actual feedback to de-
termine how accurate their recall was. In many studies (DeMonbreun
& Craighead, 1977; Dennard & Hokanson, 1986; Gotlib, 1983; Lobitz &
Post,1979; Nelson & Craighead, 1977;Rozenskyetal., 1977),depressed
Studies comparing expectancies of success on various tasks with
depressed and nondepressed individuals have replicated these find-
ings as well (Alloy & Abramson, 1980; Alloy & Seligman, 1979;
Golin, Terrel, Weitz, & Drost, 1979; Golin, Terrell, & Johnson, 1977).
In many of these studies, the predictions of future success of
depressed and nondepressed individuals are compared on both
chance-tasks as well as tasks designed to appear skill-determined
(but are actually chance-determined), both prior to and immediately
after reinforcement or punishment. Smaller changes in expectancies
of success by nondepressed relative to depressed individuals have
been found following reinforcement or punishment in the tasks
designed to appear skill-based (Alloy & Abramson, 1980; Alloy &
Seligman, 1979). Insofar as performance is expected to improve on
skill-determined tasks, the findings that expectancies of the nonde-
pressed participants do not change as much as the depressed partic-
ipants is taken as evidence of perceptual bias in nondepressed
participants. These differences between depressed and nondepressed
participants have not been found using chance-determined tasks, where
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
performance would not be expected to improve (Alloy & Abramson,
1980;Alloy&Seligman,1979).Taken together, the aforementioned re-
sults have been interpreted by proponents of depressive realism as
evidence that the depressed individual more accurately perceives their
performance on these tasks.
1.3. Boundaries and potential functions of depressive realism
Although the above-mentioned research attests to the robustness
and generalizability of the depressive realism phenomenon, there are
studies that report circumstances under which depressive realism ef-
fects are not obtained (Ahrens, 1986; Alloy & Abramson, 1988; Alloy &
Ahrens, 1987; Benassi & Mahler, 1985; Buchwald, 1977; DeMonbreun
& Craighead, 1977; Dennard & Hokanson, 1986; Hoehn-Hyde,
Schlottman, & Rush, 1982; Loewenstein & Hokanson, 1986; Moore &
Fresco, 2007; Nelson & Craighead, 1977; Sacco & Hokanson, 1978,
1982; Siegel & Alloy, 1990; Tennen & Herzberger, 1987; Vazquez,
1987; Vestre & Caulfield, 1986; Wenzlaff & Berman, 1985). These
boundaries, in turn, suggest how depressive realism may fit into pre-
existing theory in social psychology and psychopathology. Alloy and
Abramson (1988), in their comprehensive narrative review of the
depressive realism literature, identified six boundary conditions on
depressive realism that possessed some degree of research support.
Four of these conditions refer to constraints related to situations and
two refer to constraints related to the individual.
1.3.1. Situational constraints
perceived. Although the overwhelming majority of depressive realism
research has asked participants to make judgments or otherwise report
judgments of the self versus judgments of another person between de-
pressed and nondepressed persons (Ahrens, 1986, 1991; Ahrens, Zeiss,
& Kanfer, 1988; Alloy & Abramson, 1988; Gotlib & Meltzer, 1987; Javna,
Alloy, 1990; Vazquez, 1987). Results have shown that nondepressed
participants demonstrate a positive bias in their perceptions of their
own performance, but no bias in the perceptions of the performance
of others. In addition, depressed participants demonstrate relatively re-
perceptions of others' performance (see Gotlib & Meltzer, 1987; Javna,
1981; Pyszczynski et al., 1987, and the performance of females in
Martin et al., 1984 for exceptions).
The second of the situational constraints is whether the judgment
or perception is made in public or private (Benassi & Mahler, 1985;
Sacco & Hokanson, 1978, 1982; Strack & Coyne, 1983). Findings indi-
cate that the cognitions of nondepressed individuals are more opti-
mistic in public than in private, while the cognitions of depressed
individuals are less responsive to the presence of others (see Strack
& Coyne, 1983 for an exception to this trend).
The third situational constraint is whether the perception is made
immediately or after a delay between the to-be-perceived stimulus
and when the perception is assessed. Even among studies utilizing
the recall of feedback paradigm, only three studies directly compared
immediate perceptions to those made after a delay (DeMonbreun &
Craighead, 1977; Nelson & Craighead, 1977; Wenzlaff & Berman, 1985).
Both DeMonbreun and Craighead (1977) and Nelson and Craighead
(1977) found that, while depressed participants' immediate perceptions
were typically accurate, their memories made after a delay were nega-
itive bias in both their immediate perceptions as well as their memories.
Wenzlaff and Berman (1985) found similar results, with the significant
exception that they found both the perceptions and memories of
depressed participants to be accurate.
be-perceived stimulus is ambiguous (i.e. explicitly neutral feedback) or
unambiguous (i.e. clearly positive or negative feedback or information).
Only one study has been conducted which has explicitly made this com-
parison. Dykman, Abramson, Alloy, and Hartlage (1989) evaluated the
encoding of both ambiguous and unambiguous information which were
both consistent and inconsistent with prior, deeply-held beliefs about
the self. Results indicated that only ambiguous feedback was conducive
to differential encoding by depressed and nondepressed participants.
1.3.2. Individual constraints
Alloy and Abramson (1988) also identified two constraints which
involve individual factors which have some degree of research sup-
port. The first of these constraints is the severity of the depressive dis-
order under study. Several theorists have suggested that perceptual
bias and depression may not be related in a monotonically increasing
function, where degree of bias is correlated with degree of depression
(e.g., Beck, 1986; Evans & Hollon, 1988; Ruehlman, West, & Pasahow,
1985). These authors have posited that nondepressed individuals may
be characterized by positive biases, mildly depressed individuals by
more realistic perceptions, and severely depressed individuals may be
characterized by the negative perceptual and memory biases hypothe-
sized by Beck (1967, 1976). Two studies (Dennard & Hokanson, 1986;
Loewenstein & Hokanson, 1986) which have directly addressed this
dents and bothhave found thesegroups tobe equally accurate.Howev-
er, McKendree-Smith and Scogin (2000) compared the perceptions of
bogus, neutral personality test feedback in nondepressed, mildly, and
moderately/severely depressed college students. They found that the
nondepressed and mildly depressed students rated their profiles more
positively than the moderately/severely depressed students. Unfortu-
nately, this study did not address the issue of realism, per se, as it was
impossible to determine which interpretation was the “correct” one,
given the lack of an objective comparison (i.e., the students' actual
Lastly, it is possible that perceptual bias is not caused by depressed
mood at all, but by some, as yet unidentified third variable(s) that is
correlated with depressed mood such as self-esteem (Tennen &
Herzberger, 1987, but see Crocker, Alloy, & Tabachnik-Kayne, 1988
for a failure to replicate), dysfunctional attitudes (Bynum & Scogin,
1996), or attributional style. Moore and Fresco (2007) examined the
depressive realism effect in the context of a well-validated, cognitive
diathesis–stress theory of the etiology of a subtype of depression,
hopelessness theory (Abramson, Metalsky, & Alloy, 1989). Of inter-
est is the finding that attributional accuracy was more closely related
to attributional style (both attributional accuracy and style were
measured with different instruments) than it was to symptoms of
ism, numerous studies have provided less favorable results. Even within
the seminal Alloy and Abramson's (1979) paper in which depressive re-
periment 1) failed to produce depressive realism results altogether,
while other conditions (see the noncontingency, low-density reinforce-
ment condition in Experiment 2 and Experiment 4) failed to produce
the illusion of control in nondysphoric participants. Studies assessing
the accuracy of depressed and nondepressed persons' delayed recall of
both task-performance (Craighead, Hickey, & DeMonbreun, 1979;
DeMonbreun & Craighead, 1977) and ambiguous personality feedback
(Dykman et al., 1989; Gotlib, 1983; Vestre & Caulfield, 1986) have ret-
erature examining the accuracy of recall of task-performance feedback
has returned consistently similar results for ambiguous feedback
(Craighead et al., 1979; DeMonbreun & Craighead, 1977). Depressed in-
receive and nondepressed individuals have been shown to overestimate
it (Buchwald, 1977; Wener & Rehm, 1975), illustrating bias among both
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
1.4. Critique of the depressive realism literature
In addition to this empirical inconsistency, the methodology of
some of the literature in support of a depressive realism effect has
been cogently undermined. There are three primary critiques of the
depressive realism literature.
Critique 1: lack of gold standard
Much of the research on the depressive realism effect has been crit-
icized for not including a “gold-standard” of reality with which to
compare participants' perceptions of events. This criticism seems
to call into question the “realism” of the depressive realism hypoth-
esis.Critiques of thedepressiverealismliteraturecomes from sever-
al theorists (e.g., Ackermann & DeRubeis, 1991; Alloy & Abramson,
1988; Haaga & Beck, 1995), who perceptively note that much of
the aforementioned research cannot be said to support depressive
realism unequivocally as no objective standard of reality exists
with which to compare many of the participants' ratings. Without
a “gold-standard” measure of reality, it is theoretically impossible
to state that one group or another's ratings are more or less “realis-
tic.” It should be noted that Critique 1, the lack of a gold standard of
reality, regards whether or not bias can be validly assessed, not
whether or not it is present. Experimental stimuli lacking a gold
standard are not biased, they simply cannot be said to evaluate
claims relevant to depressive realism. Bias would be demonstrated
by the perceptions of a participant to stimuli that possess an objec-
tive standard of reality. In the current investigation, a study was
said to possess an objective standard of reality to the extent that
the stimuli, being described by the participant, could be described
in an unbiased fashion at the time it was perceived. For instance,
in much of the research into the expectancies of success of de-
pressed and nondepressed persons, there is no objective standard
of realitywithwhich to compare a prediction ofthe future orexpec-
tancy atthe time that theratingismade. Whether or not thepredic-
cannot be known by the participant at the time the predictions are
made (before the prediction does or does not come to pass). As a re-
sult, other interpretations of the results of the expectancy studies
can be plausibly offered. Ackermann and DeRubeis (1991) give the
example of a nondepressed individual who may not decrease their
expectancies of success following punishment for poor perfor-
mance, thereby overestimating his/her chance of success, with the
expectation that practice will improve their future performance.
Without knowledge of how these individuals have benefited from
feedback about their performance and practice in the past, it is im-
alistic,”ornot. Itshouldbe noted,however, thatnot allresearchinto
expectancies of success fails to address this critique. Some studies
asked participants to predict their success on an explicitly-labeled,
chance-determined task with an objective probability of success
which was readily-discernable (e.g., Alloy et al., 1981, 1985; Golin
et al., 1977, 1979; Lewinsohn, Mischel, Chaplain, & Barton, 1980).
The expectation that practice will improve performance on a task
gle number on the roll of a die. Studies utilizing the judgment of
contingency task are also excellent examples of research that pro-
vides such a gold standard. Participants are asked to rate the contin-
gency between pressing a button and the illumination of a light,
while this contingency is objectively manipulated by the experi-
menter and known precisely in advance.
Studies of expectancies of success or future performance were not
classified as possessing a gold standard of reality in the current in-
vestigation; however, this is not to say that these studies have not
made important contributions to the study of depression. The
study of expectations of future positive events has important im-
plications for hopelessness, suicide, and risk for future episodes
of depression. The issue of excluding expectancy studies given
their importance to the field of depression raises the related
issue of how the topics “depressive realism” and “cognitive thera-
py of depression” are related. It is important to recognize that
these two topics are related, and not identical; part of the interest
of depressive realism lies in the fact that its predictions run oppo-
site to those of cognitive therapy of depression. However, Beck's
theory is much more expansive than depressive realism. It can-
vasses not only the presence of cognitive and perceptual biases
in the depression, but also how such biases are causal to depres-
sive disorder, and how alleviating such biases results in alleviation
of the disorder. A meta-analysis attempting to cover every study of
relevance to such a theory, even if only constrained to studies
using depressed samples, would be lengthy indeed. Inclusion of
expectancy studies may be argued on the pragmatic grounds of
their importance to the field of cognitive therapy. However, this
argument conflates depressive realism and cognitive therapy of
Critique 2: inadequate assessment of depression
The ability of self-report measures to validly assess clinical depres-
sion has also been called into question (Kendall, Hollon, Beck,
Hammen, & Ingram, 1987). Other critiques of the depressive realism
literature (Dobson & Franche, 1989; Haaga & Beck, 1995) highlight
the fact that most of the studies that compose this literature use
self-report measures, as opposed to structured clinical interviews,
to assess whether participants are “depressed” or “nondepressed.”
As a result, this criticism would seem to call into question whether
the depressive realism phenomenon really concerns “depression” at
all. Some have suggested that these individuals should be labeled as
depressed as clinical depression is predicated on several criteria not
captured by self-report measures of depression (e.g., functional im-
pairment; Kendall et al., 1987). In addition, self-report measures of
depression are ineffective at the differential diagnosis of major de-
pressive disorder and dysthymia,the conditionsof interest, from re-
lated disorders, such as bipolar disorder. Individuals with bipolar
disorder would alsobepredicted to score highly on self-report mea-
sures of depression while in the depressive phase of their illness. As
a result, it is possible that many of the participants labeled in past
studies of depressive realism may not have suffered from depres-
sion, per se. Despite the aforementioned critique, however, research
which has investigated depressive realism claims in both dysphoric
and clinically depressed participants (Dunn, Dalgleish, Lawrence, &
Ogilvie, 2007) have found similar positive biases in both groups.
Critique 3: limited external validity
Some theorists have critiqued the use of the judgment of contin-
gency task or other laboratory tasks to assess the realism in
people's perceptions of events (Dobson & Franche, 1989; Haaga
& Beck, 1995). Systematic variation in experimental findings has
been noted seemingly to indicate that more robust depressive re-
alism effects are found in less externally valid, laboratory tasks. In ad-
dition, evidence of perceptual bias in depressed participants has been
2We are thankful to an anonymous reviewer for making this comment.
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
found in tasks that more closely mimic the judgments people make
outside of the laboratory (Dobson & Franche, 1989; Moore & Fresco,
2007). This finding implies that the depressive realism effect may
merely be an artifact of a particular type of task, or constrained to lab-
ical artifact than a clinically-useful phenomenon.
1.5. The present study
Although previous reviews of the depressive realism literature
(Dobson & Franche, 1989) have attempted to resolve the empirical het-
erogeneity in obtained results, a largely qualitative, “vote-counting”
of studies finding in favor of or against a particular hypothesis is tallied,
ally, vote-counting relies exclusively on statistical significance and there-
fore ignores the size of the effects obtained in various studies. As a result,
it has been criticized as more likely to result in biased conclusions than
those based on more quantitative methods of research synthesis
(Bangert-Drowns, 1986; Glass, McGaw, & Smith, 1981). In addition, no
previous attempt to review the depressive realism literature has
accounted for the three critiques mentioned above. The current study
sought to quantitatively synthesize the literature on depressive realism
with the hopes of resolving the empirical heterogeneity of findings
while at the same time addressing the three aforementioned critiques.
1. Consistent with expectations from the depressive realism hypothesis,
effects averaged across studies will show less perceptual/attentional
bias in dysphoric/depressed versus nondysphoric/nondepressed
2. Examination of the direction of bias in dysphoric/depressed and
nondysphoric/nondepressed groups in isolation from one another
will indicate that nondysphoric/nondepressed individuals will be
biased toward positive stimuli, whereas dysphoric/depressed indi-
viduals will not evidence any such bias (consistent with the findings
of depressive realism).
depressive realism effects than studies that do not (see Critique 1).3
No research has yet been conducted which has quantitatively evalu-
ated the impact of this variable on the depressive realism effect. As a
result, this hypothesis is largely exploratory. However, it is felt that
Critique 1 is the most theoretically substantive of those listed
above and has been included for this reason.
alism effect (see Critique 2). Specifically, studies that utilize struc-
tured clinical interview will produce larger depressive realism
effects than studies that utilize self-report, as it is thought that the
former will result in more homogenous depressed/nondepressed
groups (thereby increasing resulting effect sizes).
5. Theexternalvalidityof thestudywill serveasa moderator of thede-
pressive realism effect (see Critique 3). Dobson and Franche (1989)
noted that much of the support for depressive realism came in the
ception outside of the laboratory (e.g., the judgment of contingency
task). Studies which lack external validity would be expected to
make this sacrifice at the expense of increase internal validity. We
would expect that this increased control for extraneous variables
would result in reduced error variance and larger depressive realism
will produce larger differences between dysphoric/depressed and
nondysphoric/nondepressed individuals and, therefore, larger de-
pressive realism effects.
Although it would have been ideal to evaluate the validity of the six
boundary conditions on depressive realism mentioned above, several
factors prevented these analyses from being statistically and methodo-
logically feasible. For the self- versus other-reference and public versus
private conditions, the majority of the research conducted does not ade-
quately address Critique 1. Most of the authors investigating the percep-
tion of self versus other were primarily interested in relative differences
on this variable. As a result, establishing which version of the percept
was “right” (self or other) was not a primary aim of this research. With
regard to theliterature evaluating thedepressive realism effect in public
versus private conditions, only three studies have been conducted mak-
ing this comparison. Of these three studies, only two studies addressed
Critique 1 and, of these two studies, information necessary to be useful
in this meta-analysis could not be obtained for one of them. A similar
biguous stimuli and severity of depression. With regard to the examina-
tion of ambiguous versus unambiguous stimuli, only one study was
found. Two studies have examined the relationship between severity
of depression and the depressive realism effect. However, only one of
these studies adequately addresses Critique 1 and, lamentably, informa-
tion necessary to be useful in this meta-analysis could not be obtained
from it. While sufficient number of studies have been conducted using
both immediate and delayed perceptions, this hypothesis would be al-
adigm to other methodological paradigms. This paradigm is primarily
differentiated from the self-evaluation of task performance paradigm
by the delayed nature of the perception in question. Because the effects
adigm under which it was evaluated, a comparison of immediate and
delayed perceptions was not included in the present investigation.
2.1. Search procedure
The current investigation attempted to obtain data from as many
studies relevant to depressive realism as possible. However, it was
outside the scope of this study to attempt to canvass certain closely-
related research areas. Studies utilizing the emotional Stroop and
dot probe tasks in depressed and nondepressed individuals were
not included in the current investigation. This exclusion was made
on practical grounds as these studies could, and have (cf. MacLeod,
Mathews, & Tata, 1986), composed their own, quite voluminous
meta-analysis. The current investigation also did not examine the to-
tality of studies examining memory biases in depression. This was
done because much of the research examining memory biases does
not attempt to directly evaluate depressive realism. Much of this litera-
ture attempts to demonstrate that depressed individuals preferentially
recall negatively-valenced material and nondepressed individuals pref-
erentially recall positively-valenced information. This paradigm at-
tempts to evaluate differences between groups, but not the systematic
biases that are the hallmark of depressive realism. In other words, this
paradigm assumes that neither group is more biased, simply that both
are biased equally under differing circumstances. In circumstances
where this type of paradigm is not utilized (e.g., the recall of feedback
paradigm), these studies were included.
Relevant studies were located by first conducting a search of Psy-
cINFO using the search terms “depressive realism,” “illusion of control,”
“cognitive distortion,” and “judgmentof contingency.” Relevant articles
were also selected via a thorough search of studies cited in already-
3It should be noted that the aforementioned hypothesis merely seeks to evaluate
whether an objective standard of reality is a moderator of the depressive realism effect.
It does not seek to quantitatively evaluate Critique 1. Critiques 2 and 3 argue that poor
assessment and lack of external validity moderate depressive realism effects. Critique
1, on the other hand, argues that an objective standard of reality is an absolute, theo-
retical necessity when evaluating theory relevant to depressive realism, regardless of
whether or not this variable exerts any influence on the results of said evaluation.
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
located articles. This latter selection method allowed for detection of
unpublished sources (e.g. theses, dissertations, conference presenta-
tions) that are more likely to report results which are not statistically
significant; addressing the so-called “file drawer problem” (Rosenthal,
1979). Relevant articles were defined as any study that: (1) could be
said to examine perceptual accuracy and (2) did so via comparison of
ipants. With regard to the former criteria, we used the rather liberal
threshold of any study that purported to investigate bias, as defined in
a non-relativistic manner. This criterion primarily excluded studies
which examined perceptual differences and made no claims about ac-
curacy, such as the memory bias studies described above.
2.2. Coding procedure
Each study was coded as to: whether the dependent variable(s)
could be compared to an objective standard of reality, how depres-
sion was assessed, which methodological paradigm was used, and
the degree to which this method was externally valid. Studies
where the dependent variable used to assess realism was compared
to an objective standard of reality (and, therefore, addressed Critique
1) were compared to studies that did not utilize an objective standard
of reality to determine what influence this potential moderator vari-
able has on the magnitude of depressive realism results obtained.
Studies which examined the differential expectancies or predictions
of future success on a skill-determined task(s) by depressed and non-
depressed individuals did not utilize an objective standard of reality.
As a result, these studies did not address Critique 11 and were catego-
rized accordingly. Studies, which compared self-perceptions to the
perception of others in depressed and nondepressed participants
(i.e., without attempting to determine if either of these perceptions
were more realistic or objective) were also coded as not having
addressed Critique 1. As mentioned previously, studies which do not
address Critique 1, and do not possess an objective standard of reality,
cannot be said to evaluate depressive realism, unequivocally. Insofar
as studies which do not address Critique 1 are not directly relevant to
the depressive realism literature, only studies whichaddressedCritique
1 were utilized in the evaluation of our hypotheses (with the obvious
exception of Hypothesis 3).
To address Critique 2 (that depressive realism studies really assess
dysphoria instead of depression), studies that address Critique 1 were
coded as to how depression was assessed. Studies that utilized clinical
interview were compared to studies that, instead, utilized just self-
report, to determine whether method of assessment of depression
served as a moderator variable of the depressive realism effect. To ad-
dress Critique 3 (that the depressive realism effect may not be repli-
cable outside of the laboratory), studies that satisfy Critique 1 were
also coded on the degree of external validity present in the dependent
variable (High versus Low) to determine the influence of this moder-
ator. Studies where the experimental task closely mimicked judg-
ments made outside of the laboratory would be rated “High.” How a
study was coded was a function of both aspects of the context and
methodology (stimuli presented via computer versus interaction
with a confederate) as well as the nature of the variable itself. In the
case of a participant asked to judge their performance on a task in
the presence of objective feedback, is the task one that the participant
would be likely to encounter outside of the experiment? An example
of a research design that was coded as high in external validity is a
study that used the participants' ratings of their performance in a so-
cial interaction that they were not informed was part of the study. An
example of a research design that was coded as low in external validity
was the judgment of contingency paradigm. In addition to addressing
these three critiques, the experimental methodology used in a particu-
lar study was coded (judgment of contingency, recall of feedback, and
evaluation of performance) to determine the potential of this variable
as a moderator of the depressive realism effect. All studies submitted
to statistical analysis in the current investigation (n=75) were coded
by three trained raters. Raters coded practice articles until their ratings
were determined to match those of a criterion coder (the first author).
Adequate inter-rater reliability was obtained for whether the study
possessed an objective standard of reality (intraclass correlation
[ICC]=.87), method of assessment (ICC=.88), methodological par-
adigm (ICC=.91), and the degree to which this method was exter-
nally valid (ICC=.87).
2.3. Statistical procedure
For studies coded as addressing Critique 1, all mean raw subjective
scores for both dysphoric/depressed and nondysphoric/nondepressed
groups were subtracted from the objective scores, which were pro-
vided in the text of the studies themselves. Therefore, a score of
zero indicates purely objective responding, while increasingly nega-
tive scores indicate negative bias, and increasingly positive scores in-
dicate positive bias. For example, a group whose mean judgment of
contingency score on the judgment of contingency task was 40,
when the experimenter-determined contingency was 75, would
have a mean difference score of −35. The scores' negative sign indi-
cates that the judgment of this event was more negative than the
event itself, while the absolute value indicates the degree of bias.
For studies that assessed the accuracy of participants' recall for posi-
tive and negative stimuli (making a score of 100% indicate perfect ac-
curacy), the mean scores for negative stimuli were subtracted from
the means for positive stimuli. This difference score was used to
make the results of all studies interpretable in the same manner, as
a score of zero would indicate evenhanded accuracy, increasingly
negative scores would indicate preference for negative stimuli (dem-
onstrating a negative bias), and increasingly positive scores would in-
dicate preference for positive stimuli (demonstrating a positive bias).
For example, a group that recalled negative stimuli correctly an average
of 50% of the time and positive stimuli an average of 20% of the time
negative stimuli. Effect size statistics (Cohen's d) were then computed by
the pooled standard deviation.
The d statistic has been critiqued for being a biased estimator of the
true population effect size in smaller samples (Hedges, 1981), therefore,
a correction factor was applied according to the suggestions of Hedges
(1981). Note that all descriptive statistics listed in the current investiga-
tion have been corrected for this sampling error. Using this corrected d
statistic,a small effect sizewould indicate that bothgroups were equally
accurate in their perceptions, while increasing positive effects indicate
relatively higher degrees of accuracy in the nondysphoric/nondepressed
group (contrary to predictions of depressive realism) and increasingly
negative effect sizes indicate relatively higher degrees of accuracy in
the dysphoric/depressed group (consistent with predictions from de-
If a single study possessed multiple, relevant dependent variables, a
weighted average was computed (composed of the effect sizes of the
dependent variables within a study). This was done to address critique
that studies that use multiple effect size statistics (data points) from a
single study violate the independence of observation assumptions
of much of inferential statistics (Glass et al., 1981; Rosenthal, 1991).
Weighted averages were computed using the random-effects proce-
dure outlined by Hedges and Vevea (1998). Random-effects analyses,
are invariant across studies (e.g., Hedges & Vevea, 1998; Schmidt, Oh, &
Hayes, 2009). As a result, findings from random-effects analyses can be
more readily generalized to participants that were not included in the
studies being analyzed. However, the trade-off of this increased exter-
nal validity is the decreased power of these statistics. Circumstances
where population parameters would be presumed to vary across
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
studies include (but are not limited to) where an unmeasured modera-
tor variable is present in the collection of studies or if measurement
error is highly variable across studies (Schmidt et al., 2009). The use
of fixed-effects analyses has been criticized on the grounds that most
meta-analyses fall into at least one of these two circumstances and are
concerned with generalization to studies not included in the meta-
analysis itself (e.g., Field, 2003; Hedges, 1994; Hedges & Vevea, 1998;
Hunter & Schmidt, 2000; Raudenbush, 1994; Rosenthal, 1991). The
not all of the studies relevant to depressive realism could be included
(making generalization to these studies a significant strength), and sig-
ies (see below). Given these three conditions, and evidence that
suggests that erroneously narrow confidence intervals and inflated
Type I Error results from the inappropriate use of fixed effects analyses
(e.g., Field, 2003; Hedges, 1994; Hedges & Vevea, 1998; Hunter &
Schmidt, 2000; Raudenbush, 1994; Rosenthal, 1991; Schmidt et al.,
2009), random effects analyses were used exclusively in the current
Calculating effect sizes by subtracting nondysphoric/nondepressed
group scores from those of the dysphoric/depressed group provided
an index of the degree of perceptual accuracy in one group relative to
the other. However, this approach does not provide information on
how each group is biased in an absolute sense, positively or negatively.
One-sample t-tests were computed from the signed difference scores
mentioned above for the dysphoric/depressed and nondysphoric/
nondepressed groups individually, which were then converted into
corrected effect size statistics (Cohen's d) using a supplemental formu-
la.4A score of zero would indicate purely objective responding, while
increasingly negative scores indicate negative bias, and increasingly
positive scores indicate positive bias.
To evaluate the presence of a moderator random-effects analyses
were again used. The Q-statistic (Hedges & Olkin, 1985) was utilized
as the random-effects statistic and is used as an indication of the de-
gree of variability in effect sizes in meta-analysis. Similar to the
F-statistic in ANOVA, Q-statistics are calculated to provide estimates
of the degree of variance within the levels of the moderator (Qw) as
well as between them (QB). To evaluate the presence of publication
bias, we utilized Rosenthal's (1979) “File Drawer” Test or Fail Safe N
(FSN). The results of this test provide an indication of the number of
studies demonstrating statistically nonsignificant results that would
have to exist in “file drawers” to reduce a particular effect to non-
significance. Whenever a mean effect size is presented below, we
have also included the value for the FSN.
A total of 121 studies were located that were relevantto the depres-
sive realism literature and made at least one comparison between dys-
phoric/depressed and nondysphoric/nondepressed groups. Of the 121
total studies, 46 studies (38% of the total) did not provide sufficient in-
formation for effect size statistics to be calculated. These 46 studies fell
into two types: (1) the authors could not be contacted to provide the
missing information (13 studies, 28% of studies with insufficient infor-
mation) or (2) the authors no longer possessed such information (33
studies, 72% of studies with insufficient information). The large number
of missing studies is an unfortunate consequence that much of the de-
pressive realism literature was conducted in response to the Alloy and
Abramson's (1979) manuscript, which was prior to the advent of
personal computers, and the ease of data storage and retrieval that
resulted from their use. For studies that did not possess sufficient infor-
mation to calculate effect size statistics (including the unpublished
sources mentioned above), multiple attempts were made to contact
any and all authors for whom contact information could be obtained.
Of the 75 studies remaining (see Table 1 for a complete list of studies
and effect sizes), 36 studies (48%) addressed Critique 1. Of these 36
studies, 15 studies utilized the judgment of contingency task (42%),
12 studies utilized the recall of feedback paradigm (33%), and 9 asked
the participants to make evaluations of their performance (25%). The
36 studies that addressed Critique 1 comprise 4108 participants (ap-
proximately 66% female) from across the US and Canada, as well as
from England, Spain, and Israel. Unfortunately, data on age and race
were provided for such a small number of studies that it precluded ex-
amination of these variables.
In any research endeavor involving inferential statistics, random
sampling is an important prerequisite in making generalizations
from the particular participants sampled to the population about
which the researcher wishes to draw conclusions. In meta-analysis,
studies themselves, rather than participants, are the unit of analysis.
Therefore, random sampling involves randomly sampling studies
from the population of all relevant research articles. The difficulty in
random sampling in meta-analysis lies in the tendency for studies
with nonsignificant effects, dissertations, unfinished conference pre-
sentations, etc., not to be published; the so-called File Drawer Effect
or publication bias. This makes sophisticated tests of publication
bias a necessity in meta-analysis. To test for publication bias, we uti-
lized the Duval and Tweedie (2000a, 2000b) trim and fill procedure.
In this method, the inverse of study variances are plotted on the y-
axis (such that increasing values indicate decreasing variances),
while the corrected study-level effect size is plotted on the x-axis
(see Figs. 1 and 2). Lack of symmetry in the plots is indicative of pub-
lication bias. For example, an asymmetrical plot due to a truncated
right tail would indicate a lack of studies with larger effects. Analysis
of the plot of all 75 studies (see Fig. 1) revealed that only 4 studies
needed to be trimmed to correct for publication bias. Fig. 1 is left-
skewed and is illustrative of a lack of studies with large variances
and large effects that are contrary to depressive realism. The trim
and fill procedure can also be used to correct for this publication
bias where it is detected by mirror-reflecting outliers, adding this
projected data, and recalculating relevant means. This is done itera-
tively, beginning with the largest outliers, until the corrected plot
does not indicate publication bias. Correcting for these 4 studies chan-
ged the mean effect size from −.09 to −.07. The trim and fill procedure
indicated that no studies are needed to be trimmed to correct for publi-
cation bias in the subset of data addressing Critique 1, the data upon
which almost all of our analyses were conducted (see Fig. 2).
Hypothesis 1 predicted that dysphoric/depressed participants
would illustrate a smaller degree of bias than nondysphoric/nonde-
pressed participants. Consistent with expectations from the depres-
sive realism hypothesis, dysphoric/depressed individuals illustrated
a smaller degree of bias than nondysphoric/nondepressed individuals
(weighted mean d=−.07, SD=.46, FSN=4283). However this find-
ing was below Cohen's (1992) convention for a small effect. In addi-
tion, the large standard deviation suggests that this mean result
might not adequately characterize a substantial portion of the total
literature. The results obtained using effect sizes calculated by sub-
tracting the nondysphoric/nondepressed group mean from the dys-
phoric/depressed group mean, while useful, can only speak to the
amount of bias that dysphoric/depressed individuals possess relative
to nondysphoric/nondepressed individuals and does not address if the
perceptions of either group are biased in an absolute sense. Hypothesis
2 addressed this point and posited that while dysphoric/depressed
4We are thankful to Larry Hedges for the following, helpful information (L. Hedges,
personal communication, August 30, 2007). To obtain QB, it is necessary to calculate a
constant c, which is defined as 1–(3/(4 m−1)), where m = the degrees of freedom in
the standard deviation. In the case of computing c with only one sample, m=n−1
and, therefore, c=1−(3/(4n−5)). Similarly, v also needs to be similarly adjusted
for use with one sample. The adjusted formula is, as follows: v=(1/n)+(d2/2n).
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
Effect sizes, sample weights, and coded variables of depressive realism studies.
Abramson, Alloy, & Rosoff (1981)
Alloy & Abramson (1979)
Blanco, Matute, & Vadillo (2009)
Bryson, Doan, & Pasquali (1984)
Carson, Hollon, & Shelton (2010)
Dobson & Pusch (1995)
Ford & Neale (1985)
Kapci & Cramer (1999)
Martin, Abramson, & Alloy (1984)
Mikulineer, Gerber, & Weisenberg (1990)
Msetfi, Murphy, & Simpson (2007)
Msetfi, Murphy, & Simpson, & Kombrot (2005)
Presson & Benassi (2003)
Craighead, Hickey, & DeMonbreun (1979)
Derry & Kuiper (1981)
Dobson & Shaw (1981)
Dykman, Abramson, & Albright (1991)
Johnson, Petzel, Hartney, & Morgan (1983)
Nelson & Craighead (1977)
Puseh, Dobson, Ardo, & Murphy (1998)
Roth & Rehm (1998)
Bruce & Arnett (2004)
Bynum & Scogin (1996)
Johnson & DiLorenzo (1998)
Stone, Dodrill, & Johnson (2001)
Strack & Coyne (1983)
Ahrens, Zeiss, & Kanfer (1998)
Cane & Gotlib (1985)
Crocker, Alloy, & Tabachnik-Kayne (1988)
DeMonbreun & Craighead (1977)
Dunning & Story (1991)
Dykman, Abramson, Alloy, & Hartlage (1989)
Dykman, Horowitz, Abramson, Usher (1991)
Finkel, Glass, & Merluzzi (1982)
Garber & Hollon (1980)
Glass, McKnight, & Valdimarsdottir (1993)
Gotlib & Meltzer (1987)
Hammen & Krantz (1976)
Hancock, Moffoot, & O’Carroll (1996)
Kapci & Cramer (1998)
Krantz & Gallagher-Thompson (1990)
Loeb, Beek, & Diggory (1971)
Loewenstein & Hokanson (1986)
Abramson, Garber, Edwards, & Seligman (1978)
Dunn, Dalgleish, Lawrence, & Ogilvie (2007)
Harkness, Sabbagh, Jacobson, Chowdrey, &
External validity*Avg. dw n (% female)da
Margo, Greenberg, Fisher, & Dewan (1993)
McKendree-Smith & Scogin (2000)
McNamara & Hackett (1986)
Miller & Seligman (1973)
Miller & Seligman (1976)
Miller, Seligman & Kurlander (1975)
Pacini, Muir, & Epstein (1998)
Pyszezynski, Holt & Greenberg (1987)
Rosenfarb, Burker, Morris, & Cush (1993)
Sacco & Hokanson (1978)
Sacco & Hokanson (1982)
Stone & Glass (1986)
Strunk & Adler (2009)
Strunk, Lopez, & DeRubeis (2006)
Vestre & Caulfield (1986)
Whitton, Larson, & Hauser (2008)
Note: Average weighted effect across all studies = -.10; Statistical analyses were only conducted on this variable for studies which addressed Critique 1; This variable was only coded for studies which were evaluated for the presence of moderator variables (i.e, addressed Critique 1); Avg. d = Average effect size ( Cohen’s d) comparing perceptual accuracy of depressed/dysphoric
and nondepressed/nondysphoric participants; w = Inverse variance weight; JOC = Judgment of Contingency; ROF = Recall of Feedback; EOP = Evaluation of Performance; EXP = Expectancies of Success; SOC = Social Comparison; OTR = Other; dd = Average d for depressed participants; ded = Average d for nondepressed participants; * = Outlier not included in analysis; NA = Gender
information not available; Small effect = .20, Medium effect = .50, Large effect = .80.
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
participants would not evidence any significant bias, nondysphoric/
nondepressed participants would demonstrate a bias for positively-
valenced stimuli. Analyses that examined each group individually indi-
cated that individuals in the dysphoric/depressed group tended to be
biased optimistically (weighted mean d=.14, SD=2.42, FSN=8347),
however, this result was less than a small effect. Nondysphoric/nonde-
pressed individuals also illustrated an optimistic bias, although to a
greater extent (weighted mean d=.29, SD=2.53, FSN=4777), and
exceeded the convention for a small effect. Given these findings, Hy-
pothesis 2 can be said to be partially supported. The findings that non-
dysphoric/nondepressed individuals evidence a larger degree of
absolute bias and are biased positively, are consistent with both Hy-
pothesis 2 and the expectations of depressive realism. However, that
both groups demonstrate a positive bias is not consistent with it. In ad-
dition, the large variability present here both requires caution in over-
interpreting the above results and suggests the presence of moderator
variables, which are discussed below.
Hypothesis 3 stated that studies that utilized an objective standard of
reality, and thereby adequately addressed Critique 1, would produce
larger effect sizes than studies that did not. This variable did serve as a
significant moderator of the depressive realism effect (QB[df=1]=
6.87, p=.0088, k [number of studies]=75, total n=7305). Examination
of average effects for both studies adequately addressing Critique 1
(weighted mean d=−.03, SD=.41, FSN=868) as well as those studies
that did not (weighted mean d=−.15, SD=.51, FSN=1245) indicated
that both types of studies found depressive realism effects. However,
counter to expectations, this effect was much stronger in studies lower
in methodological quality.
Hypothesis 4 stated that studies that utilize structured clinical inter-
view would produce larger effects than studies that utilize self-report.
Method of assessment influenced whether depressive realism effects
were found (QB[df=1]=7.57, p=.0059, k=36, total n=4108). Con-
trary to prediction, studies utilizing self-report were more likely to find
depressiverealism effects (weighted
ed mean d=.16, SD=.48, FSN=10).
Finally, Hypothesis 5 stated that studies that more readily general-
ized outside of the laboratory would be less likely to produce depres-
sive realism effects. Results indicated that the external validity of the
study did serve as a significant moderator of the depressive realism
effect (QB[df=1]=32.80, pb.0001, k=36, total n=4108). Contrary
nal validity was almost identical to studies high on external validity
(weighted mean d=−.03, SD=.38, FSN=76, and weighted mean
d=−.02, SD=.48, FSN=357, respectively).
Exploratory analyses were conducted using methodological para-
digm (judgment of contingency, recall of feedback, and evaluation
of performance) as a moderator of the depressive realism effect as
no prior research has been done on this topic and there was little the-
ory available to guide the formation of specific hypotheses. Methodol-
ogy type was found to be a significant moderator (QB[df=2]=19.10,
p=.00007, k=36, total n=4108) and the results were, therefore,
decomposed further via examination of the size of effects associated
with the four major methodological paradigms used in the depressive
realism literature. Both relative bias (effect sizes calculated using dys-
phoric/depressed and nondysphoric/nondepressed groups) and abso-
lute bias (one-sample t-tests converted to effect size statistics) were
examined. Surprisingly, results from studies using the judgment of
contingency task only demonstrated a small overall depressive real-
ism effect (weighted mean d=−.09, SD=.37, FSN=96), despite
the fact that this was the paradigm in which the depressive realism
effect was first demonstrated (Alloy & Abramson, 1979). Examination
of depressed and nondepressed participants separately, to determine
the degree of absolute bias, indicated that both depressed and nonde-
pressed participants overestimated the degree of contingency to the
same extent (weighted mean d=.53, SD=2.26, FSN=421 and .60,
SD=2.32, FSN=92, respectively). Both of these results exceed the
convention for a large effect. The depressive realism effect was exam-
ined using this paradigm withvarying degrees of objective contingency
between pressing the button and the onset of the light (−50% to 100%;
negative contingencies represent button pressing resulting in the sup-
pression of illumination). Exploratory analyses were conducted evalu-
ating degree of contingency as a potential moderator of the depressive
realism effect. Unfortunately, the small number of judgment of contin-
gency studies using each of these degrees of contingency individually
necessitated aggregation into groups. Therefore, contingency was sepa-
rated into Low (−50–49%) and High (50%–100%) groups. This division
High=9 studies).5Interestingly, studies using a low pre-determined
contingency produced results that are more consistent with depressive
realism (weighted mean d=−.20, SD=.71, FSN=84) than studies
using a high contingency (weighted mean d=.03, SD=.42, FSN=7).
5The total number of studies using low and high degrees of contingency (21) is
greater than the total number of judgment of contingency studies (14) because many
studies evaluated multiple degrees of contingency. However, in cases where one study
had multiple effect sizes that fit into either category, these effect sizes were averaged
so that studies did not contribute multiple data points to each group.
Effects size (d)
Fig. 1. Funnel plot of all effect sizes as a function of inverse-variance weights (w). N=75.
Effect size (d)
Fig. 2. Funnel plot of effect sizes for studies which address Critique 1 as a function of
inverse-variance weights (w). N=36.
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
This difference, while seemingly small, was significant as degree of con-
ies utilizing the judgment of contingency task (QB [df=1]=16.91,
p=.00004, k=23, total n=1588).
Results from the evaluation of performance studies indicated
slightly more bias among depressed/dysphoric participants (weight-
ed mean d=.14, SD=.50, FSN=79). Recall of feedback studies
were more equivocal in their results (weighted mean d=−.03,
SD=.40, FSN=39). However, the results from both of these method-
ological paradigms corresponded to less than small effects. Depressed
participants in the evaluation of performance paradigm were rela-
tively evenhanded and evidenced only a small negative bias (weight-
ed mean d=−.06, SD=1.07, FSN=252), while the nondepressed
subjects possessed a small, but positive bias (weighted mean
d=.14, SD=1.40, FSN=258). Similar results were obtained in the
recall of feedback studies where depressed participants were relative-
ly evenhanded and evidenced only a small negative bias (weighted
mean d=−.10, SD=3.12, FSN=337), whereas nondepressed partic-
ipants possessed a small, but positive bias (weighted mean d=.14,
The current investigation serves as the first attempt to quantitatively
summarize and investigate the depressive realism literature. Although
the results averaging across all studies addressing Critique 1 were gener-
effect was small. However, the large degree of variability in the size of
the effects obtained by the various studies in the depressive realism lit-
erature (QTotal=493.89 [df=74], pb.001, SD=.72, range: −3.59–.99)
resulted in the small depressive realism effect obtained when studies
were averaged. Dysphoric/depressed participants were found to
be relatively evenhanded in their perceptions, while nondysphoric/
nondepressed participants evidenced a more substantial positive bias.
Substantial variability was also found among these groups, suggesting
caution in interpreting these results, as well as the presence of moder-
ator variables, discussed below.
The manner in which Hypothesis 2 was evaluated in the current
investigation is worthy of comment. Bias was investigated both in
dysphoric/depressed and nondysphoric/nondepressed groups relative
to one another, as well as compared to an absolute standard. Much of
the literature on depressive realism has not differentiated between
these two methods of evaluating the theory. Past research (e.g.,
Dobson & Franche, 1989; Dunn et al., 2007) has noted the importance
of assessing both perceptions in one group relative to another (relative
bias) andcomparingonegroup'sperceptionstoanabsolute standard to
reality (absolute bias). We echo their suggestion that future investiga-
tions of depressive realism specify the type of bias, relative, absolute,
or somewhere between the two, that is being predicted. The type of
bias assessed has important implications for the theory that is being in-
vestigated. We argue that two versionsof depressive realismhave been
implicitly studied. We propose that the version of depressive realism
which posits only relative bias be referred to as weak depressive real-
ism. Weak depressive realism posits merely that depressed/dysphoric
participants demonstrate less bias than nondepressed/nondysphoric
participants. We propose that the version of depressive realism that
makes more restrictive claims and posits both relative and absolute
bias be referred to as strong depressive realism. This version of the de-
pressive realism hypothesis posits both that depressed/dysphoric par-
ticipants demonstrate a lack of significant positive or negative bias
and demonstrate less bias than nondepressed/nondysphoric partici-
pants. Additionally, an intermediate version of depressive realism
quire that depressed/dysphoric participants demonstrate less bias than
nondepressed/nondysphoric participants and also that the bias in both
groups be positive. The current investigation is an example of where
making this differentiation is significant and not doing so could lead
to confusion. Our results are partially supportive of strong depressive
realism (which is the view we described in our hypotheses above),
but fully supportive of an intermediate or weak version of depressive
realism. However, this degree of support for depressive realism should
be interpreted in light of the results of our moderation analyses.
An attempt was made to model the extent of the variability in the re-
identified moderators of the depressive realism effect. These analyses in-
absence of an objective standard of reality and when self-report (as op-
posed to clinical interview) was used to assess level of dysphoria. Analy-
ses were also conducted that suggested that depressive realism effects
were more equivocal in studies utilizing the judgment of contingency,
evaluation of performance, and recall of feedback paradigms. This result,
the lack of a strong depressive realism effect in the judgment of contin-
gency paradigm, where depressive realism has been more frequently
evaluated, was particularly surprising. Additional analyses were con-
ducted attempting to model the heterogeneity in studies utilizing the
judgmentof contingency paradigm whereit was discovered thatdepres-
interest and is relevant to conjectures researchers have made since the
beginning of research into depressive realism (Alloy & Abramson, 1979;
ity of depressed participants to accurately judge zero contingency condi-
tions is the result of these conditions matching their preconceptions
about their relationship to the world (Alloy & Abramson, 1979; Langer,
1975; Msetfi et al., 2005). That is to say, depressed individuals do not be-
lieve that their actions have any influence on events and it is coincidence
that accuracy in a zero contingency condition corresponds to this view.
However, our data are too limited to support the assertion that depres-
tingency task are merely an artifact. For example, due to the small
number of judgment of contingency studies overall (n=15), High and
Lowcontingency groups had tobe created usinga mediansplit. This arti-
ficial dichotomization may have resulted in the small differences ob-
served between the two groups by creating groups that were not
homogenous with regard to their performance on the judgment of con-
tingency task. Additional research will need to be conducted to experi-
mentally determine if this is the case. In particular, research is needed
examining judgments of contingency using levels of contingency higher
than zero. Of the 15 articles found which examined judgments of contin-
gency, 13 of these used a zero contingency condition, and 5 did so
Unfortunately, varying the nature of the “reality” that participants
are asked to report upon has only been attempting using the judg-
ment of contingency task. It is possible, for example, that depressive
realism effects may be constrained to recall of feedback studies
where negative feedback was given. However, the valence of feed-
back has never, to our knowledge, been systematically varied to de-
termine its influence on whether depressive realism effects were
obtained. Future research in depressive realism should focus on ex-
amining to what extent the match between participant schemata
and “reality” may underlie the depressive realism effect, which
would involve systematic variation of important aspects of “reality.”
Although the aggregated results of the current investigation are
certainly suggestive that the depressive realism effect may be con-
strained to a very particular set of circumstances, they should not
be interpreted as suggesting that the depressive realism effect is not
a valid phenomenon. There were only a limited number of studies uti-
lizing clinical interview (n=4), externally valid stimuli (n=11), and
particular research paradigms (average n=12, range 9–14).The small
number of studies using externally valid research designs is an unfor-
tunate consequence on the popularity of the JOCT among depressive
realism researchers and the relative importance placed on internal
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
validity in the early investigations of a topic. The results of the current
investigation suggest that future investigations of depressive realism
should utilize structured clinical interview and externally valid designs.
Future research should also attempt to collect data from participants
displayingarangeofseverity insymptomsof depression.As mentioned
previously, the potential for a curvilinear relationship between depres-
sion and perceptual accuracy has only been adequately evaluated once
before (and so could not be examined in the current study). The ques-
fering from a mood disorder, or outside of the laboratory at all, loom
large. In addition, there was a statistically significant degree of unac-
counted for variance within each of the levels of our moderator vari-
ables (i.e., QW). Future research will hopefully detect variables that
can successfully model and account for this variability. However, for
this increased accuracy in statistical prediction to occur, not only will
more researches have to be done, but a consistent set of stimuli will
have to be developed. Even within the judgment of contingency litera-
ture, which is centered on a particular task, the judgment of contingency
several forms of a computerized task. This lack of consistency in stimuli
no doubt partially accounts for the vast heterogeneity in the results of
judgment of contingency studies and other depressive realism studies.
It is difficult to attempt to theoretically model unaccounted for variance
when the variables making up that variance are not consistent from
study to study.
Anotherfactor thathasbeen under-investigatedinthedepressive
realism literature is the role of comorbid anxiety. Numerous studies
have demonstrated rates of comorbidity that are alarmingly high
(e.g., Brawman-Mintzer at al., 1993; Brown et al., 2001; Kessler et al.,
2005). Given this degree of overlap, it is possible that the depressive re-
alism phenomena might not be specific to depression or, at worst, be
ly evaluated depressive realism in the context of the tripartite model of
mood and anxiety disorders (Clark & Watson, 1991; Clark, Watson, &
Mineka, 1994). The tripartite model proposes that mood and anxiety dis-
orders are best represented by symptom dimensions that are particu-
lar to each (i.e., low positive affect/anhedonia and anxious arousal,
respectively) and a non-specific ‘general distress’ dimension. Dunn et
atively related to symptoms of anhedonia and unrelated to the anxious
arousal dimension. Future research should attempt to replicate this
work and extend it by supplementing the assessment of depression
and anxiety with structured interview. In addition, future research
should examine the relationship between comorbid anxiety and depres-
sive realism using many of the various research paradigms mentioned
The results of the current investigation also have clinically-
relevant theoretical implications. These results can be used to explain
the dichotomy between researchers finding statistically significant
depressive realism effects while practitioners fail to notice such ef-
fects in their clients. It is possible that the depressive realism effect
is not present under conditions normally encountered in therapy, as
a result of the effect of some moderator variable(s). This hypothesis
is supported by the fact that almost all of the levels of the various
moderator variables examined above possessed a significant degree
of variability. Potential moderator variables that have not been ade-
quately investigated are whether participant responses refer to self
versus other, responses made in public versus private settings, the
level of ambiguity of the stimuli used, severity of depression, and
al from memory). With regard to biases in information processing, it is
individuals relative to those suffering from depression. However, this
bias may not only disappear, but reverse itself, in processes that occur
later in information processing (during memory encoding or retrieval).
Attentional bias research has found that nondepressed individuals
reliably evidence either a bias toward stimuli likely to result in a positive
mood or away from stimuli likely to result in a negative mood, and that
depressed individuals show no such bias (Gotlib, McLachlan, & Katz,
1988; McCabe & Gotlib, 1995; McCabe & Toman, 2000; McCabe, Gotlib,
(Goddard, Dritschel, & Burton, 1996; Kuyken & Brewin, 1995; Kuyken &
Dalgleish, 1995; Puffet, Jehin-Marchot, Timsit-Berthier, & Timsit, 1991),
as well as memory recall research in general (Bradley & Mathews,
1983; Derry & Kuiper, 1981; Gilboa & Gotlib, 1997; MacLeod et al.,
1986), finds that depressed individuals show a preference for
negatively-valenced, self-referent information and that this bias reliably
predicts the occurrence of depressive symptoms, introducing the possi-
before or during memory recall. While biased memory recall may be
more salient or noticeable therapeutically than attentional biases, it
would not be surprising that clinicians do not make note of the de-
pressive realism effect. Evidence that nondysphoric/nondepressed in-
dividuals evidence generally positively-biased perceptions relative to
dysphoric/depressed individuals also lends support to the notion that
perceptual bias may serve to protect an individual from the occurrence
of depression (Alloy & Clements, 1992). It is also possible that the po-
tential positivity bias present in the nondysphoric/nondepressed ac-
counts for why depressive realism effects may not be noticed in
therapy. To the extent that therapists are euthymic, they may tend to
pathologize their dysphoric/depressed clients, whose outlook is so
much more negative than their own.2
Research demonstrating the potentially protective function of posi-
tively biased perceptions also highlights another under-investigated
area in depressive realism: the function of perceptual bias. While the
study by Alloy and Clements (1992) frames the question in a dichoto-
mous fashion, bias is either good or bad, it is possible that the value of
a positive/negative perceptual bias may depend on the situation. If
demonstrated, this possibility suggests that successful therapy could
consist of either alteration of trait-like cognitive structures (as in tradi-
tion cognitive therapy) or a change in context.
The current investigation, and the depressive realism literature as
a whole, is relevant to the question of how cognitive therapy results
in reductions in depression. The theory behind cognitive therapy
(Beck et al., 1979) suggests that the cognitions of depressed individuals
are negatively biased and making these thoughts more realistic is the
process by which the therapy exerts its effect. Depressive realism sug-
gests that depressed individuals are more accurate in their perceptions
and, by extension, calls into question how cognitive therapy works.
While it would be tempting to frame the results of the current study
as supporting one theory versus the other, it would be more accurate
to see it as a beginning in defining the boundaries of the depressive re-
alism phenomena. Future research will be needed to elaborate on the
relevance of depressive realism for etiological models of depression
(e.g., Moore & Fresco, 2007) and for therapy. Specifically, past research
has utilized process measures which do not attempt to evaluate the ac-
curacy of client thoughts (e.g., DeRubeis et al., 1990; Dimidjian et al.,
2006; Jacobson et al., 1996). Future work could utilize methodologies
relevant to both depressive realism and mediators of cognitive therapy,
such as the one used in Moore and Fresco (2007), to evaluate if cogni-
tive therapy works by making thoughts more realistic or just making
them more positive.
Although the current investigation provides the first foray into
quantitative review of a theoretically rich literature, some limitations
of the current design warrant mention and effect the conclusions that
can be drawn from it. First, slightly fewer than 40% of the total studies
relevant to the depressive realism literature could not be obtained
and submitted to analysis. A large amount of missing data is inevita-
ble given the age of the depressive realism literature. Our analyses in-
dicated that these missing data were randomly distributed and were
not likely to significantly influence the results of our subsequent anal-
yses. Nonetheless, it is possible that different results would be
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
obtained had this missing data been available. One important variable
race, which precluded analyses of the validity of depressive realism in
different racial/ethnic groups.
Despite the aforementioned limitations, the current investigation
provides insight into deficits in the depressive realism literature as a
whole, as well as potentially fruitful areas in need of further investi-
gation. Future depressive realism studies would benefit from more at-
tention given to generalizing depressive realism effects outside of the
laboratory and identifying depressed and nondepressed individuals
using structured clinical interview. In addition, more research should
be conducted using treatment-seeking samples. While sample char-
acteristics were not formally under study in the current investigation,
it should be noted that only 11 of the 36 studies in question utilized a
treatment-seeking sample. This vastly limited the conclusions that
could be drawn about the depressive realism phenomenon in this
population. Finally, more research in depressive realism in general,
and work in developing widely-accepted stimuli in particular, may
help to model the large degree of variability in the results of depressive
realism studies that were observed. The influence of other variables
which may significantly alter the depressive realism effect, such as the
degree of contingency in judgment of contingency studies and type of
information processing tapped, should also be explored. If nothing
else, the current investigation highlights that the depressive realism ef-
fect is far from universal. The questionfor future research then becomes:
sive realism effect valid?
*Abramson, L. Y., Alloy, L. B., & Rosoff, R. (1981). Depression and the generation of com-
plex hypotheses in the judgment of contingency. Behaviour Research and Therapy,
*Abramson, L. Y., Garber, J., Edwards, N. B., & Seligman, M. E. P. (1978). Expectancy
changes in depression and schizophrenia. Journal of Abnormal Psychology, 87,
Abramson, L. Y., Metalsky, G. I., & Alloy, L. B. (1989). Hopelessness depression: A
theory-based subtype of depression. Psychological Review, 96, 358–372.
Ackermann, R., & DeRubeis, R. J. (1991). Is depressive realism real? Clinical Psychology
Review, 11, 565–584.
Ahrens, A. H. (1986). Choice of social comparison targets by depressed and nonde-
pressed students. Unpublished doctoral dissertation, Stanford University, Stanford,
*Ahrens, A. H. (1991). Dysphoria and social comparison: Combining information re-
garding others' performances. Journal of Social and Clinical Psychology, 10, 190–205.
*Ahrens, A. H., Zeiss, A. M., & Kanfer, R. (1988). Depressive deficits in interpersonal
standards, self-efficacy, and social comparison. Cognitive Therapy and Research,
*Alloy, L. B., & Abramson, L. Y. (1979). Judgment of contingency in depressed and non-
depressed students: Sadder but wiser? Journal of Experimental Psychology: General,
Alloy, L. B., & Abramson, L. Y. (1980). The cognitive component of human helplessness
and depression: A critical analysis. In J. Garber, & M. E. P. Seligman (Eds.), Human
helplessness: Theory and application (pp. 59–70). New York: Academic Press.
Alloy, L. B., & Abramson, L. Y. (1988). Depressive realism: Four theoretical perspectives.
In L. B. Alloy (Ed.), Cognitive processes in depression (pp. 223–265). New York:
Alloy, L. B., Abramson, L. Y., & Kossman, D. (1985). The judgment of predictability in de-
pressed and nondepressed college students. In F. R. Brush, & J. B. Overmier (Eds.),
Affect, conditioning and cognition: Essays on the determinants of behavior
(pp. 229–246). Hillsdale, NJ: Erlbaum.
Alloy, L. B., Abramson, L. Y., & Viscusi, D. (1981). Induced mood and the illusion of control.
Journal of Personality and Social Psychology, 41, 1129–1140.
Alloy, L. B., & Ahrens, A. H. (1987). Depression and pessimism for the future: Biased use
of statistically relevant information in predictions for self versus others. Journal of
Personality and Social Psychology, 52, 366–378.
Alloy, L. B., & Clements, C. M. (1992). Illusion of control: Invulnerability to negative affect
and depressive symptoms after laboratory and natural stressors. Journal of Abnormal
Psychology, 101, 234–245.
Alloy, L. B., & Seligman, M. E. P. (1979). In G. H. Bower (Ed.), On the cognitive component
of learned helplessness and depression. The psychology of learning and motivation,
Vol. 13. (pp. 219–276) New York: Academic Press.
*Andersen, S. M. (1990). The inevitability of future suffering: The role of depressive
predictive certainty in depression. Social Cognition, 8, 203–228.
Bangert-Drowns, R. L. (1986). Review of developments in meta-analytic methods.
Psychological Bulletin, 99, 388–399.
Beck, A. T. (1967). Depression: Clinical, experimental, and theoretical aspects. New York:
Harper & Row.
Beck, A. T. (1976). Cognitive therapy and emotional disorders. New York: International
Beck, A. T. (1986). Cognitive therapy, behavior therapy, psychoanalysis, and pharmaco-
therapy: The cognitive continuum. In J. B. W. William, & R. L. Spitzer (Eds.), Psycho-
therapy research: Where are we and where should we go? (pp. 114–134). New York:
Beck, A. T. (1987). Cognitive models of depression. Journal of Cognitive Psychotherapy:
An International Quarterly, 1, 5–37.
Beck, A. T., Rush, A. J., Shaw, B. F., & Emery, G. (1979). Cognitive therapy of depression.
New York: Guilford Press.
Benassi, V. A., & Mahler, H. I. M. (1985). Contingency judgments by depressed college
students: Sadder but not always wiser. Journal of Personality and Social Psychology,
*Beyer, S. (2002). The effects of gender, dysphoria, and performance feedback on the
accuracy of self-evaluations. Sex Roles, 47, 453–464.
Blackburn, I., & Moorhead, S. (2001). Depression. In W. L. Lyddon, & J. V. Jones (Eds.),
Empirically supported cognitive therapies: Current and future applications. New
*Blanco, F., Matute, H., & Vadillo, M. A. (2009). Depressive realism: Wiser or quieter?
Psychological Record, 59, 551–562.
Bradley, B., & Mathews, A. (1983). Negative self-schemata in clinical depression. British
Journal of Clinical Psychology, 22, 173–181.
Brawman-Mintzer, O., Lydiard, R. B., Emmanuel, N., Payeur, R., Johnson, M., Roberts, J.,
Jarrell, M. P., & Ballenger, J. C. (1993). Psychiatric commorbidity in patients with
generalized anxiety disorder. The American Journal of Psychiatry, 150, 1216–1218.
Brown, T. A., Campbell, L. A., Lehman, C. L., Grisham, J. R., & Mancill, R. B. (2001). Cur-
rent and lifetime comorbidity of the DSM-IV anxiety and mood disorders in a
large clinical sample. Journal of Abnormal Psychology, 110, 585–599.
*Bruce, J. M., & Arnett, P. A. (2004). Self-reported everyday memory and depression in
patients with multiple sclerosis. Journal of Clinical and Experimental Neuropsychology,
*Bryson,S.E., Doan, B.D., & Pasquali,P.(1984).Sadderbut wiser: A failure todemonstrate
that mood influences judgment of control. Canadian Journal of Behavioral Science, 16,
Buchwald, A. M. (1977). Depressive mood and estimates of reinforcement frequency.
Journal of Abnormal Psychology, 86, 443–446.
*Bynum, J., & Scogin, F. (1996). The impact of dysfunctional attitudes on depressive re-
alism. Journal of Social and Clinical Psychology, 15, 305–317.
*Cane, D. B., & Gotlib, I. H. (1985). Depression and the effects of positive and negative
feedback on expectations, evaluations, and performance. Cognitive Therapy and
Research, 9, 145–160.
*Carson, R. C., Hollon, S. D., & Shelton, R. C. (2010). Depressive realism and clinical de-
pression. Behaviour Research and Therapy, 48, 257–265.
Clark, L. A., & Watson, D. (1991). Theoretical and empirical issues in differentiating de-
pression from anxiety. In A. Becker, & J. Kleinman (Eds.), Psychosocial Aspects of De-
pression (pp. 39–65). Hillsdale, NJ: Erlbaum.
Clark, L. A., Watson, D., & Mineka, S. (1994). Temperament, personality, and the mood
and anxiety disorders. Journal of Abnormal Psychology, 103, 103–116.
*Cobbs,D.L.(1990). Judgmentofcontingencyandthe cognitive functioningofclinical de-
pressives. Unpublished doctoral dissertation, University of North Texas, Denton, TX.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
*Craighead, W. E., Hickey, K. S., & DeMonbreun, B. G. (1979). Distortion of perception
and recall of neutral feedback in depression. Cognitive Therapy and Research, 3,
*Crocker, J., Alloy, L. B., & Tabachnik-Kayne, N. T. (1988). Attributional style, depression
and perceptions of consensus for events. Journal of Personality and Social Psycholo-
gy, 54, 840–846.
*DeMonbreun, B. G., & Craighead, W. E. (1977). Distortion of perception and recall of
Dennard, D. O., & Hokanson, J. E. (1986). Performance on two cognitive tasks by dys-
phoric and nondysphoric students. Cognitive Therapy and Research, 10, 377–386.
*Derry, P. A., & Kuiper, N. A. (1981). Schematic processing and self-reference in clinical
depression. Journal of Abnormal Psychology, 90, 286–297.
DeRubeis, R. J., & Crits-Cristoph, P. (1998). Empirically supported individual and group
psychological treatments for adult mental disorders. Journal of Consulting and Clin-
ical Psychology, 66, 37–52.
DeRubeis, R. J., Evans, M. D., Hollon, S. D., Garvey, M. J., Grove, W. M., & Tuason, V. B.
(1990). How does cognitive therapy work? Cognitive change and symptom change
in cognitive therapy and pharmacotherapy for depression. Journal of Consulting and
Clinical Psychology, 58, 862–869.
Dimidjian, S., Hollon, S. D., Dobson, K. S., Schmaling, K. B., Kohlenberg, R. J., Addis, M. E.,
Gallop, R., McGlinchey, J. B., Markley, D. K., Gollan, J. K., Atkins, D. C., Dunner, D. L., &
Jacobson, N. S. (2006). Randomized trial of behavioral activation, cognitive thera-
py, and antidepressant medication in the treatment of adults with major depres-
sion. Journal of Consulting and Clinical Psychology, 74, 658–670.
*Dobson, K. S., & Shaw, B. F. (1981). The effect of self-correction on cognitive distor-
tions in depression. Cognitive Therapy and Research, 5, 391–403.
Dobson, K. S., & Franche, R. L. (1989). A conceptual and empirical review of
the depressive realism hypothesis. Canadian Journal of Behavioral Science, 21,
6References marked by an asterisk indicate studies included in the meta-analysis.
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
*Dobson, K. S., & Pusch, D. (1995). A test of the depressive realism hypothesis in clinically
depressed subjects. Cognitive Therapy and Research, 19, 179–194.
*Dunn, B. D., Dalgleish, T., Lawrence, A. D., & Ogilvie, A. D. (2007). The accuracy of self-
monitoring and its relationship to self-focused attention in dysphoria and clinical
depression. Journal of Abnormal Psychology, 116, 1–15.
*Dunn, B. D., Stefanovitch, I., Buchan, K., Lawrence, A. D., & Dalgleish, T. (2009). A re-
duction in positive self-judgment bias is uniquely related to the anhedonic symp-
toms of depression. Behaviour Research and Therapy, 47, 374–381.
*Dunning, D., & Story, A. L. (1991). Depression, realism, and the overconfidence effect: Are
the sadder wiser when predicting future actions and events? Journal of Personality and
Social Psychology, 61, 521–532.
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-plot-based method of
testing and adjusting for publication bias in meta-analysis. Biometrics, 56, 455–463.
Duval, S., & Tweedie, R. (2000). A nonparametric “trim and fill” method of accounting
for publication bias in meta-analysis. Journal of the American Statistical Association,
*Dykman, B.M., Abramson, L.Y., & Albright,J. S.(1991). Effects of ascending and descending
patterns of success upon dysphoric and nondysphoric subjects' encoding, recall, and
predictions of future success. Cognitive Therapy and Research, 15, 179–199.
*Dykman, B. M., Abramson, L. Y., Alloy, L. B., & Hartlage, S. (1989). Processing of ambigu-
ous and unambiguous feedback by depressed and nondepressed college students:
Schematic biases and their implications for depressive realism. Journal of Personality
and Social Psychology, 56, 431–445.
*Dykman, B. M., Horowitz, L. M., Abramson, L. Y., & Usher, M. (1991). Schematic and
situational determinants of depressed and nondepressed students' interpreta-
tion of feedback. Journal of Abnormal Psychology, 100, 45–55.
Evans, M. D., & Hollon, S. D. (1988). Patterns of personal and causal inference: Implications
for the cognitive therapy of depression. In J. Garber, & M. E. P. Seligman (Eds.), Human
helplessness: Theory and application (pp. 344–377). New York: Academic Press.
Field, A. P. (2003). The problem in using fixed-effects models of meta-analysis on real
world data. Understanding Statistics, 2, 77–96.
*Finkel, C. B., Glass, C. R., & Merluzzi, T. V. (1982). Differential discrimination of self-
referent statements by depressives and nondepressives. Cognitive Therapy and
Research, 6, 173–183.
Personality and Social Psychology, 49, 1330–1336.
*Garber, J., & Hollon, S. D. (1980). Universal versus personal helplessness in depression:
Gilboa, E., & Gotlib, I. H. (1997). Cognitive biases and affect persistence in previously
dysphoric and never-dysphoric individuals. Cognition and Emotion, 11, 517–538.
Glass, G. V., McGaw, B., & Smith, M. L. (1981). Meta-analysis in social research. Beverly
Hills, CA: Sage Publications.
*Glass, D. C., McKnight, J. D., & Valdimarsdottir, H. (1993). Depression, burnout, and
perceptions of control in hospital nurses. Journal of Consulting and Clinical Psychology,
Goddard, L., Dritschel, B., & Burton, A. (1996). Role of autobiographical memory in so-
cial problem solving and depression. Journal of Abnormal Psychology, 105, 609–616.
Golin, S., Terrel, F., Weitz, J., & Drost, P. L. (1979). The illusion of control among de-
pressed patients. Journal of Abnormal Psychology, 88, 454–457.
Golin, S., Terrell, F., & Johnson, B. (1977). Depression and the illusion of control. Journal
of Abnormal Psychology, 86, 440–442.
*Gotlib, I. H. (1981). Self-reinforcement and recall: Differential deficits in depressed and
nondepressed psychiatric patients. Journal of Abnormal Psychology, 90, 521–530.
*Gotlib, I. H. (1982). Self-reinforcement and depression in interpersonal interaction:
The role of performance level. Journal of Abnormal Psychology, 91, 3–13.
*Gotlib, I. H. (1983). Perception and recall of interpersonal feedback: Negative bias in
depression. Cognitive Therapy and Research, 7, 399–412.
Gotlib, I. H., McLachlan, A. L., & Katz, A. N. (1988). Bias in visual attention in depressed
and nondepressed individuals. Cognition and Emotion, 2, 185–200.
*Gotlib, I. H., & Meltzer, S. J. (1987). Depression and the perception of social skill in dyadic
interaction. Cognitive Therapy and Research, 11, 41–54.
of depression in 1990. The Journal of Clinical Psychiatry, 54, 405–426.
Haaga, D. A. F., & Beck, A. T. (1995). Perspectives on depressive realism: Implications
for cognitive theory of depression. Behaviour Research and Therapy, 33, 41–48.
*Hammen, C. L., & Krantz, S. (1976). Effect of success and failure on depressive cogni-
tions. Journal of Abnormal Psychology, 85, 577–586.
*Hancock, J. A., Moffoot, A. P. R., & O'Carroll, R. E. (1996). “Depressive realism” assessed
via confidence in decision-making. Cognitive Neuropsychiatry, 1, 213–220.
*Harkness, K. L., Sabbagh, M. A., Jacobson, J. A., Chowdrey, N. K., & Chen, T. (2005). En-
hanced accuracy of mental state decoding in dysphoric college students. Cognition
and Emotion, 19, 999–1025.
Hedges, L. V. (1981). Distribution theory for Glass' estimator of effect size and related
estimators. Journal of Educational Statistics, 6, 107–128.
Hedges, L. V. (1994). Statistical considerations. In H. Cooper, & L. V. Hedges (Eds.), The
handbook of research synthesis (pp. 28–38). New York: Russell Sage.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis. San Diego, CA: Ac-
Hedges, L. V., & Vevea, J. L. (1998). Fixed and random-effects models in meta-analysis.
Psychological Methods, 3, 486–504.
Hoehn-Hyde, D., Schlottman, R. S., & Rush, A. J. (1982). Perception of social interactions
in depressed psychiatric patients. Journal of Consulting and Clinical Psychology, 50,
Hunter, J. E., & Schmidt, F. L. (2000). Fixed-effects vs. random-effects meta-analysis
models: Implications for cumulative research knowledge. International Journal of
Selection and Assessment, 8, 275–292.
Jacobson, N. S., Dobson, K. S., Truax, P. A., Addis, M. E., Koerner, K., Gollan, J. K., Gortner,
E., & Prince, S. E. (1996). A component analysis of cognitive-behavioral treatment
for depression. Journal of Consulting and Clinical Psychology, 64, 295–304.
*Javna, C. D. (1981). Depressed and nondepressed college students' interpretations of
and memory for feedback about self and others. Unpublished doctoral dissertation,
Ohio State University, Columbus, OH.
*Johnson, T. J., & DiLorenzo, T. M. (1998). Social information processing biases in de-
pressed and nondepressed college students. Journal of Social Behavior & Personality,
*Johnson, J. E., Petzel, T. P., Hartney, L. M., & Morgan, R. A. (1983). Recall of importance
ratings of completed and uncompleted tasks as a function of depression. Cognitive
Therapy and Research, 7, 51–56.
*Kapci, E. G., & Cramer, D. (1998). The accuracy of dysphoric and nondepressed groups'
predictions of life events. Journal of Psychology: Interdisciplinary and Applied, 132,
*Kapci, E. G., & Cramer, D. (1999). Judgement of control revisited: Are the depressed re-
alistic or pessimistic? Counselling Psychology Quarterly, 12, 95–105.
Kendall, P. C., Hollon, S. D., Beck, A. T., Hammen, C. L., & Ingram, R. E. (1987). Issues and
recommendations regarding use of the Beck Depression Inventory. Cognitive Therapy
and Research, 11, 289–299.
Kessler, R. C., Berglund, P., Demler, O., Jin, R., Koretz, D., Merikangas, K. R., et al. (2003).
The epidemiology of major depressive disorder: Results from the National Comor-
bidity Survey Replication (NCS-R). Journal of the American Medical Association, 289,
Kessler, R. C., Chiu, W. T., Demler, O., & Walters, E. E. (2005). Prevalence, severity, and
comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey
Replication. Archives of General Psychiatry, 62, 617–627.
*Klein, D. C. (1975). Reversal of performance deficits and perceptions of response-
reinforcement independence associated with learned helplessness and depression.
Unpublished doctoral dissertation, University of Pennsylvania, Philadelphia, PA.
*Krantz, S. E., & Gallagher-Thompson, D. (1990). Depression and information valence
influence depressive cognition. Cognitive Therapy and Research, 14, 95–108.
Kuyken, W., & Brewin, C. R. (1995). Autobiographical memory functioning in depres-
sion and reports of early abuse. Journal of Abnormal Psychology, 104, 585–591.
Kuyken, W., & Dalgleish, T. (1995). Autobiographical memory and depression. British
Journal of Clinical Psychology, 34, 89–92.
Langer, E. J. (1975). The illusion of control. Journal of Personality and Social Psychology,
Lewinsohn, P. M., Mischel, W., Chaplain, W., & Barton, R. (1980). Social competence and
depression: The role of illusory self-perceptions? Journal of Abnormal Psychology,
Lobitz, W. C., & Post, R. D. (1979). Parameters of self-reinforcement and depression.
Journal of Abnormal Psychology, 88, 33–41.
*Loeb, A., Beck, A. T., & Diggory, J. C. (1971). Differential effects of success and failure on
depressed and nondepressed patients. The Journal of Nervous and Mental Disease,
*Loewenstein, D. A., & Hokanson, J. E. (1986). The processing of social information by
mildly and moderately dysphoric college students. Cognitive Therapy and Research,
Longmore, R. J., & Worrell, M. (2007). Do we need to challenge thoughts in cognitive
behavioral therapy? Clinical Psychology Review, 27, 173–187.
*Lovejoy, M. C. (1991). Maternal depression: Effects on social cognition and behavior in
parent–child interactions. Journal of Abnormal Child Psychology, 19, 693–706.
MacLeod, C., Mathews, A., & Tata, P. (1986). Attentional biases in emotional disorders.
Journal of Abnormal Psychology, 95, 15–20.
*Margo, G. M., Greenberg, R. P., Fisher, S., & Dewan, M. (1993). A direct comparison of
the defense mechanisms of nondepressed people and depressed psychiatric inpa-
tients. Comprehensive Psychiatry, 34, 65–69.
*Martin, D. J., Abramson, L. Y., & Alloy, L. B. (1984). The illusion of control for self and
others in depressed and nondepressed college students. Journal of Personality and
Social Psychology, 46, 125–136.
McCabe, S. B., & Gotlib, I. H. (1995). Selective attention and clinical depression: Perfor-
mance on a deployment-of-attention task. Journal of Abnormal Psychology, 104,
McCabe, S. B., Gotlib, I. H., & Martin, R. A. (2000). Cognitive vulnerability to depression:
Deployment-of-attention as a function of history of depression and current mood
state. Cognitive Therapy and Research, 24, 427–444.
McCabe, S. B., & Toman, P. E. (2000). Stimulus exposure duration in a deployment-of-
attention task: Effects on dysphoric, recently dysphoric, and nondysphoric individuals.
Cognition and Emotion, 14, 125–142.
*McKendree-Smith, N., & Scogin, F. (2000). Depressive realism: Effects of depression
severity and interpretation time. Journal of Clinical Psychology, 56, 1601–1608.
*McNamara, K., & Hackett, G. (1986). Gender, sex-type and cognitive distortion: Self-
perceptions of social competence among mild depressives. Social Behavior and
Personality, 14, 113–121.
*Mikulincer, M., Gerber, H., & Weisenberg, M. (1990). Judgment of control and depres-
sion: The role of self-esteem threat and self-focused attention. Cognitive Therapy
and Research, 14, 589–608.
*Miller, W. R., & Seligman, M. E. P. (1973). Depression and the perception of reinforce-
ment. Journal of Abnormal Psychology, 82, 62–73.
*Miller, W. R., & Seligman, M. E. P. (1976). Learned helplessness, depression, and the
perception of reinforcement. Behaviour Research and Therapy, 14, 7–17.
*Miller, W. R., Seligman, M. E. P., & Kurlander, H. M. (1975). Learned helplessness,
depression, and anxiety. The Journal of Nervous and Mental Disease, 161, 347–357.
Moore, M. T., & Fresco, D. M. (2007). Depressive realism and attributional style: Impli-
cations for individuals at risk for depression. Behavior Therapy, 38, 144–154.
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509
*Moretti, M. M. (1985). Processing positive and negative stimuli in depression. Download full-text
Unpublished doctoral dissertation, Simon Fraser University, Vancouver, BC, Canada.
*Msetfi, R. M., Murphy, R. A., & Simpson, J. (2007). Depressive realism and the effect of
inter-trial interval on judgments of zero, positive and negative contingencies.
Quarterly Journal of Experimental Psychology, 60, 461–481.
*Msetfi, R. M., Murphy, R. A., Simpson, J., & Kornbrot, D. E. (2005). Depressive realism
and outcome density bias in contingency judgments: The effect of the context
and intertrial interval. Journal of Experimental Psychology: General, 134, 10–22.
Musson, R. F., & Alloy, L. B. (1987). Depression, self-consciousness, and judgments of
control: A test of the self-focused attention hypothesis. Unpublished manuscript,
Northwestern University, Evanston, IL.
*Nelson, R. E., & Craighead, W. E. (1977). Selective recall of positive and negative feedback,
self-control behaviors and depression. Journal of Abnormal Psychology, 86, 379–388.
*Pacini, R., Muir, F., & Epstein, S. (1998). Depressive realism from the perspective of
cognitive-experiential self-theory. Journal of Personality and Social Psychology, 74,
*Presson, P. K., & Benassi, V. A. (2003). Are depressive symptoms positively or negatively
associated with the illusion of control? Social Behavior and Personality, 31, 483–495.
Puffet, A., Jehin-Marchot, D., Timsit-Berthier, M., & Timsit, M. (1991). Autobiographical
memory and major depressive states. European Psychiatry, 6, 141–145.
*Pusch, D., Dobson, K. S., Ardo, K., & Murphy, T. (1998). The relationships between
sociotropic and autonomous personality styles and depressive realism in dysphoric
and nondysphoric university students. Canadian Journal of Behavioural Science, 30,
*Pyszczynski, T., Holt, K., & Greenberg, J. (1987). Depression, self-focused attention, and
expectancies for positive and negative future life events for self and others. Journal
of Personality and Social Psychology, 52, 994–1001.
Raudenbush, S. W. (1994). Analyzing effect sizes: Random effects models. In H. Cooper,
& L. V. Hedges (Eds.), The handbook of research synthesis (pp. 295–315). New York:
*Rosenfarb, I. S., Burker, E. J., Morris, S. A., & Cush, D. T. (1993). Effects of changing con-
tingencies on the behavior of depressed and nondepressed individuals. Journal of
Abnormal Psychology, 102, 642–646.
Rosenthal, R. (1979). The file drawer problem and tolerance for null results. Psycholog-
ical Bulletin, 86, 638–641.
Rosenthal, R. (1991). Meta-analytic procedures for social research (2nd ed.). Newbury
Park, CA: Sage.
*Roth, D., & Rehm, L. P. (1980). Relationships among self-monitoring processes. Mem-
ory, and depression. Cognitive Therapy and Research, 4, 149–157.
Rozensky, R. H., Rehm, L. P., Pry, G., & Roth, D. (1977). Depression and self-
reinforcement behavior in hospitalized patients. Journal of Behavior Therapy and
Experimental Psychiatry, 8, 35–38.
Ruehlman, L. S., West, S. G., & Pasahow, R. J. (1985). Depression and evaluative schema-
ta. Journal of Personality, 53, 46–92.
*Sacco, W. P., & Hokanson, J. E. (1978). Expectations of success and anagram perfor-
mance of depressives in a public and private setting. Journal of Abnormal Psycholo-
gy, 87, 122–130.
*Sacco, W. P., & Hokanson, J. E. (1982). Depression and self-reinforcement in a public
and private setting. Journal of Personality and Social Psychology, 42, 377–385.
Schmidt, F. L., Oh, I., & Hayes, T. L. (2009). Fixed- versus random-effects models in
meta-analysis: Model properties and an empirical comparison of differences in re-
sults. British Journal of Mathematical and Statistical Psychology, 62, 97–128.
Siegel, S. J., & Alloy, L. B. (1990). Interpersonal perceptions and consequences of
depressive-significant other relationships: A naturalistic study of college room-
mates. Journal of Abnormal Psychology, 99, 361–373.
*Stone, A. L., & Glass, C. R. (1986). Cognitive distortion of social feedback in depression.
Journal of Social and Clinical Psychology, 4, 179–188.
*Stone, E. R., Dodrill, C. L., & Johnson, N. (2004). Depressive cognition: A test of the de-
pressive realism versus negativity using general knowledge questions. The Journal
of Psychology, 135, 583–602.
*Strack, S., & Coyne, J. C. (1983). Social confirmation of dysphoria: Shared and private
reactions. Journal of Personality and Social Psychology, 44, 798–806.
*Strunk, D. R., & Adler, A. D. (2009). Cognitive biases in three predictive tasks: A test of
the cognitive model of depression. Behaviour Research and Therapy, 47, 34–40.
*Strunk, D. R., Lopez, H., & DeRubeis, R. J. (2006). Depressive symptoms are associated
with unrealistic negative predictions of future life events. Behaviour Research and
Therapy, 44, 875–896.
Tennen, H., & Herzberger, S. (1987). Depression, self-esteem, and the absence of self-
protective attributional biases. Journal of Personality and Social Psychology, 52, 72–80.
*Vazquez, C. V. (1987). Judgment of contingency: Cognitive biases in depressed
and nondepressed subjects. Journal of Personality and Social Psychology, 52,
*Vestre, N. D., & Caulfield, B. P. (1986). Perception of neutral personality descriptors by
depressed and nondepressed subjects. Cognitive Therapy and Research, 10, 31–36.
Wener, A. E., & Rehm, L. P. (1975). Depressive affect: A test of behavioral hypothesis.
Journal of Abnormal Psychology, 84, 221–227.
*Wenzlaff, R. M. (1984). Depression and judgments of personal feedback: Accuracy at
the expense of self-enhancement? Unpublished doctoral dissertation, University
of Texas, Austin.
Wenzlaff, R. M., & Berman, J. S. (1985, August). Judgmental accuracy in depression.
Paper presented at the meeting of the American Psychological Association, Los Angeles.
*Whitton, S. W., Larson, J. J., & Hauser, S. T. (2008). Depressive symptoms and bias in
perceived social competence among young adults. Journal of Clinical Psychology,
M.T. Moore, D.M. Fresco / Clinical Psychology Review 32 (2012) 496–509