Personality and Individual Differences 54 (2013) 692697

Contents lists available at SciVerse ScienceDirect

Personality and Individual Differences

journal homepage: www.elsevier.com/locate/paid

Perfectionism, discrepancies, and depression: Testing the perfectionism social

disconnection model in a short-term, four-wave longitudinal study
Simon B. Sherry a,b,, Anna L. MacKinnon a, Kristin-Lee Fossum a, Martin M. Antony c, Sherry H. Stewart a,b,
Dayna L. Sherry d, Logan J. Nealis a, Aislin R. Mushquash a
a

Department of Psychology, Dalhousie University, 1355 Oxford Street, Halifax, Nova Scotia, Canada B3H4R2
Department of Psychiatry, Dalhousie University, 5909 Veterans Memorial Lane, Halifax, Nova Scotia, Canada B3H2E2
Department of Psychology, Ryerson University, 350 Victoria Street, Toronto, Ontario, Canada M5B2K3
d
Queen Elizabeth II Health Sciences Centre, 1276 South Park Street, Halifax, Nova Scotia, Canada B3H2Y9
b
c

a r t i c l e

i n f o

Article history:
Received 23 May 2012
Received in revised form 17 November 2012
Accepted 21 November 2012
Available online 20 December 2012
Keywords:
Perfectionism
Discrepancies
Social maladjustment
Depression

a b s t r a c t
Perfectionistic concerns (i.e., negative reactions to failures, exaggerated concerns over others criticism
and expectations, and nagging self-doubts) are linked to social disconnection and depressive symptoms.
According to the perfectionism social disconnection model, perfectionistic concerns contribute to social
disconnection (i.e., feeling rejected, excluded, and unwanted by others) which, subsequently, contributes
to depressive symptoms. The social world is replete with chances for interpretations. In interpreting their
social worlds, people high in perfectionistic concerns tend to perceive interpersonal discrepancies, a distressing form of social disconnection that involves perceptions of others as dissatised with them and as
disapproving of them. These interpretations are also conceptualized as having depressing consequences
for people high in perfectionistic concerns. This study tested whether perceived interpersonal discrepancies mediate the relation between perfectionistic concerns and depressive symptoms; 240 participants
were recruited and this mediational model was tested with a four-wave, 4-week longitudinal design.
Structural equation modeling with bootstrapped tests of mediation indicated the perfectionistic concerns-depressive symptoms relationship was mediated by interpersonal discrepancies (even after controlling for perfectionistic strivings). People high in perfectionistic concerns perceive others as
dissatised with them and as disapproving of them. Feeling rejected, excluded, and unwanted by others,
people high in perfectionistic concerns are vulnerable to depression.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
Perfectionism is a risk factor for and maintenance factor in emotional distress (Stoeber & Yang, 2010). In particular, perfectionism
is linked with depressive symptomseven after controlling for
well-established contributors to depressive symptoms such as
neuroticism (Sherry & Hall, 2009). Although perfectionism plays
an important role in depressive symptoms, the mechanisms
through which perfectionism inuences depressive symptoms
need explication.
1.1. Improving research on perfectionism and depressive symptoms
Notable limitations exist in research on perfectionism and
depressive symptoms. Most research in this area uses cross-sectional or two-wave longitudinal designs. Cross-sectional designs
Corresponding author at: Department of Psychology, Dalhousie University,
1355 Oxford Street, Halifax, Nova Scotia, Canada B3H4R2. Tel.: +1 902 494 8070;
fax: +1 902 494 6585.
E-mail address: [email protected] (S.B. Sherry).
0191-8869/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.paid.2012.11.017

prevent causal inferences, since they fail to address directionality

or temporal precedence. Two-wave longitudinal designs capture
a narrow slice of change and often involve temporal confounding
(e.g., a mediator and a criterion both measured at Wave 2). Collecting three or more waves of data better captures changes over time
and, for mediational designs, permits assessment of temporally
independent predictors, mediators, and outcomes. In the present
study, we used a four-wave longitudinal design that enabled stronger causal inferences.
Longitudinal studies with widely spaced measurement intervals
(e.g., 12 months) test long-term relationships, but may miss important experiences due to infrequent sampling. Our study involved
weekly measurement intervals, which minimize recall bias. Data
collected over shorter intervals may also better capture short-term
micro-transactions between participants and their social environments. This is important as social phenomena were key to our
study. Short-term, multiwave longitudinal designs also increase
reliability by repeatedly assessing events.
Studies show links among perfectionism and depressive symptoms, perfectionism and social problems, and social problems and

S.B. Sherry et al. / Personality and Individual Differences 54 (2013) 692697

depressive symptoms (Hewitt, Flett, Sherry, & Caelian, 2006). However, research seldom integrates these ndings (for exceptions, see
Cox, Clara, & Enns, 2009; Dunkley, Sanislow, Grilo, & McGlashan,
2006). In contrast, our study aimed to provide an integrative model
explaining how perfectionistic concerns (i.e., negative reactions to
failures, exaggerated concerns over others criticism and expectations, and nagging self-doubts) contribute to depressive symptoms
via social problems.
Whereas past perfectionism research often relied on single indicators of study constructs (Sherry, Law, Hewitt, Flett, & Besser,
2008), our study measured constructs with latent variables each
composed of multiple indicators. For example, we operationalized
perfectionism using a perfectionistic concerns latent variable. This
latent variable represents a theoretical and empirical synthesis of
several well-researched dimensions of perfectionism (Dunkley,
Blankstein, Halsall, Williams, & Winkworth, 2000; Stoeber & Otto,
2006). Latent variables also provide a more encompassing assessment of constructs that is not reliant on the properties of any
one single scale. Moreover, latent variables result in more reliable
statistical estimates, take measurement error into account, and allow tests of models with multiple predictors, mediators, and outcomes (Kline, 2005).
We address the above limitations by testing an integrative theoretical model using latent variables and a short-term, four-wave
longitudinal design. Our study thus represents a needed contribution to research on perfectionism and depressive symptoms.
1.2. Perfectionism social disconnection model (PSDM)
The social world is nuanced and replete with opportunities for
interpretations. For example, a supervisor asking a supervisee a
seemingly straight forward questionHow is your thesis
going?may be interpreted in many ways, including a benign
question from a supportive supervisor or a pressure-lled criticism
from a dissatised supervisor. Considered from this perspective,
our social worlds are heavily inuenced by interpretationsand
the interpretations we make can profoundly inuence our mood.
According to the PSDM (Hewitt et al., 2006; see Cox et al., 2009),
people high in perfectionistic concerns are vulnerable to depressive symptoms because they experience social disconnection (i.e.,
feeling rejected, excluded, and unwanted by others). Whether a
person feels a sense of connectedness (e.g., care, nurturance, and
support) or disconnection (e.g., judgment, criticism, and rejection)
depends heavily upon interpretations. People high in perfectionistic concerns tend to perceive themselves as not accepted or not
belonging and to see themselves as consistently falling short of
others expectations (Hewitt et al., 2006; Stoeber, 2012). Feeling
others are dissatised with them and disapproving of them also
appears to have depressing consequences for people high in perfectionistic concerns (Sherry et al., 2008). The sense of close connection to others that is essential to well-being is elusive for
people high in perfectionistic concerns.
Given a social world lled with ambiguities, personality traits
(e.g., perfectionistic concerns) may guide social cognition (i.e.,
how people attend to and interpret social information in their efforts to understand their social worlds). Perfectionistic concerns involve a chronic, dispositional tendency to view others as
hypercritical, demanding, and intolerant of mistakes (Dunkley
et al., 2006; Hewitt et al., 2006). This disposition may inuence
how people high in perfectionistic concerns interpret and make
meaning of their social worlds.
It seems people high in perfectionistic concerns are often
defective detectives when interpreting their social worlds
prone to experiencing interpersonal discrepancies, a form of social
cognition where they conclude they are letting others down.
Consistent with this assertion, perfectionistic concerns and

693

interpersonal discrepancies are linked (Sherry & Hall, 2009). In fact,

across various contexts and occasions, interpersonal discrepancies
appear to be a prototypic form of social cognition for people high in
perfectionistic concerns. This tendency to see others as dissatised
with them and as disapproving of them leaves people high in perfectionistic concerns in a bind: On one hand they want approval,
acceptance, and belonging, but on the other hand they consistently
perceive the oppositedisapproval, rejection, and criticism
(Hewitt et al., 2006). Interpersonal discrepancies may thus be a
particularly depressing form of social disconnection for people
high in perfectionistic concerns, especially given their fear of
evaluation and need for approval (Hewitt & Flett, 1991).
1.3. Hypotheses
We hypothesized perfectionistic concerns would indirectly affect depressive symptoms through social disconnection. Specically, we hypothesized perfectionistic concerns would contribute
to social disconnection which, in turn, would contribute to depressive symptoms. Social disconnection was operationalized as interpersonal discrepancies (see Fig. 1). We also based this hypothesis
on two preliminary cross-sectional pilot studies (Sherry et al.,
2008, 2012), where a similar pattern of indirect effects was predicted and observed.
We hypothesized all paths in the PSDM would remain virtually
unchanged after controlling for perfectionistic strivings (i.e., ceaselessly demanding perfection of oneself). Shafran, Cooper, and
Fairburn (2002) propose perfectionistic strivings play a key role
in the perfectionism-psychopathology link and suggest research
should focus on strictly intrapersonal models of perfectionism. In
contrast, perfectionistic concerns represent personality traits
where interpersonal concerns (e.g., social-evaluative fears) gure
prominently. Given Shafran et al.s (2002) call to use an entirely
intrapersonal model of perfectionism, it is important to test if
perfectionistic concerns add incrementally to our understanding
of depressive symptoms beyond perfectionistic strivings.
2. Method
2.1. Participants
A sample of 240 undergraduates (200 women; 40 men) was recruited from psychology courses at Dalhousie University. Participants averaged 20.00 years of age (SD = 3.23) and were primarily
in rst (42.9%), second (20.4%), and third (20.8%) year of university.
Participants reported living in Canada for an average of 18.37 years
(SD = 5.84) and were predominantly (86.7%) of European descent.
Most participants were single (47.5%) or dating (40.0%), with
7.5% cohabitating, 3.8% married, and 0.4% separated.
2.2. Measures
2.2.1. Perfectionistic concerns
The latent perfectionistic concerns variable was measured using
three observed indicators and 13 items: A 5-item short form of the
concern over mistakes subscale (COM; e.g., If I fail at work/school,
I am a failure as a person) of the Frost Multidimensional Perfectionism Scale (FMPS; Cox, Enns, & Clara, 2002; Frost, Marten, Lahart, &
Rosenblate, 1990), the 4-item doubts about actions subscale (DAA;
e.g., I usually have doubts about the simple everyday things I do)
of the FMPS, and the 4-item interpersonal perceptions subscale (IP)
of the FMPS. We selected these three manifest indicators for our latent perfectionistic concerns variable as theory and evidence suggest they measure core cognitive, behavioral, and interpersonal
features of perfectionistic concerns (e.g., Dunkley et al., 2000).

694

S.B. Sherry et al. / Personality and Individual Differences 54 (2013) 692697

.93
Perfectionistic
concerns
(Wave 1)

.79

.72
Interpersonal
discrepancies
(Wave 1)

.67

.96

.68

.17
.10

Depressive
symptoms
(Wave 1)

.76

Perfectionistic
concerns
(Wave 2)

.73
Interpersonal
discrepancies
(Wave 2)

.69
Depressive
symptoms
(Wave 2)

.96
.98
.77
.17
.11
.79

Perfectionistic
concerns
(Wave 3)

.84
Interpersonal
discrepancies
(Wave 3)

.72
Depressive
symptoms
(Wave 3)

.92
.96
.76
.15
.11
.77

Perfectionistic
concerns
(Wave 4)

.82
Interpersonal
discrepancies
(Wave 4)

.69
Depressive
symptoms
(Wave 4)

Fig. 1. Structural model for the perfectionism social disconnection model. Ovals represent latent variables. Double-headed black arrows represent signicant latent
correlations (p < .05). Single-headed black arrows represent signicant paths (p < .05). Standardized path coefcients appear in bold. Italicized numbers (e.g., .69) appearing in
the upper right hand of endogenous variables (e.g., Depressive symptoms [Wave 4]) represent the proportion of variance accounted for by associated exogenous variables.

The COM was changed into a 5-item scale by Cox et al. (2002) by
selecting the ve highest factor loadings from Frost et al.s original,
9-item COM. Cox et al. found the psychometric properties of the
COM improved after it was reduced from 9 to 5 items. The COM
is comprised of items 9, 13, 14, 23, and 34 from Frost et al.s original, 9-item COM. The DAA subscale was not modied. The IP is a
modied (but not shortened in terms of scale length) version of
the parental perceptions subscale of the FMPS.1 For example, As
a child, I was punished for doing things less than perfect on the
parental perceptions subscale was modied to Others punish me
for doing things less than perfect on the IP. Modifying the IP in this
way enabled us to assess perfectionistic concerns apart from childhood history. The COM, DAA, and IP are rated on a 5-point scale from
1 (strongly disagree) to 5 (strongly agree). Evidence supports the
reliability and validity of these subscales (Cox et al., 2002; Frost,
Heimberg, Holt, & Mattia, 1993; Sherry & Hall, 2009). Sherry
(2012) found the COM and IP have high alpha reliabilities (.86 and
.81 respectively) and strong correlations (rs = .91 and .68 respectively) with their original subscales. The alpha reliability for the
DAA was .99 in Sherry.
2.2.2. Perfectionistic strivings
Perfectionistic strivings were measured using a 4-item short
form of the personal standards subscale (PS; e.g., I set higher goals
than others) of the FMPS. The PS is rated on a 5-point scale from 1
(strongly disagree) to 5 (strongly agree). The reliability and validity
of the PS is supported by research (Cox et al., 2002). In Sherry
(2012), the PS had high alpha reliability (.87) and a strong correlation with the original subscale (r = .98).
2.2.3. Interpersonal discrepancies
The latent interpersonal discrepancies variable was measured
using three observed indicators and 14 items: A 5-item short form
of the interpersonal discrepancies subscale (e.g., I felt I had disappointed others.) of the Reconstructed Depressive Experiences Questionnaire (RDEQ-ID; Bagby, Parker, Joffe, & Buis, 1994), the 5-item
interpersonal discrepancies subscale (e.g., Were you unable to
reach goals others have imposed on you?) of the Multidimensional
Discrepancies Inventory (MDI-ID; Flett & Hewitt, 2012), and the 4item interpersonal discrepancies subscale (e.g., My best was not
good enough for others.) of the Almost Perfect Scale-Revised
(APS-R-ID; Slaney, Rice, Mobley, Trippi, & Ashby, 2001; see Sherry
1
A cross-sectional study was conducted to assess the psychometric properties of
modied scales we used. This study involved 119 undergraduates (79.0% women).
Participants averaged 20.58 years of age (SD = 4.33). This study is referenced as Sherry
(2012).

& Hall, 2009). We selected these three manifest indicators for our
latent interpersonal discrepancies variable as research suggests
they adequately measure central features of the interpersonal discrepancies construct and cohere together in a valid way (e.g.,
Mushquash & Sherry, 2012). Consistent with our study design,
measures of interpersonal discrepancies were modied to have a
short-term timeframe (i.e., during the past 7 days). The RDEQ-ID
and APS-R-ID are rated on a 7-point scale from 1 (strongly disagree)
to 7 (strongly agree). The MDI-ID is rated on a 4-point scale from 1
(not at all) to 4 (very much). Studies support the reliability and
validity of these subscales (Mushquash & Sherry, 2012). Sherry
(2012) found the RDEQ-ID, MDI-ID, and APS-R-ID have high alpha
reliabilities (.85, .85, .91 respectively) and correlate strongly with
their original subscales (rs = .68, .63, .58 respectively). Discriminant validity analyses indicate interpersonal discrepancies and
perfectionistic concerns are best represented as distinct constructs
(Mushquash & Sherry, 2012). Such results suggest that feeling one
has not achieved others standards (interpersonal discrepancies) is
different from negative reactions to failures, concerns over others
criticism, and nagging self-doubts (perfectionistic concerns).

2.2.4. Depressive symptoms

The latent depressive symptoms variable was measured using
three observed indicators and 24 items: A 4-item short form of
the depression subscale (e.g., sad) of the Prole of Mood States
(POMS-D; McNair, Lorr, & Droppleman, 1992), the 13-item depression subscale (e.g., Feeling hopeless about the future) of the
Symptom Checklist-90-R (SCL-90-D; Derogatis, 1994), and the
7-item depression subscale (e.g., I felt that I had nothing to look
forward to.) of the 21-item Depression Anxiety and Stress Scale
(DASS-D; Lovibond & Lovibond, 1995). The SCL-90-D and DASS-D
were not altered. We selected these three manifest indicators for
our latent depressive symptoms variable as they measure core
affective, cognitive, interpersonal, and physiological features of
depressive symptoms (e.g., Barker-Collo, 2003). All three manifest
indicators are needed to adequately represent these core features.
A 7-day timeframe was used for measures of depressive symptoms. The POMS-D and SCL-90-D are rated on a 5-point scale from
0 (not at all) to 4 (extremely). The DASS-D is rated on a 4-point scale
from 0 (did not apply to me at all) to 3 (applied to me very much, or
most of the time). Research supports the reliability and validity of
all three subscales (Antony, Bieling, Cox, Enns, & Swinson, 1998).
Sherry (2012) found the POMS-D has high alpha reliabilities (.84)
and is strongly correlated with the original subscale (r = .87). In
Sherry, the alpha reliabilities for the SCL-90-D and the DASS-D
were .92 and .89 respectively.

S.B. Sherry et al. / Personality and Individual Differences 54 (2013) 692697

695

Dalhousie Universitys Research Ethics Board approved this

study. Participants completed questionnaires once a week for four
consecutive weeks. After Wave 4, participants received $10 and a
3.0% bonus credit.
In total, 240 (100%), 238 (99.2%), 230 (95.8%), and 232 (96.7%)
participants completed Waves 1, 2, 3, and 4, respectively. Waves
2, 3, and 4, occurred an average of 7.04 (SD = 0.44), 14.13
(SD = 0.66), and 21.20 (SD = 1.23) days after Wave 1, respectively.

Perfectionistic strivings signicantly correlated with all but one

of the variables in the PSDM (i.e., Wave 1 perfectionistic strivings
was not signicantly correlated with Wave 3 depressive symptoms). These correlations suggested perfectionistic strivings are a
suitable covariate. Demographics were not included as covariates,
as none correlated with the variables in the PSDM. High testretest
correlations for perfectionistic concerns (rs = .82.92), perfectionistic
strivings
(rs = .81.92),
interpersonal
discrepancies
(rs = .66 .80), and depressive symptoms (rs = .66.78) supported
the reliability of the measures (see Table 2).

2.4. Data analysis

3.3. SEM

Conrmatory factor analysis (CFA) tested the measurement

model for the perfectionism social disconnection model (PSDM).
Structural equation modeling (SEM) tested the structural model
for the PSDM. Bootstrapping analyses tested the mediational
hypothesis in this structural model. Incremental validity analyses
tested if paths in the PSDM remained signicant when controlling
for perfectionistic strivings.

SEM analyses used AMOS 7.0. A v2/df ratio around 2, a comparative t index (CFI) and an incremental t index (IFI) around .95,
and a root-mean-square error of approximation (RMSEA) around
.06 indicate good model t (Kline, 2005). RMSEA values are reported with 90% condence intervals (CI). The Akaike information
criterion (AIC) was used in model comparisons. Smaller AIC values
indicate greater parsimony and better t.

3. Results

3.4. Measurement model for the perfectionism social disconnection

model (PSDM)

2.3. Procedure

3.1. Descriptive statistics

Means and standard deviations for manifest variables (see Table 1) were computed by aggregating across all waves. Aggregated
means from our study fell within one standard deviation of means
from studies involving similar samples (e.g., Sherry & Hall, 2009).
This suggests means from our study are consistent with research
involving comparable samples.
3.2. Alpha reliabilities and bivariate correlations
Manifest indicators of latent variables were standardized (i.e.,
transformed into z-scores) and summed prior to analyses. For
example, the RDEQ-ID was standardized, the MDI-ID was standardized, and the APS-R-ID was standardized; these subscales
were then summed to create the interpersonal discrepancies variable in Table 2. Alpha reliabilities were adequate across all four
waves: perfectionistic concerns (.89.92), perfectionistic strivings
(.87.89), interpersonal discrepancies (.92.94), and depressive
symptoms (.94.95).
Across all waves, (a) perfectionistic concerns were signicantly
correlated with interpersonal discrepancies and depressive symptoms and (b) interpersonal discrepancies and depressive symptoms were signicantly correlated with each other (see Table 2).

Table 1
Means and standard deviations for manifest variables.
M

SD

Perfectionistic concerns
FMPS concern over mistakes
FMPS doubts about actions
FMPS interpersonal perceptions

10.01
8.39
4.27

9.57
4.81
3.57

Perfectionistic strivings
FMPS personal standards

12.93

4.27

Interpersonal discrepancies
RDEQ interpersonal discrepancies
MDI interpersonal discrepancies
APS-R interpersonal discrepancies

12.68
8.63
3.43

8.07
6.72
5.64

3.17
6.11
9.81

11.33
3.08
7.50

Depressive symptoms
POMS-D depressive symptoms
SCL-90-D depressive symptoms
DASS-D depressive symptoms

The measurement model involved the 12 latent variables

shown in Fig. 1 (i.e., four perfectionistic concerns latent variables,
four interpersonal discrepancies latent variables, and four depressive symptoms latent variables). CFA indicated this model had
acceptable t: v2(492, N = 240) = 801.84, p = .000; v2/df = 1.63;
CFI = .97; IFI = .97; RMSEA = .05 (90% CI: .05, .06). This model also
had signicant (p < .001) standardized factor loadings for manifest
indicators. Factor loadings ranged from .60 to .82 for perfectionistic
concerns, .86 to .96 for interpersonal discrepancies, and .84 to .97
for depressive symptoms.
3.5. Structural model for the perfectionism social disconnection model
(PSDM)
3.5.1. Model testing
Fit indices suggested the structural model for the PSDM (see
Fig. 1) t the data well: v2(157, N = 240) = 312.6, p = .00; v2/
df = 1.99; CFI = .97; IFI = .97; RMSEA = .06 (90% CI: .05, .07);
AIC = 1297.97. All autoregressive path coefcients were signicant
(see Fig. 1). For example, the perfectionistic concerns latent variable at Wave 1 signicantly predicted the perfectionistic concerns
latent variable at Wave 2; the perfectionistic concerns latent variable at Wave 2 signicantly predicted the perfectionistic concerns
latent variable at Wave 3; and the perfectionistic concerns latent
variable at Wave 3 signicantly predicted the perfectionistic concerns latent variable at Wave 4. As Fig. 1 shows, paths from the
perfectionistic concerns latent variable to the interpersonal discrepancies latent variable and paths from the interpersonal discrepancies latent variable to the depressive symptoms latent
variable were signicant. Paths in the PSDM were therefore congruent with the hypothesized structural model for the PSDM.
3.5.2. Mediational analyses
Analysis of the hypothesized indirect effects in the PSDM was
conducted using bootstrapping (Kline, 2005). Random sampling
with replacement was used to make 20,000 (n = 240) bootstrap
samples. Bootstrap estimates indicated the hypothesized indirect
effect in the PSDM was signicant B = .26, [95% CI: .08, .60], and
SE = .12. That is, the indirect effect of perfectionistic concerns on
depressive symptoms through interpersonal discrepancies was
signicant.

696

S.B. Sherry et al. / Personality and Individual Differences 54 (2013) 692697

Table 2
Bivariate correlations.
Manifest variables

Wave 1
1. Perfectionistic concerns
2. Perfectionistic strivings
3. Inter. discrepancies
4. Depressive symptoms
Wave 2
5. Perfectionistic concerns
6. Perfectionistic strivings
7. Inter. discrepancies
8. Depressive symptoms

Wave 1

Wave 2

Wave 3

Wave 4

10

11

12

13

14

15

16

0.36

0.6
0.18

0.56
0.18
0.62

0.86
0.35
0.62
0.55

0.38
0.85
0.21
0.23

0.63
0.18
0.68
0.52

0.56
0.22
0.57
0.73

0.83
0.35
0.57
0.5

0.4
0.84
0.19
0.24

0.65
0.2
0.66
0.51

0.47
0.06
0.46
0.7

0.82
0.35
0.59
0.53

0.36
0.81
0.2
0.25

0.62
0.18
0.66
0.5

0.48
0.17
0.54
0.66

0.46

0.72
0.23

0.6
0.28
0.65

0.91
0.45
0.67
0.58

0.46
0.9
0.26
0.31

0.68
0.23
0.76
0.57

0.47
0.13
0.48
0.78

0.89
0.43
0.69
0.6

0.44
0.89
0.26
0.32

0.69
0.23
0.72
0.56

0.48
0.23
0.5
0.72

0.49

0.71
0.27

0.5
0.16
0.63

0.92
0.44
0.72
0.51

0.46
0.92
0.28
0.18

0.69
0.25
0.8
0.52

0.48
0.23
0.58
0.78

0.47

0.71
0.28

0.53
0.22
0.63

Wave 3
9. Perfectionistic concerns
10. Perfectionistic strivings
11. Inter. discrepancies
12. Depressive symptoms
Wave 4
13. Perfectionistic concerns
14. Perfectionistic strivings
15. Inter. discrepancies
16. Depressive symptoms

Note: Inter = interpersonal. Testretest correlations are bolded. Bivariate correlations P .13 are signicant (p < .05).

3.5.3. Incremental validity

Perfectionistic strivings at Wave 1 were added to Fig. 1 as a
covariate. As hypothesized, paths in the structural model for the
PSDM were virtually unchanged in terms of magnitude and significance after controlling for perfectionistic strivings at Wave 1.
When added to the structural model shown in Fig. 1, perfectionistic
strivings were not signicantly related to perfectionistic concerns
(b = .01, p > .05) or depressive symptoms (b = .01, p > .05). However, perfectionistic strivings were negatively and signicantly related to interpersonal discrepancies (b = .06, p < .05). The t of the
structural model was eroded when perfectionistic strivings were
included in Fig. 1 (AIC = 1477.48). The PSDM was not improved
by including perfectionistic strivings and paths in the PSDM were
still signicant once perfectionistic strivings were taken into
account.

4. Discussion
The perfectionism social disconnection model (PSDM) relates
perfectionistic concerns to depressive symptoms through interpersonal discrepancies. Analyses indicated the measurement model
for the PSDM t the data well. The structural model for the PSDM
t the data well and all hypothesized paths were signicant. Interpersonal discrepancies also mediated the link between perfectionistic concerns and depressive symptoms in the hypothesized
manner. All paths in the PSDM were virtually unaltered after controlling for perfectionistic strivings, thus supporting our
hypotheses.

4.1. Structural model for the perfectionism social disconnection model

(PSDM)
Our study claried one mechanism through which perfectionistic concerns inuence depressive symptoms: perceiving other people as dissatised and as disapproving. Consistent with the PSDM,
perfectionistic concerns indirectly affected depressive symptoms
through interpersonal discrepancies. These results are congruent
with past tests of the PSDM that involved cross-sectional designs
(Sherry et al., 2008, 2012). Our study also improves on these two

studies by stringently testing the PSDM using a longitudinal design

that sheds light on directionality and temporal precedence.
Attempting to understand the social world, and ones unique
place in it, relies greatly on interpretations. Personality traits can
shape the interpretations we make, and the interpretations we
make can inuence our mood. Across various contexts and occasions, people high in perfectionistic concerns appear to interpret
diverse social stimuli similarly and negatively (Sherry & Hall,
2009). In particular, our study suggests these people interpret others in their social worlds as dissatised with them and disapproving of them. To people high in perfectionistic concerns, an
innocuous gesture, benign question, or supportive comment may
all seem loaded with dissatisfaction or disapproval. Perfectionistic
concerns involve viewing others as hypercritical, demanding, and
intolerant of mistakes (Hewitt et al., 2006); thus it seems tting
people high in perfectionistic concerns report they are continually
letting others down. From this perspective, depressive symptoms
may be brought on by the powerful sense of others dissatisfaction
and disapproval experienced by people high in perfectionistic concerns. Indeed, perfectionistic concerns may thwart the feelings of
connection to and acceptance by others that are crucial to emotional well-being.
4.2. Incremental validity
As hypothesized, results supported the incremental validity of
the PSDM beyond a relevant covariate, perfectionistic strivings.
These ndings are important given Shafran et al.s (2002) calls to
return to an exclusively intrapersonal model of perfectionism. In
contrast to Shafran et al.s assertion, our results suggest an important role for perfectionistic concerns (a predominantly interpersonal trait) beyond perfectionistic strivings (an exclusively
intrapersonal trait) when seeking to understand depressive
symptoms.
4.3. Limitations
Our study was limited by the use of self-report measures, which
may lead to biased responses (e.g., defensiveness). Moreover, using
a short-term time lag between measurement occasions may have

S.B. Sherry et al. / Personality and Individual Differences 54 (2013) 692697

lead to overestimation of stability effects and underestimation of

other effects (e.g., paths from perfectionistic strivings to depressive
symptoms). Our use of modied scales is another limitation; less is
known about the psychometrics of our modied scales.
5. Conclusions
Our study integrates perfectionistic concerns, interpersonal discrepancies, and depressive symptoms into the perfectionism social
disconnection model, a coherent theoretical framework for understanding why people high in perfectionistic concerns are vulnerable to depressive symptoms. As they go about their lives, people
who are high in perfectionistic concerns tend to perceive others
as dissatised with them and as disapproving of them. Feeling rejected, excluded, and unwanted by others, people high in perfectionistic concerns are vulnerable to depression.
Acknowledgments
This manuscript was supported by a Grant from the Social Sciences and Humanities Research Council of Canada. Simon Sherry
was funded by the H.J. Eysenck Memorial Fund Award. Fund
sources were not involved in any aspect of this manuscript.
References
Antony, M., Bieling, P., Cox, B., Enns, M., & Swinson, R. (1998). Psychometric
properties of the 42-item and 21-item versions of the depression anxiety stress
scales. Psychological Assessment, 10, 176181. http://dx.doi.org/10.1037/10403590.10.2.176.
Bagby, R., Parker, J., Joffe, R., & Buis, T. (1994). Reconstruction and validation of the
Depressive Experiences Questionnaire. Assessment, 1, 5968. http://dx.doi.org/
10.1177/1073191194001001009.
Barker-Collo, S. (2003). Culture and validity of the symptom checklist-90-revised
and prole of mood states. Cultural Diversity and Ethnic Minority Psychology, 9,
185196. http://dx.doi.org/10.1037/1099-9809.9.2.185.
Cox, B., Clara, I., & Enns, M. (2009). Self-criticism, maladaptive perfectionism, and
depression symptoms in a community sample. Journal of Cognitive
Psychotherapy, 23, 336349. http://dx.doi.org/10.1891/0889-8391.23.4.336.
Cox, B., Enns, M., & Clara, I. (2002). The multidimensional structure of perfectionism
in clinically distressed and college student samples. Psychological Assessment,
14, 365373. http://dx.doi.org/10.1037/1040-3590.14.3.365.
Derogatis, L. (1994). BSI brief symptom inventory, administration, scoring, and
procedures manual. Minneapolis, MN: National Computer Systems.
Dunkley, D., Blankstein, K., Halsall, J., Williams, M., & Winkworth, G. (2000). The
relation between perfectionism and distress. Journal of Counseling Psychology,
47, 437453. http://dx.doi.org/10.1037/0022-0167.47.4.437.

697

Dunkley, D., Sanislow, C., Grilo, C., & McGlashan, T. (2006). Perfectionism and
depressive symptoms 3 years later. Comprehensive Psychiatry, 47, 106115.
http://dx.doi.org/10.1016/j.comppsych.2005.06.003.
Flett, G., Hewitt, P. (2012). The multidimensional discrepancy inventory. Unpublished
manuscript.
Frost, R., Heimberg, R., Holt, C., & Mattia, J. (1993). A comparison of two measures of
perfectionism. Personality and Individual Differences, 14, 119126. http://
dx.doi.org/10.1016/0191-8869(93)90181-2.
Frost, R., Marten, P., Lahart, C., & Rosenblate, R. (1990). The dimensions of
perfectionism. Cognitive Therapy and Research, 14, 449468. http://dx.doi.org/
10.1007/BF01172967.
Hewitt, P., & Flett, G. (1991). Perfectionism in the self and social contexts. Journal of
Personality and Social Psychology, 60, 456470. http://dx.doi.org/10.1037/00223514.60.3.456.
Hewitt, P., Flett, G., Sherry, S., & Caelian, C. (2006). Trait perfectionism dimensions
and suicide behavior. In T. Ellis (Ed.), Cognition and suicide (pp. 215235).
Washington, DC: American Psychological Association.
Kline, R. (2005). Principles and practice of structural equation modeling. New York, NY:
Guilford.
Lovibond, P., & Lovibond, S. (1995). The structure of negative emotional states.
Behaviour Research and Therapy, 33, 335343. http://dx.doi.org/10.1016/00057967(94)00075-U.
McNair, D., Lorr, M., & Droppleman, L. (1992). EdiTS manual for the prole of mood
States. San Diego, CA: Educational Testing Service.
Mushquash, A., & Sherry, S. (2012). Understanding the socially prescribed
perfectionists cycle of self-defeat. Journal of Research in Psychology, 46,
700709. http://dx.doi.org/10.1016/j.jrp. 2012.08.006.
Shafran, R., Cooper, Z., & Fairburn, C. (2002). Clinical perfectionism: A cognitivebehavioural analysis. Behaviour Research and Therapy, 40, 773791. http://
dx.doi.org/10.1016/S0005-7967(01)00059-6.
Sherry, S. (2012). Psychometric data. Unpublished raw data.
Sherry, S., & Hall, P. (2009). The perfectionism model of binge eating. Journal of
Personality and Social Psychology, 96, 690709. http://dx.doi.org/10.1037/
a0014528.
Sherry, S., Hewitt, P., Stewart, S., Mackinnon, A., Mushquash, A., Flett, G., et al.
(2012). Social disconnection and hazardous drinking mediate the link between
perfectionistic attitudes and depressive symptoms. Journal of Psychopathology
and Behavioral Assessment, 34, 370381. http://dx.doi.org/10.1007/s10862-0129291-8.
Sherry, S., Law, A., Hewitt, P., Flett, G., & Besser, A. (2008). Social support as a
mediator of the relationship between perfectionism and depression. Personality
and
Individual
Differences,
45,
339344.
http://dx.doi.org/10.1016/
j.paid.2008.05.001.
Slaney, R., Rice, K., Mobley, M., Trippi, J., & Ashby, J. (2001). The revised almost
perfect scale. Measurement and Evaluation in Counseling and Development, 34,
130145.
Stoeber, J. (2012). Dyadic perfectionism in romantic relationships. Personality and
Individual
Differences,
53,
300305.
http://dx.doi.org/10.1016/
j.paid.2012.04.002.
Stoeber, J., & Otto, K. (2006). Positive conceptions of perfectionism. Personality and
Social
Psychology
Review,
10,
295319.
http://dx.doi.org/10.1207/
s15327957pspr1004_2.
Stoeber, J., & Yang, H. (2010). Perfectionism and emotional reactions to perfect and
awed achievements. Personality and Individual Differences, 49, 246251. http://
dx.doi.org/10.1016/j.paid.2010.03.044.