Sironica, A. y Reeve, B. (2015)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Psychological Assessment © 2015 American Psychological Association

2015, Vol. 27, No. 4, 1471–1483 1040-3590/15/$12.00 http://dx.doi.org/10.1037/pas0000137

A Combined Analysis of the Frost Multidimensional Perfectionism Scale


(FMPS), Child and Adolescent Perfectionism Scale (CAPS), and Almost
Perfect Scale—Revised (APS-R): Different Perfectionist Profiles in
Adolescent High School Students
Amanda Sironic and Robert A. Reeve
University of Melbourne

To investigate differences and similarities in the dimensional constructs of the Frost Multidimensional
Perfectionism Scale (FMPS; Frost, Marten, Lahart, & Rosenblate, 1990), Child and Adolescent Perfectionism
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Scale (CAPS; Flett, Hewitt, Boucher, Davidson, & Munro, 2000), and Almost Perfect Scale—Revised
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(APS-R; Slaney, Rice, Mobley, Trippi, & Ashby, 2001), 938 high school students completed the 3 perfec-
tionism questionnaires, as well as the Depression Anxiety Stress Scales (DASS; Lovibond & Lovibond, 1995).
Preliminary analyses revealed commonly observed factor structures for each perfectionism questionnaire.
Exploratory factor analysis of item responses from the questionnaires (combined) yielded a 4-factor solution
(factors were labeled High Personal Standards, Concerns, Doubts and Discrepancy, Externally Motivated
Perfectionism, and Organization and Order). A latent class analysis of individuals’ mean ratings on each of
the 4 factors yielded a 6-class solution. Three of the 6 classes represented perfectionist subgroups (labeled
adaptive perfectionist, externally motivated maladaptive perfectionist, and mixed maladaptive perfectionist),
and 3 represented nonperfectionist subgroups (labeled nonperfectionist A, nonperfectionist B, and order and
organization nonperfectionist). Each of the 6 subgroups was meaningfully associated with the DASS. Findings
showed that 3 out of 10 students were classified as maladaptive perfectionists, and maladaptive perfectionists
were more prevalent than adaptive perfectionists. In sum, it is evident that combined ratings from the FMPS,
CAPS, and APS-R offer a meaningful characterization of perfectionism.

Keywords: perfectionism, perfectionist, adolescence, subgroup, depression, anxiety, stress

Supplemental materials: http://dx.doi.org/10.1037/pas0000137.supp

Perfectionism is widely regarded as a multidimensional con- tualization of perfectionism may emerge by (a) subjecting items
struct—a claim reflected in the structure of contemporary perfec- from the three perfectionism questionnaires to factor analysis to
tionism questionnaires, which include two or more dimensions in identify higher-order perfectionism dimensions, (b) establishing
their design. An issue of some importance is the degree to which meaningful perfectionist profiles from combinations of higher-
within and across questionnaire dimensions represent unique or order perfectionism dimensions, and (c) examining how emergent
similar constructs. This issue is particularly salient because a perfectionist profiles are associated with differences in psychopa-
growing number of researchers are using combinations of ques- thology.
tionnaire dimensions to identify different perfectionist subgroups
that, in turn, are hypothesized to be associated with differences in Perfectionism Questionnaires and Their Dimensions:
psychopathology (Gaudreau & Thompson, 2010; Rice & Ashby, Similarities and Differences
2007; Rice, Ashby, & Gilman, 2011; Sironic & Reeve, 2012). In
the current study, we examine the uniqueness and similarity of The FMPS, HMPS, and APS-R are commonly used perfection-
dimensions from three frequently used perfectionism question- ism questionnaires. Frost et al.’s (1990) FMPS closely aligns to
naires (i.e., the Frost Multidimensional Perfectionism Scale characteristics ascribed to perfectionism in school students (Hawk-
[FMPS], Frost, Marten, Lahart, & Rosenblate, 1990; the Hewitt ins, Watt, & Sinclair, 2006) and comprises six dimensions: Per-
Multidimensional Perfectionism Scale [HMPS], Hewitt & Flett, sonal Standards (setting self-determined high standards for per-
19911; and the Almost Perfect Scale—Revised [APS-R], Slaney, sonal performance), Concern Over Mistakes (a tendency to be
Rice, Mobley, Trippi, & Ashby, 2001). A more integrated concep- overly critical of one’s performance), Doubts About Actions (a
tendency to have doubts about the quality of one’s performance),
Organization (placing an emphasis on order and precision), Paren-
tal Expectations (placing considerable value on parents’ expecta-
This article was published Online First May 18, 2015. tions), and Parental Criticism (placing considerable value on par-
Amanda Sironic and Robert A. Reeve, Melbourne School of Psycho-
logical Sciences, University of Melbourne.
Correspondence concerning this article should be addressed to Amanda 1
The child and adolescent version of the HMPS, the Child and Adoles-
Sironic or Robert A. Reeve, Melbourne School of Psychological Sciences, cent Perfectionism Scale (CAPS; Flett, Hewitt, Boucher, Davidson, &
University of Melbourne, VIC 3010, Australia. E-mail: amandasironic@ Munro, 2000), was used in this study because research was conducted on
gmail.com or r.reeve@unimelb.edu.au an adolescent sample.

1471
1472 SIRONIC AND REEVE

ents’ disapproval and criticism). Hewitt and Flett (1991) share predicted an internal locus of control; and there were no sig-
Frost et al.’s intra- and interpersonal conceptualization of perfec- nificant predictions with the Order/Organization factor.
tionism. Unlike Frost et al., however, Hewitt and Flett argue that Rice et al. (2005) conducted confirmatory factor analysis (CFA)
perfectionism is best conceptualized as either internally and/or on 241 undergraduate students’ responses on the FMPS, HMPS,
externally motivated. The HMPS comprises three dimensions: and APS-R subscales: a four-factor model yielded a superior fit to
Self-Oriented Perfectionism (holding high standards that originate a three-factor model (Comparative Fit Index [CFI] ⫽ 0.88 and
from the self), Socially Prescribed Perfectionism (adopting high 0.67, respectively). Rice et al. labeled the four factors Adaptive
standards perceived as coming from others), and Other-Oriented Perfectionism (FMPS Personal Standards, HMPS Self-Oriented
Perfectionism (expecting others to hold high standards and be Perfectionism, and APS-R High Standards), Maladaptive Perfec-
perfect). Finally, Slaney et al.’s (2001) APS-R was developed from tionism (FMPS Doubts About Actions, FMPS Concern Over Mis-
a counseling perspective and, unlike the FMPS and HMPS, is a takes, and APS-R Discrepancy), Parental/Social Influences (FMPS
strictly intrapersonal measure of perfectionism (i.e., it does not Parental Criticism, FMPS Parental Expectations, and HMPS So-
assess perfectionism as an externally motivated construct). The cially Prescribed Perfectionism), and Organization/Order (HMPS
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

APS-R comprises three dimensions: High Standards (setting self- Organization and APS-R Order). Rice et al.’s findings suggest that
This document is copyrighted by the American Psychological Association or one of its allied publishers.

determined high standards), Order (striving for order and organi- parental and social performance expectations are significant pre-
zation), and Discrepancy (experiencing inconsistency between dictors of adaptive and maladaptive perfectionism. However, they
one’s self-determined standards and one’s actual performance). argue that although parental and/or social forces may influence
Investigations into perfectionism questionnaire dimensions have perfectionism, they should not be regarded as an index of perfec-
largely focused on distinguishing between positive and negative tionism per se. Shafran and Mansell (2001) agree with this claim,
forms of perfectionism and variable-centered analyses. For in- arguing that socially prescribed beliefs (e.g., beliefs assessed by
stance, Frost, Heimberg, Holt, Mattia, and Neubauer (1993) ad- HMPS Socially Prescribed Perfectionism) are not integral to the
ministered the FMPS and HMPS to 553 undergraduate psychology “classical concept of perfectionism” (p. 886). Moreover, they
students and subjected questionnaire item responses to factor anal- suggest that the parental dimensions of the FMPS are retrospective
ysis to yield two factors: Positive Striving (HMPS Self-Oriented and provide no indication of whether an individual is currently a
perfectionist. While these claims may be relevant to adults, social
Perfectionism, HMPS Other-Oriented Perfectionism, FMPS Per-
influences are paramount in adolescence (Flett, Hewitt, Oliver, &
sonal Standards, and FMPS Organization) and Maladaptive Eval-
Macdonald, 2002). In fact, the expectations of significant others
uative Concerns (HMPS Socially Prescribed Perfectionism, FMPS
are often considered more important than self-expectation in the
Concern Over Mistakes, FMPS Doubts About Actions, FMPS
adolescent developmental period (Muuss, 2006). Arguably, exter-
Parental Expectations, and FMPS Parental Criticism). Maladaptive
nally motivated perfectionism may be important in conceptualiz-
Evaluation Concerns was related to high levels of negative affect
ing adolescent perfectionism.
and depression and low levels of self-esteem (and unrelated to
positive affect), while Positive Striving was related to high levels
of positive affect (and unrelated to negative affect, depression, and Using Perfectionism Dimensions to Develop
self-esteem). Frost et al.’s findings support the view that perfec- Perfectionist Profiles
tionism is best represented by a positive and a negative dimension; While factor analyses have revealed positive and negative di-
a pattern of findings that has since been replicated by other mensions of perfectionism, a growing body of research has used
researchers (see Stoeber & Otto, 2006, for a review). combinations of perfectionism dimensions to identify or create
To date, only two studies have jointly analyzed the HMPS, different subgroups of perfectionists. The focus on a person-
FMPS, and APS-R (Rice, Lopez, & Vergara, 2005; Suddarth & centered approach (compared with a variable-centered approach)
Slaney, 2001). Suddarth and Slaney administered the FMPS, has various advantages. For instance, a person-centered approach
HMPS, and APS-R to 196 undergraduate students (151 women, 45 makes sense of the fact that ratings on one dimension may have a
men) and subjected mean responses from the 12 subscales to different overall meaning depending on how they relate to ratings
exploratory factor analysis (EFA). A three-factor solution emerged on other dimensions. Moreover, the ability to view patterns and
from a principal-components analysis. The first factor was labeled individual differences provides useful information relevant to clin-
Maladaptive Perfectionism (FMPS Doubts About Actions, FMPS ical and counseling presentations, and assists in the development
Concern Over Mistakes, FMPS Parental Criticism, FMPS Parental of more specific and targeted treatment plans.
Expectations, HMPS Socially Prescribed Perfectionism, and Lundh, Saboonchi, and Wangby (2008) provide support for a
APS-R Discrepancy) and accounted for 35.5% of the variance. The person-centered approach (over a variable-centered approach) in
second factor was labeled Adaptive Perfectionism (FMPS Per- their study examining clinically significant perfectionism. On the
sonal Standards, HMPS Self-Oriented Perfectionism, HMPS basis of cluster analysis of FMPS item responses, Lundh et al.
Other-Oriented Perfectionism, and APS-R High Standards) and identified 11 subgroups that differentiated clinical and nonclinical
accounted for 21.8% of the variance. The third factor was samples. In particular, they identified three clinically significant
labeled Order/Organization (FMPS Organization and APS-R perfectionist subgroups (characterized by high ratings on Concern
Order) and accounted for 10.6% of the variance. Suddarth and Over Mistakes, Doubts About Actions, Personal Standards, and
Slaney found the three factors were differentially associated Organization dimensions) and one adaptive perfectionist subgroup
with locus of control, anxiety, and psychological distress. Mal- (characterized by high ratings on the Personal Standards dimen-
adaptive Perfectionism predicted an external locus of control, sion, although lower than in the three clinically significant clusters,
anxiety, and psychological distress; Adaptive Perfectionism high ratings on the Organization dimension, and low ratings on the
PERFECTIONIST PROFILES IN ADOLESCENTS 1473

remaining FMPS dimensions). Lundh et al. suggest that their three which they rejected in favor of a three-cluster solution primarily
clinically significant subgroups provide evidence against using a because the latter was more consistent with extant conceptualiza-
variable-centered approach, in which perfectionists are represented tions of perfectionism.
in terms of ratings on isolated higher-order perfectionism dimen- Other researchers have supported a fourth APS-R perfectionist
sions (e.g., Frost et al., 1993—Adaptive Perfectionism and Mal- subgroup (Sironic & Reeve, 2012; Wang, Slaney, & Rice, 2007).
adaptive Perfectionism dimensions). They argue that if isolated Sironic and Reeve suggested that individuals who exhibit low
perfectionism dimensions actually represent adaptive or maladap- HS/high Disc profiles (i.e., rate themselves as having low stan-
tive dimensions, “adaptive perfectionism” (as measured by the dards but as experiencing high discrepancy) are perfectionists,
Personal Standards and Organization dimensions) should counter- arguing that they may lack insight into their perfectionism. They
act or “buffer” the negative effects of “maladaptive perfectionism” identified four subgroups from a cluster analysis of 17-year-old
(as measured by the Concern Over Mistakes and Doubts About high school students’ APS-R ratings: adaptive perfectionists (high
Actions dimensions). This buffering hypothesis was not supported HS/low Disc), high standards maladaptive perfectionists (high
by Lundh et al.’s clinically significant perfectionist subgroup pro- HS/high Disc), low standards maladaptive perfectionists (low HS/
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

files. Instead, their findings show that high ratings on the Personal high Disc) and nonperfectionists (low HS/low Disc). The high
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Standards dimension are adaptive when combined with the accep- standards maladaptive perfectionists and low standards maladap-
tance of nonperfection (i.e., low ratings on Concern Over Mistakes tive perfectionists subgroups exhibited higher depression ratings
and Doubts About Actions dimensions), but maladaptive when than the adaptive perfectionists subgroup, and the low standards
combined with an inability to accept mistakes and shortcomings maladaptive perfectionist subgroup claimed to use self-regulation
(i.e., high ratings on Concern Over Mistakes and Doubts About learning strategies less often than the adaptive perfectionist and
Actions dimensions). Overall, it is important that ratings on one high standards maladaptive perfectionist subgroups. Sironic and
dimension are examined in relation to ratings on other dimensions Reeve interpreted these findings as suggesting that the low stan-
to understand their overall meaning. dards maladaptive perfectionist subgroup’s poor self-regulatory
Perfectionism questionnaire dimensions have been combined in abilities reflect a lack of insight.
different ways and using different methodological procedures to Wang et al. (2007) also identified four APS-R subgroups from
investigate the significance of categories. For instance, Gaudreau a cluster analysis of Taiwanese university students’ APS-R ratings.
and Thompson (2010) used responses from shortened versions of They suggested that their fourth subgroup (i.e., a low HS/high Disc
the FMPS and HMPS to argue for a four-subgroup model of subgroup) makes sense in a collectivist culture, claiming that
perfectionism. On the basis of an EFA, they identified two higher- individuals may not set high standards for themselves, but are
order dimensions: Evaulative Concerns Perfectionism (ECP; perfectionists because they strive to meet the expectations of
FMPS Concern Over Mistakes, FMPS Doubts About Actions, significant others (e.g., parents, teachers, peers). Wang et al.’s low
FMPS Parental Pressure, and HMPS Socially Prescribed Perfec- HS/high Disc subgroup exhibited lower ratings on trait anxiety,
tionism) and Personal Standards Perfectionism (PSP; FMPS Per- individual-oriented achievement orientation and depression com-
sonal Standards and HMPS Self-Oriented Perfectionism). Gaud- pared with the high HS/high Disc subgroup, and the low HS/high
reau and Thompson proposed a 2 ⫻ 2 model of perfectionism by Disc subgroup had significantly higher scores on trait anxiety, state
crossing high and low ECP and PSP scores. Perfectionist sub- anxiety and depression, and lower scores on self-esteem, compared
groups were based on standardized low (⫺1 SD) and high (⫹1 SD) with the high HS/low Disc subgroup. Wang et al.’s interpretation
ECP and PSP means, resulting in nonperfectionism (low ECP/low implicates external and/or cultural factors in perfectionism. How-
PSP), pure personal standards perfectionism (low ECP/high PSP), ever, they did not measure external factors—the APS-R is an
pure evaluative concerns perfectionism (high ECP/low PSP), and intrapersonal measure of perfectionism and the Discrepancy di-
mixed perfectionism (high ECP/high PSP) subgroups. They found mension does not assess the expectations of others. It is possible
that the four subgroups were differentially associated with inter- that an external index of perfectionism (e.g., the interpersonal
nalization (academic self-determination) and adjustment (positive dimensions of the FMPS or HMPS) may provide a more direct
and negative affect and academic satisfaction). The pure personal assessment of externally motivated perfectionism.
standards perfectionism subgroup was associated with the most
positive outcomes and the pure evaluative concerns perfectionism
The Present Study
subgroup the most negative.
Some researchers have analyzed the APS-R dimensions to iden- In the current study, 938 14- to 18-year-olds completed the
tify different subgroups of perfectionists. Using cluster analysis, FMPS, Child and Adolescent Perfectionism Scale (CAPS; Flett,
Rice and colleagues (Rice & Ashby, 2007; Rice et al., 2011) Hewitt, Boucher, Davidson, & Munro, 2000), APS-R, and the
argued for three subgroups on the basis of the Discrepancy (Disc) Depression Anxiety Stress Scales (DASS; Lovibond & Lovibond,
and High Standards (HS) subscales of the APS-R. Rice and col- 1995). The aim was to determine whether we could identify
leagues distinguished between adaptive perfectionists (high HS/ different perfectionist subgroups in a sample of Australian adoles-
low Disc), maladaptive perfectionists (high HS/high Disc), and cents and to establish how these subgroups are associated with
nonperfectionists (low HS/low Disc). Rice et al. (2011)’s maladap- self-reported depression, anxiety, and stress symptoms. We fo-
tive perfectionists rated themselves as experiencing higher levels cused on adolescent students for three reasons. First, perfectionism
of depression, anxiety, and social stress, and lower levels of and its sequelae tend to be manifest in the mid- to late-adolescent
self-adequacy, self-esteem, and satisfaction with life, compared period, possibly because of stress associated with high school
with adaptive perfectionists and nonperfectionists. It is worth (Herman, Wang, Trotter, Reinke, & Ialongo, 2013; Moshman,
noting that Rice et al. (2011) also found a four-cluster solution, 2005). Second, many psychopathologies emerge in the adolescent
1474 SIRONIC AND REEVE

period and, if untreated, persist throughout the life span. Third, The procedure for data collection was the same for each school.
given perfectionism is linked to some adolescent psychopatholo- A member of the research team distributed the four questionnaires
gies, a more complete understanding of perfectionism may provide to small groups of students. Students were seated so they could not
greater insight into adolescent mental health. It is also worth noting confer with each other; they completed the questionnaires in
that relatively little research has investigated perfectionism in approximately half an hour.
Australian children and adolescents (however, see Einstein, Lovi-
bond, & Gaston, 2000; Hawkins et al., 2006; Huggins, Davis,
Rooney, & Kane, 2008; Kornblum & Ainley, 2005; Mitchell, Measures
Broeren, Newall, & Hudson, 2013; Sironic & Reeve, 2012). It FMPS. The FMPS (Frost, Marten, Lahart, & Rosenblate,
would seem important to identify similarities and differences in 1990) is a 35-item self-report questionnaire that comprises six
perfectionism across cultures, especially given the transcultural subscales: Concern Over Mistakes (nine items), Personal Stan-
use of common perfectionism measures. dards (seven items), Parental Criticism (four items), Parental Ex-
We (1) analyze the factor structure of each of the three perfec-
pectations (five items), Doubts About Actions (four items), and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tionism questionnaires to establish test score validity in our sam-


Organization (six items).2 Items are rated on a five-point scale
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ple, (2) conduct EFA and CFA on the three perfectionism tests
(1 ⫽ disagree strongly and 5 ⫽ agree strongly). FMPS question-
(combined) to uncover higher-order perfectionism dimensions, (3)
naire item scores report good internal reliability with Cronbach
use latent class analysis (LCA) to identify different perfectionist
alphas ranging from 0.70 to 0.93 (Frost et al., 1990; Rice, Ashby,
subgroups based on individuals’ mean ratings of higher-order
& Slaney, 1998; Rice & Mirzadeh, 2000) and good test–retest
perfectionism dimensions, and (4) analyze the relationship be-
reliability across a 10-week period, ranging from 0.63 to 0.88
tween subgroup membership and psychopathology via self-
(Rice & Dellwo, 2002). The subscales and higher-order dimen-
reported DASS ratings.
sions also relate in expected directions to other measures (Frost et
Insofar as we are able to replicate the factor structures of the
al., 1993; Rice et al., 1998).
three perfectionist inventories, we expect to identify at least adap-
CAPS. The CAPS (Flett, Hewitt, Boucher, Davidson, & Mu-
tive and maladaptive perfectionist subgroups, as well as a nonper-
nro, 2000) is a 22-item self-report measure, which is closely
fectionist subgroup. In addition, because we include the FMPS,
patterned after its adult equivalent, the HMPS (Hewitt & Flett,
CAPS, and APS-R, we expect to identify a purely externally
motivated subgroup of maladaptive perfectionists. Also of interest 1991). Unlike its adult equivalent, the CAPS does not include the
is whether different perfectionist profiles differ across different Other-Oriented Perfectionism dimension, which assesses the inter-
grade and gender groups. nally motivated belief that it is important for others to strive for
perfection and be perfect. The CAPS comprises two dimensions:
Self-Oriented Perfectionism (12 items), and Socially Prescribed
Method Perfectionism (10 items). Three items require reversing (items 3, 9,
and 18). Participants respond to the items using a five-point scale
(where 1 ⫽ false—not at all true of me, and 5 ⫽ very true of me),
Participants
and higher scores reflect greater perfectionism. The CAPS ques-
Nine hundred thirty-eight high school students attending four tionnaire scores demonstrate good internal reliability, reporting
high schools in middle-class suburbs of a large Australian city Cronbach’s alpha levels of 0.85 and 0.86 for Self-Oriented and
participated. The sample comprised Grade 9 (18 male, 137 Socially Prescribed Perfectionism, respectively (Hewitt et al.,
female: M ⫽ 14.79 years, SD ⫽ 0.33), Grade 10 (93 male, 159 2002).
female: M ⫽ 15.67 years, SD ⫽ 1.04), Grade 11 (102 male, APS-R. The APS-R (Slaney et al., 2001) is a 23-item self-
147 female: M ⫽ 16.73 years, SD ⫽ 0.34), and Grade 12 (final report questionnaire designed to measure internally motivated per-
year of high school; 142 male, 140 female: M ⫽ 17.83 years, fectionism. Participants respond to questions on a seven-point
SD ⫽ 0.43) students. Of those students invited to participate, scale (1 ⫽ strongly disagree and 7 ⫽ strongly agree). The APS–R
approximately 70% did so. has three subscales: High Standards (seven items), Order (four
items), and Discrepancy (12 items). Principal components factor
analysis supports the three subscales (Slaney et al., 2001). The
Procedure reliability and validity of the APS–R questionnaire scores is ac-
Ethical approval to conduct the research was obtained from the ceptable. Internal consistency coefficients range from 0.85 to 0.92
authors’ university’s human research ethics committee and the (Slaney et al., 2001). Concurrent validity has been demonstrated
high schools’ research ethics authorities. Invitations for schools to through significant correlations with other perfectionism measures
participate in the study were sent to the head of school manage- and theoretically related constructs (Ashby & Rice, 2002; Slaney
ment, who nominated a senior staff member to liaise between the et al., 2001). Factor intercorrelations are appropriately small. The
school and research group. Separate student and guardian research correlation between High Standards and Order was 0.42, the
description statements and consent forms were distributed. Stu- correlation between High Standards and Discrepancy was 0.12,
dents were advised that participation was voluntary, and that
should they wish to withdraw at any stage, or to withdraw any 2
It should be noted that DiBartolo, Frost, Chang, LaSota, and Grills
unprocessed data, this would happen without prejudice. Students (2004) suggest that two items from the FMPS Personal Standards dimen-
were also advised that information collected was confidential and sion relate to self-worth (items 4 and 6) and should be removed; we
only accessible to researchers, subject to legal limitations. retained the full set of items, however.
PERFECTIONIST PROFILES IN ADOLESCENTS 1475

and the correlation between Order and Discrepancy was 0.03 and incrementally increases the number of latent classes until there
(Slaney et al., 2001). is no further improvement in the model fit.
DASS. The DASS (Lovibond & Lovibond, 1995) is a 32-item For latent class models, model fit is determined by fit indices
self-report questionnaire that comprises three scales: Depression, such as the log likelihood ratio (LL), the Akaike information
Anxiety, and Stress. Each scale contains 14 items. Participants are criterion (AIC), the Bayesian information criterion (BIC), and the
asked to use four-point severity/frequency scales to rate the extent sample-size adjusted BIC (aBIC). The smaller the value of the fit
to which they have experienced each emotional state over the past index, the better the model fit. Entropy values are also important
week (where 0 ⫽ did not apply to me at all and 3 ⫽ applied to me indicators of relative fit. A model’s entropy value describes the
very much, or most of the time). Factor analysis supports the level of classification uncertainty. Entropy values can range from
three-factor structure of the DASS (Antony, Bieling, Cox, Enns, & 0 to 1.0 and entropy values greater than 0.8 represent better fitting
Swinson, 1998). Internal consistency for the DASS questionnaire models—that is, models with more certainty of correct classifica-
scores is excellent, with Cronbach’s alphas of 0.97, 0.92, and 0.95 tion.
for the Depression, Anxiety, and Stress subscales, respectively
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(Antony et al., 1998). Concurrent validity of test scores with other


This document is copyrighted by the American Psychological Association or one of its allied publishers.

measures of anxiety and depression are moderate to high (Antony Results


et al., 1998). Descriptive data analyses were conducted using SPSS Version
21.0 (2012) and subsequent analyses conducted using MPlus Ver-
sion 7 (2012). Subscale scores were examined for outliers and
Statistical Analyses
issues with univariate normality. There were 77 extreme ratings (z
Dimensions of perfectionism. A combination of exploratory score ⬎3.29) identified from a possible 114,436 ratings, and 62 of
and confirmatory procedures was employed to identify latent di- these extreme ratings were from the same two subscale items (i.e.,
mensions of perfectionism from the (combined) FMPS, CAPS, and DASS Anxiety subscale items 23 and 36). As a consequence,
APS-R perfectionist questionnaires. An EFA was first conducted univariate outliers were not removed from the data set. Values for
and model fit statistics were used to help ascertain the best factor skewness (range ⫽ ⫺0.54 to 1.36) and kurtosis (range ⫽ ⫺0.44 to
structure (i.e., the best representative number of factors or dimen- 1.47) for all subscales were within acceptable ranges, as recom-
sions). A delta parameterization and robust adjusted weighted least mended by West, Finch, and Curran (1995). Multivariate normal-
square estimator were applied to a matrix of polychoric correla- ity was examined following procedures described by Tabachnick
tions; polychoric correlations were used to account for the ordinal and Fidell (2007). Based on Mahalanobis’ distance, 15 cases were
nature of the observed variables. We chose to use an oblique identified as multivariate outliers (p ⬍ .001) and removed from the
rotation (Geomin) because we anticipated factors would be corre- sample. Descriptive statistics are reported in Table 1. Internal
lated. Subsequently, a CFA was conducted with items retained consistency coefficients were good to excellent for all subscales
from the best EFA factor solution. (i.e., Cronbach’s alpha range ⫽ 0.82 to 0.94), except for the FMPS
For the factor analytic models, model fit was determined by the Doubts About Actions subscale, which showed relatively low test
chi-square difference test, the CFI, the Tucker-Lewis Index (TLI) score reliability (Cronbach’s alpha ⫽ .68). Correlations between
and the root mean square error of approximation (RMSEA). While all measures are reported in Table 2. (Please note that gender-
the chi-square difference test is commonly used, some question specific descriptive statistics and correlations are available online
whether this index should be used given its sensitivity to trivial in the supplemental material.)
deviations of fit and large sample sizes (Cheung & Rensvold, Correlations between dimensions provided evidence for concur-
2002). For the CFI and TLI, values greater than 0.95 are preferred, rent and divergent validity of test scores for the three perfectionism
but values close to 0.90 are considered acceptable. Cheung and questionnaires. For instance, very strong correlations emerged
Rensvold suggest looking at changes in the CFI, where changes of between the FMPS Personal Standards dimension, and the CAPS
0.01 or less suggest the invariance hypothesis should not be Self-Oriented Perfectionism and APS-R High Standards dimen-
rejected, changes between 0.01 and 0.02 suggest differences may sions (r ⫽ .75 and r ⫽ .82, respectively), suggesting that these
exist, and changes greater than 0.02 represent definite differences dimensions share overlapping variance. Similarly strong correla-
between models. For the RMSEA, values of below 0.05 are pre- tions emerged between the APS-R Discrepancy dimension, and the
ferred and values below 0.08 are considered acceptable. FMPS Concern Over Mistakes and FMPS Doubts About Actions
Latent classes of perfectionists. LCA was used to determine dimensions (r ⫽ .70 and r ⫽ .70, respectively). In contrast,
whether profiles of perfectionists could be identified based on generally lower correlations emerged between internally motivated
factors identified via exploratory and confirmatory factor analytic high standards dimensions such as the APS-R High Standards
techniques. LCA is a person-centered approach that groups indi- dimension, and performance concerns dimensions such as the
viduals into categories (latent classes) according to patterns of FMPS Concern Over Mistakes, FMPS Doubts About Actions, and
responses. We examined patterns of responses based on factor APS-R Discrepancy dimensions (r ⫽ .46, r ⫽ .26, and r ⫽ .37,
scores. Latent classes comprise individuals who are similar to each respectively).
other but who differ from individuals in other classes. Researchers EFA was conducted on each perfectionism questionnaire indi-
can examine models (i.e., different numbers of latent classes) and vidually to assess for test score validity in our Australian adoles-
compare model fit statistics. The aim is to identify the smallest cent sample. Findings support a four-factor solution for the FMPS,
number of latent classes that best describe associations among a two-factor solution for the CAPS, and a three-factor solution for
observed variables. The LCA process starts with a one-class model the APS-R in our 14- to 17-year-old high school students (see
1476 SIRONIC AND REEVE

Table 1
Descriptive Statistics for All Measures

Measure Minimum Maximum M SD Cronbach ␣

FMPS
Personal Standards 7.00 36.00 21.62 5.59 0.85
Concern Over Mistakes 9.00 43.00 19.32 6.87 0.88
Doubts About Actions 4.00 20.00 11.16 3.14 0.68
Parental Expectations 5.00 25.00 13.96 4.59 0.85
Parental Criticism 3.00 20.00 8.30 3.65 0.82
Organization 6.00 32.00 21.10 5.52 0.93
CAPS
Self-Oriented Perfectionism 12.00 60.00 36.45 8.16 0.87
Socially Prescribed Perfectionism 10.00 49.00 26.05 7.39 0.86
APS-R
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

High Standards 10.00 49.00 35.20 8.03 0.88


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Discrepancy 12.00 84.00 44.21 14.50 0.92


Order 4.00 28.00 18.21 4.99 0.83
DASS
Depression 0.00 42.00 9.05 9.71 0.94
Anxiety 0.00 42.00 8.66 8.15 0.90
Stress 0.00 42.00 12.73 9.66 0.93
Note. FMPS ⫽ Frost Multidimensional Perfectionism Scale; CAPS ⫽ Child and Adolescent Perfectionism
Scale; APS-R ⫽ Almost Perfect Scale—Revised; DASS ⫽ Depression Anxiety Stress Scales.

online supplemental material for details and discussion regarding to improve goodness-of-fit indices, solutions with greater than four
factor analysis for each individual perfectionism questionnaire). factors were difficult to interpret and inconsistent with findings
reported in the perfectionism literature. The four-factor solution
Factor Analytic Models of Perfectionism was consistent with previous findings (e.g., Rice et al., 2005) and
provided an acceptable fit of the data (CFI ⫽ 0.92, TLI ⫽ .92,
EFA. Items from the three perfectionism inventories were
RMSEA ⫽ 0.05; see online supplemental material for model fit
subjected to a joint EFA. To interpret the EFA, we first examined
eigenvalues and found that 11 were greater than one: 23.88, 10.10, statistics for alternative EFA factor solutions).
4.53, 4.32, 1.99, 1.61, 1.46, 1.25, 1.18, 1.14, and 1.10. The scree- Factor loadings for the four-factor solution are presented in the
test indicated that four factors should be retained. Parallel analysis online supplemental material. Factor 1 (labeled High Personal
indicated a seven-factor solution because the eighth eigenvalue Standards) comprised items from the FMPS Personal Standards,
(1.25) was lower than the 95th percentile of the eighth random CAPS Self-Oriented Perfectionism, and the APS-R High Stan-
eigenvalue (1.43). We tested models ranging from three- to eight- dards dimensions. Factor 2 (labeled Concerns, Doubts, and Dis-
factor solutions. While increasing the number of factors continued crepancy) comprised items from the FMPS Doubts About Actions,

Table 2
Bivariate Correlations Between Frost Multidimensional Perfectionism Scale (FMPS), Child and Adolescent Perfectionism Scale
(CAPS), Almost Perfect Scale—Revised (APS-R), and Depression Anxiety Stress Scales (DASS) Subscales

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. FMPS-PS —
2. FMPS-CM .56ⴱ —
3. FMPS-DA .34ⴱ .59ⴱ —
4. FMPS-PE .43ⴱ .49ⴱ .30ⴱ —
5. FMPS-PC .21ⴱ .49ⴱ .40ⴱ .71ⴱ —
6. FMPS-O .35ⴱ .10ⴱ .10ⴱ .05 ⫺.03 —
7. CAPS-SOP .75ⴱ .65ⴱ .39ⴱ .42ⴱ .26ⴱ .35ⴱ —
8. CAPS-SPP .40ⴱ .56ⴱ .40ⴱ .71ⴱ .64ⴱ .06 .46ⴱ —
9. APS-R-HS .82ⴱ .46ⴱ .26ⴱ .38ⴱ .13ⴱ .35ⴱ .73ⴱ .31ⴱ —
10. APS-R-Disc .38ⴱ .70ⴱ .70ⴱ .40ⴱ .52ⴱ .05 .48ⴱ .49ⴱ .37ⴱ —
11. APS-R-Ord .43ⴱ .18ⴱ .17ⴱ .12ⴱ .02 .86ⴱ .42ⴱ .12ⴱ .48ⴱ .14ⴱ —
12. DASS-D .12ⴱ .47ⴱ .45ⴱ .23ⴱ .36ⴱ ⫺.08ⴱ .19ⴱ .32ⴱ .08ⴱ .55ⴱ ⫺.04 —
13. DASS-A .15ⴱ .41ⴱ .44ⴱ .18ⴱ .31ⴱ .00 .21ⴱ .31ⴱ .10ⴱ .46ⴱ .03 .73ⴱ —
14. DASS-S .22ⴱ .47ⴱ .48ⴱ .22ⴱ .31ⴱ .01 .28ⴱ .34ⴱ .18ⴱ .53ⴱ .05 .79ⴱ .82ⴱ —
Note. PS ⫽ Personal Standards; CM ⫽ Concern Over Mistakes; DA ⫽ Doubts About Actions; PE ⫽ Parental Expectations; PC ⫽ Parental Criticism;
O ⫽ Organization; SOP ⫽ Self-Oriented Perfectionism; SPP ⫽ Socially Prescribed Perfectionism; HS ⫽ High Standards; Disc ⫽ Discrepancy; Ord ⫽
Order; D ⫽ Depression; A ⫽ Anxiety; S ⫽ Stress.

p ⬍ .01.
PERFECTIONIST PROFILES IN ADOLESCENTS 1477

FMPS Concern Over Mistakes, and APS-R Discrepancy dimen- to zero. Finally, Model 4 combined adjustments made in Model 2
sions. Factor 3 (labeled Externally Motivated Perfectionism) com- and Model 3. Table 3 presents model fit indices for the four
prised items from the FMPS Parental Criticism, FMPS Parental four-factor models. Model 4 showed the best fit to the data (CFI ⫽
Expectations, and CAPS Socially Prescribed Perfectionism dimen- 0.90, TLI ⫽ .90, RMSEA ⫽ 0.06), and changes in the CFI suggest
sions. Finally, Factor 4 (labeled Organization and Order) com- that there are differences between Model 4 and Models 1, 2 and 3
prised items from FMPS Organization and APS-R Order dimen- (⌬CFI ⫽ 0.07, 0.03, and 0.03, respectively). In Model 4, there
sions. All items showed factor loadings greater than 0.32 and were were no modification indices to suggest that error covariance
retained, as recommended by Costello and Osborne (2005). between items should be freed.
It should be noted that two items loaded most strongly onto
unexpected factors. Item FMPS-18 (“I hate being less than the best
at things”) from the FMPS Concern Over Mistakes subscale, and LCA and Perfectionist Profiles
item CAPS-19 (“I am always expected to do better than others”) To examine whether perfectionists could be classified into sub-
from the CAPS Socially Prescribed Perfectionism subscale, loaded groups, LCA was employed using factor means from the best
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

most strongly onto the High Personal Standards factor. Given the four-factor model, Model 4. Table 4 includes fit information (log
This document is copyrighted by the American Psychological Association or one of its allied publishers.

differences in cross loadings between actual and expected factors likelihood ratio, BIC, AIC, aBIC, and entropy values) for LCA
were small (i.e., 0.12 and 0.02, respectively), these items were models with two through to eight classes. While the AIC and aBIC
specified to load onto their expected factors in the subsequent model fit indices continued to reduce with increasing classes, the
CFA. smallest BIC value was observed for the six-class model. Given
Examination of individual items revealed a further anomaly— our focus was on a parsimonious and clinically meaningful model,
10 items (CAPS 10, CAPS 15, CAPS 19, CAPS 22, FMPS 13, and the reductions in aBIC and AIC values after the six-class
FMPS 14, FMPS 18, FMPS 22, FMPS 35, and APS-R 15) exhib- model were relatively small, we retained the six-class solution.
ited cross loadings of 0.50 or higher. Costello and Osborne (2005) Moreover, the six-class solution yielded adequately sized classes
suggest that items with cross-loadings of 0.50 or higher should be with good probabilities of class membership (see online supple-
dropped from analyses. This recommendation was explored in the mental material for details).
CFA models (i.e., Model 2 and Model 4, as discussed below). The six subgroups were differentiated according to mean factor
It should also be noted that not all FMPS Concern Over Mis- scores on each of the four higher-order perfectionism dimensions
takes items are internally motivated—items 21, 25, and 34 make (see Table 5 for subgroup profiles). Factor scores are not standard-
reference to other people’s expectations and reactions to mistakes. ized and should be interpreted relative to each other rather than
While FMPS items 21, 25, and 34 loaded onto the expected compared with a mean endorsement of each factor, as with stan-
Concerns, Doubts, and Discrepancy higher-order dimension, this dardized scores. The six subgroups are labeled herein as:
observation explains why the three items also showed high sec-
ondary loadings on the Externally Motivated Perfectionism dimen- 1. Adaptive Perfectionist (AP; n ⫽ 208; 93 male, 115
sion (0.48, 0.47, and 0.46, respectively). female): Students in this subgroup strive for perfection
CFA. CFA of the best-fitting EFA model (i.e., the four-factor and set high standards for themselves, but accept results
model) was conducted. In Model 1, all items were specified to load that are not perfect. This subgroup do not report external
on their respective factors and correlations between factors were criticism or external pressures to perform and rate them-
estimated freely. Three alternative four-factor models were also selves as valuing order and organization.
investigated. Model 2 followed recommendations made by
Costello and Osborne (2005) in which items with cross-loadings 2. Externally Motivated Maladaptive Perfectionist (EMMP;
greater than or equal to 0.50 were dropped. Model 3 took into n ⫽ 218; 110 male, 108 female): Students in this sub-
account correlations between factors in Model 1 (see online sup- group do not set high standards for themselves; rather,
plemental material for correlations). Results suggest that Factor 4 they report experiencing perfectionism prescribed by sig-
(Organization and Order) correlates relatively poorly with Factor 2 nificant others. In addition, they are dissatisfied with their
(Concerns, Doubts, and Discrepancy) and Factor 3 (Externally efforts and experience concern and doubts about their
Motivated Perfectionism; r ⫽ .09 and r ⫽ .06, respectively). These performance. This subgroup is not concerned about order
poor correlations were specified in Model 3— correlations were set or organization.

Table 3
Four Proposed Four-Factor Confirmatory Factor Analysis Models: Model Fit Indices Using Weighted Least Square Estimation
Applied to Polychoric Correlations

Four-factor models ␹2 df CFI ⌬CFI TLI RMSEA

Model 1 (baseline model) 16,924.67 3,074 0.83 — 0.83 0.07


Model 2 with 10 deleted items 12,019.87 2,339 0.87 0.04 0.86 0.07
Model 3 with two factor correlations set to zero 13,438.14 3,076 0.87 0.00 0.87 0.06
Model 4 with 10 deleted items and two factor correlations set to zero 9,480.08 2,341 0.90 0.03 0.90 0.06
Note. ␹2 ⫽ chi-square difference test; df ⫽ degrees of freedom; CFI ⫽ Comparative Fit Index; TLI ⫽ Tucker-Lewis Index; RMSEA ⫽ root mean square
error of approximation.
1478 SIRONIC AND REEVE

Table 4
Summary of Fit Indices for Latent Class Models

Number of classes LL AIC BIC aBIC npar Entropy

2 ⫺4,183.32 8,392.64 8,455.40 8,414.12 13 0.66


3 ⫺4,078.01 8,192.02 8,278.92 8,221.75 18 0.73
4 ⫺4,040.22 8,126.45 8,237.48 8,164.44 23 0.70
5 ⫺3,980.95 8,017.90 8,153.07 8,064.15 28 0.71
6 ⫺3,960.29 7,986.57 8,145.89 8,041.09 33 0.72
7 ⫺3,945.75 7,967.50 8,150.95 8,030.26 38 0.70
8 ⫺3,931.83 7,949.67 8,157.25 8,020.69 43 0.71
Note. LL ⫽ log likelihood ratio; AIC ⫽ Akaike information criterion; BIC ⫽ Bayesian information criterion; aBIC ⫽ sample-size adjusted BIC; npar ⫽
number of parameters.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

3. Mixed Maladaptive Perfectionist (MMP; n ⫽ 61; 26 (We also include means and standard deviations for each subgroup
male, 35 female): Students in this subgroup endorse on the original FMPS, CAPS, and APS-R perfectionism dimen-
the highest ratings on all four factors compared with sions in the online supplemental material).
other subgroups. Mixed maladaptive perfectionists are
motivated by a mixture of their own high standards and Grade and Gender as Predictors of Class Membership
the expectations of others. They also experience
We used a multinomial logistical regression analyses with latent
doubts and concerns about their performance and feel
variables to investigate whether grade and gender predicted class
distressed when expectations are not met. This sub-
membership. We utilized the three-step modeling procedure (As-
group also expressed a strong desire for order and
parouhov & Muthén, 2013; Feingold, Tiberio, & Capaldi, 2014;
organization.
Vermunt, 2010). The three-step procedure first conducts LCA
4. Order and Organization Nonperfectionist (OONP; n ⫽ without adding covariates. In the second step, it proceeds to
27; 2 male, 25 female): Students in this subgroup do identify the most likely class memberships from posterior proba-
not set high standards for themselves, nor do they feel bilities of the LCA and classification uncertainty rates. In the third
that others impose high standards on them; they are step, the most likely class membership variables are analyzed
not dissatisfied with their efforts or concerned about together with auxiliary variables (grade or gender in this case)
their performance. However, students in this group while accounting for the measurement error in classification. As
rate themselves as strongly valuing order and organi- with all forms of multinomial logistical regression analyses, one
zation. group is designated the reference group. The analysis then
compares the probability of an observation belonging to each
5. Nonperfectionist A (NPA; n ⫽ 352; 108 male, 244 nonreference group (compared with the reference group; e.g.,
female): Students in this subgroup exhibit low ratings six latent classes; AP subgroup as the reference group vs. five
on each of the four factors. They do not set high other subgroups as the nonreference groups) across levels of the
standards for themselves or feel that others impose independent variable (e.g., gender; male reference group vs.
high standards on them, and they are not concerned female nonreference group). Analyses yield regression coeffi-
about their performance or dissatisfied with their ef- cients and odds ratios, which are used to establish predictor
forts. They also rate themselves as being unconcerned effects.
by order and organization. A complete set of multinomial logistical regression analyses
was conducted to examine whether grade predicted subgroup
6. Nonperfectionist B (NPB; n ⫽ 57; 10 male, 47 fe- membership. There were no significant predictor effects (p ⬍
male): Students in this subgroup have a very similar .05) for grade across all levels of grade. We also examined
profile to NPA. Their ratings are significantly lower whether gender predicted latent class membership. A summary
than ratings made by the NPA subgroup, however. of these analyses is presented in Table 6. Findings suggest that

Table 5
Factor Score Means and Standard Deviations for the Four Factors in the Six-Class Solution

High Personal Concerns, Doubts, Externally Motivated


Variable Standards and Discrepancy Perfectionism Order and Organization

Adaptive perfectionist 0.45 (0.09) ⫺0.03 (0.08) 0.03 (0.07) 0.64 (0.12)
Externally motivated maladaptive perfectionist 0.32 (0.09) 0.67 (0.08) 0.64 (0.07) ⫺0.30 (0.08)
Mixed maladaptive perfectionist 1.42 (0.11) 0.99 (0.12) 0.95 (0.11) 1.06 (0.18)
Order and organization nonperfectionist 0.17 (0.35) ⫺1.02 (0.24) ⫺.088 (0.14) 0.99 (0.38)
Nonperfectionist A ⫺0.50 (0.08) ⫺0.26 (0.08) ⫺0.30 (0.08) ⫺0.36 (0.08)
Nonperfectionist B ⫺1.47 (0.12) ⫺1.23 (0.18) ⫺1.12 (0.12) ⫺0.79 (0.15)
PERFECTIONIST PROFILES IN ADOLESCENTS 1479

Table 6
Gender (Males as Reference Group) as a Predictor of Membership in the Six Latent Classes With Each Class as a Reference Group
(Featuring Regression Coefficients and Odds Ratios With Significant Predictor Effects [p ⬍ .05] Only)

Externally motivated
Gender (males as Adaptive maladaptive Mixed maladaptive Order and organization Nonperfectionist Nonperfectionist
reference group) perfectionist perfectionist perfectionist Nonperfectionist A B

Regression coefficient Reference group ns ns 15.54 0.77 1.66


Odds ratio 5.61 ⫻ 107 2.16 5.26
Regression coefficient Reference group ns 15.80 1.03 1.92
Odds ratio 7.28 ⫻ 107 2.80 6.82
Regression coefficient Reference group 15.28 ns 1.40
Odds ratio 4.33 ⫻ 107 4.06
Regression coefficient Reference group ⫺14.77 ⫺13.88
Odds ratio 3.85 ⫻ 10–7 9.38 ⫻ 10–7
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Regression coefficient Reference group ns


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Odds ratio
Regression coefficient Reference group
Odds ratio
Note. ns ⫽ gender is a nonsignificant predictor of class membership for the specified reference group.

females were more likely (compared with males) to be in the When examining differences (p ⬍ .05), the MMP and EMMP
OONP and NPB subgroups compared with the AP, EMMP, or subgroups consistently reported significantly higher levels of de-
MMP subgroups. Females were also more likely (compared pression, anxiety, and stress than the AP and three nonperfectionist
with males) to be in the NPA subgroup compared with the AP subgroups, providing support for their “maladaptive” label. While
or EMMP subgroups, but less likely (compared with males) to the MMP and EMMP subgroups did not significantly differ in their
be in the NPA or NPB subgroups compared with the OONP depression and anxiety ratings, students in the MMP subgroup did
subgroup. rate themselves as experiencing significantly higher levels of stress
compared with students in the EMMP subgroup. The AP and NPA
Differences in Psychopathology subgroups had significantly higher ratings on depression, anxiety
and stress than both the NPB and OONP subgroups. No difference
To determine whether the perfectionist subgroups differed in was observed between the AP and NPA subgroups in their depres-
depression, anxiety and stress ratings on the DASS, we used the sion, anxiety and stress ratings.
auxiliary analysis option in MPlus to predict subgroup member-
ship. This analytic procedure involves a series of pairwise Wald
Discussion
tests to assess the equality of means (or otherwise) across latent
classes (subgroups) based on posterior probabilities (Asparouhov, The aim of the present study was to determine whether mean-
2007). This is a conservative approach to analyzing multiple ingfully different perfectionism subgroups could be identified on
predictors in mixture models. Table 7 presents means and standard the basis of a joint analysis of the FMPS, CAPS, and APS-R
deviations for the DASS for each of the six subgroups, including perfectionism questionnaires. Six findings are of note. First, factor
pairwise contrasts (p ⬍ .05). The same pattern of findings was analyses revealed commonly observed factor structures for each
found for ratings on depression, anxiety, and stress across the six perfectionism questionnaire in a large adolescent sample (n ⫽
subgroups, namely, MMP ⬎ EMMP ⬎ AP ⬎ NPA ⬎ NPB ⬎ 923). Second, when items from all questionnaires were included in
OONP. a single EFA, four factors emerged: High Personal Standards;

Table 7
Means and Standard Errors on the Depression Anxiety Stress Scales (DASS) for the Six-Class Solution, Including Pairwise Contrasts
(p ⬍ .05)

Externally
motivated Mixed Order and
Adaptive maladaptive maladaptive organization
perfectionist perfectionist perfectionist nonperfectionist Nonperfectionist Nonperfectionist
(N ⫽ 208) (N ⫽ 218) (N ⫽ 61) (N ⫽ 27) A (N ⫽ 352) B (N ⫽ 57)
DASS M SE M SE M SE M SE M SE M SE

Depression 6.93a 0.61 14.51b 0.85 15.74b 1.65 2.80c 1.04 7.09a 0.50 4.42c 0.81
Anxiety 7.64a 0.56 12.26b 0.70 13.87b 1.38 4.07c 1.17 7.13a 0.44 5.03c 0.71
Stress 11.60a 0.67 17.31b 0.77 21.29c 1.46 7.34d 1.42 10.37a 0.52 7.45d 1.07
Note. Means that were significantly different (p ⬍ .05) are based on a series of pairwise Wald tests are indicated by different lettered subscripts.
1480 SIRONIC AND REEVE

Concerns, Doubts, and Discrepancy; Externally Motivated Perfec- The MMP subgroup has been identified previously (Boone et
tionism; and Order and Organization. Third, CFA identified the al., 2010; Rice & Ashby, 2007; Rice et al., 2011; Sironic & Reeve,
best fitting four-factor solution. This four-factor solution is inter- 2012; Wang et al., 2007). Individuals in the MMP subgroup
esting because it supports a more differentiated view of maladap- experience both internally and externally motivated perfection-
tive perfectionism. Our findings suggest that Externally Motivated ism—they set high standards for themselves and adopt high stan-
Perfectionism is an important factor in its own right, and not an dards that they perceive as imposed by others. They also experi-
aspect of a broader “Maladaptive Evaluative Concerns” factor. ence a discrepancy between the standards they set and achieved
Fourth, LCA of participants’ mean factor scores on the four outcomes, show concerns and doubts about their performance, and
higher-order perfectionism dimensions identified a six-class solu- value order and organization.
tion, representing three perfectionist profiles and three nonperfec- The MMP and EMMP subgroups report significantly higher
tionist profiles. We labeled the six subgroups: (1) AP, (2) EMMP, levels of self-reported depression, anxiety, and stress compared
(3) MMP, (4) OONP, (5) NPA, and (6) NPB. Fifth, the six with the AP and nonperfectionist subgroups. It is interesting that
subgroups were related to self-reported depression, anxiety, and while the MMP and EMMP subgroups did not differ in their
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

stress symptoms in expected ways. Sixth, perfectionist subgroups depression and anxiety ratings, the MMP subgroup reported sig-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

did not vary as a function of grade, but did vary as a function of nificantly greater levels of stress than the EMMP subgroup. This
gender. The findings suggest that females were more likely (than finding has implications for the MMP subgroup’s need for stress
males) to belong to the OONP and NPB subgroups compared with counseling and appears fitting given diathesis-stress models of
the three perfectionist subgroups; females were also more likely perfectionism, which maintain that perfectionists exposed to stress
(than males) to be in the NPA subgroup compared with the AP and are vulnerable to psychological distress (Chang & Rand, 2000;
EMMP subgroups but less likely (than males) to be in the NPA and Cheng, 2001; Enns, Cox, & Clara, 2005; Flett, Hewitt, Blankstein,
NPB subgroups compared with the OONP subgroup. In sum, it is & Mosher, 1995; O’Connor, Rasmussen, & Hawton, 2010).
evident that meaningful perfectionist subgroups can be identified Although some may argue that the MMP subgroup experience a
from combinations of higher-order perfectionism dimensions de- more extreme form of perfectionism simply because of their higher
rived from the FMPS, CAPS, and APS-R. ratings on the High Personal Standards factor, it is important to
The AP profile has been identified previously (Boone, Soenens, consider that students in the MMP subgroup rated themselves higher
Braet, & Goossens, 2010; Rice & Ashby, 2007; Rice et al., 2011; not only on the High Personal Standards factor but on all four
Sironic & Reeve, 2012; Wang et al., 2007). Students in the AP perfectionism factors, compared with other perfectionist subgroups. A
subgroup set high standards for themselves and are mostly satis- person-centered approach suggests that ratings on one dimension may
fied by their efforts. As expected, the AP subgroup reported lower have a different overall meaning depending on how they relate to
levels of depression, anxiety and stress, compared with the two ratings on other dimensions. Indeed, students in the EMMP subgroup
other perfectionist subgroups (Rice et al., 2011; Sironic & Reeve, rated themselves as experiencing the same levels of depression and
2012; Wang et al., 2007). Moreover, there were no self-reported anxiety as students in the MMP subgroup, despite having lower
depression, anxiety, or stress symptom differences between the AP self-ratings on each of the four higher order perfectionism dimen-
and NPA subgroups. sions. We suggest that it is the particular overlap of (or pattern of
The EMMP profile is characterized by individuals who do not scores on) higher-order perfectionism dimensions that make maladap-
set high standards for themselves, but still experience high levels tive perfectionists “maladaptive” in nature and more likely to expe-
of discrepancy between their expectations and performance. This rience psychopathology, compared with the AP and nonperfectionist
perfectionist profile is somewhat controversial. Some researchers perfectionist subgroups. While ratings on individual perfectionism
who have analyzed the APS-R have questioned whether a low dimensions are important, they may be misleading (in understanding
Personal Standards/high Discrepancy profile represents a perfec- perfectionists) when viewed in isolation. This is not to suggest that
tionist subgroup per se because its members do not describe research following a variable-centered approach, which examines
themselves as possessing high standards (Rice & Ashby, 2007; perfectionism dimensions in isolation, is not important. Indeed, we
Rice et al., 2011). However, other researchers have argued that acknowledge that perfectionism dimensions are important in their
these individuals may lack insight into their perfectionism (Sironic own right and helpful in understanding adaptive and maladaptive
& Reeve, 2012) or experience externally motivated high standards forms of perfectionism (see Mitchell et al., 2013; Nobel, Manassis, &
(Wang et al., 2007). By including measures of externally moti- Wilansky-Trainor, 2012). Nevertheless, a definition of perfectionists
vated perfectionism from the FMPS and the CAPS, we were able contingent on high ratings on one dimension (e.g., High Personal
to examine the relationship between APS-R perfectionist profiles Standards) is inconsistent with a person-centered conceptualization.
(e.g., low Personal Standards/high Discrepancy) and externally Our findings support the work of Lundh et al. (2008). Lundh et
motivated perfectionism. Our findings support the claim that the al. supported a person-centered approach (compared with a
APS-R low Personal Standards/high Discrepancy subgroup is an variable-centered approach) by identifying three clinically signif-
externally motivated perfectionist subgroup (represented in this icant perfectionist subgroups, which were characterized by high
study by the EMMP subgroup)—they do not set high standards for ratings on the FMPS Concern Over Mistakes, Doubts About Ac-
themselves but likely adopt the standards of parents, teachers, and tions, Personal Standards, and Organization dimensions. These
significant others, and experience concerns, doubts, and discrep- three groups are similar to our MMP subgroup in that they show
ancy based on the standards set by others. It should be noted, high internally motivated standards and experience concerns and
however, that students in the EMMP subgroup do not endorse doubts about their performance. It is interesting that two of Lundh
greater levels of externally motivated perfectionism than students et al.’s three clinically significant subgroups showed elevated
in the MMP subgroup. ratings on the FMPS Parental Criticism and Parental Expectations
PERFECTIONIST PROFILES IN ADOLESCENTS 1481

dimensions, while the remaining third subgroup did not. The two day-to-day tasks of meeting and evaluating standards may relate to
subgroups with higher ratings on the Parental Criticism and Pa- how they value order and organization. Moreover, the Order and
rental Expectations dimensions reported greater depression than Organization higher-order dimension helps differentiate between
the subgroup that rated themselves as low on Parental Criticism the three perfectionist subgroups. Students in the MMP subgroup
and Parental Expectations dimensions. This supports our claim that rate themselves highest on the Order and Organization dimension
externally motivated perfectionism may be important in under- followed by students in the AP and then EMMP subgroups. These
standing perfectionism-related psychopathology. It should also be findings suggest that organization and order dimensions in perfec-
noted that Lundh et al. briefly refer to a fourth possible maladap- tionism may warrant further consideration.
tive perfectionist subgroup, which they deemed as clinically sig- Furthermore, some suggest that interpersonal dimensions of
nificant. Individuals in this subgroup rated themselves as high on perfectionism (e.g., FMPS Parental Criticism, FMPS Parental Ex-
the Parental Criticism and Parental Expectations dimensions, mod- pectations, HMPS Socially Prescribed Perfectionism) may be bet-
erate on the Concern Over Mistakes dimension, and relatively low ter conceptualized as developmental antecedents or correlates of
on the Personal Standards dimension, compared with the other perfectionism and should not be included in perfectionism-related
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

clinically significant perfectionist subgroups. Lundh et al. did not analyses (Rice et al., 2005; Stoeber & Otto, 2006). However, it
This document is copyrighted by the American Psychological Association or one of its allied publishers.

consider this subgroup to be a perfectionist subgroup because its would seem unwise to disregard measures of interpersonal influ-
members showed no evidence of elevated self-motivated personal ence that may be associated with the development of perfection-
standards. We suggest that this fourth subgroup is similar to our ism, especially in adolescence. Moreover, our findings suggest that
EMMP subgroup, in which perfectionism is externally motivated. there is a group of perfectionists that is mostly externally moti-
The data show that approximately 50% of adolescents are clas- vated (i.e., individuals who experience relatively little internally
sified as perfectionists. While adaptive and maladaptive perfec- motivated perfectionism). This group may be unique to an adoles-
tionists are both highly prevalent, our findings show that adoles- cent sample, given the attention adolescents pay to parents and
cents appear more likely to possess a maladaptive form of significant others at this developmental stage.
perfectionism and be at an increased risk of experiencing psycho- While our findings suggest that perfectionism subgroups remain
pathology. Recent research, for example, suggests that maladap- stable across the high school years, caution should be exercised in
tive perfectionists may be more at risk of self-harm or suicide
drawing applied inferences about these subgroups from our cross-
(Hewitt, Caelian, Chen, & Flett, 2014). It is evident that more
sectional data. Adolescence is a unique developmental period
research is needed to understand the health and wellbeing of
affected by multiple influences, and longitudinal research is re-
adolescents with maladaptive perfectionism. In addition, given the
quired to understand the stability and/or change in subgroup mem-
known risk factors associated with perfectionisms per se, early
bership over time. It is worth noting that several studies have
intervention programs that identify and treat adolescent perfection-
assessed perfectionism longitudinally and suggest that it is rela-
ism would seem critical. It should be noted that the EMMP
tively stable (Nilsson, Sundbom, & Hägglöf, 2008; O’Connor et
subgroup (n ⫽ 218) is much larger than the MMP subgroup (n ⫽
al., 2009; Rice & Aldea, 2006). Nevertheless, a better understand-
61) that experiences both internally- and externally motivated
ing of the stability (or otherwise) of “externally motivated mal-
perfectionism. We suggest that this pattern of findings is evidence
adaptive” adolescent perfectionists as they become adults may
for the importance of significant others (e.g., parents, teachers,
peers) in the life of adolescents. The finding has implications for clarify the status of this particular perfectionist subgroup.
understanding how negative forms of perfectionism are manifest. It is also worth noting that perfectionist profiles similar to those
From a clinical or a counseling perspective, maladaptive perfec- identified in the present study occur across cultures (Boone et al.,
tionism in the adolescent period may require a broad based inter- 2010; Rice et al., 2011; Sironic & Reeve, 2012; Wang et al., 2007).
vention that involves not only students, but also parents, teachers, The latter point is particularly relevant in light of Wang et al.’s
and significant others. suggestion that “externally motivated maladaptive perfectionists”
The NPA and NPB subgroups exhibited low ratings on each of might be unique to collectivist cultures, such as Taiwan. On the basis
the four perfectionism dimensions and, the NPB subgroup, with of our findings it seems likely that the EMMP profile is common in
the OONP subgroup, reported the lowest ratings of psychopathol- Australia, Belgium, the United States, and Taiwan, at least.
ogy. Individuals with an OONP profile score high on Organization In the present study, we used self-ratings to assess perfectionism
and Order and low on remaining higher-order dimensions. It is and psychopathology. Even though self-reports are commonly
interesting to note that the OONP and MMP profiles were com- used in perfectionism and clinical/counseling research, the relative
parable in their scores on the Organization and Order dimension. absence of more direct observational measures is worrying (Her-
While some consider order and organization central to perfection- man et al., 2013). In particular, it would seem important to inves-
ism, others disagree (Frost et al., 1990; Rice & Asbhy, 2007; tigate the degree to which perceptions of externally motivated
Stoeber & Otto, 2006). Indeed, some researchers do not include the perfectionists actually reflect the behavior/pressures of significant
APS-R Order and FMPS Organization dimensions in their analy- others. Multiple informant report methods (e.g., parent-report,
ses of perfectionism (Egan, Wade, & Shafran, 2011; Rice & teacher-report, clinical interview) would be helpful in this regard.
Ashby, 2007), arguing that it does not form a core dimension of We also note that even though different psychopathologies (i.e.,
perfectionism. We included all dimensions from the FMPS, CAPS, depression, anxiety, and stress) are associated with different sub-
and APS-R in our analyses, which show that the Order and groups, this does not imply a causal link between perfectionist
Organization factor is significantly correlated with the High Per- subgroup membership and different psychopathologies. In addi-
sonal Standards and Concerns, Doubts, and Discrepancy factors. It tion, we acknowledge that while perfectionism questionnaires
seems likely that the way in which perfectionists tackle their were designed to identify perfectionism, other psychological fac-
1482 SIRONIC AND REEVE

tors or pathologies may, of course, be at play and assessed by & Development, 80, 197–203. http://dx.doi.org/10.1002/j.1556-6678
questionnaire items. We understand there are no “pure” test items. .2002.tb00183.x
The three nonperfectionist subgroups, including the relatively small Asparouhov, T. (2007). Wald test of mean equality for potential latent class
OONP subgroup (comprising only 27 individuals), may also be con- predictors in mixture modeling. Technical appendix. In Mplus user’s
sidered a limitation of the current study. We believe it is important to guide (6th ed.). Los Angeles, CA: Muthén & Muthén. Retrieved from
https://www.statmodel.com/download/usersguide/Mplus%20Users%
retain these three nonperfectionist subgroups because they emerged
20Guide%20v6.pdf
from our analysis, and an individual in any one of these groups is Asparouhov, T., & Muthén, B. O. (2013). Auxiliary variables in mixture
deemed to have a different pattern of scores (across the four higher modeling: A 3-step approach using Mplus (Mplus Web Notes: No. 15,
order perfectionism dimensions) compared with an individual from Version 6). Retrieved from http://statmodel.com/examples/webnotes/
any of the other (five) subgroups. We were reluctant to collapse AuxMixture_submitted_corrected_webnote
subgroups. Moreover, there were differences between nonperfection- Boone, L., Soenens, B., Braet, C., & Goossens, L. (2010). An empirical
ist subgroups. The NPA subgroup showed higher levels of depression, typology of perfectionism in early-to-mid adolescents and its relation
anxiety, and stress than the NPB and OONP subgroups (and similar with eating disorder symptoms. Behaviour Research and Therapy, 48,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

levels of depression, anxiety, and stress when compared with the AP 686 – 691. http://dx.doi.org/10.1016/j.brat.2010.03.022
This document is copyrighted by the American Psychological Association or one of its allied publishers.

subgroup). From a research information perspective, we suggest that Chang, E. C., & Rand, K. L. (2000). Perfectionism as a predictor of
is possible that similar subgroups will emerge in other research and subsequent adjustment: Evidence for a specific diathesis–stress mecha-
nism among college students. Journal of Counseling Psychology, 47,
our findings will provide comparison information. Moreover, model
129 –137. http://dx.doi.org/10.1037/0022-0167.47.1.129
fit statistics suggest that our six-subgroup model is acceptable. We do
Cheng, S. K. (2001). Life stress, problem solving, perfectionism, and
not, however, claim that our model cannot be improved. We encour- depressive symptoms in Chinese. Cognitive Therapy and Research, 25,
age future research to adopt a person-centered approach and identify 303–310. http://dx.doi.org/10.1023/A:1010788513083
improved models of perfectionists using various perfectionism dimen- Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit
sions. indexes for testing measurement invariance. Structural Equation Mod-
Future research with adults may also choose to include the eling, 9, 233–255. http://dx.doi.org/10.1207/S15328007SEM0902_5
HMPS Other-Oriented Perfectionism dimension, which is not part Costello, A. B., & Osborne, J. W. (2005). Best practices in exploratory
of the child and adolescent version of the HMPS (i.e., the CAPS). factor analysis: Four recommendations for getting the most from your
Recent research suggests that Other-Oriented Perfectionism is an analysis. Practical Assessment, Research & Evaluation, 10, 1–9.
important form of perfectionism associated with antisocial and DiBartolo, P. M., Frost, R. O., Chang, P., LaSota, M., & Grills, A. E.
narcissistic personality characteristics (Sherry, Gralnick, Hewitt, (2004). Shedding light on the relationship between personal standards
and psychopathology: The case for contingent self-worth. Journal of
Sherry, & Flett, 2014; Stoeber, 2014). The inclusion on this
Rational-Emotive & Cognitive-Behavior Therapy, 22, 241–254.
perfectionism dimension in future person-centered perfectionism
Egan, S. J., Wade, T. D., & Shafran, R. (2011). Perfectionism as a
research may lead to the identification of another perfectionist transdiagnostic process: A clinical review. Clinical Psychology Review,
subgroup, in which members expect others to be perfect, and are 31, 203–212. http://dx.doi.org/10.1016/j.cpr.2010.04.009
highly critical of those who fail to meet their expectations. Einstein, D. A., Lovibond, P. F., & Gaston, J. E. (2000). Relationship
between perfectionism and emotional symptoms in an adolescent sam-
Conclusion ple. Australian Journal of Psychology, 52, 89 –93. http://dx.doi.org/
10.1080/00049530008255373
In the present study, we analyzed differences and similarities in Enns, M. W., Cox, B. J., & Clara, I. P. (2005). Perfectionism and neurot-
the dimensional constructs of three widely used perfectionism icism: A longitudinal study of specific vulnerability and diathesis-stress
questionnaires to provide a broader understanding of perfection- models. Cognitive Therapy and Research, 29, 463– 482. http://dx.doi
ism. Consistent with previous research, we identified three perfec- .org/10.1007/s10608-005-2843-04
tionist subgroups (i.e., AP, EMMP, and MMP subgroups), and Feingold, A., Tiberio, S. S., & Capaldi, D. M. (2014). New approaches for
subgroup membership was meaningfully related to self-reported examining associations with latent categorical variables: Applications to
substance abuse and aggression. Psychology of Addictive Behaviors, 28,
depression, anxiety, and stress. The finding of two maladaptive
257–267. http://dx.doi.org/10.1037/a0031487
perfectionist subgroups experiencing externally motivated perfec-
Flett, G. L., Hewitt, P. L., Blankstein, K. R., & Mosher, S. W. (1995).
tionism (30% of our entire sample, and 56% of all perfectionists) Perfectionism, life events, and depressive symptoms: A test of a
suggests that externally motivated perfectionism is common in diathesis-stress model. Current Psychology, 14, 112–137. http://dx.doi
adolescence and may require treatment regimens that involve not .org/10.1007/BF02686885
only students, but also parents, teachers, and significant others. Flett, G. L., Hewitt, P. L., Boucher, D. J., Davidson, L. A., & Munro, Y.
Whether these students remain externally motivated perfectionists, (2000). The Child–Adolescent Perfectionism Scale: Development, vali-
or indeed perfectionists, over time, is a matter for further research. dation, and association with adjustment. Unpublished manuscript, York
University, Toronto, Ontario, Canada.
Flett, G. L., Hewitt, P. L., Oliver, J. M., & Macdonald, S. (2002). Perfec-
References
tionism in children and their parents: A developmental analysis. In G. L.
Antony, M. M., Bieling, P. J., Cox, B. J., Enns, M. W., & Swinson, R. P. Flett & P. L. Hewitt (Eds.), Perfectionism: Theory, research and treat-
(1998). Psychometric properties of the 42-item and 21-item versions of ment (pp. 89 –132). Washington, DC: American Psychological Associ-
the Depression Anxiety Stress Scales (DASS) in clinical groups and a ation. http://dx.doi.org/10.1037/10458-004
community sample. Psychological Assessment, 10, 176 –181. http://dx Frost, R. O., Heimberg, R. G., Holt, J. I., Mattia, J. I., & Neubauer, A. L.
.doi.org/10.1037/1040-3590.10.2.176 (1993). A comparison of two measures of perfectionism. Personality
Ashby, J. S., & Rice, K. G. (2002). Perfectionism, dysfunctional attitudes, and Individual Differences, 14, 119 –126. http://dx.doi.org/10.1016/
and self-esteem: A structural equations analysis. Journal of Counseling 0191-8869(93)90181-2
PERFECTIONIST PROFILES IN ADOLESCENTS 1483

Frost, R. O., Marten, P., Lahart, C., & Rosenblate, R. (1990). The dimen- Rice, K. G., & Aldea, M. A. (2006). State dependence and trait stability of
sions of perfectionism. Cognitive Therapy and Research, 14, 449 – 468. perfectionism: A short-term longitudinal study. Journal of Counseling
http://dx.doi.org/10.1007/BF01172967 Psychology, 53, 205–213. http://dx.doi.org/10.1037/0022-0167.53.2.205
Gaudreau, P., & Thompson, A. (2010). Testing a 2 ⫻ 2 model of dispo- Rice, K. G., & Ashby, J. S. (2007). An efficient method for classifying
sitional perfectionism. Personality and Individual Differences, 48, 532– perfectionists. Journal of Counseling Psychology, 54, 72– 85. http://dx
537. http://dx.doi.org/10.1016/j.paid.2009.11.031 .doi.org/10.1037/0022-0167.54.1.72
Hawkins, C. C., Watt, H. M. G., & Sinclair, K. E. (2006). Psychometric Rice, K. G., Ashby, J. S., & Gilman, R. (2011). Classifying adolescent
properties of the Frost Multidimensional Perfectionism Scale with Aus- perfectionists. Psychological Assessment, 23, 563–577. http://dx.doi.org/
tralian adolescent girls: Clarification of multidimensionality and perfec- 10.1037/a0022482
tionist typology. Educational and Psychological Measurement, 66, Rice, K. G., Ashby, J. S., & Slaney, R. B. (1998). Self-esteem as a mediator
1001–1022. http://dx.doi.org/10.1177/0013164405285909 between perfectionism and depression: A structural equations analysis.
Herman, K. C., Wang, K., Trotter, R., Reinke, W. M., & Ialongo, N. Journal of Counseling Psychology, 45, 304 –314. http://dx.doi.org/
10.1037/0022-0167.45.3.304
(2013). Developmental trajectories of maladaptive perfectionism among
Rice, K. G., & Dellwo, J. P. (2002). Perfectionism and self-development:
African American adolescents. Child Development, 84, 1633–1650.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Implications for college adjustment. Journal of Counseling & Develop-


http://dx.doi.org/10.1111/cdev.12078
ment, 80, 188 –196. http://dx.doi.org/10.1002/j.1556-6678.2002
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Hewitt, P. L., Caelian, C. F., Chen, C., & Flett, G. L. (2014). Perfectionism,
.tb00182.x
stress, daily hassles, hopelessness, and suicide potential in depressed
Rice, K. G., Lopez, F. G., & Vergara, D. (2005). Parental/social influences
psychiatric adolescents. Journal of Psychopathology and Behavioral
on perfectionism and adult attachment orientations. Journal of Social
Assessment, 36, 663– 674. http://dx.doi.org/10.1007/s10862-014-9427-0 and Clinical Psychology, 24, 580 – 605. http://dx.doi.org/10.1521/jscp
Hewitt, P. L., Caelian, C. F., Flett, G. L., Sherry, S. B., Collins, L., & .2005.24.4.580
Flynn, C. A. (2002). Perfectionism in children: Associations with de- Rice, K. G., & Mirzadeh, S. A. (2000). Perfectionism, attachment, and
pression, anxiety and anger. Personality and Individual Differences, 32, adjustment. Journal of Counseling Psychology, 47, 238 –250. http://dx
1049 –1061. http://dx.doi.org/10.1016/S0191-8869(01)00109-X .doi.org/10.1037/0022-0167.47.2.238
Hewitt, P. L., & Flett, G. L. (1991). Perfectionism in the self and social Shafran, R., & Mansell, W. (2001). Perfectionism and psychopathology: A
contexts: Conceptualization, assessment, and association with psycho- review of research and treatment. Clinical Psychology Review, 21,
pathology. Journal of Personality and Social Psychology, 60, 456 – 470. 879 –906. http://dx.doi.org/10.1016/S0272-7358(00)00072-6
http://dx.doi.org/10.1037/0022-3514.60.3.456 Sherry, S. B., Gralnick, T. M., Hewitt, P. L., Sherry, D. L., & Flett, G. L.
Huggins, L., Davis, M. C., Rooney, R., & Kane, R. (2008). Socially (2014). Perfectionism and narcissism: Testing unique relationships and
prescribed and self-oriented perfectionism as predictors of depressive gender differences. Personality and Individual Differences, 61, 52–56.
diagnosis in preadolescents. Australian Journal of Guidance and Coun- http://dx.doi.org/10.1016/j.paid.2014.01.007
selling, 18, 182–194. http://dx.doi.org/10.1375/ajgc.18.2.182 Sironic, A., & Reeve, R. A. (2012). More evidence for four perfectionism
Kornblum, M., & Ainley, M. (2005). Perfectionism and the gifted: A study subgroups. Personality and Individual Differences, 53, 437– 442. http://
of an Australian school sample. International Education Journal, 6, dx.doi.org/10.1016/j.paid.2012.04.003
232–239. Slaney, R. B., Rice, K. G., Mobley, M., Trippi, J., & Ashby, J. S. (2001).
Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the Depression The Revised Almost Perfect Scale. Measurement and Evaluation in
Anxiety Stress Scales. Sydney, Australia: Psychology Foundation. Counseling and Development, 34, 130 –145.
Lundh, L.-G., Saboonchi, F., & Wangby, M. (2008). The role of personal Stoeber, J. (2014). How other-oriented perfectionism differs from self-
standards in clinically significant perfectionism. A person-oriented ap- oriented and socially prescribed perfectionism. Journal of Psychopathol-
proach to the study of patterns of perfectionism. Cognitive Therapy and ogy and Behavioral Assessment, 36, 329 –338. http://dx.doi.org/10.1007/
Research, 32, 333–350. http://dx.doi.org/10.1007/s10608-006-9109-7 s10862-013-9397-7
Mitchell, J. H., Broeren, S., Newall, C., & Hudson, J. L. (2013). An Stoeber, J., & Otto, K. (2006). Positive conceptions of perfectionism:
experimental manipulation of maternal perfectionistic anxious rearing Approaches, evidence, challenges. Personality and Social Psychology
Review, 10, 295–319. http://dx.doi.org/10.1207/s15327957pspr1004_2
behaviors with anxious and non-anxious children. Journal of Experi-
Suddarth, B. H., & Slaney, R. B. (2001). An investigation of the dimen-
mental Child Psychology, 116, 1–18. http://dx.doi.org/10.1016/j.jecp
sions of perfectionism in college students. Measurement and Evaluation
.2012.12.006
in Counseling and Development, 34, 157–165.
Moshman, D. (2005). Adolescent psychological development: Rationality,
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th
morality, and identity (2nd ed.). Mahwah, NJ: Erlbaum.
ed.). New York, NY: Allyn & Bacon.
Muuss, R. E. (2006). Theories of adolescence (6th ed.). New York, NY:
Vermunt, J. K. (2010). Latent class modeling with covariates: Two im-
McGraw-Hill Publishers. proved three-step approaches. Political Analysis, 18, 450 – 469. http://
Nilsson, K., Sundbom, E., & Hägglöf, B. (2008). A longitudinal study of dx.doi.org/10.1093/pan/mpq025
perfectionism in adolescent onset anorexia nervosa-restricting type. Eu- Wang, T. W., Slaney, R. B., & Rice, K. G. (2007). Perfectionism in
ropean Eating Disorders Review, 16, 386 –394. http://dx.doi.org/ Chinese university students from Taiwan: A study of psychological
10.1002/erv.850 well-being and achievement motivation. Personality and Individual Dif-
Nobel, R., Manassis, K., & Wilansky-Traynor, P. (2012). The role of ferences, 42, 1279 –1290. http://dx.doi.org/10.1016/j.paid.2006.10.006
perfectionism in relation to an intervention to reduce anxious and de- West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models
pressive symptoms in children. Journal of Rational-Emotive & with non-normal variables: Problems and remedies. In R. Hoyle (Ed.),
Cognitive-Behavior Therapy, 30, 77–90. http://dx.doi.org/10.1007/ Structural equation modeling: Concepts, issues, and applications (pp.
s10942-011-0133-5 56 –75). Thousand Oaks, CA: Sage.
O’Connor, R. C., Rasmussen, S., & Hawton, K. (2010). Predicting depres-
sion, anxiety and self-harm in adolescents: The role of perfectionism and Received November 4, 2013
acute life stress. Behaviour Research and Therapy, 48, 52–59. http://dx Revision received March 10, 2015
.doi.org/10.1016/j.brat.2009.09.008 Accepted March 11, 2015 䡲

You might also like