Advertisement

Cognitive Therapy and Research

, Volume 32, Issue 3, pp 333–350 | Cite as

The Role of Personal Standards in Clinically Significant Perfectionism. A Person-Oriented Approach to the Study of Patterns of Perfectionism

  • Lars-Gunnar Lundh
  • Fredrik Saboonchi
  • Margit Wångby
Original Article

Abstract

Clinically significant perfectionism is defined as patterns of perfectionism which are over-represented in clinical samples and under-represented in non-clinical samples. The present study contrasted two hypotheses about what characterizes clinically significant perfectionism: the two-factor theory and perfectionism/acceptance theory. First, a person-oriented approach by means of cluster analysis was used to identify typical patterns of perfectionism. These clusters were then cross-tabulated with two clinical samples (patients with social phobia and patients with panic disorder) and a non-clinical sample. The results showed that patterns of clinically significant perfectionism combined high Concern over Mistakes (CM) and Doubts about Action (DA) with high Personal Standards (PS) (and to a lesser extent also high Organization)––which is consistent with perfectionism/acceptance theory, but at odds with the two-factor theory. The results illustrate the value of a person-oriented methodological approach as a complement to the traditional variable-oriented approach.

Keywords

Perfectionism Social phobia Panic disorder Depression Person-oriented approach 

Introduction

The topic of perfectionism and its association with emotional health and psychopathology has become the focus of considerable research since the 90s, when four different research groups (Frost, Marten, Lahart, & Rosenblate, 1990; Hewitt & Flett, 1991; Slaney, Rice, Mobly, Trippi, & Ashby, 2001; Terry-Short, Owens, Slade, & Dewey, 1995) introduced multidimensional conceptions of perfectionism, with validated instruments for the measurement of different dimensions of perfectionism. An important implication of the multidimensional conception of perfectionism is that it opens for the possibility that some aspects or patterns of perfectionism are associated with psychopathology, whereas others are not––and even for the possibility that there may be aspects or patterns of perfectionism which are positively associated with health and well-being. Consequently, various researchers during the later part of 1990s came to distinguish between positive and negative perfectionism (Slade & Owens, 1998), active versus passive perfectionism (Adkins & Parker, 1996; Lynd-Stevenson & Hearne, 1999) and adaptive versus maladaptive perfectionism (Rice et al., 1998). Similarly, Frost, Heimberg, Holt, Mattia, and Neubauer (1993) factor analyzed Frost et al.’s (1990) and Hewitt and Flett’s (1991) multidimensional perfectionism scales and identified two factors, Maladaptive Evaluation Concerns and Positive Striving, of which the first correlated positively with depression and negative affect, and the second correlated positively with positive affect.

In the present study, we were interested in a particular variety of maladaptive perfectionism, which we labeled clinically significant perfectionism, and operationalized as patterns of perfectionism which are over-represented in clinical samples and under-represented in non-clinical samples. We used Frost et al.’s (1990) Multidimensional Perfectionism Scale (MPS), and a person-oriented methodological approach by means of cluster analysis, to test two different hypotheses about clinically significant perfectionism: the two-factor theory and perfectionism/acceptance theory. In the first theoretical part of this paper, these two theories are described and are shown to lead to different predictions about what characterizes maladaptive patterns of perfectionism. The rationale for using a person-oriented methodological approach as a complement to a variable-oriented approach is then outlined, especially for testing the competing predictions from perfectionism/acceptance theory and the two-factor theory.

Frost et al.’s (1990) MPS has six different subscales: Personal Standards (PS), Concern over Mistakes (CM), Doubts about Action (DA), Parental Expectations (PE), Parental Criticism (PC), and Organization (O). Research indicates that CM and DA (and to a lesser extent PE and PC) correlate positively with anxiety and depression (e.g., Frost et al., 1993; Kawamura, Hunt, Frost, & DiBartolo, 2001), whereas PS and O generally do not. That is, to be highly concerned over mistakes and to doubt the quality of one’s actions seem to be dysfunctional aspects of perfectionism, whereas the setting of high personal standards and emphasizing the value of order and organization are perhaps not. Accordingly, in terms of Frost et al.’s (1993) factor analysis, CM, DA, PC and PE loaded on the maladaptive factor, whereas PS and O loaded on the adaptive factor. Using confirmatory factor analysis, Bieling, Israeli and Antony (2004) found that this two-factor model was a better fit to the data than a unitary perfectionism model. Similar findings were reported by Stumpf and Parker (2000), who found two orthogonal higher-order MPS factors, that were interpreted as healthy and unhealthy perfectionism, and characterized as “two largely independent dimensions showing different patterns of correlations with other personality variables” (p. 849). This notion will here be referred to as the two-factor theory.

This two-factor theory is open to at least two kinds of critique. First, it can be questioned whether adaptive and maladaptive perfectionism are in fact two separate dimensions which correlate in opposite ways with positive and negative outcomes. According to Flett and Hewitt (2002), the notion of a positive, healthy, adaptive form of perfectionism has been too uncritically accepted, and a review of the literature shows several findings that are difficult to reconcile with the two-factor model (e.g., Cox, Enns, & Clara, 2002; Enns, Cox, Sareen & Freeman, 2001; Kawamura et al., 2001).

Second, the two-factor theory can be criticized for its implications about what constitutes maladaptive patterns of perfectionism. It is important to note that as soon as we generalize about persons, we move from the level of dimensions and their correlates to a more complex level of patterns of values on these different dimensions, and there is no easy translation from the one level to the other (Bergman & Magnusson, 1997). For example, what are we to predict from various combinations of scores on the two different MPS factors, like subgroups of individuals who score high on both factors, or low on both factors, or high on one factor and low on the other? If the two factors are interpreted as “adaptive” and “maladaptive” perfectionism, it seems reasonable to expect that high levels of “adaptive perfectionism” (as measured by the PS and O) should counteract or “buffer” the negative effects of high levels of “maladaptive perfectionism” (as measured by the CM and DA). That is, if adaptive and maladaptive perfectionism are independent factors, with opposite effects on health, the combination of high scores on measures of both should balance each other out, so that people with this particular pattern of perfectionism (i.e., high scores on CM, DA, PS, and O) should suffer less from psychopathology than people who score high only on the “maladaptive perfectionism” dimension (i.e., high scores CM and DA, in the absence of high scores on PS and O). This implication is clearly formulated by Bieling et al. (2004, p. 1383), who assume that the most dysfunctional pattern of perfectionism is the combination of high scores on the maladaptive factor and low score on the adaptive factor. Haase and Prapavessis (2004) similarly suggest that “high levels of Positive Perfectionism may ‘buffer’ relations between Negative Perfectionism and psychopathology” (p. 1737).

An alternative to the two-factor theory is perfectionism/acceptance theory (Lundh, 2004), according to which high personal standards or other strivings for perfection are adaptive when combined with the acceptance of non-perfection (i.e., the acceptance of various kinds of failures, mistakes, and shortcomings), but maladaptive when combined with an inability to accept failures, mistakes, and shortcomings. On the MPS, high PS and O represent perfectionistic strivings, whereas high CM and DA can be seen as markers for an inability to accept non-perfection. When high PS and high O are combined with high CM and DA, adaptive perfectionistic strivings are transformed into maladaptive perfectionistic demands. In opposition to the two-factor theory, therefore, perfectionism/acceptance theory predicts that the most maladaptive forms of perfectionism is a pattern of high scores on CM and DA in combination with high scores on PS and/or O.

Frost et al. (1990) hinted at this possibility when they suggested that “the relationship between Personal Standards and Concern over Mistakes needs further clarification. It may be, for instance, that high Personal Standards are associated with psychopathology only among people who are high in Concern over Mistakes” (p. 466–467). Importantly, this comment illustrates the possibility that high scores on one variable (in this case PS) may have different meaning depending on the pattern (“Gestalt”) that it is part of, and thereby illustrates the need for methodological approaches that focus on patterns, rather than focusing “atomistically” on single variables (Bergman & Magnusson, 1997). However, this lead was never followed by Frost et al.

Perfectionism/acceptance theory is consistent with Alden, Ryder and Melling’s (2002) two-component model of pathological perfectionism, according to which high standards will or will not be pathological depending on the presence of maladaptive self-appraisal. It is also consistent with Shafran, Cooper and Fairburn’s (2002) notion that clinical perfectionism is characterized by an “overdependence of self-evaluation on the determined pursuit (and achievement) of self-imposed personally demanding standards of performance in at least one salient domain, despite the occurrence of adverse consequences” (p. 773). According to this definition, “personally demanding standards of performance” are an intrinsic part of maladaptive perfectionism. Accordingly, very high scores not only on the “maladaptive” perfectionism scales of the MPS but also on the supposedly “adaptive” ones (i.e., PS and O) have been reported in patients with anorexia nervosa (Bastiani, Rao, Weltzin, & Kaye, 1995; Halmi et al., 2000; Srinivasagam et al., 1995).

These results also accord with Hamachek’s (1978) distinction between normal and neurotic perfectionism. “Normal perfectionism”, in this perspective, means to set high standards for oneself and to “derive a very real sense of pleasure from the labors of a painstaking effort”, and yet “feel free to be less precise as the situation permits” (p. 27). Neurotic perfectionism, on the other hand, means to set high standards and allow little latitude for making mistakes. That is, in Hamachek’s perspective, high personal standards are common to both adaptive and maladaptive perfectionism. In this perspective, it is interesting to note that Davis (1997) operationalized normal and neurotic perfectionism as separate dimensions––i.e., consistent with two-factor theorizing––and arrived at results consistent with perfectionism/acceptance theory. Davis studied the relation between these two dimensions of perfectionism and body esteem in bulimic and anorexic patients. The interesting thing about her results is that, when testing the interaction between “normal” and “neurotic” perfectionism in a multiple regression analysis, she found that “normal” perfectionism was positively related to body esteem when neurotic perfectionism was low, but at higher levels of neurotic perfectionism the slope of this relationship became flatter, and at the highest level of neurotic perfectionism it actually reversed into a negative relationship. That is, the most dysfunctional pattern of perfectionism was the combination of high scores on both “neurotic” and “normal” perfectionism.

From a clinical perspective, Shafran et al. (2002) have argued that the multidimensional conception of perfectionism has not contributed to any substantial advances in the understanding of clinical perfectionism. From our perspective, a similar critique can be formulated on the basis that all research on clinical samples that has been carried out with the multidimensional conception of perfectionism is variable-oriented––that is, it focuses on dimensions of perfectionism and their correlates. This is clearly at variance with the person-oriented interest in perfectionism as patterns of thinking and behaving in individual patients, which is of interest to the clinical psychologist. To identify two independent dimensions of perfectionism by means of factor analysis, and then to label them “adaptive” and “maladaptive” because of their different correlates is very different from showing that there are both adaptive and maladaptive perfectionists. The latter requires the demonstration of groups of individuals with different profiles of values on multidimensional measures of perfectionism, which are associated with different kinds of outcomes. The person-oriented approach that is pursued here focuses on such profiles of values on the MPS subscales, and compares their presence in clinical and non-clinical samples. The method that is used is cluster analysis, which starts by comparing individual profiles of values on the MPS subscales and then categorizes individuals with similar profiles into clusters in an iterative process that serves to find an optimal cluster solution.

Two studies have used cluster analysis to analyze patterns of MPS perfectionism in student samples (Parker, 1997; Rice & Mirzadeh, 2000). Both identified three clusters: maladaptive perfectionists, adaptive perfectionists, and non-perfectionists. Interestingly, in both studies the maladaptive perfectionists scored higher than the healthy perfectionists on five of six MPS scales, including Personal Standards. These results are clearly more compatible with perfectionism/acceptance theory than with the two-factor theory. No previous study, however, has used cluster analysis to compare clinical and non-clinical samples.

A further difference relative to the previous cluster analytic studies of perfectionism is that we used another set of criteria (Bergman, 1998) to choose the optimal cluster solution, and an explicit procedure (Monte Carlo simulations to create random data for comparison) to test the statistical significance of the cluster solution (Bergman, Magnusson, & El-Khouri, 2003). The latter is important, because cluster analysis has been criticized for producing cluster solutions to any set of data, including randomly generated data. Further, previous cluster analytic studies have reported neither the explained variance of the cluster solutions nor the homogeneity coefficients of the clusters––in the present study we used software (Bergman & El-Khouri, 2002) that made these computations possible.

Testing the hypotheses about clinically significant perfectionism requires a comparison between clinical samples and non-clinical samples, and that the clinical samples involve disorders which are associated with elevated levels of perfectionism. The clinical samples in the present study were patients with social phobia, and patients with panic disorder and agoraphobia. Social phobia has been repeatedly found to be associated with elevated scores on CM, DA and PC (Antony, Purdon, Huta, & Swindon, 1998; Juster et al., 1996; Saboonchi, Lundh, & Öst, 1999), and panic disorder has been found to be consistently associated with elevated scores at least on CM (although not as high as patients with social phobia), and less consistently on DA and PC (Antony et al., 1998; Saboonchi et al., 1999). Theoretical models clearly posit a role for perfectionistic beliefs in social anxiety (e.g., Alden et al., 2002), and it may even be argued that there is a conceptual overlap between the fear of social evaluation that is typical of social anxiety and perfectionistic concerns over mistakes and doubts about action. Although theoretical models of panic disorder generally do not include perfectionism as a factor, evidence shows that CM and DA correlate with agoraphobic fears and fears of illness (Saboonchi & Lundh, 1997), and that perfectionistic cognitions predict anxiety sensitivity (Flett, Greene, & Hewitt, 2004), which is a risk factor for panic disorder (McNally, 2002). Theoretically, it is possible both that perfectionistic demands about being free from anxiety-related symptoms contribute to an increased anxiety sensitivity, and that high anxiety sensitivity may foster demands about being perfectly free from anxiety-related symptoms (Ellis, 2002). Anyway, both previous research and theoretical considerations suggest that patterns of maladaptive perfectionism should be over-represented in both social phobia and panic disorder with agoraphobia.

No study has shown elevated scores on PS or O in patients with social phobia or panic disorder. Despite these results with a variable-oriented approach, perfectionism/acceptance theory predicts that patterns of clinically significant perfectionism among patients with these disorders should be characterized by elevated scores not only on CM and DA, but also on PS and/or O.

To summarize, the present study had two main purposes: (1) to identify typical patterns of perfectionism on the MPS, and (2) to test the hypothesis, derived from perfectionism/acceptance theory, that clinically significant perfectionism is characterized by patterns that combine high CM and DA with high PS and O (i.e., these patterns are over-represented in clinical samples, under-represented in non-clinical samples, and are associated with elevated scores on depression). This hypothesis is opposite to that derived from the two-factor theory, according to which patterns of clinically significant perfectionism should combine high CM and DA with low PS and O. An additional purpose was (3) to test the hypothesis (which is common to both the two-factor theory and perfectionism/acceptance theory) that there is also an adaptive form of perfectionism, which is characterized by high scores on PS and/or O and low scores on the other MPS scales, and which is under-represented in clinical samples, over-represented in non-clinical samples, and/or associated with low scores on depression.

Method

Participants1

The participants were 78 patients with social phobia, 67 patients with panic disorder and agoraphobia, and 432 non-clinical subjects. Both clinical groups were patients who had been referred to treatment projects with cognitive-behaviour therapy at the Department of Psychology, Stockholm University. All patients were diagnosed either according to DSM-III-R or DSM-IV criteria (American Psychiatric Association, 1987, 1994) by clinically trained psychologists using either the revised version of the Anxiety Disorders Interview Schedule (ADIS-R; DiNardo & Barlow, 1988) or the Anxiety Disorders Interview Schedule-IV (Brown, DiNardo, & Barlow, 1994). Potential participants were excluded if they received a diagnosis of primary depression, bipolar disorder, psychotic disorder, or active drug or alcohol dependence (For more information about the treatment studies and the excluded patients, see Öst, Thulin, & Ramnerö (2004) and Lundh & Öst (2001)).

The non-clinical participants were mostly students who participated in different research projects at the Department of Psychology, Stockholm University, the School of Social and Health Sciences, Halmstad University, and the Department of Social Sciences at Mälardalen University in Eskilstuna and Västerås, Sweden. They were recruited in two ways: (a) via advertisements in a living area not far from Stockholm University; these participants were paid 100 Swedish crowns for their participation, and (b) from undergraduate students who participated for course credits.

Materials

Multidimensional Perfectionism Scale (MPS; Frost et al., 1990). This instrument, which was developed by Frost et al. (1990), contains 35 items in the form of statements with a Likert type 5-points response format ranging from strongly disagree to strongly agree. The 35 items are divided into six subscales tapping six dimensions of perfectionism: Concern over Mistakes (CM, 9 items), Personal Standards (PS, 7 items). Parental Expectations (PE, 5 items), Parental Criticism (PC, 4 items), Doubts about Action (DA, 4 items), and Organization (O, 6 items). Cronbach’s alphas for the Swedish version of the MPS in the present study were: CM α = .87, PS α = .78, PE α = .87, PC α = .87, DA α = .76, and O α = .86.

Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). The BDI is a 21-item self-report questionnaire that assesses the presence and severity of cognitive, behavioral, affective and somatic symptoms of depression. Each item consists of a 0 to 3 self-rating, with increasing scores indicating greater symptom severity. The BDI is internally consistent, has high test–retest reliability, and has been shown to be valid with both psychiatric and normal samples (Beck, Steer, & Garbin, 1988). BDI data was available for all 145 patients and for 165 individuals from the non-clinical sample.

Procedure

All patients filled out the questionnaires individually, after having been interviewed, but before they started treatment. The questionnaires were filled out at the beginning of a session that involved various cognitive laboratory tasks.

Cluster analysis

Cluster analysis was used to group the individuals on the basis of their different profiles of scores on the MPS, according to the LICUR procedure (Bergman, 1998). This was done in four steps. First, 10 outliers were identified by means of the residue procedure in the statistical package for pattern-oriented analyses SLEIPNER 2.1 (Bergman & El-Khouri, 2002). Second, Wards’s hierarchical clustering method was applied, which starts by considering each individual case as a separate cluster. At each subsequent step, the two clusters are then merged that result in the smallest increase in the overall error sum of squares (ESS). The purpose of this is to provide as homogenous clusters as possible, and the process continues until the optimal cluster solution is identified. Four criteria presented by Bergman (1998) were used to find the optimal cluster solution: (1) theoretical meaningfulness of the cluster solution; (2) if a distinct drop in the explained error sum of squares (EESS) occurs when a cluster solution is extracted this may imply that two not so similar clusters were merged to a non-optimal cluster solution; (3) the number of clusters should not be more than 15 and should not be expected to be less than five; (4) the size of the EESS for the chosen cluster solution should preferably not be less than 67%, and at the very least exceed 50%. In addition, the homogeneity coefficient of each cluster should preferably be <1. Third, a data simulation was undertaken to verify that the explained ESS was higher than what could be expected on a random data set with the same general properties as the “real” data set. Fourth, a non-hierarchical relocation procedure (Wishart, 1987) was carried out in order to improve the homogeneity of the clusters and to increase the variance explained by the cluster solution.

Results

Demographic data

The non-clinical sample (mean age 27.1, SD = 7.8) was significantly younger than the patients with social phobia (mean age 31.6, SD = 8.5), and panic disorder with agoraphobia (mean age 35.8, SD = 9.7), but the groups did not differ significantly in terms of gender distribution (the ratio of women to men was 65/13 for the social phobia group, 46/21 for the panic patients, and 331/100 for the non-clinical sample, χ 2(2) = 4.35, P > .11). A one-way ANOVA with Tukey post-hoc tests showed that both the social phobia and the panic disorder groups scored higher than the non-clinical sample on the BDI, 15.6 (SD = 8.8) and 13.7 (SD = 8.7) vs. 7.0 (SD = 6.8), respectively, although the patient groups did not differ significantly.

Cluster analysis

First, 10 outliers were identified by means of the residue procedure in the statistical package for pattern-oriented analyses SLEIPNER 2.1 (Bergman & El-Khouri, 2002) and excluded, thus leaving 567 individuals. Second, Wards’s hierarchical clustering method was applied to find the optimal cluster solution. The application of Bergman’s (1998) criteria resulted in the choice of an 11-cluster solution, which explained 61.29% of the total error sum of squares (ESS); choosing the 10-cluster solution would have caused a distinct drop in the explained ESS to 59.63. Third, a data simulation showed that the explained ESS of the cluster solution was significantly higher than expected by chance (p < .0001). Fourth, a non-hierarchical relocation procedure (Wishart, 1987) was carried out in order to improve the homogeneity of the clusters and to increase the variance explained by the cluster solution. This procedure resulted in an 11-cluster solution that, after the relocation procedure, was found to explain 65.85% of the variance, with all clusters having a homogeneity coefficient of <1. Table 1 shows the profiles of the MPS scale means for the clusters, together with the means for the whole group, and Fig. 1 shows the profiles of z-scores for each cluster (defined as the differences between the cluster mean and the total group mean, divided by the SD of the total group).
Table 1

Profiles of perfectionism: 11-cluster solution

Cluster

N

CM

DA

PE

PC

PS

O

M (SD)

M (SD)

M (SD)

M (SD)

M (SD)

M (SD)

High Perfectionism 1

19

30.4 (7.4)

10. 9 (2.0)

20.0 (2.7)

16.7 (2.2)

27.2 (3.8)

24.4 (2.5)

High Perfectionism 2

41

32.5 (5.4)

12.9 (2.2)

13.9 (2.5)

9.3 (2. 6)

25.6 (3.9)

22.2 (3.6)

High Present Perfectionism

52

28.1 (5.5)

11.5 (2.8)

7.9 (1.7)

6.0 (2.1)

26.0 (3.9)

25.0 (3.1)

High Historical Perfectionism

42

26.5 (5.1)

8.6 (2.4)

17.4 (2.7)

12.8 (2.3)

22.2 (3.3)

16.7 (4.1)

High Parental Expect.

43

20.4 (4.8)

7.5 (2.3)

15.2 (2.9)

7.8 (2.2)

24.5 (2.7)

22.2 (3.5)

High Doubts about Action

56

22.1 (4.6)

11.8 (2.1)

7.9 (2.3)

6.8 (2.3)

17.0 (3.2)

16.8 (2.7)

High Organization/PS

59

17.4 (4.6)

6.9 (2.0)

7.3 (1.9)

4.9 (1.6)

23.6 (3.3)

25.5 (2.7)

Low Organization

74

22.5 (4.5)

7.2 (1.8)

8.0 (2.2)

5.0 (1.3)

22.6 (3.1)

16.5 (3.3)

Normal Perfectionism

49

17.8 (3.7)

7.2 (1.5)

11.1 (2.6)

8.5 (2.7)

17.5 (2.7)

20.7 (2.6)

Low Perfectionism

76

13.6 (3.2)

6.3 (1.9)

7.0 (1.8)

4.6 (1.0)

14.8 (2.8)

21.9 (2.9)

Non-Perfectionism

56

12.9 (3.3)

5.1 (1.3)

7.3 (2.1)

4.7 (1.3)

13.8 (4.0)

13.1 (3.2)

Alla

577

21.2 (7.7)

8.4 (3.2)

10.2 (4.6)

7.1 (3.7)

20.7 (5.7)

20.1 (5.1)

aIncluding 10 outliers who were not classified in any cluster

Fig. 1

Profiles of the 11 MPS clusters, in terms of z-scores (where z = 0 corresponds to the sample mean on each MPS scale). CM = Concern over Mistakes; DA = Doubts about Action; PE = Parental Expectations; PC = Parental Criticism; PS = Personal Standards; and O = Organization

The cluster analysis identified two clusters with generally high perfectionism scores. First, there was a small cluster (HighPerf1) of 19 individuals with a total mean MPS score of 129.6, including very high scores especially on PE and PC (effect sizes z > 2) but also on CM and PS (z > 1), and relatively high scores (z > .78) also on DA and O. Second, there was a larger cluster (HighPerf2) of 41 individuals with a total mean MPS score of 116.3, including very high scores on CM and DA (z > 1.4), relatively high scores on PS and PE (z > 0.8) and PC (z > 0,6), and slightly elevated scores (z > .4) on O.

Unexpectedly, the cluster analysis also identified two moderately high perfectionism clusters that differed in terms of a dimension that has not been much discussed in the literature: “present” versus “historical” perfectionism. First, there was a High Present Perfectionism cluster (HighPres, 52 individuals, mean total MPS score 104.5), characterized by high scores on the four perfectionism scales that refer to present functioning (CM, DA, PS and O, all z > .9), but low scores on the “historical” scales PE and PC (<the total group mean). Second, there was a High Historical Perfectionism cluster (HighHist, 42 individuals, mean total MPS score 104.2), which scored high specifically on PE and PC (both z > 1), and moderately high on CM (z = .69), but close to the group mean on PS and DA, and well below the group mean on O.

Also somewhat unexpectedly, two clusters were identified with high scores specifically on only one MPS scale. One of these was characterized by High Parental Expectations (HighPE, 43 individuals, z = 1.11 on PE, mean total MPS score 97.5). The other was a cluster with High Doubts about Action (HighDA, 56 individuals, z = 1.06 on DA, mean total MPS score 82.4).

There was also a relatively large cluster which conformed fairly well to the profile that is predicted to be characteristic of people with adaptive perfectionism. This cluster (with 59 individuals and a mean total MPS score of 85.5) scored high on Organization (z = 1.06) and moderately high (z = .51) on Personal Standards, while scoring well below the group mean on all other scales, and was therefore termed High Organization and Personal Standards (HighOrgPs).

Finally, there were four rather large clusters with low or average scores on all or most MPS scales. One cluster (with 56 individuals and a mean total MPS score of 57.4) scored consistently low on all MPS scales (z < −1 on CM, DA, PS and O, and z < −.5 on PC and PE), and was referred to as Non-Perfectionists” (NonPerf). Another cluster, actually the largest one (with 76 individuals and a mean total MPS score of 68.2), scored low (z < −.5) on all MPS subscales except Organization (z = .35), and was referred to as Low Perfectionism (LowPerf). The final two clusters scored relatively close to the total group mean on all MPS scales, while at the same time showing different profiles of perfectionism scores. One of these (with 74 individuals and a mean total MPS score of 81.8) was most clearly characterized by its low score on Organization (z = −.70) and was therefore labeled Low Organization (LowOrg). The other one (with 49 individuals and a mean total MPS score of 82.6) was labeled Normal Perfectionism (NormPerf), because it scored relatively close to the group mean on all scales, except for PS and CM (both z < −.4).

One-way ANOVA with the 11 clusters as the independent variable showed that the clusters differed significantly on age, F(10) = 4.05, P < .0001. Tukey post-hoc tests showed that the HighPerf1 cluster (mean age 38.5, SD = 11.3) was significantly older than all the other clusters, which did not differ significantly in age (means ranging from 26.6 to 31.3 years). The 11 clusters did not differ significantly in terms of gender distribution, χ2(10) = 17.3, P > .06), although the men were relatively most represented in the HighDA cluster (22 of 56), and the women relatively most represented in the HighPerf2 cluster (38 of 41).

The over- and under-representation of the clusters in the different samples

Table 2 shows a cross-tabulation of perfectionism clusters and the three samples. To test the hypotheses, the observed frequency in each cell was compared with the frequency that should be expected by chance alone, and one-tailed probabilities were computed according to the fixed-margins model using EXACON (Bergman & El-Khouri, 1987).
Table 2

Cross-tabulation of groups and perfectionism clusters, with a comparison between observed and expected frequencies in each cell (expected frequencies in parentheses)

 

Social phobia (n = 73)

Panic disorder with agoraphobia (n = 65)

Non-clinical group (n = 429)

High Perfectionism 1

6 (2.4)*

6 (2.2)*

7 (14.4)***

High Perfectionism 2

12 (5.3)**

2 (4.7)

27 (31.0)

High Present Perfectionism

15 (6.7)***

7 (6.0)

30 (39.3)**

High Historical Perfectionism

11 (5.4)*

9 (4.8)*

22 (31.8)***

High Parental Expectations

2 (5.5)

4 (4.9)

37 (32.5)

High Doubts about Action

10 (7.2)

4 (6.4)

42 (42.4)

High Organization/Personal Stand

0 (7.6)***

4 (6.8)

55 (44.6)***

Low Organization

6 (9.5)

9 (8.5)

59 (56.0)

Normal Perfectionism

4 (6.3)

7 (5.6)

38 (37.1)

Low Perfectionism

5 (9.8)

4 (8.7)*

67 (57.5)**

Non-Perfectionism

2 (7.2)*

9 (6.4)

45 (42.4)

* p < .05; ** p < .01; *** p < .001

Of the 11 clusters, one (HighPerf1) fulfilled all criteria for being “clinically significant”: it was over-represented in the social phobia sample (observed frequency 6, expected frequency 2.45, χ 2 = 5.16, p = .026) and in the panic disorder sample (observed frequency 6, expected frequency 2.18, χ 2 = 6.71, p = .015), and under-represented in the non-clinical sample (observed frequency 7, expected frequency 14.38, χ 2 = 3.78, p = .0003). Two other clusters (HighPerf2 and High Present Perfectionism) also fulfilled some criteria for being clinically significant: High Present Perfectionism (HighPres) was over-represented in the social phobia sample (observed frequency 15, expected frequency 6.69, χ 2 = 10.30, p = .001) and under-represented in the non-clinical sample (observed frequency 30, expected frequency 39.34, χ 2 = 2.22, P = .002); and HighPerf2 was over-represented in the social phobia sample (observed frequency 12, expected frequency 5.28, χ 2 = 8.56, p = .003). All three of these clusters were characterized by high scores on CM and DA in combination with high scores on PS and O, which is consistent with perfectionism/acceptance theory, but inconsistent with the two-factor theory.

With regard to the hypothesis of adaptive perfectionism, one cluster (HighOrgPS) conformed to the expected profile (i.e., high scores on PS and O, in combination with low scores on the other scales), and was found to be under-represented (in fact, unrepresented) in the social phobia sample (observed frequency 0, expected frequency 7.69, χ 2 = 7.60, P = .0002), and over-represented in the non-clinical sample (observed frequency 55, expected frequency 44.64, χ 2 = 2.40, P = .0003).

With regard to the exploratory part of the analysis (i.e. for the 21 cells where no a priori hypotheses were formulated), Bonferroni correction was used to adjust the alpha level to .05/21 = .0024. On this criterion, only one additional effect was statistically significant: the High Historical Perfectionism cluster (HighHist) was under-represented in the non-clinical sample (observed frequency 22, expected frequency 31.78, χ 2 = 3.01, p = .0005). In addition, however, it also tended to be over-represented both among the social phobics (observed frequency 11, expected frequency 5.41, χ 2 = 5.78, p = .012), and the panic patients (observed frequency 9, expected frequency 4.81, χ 2 = 3.64, p = .039).

Comparison between the clusters on depression

One-way ANOVA with the 11 cluster solution as the independent variable showed that the clusters differed significantly on BDI, F(10) = 6.69, P < .0001. The means and standard deviations for their scores on the BDI are shown in Table 3. Tukey post-hoc tests showed that the HighPerf1, HighPerf2, HighHist, and HighDA clusters scored higher on depression than other clusters, whereas the HighOrgPS, NormPerf, LowPerf and NonPerf clusters scored lower than other clusters. The HighPres, HighPE and LowOrg clusters did not differ significantly from any of the other clusters (for more detailed information about which comparisons were significant, see Table 3).
Table 3

Comparison between the clusters on the BDI

Cluster

N

BDI

M (SD)

P (Tukey post-hoc tests)

HighPerf1

16

16.19 (11.69)

>NormPerf, HighOrgPS, NonPerf, LowPerf

HighPerf2

22

16.50 (8.10)

>NormPerf, HighOrgPS, NonPerf, LowPerf

HighPres

35

10.89 (8.02)

 

HighHist

29

15.10 (8.88)

>HighOrgPS, NonPerf, LowPerf

HighPE

18

11.33 (8.52)

 

HighDA

31

12.68 (7.67)

>NonPerf

HighOrgPS

26

6.65 (8.26)

<HighPerf1, HighPerf2, HighHist

LowOrg

33

10.21 (7.68)

 

NormPerf

32

8.59 (8.82)

<HighPerf1, HighPerf2

LowPerf

37

6.32 (5.53)

<HighPerf1, HighPerf2, HighHist

NonPerf

31

4.87 (5.16)

<HighPerf1, HighPerf2, HighHist, HighDA

All

310

10.35 (8.63)

 

The residue

Cluster analysis is highly sensitive to single individuals with extreme and unique profile patterns, and the presence of a few cases that lie outside the general patterns may appreciably alter the total cluster solution; such outliers were therefore excluded from the analysis. Further, individuals with unique patterns of this kind should not be forced into a classification; rather, it may be of theoretical interest to identify individuals with “unique” patterns and to analyze them separately (Bergman, 1988). In the present study, a residue of 10 outliers (five from the social phobia group, two from the panic disorder group, and three from the non-clinical group) with unique patterns of MPS scores was identified by means of the standard residue procedure in the statistical package for pattern-oriented analyses SLEIPNER 2.1 (Bergman & El-Khouri, 2002). It is interesting to note that of these 10 outliers, nine individuals scored very high on most MPS scales (total MPS scores ranging from 104 to 150), whereas one scored very low (71); and that five of these very-high-perfectionism individuals were from the social phobia group.

Discussion

The present results on clinically significant perfectionism are clearly more consistent with perfectionism/acceptance theory than with the two-factor theory. The cluster analysis identified three high perfectionism clusters (HighPerf1, HighPerf2, and HighPres) which represented varieties of clinically significant perfectionism, in the sense that they were overrepresented among the social phobia patients and associated with higher scores on depression; one of these was also over-represented among the panic disorder patients (HighPerf1), and two were underrepresented in the non-clinical sample (HighPerf1 and HighPres). All three of these clusters were characterized by high scores on Personal Standards (PS; z > 0.8), which is consistent with predictions from perfectionism/acceptance theory, but at odds with predictions from the two-factor theory. Also at odds with the two-factor theory, no cluster conformed to the pattern of perfectionism (high CM and DA, in combination with low PS and O) that should be most maladaptive according to the this theory (e.g., Bieling et al., 2004).

The cluster analysis also identified a fourth cluster, HighHist, which could be seen as maladaptive, because it was under-represented in the non-clinical sample and associated with higher degrees of depression, and also tended to be over-represented in both clinical samples. The HighHist cluster scored high specifically on “historical” perfectionism (i.e., PE and PC), but also had moderately high scores on CM, although not on PS. In terms of Shafran et al.’s (2002) model, this is not a high perfectionism cluster, because it shows no evidence of high self-imposed personal standards. In terms of Hewitt and Flett’s (1991) model, however, maladaptive perfectionism need not necessarily involve self-imposed perfectionistic demands but may alternatively involve the experience of perfectionistic demands from others (“Socially Prescribed Perfectionism”); the HighHist cluster may be an example of this.

The cluster analysis also identified a High Parental Expectations cluster (HighPE), which was not associated with psychopathology. The HighPE cluster differed from the HighHist cluster by scoring high only on PE and not on PC. This suggests that high parental expectations as such, without high parental criticism, are not associated with psychopathology. One possibility is that high PE serves a similar role in socially prescribed perfectionism as high PS does in self-imposed perfectionism––that is, high PE is maladaptive only in connection with elevated scores on other dimensions of perfectionism. From the perspective of perfectionism/acceptance theory (Lundh, 2004), high PE but not high PC are compatible with parental acceptance of non-perfection.

The analysis also identified a cluster characterized by High Doubts about Action, which was associated with elevated scores on depression but showed no other evidence of clinically significant perfectionism. This does not preclude the possibility, however, that this pattern of perfectionism may possibly be associated with other varieties of psychopathology.

With regard to adaptive perfectionism, the analysis identified one cluster that at least partly conformed to the expected profile characteristic of people with adaptive perfectionism. This cluster was termed HighOrgPS, because it was characterized by high O, moderately high scores on PS (although, it may be noted, lower than in the three maladaptive perfectionism clusters), and low scores on all other MPS scales. This cluster was “adaptive” in the sense that it was overrepresented in the non-clinical sample, unrepresented in the social phobia sample, and scored low on depression. We have no evidence, however, that this cluster was more adaptive than the clusters that were characterized by non-perfectionism (i.e., low scores on all or almost all MPS scales, as seen in the NonPerf and LowPerf clusters), or “normal” perfectionism (average scores on MPS scales). Conclusions on this matter must await research which includes measures also of positive affect, health, and well-being.

The present results clearly suggest a different picture of the role of high personal standards in maladaptive perfectionism than that obtained in previous studies with a variable-oriented approach. As seen in Table 1, of the four clusters with the highest PS scores (HighPerf1, HighPerf2, HighPres, and HighOrgPS), the three former were overrepresented among the social phobics (and two of them underrepresented in the non-clinical sample), whereas the fourth one was overrepresented in the non-clinical sample and totally unrepresented in the social phobia sample. In other words, whereas social phobics who scored high on PS tended to do this in combination with high scores on the more typical “maladaptive” MPS subscales, the non-clinical participants who scored high on PS did this in the relative absence of high scores on the more typical “maladaptive” MPS subscales. That is, high PS scores in social phobics seem to be part of a completely different “Gestalt”, as compared with high PS scores in non-clinical participants. This seems to be a highly illustrative example of how a person-oriented approach can bring new knowledge that is complementary to the results that are gained from a traditional variable-oriented approach.

A similar argument can be made for the Organization dimension. Although high O in itself merely indicates that the individual puts a high premium on organization and orderliness, it seems that when this motive is combined with a lack of acceptance of possible failures in these strivings (as seen, for example, in high CM and DA) they take on the additional meaning of a must that cannot be ignored without feeling like a failure. The four clusters with the highest PS scores––that is, the three most maladaptive clusters (HighPerf1, HighPerf2, HighPres) and the adaptive perfectionism cluster (HighOrgPS)––also had the highest scores on O. This, again, illustrates a pattern of results that is hard to discover by variable-oriented approaches. That is, the meaning of high O obviously differs radically depending on the kind of pattern it is part of.

The present results corroborate findings from previous research that maladaptive perfectionism seems to be more of a problem in social phobia than in panic disorder. Still, far from all social phobics showed a pattern of maladaptive perfectionism. In fact, a sizeable number of social phobics (23%), and an even larger number of panic disorder patients (51%), were classified in the five clusters that were characterized by adaptive perfectionism (HighOrgPS), normal perfectionism (NormPerf and LowOrg), low perfectionism (LowPerf), and non-perfectionism (NonPerf). An interesting question is if these different patterns of perfectionism have implications for the causal processes involved in these cases of anxiety disorders, and for the treatment of these patients. One preliminary result which speaks in this direction is Lundh and Öst’s (2001) finding of a pattern of high PS in combination with high CM and DA in a small subgroup of patients with social phobia who responded less well to cognitive-behavioural treatment.

An unexpected result was the identification of two high MPS clusters that were on opposite ends of a present/historical dimension: (a) A cluster (HighPres) with elevated scores on all “present perfectionism” scales (i.e., CM, DA, PS, and O) in combination with low PE and PC; and (b) a “historical perfectionism” cluster (HighHist) with very high PE and PC. This dissociation between groups that score high on present and historical perfectionism has, to our knowledge, not been observed earlier in the literature. The elevated BDI scores in the HighHist cluster might indicate either that parental perfectionistic demands represent a risk factor for the development of depression, or that depressed people tend to interpret past relationships in overly negative terms––because of the correlational nature of the present study, there is no way of concluding which interpretation is the most probable one from the present data.

Previously published cluster analyses of patterns of perfectionism with the MPS in non-clinical samples (Parker, 1997; Rice & Mirzadeh, 2000) have arrived at three-cluster solutions, with three theoretically meaningful clusters: adaptive perfectionists, maladaptive perfectionists, and non-perfectionists. In contrast to these results, the present study produced an 11-cluster solution, based on Bergman’s (1998) criteria. It is difficult to evaluate the three-cluster solutions from previous studies, because they neither describe the explained error sum of squares (EESS) for their cluster solutions, nor the homogeneity coefficients of the clusters. In fact, a three-cluster solution for our data would have looked partly similar to the three-cluster solutions described in the previous studies. However, this cluster solution would only have explained 35.8% of the total error sum of squares, and both the maladaptive and adaptive/normal clusters would have suffered from low homogeneity.

On the other hand, it might be argued that an 11-cluster solution is too complex and over-differentiated. Here it is important to emphasize that the present findings do not imply the existence of 11 “natural types” of patterns on perfectionism measures. In the present study, on the basis of Bergman’s (1998) criteria, the 11-cluster solution was the most optimal, but it remains to be seen if similar analyses with the same criteria in other samples produce similar cluster solutions. What is of most theoretical importance, for the present purposes, however, is not the exact number of clusters that were identified, but the hypotheses based on perfectionism/acceptance theory that were tested and supported by the present analysis.

Several limitations of the present study should be acknowledged. First, the study used no measure of positive affect, health or any other aspect of adaptive functioning, which would be required to draw stronger conclusion about adaptive patterns of perfectionism. Second, the study involved no measure of trait anxiety, so we could not compare the degree of anxiety of the clinical groups with the non-clinical sample. Third, the non-clinical sample was significantly younger than the two patient samples, and the cluster with the highest overall scores on perfectionism was older than the other clusters. Would the HighPerf1 cluster still be under-represented in the non-clinical comparison group if it had been age-matched with the patients? Fourth, the non-clinical sample was not systematically screened for psychopathology; that is, it may have contained participants with various mental disorders, including social phobia and panic disorder. Sixty-four participants from the non-clinical sample were categorized in the three maladaptive perfectionism clusters (High Perfectionism 1, High Perfectionism 2, High Present Perfectionism), and it would have been interesting to see to what extent these individuals fulfilled criteria for mental disorders (e.g., anxiety disorders, depression, or eating disorders). Still, this limitation of the present study would be expected to make it harder to find differences in patterns of perfectionism between the clinical samples and the non-clinical sample, and can hardly be used to question the direction of the differences that were found. Finally, some additional limitations are the uneven gender distribution in the present study, and the mono-method approach to assessment (i.e., using only self-report measures of perfectionism and depression).

To summarize, the present study illustrates an approach to psychopathology that is probably too seldom used. Whereas almost all research in this area focuses on single variables, the present study focuses on patterns of values on related variables. This makes it possible to study the association between different kinds of psychopathology and patterns of personality, thereby coming closer to the “real person”, and making research with multidimensional measures of perfectionism more relevant to the study of what Shafran et al. (2002) refer to as clinical perfectionism. This approach may also be interesting to pursue in treatment research––that is, to study the importance for treatment outcome not only of individual predictor variables, but also of patterns of values on such variables.

Footnotes

  1. 1.

    A subset of the data from this study were presented in another paper (Saboonchi et al., 1999) that compared patients with social phobia, panic disorder and agoraphobia, and a non-clinical sample, in a more traditional variable-oriented approach.

Notes

Acknowledgement

We would like to thank Professor Lars R. Bergman for highly valuable comments on the cluster analytic parts of the study.

References

  1. Adkins, K. K., & Parker, W. (1996). Perfectionism and suicidal preoccupation. Journal of Personality, 64, 529–543.PubMedCrossRefGoogle Scholar
  2. Alden, L. E., Ryder, A. G., & Mellings, T. M. B. (2002). In: G. L. Flett & P. L. Hewitt (Eds.), Perfectionism. Theory, research, and treatment (pp. 373–391). Washington, D.C.: American Psychological Association.Google Scholar
  3. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author.Google Scholar
  4. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.Google Scholar
  5. Antony, M. M., Purdon, C. L., Huta, V., & Swindon, R. P. (1998). Dimensions of perfectionism across the anxiety disorders. Behaviour Research and Therapy, 36, 1143–1154.PubMedCrossRefGoogle Scholar
  6. Bastiani, A. M., Rao, R., Weltzin, T., & Kaye, W. H. (1995). Perfectionism in anorexia nervosa. International Journal of Eating Disorders, 17, 147–152.PubMedCrossRefGoogle Scholar
  7. Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five year of evaluation. Clinical Psychology Review, 8, 77–100.CrossRefGoogle Scholar
  8. Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561–571.PubMedGoogle Scholar
  9. Bergman, L. R. (1988). You can’t classify all of the people all of the time. Multivariate Behavioral Research, 23, 425–441.CrossRefGoogle Scholar
  10. Bergman, L. R. (1998). A pattern-oriented approach to studying individual development: Snapshots and processes. In R. B. Cairns, L. R. Bergman, & J. Kagan (Eds.), Methods and models for studying the individual (pp. 82–31). Thousand Oaks.Google Scholar
  11. Bergman, L. R., & El-Khouri, B. M. (1987). EXACON: A fortran 77 program for the exact analysis of single cells in a contingency table. Educational and Psychological Measurement, 47, 155–161.CrossRefGoogle Scholar
  12. Bergman, L. R. & El-Khouri, B. M. (2002). SLEIPNER. A statistical package for pattern-oriented analyses. Version 2.1. Stockholm: Stockholm University, Department of Psychology. www.psychology.su.se/sleipner.Google Scholar
  13. Bergman, L. R., & Magnusson, D. (1997). A person-oriented approach in research on developmental psychopathology. Development and Psychopathology, 9, 291–319.PubMedCrossRefGoogle Scholar
  14. Bergman, L. R., Magnusson, D., & El-Khouri, B. M. (2003). Studying individual development in an interindividual context. A person-oriented approach. Mahwah, NJ: Lawrence Erlbaum.Google Scholar
  15. Bieling, P. J., Israeli, A. L., & Antony, M. M. (2004). Is perfectionism good, bad, or both? Examining models of the perfectionism construct. Personality and Individual Differences, 36, 1373–1385.CrossRefGoogle Scholar
  16. Brown, T. A., DiNardo, P. A., & Barlow, D. H. (1994). Anxiety disorders interview schedule for DSM-IV. Albany, N.Y.: Center for Stress and Anxiety Disorders.Google Scholar
  17. Cox, B. J., Enns, M. W., & Clara, I. (2002). The multidimensional structure of perfectionism in clinically distressed and college student samples. Psychological Assessment, 14, 365–373.PubMedCrossRefGoogle Scholar
  18. Davis, C. (1997). Normal and neurotic perfectionism in eating disorders: An interactive model. International Journal of Eating Disorders, 22, 421–426.PubMedCrossRefGoogle Scholar
  19. DiNardo, P. A., & Barlow, D. H. (1988). Anxiety disorders interview schedule-revised. Albany, N.Y.: Center for Stress and Anxiety Disorders.Google Scholar
  20. Ellis, A. (2002). The role of irrational beliefs in perfectionism. In: G. L. Flett & P. L. Hewitt (Eds.), Perfectionism. Theory, research, and treatment (pp. 217–229). Washington, D.C.: American Psychological Association.Google Scholar
  21. Enns, M. W., Cox, B. J., Sareen, J., & Freeman, P. (2001). Adaptive and maladaptive perfectionism in medical students: A longitudinal investigation. Medical Education, 35, 1034–1042.PubMedCrossRefGoogle Scholar
  22. Flett, G. L., Greene, A., & Hewitt, P. L. (2004). Dimensions of perfectionism and anxiety sensitivity. Journal of Rational-Emotive and Cognitive-Behavior Therapy, 22, 39–57.CrossRefGoogle Scholar
  23. Flett, G. L., & Hewitt, P. L. (2002). Perfectionism and maladjustment: An overview of theoretical, definitional, and treatment issues. In G. L. Flett, & P. L. Hewitt (Eds.), Perfectionism. Theory, research, and treatment (pp. 5–32). Washington, D.C.: American Psychological Association.Google Scholar
  24. Frost, R. O., Heimberg, R. G., Holt, C. S., Mattia, J. I., & Neubauer, L. A. (1993). A comparison of two measures of perfectionism. Personality and Individual Differences, 14, 119–126.CrossRefGoogle Scholar
  25. Frost, R. O., Marten, P., Lahart, C., & Rosenblate, R. (1990). The dimensions of perfectionism. Cognitive Therapy and Research, 14, 449–468.CrossRefGoogle Scholar
  26. Haase, A. M., & Prapavessis, H. (2004). Assessing the factor structure and composition of the positive and negative perfectionism scale in sport. Personality and Individual Differences, 36, 1725–1740.CrossRefGoogle Scholar
  27. Halmi, K. A., Sunday, S. P., Stober, M., Kaplan, A., Woodside, D. B., Fichler, M., et al. (2000). Perfectionism in anorexia nervosa: Variations by clinical subtype, obsessionality, and pathological eating behavior. American Journal of Psychiatry, 157, 1799–1805.PubMedCrossRefGoogle Scholar
  28. Hamachek, D. E. (1978). Psychodynamics of normal and neurotic perfectionism. Psychology, 15, 27–33.Google Scholar
  29. Hewitt, P. L., & Flett, G. L. (1991). Perfectionism in the self and social contexts: Conceptualization, assessment, and association with psychopathology. Journal of Personality and Social Psychology, 60, 456–470.PubMedCrossRefGoogle Scholar
  30. Juster, H. R., Frost, R. O., Heimberg, R. G., Holt, C. S., Mattia, J. I., & Faccenda, K. (1996). Social phobia and perfectionism. Personality and Individual Differences, 21, 403–410.CrossRefGoogle Scholar
  31. Kawamura, K. Y., Hunt, S. L., Frost, R. O., & DiBartolo, P. M. (2001). Perfectionism, anxiety, and depression: Are the relationships independent? Cognitive Therapy and Research, 25, 291–301.CrossRefGoogle Scholar
  32. Lundh, L. G. (2004). Perfectionism and acceptance. Journal of Rational-Emotive and Cognitive-Behavior Therapy, 22, 251–265.CrossRefGoogle Scholar
  33. Lundh, L. G., & Öst, L. G. (2001). Attentional bias, self-consciousness, and perfectionism in social phobia before and after cognitive-behaviour therapy. Scandinavian Journal of Behaviour Therapy, 30, 4–16.CrossRefGoogle Scholar
  34. Lynd-Stevenson, R. M., & Hearne, C. M. (1999). Perfectionism and depressive affect: The pros and cons of being a perfectionist. Personality and Individual Differences, 26, 549–562.CrossRefGoogle Scholar
  35. McNally, R. J. (2002). Anxiety sensitivity and panic disorder. Biological Psychiatry, 52, 938–946.PubMedCrossRefGoogle Scholar
  36. Öst, L. G., Thulin, U., & Ramnerö, J. (2004). Cognitive behavior therapy vs exposure in vivo in the treatment of panic disorder with agrophobia. Behaviour Research and Therapy, 42, 1105–1127.PubMedCrossRefGoogle Scholar
  37. Parker, W. D. (1997). An empirical typology of perfectionism in academically talented children. American Educational Research Journal, 34, 545–562.Google Scholar
  38. Rice, K. G., Ashby, J. S., & Slaney, R. B. (1998). Self-esteem as a mediator between perfectionism and depression: A structural equations analysis. Journal of Counseling Psychology, 45, 304–314.CrossRefGoogle Scholar
  39. Rice, K. G., & Mirzadeh, S. A. (2000). Perfectionism, attachment, and adjustment. Journal of Counseling Psychology, 47, 238–250.CrossRefGoogle Scholar
  40. Saboonchi, F., & Lundh, L. G. (1997). Perfectionism, self-consciousness, and anxiety. Personality and Individual Differences, 22, 921–928.CrossRefGoogle Scholar
  41. Saboonchi, F., Lundh, L. G., & Öst, L. G. (1999). Perfectionism and self-consciousness in social phobia and panic disorder with agoraphobia. Behaviour Research and Therapy, 37, 799–808.PubMedCrossRefGoogle Scholar
  42. Shafran, R., Cooper, Z., & Fairburn, C. G. (2002). Clinical perfectionism: A cognitive-behavioural analysis. Behaviour Research and Therapy, 40, 773–791.PubMedCrossRefGoogle Scholar
  43. Slade, P. D., & Owens, B. G. (1998). A dual process model of perfectionism based on reinforcement theory. Behavior Modification, 22, 372–390.PubMedCrossRefGoogle Scholar
  44. Slaney, R. B., Rice, K. G., Mobly, M., Trippi, J., & Ashby, J. S. (2001). The revised almost perfect scale. Measurement and Evaluation in Counseling and Development, 34, 130–145.Google Scholar
  45. Srinivasagam, N. M., Kaye, W. H., Plotnicov, K. H., Greeno, C., Weltzin, T. E., & Rao, R. (1995). Persistent perfectionism, symmetry, and exactness after long-term recovery from anorexia nervosa. American Journal of Psychiatry, 152, 1630–1634.PubMedGoogle Scholar
  46. Stumpf, H., & Parker, W. D. (2000). A hierarchical structural analysis of perfectionism and its relation to other personality characteristics. Personality and Individual Differences, 28, 837–852.CrossRefGoogle Scholar
  47. Terry-Short, L. A., Owens, R. G., Slade, P. D., & Dewey, M. E. (1995). Positive and negative perfectionism. Personality and Individual Differences, 18, 663–668.CrossRefGoogle Scholar
  48. Wishart, D. (1987). CLUSTAN. User manual. Cluster analysis software. Computing Laboratory, University of St. Andrews.Google Scholar

Copyright information

© Springer Science+Business Media, LLC 2007

Authors and Affiliations

  • Lars-Gunnar Lundh
    • 1
  • Fredrik Saboonchi
    • 2
  • Margit Wångby
    • 1
  1. 1.Department of PsychologyLund UniversityLundSweden
  2. 2.H. M. Queen Sophia University College of NursingStockholmSweden

Personalised recommendations