Differentiating Adaptive and Maladaptive Perfectionism on the MMPI-2 and MIPS Revised

Although conceptualizations of perfectionism have emphasized adaptive as well as maladaptive expressions of the construct, how these different dimensions or types of perfectionists might be reflected in comprehensive personality assessment instruments is unknown. An initial sample of 267 university students completed the Minnesota Multiphasic Personality Inventory–2 (MMPI–2; Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 2001), Millon Index of Personality Styles Revised (MIPS–R; Millon, 2004), and Almost Perfect Scale–Revised (Slaney, Mobley, Trippi, Ashby, & Johnson, 1996). Analyses indicated that dimensions and types of perfectionism were associated, in expected directions, with select scores on the MMPI–2 and MIPS–R.

[21] defined perfectionism as "the striving for flawlessness, and extreme perfectionists are people who want to be perfect in all aspects of their lives" (p. 5). Flett and Hewitt went on to acknowledge that different researchers define perfectionism in different ways. The "unidimensional camp" ([21], p. 10) and the "clinical perfectionism" ([41], p. 778) camp have understood perfectionism as a set of beliefs or performance expectations that are problematic in one or more salient life domains. The multidimensional camp has argued that perfectionism is comprised of multiple personal dimensions, or combinations of personal and interpersonal dimensions, that provide important descriptions of the construct not captured in unidimensional conceptualizations.

Slaney, Rice, and Ashby (2002) and Slaney, Rice, Mobley, Trippi, and Ashby (2001) have argued that perfectionism should be understood primarily as very high standards or expectations for personal performance, operationalized as the High Standards subscale on their Almost Perfect Scale–Revised (APS–R; Slaney, Mobley, Trippi, Ashby, & Johnson, 1996). Because high performance expectations could be indicative of a functional or positive pursuit of excellence, [43] argued that it is important to differentiate what might be adaptive/healthy/functional perfectionism from maladaptive/unhealthy/dysfunctional perfectionism. [43] suggested the key differentiating factor between a healthy form of perfectionism and unhealthy perfectionism is "the perception that one consistently fails to meet the high standards one has set of oneself," a dimension referred to as "discrepancy" (p. 69). [42] operationalized the perceived gap between performance standards and one's self-evaluation of adequacy in meeting those standards as the Discrepancy subscale on the APS–R. Essentially, maladaptive/unhealthy perfectionists were conceptualized as people who have very high personal performance standards (elevated High Standards) and at the same time characteristically tend to be self-critical in their self-evaluations or perceived adequacy in meeting their standards (elevated Discrepancy). In contrast, adaptive perfectionists tend to have low, negligible levels of self-criticism (low Discrepancy), although their personal standards for performance would be comparable to those of maladaptive perfectionists (elevated High Standards). [35] used frequency distributions, cluster analysis, discriminant function analysis, and receiver operating curve analyses to develop an empirically supported method of classifying perfectionists based on Slaney et al.'s (1996, 2001) conceptualization and operationalization of perfectionism. A primary interest in this study involves construct validity evidence for the APS–R. Therefore, although several other measures and conceptual models of perfectionism have been advanced, in this study, we emphasize the [43] model and the Discrepancy and High Standards subscales derived from the APS–R.

Multidimensional studies of perfectionism often have provided support for a dimension of positive perfectionistic strivings or healthy perfectionism in addition to a dimension of negative perfectionistic concerns. As reported by [45], in studies when the overlap between the two dimensions is controlled, the perfectionistic strivings factor has been positively associated with a host of desired psychological and academic outcomes such as self-esteem, life satisfaction, positive affect, state pride, emotional regulation, exam performance and academic achievement, study behavior, school motivation, career decision-making self-efficacy, active coping, secure adult attachment, interpersonal adjustment, and perceived social support. In contrast, the dimension of perfectionistic concerns has been linked to numerous indicators of maladjustment and problematic functioning including depression, suicidal ideation, anxiety, stress, eating disorders, emotional dysregulation, recurrent physical pain and other medical problems, insecure adult attachment, marital and premarital difficulties, and less desirable academic performance ([12]).

Perfectionistic characteristics have also been correlated with other major personality dimensions such as those derived from the Five-factor model (FFM; Costa & McCrae, 1992) of personality. For example, the FFM dimension of conscientiousness positively correlates with the APS–R High Standards subscale (r =.46–.48), whereas the FFM neuroticism dimension positively correlates with Discrepancy (r =.59–.65; Rice, Ashby, & Slaney, 2007; for studies that have used other perfectionism measures resulting in similar findings, see Dunkley, Blankstein, Zuroff, Lecce, & Hui, 2006; Enns & Cox, 2002).

Discrepancy scores on the APS–R overlap with the FFM neuroticism dimension, and High Standards scores on the APS–R overlap with conscientiousness, but it is important to point out that these or similar perfectionism dimensions have accounted for additional variance in outcome criteria over and above variance attributable to relevant FFM dimensions. For example, after accounting for neuroticism and conscientiousness, an additional 9% to 13% of additional variance in self-esteem was attributed to APS–R High Standards and Discrepancy scores ([35]), whereas another 3% to 5% of variance in depression scores was attributable to self-critical perfectionism after accounting for neuroticism ([17]; [18]). Conceptually, it may be more accurate to think of these perfectionism characteristics as facets or lower order dimensions associated with, or influenced by higher order constructs from the FFM ([46]). Indeed, findings from other studies have indicated that a sharper focus on lower order personality traits rather the FFM's higher order dispositions may prove more fruitful in clarifying personality-behavior links. There is now convincing evidence that narrow, lower order personality traits provide substantial incremental validity in the prediction of a variety of behaviors above and beyond what can be attributed to the higher order FFM dimensions ([33]).

Despite refinements in conceptualizations and measures of perfectionism, interpretive texts for major personality assessment instruments continue to retain a unidimensional interpretation of perfectionism that is potentially less informative than a multidimensional conceptualization. These sources refer simply to perfectionism in interpretation and diagnostic guidelines without recognizing that perfectionism could be multidimensional and could reflect adaptive or positive characteristics as well as the negative or maladaptive characteristics. For example, elevated scores in the clinical (i.e., pathological) range on some scales from the Minnesota Multiphasic Personality Inventory–2 (MMPI–2; Butcher, Dahlstrom, Graham, Tellegen, & Kaemmer, 2001) have been interpreted as indicative of perfectionism. [24] described people with elevations on the MMPI–2 Scale 7 (Psychasthenia) as having "high standards of behavior and performance for themselves and others. They are likely to be perfectionist [sic] and conscientious; they may feel guilty about not living up to their own standards and depressed about falling short of goals" (p. 81). Although the latter aspects of this interpretation are consistent with maladaptive perfectionism, which could reasonably result in an elevation of Scale 7 on the MMPI–2, the former aspects of the interpretation could be describing adaptive or healthy perfectionistic qualities (see also [32], pp. 160–161).

Therefore, one purpose of this study was to examine the association between dimensions of perfectionism derived from the APS–R ([42]) and two comprehensive personality measures, the MMPI–2 and the Millon Index of Personality Styles Revised (MIPS–R; Millon, 2004). Both of the comprehensive instruments measure some aspect of perfectionism or a related relevant construct. Although the MMPI–2 contains numerous scales to measure a wide range of adjustment and personality characteristics, several scales are thought to aid in the specific assessment of perfectionistic personality characteristics and related qualities. The scales of most relevance, based on constructs assessed, to this study include two of the basic Clinical scales (Scale 2, Depression, and Scale 7, Psychasthenia) and their Restructured Clinical (RC) scale counterparts (RC2, Low Positive Emotions, and RC7, Dysfunctional Negative Emotions). Among other things, Scales 2 and RC2 tap personality characteristics such as self-deprecation and pessimism (likely associated with maladaptive perfectionism) and minimal to absent achievement motivation (likely associated, inversely, with adaptive perfectionism). Perfectionistic personality characteristics have been described for elevations of Scales 7 and RC7 as well. In addition to Graham's (2006) description, [32] noted that individuals with elevations on Scale 7 "are vulnerable to failures to live up to their own rigid expectations" (p. 161). Because the MMPI–2 is a problem and psychopathology-oriented instrument, we expected that Discrepancy would be positively associated with each of the preceding scales, and correspondingly that maladaptive perfectionists would score higher than adaptive perfectionists on these scales. In contrast, because a dimensional assessment of high performance expectations could be expected to be positively associated with well-being (in the case of adaptive perfectionism) or associated with maladjustment (in the case of maladaptive perfectionism), we could not posit a specific, directional hypothesis for associations with High Standards.

Scale development research has also produced a set of five, higher order personality dimensions tapped by the MMPI–2 that are relevant to this study. The MMPI–2 Personality Psychopathology Five (PSY–5; Harkness, McNulty, & Ben-Porath, 1995) scales are labeled Aggressiveness, Psychoticism, Disconstraint (DISC), Negative Emotionality/Neuroticism (NEGE), and Introversion/Low Positive Emotionality (INTR). The scales were developed from studies of normal personality and personality disorders, consistent with this study's interest in adaptive as well as maladaptive aspects of perfectionism. Egger, De Mey, Derksen, and van der Staak (2003) reported positive correlations in the moderate to large range between Neuroticism and NEGE and INTR and negative correlations between Conscientiousness and NEGE and INTR. There would be a conceptual warrant to expect an association between DISC scores and Conscientiousness because lower scores on DISC would be consistent with being a careful, nonimpulsive, constrained rule follower ([8]). Bagby, Ryder, Ben-Dat, Bacchiochi, and Parker (2002) reported a positive correlation between NEGE scores and a measure of obsessive–compulsive personality disorder (OCPD). The INTR scores were also correlated with OCPD scores in the [3] study, but the correlation with DISC scores was not significant.

In this study, we expected Discrepancy scores, as assessed by the APS–R, to be associated with both NEGE and INTR, and High Standards scores to be associated (inversely) with DISC. We also explored the association between APS–R Discrepancy and High Standards dimensions with two MMPI–2 Content scales (DEP, Depression, LSE, Low Self-Esteem) after partialling any effects associated with NEGE and INTR to determine whether the perfectionism dimensions accounted for additional variation after controlling for higher order personality dimensions akin to the FFM. In comparisons of groups of perfectionists, we expected maladaptive perfectionists to have a substantially more problematic pattern of MMPI–2 score elevations compared with adaptive perfectionists, although both adaptive and maladaptive perfectionists would be defined as having very high personal standards.

Similar to the MMPI–2, there are several MIPS–R scales that seem especially attuned to perfectionistic characteristics. For example, the description of the Pain-Avoiding Motivating Style scale indicates that high scorers perceive "the past as having been personally troubling, they always seem to be waiting for something else to go wrong, and feel that things are likely to go from bad to worse. They are easily upset by minor concerns and disappointments" ([31], p. 3).

This description is similar to qualitatively derived portrayals of maladaptive perfectionists (e.g., [37]); therefore, in this study, we expected Discrepancy scores on the APS–R to be positively associated with Pain-Avoiding scores. Likewise, maladaptive perfectionists would be expected to score higher on this dimension than adaptive perfectionists. People with elevations in the Conservation-Seeking Thinking Style are described as individuals who "transform new knowledge in line with what is known and are careful, if not perfectionistic, in arranging even minor details. As a result, they are seen by others as orderly, conservative, and traditional" ([31], p. 3). Elevations on the Dutiful/Conforming Behaving style scale are consistent with people who, among other things, "tend to behave in a formal and proper manner in social situation, and are unlikely to be self-expressive or to act spontaneously" ([31], p. 4). [31] went on to describe the Dutiful/Conforming style as consistent with "Leary's responsible-hypernormal personality, with its ideal of proper, conventional, orderly, and perfectionistic behavior" (p. 26). Thus, these latter two dimensions on the MIPS–R appear to tap more characteristics that would be positively associated with High Standards scores on the APS–R. These MIPS–R dimensions appear to be adaptive and not necessarily problematic characteristics and likely would be elevated among adaptive perfectionists compared with maladaptive perfectionists. The MIPS–R also contains a Clinical Index, described as a "rough gauge of a client's mental health status" ([31], p. 75); lower scores correspond with more significant psychological distress. The Clinical Index is expected to be inversely associated with Discrepancy scores. Maladaptive perfectionists are expected to score lower on this index than adaptive perfectionists.

A total of 267 undergraduate students (91 men, 176 women) participated in the study. Ages of participants ranged from 18 to 27 years (M = 18.84, SD = 1.15); 80% of the sample was 18 or 19 years old, and only three participants were over the age of 22. Students attended a large, public, land-grant university in the southeastern United States. In this sample, 56% of the students identified themselves as White/European American, 15% were Black/African American, 15% were Hispanic/Latino/Mexican American, 7% were Asian/Asian American, 5% identified themselves as Multicultural Mixed Race, and 2% reported "other." This racial/ethnic distribution approximates the population of undergraduate students on the selected campus; approximately 34% of the students represent U.S. racial/ethnic minorities. Although students were recruited from undergraduate psychology courses, 18 different colleges in the university were represented in the sample. The largest group of the students (36%) indicated their major would be within liberal arts and sciences majors, another 10% were in health and human performance, 8% were in journalism, 7% were in agriculture and life sciences, 6% were business, 6% were in nursing, and the following added another 5% each to the distribution: education, engineering, and other health professions. Approximately 5% indicated they had not yet decided on a college or major.

Participants completed the 23-item APS–R ([42]), which is a self-report measure of three perfectionism dimensions: Discrepancy, High Standards, and Order. The Discrepancy subscale (12 items) measures the perceived difference between one's personal performance expectations and performance self-evaluations (e.g., "I am not satisfied even when I know I have done my best."). The High Standards subscale (7 items) taps one's level of performance expectations (e.g., "I set very high standards for myself."). The Order subscale (4 items) measures one's preference for structure and neatness (e.g., "Neatness is important to me.") and is less relevant when using the APS–R to create adaptive, maladaptive, and nonperfectionist typologies or dimensions ([35]; [45]). The direction of scoring is such that higher subscale scores indicate higher personal standards and larger discrepancies between expected performance and evaluation of meeting self-imposed standards. Numerous studies have provided psychometric support for the APS–R, especially in its use with college student samples (e.g., [35]; [44]; [48]).

The MMPI–2 ([8]) consists of 567 true–false items designed for the assessment of numerous personality and adjustment indicators. The MMPI–2 is the most widely used clinical personality instrument and contains numerous scales to measure adjustment and personality characteristics. Research supports the reliability and validity of most MMPI–2 scales ([24]). For the scales of interest in this study (Scales 2 and 7, RC2 and RC7, the PSY–5 scales, and DEP and LSE), previous studies have yielded test–retest reliability coefficients ranging from approximately.65 to.85 for the scales, in comparable samples most relevant to this study ([8]; [10]; [22]; [49]). Several studies have supported the use of the MMPI–2 with nonclinical college student samples (e.g., [9]; [10]; [29]).

The MIPS–R ([31]) is a 180-item measure designed to assess "personality styles of normally functioning adults between the ages of 18 and 65+" ([31], p. 1). The measure serves as an operationalization of Millon's (2004) personality theory, which draws from psychodynamic, evolutionary, and social learning principles to organize personality as a function of interwoven motivating, thinking, and behaving styles. Among other things, the MIPS–R is described as a measure useful to "normality-oriented research investigators" ([31], p. x).

The MIPS–R contains 12 pairs of bipolar scales. Three pairs of scales correspond to Motivating Styles, designed to reflect patterns of reinforcement from the environment. Four pairs of scales assess Thinking Styles, reflecting styles of processing information. Five pairs of scales measure Behaving Styles, or how an individual typically interacts with others. For this study, we analyzed only results pertaining to the Pain-Avoiding Motivating Style, Conservation-Seeking Thinking Style, and Dutiful/Conforming Behaving Style. In addition to scales measuring theoretically derived personality dimensions, the MIPS–R contains three scales to detect distorted responding as well as a Clinical Index, which is described as a "rough gauge of a client's mental health status" ([31], p. 75). Lower scores on the Clinical Index suggest the respondent might be experiencing significant psychological distress. The MIPS–R has two normative samples. The college sample (N = 1,600) is most relevant to this study and was similar in age to our study sample. MIPS–R scores in the college student sample demonstrated adequate reliability. Scores on the MIPS–R and the original MIPS ([30]) have been shown to correlate with other personality scales as well as with measures of adjustment and mental health ([31]).

The study was approved by the university's institutional review board. Participants were recruited from the General Psychology research pool as well as from other undergraduate psychology courses. Participants enrolled in general psychology receive course credit for participating in research or can obtain the same level of credit for completing alternative activities. Participants in other undergraduate psychology courses received extra credit for participating in the research, although the amount of that credit would likely vary by instructors. Potential participants indicated their interest by signing up for a single experimental session. No more than 20 students participated during a single session, and all sessions were scheduled in classrooms with seating for 30 to 50 students. After the initial informed consent process, participants were provided with a packet of questionnaires and instructions for completing the questionnaires were read to the students. We assigned one of three different sequences of the measures randomly to each participant (in later preliminary analyses, there were no significant mean differences resulting from order effects). The research assistant remained in the classroom to answer any questions and monitor the session, which typically lasted 90 min.

There was no imputation for missing values at the item level. We first examined the associations between perfectionism and two comprehensive measures of personality with Pearson correlations between the APS–R Discrepancy and High Standards subscales and the most relevant subscales from both the MMPI–2 and the MIPS–R. To determine whether perfectionism dimensions accounted for additional predictive value in adjustment measures above and beyond higher order personality constructs, we conducted hierarchical multiple regression analyses in which we entered theoretically relevant personality constructs (measured by MMPI–2 PSY–5 scales) first in the equations, followed by the APS–R perfectionism dimensions. In these analyses, we used the MMPI–2 DEP and INTR Content scales as dependent variables.

For the final set of analyses, we classified participants into adaptive, maladaptive, and nonperfectionist categories on the basis of APS–R classification scoring rules ([35]). Briefly, those rules result in participants being classified as perfectionists if they have extremely elevated High Standards scores. Then, to further differentiate that group, participants with at least moderately high Discrepancy raw scores would be classified as maladaptive perfectionists, whereas those with low Discrepancy scores would be classified as adaptive. We conducted separate multivariate analyses of variance (MANOVAs), first with relevant scales from the MMPI–2 and then with relevant scales from the MIPS. We probed significant results with follow-up analyses of variance (ANOVAs) and the Ryan–Einot–Gabriel–Welsch (REGWQ) technique for pairwise comparisons.

As an initial step, and following recommendations from [8] for respondents in nonclinical settings and other research on nonclinical samples (e.g., [3]), we examined MMPI–2 Validity scales and included participants who obtained a raw score of less than 30 on the Cannot Say scale (CNS; items omitted or double marked) and who met the following criteria for T scores: Variable or True Response Inconsistency scales (VRIN or TRIN < 80), Infrequency (F < 80), Back F scale (FB < 90) and FB was less than 30 T-score points different from F (FB 30T ≥ F and FB ≥ 90T), Infrequency-Psychopathology (FP < 100), Lie scale (L < 80), Correction (K < 75), and Superlative Self-Presentation (S < 75). A total of 222 met these inclusion criteria. The elimination of 17% of the respondents is comparable to other MMPI–2 research on nonclinical samples (e.g., [3]). On the MIPS–R, inclusion criteria (see [31]) were that responses occurred to at least 90% of the items (less than 18 missing responses) and the Consistency score was ≥ 3. The test manual does not provide definitive score cutoffs for Positive Impression and Negative Impression scores. Another 21 participants did not meet the inclusion criteria for the MIPS–R, resulting in a usable sample of 201 participants for the remaining analyses (61 men, 140 women).

Descriptive sample statistics appear in Table 1. Not all possible scales are presented; only the most relevant MMPI–2 and MIPS–R scores given the purposes of this study are displayed. MMPI–2 scores are without K corrections. Non-K-corrected scores were used based on recommendations from [9] and research indicating that, in settings where defensiveness is less likely, K does not appear to act as a suppressor and therefore Clinical scales do not require the K correction ([4]). Means, standard deviations, and internal consistency estimates for all scores were comparable to other studies or test manual data on college student samples (e.g., [5]; [26]; [29]; [31]; [47]; [51]).

Table 1 Descriptive statistics.

The correlation between the APS–R Discrepancy and High Standards subscales scores was comparable to findings in other studies of college students, r = −.03. Correlations between the APS–R scores and the selected MMPI–2 and MIPS–R scores are presented in Table 2. In the analyses involving MMPI–2 scores, all but one of the correlations with Discrepancy were statistically significant, positive, and in the medium to large range in terms of effect sizes. All but two of the correlations between High Standards scores and MMPI–2 scales were significant and negative, with effect sizes in the small to medium range. The overall pattern of correlations suggests the problems associated with Discrepancy were rather pervasive. In contrast, High Standards scores were indicative of either better psychological functioning or perhaps the absence of problematic personality characteristics and maladjustment.

Table 2 Correlations between APS–R subscales and select MMPI–2 and MIPS–R scales.

To address the question of whether the perfectionism dimensions contribute substantially to personality or adjustment over and above what can be attributable to higher order personality factors, we used hierarchical regression analyses to examine incremental variance in the MMPI–2 DEP and LSE Content scales explained by the APS–R scores after controlling for the PSY–5 NEGE and INTR scores. Discrepancy (β =.22, p <.001) and High Standards (β = −.09, p <.05) combined to account for significant additional variation in DEP scores after controlling for the NEGE and INTR scores; R2 =.04, p <.001. Nearly identical results were obtained for LSE: Discrepancy (β =.21, p <.001) and High Standards (β = −.13, p <.05) combined to account for significant additional variation after partialling effects for NEGE and INTR scores; R2 =.04, p <.001.

The mostly small effect sizes for correlations involving High Standards might be the result of High Standards performing "double duty" in personality and adjustment. In and of themselves, higher performance expectations may not be problematic, and may even be beneficial, unless high standards are joined with a persistent tendency to be self-critical or to perceive a painful gap between expectations and self-evaluation of performance (i.e., Discrepancy). Therefore, we classified participants into adaptive and maladaptive perfectionist categories on the basis of APS–R classification scoring rules ([2]; [35]). We classified Students with High Standards raw scores ≥ 42 as perfectionists. Within that group, we classified students with Discrepancy raw scores ≥ 42 as maladaptive (n = 46; 10 men and 36 women); and we classified those with Discrepancy scores < 42 as adaptive (n = 67; 15 men and 52 women). We categorized those with low scores on High Standards as nonperfectionists (n = 88; 36 men and 52 women). There was a significant difference in the distribution of men and women across these categories, χ2(2, _I_N_i_ = 201) = 8.26, p <.05. About one third of the adaptive and maladaptive perfectionists were men, but nearly 70% of the nonperfectionists were men. For descriptive purposes, average MMPI–2 scores for all three groups are displayed in Table 3.

Table 3 Comparison of perfectionist groups on MMPI–2 scales.

We conducted three MANOVAs for the three sets of MMPI–2 dependent scores (Scales 2 and 7, RC2 and RC7, and three PSY–5 scales). We followed significant multivariate effects by univariate analyses, and we further probed them using the REGWQ test for pairwise comparisons, a post hoc procedure that "modifies Tukey's test to make it more powerful without allowing αEW to creep above whatever value (usually.05) is set for the test" ([13], p. 370).

There was a significant multivariate effect in the analysis of the MMPI–2 Scales 2 and 7, Wilks's Λ = 0.790, F(4, 394) = 12.32, p <.0005. Univariate analyses revealed significant differences (p <.0005) between the perfectionism groups: for Scale 2 (Depression), F(2, 198) = 19.16; and for Scale 7 (Psychasthenia), F(2, 198) = 18.80. Post hoc tests revealed significant differences between all three groups on both of the MMPI–2 scales, with the maladaptive perfectionists scoring the highest and the adaptive perfectionists scoring the lowest (see Table 3). Effect sizes of the differences between the adaptive and maladaptive perfectionists were both over a full standard deviation, d = 1.21 and 1.12 for Scale 2 and 7, respectively.

Although none of the average scores exceeded clinically significant cutoffs in this nonclinical sample, it is interesting to note that most (76%) of the maladaptive perfectionists were disproportionately overrepresented among participants who scored in the high (defined on the MMPI–2 as T score ≥ 65) or moderate (T score from 55–64) range on MMPI–2 Scale 2; about 16% of the adaptive perfectionists and 39% of the nonperfectionists scored within these ranges, χ2(6, _I_N_i_ = 201) = 47.87, p <.0005. There was also significant imbalance in the frequencies of perfectionists and nonperfectionists for Scale 7 groupings, χ2(6, _I_N_i_ = 201) = 34.65, p <.0005. About 74% of the maladaptives scored within the cutoffs for moderate or high ranges on Scale 7, whereas 28% of the adaptives and 56% of the nonperfectionists scored in those ranges.

Significant differences between perfectionism groups emerged in the analysis of the RC scales of interest (RC2 and RC7), Wilks's Λ = 0.817, F(4, 394) = 10.45, p <.0005 (see Table 3). Univariate tests confirmed significant differences for each of the two scales: F(2, 198) = 18.16 and F(2, 198) = 8.91, for the RC2 and RC7 scales, respectively, p <.0005. Post hoc tests revealed the same pattern of differences for both scales in that adaptive perfectionists scored significantly lower than the other two groups, with no significant differences emerging between maladaptive perfectionists and nonperfectionists. Effect sizes for the differences between the adaptive and maladaptive perfectionists were large, d = 1.07 and 0.83 for RC2 and RC7, respectively.

There were significant differences between groups on the three PSY–5 Scales of interest, Wilks's Λ = 0.760, F(6, 392) = 9.63, p <.0005 (see Table 3). Significant univariate differences emerged for each of the scales: DISC, F(2, 198) = 5.12, p <.01; NEGE, F(2, 198) = 13.67, p <.0005; INTR, F(2, 198) = 11.84, p <.0005. There was no significant difference between adaptive and maladaptive perfectionists on DISC scores, and both groups had lower average scores than nonperfectionists. Maladaptive perfectionists were significantly higher than the other groups on NEGE followed by nonperfectionists and then adaptive perfectionists (the difference between adaptive and maladaptive perfectionists was large, d = 1.00). On the INTR, there were no significant differences between maladaptive perfectionists and nonperfectionists, with adaptive perfectionists scoring significantly lower than the other two groups. Again, the effect size of the difference in INTR scores between adaptive and maladaptive perfectionists was large, d = 0.88.

There was a large, positive correlation between Discrepancy scores and the MIPS–R Pain-Avoiding Motivational Style (see Table 2). The correlations between Discrepancy and the Conservation-Seeking and Dutiful/Conforming scores were significant, negative, and small in terms of effect sizes. A different pattern emerged with High Standards scores. Large, significant, and positive correlations emerged between High Standards and Conservation-Seeking and Dutiful/Conforming scores, with a much smaller, albeit statistically significant, association observed with the Pain-Avoiding Motivational Style. The two perfectionism dimensions also related differently to the Clinical Index. High Standards was positively and moderately associated with Clinical Index scores (indicating higher standards were associated with better mental health), and Discrepancy was substantially and inversely associated with the Clinical Index.

We next divided participants into three perfectionist groups (as outlined previously) and analyzed group differences in the three MIPS–R scores. Descriptive statistics for these analyses appear in Table 4. We conducted a MANOVA for the three MIPS–R scales, and we conducted a separate univariate ANOVA on the Clinical Index score (the Clinical Index is based on the MIPS–R scales and as a nonindependent variable; it requires a separate analysis). The MANOVA was significant, Wilks's Λ = 0.682, F(6, 392) = 13.76, p <.0005, as were follow-up univariate ANOVAs. Significant between group effects emerged for Pain-Avoiding, Conservation-Seeking, and Dutiful/Conforming, F(2, 198) = 19.10, 21.45, and 18.08, p <.0005, respectively. Post hoc tests revealed significant differences between all three groups on the Pain-Avoiding scale, with adaptive perfectionists scoring lowest and maladaptive perfectionists scoring highest; the size of the difference between adaptives and maladaptives was quite large, d = 1.25. On the Conservation-Seeking and Dutiful/Conforming scales, both perfectionist groups scored comparably, and both had significantly higher average scores compared with nonperfectionists. The effect size of the difference between the perfectionists (combined adaptive and maladaptive) and nonperfectionists was large, d = 0.88 and 0.80 for Conservation-Seeking and Dutiful/Conforming, respectively. There were also significant differences between groups on the Clinical Index, F(2, 198) = 24.19, p <.0005, with post hoc tests revealing substantially greater mental health functioning for the adaptive perfectionists compared with the other two groups. The difference between adaptive and maladaptive perfectionists was large, d = 1.45. No significant differences between maladaptive perfectionists and nonperfectionists were observed in the analysis of Clinical Index scores (see Table 4).

Table 4 Comparison of perfectionist groups on MIPS–R scales.

One major purpose of this study was to examine the association between perfectionism, as measured by the APS–R, and scales from two personality assessment instruments to better understand how adaptive and maladaptive perfectionism might be detected at the level of comprehensive personality assessment. This was accomplished in several ways. First, we correlated scores from APS–R subscales with specific scales from the MMPI–2 and MIPS–R to determine relationships between the measures. We conducted regression analyses to examine the unique effects of perfectionism above and beyond higher order personality factors derived from the MMPI–2 (PSY–5 scales). Finally, we compared participants and MMPI–2 and MIPS–R scores between groups of adaptive perfectionists, maladaptive perfectionists, and nonperfectionists.

Because the MMPI–2 is most often used as a measure of clinical problems and psychopathology, we expected elevated scores on the MMPI–2 to be more indicative of maladaptive than adaptive perfectionism. Correlations between the APS–R subscales and the MMPI–2 scales supported this hypothesis. APS–R Discrepancy scores were substantially correlated with most of the targeted MMPI–2 scales, with directions of effects suggestive of problematic personality characteristics or maladjustment. Because scores on the Discrepancy scale of the APS–R are used to differentiate adaptive from maladaptive perfectionism, this finding is particularly significant. Consistent with past research that has indicated the harmful effects of perfectionistic concerns and maladaptive perfectionism ([34]), these results suggest that individuals with high Discrepancy scores likely experience depression, negative emotionality, and psychological tension.

By contrast, High Standards scores were inversely, if at all, correlated with the selected MMPI–2 scales, and the directions of effects suggested either no substantial relationship or an association indicative of adaptive psychological functioning. These contrasting patterns of correlations may shed some light on the dimensional nature of perfectionism. The results indicate that high standards, a defining feature of individuals labeled perfectionists, may be associated with high self-esteem, more positive emotions, less psychological tension, and overall indicates better psychological health. That is, perhaps one of the most essential elements of perfectionism (very high performance expectations) is, in itself, more closely related to indicators of healthy adjustment and personality characteristics than unhealthy adjustment and personality.

The findings lend support to a growing literature that has favored attention to what might be positive dimensions of perfectionism (see [45], for a review) as contrasted with the detrimental aspects of high performance expectations. Conceptualizations emphasizing the adaptive nature of perfectionistic high standards postulate that these standards become problematic when an individual perceives a large discrepancy between standards and performance ([43]). This distinction between maladaptive and adaptive perfectionism was supported by comparisons of perfectionist groups in this study. Significant differences emerged on all of the targeted scales after classifying participants into adaptive perfectionist, maladaptive perfectionist, and nonperfectionist categories.

This study also aimed to aid in the detection of perfectionism at the level of normal personality assessment. We expected that both adaptive and maladaptive aspects of perfectionism would be differentially related to MIPS–R scales in a manner consistent with the scales' conceptual definitions. This was evidenced by correlations between APS–R subscale scores and MIPS–R subscales. In particular, Discrepancy scores had a strong positive association with the MIPS–R Pain-Avoiding Motivating Style and a strong inverse association with the Clinical Index (indicating poorer mental health), whereas High Standards had a strong positive association with Conservation-Seeking and Dutiful/Conforming and a moderate positive correlation with the Clinical Index (indicating better mental health).

Comparison of perfectionist groups enabled us to identify scale elevations indicative of maladaptive versus adaptive aspects of perfectionism. Specifically, both maladaptive and adaptive perfectionists demonstrated higher scores on a Conservation-Seeking cognitive style and a Dutiful/Conforming behaving style than nonperfectionists. Results indicated that both groups are likely to adhere to traditional values and engage in orderly "rule-following" behavior. They are likely to be careful in arranging details and are somewhat less likely to act spontaneously ([31]). Furthermore, these results indicate that these scales may involve global aspects of perfectionism rather than maladaptive characteristics. Characteristics measured by these scales may be helpful in distinguishing perfectionists from nonperfectionists but do not appear to reflect problems that necessitate intervention. This interpretation should be taken into account during the process of assessment and treatment planning.

Although scores on the MIPS–R indicate that adaptive and maladaptive perfectionists share some important personality characteristics, these groups were distinguishable from one another by their scores on the Pain-Avoiding motivating style. High scorers on this scale are more motivated to avoid painful experiences than actively seek pleasurable experiences. Maladaptive perfectionists scored significantly higher than both nonperfectionists and adaptive perfectionists on this dimension. This suggests that maladaptive perfectionists are alert to threats to emotional and physical security including negative feedback. They may appear inhibited, restrained, and worrisome and may disengage from relationships and experiences to avoid painful outcomes. Adaptive perfectionists, on the other hand, appear to be motivated by positive goals and seek to enhance their lives through experiences and challenges that magnify their vitality. They are likely to be more active in shaping their circumstances to increase such experiences. This distinction is consistent with conceptualizations of perfectionism as driven either by positive or negative reinforcement ([50]). According to this conceptualization, although the outward behavior (setting high standards, achievement-oriented striving) of many perfectionists may look similar, some perfectionists are motivated to achieve positive reinforcement, whereas others are motivated to avoid failure. Psychological assessment aimed at distinguishing the differing patterns of motivation behind otherwise similar behavior may assist with the identification of potentially problematic types of perfectionism.

This study substantially extends previous research on the measurement of perfectionism. Although interpretive statements relevant to major personality assessment instruments reference a unidimensional perfectionism characteristic, the results of this study suggest that interpretations of these scale elevations could be informed by a multidimensional conceptualization of perfectionism. Elevations on certain scales, such as the MMPI–2 Clinical scales, seem to be influenced by maladaptive aspects of perfectionism; other scales (such as the MIPS–R Dutiful/Conforming Behaving Style) appear to tap into more global aspects of perfectionism that are not, in and of themselves, problematic. When examining dimensions of normal personality, both the MIPS–R Dutiful/Conforming scale and the Conservation-Seeking scale appear to measure characteristics that separate perfectionists from nonperfectionists; whereas the Pain-Avoiding Motivating Style appears to measure a more maladaptive aspect of perfectionism. Individuals with elevations on the Dutiful/Conforming and Conservation-Seeking scales are likely to exhibit clinical problems (as evidenced by lower scores on the Clinical Index and elevated MMPI–2 scales) when they also score highly on the Pain-Avoiding Motivating Style. This indicates that although elevations on scales thought to tap into perfectionism are typically interpreted using a unidimensional conceptualization, results may be more clinically useful when interpreted in terms of both adaptive and maladaptive dimensions of perfectionism.

As with any empirical study, there are limitations to keep in mind when considering this study's implications. Self-reported data were collected at a single time point in this correlational study, making it impossible to draw firm conclusions related to causality. For instance, it is possible that psychological distress (source of elevation on the MMPI–2 Clinical scales) leads people to perceive more discrepancy between their standards and their performance. A substantial body of research has conceptualized perfectionism as trait like and relatively stable over time ([21]), with several perfectionism-adjustment associations having also shown considerable stability across time ([15]; [34]; [34]). Thus, patterns of association between perfectionism and other dimensions of personality (and adjustment) from the MMPI–2 and MIPS–R could reasonably be presumed to be stable as well, although future replications should test that assumption. Indeed, research on personality trait consistency generally finds stability effects in the r =.50 to.70 range for major personality dimensions, with those correlations being relatively lower for younger samples and considerably higher for older adults ([39]; [52]). An extension of this study would be to examine the longitudinal patterning of perfectionism and perfectionism-linked indicators assessed with small personality measures (e.g., the APS–R) and comprehensive personality measures (e.g., MMPI–2), and the stability of perfectionism-adjustment associations over longer time frames or as a function of treatment effects over time (e.g., [53]).

Because in this study we used a sample comprised exclusively of college student participants, the broader generalizability of these findings may be limited. However, there are several reasons to consider this sample selection as a reasonable trade-off. There appears to be a rather high base rate of perfectionism in college settings ([35]). Perfectionism is also a relatively frequent reason for college students to seek treatment at university counseling centers ([11]). In fact, recent epidemiological research has indicated that 8% of young adult college students likely have OCPD, making it the most frequently occurring personality disorder within that population ([6]). According to the Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; American Psychiatric Association, 2000), one of the defining features of OCPD is maladaptive perfectionism. Although it can sometimes be problematic to conduct MMPI research with nonclinical populations, [9] concluded that these issues were minimal when the research was targeting a normal range of personality characteristics and/or disorders that have a high base rate, and this study appears to meet both those criteria. Similarly, although the higher proportion of females in the study's sample is somewhat of a limitation, this balance is similar to the population found in many university counseling centers ([16]; [23]). Of course, replication and extension of this study through data gathered in counseling centers or other clinical contexts would be important for future research.

Another possible limitation concerns the categorical classification of participants as adaptive perfectionists, maladaptive perfectionists, or nonperfectionists. Recent research that used other measures of perfectionism suggested that the latent structure of perfectionism is more likely to be dimensional than taxonic ([7]). Nevertheless, as noted by Broman-Fulks et al., findings that have supported the dimensionality of perfectionism have been based on measures that were not developed nor intended to identify types of perfectionists; to our knowledge, there is no published research applying taxometric methods such as those described by Ruscio, Haslam, and [40] to APS–R data. Although effect sizes were substantial for most of the differences observed between the three groups in this study, future research using taxometric methods with APS–R data will be useful in evaluating the appropriateness, accuracy, and clinical utility of classifying people according to APS–R cut scores.

Finally, in the broader field of personality research, some scholars ([27]; [27]) have suggested that the adaptiveness of personality characteristics involves an interaction between a relatively stable individual trait- and situation-specific factors. That is, a personality characteristic thought to be maladaptive (in this case, the tendency to perceive a discrepancy between one's standards and one's performance) may result in behavior that is actually adaptive in certain situations. Although the results of this study and other studies have seemed to support the maladaptiveness of elevated discrepancy scores across life domains, future research using multiple situation-specific measures could further clarify the construct of maladaptive perfectionism in other person–environment contexts.

Despite limitations, the findings of this study extend the literature on personality assessment in important directions. Associations were identified between perfectionism and other, related qualities and adjustment indicators derived from comprehensive personality measures. The study distinguished between those scales and scores that were characteristic of healthy (adaptive) aspects of perfectionism and those that were characteristic of unhealthy (maladaptive) perfectionism. In doing so, we highlighted the broader psychological and behavioral characteristics associated with perfectionism, and we advanced additional validity support for the APS–R. These findings may be especially helpful to clinicians hoping to identify perfectionism at the level of comprehensive psychological assessment and may substantially inform both interpretation and subsequent intervention.

We thank Dan Knippel, Brooke Watson, Melina Sevlever, Natalie Arcario, Davina Craig, Nicole Critelli, Paula Garces, Joey Hannah, Alexzandria May, Damita Miller, Jennifer Langille, Theresa Layug, Christine Lee, Lori Nabors, Michelle Nagri, Ashley Taylor, and Huan Ye for their assistance with this study. Preliminary results from this study were presented at the Annual Convention of the American Psychological Association in August 2007.