Psychological Distress and the Frequency of Perfectionistic …€¦ · Blankstein and Lisa Gray,...

Journal of Personality and Social Psychology1998, Vol. 75, No. 5, 1363-1381

Copyright 1998 by the American Psychological Association, Inc.0022-3514/98/13.00

Psychological Distress and the Frequency of Perfectionistic Thinking

Gordon L. FlettYork University

Paul L. HewittUniversity of British Columbia

Kirk R. Blankstein and Lisa GrayUniversity of Toronto at Mississauga

Five studies tested the hypotheses that there are individual differences in the frequency of automaticthoughts involving perfectionism and that these thoughts are associated with psychological distress.Research with the Perfectionism Cognitions Inventory (PCI) established that this new measure hasadequate psychometric properties, and high PCI scorers tend to spontaneously report perfectionisticthoughts in naturalistic situations. Additional research confirmed that frequent perfectionism thoughtsaccount for unique variance in distress, over and above variance predicted by standard measures ofnegative automatic thoughts and trait perfectionism measures. Overall, the findings support the viewthat personality traits involved in depression and anxiety have a cognitive component involvingruminative thoughts and that activation of this cognitive personality component contributes to distress.

Over the past two decades, there have been several importantdevelopments in research and theory on the role of automaticthoughts and related cognitive processes in emotion distress.Initial work testing predictions from Beck's (1967) cognitivemodel found that a preponderance of negative automaticthoughts was associated with depression (see Hollon & Kendall,1980). This work was supplemented by theoretical accounts andinvestigations of the role of an absence of positive thoughts indepression (Ingram & Wisnicki, 1988; Kendall, Howard, &Hays, 1989; Schwartz & Garamoni, 1986). More recently, No-len-Hoeksema and her colleagues have focused on a ruminativeresponse style as a coping tendency that contributes to the persis-tence of depression (Nolen-Hoeksema, Parker, & Larson, 1994).

In the current article, we reexamine the role of ruminativeprocesses in psychological distress by considering the argumentthat certain personality traits can also be assessed in terms offrequency of thinking. Specifically, the present research investi-gated the possibility that there are individual differences in the

Gordon L. Flett, Department of Psychology, Tfork University, Toronto,Ontario, Canada; Paul L. Hewitt, Department of Psychology, Universityof British Columbia, Vancouver, British Columbia, Canada; Kirk R.Blankstein and Lisa Gray, Department of Psychology, University of To-ronto at Mississauga, Ontario, Canada.

This research was supported by Grants 410-89-0335, 410-91-8056,410-91-1690, and 410-93-1256 from the Social Sciences and HumanitiesResearch Council of Canada, as well as by internal grants from theResearch and Program Evaluation Committee, Brockville PsychiatricHospital, and the University of Toronto.

We thank Marjorie Cousins, Caroline Fitzpatrick, Spomenka Koledin,Lawrence Levy, Silvana Macdonald, Thomas Martin, Shawn Mosher,Lois Ritchie, Dana Sawatzky, and Wendy Tumbull for their assistancein completing the research. .

Correspondence concerning this article should be addressed to GordonL. Flett, Department of Psychology, "York University, 4700 Keele Street,Tbronto, Ontario, Canada M3J 1P3. Electronic mail may be sent [email protected].

frequency of perfectionistic cognitions and that the tendency toexperience frequent perfectionism thoughts is associated withelevated levels of psychological distress.

The notion that perfectionists are characterized by frequentcognitions about the attainment of ideal standards is consistentwith descriptions of the cognitive preoccupations of introjectiveindividuals (Blatt & Shichman, 1983). Blatt and Shichman(1983) observed that introjective individuals ruminate exces-sively about failures to meet personal standards and maintain asense of control, whereas dependent, anaclitic individuals arebelieved to experience frequent thoughts about interpersonal ex-periences. A tendency to ruminate about the inability to attainperfectionistic standards is also in keeping with evidence indi-cating that the ideal self functions as a self-schema that facili-tates the recall of perfectionistic content (Hewitt & Genest,1990). According to Hewitt and Genest (1990), the ideal self-schema is especially likely to encode and process informationthat indicates that perfection has not been obtained.

Several additional sources converge to suggest that certainperfectionists should indeed be characterized by frequent rumi-nations that reflect the attainment of standards. For example, ata descriptive level, clinical accounts suggest that perfectionistsfrequently experience thoughts with perfectionistic themes (e.g.,Barrow & Moore, 1983; Guidano & Liotti, 1983). Empiricalevidence has been provided by Frost and associates. Frost andHenderson (1991) had a sample of 40 female athletes completevarious measures, including the multidimensional perfectionisminventory developed by Frost, Marten, Lahart, and Rosenblate(1990). These women also provided structured reports of theirthoughts 24 hr before competition. The authors found that per-fectionists reported a greater propensity to experience such cog-nitive reactions as "Images of me making a mistake clog mymind" and " I dream of being perfect." In a subsequent study,Frost et al. (1997) asked students to monitor their daily mis-takes. They found that perfectionists reported a greater fre-quency of thoughts about highly important mistakes.

1363

1364 FLETT, HEWITT, BLANKSTEIN, AND GRAY

In addition to descriptive and empirical evidence, theories ofgoal-directedness and cognitive processing (e.g., Klinger, 1977;Kuhl, 1984; Martin & lesser, 1989) and theories of stress andcognition (Pennebaker, 1989) suggest that perfectionism shouldhave a ruminative component. The link between perfectionismand ruminative thought can be analyzed within the context ofincompleteness theories of rumination (for a review, see Gold &Wegner, 1995). For example, Martin and lesser (1989) arguedthat rumination occurs after the frustration of not reaching asal ient goal, especially if a person has reached the stage of *' end-state thinking." According to these authors, end-state thinkingoccurs when "individuals think not about different ways ofattaining the goal, but about the goal objects themselves and thefeelings associated with them" (Martin & Tessei; 1989, p. 314).Ruminative thoughts and end-state thinking tend to emerge whena person cannot engage in instrumental behaviors that will movethe person closer to attaining his or her goal (Martin & Tesser,1989; Mclntosh, Harlow, & Martin, 1995). In the case of perfec-tionists, this would be especially likely to happen when it beginsto become apparent that perfection will not be attained. A similarformulation was provided by Klinger (1977), who argued thatan individual is likely to ruminate about goals that involve ahigh level of commitment but are beyond his or her reach (alsosee Klinger, 1996). According to these theoretical accounts, per-fectionists should be especially prone to rumination involvingthe attainment of their goals. Because perfection is an impossiblegoal to attain, yet the perfectionist is highly focused on andcommitted to his or her goals (Hewitt & Flett, 1991b), it standsto reason that perfectionists should experience numerousthoughts reflecting their failure to attain perfection, along withcounterfactual thoughts about "what might have been." Thisform of thinking may be conceptually similar to the negativeintrospection, ruminative processes, and analytical, high-levelthinking about the self that characterize individuals with persis-tent, elevated depression and high trait levels of negative affec-tivity (see Pennebaker, 1989; Watson & Clark, 1984).

Finally, the possibility that the perfectionism construct hasa cognitive element is in keeping with the vast literature onachievement motivation. Perfectionism is typically regarded asan achievement-based construct that involves the tenacious pur-suit of goals of personal importance (Hewitt & Flett, 1993b). Anextensive literature on achievement motivation and its correlatesindicates that cognitive imagery involving the attainment or non-attainment of goals is a key element of achievement motivation(see McClelland, 1961).

The current research was designed to extend the existingliterature in two respects. At present, although there has beensome initial work on cognitive processing of perfectionistic in-formation (Hewitt & Genest, 1990), there have been no attemptsto examine the frequency of perfectionistic cognitions as a per-sonality variable. Existing measures of perfectionism (e.g.,Burns, 1980; Frost et al., 1990; Hewitt & Flett, 1991b; John-son & Slaney, 1996) focus more globally on the degree of traitlevels of perfectionism with no attempt to measure levels ofongoing cognitive activity involving perfectionism and goal at-tainment, despite observations that perfectionists are preoccu-pied with frequent automatic thoughts and images involving theneed to be perfect (Frost & Henderson, 1991; Moore & Barrow,1986). Consequently, one goal of the present research was to

develop a reliable and valid measure of automatic perfectionisticthoughts. Such a measure would represent an indirect assess-ment of the frequency with which individuals make evaluativecomparisons at the cognitive level between the ideal, perfection-istic self and the current self or the current situation. A preoccu-pation with ruminative thoughts about the self is believed toreflect an underlying tendency to focus analytically on currentaspects of the self and the environment rather than goal-directedmodes of action (Kuhl, 1984).

A related purpose of the present research was to use thismeasure to test the hypothesis that frequent perfectionistic think-ing is associated significantly with psychological distress interms of elevated symptoms of depression and anxiety. Severalstudies have indicated that trait levels of perfectionism are asso-ciated with depression and distress in clinical and subclinicalsamples (e.g., Flett, Hewitt, Blankstein, & Mosher, 1991; Frostet al., 1990; Hankin, Roberts, & Gotlib, 1997; Hewitt & Flett,1991a, 1993a; Hewitt, Flett, & Ediger, 1996), leading majortheorists such as Blatt (1995) to describe perfectionism ashighly destructive. A tendency to engage in excessive perfection-istic thinking should be associated with depressive tendenciesbecause constant rumination about perfectionistic standardswould serve to highlight possible discrepancies between theactual self and the perfectionistic ideal self and could lead toheightened levels of self-focused attention and self-reflection.In addition, a tendency to engage in frequent perfectionisticthinking could promote cognitive biases such as "all or none"thinking (Burns, 1980). Finally, according to the availabilityheuristic (Tversky & Kahneman, 1973), frequent perfectionisticthinking could increase the likelihood that perfectionists willconsistently use stringent criteria when evaluating themselves.

Individual differences in the frequency of perfectionistic cog-nitions were investigated in a series of five studies. This researchaddressed the following issues: (a) Are there personality differ-ences in perfectionistic thought frequency, and can these differ-ences be assessed in a reliable manner? (b) Is there a relationbetween individual differences in perfectionistic thought fre-quency and poor psychological adjustment? and (c) If there isindeed a link between perfectionism cognitions and maladjust-ment, can levels of perfectionistic thought frequency accountfor unique variance in depressive symptomatology over andabove trait dimensions of perfectionism, neuroticism, and gen-eral measures of automatic thoughts?

Study 1

In Study 1, the primary focus was on developing the newmeasure—the Perfectionism Cognitions Inventory (PCI) —with the caveat that our main purpose was to understand moreabout the nature of perfectionism as a personality construct.This new measure was developed using the construct validationapproach (Jackson, 1970).

The participants in Study 1 completed the AutomaticThoughts Questionnaire (ATQ; Hollon & Kendall, 1980) and theAutomatic Thoughts Questionnaire-Positive (ATQ-P; Ingram &Wisnicki, 1988), in addition to the PCI. These measures wereincluded primarily to test the hypothesis that a high frequencyof perfectionism cognitions would be associated with a highlevel of negative automatic thoughts about the self. An associa-

PERFECTIONISTIC THINKING AND DISTRESS 1365

tion between perfectionistic thoughts and negative self-thoughtswould be in keeping with our belief that perfectionism cogni-tions arise, in part, from a concern that the present self is dis-crepant in a negative sense from the ideal self. In addition, thereare other indications that perfectionists are high in self-criticism(Blatt, 1995; Hewitt & Flett, 1990; Hewitt & Flett, 1993a) andthat they tend to be dissatisfied with their degree of goal attain-ment (Mor, Day, Flett, & Hewitt, 1995).

Finally, the participants in this study also completed a mea-sure of depressive symptomatology. In addition to allowing ex-amination of the relation between perfectionistic thought fre-quency and self-reported distress, the inclusion of this measureallowed us to test whether individual differences in perfection-istic thought frequency could account for unique variance indepression scores after taking into account variance due to theexisting measures of automatic thoughts.

Method

Materials and Procedure

We generated items for the PCI independently on the basis of availableliterature, experience with perfectionists, and general understanding ofthe perfectionism construct. Items were generated to reflect direct refer-ences to perfectionism (e.g., I should be perfect) and to upward strivingdefined in absolute terms (e.g., I can always do better, even if thingsare almost perfect) and in relative terms involving social comparison(e.g., I have to be the best). The social comparison content reflectedthe competitive orientation exhibited by extreme perfectionists (Flett,Hewitt, Blankstein, & Dynin, 1994). Additional PCI items tapped anindividual's awareness of being imperfect and not attaining high goals(e.g., Why can't I be perfect?). In total, 55 items were constructed.These items were corrected for clarity, duplicates were deleted, andsome items were rephrased. This resulted in a pool of 36 potential items.

Participants

Sample 1. hi the initial phase of this study, a sample of 234 under-graduate students (86 men and 148 women) in a lst-year psychologycourse were administered the 36 items. The mean age of the participantswas 21.23 years. The instructions were patterned after the previous ATQmeasures. Specifically, participants were told the following:

Listed below are a variety of thoughts about perfectionism thatsometimes pop into people's heads. Please read each thought andindicate how frequently, if at all, the thought occurred to you overthe last week. Please read each item carefully and circle the appro-priate number, using the scale below.

Participants made ratings varying from not at all (0) to all of the time(4) . Participants also completed the ATQ, the ATQ-P, and the BeckDepression Inventory (BDI; Beck, Rush, Shaw, & Emery, 1979).

The ATQ (Hollon & Kendall, 1980) is a 30-item scale that presentsindividuals with a series of negative thoughts. Respondents must indicatethe frequency with which they have experienced these negative thoughtsduring the past week.

The ATQ-P (Ingram & Wisnicki, 1988) consists of a series of 30positive thoughts. The ATQ-P was completed with respect to the thoughtsexperienced during the previous week. The BDI (Beck et al., 1979) isa 21-item instrument measuring the behavioral, cognitive, motivational,and vegetative symptoms of depression.

Sample 2. In the subsequent phase of this study, a sample of 747students (208 men and 539 women) from an introductory psychology

class at the University of Winnipeg completed the PCI as part of a largemass assessment Their mean age was 20.2 years. The version of thePCI used in this study was the final version of the measure, which isdescribed subsequently. The main purpose of Sample 2 was to assessthe structure of the PCI in a large sample.

Results

Psychometric Analyses and Item Reduction

Initially, analyses were performed to select the best itemsfrom the larger pool of items administered to the participantsin Sample 1. Following recommendations by Jackson (1970),items were retained if the mean item endorsement was not toolow (greater than 0.5) or not too high (lower than 3.5). Elevenitems with excessively high or low endorsements were droppedfrom further consideration. The remaining items were subjectedto a principal-components analysis. It was found that all 25items loaded .61 or higher on a large first component that ac-counted for 51.3% of the variance (eigenvalue = 12.83). Twosmaller components also emerged and accounted for 6.6% and5.0% of the variance. Because the items making up these compo-nents were complex in that they had significant loadings on thefirst component as well, it was decided to focus on the largefirst component.

The alpha coefficient (Cronbach, 1951) of the 25 PCI itemsin Sample 1 was .96. The mean interitem correlation was .49.This value approached but did not exceed the cutoff point of.50 that Briggs and Cheek (1986) identified as an indicationthat scale items are overly redundant.

There was no gender difference in mean PCI scores. Therespective means for men and women were 35.41 (SD ~ 23.19)and 38.25 (SD = 22.01).

The replicability of the previous results was investigated byconducting a principal-components analysis on the responses tothe 25 PCI items that were administered to the large sample of747 students from the University of Winnipeg. This analysisconfirmed that the scale consists of one large component withloadings ranging from .38 to .75, most loadings being .50 orgreater The 25 PCI items are displayed in Table 1, along withtheir loadings, means, and standard deviations. The eigenvaluefor this component was 9.39, and it accounted for 37.6% of thevariance. Overall, the results indicated that the PCI is essentiallyunidimensional. Three smaller components with eigenvalues of1.75, 1.48, and 1.23 were obtained, but each component hadfew items with significant loadings of .40 or greater, and theseitems had significant loadings on the first component as well asone of the other components. The components were not rotatedbecause we did not hypothesize any underlying causal structureinvolving two or more components.1

The respective means for men and women were 43.08 (SD= 17.98) and 43.91 (SD = 19.18). This difference was notstatistically significant.

Correlational Analyses

In the next set of analyses, we examined how perfectionismcognitions related to other measures of automatic cognitions

1 Highly similar results were obtained when separate factor analyseswere conducted on the data provided by women and men. These resultsare available from Gordon L. Flett on request.


Table 1Factor Loadings for Perfectionism Cognitions Inventory(PCI) Items From Sample 2, Study 1

Item

1. Why can't I be perfect?2. I need to do better3. I should be perfect4. I should never make the same

mistake twice5. I've got to keep working on my goals6. I have to be the best7. I should be doing more8. I can't stand to make mistakes9. I have to work hard all the time

10. No matter how much I do, it's neverenough

11. People expect me to be perfect12. I must be efficient at all times13. My goals are very high14. I can always do better, even if things

are almost perfect15. I expect to be perfect16. Why can't things be perfect?17. My work has to be superior18. It would be great if everything in my

life were perfect19. My work should be flawless20. Things are seldom ideal21. How well am I doing?22. I can't do this perfectly23. I certainly have high standards24. Maybe I should lower my goals25. I am too much of a perfectionist

Factor loading

.58

.57

.70

.60

.51

.67

.48

.70

.63

.67

.64

.69

.64

.66

.72

.63

.72

.58

.75

.47

.50

.46

.63

.38

.58

M

1.142.830.93

1.932.941.692.661.852.13

1.711.191.702.17

1.620.921.531.58

1.941.391.862.541.731.890.931.07

SD

1.201.031.19

1.281.061.311.881.251.22

1.371.281.231.31

1.351.181.361.28

1.391.191.241.131.171.271.141.22

Note. N = 747. Eigenvalue = 9.39; percentage of variance = 37.6%.Values are based on a principal-components factor analysis. PCI itemresponse range from 0 to 4.

and depression. The sum of responses to the 25 items retainedon the PCI was used as the measure of perfectionism cognitions.The correlations among the measures are shown in Table 2. Itcan be seen that the PCI was correlated positively with de-pressive symptoms in the total sample, r — .52, p < .001.Similar findings tended to emerge for men and women.

As can be seen in Table 2, individuals with more frequentperfectionism cognitions also reported more frequent negativeand positive automatic thoughts. A regression analysis with PCIscores as the outcome measure and both automatic thoughtsmeasures within the same predictor block indicated that signifi-cant variance was predicted by both negative automaticthoughts, F = 203.99,/? < .01, and positive automatic thoughts,F = 23.90, p < .01, but perfectionism cognitions were linkedmost closely with negative thoughts.

Hierarchical Regression With Cognitions MeasuresPredicting Depression

Finally, a hierarchical regression analysis was performed onthe data from Sample 1 to examine the unique contribution ofPCI scores when predicting depressive symptoms. A block ofpredictors composed of the three automatic thoughts measureswas entered and accounted for a significant amount of variance,

R2 = .53, F = 85.83, p < .001. Examination of the individualpredictors showed that all three measures accounted for signifi-cant variance in symptom scores in the expected direction. Therespective F values were 75.88 for ATQ scores {p = .54), 19.88for ATQ-P scores (0 = - .21) , and 10.48 for PCI scores (0 =.21;allps < .01).

Discussion

The results of our first study supported the hypothesis thatthere are identifiable individual differences in the frequency ofperfectionism cognitions and that these differences can be mea-sured with an adequate degree of reliability and validity. Initialresearch conducted with the PCI indicated that this measure isuni dimensional and has a high level of internal consistency whenadministered to student samples.

Additional analyses confirmed that there is a significant posi-tive association between the frequency of perfectionistic think-ing and self-reports of depressive symptoms. A tendency toexperience perfectionistic cognitions may not be entirely nega-tive in that high PCI scorers reported a greater frequency ofpositive automatic thoughts. At the same time, it must be ac-knowledged that a more robust association was obtained be-tween the PCI and the ATQ measure of negative self-thoughts.

The results of the regression analysis provided perhaps theclearest demonstration of the need to examine frequent perfec-tionistic thoughts in depression. Examination of the data fromparticipants who were administered various measures of auto-matic thoughts indicated that the PCI accounted for unique vari-ance in depression symptom scores after variance due to the

Table 2Correlations Among the Perfectionism Cognitions Inventory(PCI), Automatic Thoughts Questionnaire Measures,and Depressive Symptoms: Study 1

Measure 1

1. PCI2. ATQ3. ATQ-P4. BDI

.66***

.18**

.52***

Total sample

- .07.69*** .21**


74**.15.52**

Men

-.08.73** -.26*


.60***

.20*

.53***

Women

-.06.67** -.19*

Note. Correlations were analyzed using two-tailed tests and were basedon the responses of 86 men and 148 women. ATQ = Automatic ThoughtsQuestionnaire-Negative; ATQ-P = Automatic Thoughts Questionnaire -Positive; BDI = Beck Depression Inventory.*p < .05. **p < .01. ***/? < .001.


existing measures of automatic thoughts had been taken intoaccount. Thus, the initial data indicated that the tendency toengage in frequent ruminations about perfection is a correlateof depressive symptomatology in university students, and it isnot redundant with more general measures of positive and nega-tive automatic thoughts.

Study 2

The purpose of Study 2 was to examine the psychometricproperties of the PCI in greater detail in various samples. Thefirst sample in Study 2 completed the PCI, along with measuresof response style, Type A behavior; and self-punitiveness. Re-sponse style was assessed to determine the extent to which thePCI is contaminated by a tendency to engage in impressionmanagement. The remaining measures were used to establishconcurrent validity. Perfectionists have been described as havingType A features (Flett et a l , 1994; Westra & Kuiper, 1996),with a tendency to be self-punitive (Hewitt & Flett, 1991b).Thus, the PCI should be correlated significantly with these per-sonality tendencies.

Participants in our second sample completed the PCI at twotime points separated by a 3-month interval to allow examina-tion of the temporal stability of perfectionistic thought fre-quency. Evidence of test-retest reliability would suggest thatthe tendency to experience frequent perfectionistic thoughts isa relatively persistent feature associated with perfectionism, de-spite the general belief that automatic thoughts are states thatreflect the activation of latent self-schemas (Beck, Epstein, &Harrison, 1983).

A third sample of participants completed the PCI and thesame measure of impression management given to the first sam-ple to further examine this response bias issue. Also, participantsin this sample completed a measure of imaginal processes, aswell as measures of symptom distress, negative affect, positiveaffect, and life satisfaction. The measure of imaginal processeswas included to further examine the nature of perfectionisticthinking. Imaginal processes typically involve daydreams withvarious types of content. Past research on imaginal processesand depression has shown that depressed individuals tend toreport a greater frequency of daydreams involving themes ofguilt and fear of failure (Cundiff & Gold, 1979). Because day-dreams can be conceptualized as a cognitive tendency to thinkabout idealized circumstances or preferred outcomes (Singer &Antrobus, 1972), perfectionists with high PCI scores shouldhave higher scores on measures of imaginal processes. The mea-sure of imaginal processes used in this study included a subscaletapping fear of failure imagery. It was expected that the PCIwould be associated most strongly with this aspect of imaginalprocessing because fear of failure and worry are central to theexperience of perfectionism (Flett, Hewitt, Blankstein, &Mosher, 1991; Flett, Hewitt, Endler, & Tassone, 1994; Pacht,1984), and we believe that perfectionistic cognitions are experi-enced primarily when an individual is concerned about the dis-crepancy between ideal standards and personal characteristics.

Finally, measures of symptom distress, affect, and life satis-faction were included to further examine the validity of our newmeasure and to test some important hypotheses about the natureof perfectionism cognitions. The inclusion of measures of mood

state and life satisfaction enabled us to test both the affectiveand cognitive components of subjective well-being (see Pavot &Diener, 1993). A tendency to experience frequent perfectionismcognitions should be linked with negative affect, given the linkbetween perfectionism cognitions and depression in Study 1, aswell as the fact that a key component of negative affectivity isthe tendency to ruminate about mistakes (see Watson & Clark,1984). If perfectionism cognitions arise, in part, from an aware-ness about the discrepancy between the actual self and idealstandards of perfection, then there should also be a negativeassociation between frequent perfectionism cognitions and lifesatisfaction. Life satisfaction has been described as a cognitiveprocess that involves the mental comparison of current life qual-ity with personal standards (Diener, 1984; Pavot & Diener,1993). Individuals with frequent perfectionism cognitionsshould have relatively low levels of life satisfaction becauseof the cognitive salience of perfectionistic standards and thenonattainment of these high standards.

Method

Sample and Procedure

Sample I. As noted earlier, Sample 1 was administered the PCIand related measures for the purposes of examining validity and socialdesirability. Forty-four undergraduate students (17 men and 27 women)with a mean age of 21.61 years (SD = 3.32) completed the PCI andvarious measures designed to examine the validity and degree of socialdesirability bias associated with the scale. These participants completedthe PCI, the revised Attitudes Toward Self Scale (Carver, LaVoie,Kuhl, & Ganellen, 1988), Voun T of the Jenkins Activity Survey (Jen-kins, Zyzanski, & Rosenman, 1971), and the impression managementsubscale of the Balanced Inventory of Desirable Responding (Paulhus,1984). These measures are described subsequently.

The revised Attitudes Toward Self Scale (Carver et al., 1988) is a 15-item measure of four dimensions of self-punitive tendencies: high self-standards (e.g., I set higher goals for myself than other people seemto), self-criticism (e.g., I get angry with myself if my efforts don't leadto the results I wanted), overgeneralization (e.g., When even one thinggoes wrong I begin to wonder if I can do well at anything at all), andperseveration on failure experiences (e.g., If I fail at something, I thinkabout that particular failure for a long time afterward). Past resultsindicate that overgeneralization is the component with the strongest cor-relation with dysphoria (see Flett, Hewitt, & Mittelstaedt, 1991).

The Jenkins Activity Survey is a well-known self-report measure ofType A behavior. It was included because of evidence indicating a ten-dency for Type A individuals to endorse perfectionistic standards (Flett,Hewitt, Blankstein, et al., 1994).

The Balanced Inventory of Desirable Responding impression manage-ment scale is a 20-item measure of the tendency to engage intentionallyin the deception of others (Paulhus, 1984). The inventory also includesa 20-item measure of self-deception. In Sample 1, only the impressionmanagement subscale was used. In Sample 3 (described subsequently),both the impression management and self-deception subscales were used.

Sample 2. The second sample was composed of 73 students (23men and 50 women) in introductory psychology at the University ofWinnipeg. As part of a larger prospective study, each participant com-pleted the PCI on two occasions separated by a 3-month interval.

Sample 3. The third sample consisted of 154 students (31 men and123 women) in introductory psychology at the University of Toronto atMississauga. Participants completed the PCI, the Balanced Inventory ofDesirable Responding, the short form of the Imaginal Processes Inven-tory (Singer & Antrobus, 1970), an abbreviated version of the Hopkins


Symptom Checklist (Green, Walkey, McCormick, & Taylor, 1988), thePositive Affect and Negative Affect Schedule (PANAS; Watson, Clark, &Tellegen, 1988), and the Satisfaction With Life Scale (Diener, Emmons,Larsen, & Griffin, 1985). The abbreviated Imaginal Processes Inventoryassesses lack of attentional control or mind wandering, fear of failureimagery, and positive daydreaming. The measure has adequate psycho-metric properties, including factorial validity (see Huba & Tanaka,1983-1984).

The abbreviated Hopkins Symptom Checklist provides a measure ofnegative affectivity (labeled general feelings of distress; e.g., feelingblue), a measure of somatic distress (e.g., pains in the lower part ofyour back), and a measure of performance difficulty (e.g., difficulty inspeaking when you are excited). Green et al. (1988) reported that thethree subscales represent three distinct factors with alpha coefficients(Cronbach, 1951) of .75 or greater in a student sample.

The PANAS is a brief 20-item measure with 10 adjective items thatassess positive affect (e.g., interested and proud) and 10 adjective itemsthat assess negative affect (e.g., ashamed and nervous). Participants inthis study completed the PANAS with respect to the same time frameas the PCI (i.e., the previous week). Internal consistency values of .84or higher have been reported for both affect scales, and PANAS scoreshave been shown to correlate with related measures of distress (seeWatson, Clark, & Tellegen, 1988).

The Satisfaction With Life Scale is a five-item measure of overallsatisfaction with one's life. Item content is directly relevant to concernsabout meeting ideals (e.g., In most ways my life is close to ideal).Extensive information attests to the measure's reliability and validity(see Pavot & Diena; 1993).

Results

Sample 1

Correlations computed between the PCI and the other mea-sures provided evidence of the scale* s validity. PCI scores werecorrelated positively with the Attitudes Toward Self Scale sub-scales measuring high standards, r = .55, p < .001; self-criti-cism, r = .57, p < .001; overgeneralization, r = A3,p < .001;and failure perseveration, r = .56, p < .001. In addition, PCIscores were correlated significantly with total Type A scores,as assessed by the Jenkins scale, r = .33, p < .05. There wasno significant association between the PCI and the BalancedInventory of Desirable Responding impression managementsubscale, r - —.10, ns.

Sample 2: Test-Retest Reliability

The respective means for the PCI were 43.56 (SD = 18.57)at Time 1 and 40.73 (SD = 18.01) at Time 2. The 3-monthtest-retest reliability (r) of the instrument was .67 (p < .01).

Sample 3

In terms of correlations, scores on the PCI were not associatedsignificantly with impression management, r = .13, ns, but therewas a modest negative association between perfectionism cogni-tions and self-deception, r — - .27 , p < .01. These data suggestthat high self-deceivers are less troubled by thoughts involvingthemes of the need for perfectionism.

Correlations with the Imaginal Processes Inventory providedfurther evidence of the validity of the PCI. The PCI was corre-lated positively with the fear of failure subscale, r = .42, p <

.01; the daydreaming subscale, r = .25, p < .01; and the lackof attentional control subscale, r - .22, p < .01.

Finally, the expected correlations were obtained between thePCI and the well-being measures. A greater frequency of perfec-tionism cognitions was associated with Hopkins SymptomChecklist indexes of general distress, r - .54, p < .01; perfor-mance difficulty, r ~ .51, p < .01; and somatic distress, r =.31, p < .01. Perfectionism cognitions were unrelated to positiveaffect, r — -.02, ns, but the frequency of perfectionism cogni-tions was associated with negative affect, r = .51, p < .01, andwith lower life satisfaction, r = - .32, p < .01.

Discussion

The primary goal of Study 2 was to examine the validity ofthe PCI in detail. Substantial evidence of the measure's psycho-metric properties was obtained in this study. The PCI had ade-quate test-retest reliability and was not associated with a mea-sure of impression management. Additional results indicatedthat people with more frequent perfectionism cognitions alsotended to exhibit Type A characteristics and elements of self-punitiveness. Particularly noteworthy was the positive link be-tween frequency of perfectionism cognitions and the indexes ofself-criticism and failure perseveration. The association betweenperfectionism cognitions and self-criticism supports the con-tention that perfectionism cognitions arise, in part, from a defen-sive and self-punitive concern about one's current performanceor personal characteristics being discrepant from perfectionisticstandards. Similarly, the link with failure perseveration suggeststhat perfectionism cognitions are likely to be triggered when afailure is encountered, and the impact of the failure may bemagnified by a tendency for perfectionists to engage in obsessiverumination. The link between perfectionistic thoughts and con-cern about failures was also indicated by the results involvingthe imaginal processes measures. Separate analyses with theImaginal Processes Inventory subscales revealed that high scor-ers on the PCI were especially likely to report images involvingfear of failure. This finding accords with previous evidence ofa pervasive link between perfectionism and fear of failure andfurther supports the contention that the experience of frequentperfectionistic cognitions is a response to concerns about theinability to meet salient standards.

Consistent with expectations, strong positive correlationswere found between the PCI and the Hopkins Symptom Check-list measures of general feelings of distress and performancedifficulty. The correlation between the PCI and general symp-toms of distress replicates the finding in Study 1 in that manyitems on the Hopkins Symptom Checklist general distress mea-sure tap depressive symptoms. The current study extended theearlier findings by showing that there is a more modest butsignificant correlation between the PCI and somatic difficultiesand that there is a strong association between the experience ofperfectionism cognitions and performance difficulties. Exces-sive rumination is a factor that is often associated with perfor-mance impairment and an inability to initiate instrumental activi-ties (see Morrow & Nolen-Hoeksema, 1990). Our data suggestthat perfectionism cognitions may reflect a particular type ofcognitive interference that disrupts performance in varioussettings.


Analyses with the well-being measures revealed an associa-tion between perfectionism cognitions and both the affectiveand cognitive components of psychological well-being. Analyseswith the PANAS measures showed that higher scores on thePCI were linked with current levels of negative affect but notcurrent levels of positive affect. Although perfectionism cogni-tions were not associated with state levels of positive affect,there was a significant negative association between perfec-tionism cognitions and general life satisfaction. Because sometheorists maintain that life satisfaction involves a subjectivecomparison of one's current life situation and personal stan-dards, these data are consistent with our claim that perfectionismcognitions arise, in part, from concerns about not meeting idealstandards of perfection.

Study 3

Taken together, the results of the first two studies indicatethat it is possible to measure the frequency of perfectionismcognitions with an adequate degree of reliability and validity,and it is important to do so because frequent thoughts involvingperfectionism are associated with depressive symptoms and re-lated indexes of distress. Although these data provide insightsinto the cognitive processes involved in perfectionism, someimportant questions about the nature of perfectionistic cogni-tions remain to be addressed. One issue is whether evidence canbe obtained for the existence of perfectionistic cognitions innaturalistic settings. That is, it remains to be established whetherhigh and low PCI scorers actually do differ in the thoughts thatthey spontaneously experience. This issue was tested in Study3 by having participants with varying PCI scores keep a dailyrecord of their naturally occurring dysfunctional thoughts overa 3-day period. These thoughts were then analyzed according tothe number of spontaneous references to perfectionism-relevantcontent.

A related purpose of Study 3 was to examine the extent towhich spontaneous mentions of negative and positive thoughtsabout the self are associated with perfectionism cognitions, asassessed with the PCI and the diary method. Given the linkbetween perfectionism and structured measures of self-criticaltendencies evident in the first two studies, it was expected thatindividuals with more frequent perfectionism cognitions wouldalso report a greater number of negative thoughts aboutthemselves.

Method

Sample

The sample consisted of 56 first-year psychology students (15 menand 41 women) from the University of Winnipeg. Their mean age was25.4) years.

Procedure

Participants were told that there were going to be asked to completea diary for 3 days. First, however; they were asked to complete a batteryof pretest measures, including the PCI.

Next, participants were given a detailed instruction session by PaulL. Hewitt on how to use Beck's diary measure of dysfunctional thoughts(Beck et al., 1979). The daily dysfunctional thoughts records require

individuals to first note any actual event leading to an unpleasant emotionor a stream of thoughts, daydreams, or recollections leading to an un-pleasant emotion. The individual then specifies the emotion or emotionsexperienced and lists the automatic thoughts that preceded the emotion.This instrument has been used successfully in various research projects.For example, it has been used to examine the link between depressionand spontaneous attributions (Riskind, Castellon, & Beck, 1989). Asnoted earlier, in the current study, diary content was analyzed primarilyto estimate the number of thoughts involving perfectionism, as well asthe number of negative thoughts about the self.

Scoring of Dysfunctional Thoughts

The expressed thoughts were scored by two graduate student raterswho were unaware of each individual's level of perfectionism, as as-sessed by the PCI. The raters were instructed to tabulate the number ofthoughts with perfectionism themes. In addition, the raters tabulated thenumber of positive, negative, and neutral thoughts in three content areas(self-referent, other referent, and situation referent) according to aframework used by Fichten (1986). Two of the three content areas (i.e.,self-referent and situation referent) are comparable to categories usedto analyze the thoughts expressed in achievement situations (seeBlankstem, Toner, & Flett, 1989). Content involving perfectionisticthoughts was defined broadly in a manner consistent with die contentof the PCI and with existing trait measures (Frost et al., 1990; Hewitt &Flett, 1991b). A thought was recorded as reflecting perfectionism if itinvolved a direct reference to striving for perfection, a concern withmistakes (e.g., What if I make a mistake?), or the need for organization(see Frost et al., 1990). In light of the interpersonal aspects of perfec-tionism (Frost et al., 1990; Hewitt & Flett, 1991b), thought contentinvolving high expectations from others or a concern about meetingother people's unrealistic expectations was also included. Finally, consis-tent with certain content in the PCI, thoughts were also scored as perfec-tionistic if they reflected the need to improve oneself to maximize poten-tial (i.e., I need to do better) or an awareness of the need to be moreperfect in relative terms involving social comparison (i.e., I have to bethe best). The raters were trained in the scoring of thoughts until a highdegree of agreement between ratings was obtained. As was the case inearlier research (e.g., Blankstein et al., 1989), better than 90% agreementwas obtained between raters.

Results

Overall, it was found that the participants' diaries containedan average of 14.51 thoughts over a 3-day period. The threemost frequent thought categories were negative thoughts aboutthe self (Af = 3.79), negative thoughts about other people (M =2.57), and neutral thoughts about the self (M = 2.27). Thoughtsinvolving perfectionism themes constituted a relatively smallproportion (i.e., 7.1%) of the total number of thoughts (Af =1.04).

The main correlational findings of interest are displayed inTable 3. Correlations were computed between the PCI and the10 thought categories. The resulted showed that the PCI wascorrelated positively with the number of perfectionism thoughtsexpressed, r = .41, p < .01.2 Contrary to expectations, however^

2 It should be noted that this effect was no longer significant (p <.15) when conducting partial correlations taking into account individualdifferences in the total number of thoughts reported. This outcome re-flects the fact that those who reported perfectionistic thoughts had ageneral tendency to report more thoughts in highly relevant categories


Table 3Correlations Among the Perfectionism Cognitions Inventory (PCI) andSpontaneous Thought Categories: Study 3

Measure

1. PCI2. Perfectionistic

thoughts3. Self-positive4. Self-negative5. Self-neutral6. Other positive7. Other negative8. Other neutral9. Situation positive

10. Situation negative11. Situation neutral

1

.41

.12

.11- .13- .12

.12

.06- .13- .07- .19

2

—.12.36.37

- .03.23.02.24

- .03.02

3

—-.01-.08

.27- .01- .14- .10- .16

.09

4

—.20

- .01.12

- .12- .07

.02

.13

5

—.01.10.21.27.30.09

6

—.10.08

- .05,33.60

7

—.42

-.08.08.07

8

—.17.06.10

9

—-.08

.04

10

—.35

11

—

Note. Correlations were based on the responses of 56 participants. Correlations greater than .35 aresignificant at the p < .01 level.

the PCI was not correlated significantly with the number ofthoughts in any other category.

Further analyses focused solely on diary content. It was foundthat the number of spontaneously expressed thoughts reflectingperfectionism was associated with certain other thought catego-ries. Specifically, a greater number of perfectionism thoughtswas associated with the number of negative self-thoughts, r =.36, p < .01, and neutral self-thoughts, r = .37, p < .01. Furtherexamination of the thought categories involving the self showedthat the number of positive thoughts about the self was correlatedwith the number of positive thoughts about other people, r =.27, p < .05. The only other correlations of note were thesignificant positive associations between the number of neutralself-thoughts and positive and negative thoughts about situa-tional factors (see Table 3) .

Discussion

The results of Study 3 provided additional support for thecontention that perfectionism cognitions are an identifiable com-ponent of the perfectionism construct. A diary study was con-ducted to assess the link between the PCI and perfectionismcognitions that were spontaneously produced. It was found thatthe PCI was indeed associated with the number of perfectionisticthoughts contained in the diary. In fact, some participants pro-vided thought content that very closely resembled PCI content(e.g., Why can't I be more perfect? I think about not being ableto play an instrument perfectly). The association between thePCI and diary content constitutes evidence of the predictivevalidity of our new measure.

However, some caveats must be noted. First, it must be ac-knowledged that the total number of perfectionistic thoughtsconstituted only a relatively small percentage of the number ofthoughts reported by participants even though we used a broaddefinition of perfectionistic thoughts. It may be presumed that

(i.e., self-negative thoughts, self-neutral thoughts, negative thoughtsabout others, and positive thoughts about the situation).

a greater number of perfectionistic thoughts would have beenreported if we had placed participants in an experimental situa-tion that involved an explicit focus on achievement and evalua-tive standards, perhaps with the use of priming cues, given thefact that contextual factors influence the accessibility of thoughtcontent (Clark, 1988). This procedure was not adopted becausewe wanted to establish the presence of these thoughts in a natu-ralistic setting.

Second, our attempt to replicate the association between per-fectionism cognitions and negative self-thoughts resulted in anequivocal pattern of findings. In our earlier studies, we found arobust link between perfectionism and automatic negativethoughts (see Study 1), as well as an association between per-fectionism cognitions and measures of self-criticism and persev-erative thought about failure (see Study 2) . In contrast, thecurrent study found that the PCI was not correlated significantlywith the number of negative self-thoughts that were spontane-ously expressed. However, there was a significant positive corre-lation between the number of perfectionistic thoughts that wereproduced spontaneously and the number of negative self-thoughts and neutral self-thoughts about the self.

Why did the measure of spontaneous thoughts involving neg-ative self-thoughts correlate significantly with the number ofperfectionistic thoughts when assessed with the diaries but notwith the number of perfectionistic thoughts, as assessed by thePCI? This difference could be due to substantive reasons; perfec-tionistic thoughts and negative self-thoughts may be linked moreclosely when both perfectionism thoughts and self-thoughts in-volve the same events. Several methodological factors also mayhave contributed to these findings, including the many differ-ences between structured and unstructured methods of assess-ment (Blankstein et al., 1989) and the timing of assessments.Klinger, Barta, and Maxeiner (1980) observed that people prob-ably do not consciously process material and experience discretethoughts reflecting current concerns and goals during most ofthe time that the concern or goal exists; rather, they do so whenthe goal is being abandoned. Although speculative, perhaps theperfectionism cognitions expressed in the diaries relate moreclosely to negative self-evaluative statements because these cog-

PERFECTI0N1STIC THINKING AND DISTRESS 1371

rations were expressed by the subset of perfectionists who werein the process of abandoning their goals.

Clearly, future attempts to illustrate the spontaneous presenceof perfectionistic thoughts, or thoughts related to other personal-ity traits, must be conducted with some recognition of the com-plexity of this issue. When considering the nature of measuressuch as the PCI, it is important to keep in mind the possibilitythat structured measures of automatic thoughts may actuallyreflect the salience of particular thoughts or implicit theoriesabout the thoughts rather than the frequency of thoughts (seeClark, 1988; Glass & Arnkoff, 1982).

These observations notwithstanding, it is important to reiter-ate that the main finding of this study was that perfectionismcognitions could be detected with measures other than the PCI.This study provides empirical confirmation for case descriptions(e.g., Barrow & Moore, 1983; Burns & Beck, 1978) emphasiz-ing the tendency for certain perfectionists to ruminate obses-sively about themes involving the attainment of high standards.

Study 4

The evidence thus far indicates that there are detectable indi-vidual differences in perfectionism cognitions and that, for themost part, these perfectionism cognitions are associated in theexpected manner with measures of automatic thoughts and imag-inal processes. In the final two studies, we examined furtherthe link between perfectionism cognitions and psychologicaldistress. Study 4 addressed the issue of whether the structuredmeasure of perfectionism cognitions could account for uniquevariance in depressive symptomatology after variance associatedwith multidimensional trait measures of perfectionism had beenremoved. This is important to demonstrate because a centralpremise guiding our research is that the tendency to experiencefrequent perfectionism cognitions is an important feature of theperfectionism construct that is not assessed by existing multidi-mensional trait measures. If so, then it should be the case that thePCI accounts for a significant amount of variance in depressivesymptom scores that is not accounted for by trait measures,even though the PCI and trait self-oriented perfectionism sharea focus on high standards for the self.

Research with trait perfectionism measures has found anequivocal association between self-oriented perfectionism anddepression. Whereas some studies have found a positive associa-tion, other studies have found a negative or negligible associa-tion between self-oriented perfectionism and depression (e.g.,Flett, Hewitt, Blankstein, & Mosher, 1991; Frost, Heimberg,Holt, Mattia, & Neubauer, 1993; Hewitt & Flett, 1991a, 1993a;Westra & Kuiper, 1996). Some authors have gone so far asto suggest that self-oriented perfectionism is not necessarilyassociated with poor adjustment and that it may be adaptiveunder certain circumstances (see Frost et al., 1993; Slaney,Ashby, & Trippi, 1995). When should self-oriented perfec-tionism be associated with greater susceptibility to depression?It has been suggested previously that depression is elevatedamong those self-oriented perfectionists who have experiencedrecent life stressors, especially in the achievement domain (Flett,Hewitt, Blankstein, & Mosher, 1995; Hewitt & Flett, 1993a;Hewitt et al., 1996). A related possibility is that depressivesymptoms are evident among the subset of self-oriented perfec-

tionists who experience troubling cognitions about the need toattain perfection. This would be in keeping with self-regulationand self-discrepancy models of depression (e.g., Bandura, 1986;Higgins, Bond, Klein, & Strauman, 1986) as well as descriptiveaccounts (Burns & Beck, 1978) maintaining that depressivereactions are likely to ensue when a person with high standardsbecomes cognitively aware of the fact that he or she is notmeeting these unrealistic standards.

In the current study, three separate samples of students com-pleted the PCI, along with the Multidimensional PerfectionismScale (MPS; Hewitt & Flett, 1991b) and a measure of depressivesymptoms. The measure of depressive symptomatology was var-ied among the samples to examine the link between perfec-tionism cognitions and levels of depressive symptoms with alter-native measures of psychological distress. In addition, the thirdsample of students completed both the Hewitt and Flett versionof the MPS and the Frost et al. (1990) version.

A distinguishing feature of Study 4 is that it also included apsychiatric sample (Sample 4, described subsequently). Al-though it has been suggested that the study of personality anddepression symptoms is valid and relevant in students (seeVredenburg, Flett, & Krames, 1993), it is nevertheless importantto examine the clinical relevance of the PCI in psychiatric pa-tients. A psychiatric sample was used to examine the psychomet-ric properties of the PCI when administered to individuals withmore severe levels of psychopathology. The data from theseindividuals were also used to test the hypothesis that frequentperfectionistic thinking would be associated with increased de-pressive symptomatology among psychiatric patients. In addi-tion, we investigated whether the frequency of perfectionisticthinking accounts for variance in depression scores that is notaccounted for by existing perfectionism measures. The partici-pants in the clinical sample also completed the PCI, a measureof depressive symptomatology, and both multidimensional per-fectionism inventories (Frost et al., 1990; Hewitt & Flett,1991b).

Method

Sample 1

Sample 1 consisted of 311 students (80 men and 231 women) inintroductory psychology at the University of Toronto at Mississauga.They had a mean age of 20.6 years. All participants completed the PCI,the MPS, and the Inventory to Diagnose Depression (Zimmerman &Coryell, 1987, 1988) as part of a mass assessment. The Inventory toDiagnose Depression is a self-report measure that is regarded as asignificant improvement over other measures because it includes all ofthe symptoms of major depressive disorder contained in the Diagnosticand Statistical Manual of Mental Disorders (revised third edition;DSM~1II-R; American Psychiatric Association, 1987).

The MPS is a 45-item measure designed to measure self-orientedperfectionism (e.g., One of my goals is to be perfect in everything 1 do) ,other-oriented perfectionism (e.g., If I ask someone to do something, Iexpect it to be done flawlessly), and socially prescribed perfectionism(e.g., The people around me expect me to succeed at everything I do).

Sample 2

Sample 2 consisted of 144 students (67 men and 77 women) inintroductory psychology at the University of Toronto at Mississauga.


Participants completed the PCI^ the MPS, and the BDI, as describedearlier. The mean age of the sample was 20.63 years.

Sample 3

This sample was composed of 117 students (47 men and 70 women)in introductory psychology at the University of Toronto at Mississauga.They had a mean age of 21.3 years. Participants completed the samemeasures completed by the participants in Sample 2. They also com-pleted another trait perfectionism measure (described later).

Sample 4

Sample 4 participants were 62 psychiatric patients (36 men and 26women) from Brockville Psychiatric Hospital. The sample was com-posed of 27 inpatients and 35 outpatients with a mean age of 38.1 yearsand an average of 11.50 years of education. A heterogeneous clinicalsample was chosen to ensure adequate variation in PCI scores; use ofhomogeneous clinical samples to assess personality constructs may resultin biased statistics (see Kline, 1987). The most frequent primary diagno-ses made by each patient's psychiatrist, according to DSM-IH-R criteria(American Psychiatric Association, 1987), were depression (34%) andschizophrenia (21 % ) . An additional 11 % of the patients had a personal-ity disorder. The remaining patients had a widespread variety of diagno-ses including adjustment reaction, alcohol-drug dependence, and anxi-ety disorder, although patients with organiciry were not included. Thereliability of diagnoses was not confirmed by a second clinician becausethis study focused on the responses of the sample as a whole rather thanon a comparison of patients with particular diagnoses.

A smaller sample of 28 psychiatric patients completed the PCI twicewith a 3-month interval to allow assessment of the stability of the PCIin a clinical sample. Participants were recruited randomly and paid $10for taking part. Participants were excluded from the study if they hadless than an eighdi-grade education or a current psychosis.


The participants in the third and fourth samples completed die PCI,the BDI, the MPS, and the Frost Multidimensional Perfectionism Scale(FMPS; Frost et al., 1990) in a random order. The FMPS is a 35-itemmeasure of six perfectionism components: concern over mistakes (e.g.,I should be upset if I make a mistake), personal standards (e.g., I haveextremely high goals), parental expectations (e.g., My parents set veryhigh standards for me), parental criticism (e.g., My parents never triedto understand my mistakes), doubts about actions (e.g., I usually havedoubts about the simple everyday things I do), and organization (e.g.,I am a neat person).

Results

Sample 1

The alpha coefficient of the PCI in this sample (n = 311)was estimated at .94, with a mean interitem correlation of .40.There was no significant gender difference in mean scores. Therespective means for men and women were 42.78 (SD = 20.60)and 45.22 (SD = 19.64).

Correlational analyses confirmed that the PCI was correlatedsignificantly with self-oriented perfectionism, r = .66,/? < .01;other-oriented perfectionism, r = .26, p < .01; and sociallyprescribed perfectionism, r = .35, p < .01. In terms of correla-tions with the Inventory to Diagnose Depression, significantpositive correlations were obtained between depression symp-

toms and all of the perfectionism measures. Scores on the Inven-tory to Diagnose Depression were correlated with socially pre-scribed perfectionism, r = .36, p < .01; perfectionism cogni-tions, r = .35, p < .01; self-oriented perfectionism, r = .18, p< .01; and other-oriented perfectionism, r — .13, p < .05.

A hierarchical regression analysis was conducted to establishwhether the PCI could account for unique variance in depressionsymptom scores after variance due to trait levels of perfec-tionism had been removed. A predictor block consisting of thethree MPS dimensions accounted for 13% of the variance inInventory to Diagnose Depression scores, F = 15.51,/? < .001.The only significant individual predictor within the block wassocially prescribed perfectionism, F = 34.02, p < .001, 0 =.36. Subsequent entry of the PCI showed that it accounted for6% of the remaining variance in Inventory to Diagnose Depres-sion scores, F change = 23.49, p < .001, f3 = .34, with a higherfrequency of perfectionistic cognitions being associated withgreater depression.

Sample 2

Correlations computed with the data obtained from Sample2 (« = 144) provided more evidence of the PCI's validity. ThePCI was associated significantly with self-oriented perfec-tionism, r = .69, p < .001; other-oriented perfectionism, r -.45, p < .001; and socially prescribed perfectionism, r = .46,p < .001. The pattern of correlations was similar for men andwomen.

As expected, a significant correlation was obtained betweenthe PCI and the BDI, r = .28, p < .01. Separate examinationshowed that the correlation was significant for men, r = .42, p< .01, but not for women, r = .20, ns.

The hierarchical regression with BDI score as the outcomevariable indicated that the MPS predictor block was significant,accounting for 11.6% of the variance, F = 6.10, p < .01. Theonly significant predictor within the block was socially pre-scribed perfectionism, F = 15.93, p < .01, /3 = .36. Subsequententry of the PCI showed that it accounted for an additional 4.3%of the variance, F change = 7.13, p < .01, (3 = .32.

Sample 3

The correlations between the PCI and the separate dimensionsthat make up the MPS and FMPS are shown in Table 4 forSample 3 (n = 117). There were few gender differences, soonly the results for the total samples are reported. It can be seenin Table 4 that the PCI was correlated significantly with self-oriented, other-oriented, and socially prescribed perfectionism.As for the FMPS dimensions, the PCI was also correlated sig-nificantly with all of the measures from this inventory. Thestrongest associations were between the PCI and concerns aboutmistakes and between the PCI and personal standards.

The PCI and BDI were correlated significantly among univer-sity students in Sample 3. Separate correlations also examinedthe association between depression and the subscales of the MPSand the FMPS. Analyses revealed that the BDI was correlatedsignificantly with socially prescribed perfectionism, r = .33, p< .001; doubts about actions, r = .43, p < .001; parental


Table 4Correlations Between the Perfectionism Cognitions Inventoryand Other Measures of Perfectionism and DepressionSymptoms in Patients and Students: Study 4

Measure

MPS dimensionsSelf-orientedOther orientedSocially prescribed

FMPS dimensionsConcern over mistakesHigh standardsDoubts about actionsOrganizationParental expectationsParental criticism

Distress scaleDepression symptoms

Patients

.52***

.24

.65***

.57***41***.38**.10.38**.32**

.39**

Students

.69**

.41***^7***

_g4***49**.42***25**.28**.26**

.30**

Note. Correlations were analyzed using two-tailed tests and were basedon the responses of 62 patients (36 men and 26 women) and 117 students(47 men and 70 women). MPS = Hewitt and Flett MultidimensionalPerfectionism Scale; FMPS = Frost Multidimensional PerfectionismScale.**p < .01. ***p < .001.

criticism, r = .27, p < .01; and concern over mistakes, r = .23,p < .05.

Hierarchical Regression Analyses

Additional analyses with the data from Sample 3 assessedwhether the PCI accounted for unique variance in depressionsymptom scores after removal of variance due to the MPS andthe FMPS. The results are shown in Table 5. In the first analysis,the MPS was entered as the first predictor block. The MPS blockaccounted for 18.1% of the variance in BDI scores, F = 8.34,p < .001 (see Table 5) . Examination of the individual predictorswithin the block showed that greater depressive symptomatologywas associated with higher levels of socially prescribed perfec-tionism, F — 2}.25,p < .001, and lower levels of other-orientedperfectionism, F = 6.26, p < .05. Subsequent entry of the PCIshowed that it accounted for 5.3% of the remaining variance,F change = 7.78,/? < .01.

The analysis was repeated by substituting the FMPS measuresfor the MPS. It can be seen in Table 5 that the FMPS blockaccounted for 22.2% of the variance in BDI scores, F — 5.23,p < .001, with doubts about actions being the individual pre-dictor that was significant within the block, F = 12.78, p <.001. Subsequent entry indicated that the PCI was a marginallysignificant predictor, accounting for 2.2% of the remaining vari-ance, F change = 3.17, p < .08.

Sample 4

The final analyses in this study focused on perfectionismcognitions in psychiatric patients (n = 62). Psychometric analy-ses indicated that the PCI had a high level of internal consistencyin this clinical sample. The Cronbach alpha coefficient was .95,and the mean interitem correlation was .42. The overall scale

mean was 44.68 (SD = 23.40). Consistent with the collegestudent data, there was no mean gender difference in PCI scores.The mean for men was 43.51 (SD = 22.42), and the mean forwomen was 46.35 {SD = 25.11).

As for the stability of perfectjonistic cognitions in psychiatricpatients, analyses of the data from the subsample (n - 28)showed that PCI scores at Time 1 were correlated significantlywith PCI scores at Time 2, r = .85, p < .001. This high levelof temporal stability is not uncommon for measures of automaticthoughts (see Ingram, Kendall, Siegle, Guarino, & McLaughlin,1995).

The concurrent validity of our new scale among clinical pa-tients was examined by correlating PCI scores with the scoresobtained with the two multidimensional perfectionism measures.The correlational findings are displayed in Table 4. Investigationof the correlations with the MPS dimensions in the total showedthat the PCI was correlated with both self-oriented and sociallyprescribed perfectionism. Other-oriented perfectionism yieldedthe only gender difference; PCI scores were correlated withother-oriented perfectionism for men, r — .40, p < .05, but notfor women, r = .04, ns.

Correlations were also computed between the PCI and dimen-sions represented on the Frost perfectionism measure. As canbe seen in Table 4, the PCI was correlated significantly with allof the measures derived from the FMPS, with the exception ofthe organization subscale.

Associations With Depressive Symptomatology

Examination of the MPS dimensions in the clinical sampleshowed that depression was correlated significantly with so-cially prescribed perfectionism, r = .37, p < .01. Examination

Table 5Hierarchical Regression Analyses With the MultidimensionalPerfectionism Scales and the Perfectionism CognitionsInventory Predicting Depression Symptomsin Students: Study 4

Variable

Trait blockSelf-orientedOther orientedSocially prescribed

Perfectionism cognitions

Trait blockHigh standardsConcern about mistakesDoubts about actionsOrganization

R2 .R2

change

Analysis 1

.18

.23

.18

.05

Analysis 2

.22

High parental expectationsParental criticism

Perfectionism cognitions .24

t p < .08 (marginally significant)..001.

.22

.02

*P<

Fchange

8.34***

7.78**

5.23***

3.17f

.05. **p

0

-.08-.26

.47

.40

-.11.13.35.02

-.16.23.21

< .01.

F

0.446.26*

21.25***7.78*

1.051.21

12.78***0.031.652.993.17f

***p <


of the FMPS dimensions revealed that depression symptomswere associated with concern over mistakes, r = .36, p < .01,and doubts about actions, r = .36,/? < .01. No other correlationswere significant.

The results of the hierarchical regression analyses for theclinical sample are shown in Table 6. In the first analysis, theMPS was entered as the first predictor block. The MPS blockaccounted for 29.0% of the variance in BDI scores, F = 7.92,p < .01. Once again, examination of the individual predictorswithin the block showed that greater depressive symptomatologywas associated with higher socially prescribed perfectionism, F= 20.80, p < .01, and lower other-oriented perfectionism, r =8.04, p < .01. Subsequent entry of the PCI showed that itaccounted for a marginally significant 4% of the remaining vari-ance, F change = 3.43, p < .08.

The analysis was repeated by substituting the FMPS measuresfor the MPS. It can be seen in Table 6 that the FMPS blockaccounted for 27.7% of the variance in BDI scores, F = 3.51,p < .05. Examination within the predictor block showed thatconcern about mistakes was associated with higher levels ofdepressive symptoms, F = 7.29, p < .01, but the high personalstandards component was associated with lower levels of de-pressive symptoms, F = 5.83, p < .05. Subsequent entry indi-cated that the PCI accounted for 8.0% of the remaining variance,F change = 7.28, p < .01, with greater perfectionism cognitionsassociated with greater symptoms of depression.

Discussion

The results of Study 4 provided insight into several interre-lated issues. First, our results confirmed that the PCI is associ-ated with trait indexes of perfectionism. Analyses with the traitmeasures indicated that self-oriented perfectionism was the

Table 6Hierarchical Regression Analyses With the MultidimensionalPerfectionism Scales and the Perfectionism CognitionsInventory Predicting Depression Symptomsin Patients: Study 4

Variable

Trait blockSelf-orientedOther orientedSocially prescribed


Trait blockHigh standardsConcern about mistakesDoubts about actionsOrganizationHigh parental expectationsParental criticism


Rz

R2 change

Analysis

.29 .

.33 .

Analysis

.28 .

.36 .

1

.29

,04

2

.28

.08

Fchange

7.92**

3.43t

3.51*

7.28**

0

- .09- .38

.60

.28

- .39.51.07

- .03- .02-.01

.37

F

0.488.04**

20.80**3.43f

5.83*7.29**0.210.030.020.017.28**

t p < .08 (marginally significant). * p < .05. ** p < .01.

MPS trait dimension associated most closely with reports offrequent perfectionism cognitions, whereas concern over mis-takes was the FMPS dimension associated most strongly withthe experience of perfectionism cognitions. The link with self-oriented perfectionism reflects the content of the PCI, whichfocuses primarily on self-relevant cognitions involving the needto be perfect. The link with a concern over mistakes is in keepingwith past results suggesting that concern about mistakes reflectsa tendency to ruminate and have a difficult time forgetting errors(e.g., Frost & Henderson, 1991).

Given that there is a substantial association between automaticperfectionism cognitions and trait perfectionism, a key goal ofthis study was to determine, in both subclinical and clinicalsamples, whether levels of perfectionistic thought frequencycould account for unique variance in depression symptom scoresover and above the variance accounted for by existing traitmeasures of perfectionism. The overall pattern of results indi-cated that level of automatic perfectionism thoughts is a uniquepredictor of depressive symptoms, both in students and in pa-tients. The findings with students indicated that the PCI ac-counted for significant variance in depression symptom scoresover and above the variance predicted by Hewitt and Flett's(1991b) MPS dimensions.

Regression analyses involving the FMPS trait dimensionsshowed that the PCI accounted for a marginally significantamount of unique variance in BDI scores in students (see Sam-ple 3). However, the subsequent analysis of data from psychiat-ric patients showed that the PCI did account for significantvariance in BDI scores among more severely distressed individu-als, over and above the variance predicted by the FMPS. Thus,the bulk of the available evidence indicates that the PCI is relatedto trait perfectionism dimensions, but the PCI does not overlapto the extent that it does not predict unique variance in self-reported levels of depression symptoms.

Study 5

Our final study addressed three goals. First, in addition todepressive symptoms, we also examined the link between thePCI and symptoms of anxiety. Extensive evidence indicates thatanxious individuals are especially likely to experience intrusivethoughts that interfere with their cognitive processes and goal-directed performances (see Blankstein et al., 1989; Sarason,Pierce, & Sarason, 1996). Perfectionism is probably involvedgiven the association (described earlier) between perfectionismand fear of failure imagery (see Study 2) . At the theoreticallevel, a link between perfectionism cognitions and measures ofanxiety and worry would be expected according to a model putforth by Borkovec, Metzger, and Pruzinsky (1986). Accordingto Borkovec et al. (1986), worry is experienced when there isa discrepancy between desired goals or future plans and thecurrent self or the current state of affairs. Negative emotion islikely to be experienced when there is a large discrepancy be-tween goals and the current situation, and these negative emo-tions prompt negative images of future possibilities. Perfec-tionism is highly relevant to this model because perfectionistsshould be especially likely to experience large discrepanciesbetween their goals and the current situation as a result of theirunrealistic standards.


Our second goal was to examine the link between perfec-tionism cognitions and trait neuroticism. A link between thePCI and neuroticism would be expected for several reasons, inaddition to our notion that automatic thoughts involving perfec-tionism are due, in part, to the recognition that one is not at-taining perfection. Perfectionism is associated with trait neuroti-cism (Hill, Mclntire, & Bacharach, 1997), and a focus on theinadequacies of the self is regarded as a central element ofneurotic forms of perfectionism (see Hamachek, 1978; Terry-Short, Owens, Slade, & Dewey, 1995). Thus, in the presentstudy, we investigated the link between the PCI and neuroticism,along with the related issue of whether the PCI could accountfor unique variance in anxiety and depression over and abovetrait neuroticism.

Our final goal was to examine how perfectionistic thoughtfrequency related to other cognitive response dimensions, inaddition to thought frequency. Parkinson and Rachman (1981)established that intrusive thoughts vary not only in their per-ceived frequency but also on other identifiable dimensions. Addi-tional response dimensions include the degree of sadness associ-ated with a distressing thought, the levels of worry and guiltassociated with the thought, the perceived difficulty associatedwith removing the thought, and beliefs about the self-relevanceof the thought. The dimensional measures of perceived sadness,worry, and guilt reflect the fact that cognition, affect, and moti-vation are interrelated and emotions influence or contribute tocognitive products (see Klinger, 1989).

Clark, Feldman, and Channon (1989) have illustrated thepotential usefulness of including other cognitive response di-mensions. For instance, Clark et al. (1989) reported that theresponse dimensions just outlined can be rated meaningfullywith respect to both depressive statements (e.g., thoughts orimages of being worthless) and anxious statements (e.g.,thoughts or images of a personally embarrassing, humiliating,or painful experience).

In Study 5, two samples of students were assessed in termsof levels of perfectionism cognitions and symptoms of depres-sion and anxiety. In addition, students in the first sample com-pleted a measure of trait neuroticism, whereas students in thesecond sample also responded to the depressive and anxiousstatements making up the Clark et al. (1989) measure. That is,participants rated the degree of frequency, sadness, guilt, andworry associated with these thoughts, as well as the difficultyassociated with removing the thoughts and the degree of self-relevance. It was expected that the PCI would be associatedwith depression symptoms and depressive cognitions, as wellas with anxiety and anxiety cognitions. Because depression in-volves themes of dejection about the past or one's current situa-tion, whereas anxiety involves the threat of aversive events inthe future (Beck et al., 1979; Dombeck, Siegle, & Ingram,1996), the expected pattern of results would indicate not onlythat ruminative thoughts about perfectionism involve a concernabout not meeting high standards in the past and in the currentsituation but that such thoughts are also linked with worry aboutthe inability to achieve these standards or to avoid mistakes inthe future.

Finally, consistent with our general approach throughout thisresearch, we used hierarchical regression analyses to addressthree relevant questions: (a) Do perfectionism cognitions ac-

count for significant variance in anxiety and depression overand above trait neuroticism? (b) Do perfectionism cognitionsaccount for unique variance in depressive symptomatology afterremoving variance due to depressive cognition response do-mains, as assessed by the Clark et al. (1989) measure? and(c) Do perfectionism cognitions account for unique variance insymptoms of anxiety after removing variance due to anxietycognition response domains? If so, this would further attest tothe variable's predictive usefulness.

Method

Participants

Sample 1. The participants in this sample were 69 students (54women, 14 men, and 1 person who did not indicate gender) from anintroductory psychology class at \brk University. Their mean age was20.8 years.

Sample 2. The participants were 140 first-year university students(62 men and 78 women) from an introductory psychology class at theUniversity of Toronto at Mississauga. Their mean age was 20.7 years.


Sample 1. Participants completed the PCI, the neuroticism subscaleof the Eysenck Personality Questionnaire-Revised (Eysenck &Eysenck, 1991), and the short version of the Mood and Anxiety Symp-tom Questionnaire (MASQ; Watson & Clark, 1991). The Eysenck neu-roticism scale is a trait measure of a stable, higher order factor reflectingemotional instability (Eysenck & Eysenck, 1991). The abbreviatedMASQ is a 62-item measure stemming from the tripartite model (seeWatson, Clark, & Carey, 1988). The MASQ provides two general dis-tress factors that are common to both anxiety and depression (referredto as general distress: anxiety and general distress: depression). It alsomeasures an anxious arousal factor with symptoms of somatic tensionand hyperarousal that are specific to anxiety and an anhedonic depressionfactor with items involving loss of interest and an absence of positiveaffect that are specific to depression. Initial research indicates that thevarious subscales have adequate psychometric properties (Watson et al.,1995).

Sample 2. In total, participants completed four measures in a randomorder. In addition to the PCI and the BDI, participants completed theDistressing Thoughts Questionnaire (DTQ; Clark & Hemsley, 1985) andthe Beck Anxiety Inventory (Beck, Epstein, Brown, & Steer, 1988).

The DTQ provides separate measures of distressing thoughts involv-ing themes of depression and anxiety. Individuals are provided with sixdepressive items (e.g., thoughts or images that I am a failure and thoughtsor images of why my life is not going the way I want it to) and sixanxious items (e.g., thoughts or images of a personally embarrassing,humiliating or painful experience and thoughts or images that somethingis, or may in the future, be wrong with my health). Each item is ratedon six response dimensions: frequency (i.e., How often does this thoughtor image enter your mind?), sadness (i.e., How sad or unhappy doesthis thought or image make you feel?), worry (i.e., How worried doesthis thought or image make you feel?), removal (i.e., How difficult isit for you to remove this thought or image from your mind?), guilt (i.e.,How guilty does this thought or image make you feel when it entersyour mind?), and belief (i.e., How much do you believe that this thoughtor image is actually true of you?). In total, individuals make 72 nine-point ratings of these items. The Beck Anxiety Inventory is a 21-itemmeasure of the cognitive and physiological symptoms of anxiety (Becketal., 1988).

1376 FLETX HEWITT, BLANKSTEIN, AND GRAY

Results

Sample 1

Correlational analyses confirmed that there was a substantialassociation between the PCI and trait neuroticism, r — .48, p< .001. The two measures had similar links with the MASQmeasures. The PCI was associated with general distress: depres-sion (r = .53, p < .001), genera] distress: anxiety (r = .59, p< .001), and anxious arousal (r = .56, p < .001), but the linkwith anhedonic depression was marginally significant (r — .22,p < .08). Trait neuroticism was also associated with anhedonicdepression (r — .49, p < .001), general distress: depression (r= .50,p < .001), general distress: anxiety (r = .54, p < .001),and anxious arousal {r = .40, p < .01).

Sample 2

Correlational analyses examined the association between thePCI and the measures of anxiety, depression, anxious thinking,and depressive thinking. Analyses of the data from the total sampleshowed that higher PCI scores were correlated significantly withsymptoms of depression, r ~ .39, p < .01, and anxiety, r = .50,p < .01. In addition, as illustrated in Table 7, scores on the PCIwere correlated with all of the depressive thoughts measures, withcorrelations ranging from .30 to .47. Similarly, PCI scores werecorrelated significantly with all of the anxious thoughts measures,with correlations ranging from .35 to .44. It is not altogethersurprising that the PCI was correlated with all of the DTQ de-pressive thoughts and anxious thoughts measures because the re-sults also showed that the respective subscales of the DTQ werehighly intercorrelated, with most intercorrelations greater Ihan .70.Earlier work has shown that the DTQ subscales are highly inter-

Table 7Correlations Between the Perfectionism Cognitions Inventoryand Cognitive Response Dimensions, Depression Symptoms,and Anxiety Symptoms: Study 5

Measure

Depression scalesFrequencySadnessWorryRemovalGuiltBelief

Anxiety scalesFrequencySadnessWorryRemovalGuiltBelief

Adjustment scalesDepression symptomsAnxiety symptoms

Total

.43

.47

.47

.46

.41

.30

.36

.42

.44

.41

.40

.35

.39

.50

Men

.53

.59

.59

.60

.44

.35

.44

.51

.52

.53

.47

.39

.47

.52

Women

.34

.35

.37

.35

.38

.25*

.28a

.32

.33

.31

.34

.29a

.30

.45

Note. Correlations were analyzed using two-tailed tests and were basedon the responses of 62 men and 78 women. Except as noted, correlationswere significant at/? < .01.* Significant at p < .05.

correlated (Clark, 1986). The key analyses were hierarchical re-gressions that tested whether the PCI could account for uniquevariance in adjustment accounts when entered after neuroticism(Sample 1) and DTQ measures (Sample 2).

Hierarchical Regressions: Sample 1

The MASQ general distress: depression measure was the out-come measure in the first analysis. Trait neuroticism accountedfor 24.5% of the variance, F = 21.80, p < .001. Subsequententry of the PCI indicated that it accounted for an additional10.8% of the variance, Fchange = 11.10, p < .01.

The same analysis was repeated with the MASQ anhedonicdepression measure as the outcome. Neuroticism was highlysignificant and accounted for 24.3% of the variance, F = 21.52,p < .001, but the PCI did not predict significant unique variance.

Regarding the MASQ general distress: anxiety measure, traitneuroticism accounted for 28.9% of the variance, F = 27.25, p< .001. Subsequent entry of the PCI revealed that it accountedfor an additional 14.7% of the variance, F change = 17.21, p< .001. Finally, trait neuroticism accounted for 16.0% of thevariance in anxious arousal scores, F = 12.71, p < .001, andthe PCI accounted for an additional 17.6% of the variance, Fchange = 17.52, p < .001.

Hierarchical Regressions: Sample 2

In the first analysis, the outcome measure was BDI score.The depressive thinking measures from the DTQ were enteredas a first predictor block, and they accounted for 37.3% of thevariance in BDI scores, F = 13.16, p < .01. Although the PCIwas correlated significantly with all DTQ depressive thoughtssubscales, it was still found that the PCI accounted for a signifi-cant 2% of unique variance, F change = 4.44, p < .05.

In the second analysis, Beck Anxiety Inventory scores werethe outcome measure. The anxious thinking measures from theDTQ were entered as a first predictor block, and they accountedfor 20.4% of the variance in Beck Anxiety Inventory scores, F= 5.68,p < .01. Once again, even though the PCI was correlatedsignificantly with all DTQ anxious thinking subscales and theDTQ measures were already in the equation, it was still foundthat the PCI accounted for a significant 10.8% of unique vari-ance, F change = 20.78, p < .01.

Discussion

The results of Study 5 confirmed that individuals who experi-ence frequent perfectionism cognitions also report greater levelsof both anxiety and depression symptomatology and a greaterfrequency of thoughts or images involving content related tothemes of anxiety and depression. Whereas past research withtrait measures (i.e., the FMPS) has shown that there is a strongerassociation between perfectionism and depressive symptomsthan between perfectionism and anxiety symptoms in students(Minarik & Ahrens, 1996), our data indicated that automaticthoughts involving perfectionism may have a stronger associa-tion with symptoms of anxiety than with symptoms of depres-sion, at least in student samples. For instance, the analyses in-volving the MASQ revealed that the PCI was correlated signifi-


cantly with both measures of anxiety and the general measureof depression that overlaps with anxiety. However, the PCI wasnot correlated significantly with the measure of anhedonic de-pression. Anhedonic depression involves a lack of positive af-fect, so the data from this study and from Study 2 suggest thatthe PCI is associated with the presence of general distress butnot a lack of positive affect.

As expected, data analyses confirmed that perfectionism cog-nitions are associated highly with trait neuroticism. However, theresults of the regression analyses established that perfectionismcognitions account for a substantial degree of unique variancein three of the four MASQ measures, even after taking intoaccount the overlap between the PCI and trait neuroticism.

Results involving other cognitive appraisal dimensionsshowed that perfectionism cognitions are related to various di-mensions of anxious and depressive thoughts, in addition to thefrequency dimension. Specifically, high PCI scorers also re-ported that their anxious and depressive thoughts invokedgreater levels of sadness, worry, and guilt. Moreover, thesethoughts were seen as highly self-relevant and especially diffi-cult to remove. The finding that frequent perfectionism cogni-tions have a widespread emotional impact and seem quite persis-tent to the individual experiencing such thoughts is in keepingwith recent reports of the persistence of perfectionism and thelong-lasting difficulties associated with the treatment of perfec-tionism (see Blatt, Quinlan, Pilkonis, & Shea, 1995). In lightof these difficulties, Blatt et al. (1995) have suggested that theadjustment problems of perfectionists may require long-termintervention.

Overall, the findings from Study 5 established that the per-ceived experience of frequent perfectionism cognitions is notspecific to the experience of depressive symptoms because itgeneralizes to the experience of anxious symptoms as well. Thisis consistent with the link in Study 2 between perfectionism andimages of fear of failure, as well as the link between perfec-tionism and negative affect. As noted earlier, in terms of thoughtcontent, depression is believed to involve declarative statementsabout the past or current situation (e.g., I am a failure), whereasanxiety is believed to involve a sense of dread and fearful ques-tions about the future (e.g., I wonder if I will fail?). Given thelink with symptoms of both depression and anxiety in the pres-ent study, it appears reasonable to conclude that high PCI scorersare concerned jointly with a previous or current inability toattain perfectionistic standards, as well as anticipatory fears andworries that such standards will not be met in the future.

General Discussion

The present research demonstrated that individual differencesin the perceived frequency of perfectionistic thoughts can beassessed with an adequate degree of reliability and validity inboth subclinical and clinical samples. More important, it wasestablished that reports of frequent perfectionistic thinking areassociated with higher levels of depressive symptomatology andanxiety, as well as more general forms of negative affectivityand life dissatisfaction. The maladaptive nature of perfectionisticthoughts is highlighted by the fact that perfectionism thoughtfrequency predicts significant variance in adjustment scores,even after variance due to other measures known to be associated

with increased levels of depressive and anxious symptomatologyhas been removed.

The current findings have many practical and theoretical im-plications. At the practical level, they suggest that an importanttreatment focus for certain individuals is the amelioration ofpersistent thoughts involving perfectionism. That is, treatmentprograms should not only strive to reduce overall levels of per-fectionism, but they should also focus directly on the perfection-ist's tendency to engage in excessive cognitive rumination aboutthe need to attain perfection. Cognitive-behavioral interven-tions may be particularly useful in this regard (see Burns &Beck, 1978), although, as indicated earlier, recent research byBlatt et al. (1995) attests to the possible difficulty associatedwith removing persistent perfectionistic thoughts.

The theoretical implications of the current findings are equallyimportant. As noted earlier, some initial data indicate that theideal self functions as a schema that is involved in the cognitiveprocessing of available information (Hewitt & Genest, 1990).This finding raises the interesting possibility that a key individ-ual-differences variable is the extent to which people engageactively in rumination about the attainment or failure to attainhigh standards. One interpretation of the current findings is thatcertain perfectionistic individuals are preoccupied cognitivelywith personal and social standards. These individuals may bemore likely to interpret ambiguous feedback as reflecting a clearsuccess or failure, and they may encode event outcomes at adeeper level of processing than do individuals who are lesspreoccupied with perfectionistic thinking. Although these obser-vations are admittedly speculative, one reasonable interpretationof the current findings is that perfectionists differ from nonper-fectionists in terms of degree of perfectionism as well as thecognitive salience of perfectionism. Consequently, it can be ar-gued that subsequent models of the association between perfec-tionism and emotional well-being should include both types ofvariables.

More generally, the current findings are significant becausethey highlight the importance of a broader approach to the studyof personality and emotional well-being. Current research onpersonality and depression has focused exclusively on the abso-lute level of traits such as autonomy and sociotropy (see Flett,Hewitt, Endler, & Bagby, 1995). There have been no empiricalattempts to address the possibility that these various traits alsoinclude a cognitive element that involves the frequency of cogni-tive processing or rumination, even though theoretical accounts(e.g., Klinger, 1977) suggest that personality variables have aperceivable cognitive component. That is, the autonomous per-son may not only desire autonomy and achievement; he or shemay also ruminate about personal capabilities and circum-stances or events that interfere with autonomy and achievementgoals. Also, the sociotropic, dependent person may be character-ized both by strong affiliative needs and, episodically, bythoughts about interpersonal relationships that stem perhapsfrom relational schemas (see Baldwin, 1992). Recent empiricaldevelopments are consistent with this observation. For instance,Langston and Sykes (1997) used various procedures to docu-ment the cognitive correlates of the traits in the five-factormodel. Similarly, the results of our research suggest that theperceived frequency of cognitive rumination is a central featureof certain personality traits and that there is a need to measure


individual differences in the reported frequency of specific cog-nitions. Future attempts to study depression and personalitytraits such as autonomy and sociotropy should include variablesthat measure individual differences in the degree of ruminativethought stemming from these traits.

Directions for Future Research

The present research has addressed several issues pertainingto individual differences in the frequency of perfectionisticthinking, but many important research issues remain to be inves-tigated. One line of research should further investigate the no-mological network to clarify the underlying nature of perfection-istic thinking. Research is needed to elucidate the origins andcharacteristics of perfectionistic thinking and the factors thatdistinguish those perfectionists who engage in extensive rumina-tion and those who report relatively low rumination. One possi-bility that could be tested in future research is that perhaps thereported presence of perfectionism cognitions, relative to gen-eral dispositional measures, could be used to identify those per-fectionistic individuals who are "traited" on the perfectionismconstruct. Traitedness refers to individual differences in the rele-vance or centrality of personality traits (see Britt, 1993). Indi-viduals who are traited on dimensions such as perfectionism,in comparison with those individuals who are not traited, mayexhibit a chronic tendency to ruminate about themes that arerelevant to the dimension in question. Similarly, perfectionismcognitions about the need to attain high standards may proveto be a key variable that distinguishes * 'normal1' and ' 'neurotic''perfectionists (Hamachek, 1978) and perfectionists with highversus low levels of ego involvement.

Another issue of importance is whether it is possible to applya multidimensional perspective to the study of perfectionisticthinking, in light of indications that the perfectionism constructis multidimensional at the trait level (Frost et al., 1990; Hewitt &Flett, 1991b). As discussed earlier, the current study focusedprimarily on perfectionistic thoughts involving the self. A multi-dimensional approach can be applied to the study of differentforms of perfectionism (i.e., self vs. social), and it can be usedto compare the varying role of personal and social cues intriggering perfectionistic thoughts.

Another key issue for future research involves the trait versusstate aspect of perfectionistic cognitions. Historically, negativeautomatic thoughts have usually been regarded as states (Becket al., 1983). Although it is likely that perfectionism cognitionshave a large state component, the possibility must be acknowl-edged that perfectionism cognitions reflect enduring traits forsome individuals. Regarding the stability issue, we obtainedtest-retest reliabilities of .67 in the student sample and .85 inthe psychiatric sample. If automatic thoughts are state I ike enti-ties that occur when a life situation activates the self-schema,the apparent temporal stability of the PCI suggests that theunderlying cognitive structure is activated chronically in someperfectionists who experience persistent ruminative thoughts. Akey goal of future research in our laboratory will be to examineperfectionism cognitions from a trait-state perspective to deter-mine whether these thoughts are best viewed as traits or states.

Finally, further investigation is needed to clarify the link be-tween perfectionism thinking and maladjustment. An investiga-

tion of perfectionism cognitions and other forms of negativeaffect, such as guilt, shame, and hostility, would provide furtherinsight into the nature of perfectionistic thinking. Additionalinsight would also be provided by investigations designed toestablish whether perfectionistic thinking is associated with vul-nerability to depression and anxiety (in terms of initial suscepti-bility to depression and anxiety) and the long-term persistenceof anxiety and depression in distressed individuals.

Summary

A program of research was conducted with student and clini-cal samples to demonstrate the presence of individual differ-ences in the frequency of perfectionistic cognitions. Evidencewas obtained to indicate that frequent perfectionism cognitionsare experienced by those perfectionists who are aware of thediscrepancy between their ideal standards and their actual char-acteristics. Subsequent examination of the association betweenthe frequency of perfectionism cognitions and maladjustmentconfirmed that the tendency to experience frequent thoughtsinvolving themes of perfectionism is linked with symptoms ofdepression and anxiety, and levels of perfectionism cognitionsaccount for variance in symptom scores that is not predictedby existing measures of automatic cognitions or trait levels ofperfectionism. Overall, these results suggest that the field ofpersonality and psychological adjustment may be advanced con-siderably by future theoretical and empirical attempts to assessthe excessive cognitive ruminations associated with certain per-sonality traits.

References

American Psychiatric Association. (1987). Diagnostic and statisticalmanual of mental disorders (rev. 3rd ed.). Washington, DC: Author.

Baldwin, M. W. (1992). Relational schemas and the processing of socialinformation. Psychological Bulletin, 112, 461-484.

Bandura, A. (1986). Social foundations of thought and action: A socialcognitive theory. Englewood Cliffs, NJ: Prentice Hall.

Barrow, J. C , & Moore, C. A. (1983). Group interventions with perfec-tionistic thinking. Personnel and Guidance Journal, 61, 612—6-15.

Beck, A. T. (1967). Depression: Clinical, experimental, and theoreticalaspects. New \brk: Harper & Row.

Beck, A. T., Epstein, N., Brown, C , & Steer, R. A. (1988). An inventoryfor measuring anxiety: Psychometric properties. Journal of Con-sulting and Clinical Psychology, 56, 893-897.

Beck, A. T., Epstein, N., & Harrison, R. (1983). Cognitions, attitudes,and personality dimensions in depression. British Journal of CognitivePsychotherapy, 1, 1-16.

Beck, A. T, Rush, A. J., Shaw, B. E, & Emery, G. (1979). Cognitivetherapy for depression. New "fork: Guilford Press.

Blankstein, K. R., Toner, B., & Flett, G. L. (1989). Test anxiety and thecontents of consciousness: Thought listing and endorsement measures.Journal of Research in Personality, 23, 269-286.

Blatt, S. J. (1995). The destructiveness of perfectionism: Implicationsfor the treatment of depression. American Psychologist, 50, 1003-1020.

Blatt, S. J., Quinlan, D. M., Pilkonis, P. A., & Shea, M. T. (1995). Im-pact of perfectionism and need for approval on the brief treatmentof depression: The National Institute of Mental Health Treatment ofDepression Collaborative Research Program revisited. Journal ofConsulting and Clinical Psychology, 63, 125-132.

Blatt, S.J., & Shichman, S. (1983). Two primary configurations of


psychopathology. Psychoanalysis and Contemporary Thought, 6,187-254.

Borkovec,T. D., Metzger, R. L., &Pruzinsky,T. (1986). Anxiety, worry,and the self. In L. M. Hartman & K. R. Blankstein (Eds.)i Perceptionof self in emotional disorder and psychotherapy {pp. 219-260). NewYork: Plenum.

Briggs, S. R., & Cheek^ J. M. (1986). The role of factor analysis in thedevelopment and evaluation of personality scales. Journal of Person-ality, 54, 106-148.

Britt, T. W. (1993). Metatraits: Evidence relevant to the validity ofthe construct and its implications. Journal of Personality and SocialPsychology, 65, 554-562.

Bums, D. D. (1980). Feeling good: The new mood therapy. New "fork:New American Library.

Burns, D. D., & Beck, A. T. (1978). Cognitive behavior modificationof mood disorders. In J. P. Fbreyt & D. P. Rathjen (Eds.), Cognitivebehavior therapy: Research and application (pp. 109-134). New"fork: Plenum.

Carver, C. S., LaVoie, L., Kuhl, J., & Ganellen, R. J. (1988). Cognitiveconcomitants of depression: A further examination of the roles ofgeneralization, high standards, and self-criticism. Journal of Socialand Clinical Psychology, 7, 350-365.

Clark, D. A. (1986). Cognitive-affective interaction: A test of the"specificity" and "generality" hypothesis. Cognitive Therapy andResearch, 10, 607-623.

Clark, D. A. (1988). The validity of measures of cognition: A reviewof the literature. Cognitive Therapy and Research, 12, 1-20.

Clark, D. A., Feldman, J., & Channon, S. (1989). Dysfunctional think-ing in anorexia and bulimia nervosa. Cognitive Therapy and Research,13, 377-387.

Clark, D. A., & Hemsley, D.R. (1985). Individual differences in theexperience of depressive and anxious, intrusive thoughts. BehaviorResearch & Therapy, 23. 625-633.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure oftests. Psychometrika, 16, 297-334.

Cundiff, G., & Gold, S. R. (1979). Daydreaming: A measurable con-cept. Perceptual and Motor Skills, 49, 347-353.

Diener, E. (1984). Subjective well-being. Psychological Bulletin, 95,542-575.

Diener, E., Emmons, R., Larsen, R., & Griffin, S. (1985). The Satisfac-tion With Life Scale. Journal of Personality Assessment, 49, 71-75.

Dombeck, M. J., Siegle, G. J., & Ingram, R. E. (1996). Cognitive inter-ference and coping strategies in vulnerability to negative affect. InI. G. Sarason, G. R. Pierce, & B. R. Sarason (Eds.), Cognitive inter-ference: Theories, methods, and findings (pp. 299-324). Mahwah,NJ: Erlbaum.

Eysenck, H. J., & Eysenck, S. B.G. (1991). Manual for the EysenckPersonality Scales. London: Hodder & Stoughton.

Fichten, C. S. (1986), Self, other, and situation-referent automaticthoughts: Interaction between people who have a physical disabilityand those who do not. Cognitive Therapy and Research, 10, 571 —587.

Flett, G. L., Hewitt, P. L., Blankstein, K. R., & Dynin, C. (1994). Di-mensions of perfectionism and Type A behaviour. Personality andIndividual Differences, 15, 477-485.

Flett, G.L., Hewitt, P. L., Blankstein, K. R., & Mosher, S.W. (1991).Perfectionism, self-actualization, and personal adjustment. Journal ofSocial Behavior and Personality, 6, 147-160.

Flett, G.L., Hewitt, P. L., Blankstein, K, R., & Mosher, S.W. (1995).Perfectionism, life events, and depressive symptoms: A test of a diathe-sis-stress model. Current Psychology, 14, 112-137.

Flett, G. L., Hewitt, P. L., Endler, N. S., & Bagby, R. M. (1995). Con-ceptualization and assessment of personality factors in depression.European Journal of Personality, 9, 309-350.

Flett, G. L., Hewitt, P. L., Endler, N. S., & Tassone, C. (1994). Perfec-tionism and components of state and trait anxiety. Current Psychology,13, 326-350.

Flett, G. L., Hewitt, P. L., & Mittelstaedt, W. M. (1991). Dysphoria andcomponents of self-punitiveness: A re-analysis. Cognitive Therapyand Research, 15, 201-219.

Frost, R. Q, Heimberg, R., Holt, C , Mattia, J., & Neubauer, A, (1993).A comparison of two measures of perfectionism. Personality andIndividual Differences, 14, 119-126.

Frost, R. O., & Henderson, K. J. (1991). Perfectionism and reactions toathletic competition. Journal of Sport and Exercise Psychology, 13,323-335.

Frost, R. 0., Marten, P. A., Lahart, C , & Rosenblate, R. (1990). Thedimensions of perfectionism. Cognitive Therapy and Research, 14,449-468.

Frost, R. O., Trepanier, K. L., Brown, E. J., Heimberg, R. G., Juster,H.R., Makris, G. S., & Leung, A. W. (1997). Self-monitoring ofmistakes among subjects high and low in perfectionistic concern overmistakes. Cognitive Therapy and Research, 21, 209-222.

Glass, C. R., & Arnkoff, D. B. (1982). Thinking cognitively: Selectedissues in cognitive assessment and therapy. In P. C. Kendall (Ed.)>Advances in cognitive-behavioral research and therapy (Vol. 1, pp.35-71) . New "fork: Academic Press.

Gold, D. B., & Wegner, D. M. (1995). Origins of ruminative thought:Trauma, incompleteness, nondisclosure, and suppression. Journal ofApplied Social Psychology, 25, 1245-1261.

Green, D. E., Walkey, F. H., McCormick, I. A., & Taylor, A.J.W.(1988). Development and evaluation of a 21-item version of the Hop-kins Symptom Checklist with New Zealand and United States respon-dents. Australian Journal of Psychology, 40, 61-70.

Guidano, V., & Liotti, G. (1983). Cognitive processes and emotionaldisorders. New \brk: Guilford Press.

Hamachek, D. E. (1978). Psychodynamics of normal and neurotic per-fectionism. Psychology, 15, 27-33.

Hankin, B. L., Roberts, J., & Gotlib, I. H. (1997). Elevated self-stan-dards and emotional distress during adolescence: Emotional specific-ity and gender differences. Cognitive Therapy and Research, 21, 663-680.

Hewitt, P. L., & Flett, G. L. (1990). Perfectionism and depression: Amultidimensional analysis. Journal of Social Behavior and Personal-ity, 5, 423-438.

Hewitt, P. L., & Flett, G, L. (1991a). Dimensions of perfectionism inunipolar depression. Journal of Abnormal Psychology, 100, 98-101.

Hewitt, P.L., & Flett, G.L. (1991b). Perfectionism in the self andsocial contexts: Conceptualization, assessment, and association withpsychopathology. Journal of Personality and Social Psychology, 60,456-470.

Hewitt, P. L., & Flett, G. L. (1993a). Dimensions of perfectionism, dailystress, and depression: A test of the specific vulnerability hypothesis.Journal of Abnormal Psychology, 102, 58-65.

Hewitt, P. L., & Flett, G. L. (1993b). Perfectionism and goal orientationin impulsive and suicidal behavior. In W. McCown, M. Shure, & J. L.Johnson (Eds.), The impulsive client: Theory, research, and treatment(pp. 247-263). Arlington, VA: American Psychological Association.

Hewitt, P. L., Flett, G. L., & Ediger, E. (1996). Perfectionism and de-pression: Longitudinal assessment of a specific vulnerability hypothe-sis. Journal of Abnormal Psychology, 105, 276-280.

Hewitt, P. L., & Genest, M. (1990). The ideal-self: Schematic processingof perfectionistic content in dysphoric university students. Journal ofPersonality and Social Psychology, 59, 802-808.

Higgins, E.T., Bond, R., Klein, R., & Strauman, T.J. (1986). Self-discrepancies and emotional vulnerability: How magnitude, accessi-bility, and type of discrepancy influence affect. Journal of Personalityand Social Psychology, 41, 1-15.


Hill, R. W, Mclntire, K., & Bacharach, V. R. (1997). Perfectionism andthe big five factors. Journal of Social Behavior and Personality, 12,257-270.

Hollon, S. D., & Kendall, P C . (1980). Cognitive self-statements indepression: Development of an automatic thoughts questionnaire.Cognitive Therapy and Research, 4, 383-395.

Huba, G. J., & Tanaka, J. S. (1983-1984). Confirmatory evidence forthree daydreaming factors in the Short Imaginal Processes Inventory.Imagination, Cognition, and Personality, 4, 139-148.

Ingram, R. E , Kendall, P. C , Siegle, G., Guarino, J., & McLaughlin, S.(1995). Psychometric properties of the Positive Automatic ThoughtsQuestionnaire. Psychological Assessment, 7, 495-507.

Ingram, R. E., & Wisnicki, K. S. (1988). Assessment of positive auto-matic cognition. Journal of Consulting and Clinical Psychology, 56,898-902.

Jackson, D. N. (1970). A sequential system for personality scale devel-opment. In C. D. Spielberger (Ed.), Current topics in clinical andcommunity psychology (pp. 61-96) . New Tfork: Academic Press.

Jenkins, C D . , Zyzanski, S.J., & Rosenman, R.H. (1971). Progresstoward validation of a computer scored test for the Type A coronary-prone behavior pattern. Psychosomatic Medicine, 33, 193-202.

Johnson, D. P., & Slaney, R. B. (1996). Perfectionism: Scale develop-ment and a study of perfectionistic clients in counseling. Journal ofCollege Student Development, 37, 29-41.

Kendall, P. C , Howard, B. L., & Hays, R.C. (1989). Self-referentspeech and psyehopathology: The balance of positive and negativethinking. Cognitive Therapy and Research, 13, 583-596.

Kline, P. (1987). Factor analysis and personality theory. European Jour-nal of Personality, 1, 21-36.

Klinger, E. (1977). Meaning and void: Inner experience and the incen-tive in people's lives. Minneapolis: University of Minnesota Press.

Klinger, E. (1989). Goal orientation as psychological linchpin: A com-mentary on Cantor and Kihlstrom's social intelligence and cognitiveassessments of personality. In R. S. Wyer & T. K. Srull (Eds.), Ad-vances in social cognition (Vol. 2, pp. 123-130). Hillsdale, NJ:Erlbaum.

Klinger, E. (1996). The contents of thoughts: Interference as the down-side of adaptive normal mechanisms in thought flow. In I. G. Sarason,G. R. Pierce, & B. R. Sarason (Eds.), Cognitive interference: Theo-ries, methods, and findings (pp. 3-23). Mahwah, NJ: Erlbaum.

Klinger, E., Barta, S. G., & Maxeiner, M. E. (1980). Motivational corre-lates of thought content frequency and commitment. Journal of Per-sonality and Social Psychology, 39, 1222-1237.

Kuhl, J. (1984). Volitional aspects of achievement motivation andlearned helplessness: Toward a comprehensive theory of action con-trol. In B. A. Maher (Ed.), Progress in experimental personality re-search (Vol. 13, pp. 99-171). New \brk: Academic Press.

Langston, C. A., & Sykes, W. E. (1997). Beliefs and the big five: Cogni-tive bases of broad individual difference factors in personality. Journalof Research in Personality, 31, 141-165.

Martin, L. L., & Tesser, A. (1989). Toward a motivational and structuraltheory of ruminative thought. In J. S. Uleman & J. A. Bargh (Eds.),Unintended thought (pp. 306-326). New %rk: Guilford Press.

McClelland, D. C. (1961). The achieving society, Princeton, NJ: VanNostrand.

Mclntosh, W.D., Harlow, T. R, & Martin, L.L. (1995). Linkers andnonlinkers: Goal beliefs as a moderator of the effects of everydayhassles on rumination, depression, and physical complaints. Journalof Applied Social Psychology, 25, 1231-1244.

Minarik, M. L., & Ahrens, A. H. (1996). Relations of eating behaviorand symptoms of depression and anxiety to the dimensions of perfec-tionism among undergraduate women. Cognitive Therapy and Re-search, 20, 155-169.

Moore, C. A., & Barrow, J. C. (1986). Perfectionistic thinking in univer-

sity students: Implications for individual therapy. In J. E. Talley &W. J. K. Rockwell (Eds.), Counseling and psychotherapy with collegestudents: A guide to treatment (pp. 100-112). New "fork: Praeger.

Mor, S., Day, H. I., Flett, G. L., & Hewitt, P. L. (1995). Perfectionism,control, and components of performance anxiety in professional per-formers. Cognitive Therapy and Research, 19, 207-226.

Morrow, J., & Nolen-Hoeksema, S. (1990). Effects of responses todepression on the remediation of depressive affect. Journal of Person-ality and Social Psychology, 58, 519-527.

Nolen-Hoeksema, S., Parker, L. E., & Larson, J. (1994). Ruminativecoping with depressed mood following loss. Journal of Personalityand Social Psychology, 67, 92-104.

Pacht, A. R. (1984). Reflections on perfection. American Psychologist,39, 386-390.

Parkinson, L., & Rachman, S. (1981). Part II. The nature of intrusivethoughts. Advances in Behavior Research and Therapy, 3, 101-110.

Paulhus, D. L. (1984). Two-component models of socially desirableresponding. Journal of Personality and Social Psychology, 46, 598-609.

Pavot, W., & Diener, E. (1993). Review of the Satisfaction With LifeScale. Psychological Assessment, 5, 164-172.

Pennebaker, J. W. (1989). Streams of consciousness and stress: Levelsof thinking. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought(pp. 327-350). New \brk: Guilford Press.

Riskind, J. H., Castellon, C. S., & Beck, A. X (1989). Spontaneouscausal explanations in unipolar depression and generalized anxiety:Content analysis of dysfunctional-thought diaries. Cognitive Therapyand Research, 13, 97-108.

Sarason, I. G., Pierce, G.R., & Sarason, B. R. (1996). Domains ofcognitive interference. In I. G. Sarason, G. R. Pierce, & B. R. Sarason(Eds.), Cognitive interference: Theories, methods, and findings (pp.139-152). Mahwah, NJ: Erlbaum.

Schwartz, R. M., & Garamoni, G. L. (1986). A structural model ofpositive and negative states of mind: Asymmetry in the internal dia-logue. In P. C. Kendall (Ed.), Advances in cognitive-behavioral theoryand research (Vol. 5, pp. 1-62). New York: Academic Press.

Singer, J. L., & Antrobus, J. S. (1970). Imaginal Processes Inventory.(Available from J. L. Singer, Yale University, New Haven, CT 06520)

Singer, J. L., & Antrobus, J. S. (1972). Daydreaming, imaginal pro-cesses and personality: A normative study. In P. W. Sheehan (Ed.).The function and nature of imagery (pp. 175-202). New Tfork: Aca-demic Press.

Slaney, R. B., Ashby, J. S., & Trippi, J. (1995). Perfectionism: Its mea-surement and career relevance. Journal of Career Assessment, 5, 279-297.

Terry-Short, L. A., Owens, R. G., Slade, P. D., & Dewey, M. E. (1995).Positive and negative perfectionism. Personality and Individual Dif-ferences, 18, 663-668.

Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judg-ing frequency and probability. Cognitive Psychology, 5, 207-232.

Vredenburg, K., Flett, G. L., & Krames, L. (1993). Analogue versusclinical depression: A critical re-appraisal. Psychological Bulletin,113, 327-344.

Watson, D., & Clark, L. A. (1984). Negative affectivity: The dispositionto experience aversive emotional states. Psychological Bulletin, 96,465-490.

Watson, D.f & Clark, L. A. (1991). The Mood and Anxiety SymptomQuestionnaire. Unpublished manuscript, Department of Psychology,University of Iowa, Iowa City

Watson, D., Clark, L. A., & Carey, G. (1988). Positive and negativefeedback and their relation to anxiety and depressive disorders. Jour-nal of Abnormal Psychology, 97, 346-353. »

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development andvalidation of brief measures of positive and negative affect: The

PERFECTI0N1STIC THINKING AND DISTRESS 1381

PANAS scales. Journal of Personality and Social Psychology, 54,1063-1070.

Watson, D., Claik, L. A., Weber, K., Assenheimer, J. S., Strauss, M. E., &McCormick, R. A. (1995). Testing a tripartite model: I. Evaluatingthe convergent and discriminant validity of anxiety and depressionsymptom scales. Journal of Abnormal Psychology, 104, 3-14.

Westra, H. A., & Kuiper, N. A. (1996). Communality and specificity ofdysfunctional cognitions and the prediction of four different forms ofpsychological maladjustment. Personality and Individual Differences,20, 575-588.

Zimmerman, M., & Coryell, W. (1987). The Inventory to DiagnoseDepression (IDD): A self-report scale to diagnose major depressivedisorder. Journal of Consulting and Clinical Psychology, 55, 55-59.

Zimmerman, M., & Coryell, W. (1988). The validity of a self-reportquestionnaire for diagnosing major depressive disorder. Archives ofGeneral Psychiatry, 45, 738-740.

Received December 27, 1997Revision received March 11, 1998

Accepted March 12, 1998 •

Psychological Distress and the Frequency of Perfectionistic …€¦ · Blankstein and Lisa Gray,...

Documents

Transcript of Psychological Distress and the Frequency of Perfectionistic …€¦ · Blankstein and Lisa Gray,...