Introduction

“‘How did I get to feel this way?’ she repeatedly asked herself. By asking, she hoped to transcend her depressed state; through understanding, she hoped to repair it. Instead her questions led her deeper and deeper inside herself—further away from the path that would lead to her recovery.”

  • Reported by Treynor et al. (2003, p. 247)

The dimensions of the Multidimensional Perfectionism Scale (MPS) developed by Hewitt and Flett (1991a) are differentially associated with various adaptational outcomes (see Flett and Hewitt 2002, for reviews). Coping is a major factor that has been associated with both perfectionism and outcomes such as depression and anxiety (e.g., Blankstein and Dunkley 2002). The current study examined the links between MPS components of perfectionism, a specific type of coping—Ruminative brooding, and multiple measures of current emotional distress, in university students. We further sought to determine whether perfectionism interacts with brooding to account for unique variance in these outcomes. We also investigated the role of gender differences. The study also permitted us to address the current controversy about the relative maladaptiveness or adaptiveness of perfectionism and the critical components of “clinical” perfectionism. Our findings have important implications for psychological theory, the identification of students “at-risk,” and prevention and counseling for perfectionistic students vulnerable to psychological distress.

Pernicious Effects of Perfectionism on Psychological Outcomes?

Although early theorizing and research considered perfectionism from a unidimensional perspective, most current work views it as a multidimensional cognitive-personality disposition. The Hewitt and Flett MPS (1991a) is a self-report tool based on their influential model that conceptualizes perfectionism as a three dimensional construct: Self-oriented perfectionism (SOP; high personal standards and motivation to attain perfection for oneself), socially prescribed perfectionism (SPP; perception of unrealistically high standards imposed by others), and other-oriented perfectionism (OOP; exceedingly high standards for others).

The SPP is the dimension that has been most strongly and consistently associated with negative adaptational correlates and consequences (see Flett and Hewitt 2002, for reviews). Hewitt and Flett (1991a) assumed that socially prescribed perfectionists perceive that they are failures due to criticism received from others who impose unrealistic expectations. Because socially prescribed perfectionists must meet expectations or receive disapproval, achievements become a means to win or maintain approval. SPP is clearly a concomitant of dysphoria in students and depression in psychiatric patients (e.g., Hewitt and Flett 1991a, b, 1993). SPP has also been associated with anxiety (Antony et al. 1998). SPP is also associated with trait anger and hostility (Dunkley and Blankstein 2000), psychosomatic distress (Dunkley and Blankstein 2000), and predictors of suicide risk (e.g., Blankstein et al. 2007).

The relation between SOP and negative adaptational outcomes is equivocal (see Flett and Hewitt 2002, for reviews) and some studies have reported a negative relation with depressive symptoms (e.g., Frost et al. 1993). It is also possible that there are adaptive aspects to SOP (e.g., Blankstein and Dunkley 2002; Frost et al. 1993; Klibert et al. 2005; Mills and Blankstein 2000).

Finally, OOP is also typically not strongly associated with distress (e.g., Flett et al. 1996) or is inversely related (e.g., Chang and Sanna 2001). The possibility that vulnerability associated with SOP or OOP is activated only by the presence of moderating factors (e.g., Hewitt and Flett 2002) is an important consideration discussed below.

Perfectionism and Distress: The Specificity Issue

Although it has been suggested that SPP is more broadly associated with emotional distress than SOP or OOP (e.g., Hewitt and Flett 1991a), little research has examined the specific relations between perfectionism components and multiple measures of current affect or different forms of emotional distress in the same sample. One exception is a study of children and adolescents by Hewitt et al. (2002). SPP was significantly correlated with measures of depression and anxiety, and with the expression of anger. Hewitt et al. (2002, p. 1058) concluded that since SPP and SOP “do not predict ‘pure’ symptoms of depression, anxiety, anger...they are associated with general negative affect.” However, Kawamura et al. (2001) examined the Frost version of the MPS (Frost et al. 1990) in relation to various features of trait anxiety and concluded that a social/trait/worry factor was related to “maladaptive” perfectionism independent of depression whereas depression was related to maladaptive perfectionism after controlling for three different anxiety factors. Maladaptive perfectionism was assessed using items that are strongly related to SPP (e.g., Frost et al. 1993; Dunkley et al. 2000). Similarly, Stober and Joorman (2001) controlled for general anxiety and depression and found that scores on two trait worry measures were still significantly correlated with combined maladaptive subscales from the Frost measure. In contrast, Minarek and Ahrens (1996) reported that two Frost maladaptive subscales correlated with scores on the Beck Anxiety Inventory (BAI; Beck et al. 1988) but when depression was controlled, the relations were no longer significant.

The discrepancies in the findings possibly reflect differences in the assessment of perfectionism, differences in the features of the distress measures used, and/or the fact that only one study (Hewitt et al. 2002) controlled for anger. Thus, we included multiple outcomes in the current study in order to determine whether relations between perfectionism and emotional distress are specific to different forms of distress.

Multidimensional Perfectionism and Coping

Are perfectionism dimensions and coping styles and processes related? SPP is unequivocally associated with indicators of maladaptive or a lack of adaptive coping. For example, SPP has been associated with an absence of task-oriented coping (Dunkley and Blankstein 2000), a negative problem-solving orientation (Flett et al. 1996), emotion-focused coping (Hewitt et al. 1995), absence of constructive thinking (Flett et al. 1994), and avoidance coping (Blankstein et al. 2007). SOP has been positively associated with adaptive coping tendencies, such as greater learned resourcefulness (Flett et al. 1991), positive problem-solving ability (Blankstein et al. 2007; Flett et al. 1996), constructive thinking (Flett et al. 1994), and task-oriented coping (Dunkley and Blankstein 2000), and less consistently associated with maladaptive coping strategies, such as emotion-focused coping (Hewitt et al. 1995), avoidance coping (Blankstein et al. 2007), a lack of self-acceptance in stressful situations (Flett et al. 1994), or unrelated to maladaptive coping (e.g., Dunkley and Blankstein 2000). OOP is only weakly associated with coping but has been associated with more positive appraisals of problem solving skills (Flett et al. 1996), and less seeking of support (Blankstein et al. 2007).

Taken together, current findings suggest that there is a pervasive link between SPP and maladaptive oriented coping, whereas SOP is associated with both adaptive and maladaptive coping tendencies. Further research is required to clarify the relation between OOP and coping.

Perfectionism and Ruminative Coping

Ruminative response orientation refers to a specific style or method of coping with a negative mood that involves self-focused attention (Lyubormirsky and Nolen-Hoeksema 1993). Based on Nolen-Hoeksema’s (1987) Response Styles Theory (RST), it is characterized by the tendency to passively and repetitively focus on distress symptoms or negative emotions and on the meaning and consequences of the distress (see Butler and Nolen-Hoeksema 1994; Nolen-Hoeksema et al. 1999; Nolen-Hoeksema 2000). For instance, ruminators ask themselves such things as “Why can’t I handle my problems?” Many studies have demonstrated that rumination predicts greater dysphoria or depressive symptoms in university students and psychiatric patients, predicts the onset of major depressive episodes, prolongs a depressed mood, and mediates the gender difference in depression (e.g., Just and Alloy 1997; Nolan et al. 1998; Nolen-Hoeksema 2000; Nolen-Hoeksema and Morrow 1993; Nolen-Hoeksema et al. 1998).

Spasojevic and Alloy (2001) consider rumination to be the common mediating link between various risk factors, including dispositional self-criticism, and subsequent depression. Self-criticism may be the most robust predictor of maladjustment among perfectionism components (Dunkley et al. 2006) and rumination has many correlates and consequences that are consistent with the empirical literature on maladaptive perfectionism (e.g., Blankstein and Dunkley 2002). Thus, dysphoric rumination disrupts coping and problem-solving processes through a focus on personal problems, negative tone, self-criticism, self-blame, and reduced self-confidence, optimism, and perceived control (Lyubormirsky et al. 1999). It also alienates people who could provide social support (Nolen-Hoeksema and Davis 1999).

Nolen-Hoeksema (1991) emphasized the ruminative focus on depression symptoms rather than on situations or self (see Cox et al. 2001; also see Nolen-Hoeksema et al. 1994) and, in contrast to those who conceptualize rumination as instrumental and adaptive (e.g., Martin and Tesser 1996), she views it as maladaptive (also see Papageorgiou and Wells 2003, for discussion of the relation between metacognitive beliefs and rumination). Rumination is frequently contrasted with distraction, an opposing tendency to avoid dwelling on negative emotions by deliberately focusing on pleasant thoughts or engaging in activities that divert attention in positive ways. Distraction can attenuate depressive symptoms or episodes (e.g., Morrow and Nolen-Hoeksema 1990; Nolen-Hoeksema et al. 1993).

Some researchers (e.g., Nolan et al. 1998) have questioned whether depressive rumination is a vulnerability that is specific to depression or whether it increases risk to other forms of distress, such as anxiety. Schwartz and Koenig (1996) examined the specificity issue in adolescents. Although rumination was uniquely associated with anxiety, the effect was not as strong as it was for dysphoria. They concluded that rumination is probably related to a general negative affectivity. Segerstrom et al. (2000, p. 672) noted that depressive rumination and trait worry share some common elements, especially “the idea of repetitive thought focused on negative events” (also see Papageorgiou and Wells 2003; Watkins and Mason 2002). Although they acknowledged that the associated thought processes could differ in content, Segerstrom et al. (2000) concluded that the repetitive thought common to both rumination and worry is related to both depression and anxiety. Nolen-Hoeksema (2000) reported that rumination also predicted anxiety symptoms in people with depressive disorders and hypothesized that rumination is especially characteristic of people with mixed anxiety/depressive symptoms. Similarly, Sarin et al. (2005) reported that depressive rumination predicted increases in both depressive and anxious symptoms in students. However, Morrison and O’Connor (2005) reported that rumination did not predict increased anxiety when various stress measures were first entered into regression equations. Rumination can also lead to an increase in anger (Rusting and Nolen-Hoeksema 1998). Consistent with our interpretation of the perfectionism literature, the extant findings suggest that rumination might be a general vulnerability factor for several forms of psychological distress. Therefore, we had another rationale for examining multiple distress outcomes in the current investigation.

Psychometric limitations in the measurement of depressive rumination have led to revised measures of rumination and important refinements to Nolen-Hoeksema’s (1987) theory. First, concerns have been raised about item overlap on the most widely used self-report measure of rumination, the 22-item Rumination Response Scale (Nolen-Hoeksema and Morrow 1991), and measures of depression symptoms (Cox et al. 2001; Segerstrom et al. 2000; Roberts et al. 1998; Treynor et al. 2003). Second, Treynor et al. (2003) reanalyzed the data from Nolen-Hoeksema et al. (1999) and after removing 12 potentially dysphoric contaminated items (based on overlap with Beck Depression Inventory items; BDI, Beck et al. 1961) conducted factor analyses of the ten remaining rumination items and identified two five-item factors referred to as reflection (“neutrally valenced...contemplation”) and brooding (“moody pondering”). Only brooding was associated with greater depression concurrently and in longitudinal analyses. Armey et al. (2003) identified an identical brooding component in students and concluded that brooding is the “active ingredient” of rumination. Brooding demonstrated significant relations with dysphoria and anxiety as assessed by the Mood and Anxiety Symptom Questionnaire (MASQ; D. Watson and L. A. Clark 1991, unpublished manuscript), BDI depression, and trait worry assessed by the Penn State Worry Questionnaire (PSWQ; Meyer et al. 1990).

Perfectionism and Rumination

What is the link between perfectionism and rumination? Flett et al. (2002) administered the MPS, an abbreviated Response Styles Questionnaire (see Butler and Nolen-Hoeksema 1994), and the short version of the MASQ to undergraduate students (53 women and 12 men). They reported that both SPP and SOP correlated positively and significantly with rumination. Both SPP and rumination were associated positively with three of four MASQ subscales (General Depression, General Anxiety, and Anxious Arousal). SOP was positively correlated with General Depression only.Footnote 1 However, SPP and SOP were no longer associated with the indices of psychological distress after removing the variance attributable to rumination. More recently, Burns and Fedawa (2005) examined the relations among “positive” and “negative” perfectionism as assessed using a scale developed by Terry-Short et al. (1995), rumination assessed by the Nolen-Hoeksema and Morrow (1991) measure, and dysphoria in undergraduates (129 women and 92 men). Burns and Fedawa (2005) found a moderate positive association between negative perfectionism and rumination and concluded that rumination is a main coping style for negative perfectionists.Footnote 2 , Footnote 3

Unfortunately, these two studies did not address the issue of potential contamination with depression in the rumination scale or examine a brooding factor distinct from a reflection factor. In the present study, we examined perfectionism in relation to rumination uncontaminated with distress while emphasizing relations with a moody pondering or brooding factor derived from the version of the rumination scale employed by Nolen-Hoeksema et al. (1993). In contrast to all past research, we asked participants to discriminate between rumination related to periods of dysphoria and rumination specific to anxiety/worry.

Perfectionism and Moderating Factors

Several studies have provided limited support for diathesis-stress or specific-vulnerability (congruency hypothesis) models proposed by Hewitt and Flett (1993, 2002): Perfectionists who experience high levels of stress and/or specific types of stress are vulnerable to emotional distress. Numerous other potential moderators have been examined, including coping. For example, O’Connor and O’Connor (2003) found that high SOP predicted increases in hopelessness when cognitive reconstruction coping (an adaptive coping style) was low. In a prior report on the sample employed here (Blankstein et al. 2007), we noted that each of the MPS perfectionism components interacted with specific moderators (e.g., problem-solving coping, hassles, social self-esteem, optimism, and social support) to enhance or buffer the perfectionism-suicide risk link. In the current study, we examined possible interactions between perfectionism and ruminative brooding in the prediction of emotional distress.

Perfectionism, Rumination, and Gender

We previously concluded (Blankstein and Winkworth 2004; Blankstein et al. 2007) that gender should be considered in studies of perfectionism as gender-specific patterns of influence for different components were observed for predictions of adaptational outcomes. For example, Blankstein et al. (2007) reported that SPP was associated positively with avoidance coping in women but not in men whereas SOP was associated positively with problem solving coping in men but not in women. Further, there are reported gender differences in the way that men and women cope (e.g., Hampel and Petermann 2005; Thoits 1991). We also found some gender-specific interaction effects.

Past research on perfectionism and rumination has not examined the role of gender. Because the ruminative response style is more characteristic of women than men (e.g., Nolen-Hoeksema et al. 1993; Nolan et al. 1998), we believe that it is also important to examine possible gender differences in the relations among perfectionism, brooding, and distress.

Present Study: Focus

The current study examined (a) relations between dimensions of perfectionism and ruminative and distractive coping, particularly the brooding component of rumination; (b) relations between perfectionism, brooding, and multiple measures of current emotional distress–dysphoria, anxiety, worry, and anger; (c) the issue of specificity, viz., whether different perfectionism components and ruminative brooding are more strongly associated with different forms of distress; (d) whether perfectionism adds incremental validity to the prediction of distress beyond what is accounted for by brooding, and vice-versa; (e) whether interactions between perfectionism components and brooding components augment the prediction of outcomes; (f) possible gender differences in relations, and (g) the issue of adaptive versus maladaptive and “clinical” perfectionism.

Methods

Participants

This study reports additional findings from the sample used by Blankstein et al. (2007). Participants were 205 undergraduate students (144 women; 61 men) enrolled in an Introductory Psychology course at the University of Toronto at Mississauga. They had a mean age of 22.1 years (SD = 6.5). Participants received credit toward their course grade.

Procedure

Participants completed a package of questionnaires that included measures of perfectionism, rumination-distraction, and current distress. Measures were completed in 2 h small group sessions. The relevant measures were presented in random order, intermixed with other measures that were not a focus of the present study (see Blankstein et al. 2007).

Measures

Multidimensional Perfectionism Scale

Perfectionism was assessed using the reliable and valid Hewitt and Flett (1991a) MPS. Respondents rate 45 items on a seven-point Likert scale. The 15-item subscales measure SPP (e.g., The people around me expect me to succeed at everything I do), SOP (One of my goals is to be perfect in everything I do), and OOP (e.g., If I ask someone to do something, I expect it to be done flawlessly).

Rumination and Distraction

The checklist of 16 Ruminative Responses and 16 Distracting Responses employed by Nolen-Hoeksema et al. (1993) was adapted for the present study. Participants completed two versions, presented in counterbalanced order. One version asked them to check items that “match closely what you think about or how you behave when you feel sad or depressed” whereas the other was specific to feeling “worried or anxious.” Examples of the ruminative responses include thoughts such as “I need to understand these feelings” and behaviors such as “go to my room alone and think about my feelings.” Examples of the distracting responses include thoughts such as “I will get my mind on something else other than how I feel” and behaviours such as “do something active to get my mind off my feelings.”

Consistent with the refined RST (Treynor et al. 2003), we constructed a measure of rumination unconfounded with depression or anxiety by deleting items potentially confounded with symptoms and then used exploratory factor analysis (principal components with varimax rotation) to identify the maladaptive “brooding” component. These analyses conducted separately for women and men and on responses to “sad or depressed” and “worried or anxious” versions resulted in a six-item Ruminative Brooding Checklists (see Appendices 1 and 2). Three items are identical to items on the five-item scale developed by Treynor et al. (2003). All items are brooding thoughts as opposed to ruminative behaviors. Reliabilities ranged between 0.69 and 0.73. Initial tests of construct validity determined that brooding is related to other constructs in expected ways.Footnote 4

Current Dysphoria

The Beck Depression Inventory (BDI; Beck et al. 1961; Beck and Steer 1987) is a 21-item measure of depressive symptoms, including aspects of emotional, physical, and cognitive state. Numerous studies (e.g., Beck and Steer 1987; Beck et al. 1988) have confirmed the BDI’s high internal consistency and validity.

Current Anger

The ten-item State Anger scale from the State-Trait Anger Expression Inventory (STAXI; Speilberger 1991) measures the intensity of angry feelings at a particular time (e.g., I am furious). In order to be consistent with other outcome measures employed in the present study, “the Past Week, Including Today” was chosen as the time frame, instead of the STAXI’s “Right Now.” The original State Anger scale has demonstrated excellent internal consistency and validity (Speilberger 1991). Dunkley and Blankstein (2000) reported reliability and construct validity for the modified version. In the present study, the Current Anger Inventory (CAI) had internal consistencies of 0.89 and 0.93 for the female and male samples, respectively.

Current Anxiety

Current anxiety severity was assessed using the 21-item BAI (Beck et al. 1988). Participants indicate the extent to which they have been bothered by each symptom in the previous week, including the present day. The symptoms constitute common physiological and cognitive symptoms experienced by clinically anxious people. The BAI shows better discrimination between depression and anxiety than many other widely used anxiety measures (Beck et al. 1988; Clark and Watson 1991).

Current Worry

The 16-item PSWQ (Meyer et al. 1990) measures the trait-like tendency to worry, and focuses on dimensions of frequency, controllability, and distress independent of the content (e.g., My worries overwhelm me). It is regarded as a measure of “pathological worry.” It was adapted for the present study to assess worry during “the Past Week, Including Today.” The Current Worry Inventory (CWI) consists of the PSWQ items rewritten to reflect the extent of worry over the past week. For example, the item “When I am under pressure I worry a lot” was changed to “When I was under pressure I worried a lot.” The response format was changed from a five-point scale that ranged from “Not at all typical” to “Very typical” to a four-point Likert scale to assess “extent to which you have been worrying during the Past Week, Including Today.” The new scale ranged from “Rarely or none of the time” to “Most or all of the time.” In the present study, coefficient alpha for the CWI was 0.94 and 0.93 for women and men, respectively.

Results

Descriptive Statistics

The mean and standard deviation for all variables (see Table 1) are comparable to those reported previously for student populations. Reliabilities (coefficient alpha) were acceptable for research purposes, although depressive distraction was quite low (0.65) in women.Footnote 5

Table 1 Means, standard deviation, and reliabilities (coefficient alpha) for females and males, and gender differences
Full size table

Gender differences (see Table 1) were examined using one-way (Group) multivariate analysis of variance (MANOVA) using Wilk’s criterion followed by univariate analyses (ANOVAs). Significant differences emerged for both depressive and anxious rumination and brooding as well as for anxious distraction. Women were higher than men in both rumination and brooding whether depressed or anxious, but men were higher in the use of distraction when anxious. Women also reported higher levels of current dysphoria, anxiety, and worry.

Correlations between Perfectionism Components, Rumination-distraction, and Distress

All correlations are p < 0.05, one-tailed, uncorrected, unless otherwise specified. Consistent with Cohen (1992), we considered correlations of 0.10, 0.30, and 0.50 to represent small, medium/moderate, and large relations, respectively. Depressive brooding and anxious/worried brooding were strongly inter-correlated (women, r = 0.67; men, r =0.82) but not redundant with each other. The distress measures were also moderately inter-correlated (range equals r = 0.41–0.60 for women, and r = 0.39–0.68 for men).Footnote 6

Table 2 summarizes the zero-order and partial correlations (in parentheses) between the perfectionism dimensions and rumination-distraction and distress. Partial correlation analyses examined the unique relation between specific MPS subscales and rumination-distraction and outcome variables with other perfectionism dimensions partialled out. Examination of Table 2 reveals the following: (1) there were moderately strong correlations between SPP and rumination and brooding variables for both women and men; (2) the correlations were at least as strong for both depressive and anxious brooding which is unconfounded with symptoms; (3) SPP was unrelated or only weakly related to both depressive and anxious distraction; (4) SOP was only weakly related (women) or unrelated (men) to rumination and brooding even when the influence of SPP and OOP were partialled out; (5) OOP was unrelated to rumination and brooding, even when the other perfectionism components were partialled out; (6) both SOP and OOP were unrelated to distraction; (7) SPP was moderately or strongly positively correlated with all current distress measures; (8) the correlations with distress remained highly significant when SOP and OOP were partialled out; (9) with few exceptions, SOP was moderately correlated with distress in both women and men; (10) the SOP correlations with distress generally remained significant when SPP and OOP were partialled out; and (11) OOP was unrelated to distress, even when the other perfectionism components were partialled out.

Table 2 Zero-order and partial correlations (parentheses) between perfectionism dimensions and rumination-distraction and outcome variables by gender
Full size table

Zero-order correlations between the rumination-distraction variables and the outcome variables are summarized in Table 3. Depressive rumination was positively associated with current dysphoria in both women and men. Depressive rumination was also associated with the other measures of current distress, albeit only weakly with anger especially in men. Similarly, rumination associated with feeling anxious was associated positively with all distress measures. Both depressive and anxious distraction were unrelated to distress in women but were positively associated with distress in men. As predicted, our measures of brooding, un-confounded with symptoms and reflective pondering, were positively correlated with current distress. Although depressive brooding was most strongly associated with dysphoria, both forms of brooding were associated with all forms of distress.

Table 3 Zero-order correlations between rumination-distraction and distress by gender
Full size table

Perfectionism, Ruminative Brooding, and the Prediction of Current Distress

Next, we conducted a series of hierarchical multiple regressions that examined the ability of perfectionism dimensions and brooding to predict current distress. Table 4 summarizes these results. The MPS subscales were entered in the first block followed by a block that included depressed and anxious brooding. The reverse analyses were also conducted with the brooding variables entered first followed by the perfectionism subscales.

Table 4 Hierarchical regression analyses examining unique predictive validity of MPS perfectionism dimensions and depressive and anxious brooding by gender
Full size table

Regarding effect size, in the first block, the MPS subscales accounted for 30, 25, 26, and 16% of the variance in current dysphoria, anxiety, worry, and anger, respectively, for women compared to 31, 35, 26, and 16% variance in the distress outcomes, respectively, for men. The pattern of unique predictors differed between women and men. In women, SPP was a consistent positive predictor of current distress and SOP was an additional unique predictor of both anxiety and worry. In men, the pattern of unique effects was more complex: SPP was a positive predictor of dysphoria, anxiety, and worry, SOP was a positive predictor of dysphoria, anxiety, and anger, and OOP was a negative predictor of both current dysphoria and anxiety.

In the second block, the incremental contribution of the brooding variables in the prediction of current distress was significant for women, accounting for 5–15% of additional variance. Depressive brooding was a unique predictor of dysphoria and anger whereas anxious/worried brooding predicted anxiety and worry. The picture was again different for men. The block accounted for an additional 21 and 10% of unique variance in dysphoria and anxiety and depressive brooding was a significant positive predictor of both forms of distress. However, although the block predicted significant additional variance (8 and 11%) in worry and anger, respectively, neither depressive nor anxious brooding was a unique predictor.

In a second set of regressions, brooding was entered on the first block and perfectionism was then entered in a second block to determine whether the block of MPS measures and/or individual components still predicted unique variance in each outcome after controlling for the effects of brooding. As shown in Table 5, despite the relatively large and significant effects of brooding on current distress (anger in males being the only exception), perfectionism was uniquely related to distress over and above brooding in both women and men, accounting for an additional significant 9–12% variance in women, and 15–24% unique variance in men. Consistent with the first set of regressions, with the exception of anger in men, SPP was a significant positive predictor of all forms of distress in women and men. Similarly, SOP was still a positive predictor of worry in women, and dysphoria, anxiety, and anger in men. Further, OOP remained a significant negative predictor of dysphoria in men suggesting a different relation between perfectionism components and dysphoria in men relative to women.

Table 5 Hierarchical regression analyses examining incremental validity of MPS perfectionism dimensions over and above depressive and anxious brooding by gender
Full size table

Tests of Moderator Hypotheses

In a series of regression analyses, we tested whether interactions between perfectionism subscales and brooding subscales predict unique variance in distress over and above variance predicted by components of the interactions. We used the relatively stringent strategy described by Blankstein et al. (2007), consistent with the recommendations of Frazier et al. (2004). Thus, we entered perfectionism as an initial block, brooding as a second block, and the six multiplicative terms (each of the two brooding types in interaction with the three perfectionism subscales) in the final step. The results with perfectionism in the first block and brooding in a second block (and vice-versa) were reported above. With one exception, the two types of brooding did not interact with any of the perfectionism dimensions to predict significant incremental variance in any of the forms of current distress, in either women or men. For current worry in males, the overall block was significant (17% incremental variance, p < 0.05); however, none of the within block specific predictors was significant.

Discussion

The present study sought to test predictions and explore relations among dimensions of perfectionism, ruminative and distractive coping, and multiple indicators of current distress. In the sections that follow, we review our main findings and discuss the ways in which this research advances current knowledge on the links between perfectionism components and, in particular, a brooding factor of ruminative coping, the issue of specificity in the relation between components of perfectionism and different forms of emotional distress, and whether or not certain perfectionism dimensions are relatively adaptive.

Perfectionism and Ruminative Brooding

To our knowledge, this is the first study to examine links between perfectionism and a brooding component of rumination first identified by Treynor et al. (2003). Consistent with Treynor et al. (2003), women engaged in more depressive brooding than men. Our finding that women also engage in more anxious brooding is unique. More important, SPP was moderately positively correlated with both depressive brooding and anxious brooding in both women and men. Indeed, the correlations tended to be stronger between SPP and brooding relative to the SPP-rumination correlations, despite the fact that the original rumination measure is confounded with symptoms. Further, these relations held even when the influence of the other perfectionism dimensions was partialled out. Consistent with Flett et al. (2002) who also reported a significant relation between SOP and rumination, SOP was positively correlated with both depressive and anxious brooding but the effects were significant in women only. However, when we removed the mutual overlap with the other perfectionism components, only the link between SOP and anxious brooding remained significant. OOP and brooding were unrelated.

Treynor et al. (2003) raised a key question, viz., “why some people, when they contemplate their problems and feelings of distress, are able to engage in adaptive reflection whereas others fall into brooding” (p. 257). At a trait or distal level, we believe the socially prescribed perfectionist is strongly predisposed to brood when emotionally distressed. Thus, while brooding probably affects the intensity, duration, and likelihood of recurrence of distress, some people are more likely to engage in this cognitive coping style than others, viz., people high in SPP.

Perfectionism, Brooding, and Current Distress

Past theory (e.g., Hewitt and Flett 1991b) predicts that dispositional perfectionism confers vulnerability to depression and other forms of emotional distress. The accumulated corpus of evidence shows that specific MPS components are reliably related to depression or dysphoria as well as anxiety but few studies have examined some other forms of current distress (e.g., anger or worry), or multiple measures of current affect or different forms of distress in the same sample. Further, given the association between ruminative brooding and dysphoria, anxiety, worry, and anger, it was important to determine not only the unique predictive validity of trait perfectionism but also the incremental validity of the MPS dimensions over and above depressive and anxious brooding since brooding also appears to be a traitlike cognitive vulnerability (e.g., Nolen-Hoeksema and Davis 1998; Roberts et al. 1998).

The MPS dimensions were uniquely correlated with current distress and the pattern of relations differed somewhat for women and men. The regression analyses confirmed that perfectionism is predictive of each measure of distress and the effect sizes were relatively large. SPP was clearly the strongest and most consistent predictor, controlling for the other components, the only exception being the prediction of current anger in men. However, SOP also played a significant role, predicting anxiety and worry in women and anxiety, dysphoria, and anger in men. OOP was a negative predictor of both dysphoria and anxiety but in men only. The pattern of effects suggests that the relation between perfectionism components and adaptational outcomes is possibly more complex in men than women. These are novel and, if replicated, important findings since most past research has not considered the possibility that the link between perfectionism and distress is expressed differently in men and women. It will be a task for future research to explicate gender-specific findings. Perhaps, as suggested by Klibert et al. (2005), they reflect gender-specific reactions to different roles and demands.

Despite the strong associations between perfectionism and distress, ruminative brooding accounted for unique variance when entered in a second block in the regressions. When entered first, brooding accounted for significant and substantial variance in current dysphoria, anxiety, and worry. Discriminant validity for our specific focus of brooding on either depressive or anxious/worried symptoms was evident, particularly for women: Depressive brooding was a strong predictor of dysphoria whereas anxious/worried brooding was a strong predictor of anxiety. In men, depressive brooding was a unique predictor of both dysphoria and anxiety.

The relations with current anger were quite weak, although depressive brooding was a unique predictor in women. However, we did not ask participants to endorse brooding thoughts when angry. Perhaps a specific focus on angry brooding would yield stronger relations to current anger. Gilbert et al. (2005) examined depression-focused rumination and anger-focused rumination in relation to dysphoria and found that both forms were correlated with each other and with the BDI. They did not assess current anger. Our findings are important because they confirm that brooding plays a possibly important role as a risk factor in multiple forms of current distress in addition to depression (also see Armey et al. 2003).

To our knowledge, this is the first study to find preliminary evidence for the unique predictive power of the Hewitt and Flett (1991a) multidimensional perfectionism personality risk measure when compared with the brooding cognitive vulnerability factor. Despite large effect sizes for ruminative brooding, perfectionism accounted for significant incremental variance in dysphoria, anxiety, worry, and anger, suggesting that the constructs assessed by the MPS, especially SPP, are not merely generic risk factors interchangeable with Nolen-Hoeksema’s rumination vulnerability, particularly the brooding component uncoupled from symptom overlap.

We explored whether anxious or depressive brooding moderated the association between perfectionism components and current distress. Although there is some support for moderator hypotheses, in the present instance we failed to detect meaningful significant interactions. For example, contrary to expectations the highest levels of current dysphoria were not reported by students who were high in both SOP and depressive brooding. That is not to say that such interactions would not predict the chronicity or recurrence of depression or be statistically significant and clinically meaningful in people with diagnosed mood or other disorders.

Perfectionism and the Brooding Trap

Clearly the linkage between different dimensions of perfectionism, the brooding component of rumination, and different forms of current distress is complex, and other important variables need to be considered. According to RST (Nolen-Hoeksema 1987, 1991; Nolen-Hoeksema et al. 1993) rumination has the power to exacerbate depression (and other forms of distress) because it constricts focus of attention and concentration to a set of negative thoughts associated with emotional distress, enhances the recall of negative events, and reduces adaptive coping. A vicious cycle of negative thinking and negative emotions ensues in which the person comes to believe that there is little he or she can do to master or overcome problems. The person is trapped by rumination, or in the present context—Brooding that “passive contemplation of what’s wrong in your life and how you wish it were better” (Treynor et al. 2003, p. 257), and “spirals” deeper into depression or distress. The person feels trapped by and wants to escape from these thoughts and feelings but cannot control the rumination (entrapment) (see Gilbert et al. 2005).

We believe that certain perfectionists are predisposed to fall into the brooding trap because they have stringent expectations for themselves or perceive that others have rigid and extreme expectations for them, but life rarely works out perfectly for them, either because they experience failures in various domains (e.g., for students academics and interpersonal relationships are critical), or perceive that they have failed, or are made to feel that they have failed by the attitudes and reactions of significant others.

Papageorgiou and Wells (2003) recently tested a more complex clinical metacognitive model of rumination and depression that could be applicable to perfectionists. In their model, positive beliefs about the benefits and advantages of rumination (e.g., “Ruminating about my depression helps me to understand past mistakes and failures”) motivate people to engage in rumination; however, they then appraise the process as uncontrollable and harmful, with detrimental social consequences, thus contributing to the experience of depression. Papageorgiou and Wells (2003) propose that vicious cycles of rumination, depression, and metacognition perpetuate depression and they provided evidence in support of the model. But what causes the positive metacognitive beliefs in the first place? It will be important to determine in future research if perfectionists, particularly socially prescribed and/or self-oriented perfectionists, develop metacognitive beliefs about rumination, especially a brooding component, perhaps as a consequence of perceived failures, negative evaluations from others, self-criticism, and self-blame. If so, then we can add a focus on specific strategies to change the metacognitive beliefs of distressed perfectionists.

If rumination truly is the common mediating link between depressive risk factors and depression (Spasojevic and Alloy 2001), and proves to be a critical link between perfectionism and distress, then it is important for therapists to help perfectionists interrupt this negative cycle, to break out of the ruminative trap, especially the brooding aspects of that trap, so that it no longer sustains and amplifies emotional distress. Teasdale and his colleagues (Teasdale et al. 2000, 2002) proposed that metacognitively standing back from negative thoughts to become aware of their accuracy is a key feature of cognitive therapy. “Mindfulness” allows dispassionate observation of thoughts and can short circuit cycles of negative thinking. Teasdale et al. (2002) developed a mindfulness-based cognitive therapy specifically to prevent relapse in depressed people. Treynor et al. (2003) also suggested that cognitive therapists employ strategies that provide evidence to people that they can exert control over their problems and to respond to the tendency to engage in “toxic” social comparisons, a key aspect of their brooding factor and of our own. It will be of interest to determine if these approaches can be adapted for perfectionists.

Ruminating on negative self-cognitions is, of course, a key to cognitive theories and treatment of emotional disorders (e.g., Beck 1983; Ellis 2002). Treynor et al. (2003, p. 256) proposed that brooding reflects “a passive comparison of one’s current situation with some unachieved standard.” This position is similar to an earlier theory of Carver and Scheier (1981) to the effect that rumination leads to distress when a person cannot resolve the discrepancy between current and desired states. In terms of the treatment of perfectionists, we have always emphasized the importance of using cognitive therapy to modify the underlying negative schemata, irrational beliefs, and maladaptive cognitions characteristic of many perfectionists, particularly the socially prescribed perfectionist (e.g., Blankstein et al. 2007). Perfectionists bring much more than distress associated with brooding to the consulting room and it is widely recognized that they are difficult to treat (e.g., Blatt et al. 1998; Ellis 2002). Extreme perfectionists typically engage in self-critical thinking, self-doubt, self-blame, thoughts about failure, worry about disapproval, and rumination about a perceived or actual discrepancy between performance and perceived standards or standards imposed by others (see Flett and Hewitt 2002, for reviews). Blatt (1995) argued that perfectionists also ruminate about their reputations, perceived status, and the shame related to perceived inadequate performance. It is necessary to address these other forms of rumination.

Perfectionism, Distraction, and Distress

Although it was not a central focus, we also examined the relation between perfectionism and distraction as assessed by the measure developed by Nolen-Hoeksema et al. (1993). Men reported greater use of distraction than women, although the effect was significant for depressive distraction only. Broderick and Korteland (2002) reported that both boys and girls believe that men should distract themselves from problems rather than ruminating. Thus, we should expect men university students to employ distraction as a coping strategy. However, we also found that both forms of distraction were positively associated with current distress in men, and that distraction was essentially unrelated to perfectionism. These findings were unexpected for two reasons. First, distraction is intended to be benign (see Nolen-Hoeksema 1991). Perhaps our finding is related to the way in which distraction was measured (see Nolen-Hoeksema et al. 1993) or to sample characteristics (see Schwartz and Koenig 1996).

Second, the lack of a positive association between perfectionism and distraction is surprising given the possibility that SOP can sometimes be adaptive or appear to be adaptive, and that SOP has been associated with adaptive coping (e.g., Dunkley and Blankstein 2000). We expected SOP to be positively correlated with distraction coping (assuming distraction is operationalized as, and proves to be, adaptive), particularly in men. Flett et al. (2002) also reported that a similar measure of distraction was unrelated to perfectionism.

Is Perfectionism Adaptive?

The present results also inform the debate about the adaptiveness–maladaptiveness of different perfectionism components and further highlight the importance of considering possible gender differences. In addition to the fact that SOP was independently associated with brooding in women, SOP was associated with some forms of current distress in both women and men even after controlling for the influence of SPP. Thus, in these instances SOP is clearly not adaptive. Further, the present findings do reinforce positions that consider extreme SOP to be generally debilitative (e.g., Hewitt and Flett 1991a; Shafran et al. 2002), regardless of the role that moderator variables can play in either enhancing or buffering the link between perfectionism components and distress. Although Klibert et al. (2005) proposed that it is the combination of high scores on both SPP and SOP that is associated with the most negative adaptational outcomes (also see Blatt 1995), our results suggest that “pure” SOP can be associated with distress. Controlling for the other perfectionism dimensions, SOP predicted current anxiety and current worry in women. If we had assessed dysphoria and anger only, we would have concluded incorrectly that SOP had more limited concurrent validity. As suggested by O’Connor and O’Connor (2003), perhaps SOP can be pernicious if it is not kept in check. Nonetheless, consistent with the conclusion of Stoeber and Otto (2006), the strivings of self-oriented perfectionists might generally have positive consequences but if they also have concerns about being evaluated and about making mistakes (concerns especially pertinent to university students) then they might be more predisposed to experience anxiety and worry, especially if they encounter actual or perceived failures.

Limitations

Several limitations suggest additional research directions. First, future research should examine the generalizability of the current findings in both community-based samples and clinical samples. Second, future research should assess perfectionism and brooding using more sophisticated assessment techniques, such as structured clinical interviews, peer or spouse ratings, and direct behavioral and cognitive assessments. Thought listing, concurrent verbalization strategies, and endorsement measures (see Blankstein and Segal 2001) could also be employed to assess the cognitive processes of perfectionists engaged in various tasks in which failure is naturally experienced or deliberately induced. For example, Besser et al. (2004) conducted an experiment in which they demonstrated that high SOP students had negative cognitive and affective responses to failure feedback on an ego-involving task. Third, the current study was correlational and causality cannot be inferred. Longitudinal designs are necessary. Finally, we should examine expanded models that determine factors that moderate and mediate relations among perfectionism, cognitive processes such as brooding, and different forms of distress. Limitations not withstanding, our findings further illustrate the complexity of perfectionism as a multidimensional construct and underscore the need to develop comprehensive, multidimensional, and integrated models to help us to understand the links between perfectionism and adaptational outcomes in both women and men.