Associations Between Brooding, Reflection, Depressive Symptoms ...

Download PDF
22 downloads 0 Views 150KB Size Report
Comment
Development of depression from preadolescence to young adulthood: Emerging sex differences in a 10-year longitudinal study. Journal of Abnormal Psychol-.
Journal of Clinical Child and Adolescent Psychology 2007, Vol. 36, No. 1, 56–65

Copyright © 2007 by Lawrence Erlbaum Associates, Inc.

Subtypes of Rumination in Adolescence: Associations Between Brooding, Reflection, Depressive Symptoms, and Coping Rebecca A. Burwell Westfield State College

Stephen R. Shirk University of Denver Prior research has indicated that rumination contributes to the maintenance or intensification of depressive symptoms among adults. This study examined associations between rumination and depressive symptoms in early adolescence. Using a shortterm longitudinal design, we evaluated relations between subtypes of rumination and both depressive symptoms and coping among a community sample of 168 adolescents (70 boys, 98 girls, age M = 13.58). Results provided support for brooding and self-reflective subtypes of rumination. Brooding, but not reflection, predicted the development of depressive symptoms over time, particularly for girls. Brooding was related to maladaptive disengagement coping strategies, whereas reflection was related to adaptive primary and secondary coping strategies. These results suggest that not all types of self-focus on emotion contribute to the maintenance or intensification of depressive symptoms. According to Nolen-Hoeksema (1991), rumination is a relatively stable maladaptive coping strategy that consists of “focusing on the symptoms and the possible causes and consequences of those symptoms” (p. 569) and is distinguished from problem solving by the absence of actions aimed at changing the situation or relieving distress. Individuals who dwell on their depressed mood or symptoms, as well as those who focus on their depressed mood to gain insight, are expected to show increases in depressive symptoms over time. In short, rumination has been posited as one of the processes that transforms normative distress, especially sadness, into depression (Nolen-Hoeksema, 1991, 2004). Rumination has been advanced (Nolen-Hoeksema, 1991; Nolen-Hoeksema & Girgus, 1994) as a possible explanation for both the increase in depressive symptoms as well as the emergence of gender differences in depressive symptoms during adolescence (Hankin et al., 1998; Hankin & Abramson, 2002; Ge, Lorenz, Conger, Elder, & Simons, 1994), although little is known about rumination during this critical period. Among adults, results from a number of prospective naturalistic studies have supported the role of rumination in the maintenance or exacerbation of depressive mood and symptoms, even when controlling for baseline depression (e.g., Katz & Bertelson, 1993; Nolen-

Hoeksema & Morrow, 1991). Further, rumination has been linked with the onset of depressive episodes (Just & Alloy, 1997) and predicts the severity of depression among clinical samples (Just & Alloy, 1997; Kuehner & Weber, 1999). Of the few studies that have examined rumination among adolescents, evidence for an association between rumination and depression is mixed (Abela, Vanderbilt, & Rochon, 2004; Grant & Compas, 1995; Hart & Thompson, 1996; Schwartz & Koenig, 1996), and the absence of longitudinal data has made predictive relations ambiguous. We hypothesized that rumination would prospectively predict levels of depressive symptoms and would mediate the link between initial and subsequent levels of depression. However, we also hypothesized that not all forms of self-focus are equally deleterious to emotion regulation. A growing body of research among adults has revealed that there are several subtypes of rumination (e.g., Bagby & Parker, 2001; Cox, Enns, & Taylor, 2001; Roberts, Gilboa, & Gotlib, 1998; Watkins, 2004), that self-reflection can be distinguished from focusing on symptoms (Bagby & Parker, 2001; Cox et al., 2001; Roberts et al., 1998; Trapnell & Campbell, 1999), and that these different forms of rumination are differentially related to depressive symptoms and negative affectivity (Roberts et al., 1998; Treynor, Gonzalez, & Nolen-Hoeksema, 2003; Watkins, 2004). Specifically, Treynor and colleagues found that brooding or passively focusing on symptoms predicted depression longitudinally, whereas self-reflection or actively attempting to gain insight into problems did not. Complementing these findings, experimental studies have

This research was supported by an National Institute of Mental Health grant (MH60106) to Stephen Shirk and a National Research Service Award to Rebecca Burwell. Correspondence should be addressed to Rebecca A. Burwell, Westfield State College, Department of Psychology, 577 Western Avenue, Westfield, MA 01086–1630. E-mail: [email protected]

56

RUMINATION SUBTYPES IN ADOLESCENCE

shown that individuals who focus passively or self-critically on their mood show worse outcomes than those instructed to reflect on their feelings (Watkins, 2004) or those who self-focus but evaluate reasons for their feelings (Kross, Ayduk, & Mischel, 2005). Together, these efforts to deconstruct ruminative self-focus suggest that passively and self-critically dwelling on feelings, rather than actively examining one’s emotions, undermines emotion regulation. We hypothesized that adolescent rumination would include both adaptive and maladaptive forms of self-focus and that brooding would be associated with and predictive of depressive symptoms whereas self-reflection would not. Rumination has also been advanced as a mechanism to account for the emergence of gender differences in depressive symptoms during adolescence (Nolen-Hoeksema & Girgus, 1994). Evidence from adults indicates higher levels of rumination among women than men (e.g., Butler & Nolen-Hoeksema, 1994; Katz & Bertelson, 1993; Nolen-Hoeksema, Larson, & Grayson, 1999). Among children and adolescents there is equivocal evidence for the gender difference in rumination. A number of studies have found that girls ruminate more than do boys (Broderick, 1998; Grant & Compas, 1995; Hart & Thompson, 1996; Schwartz & Koenig, 1996; Ziegert & Kistner, 2002), but a subset of those studies failed to support the hypothesis that such differences would account for gender differences in depressive symptoms (e.g., Grant & Compas, 1995; Schwartz & Koenig, 1996). In addition, several studies have not found gender differences in rumination among children and adolescents (Abela et al., 2004; Broderick & Korteland, 2004). Because we distinguish between maladaptive and adaptive forms of self-focus and in light of previous mixed results, we offer no specific prediction concerning gender differences. However, regardless of gender, we predicted that those who ruminated would report higher depressive symptoms over time. In addition to examining subtypes of rumination in adolescence and their relations with depression, we were also interested in evaluating whether passive, brooding self-focus is linked with maladaptive coping and whether an active, reflective subtype of rumination might be related to adaptive coping. Rumination is hypothesized to maintain or exacerbate depressive symptoms in part through its deleterious effect on problem solving (Nolen-Hoeksema, 2004), and a number of studies have shown that rumination or brooding is associated with poor problem solving (Abela et al., 2004; Watkins & Moulds, 2005). An important goal of our study was to determine if rumination in adolescence entails different types of self-focus, including brooding and reflection, and if these different types of selffocus are equally detrimental to adaptive coping, including problem solving, in this age group. In this article, we first explore the factor structure of the Ruminative Response Scale among adolescents.

Second, we examine concurrent and longitudinal relations between subtypes of rumination and depressive symptoms. We then examine whether gender moderates mean levels of rumination and relations with depressive symptoms and evaluate a moderated mediation model. Third, we examine whether subtypes of rumination relate differentially to coping. Method Participants Adolescents were recruited from three metropolitan public schools in the Rocky Mountain region. Researchers announced in middle-school classrooms an opportunity to participate in a research study occurring at a local university. Of the approximately 750 who learned of the study, 198 returned initial parental consent forms indicating a willingness to be contacted further for study recruitment; 168 were successfully contacted, scheduled, and eventually participated in the initial laboratory session; and the remainder decided against participating due to transportation issues, despite the fact that compensation for bus or cab fare was offered. Of the 168 participants with complete data at Time 1, 159 returned for participation at Time 2 and 127 at Time 3. Those who completed all three time points were more likely to be European American, χ2(1, N = 166) = 4.67, p = .03; 20.5% of Whites and 38.2% of non-Whites dropped out at Time 3. Due to the low number of non-European Americans, it was not possible to control for ethnicity in this study. There were no other differences between those who were missing and those with complete data at Time 3. Mean age of the sample at Time 1 was 13.58 years (SD = .52, range = 12–15); 70 (41.7%) were boys and 98 (58.3%) were girls. Hollingshead’s (1976) Four Factor Index of Social Status showed a broad range of family backgrounds from machine operators and semiskilled workers to professionals, but most families were middle class (M = 4.23, SD = .71). The majority (79.1%) of participants’ parents were married or remarried, 16.4% were separated or divorced, and 2.7% were never married. Of the sample, 76.4% were of European American descent. The remainder of the sample was made up of approximately equal numbers of African American, Hispanic American, and biracial youth. Only 2.7% of the sample was Asian American. This distribution corresponds to the racial and ethnic makeup of the metropolitan Rocky Mountain area. Measures Children’s Depression Inventory (CDI; Kovacs, 1992). The CDI consists of 27 items that assess various symptoms of depression. Higher scores indicate 57

BURWELL AND SHIRK

greater depressive symptoms. Because Time 3 data collection was Internet- and mail-based, we did not assess suicidal ideation at that time. To draw comparisons across the three time points, depression scores at all time points were computed without this item. The CDI demonstrated good internal consistency (Time 1 α = .84) and convergent validity, correlating with scores on the Child Depression Rating Scale, a semistructured depression interview (r = .69, p < .001). Ruminative Response Scale (adapted from Nolen-Hoeksema & Morrow, 1991). We adapted the original Ruminative Response Scale, part of the larger Response Styles Questionnaire, for use with an adolescent sample and sought to address the issue of criterion contamination that has been raised previously (e.g., Roberts et al., 1998; Segerstrom, Tsao, Alden, & Craske, 2000; Treynor et al., 2003). First, we replaced the word depressed with upset, consistent with modifications made in previous studies of rumination. Second, items were anchored to specific, memorable stressors (Connor-Smith, Compas, Wadsworth, Thompson, & Saltzman, 2000). To do this, participants were instructed to endorse self-relevant social stressors that occurred since the transition to high school. Finally, participants endorsed the extent to which they ruminated about feelings related to these stressors. Scores were calculated by averaging items to obtain a mean rumination score. The modified Ruminative Response Scale similarly demonstrated good internal consistency (α = .88) and validity. It is associated with the Responses to Stress Questionnaire (Connor-Smith et al., 2000) Rumination and Intrusive Thoughts subscales (r = .39, p < .001 and r = .34, p < .001). Responses to Stress Questionnaire (Connor-Smith et al., 2000). The Responses to Stress Questionnaire is a 57-item questionnaire that assesses adolescents’ behaviors in response to 10 social (e.g., “You were teased or hassled by other kids”) and 10 academic (e.g., “You’ve had too much homework”) stressors. To evaluate adaptive and maladaptive coping, we examined the three coping factors: primary control engagement coping (e.g., problem solving, emotion expression, emotion regulation), secondary control engagement coping (i.e., positive thinking, cognitive restructuring, acceptance, distraction), and voluntary disengagement (i.e., avoidance, denial, and wishful thinking). Reliability and validity of the Responses to Stress Questionnaire have been supported in a number of samples (see Connor-Smith et al., 2000). In this sample, internal reliability for primary control engagement coping, secondary control engagement, and voluntary disengagement were α =.83, .73, and .73, respectively. 58

Children’s Depression Rating Scale–Revised (CDRS–R; Poznanski & Mokros, 1999). The CDRS–R is a semistructured depression interview that has demonstrated reliability and validity among children and adolescents (e.g., Endicott, Wagner, & Wohlberg, 2002), with internal consistency ranging from .74 to .90. Interviewers, trained on the National Institute of Mental Health clinical trial standard, demonstrated very good reliability as assessed by the intraclass correlation coefficient of r = .87. Because this study involved a community sample, total raw scores were used rather than T scores to offset attenuation in variability. The 17 symptom areas were summed, with higher scores indicating greater levels of depressive symptoms and a raw score of 40 indicating symptom severity in the clinical range. Mood and Conduct Checklist (American Psychiatric Association, 1994). Embedded within this 40-item questionnaire are the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; American Psychiatric Association, 1994) criteria for depression (e.g., “My child has trouble sleeping or sleeps a lot”), with items rated on a 3-point scale of 0 (not true), 1 (somewhat or sometimes true), or 2 (very true or often true). Scores were computed by averaging severity score (0 to 2) of 11 items within the narrowband symptom domain of depression. Internal consistency was α = .74. Procedure Adolescents and their mothers came to the laboratory for the Time 1 and 2 assessments and completed measures by mail or e-mail at Time 3. University Institutional Review Board approval was obtained for all procedures; parents or guardians provided informed consent and adolescents provided assent prior to participation. Participants were assessed in the spring of their eighth-grade year (Time 1), the fall of their ninth-grade year (Time 2), and again in the spring of their ninth-grade year (Time 3). The interval between assessments was 6 to 8 months. Mothers completed the Mood and Conduct Checklist at all three time points. Adolescents completed the Ruminative Response Scale and the Responses to Stress Questionnaire at Time 2 and the CDI at all three time points and were interviewed on the CDRS–R at Times 1 and 2. Order of presentation of the foregoing measures was counterbalanced across participants. Data Imputation To improve power and provide a more adequate representation of the initial sample (Allison, 2002), data imputation procedures were utilized. Data were miss-

RUMINATION SUBTYPES IN ADOLESCENCE

ing at random; there were no differences between those missing and those with complete data on the Mood and Conduct Checklist, CDI, or CDRS–R at Time 1 or Time 2. Multiple data imputation was run using PRELIS (in LISREL 8.52; Jöreskog & Sörbom, 2002), resulting in 156 valid Time 3 CDI observations. Mothers’ reports of their child’s depressive symptoms at Time 3 were similarly imputed from variables outside the model, resulting in 150 valid observations. Results Preliminary analyses examined relations between demographic variables (gender, socioeconomic status, ethnicity) and model variables. There were no differences based on socioeconomic level or ethnic category (European American vs. non-European American) on model variables. There were no gender differences on CDI at Times 1 or 2; on mother’s report of adolescents’ depressive symptoms at Times 1, 2, or 3; on the CDRS–R at Times 1 or 2; or on the rumination composite at Time 2. However, there was a trend toward girls reporting higher CDI depressive symptoms at Time 3 compared to boys, t(154) = –1.77, p < .08; MGirls = 6.37, SD = 7.3; MBoys = 4.63, SD = 5.0. Factor Structure of the Ruminative Response Scale Among Adolescents To address our first goal, we conducted exploratory factor analysis of the adapted Ruminative Response Scale. We first used principal axis factoring to examine

variance unique to the factors; next, for the purpose of drawing comparisons with Treynor and colleagues’ (2003) findings, we conducted principal components analysis, which accounts for both unique and common variance. In both sets of analyses, we used oblique rotation to allow for correlated factors; items loaded at ≥ .40 and did not cross-load on other factors. Our primary method—principal axis factoring—resulted in two factors with eigenvalues greater than 1, accounting for a total of 36.8% of the variance. A principal components analysis yielded similar item loadings and five factors with eigenvalues greater than 1, accounting for 57% of the variance in the measure. These results are similar to Treynor and colleagues’ principal components analysis of 10 items, which resulted in two factors accounting for 50.5% of the variance. Results of the principal axis factoring and examination of the scree plot produced two factors: Brooding and Reflection, comprised of items reflecting brooding about one’s flaws and symptoms versus attempts at insight, respectively (see Table 1 for items and scale loadings). The Brooding factor was made up of six items, two of which mapped onto Treynor and colleagues’ (2003) five-item Brooding subscale. Three of the four items on the Reflection factor overlapped with Treynor and colleagues’ five-item Reflection subscale (the final item in our scale was dropped from later versions of the Ruminative Responses Questionnaire used in Treynor et al., 2003). In this study, factors were computed using mean scores across subscale items. Reliability of the scales were α = .80 and α = .74, respectively.

Table 1. Ruminative Response Scale Factor Loadings

1. Think about how alone you feel 12. Think about a situation and wish it had gone better 13. Think, “why do I have problems others don’t?” 14. Think about how upset you feel 15. Think about all your shortcomings, failings, faults, mistakes 19. Think about how angry you are with yourself 6. Analyze recent events to try to understand why you are upset 11. Write down what your are thinking about and analyze it 17. Analyze your personality to try to understand why you are upset 22. Try to understand yourself by focusing on your feelings 10. Go away by yourself and think about why feel you like this 18. Go away alone and think about your feelings 21. Isolate self and think about the reasons you’re upset 2. Think I can’t do my work because I feel so badly 3. Think about how tired and achy you feel 4. Think about how hard it is to concentrate 5. Think about how unmotivated you feel 7. Think about how you don’t feel up to doing anything anymore 8. Think, “Why can’t I get going?” 9. Think, “Why do I react like this?” 16. Think about how you don’t feel up to doing anything 20. Listen to sad music

Brooding

Reflection

.64 .50 .57 .62 .68 .42 .11 –.03 –.12 –.02 .02 –.20 .13 .13 .36 .18 .10 .10 .19 .30 .02 .26

–.02 .11 –.03 .19 –.01 .04 .66 .43 .73 .70 .16 –.03 .11 .25 –.04 .06 –.06 .14 .14 .10 –.03 –.03

59

BURWELL AND SHIRK

Relations Between Subtypes of Rumination and Depressive Symptoms We next examined relations between rumination and depressive symptoms over time. Self-reported depressive symptoms were related to mothers’ reports and clinician-rated scores (r =.32, p < .001 and r = .69, p < .001, respectively). Brooding was associated with concurrent clinician-rated CDRS–R depression (r = .51, p < .001), self-reported CDI depressive symptoms (r = .69, p < .001), and mother report of adolescent depressive symptoms (r = .26, p < .01). Brooding was also related longitudinally to self-reported CDI (r = .53, p < .001) and to Time 3 mother-reported symptoms (r = .23, p < .01). Brooding predicted a change in CDI depressive symptoms after controlling for Time 1 CDI scores, β = .26, R2 change = .05, p < .05, F(2, 155) = 95.43, p < .001, R2 = .56, but did not predict a change in the Mood and Conduct Checklist from Times 1 to 3, β = .11, R2 change = .01, p > .10, F(2, 146) = 31.90, p < .001, R2 = .30. Reflection was modestly associated with concurrent CDI scores (r = .17, p < .05) but was not significantly related to concurrent CDRS–R depression scores (r = .15, p < .10) or to mother-reported symptoms (r = .11, p > .10). Reflection was not significantly related to either self- or mother-reported symptoms longitudinally. Reflection was unrelated to change in CDI from Time 1 to Time 3, β = .05, R2 change = .00, p > .05, F(2, 155) = 78.28, p < .001, R2 = .51, or to concurrent CDI when controlling for brooding, β = .05, p > .10. When entered simultaneously into a hierarchical linear regression, brooding (β = .26, p < .001) but not reflection (β = .01, p > .05) emerged as a significant predictor of Time 3 CDI, controlling for Time 1 CDI, F(3, 155) = 63.23, p < .001, R2 = .56.

Rumination as a Mediator of Depressive Symptoms Our second goal further entailed examining the processes by which rumination relates to the development of depressive symptoms over time. To do this, we used hierarchical linear regression to evaluate whether brooding or reflection mediated the development of depressive symptoms. To assess mediation in regression, we followed guidelines put forth by Baron and Kenny (1986). In addition to significant associations between the independent variable and both the dependent variable and mediating variable (criteria 1 and 2), the mediating variable must be related to the dependent variable, controlling for the independent variable, and its inclusion into hierarchical linear regression must significantly reduce the independent variable to dependent variable relation. Zero-order correlations between Time 1 CDI and Time 3 CDI (r = .71, p < .001) and between Time 1 CDI 60

and brooding (r = .46, p < .001) support the first two criteria, respectively. When entered into the hierarchical regression, brooding significantly reduced the relation between Time 1 CDI and Time 3 CDI from r = .71 (p < .001) to r = .59 (p < .001; Sobel’s Test of the Indirect Effect: Z = 3.59, p < .001), indicating partial mediation, β = .26, R2 change = .05, p < .001, F(2, 155) = 95.43, p < .001, R2 = .59. Next, we examined whether reflection mediated the development of depressive symptoms. Neither the association between reflection at Time 2 and CDI at Time 3 (r = .16, p < .10) nor the link between Time 1 CDI and reflection (r = .14, p < .10) were significant; consequently the criteria for a mediation test were not met. Moderation Analyses We first assessed mean level gender differences in subtypes of rumination. Girls scored higher on reflection than boys, t(156) = –2.59, p < .05 (MGirls = 1.92, SD = 0.66; MBoys = 1.66, SD = 0.55), but did not differ on brooding. Reflection and Brooding factors were correlated among girls (r = .30, p < .01) but not significantly among boys (r = .11, p > .10). However, the association between brooding and reflection was not significantly different for girls and boys (Fisher’s z = 0.62, p > .10). We next examined whether gender moderated the relation between rumination and the development of depressive symptoms. Time 1 CDI was entered in the first step, gender and the Rumination subscale (i.e., brooding or reflection) in the second step, and the centered interaction term between gender and the Rumination subscale in the third step (see Table 2). There were main effects both for gender (β = .12, p < .05) and brooding (β = .25, p < .001) and an interaction between brooding and gender, β = .14, R2 change = .02, p < .01, F(4, 155) = 54.04, p < .001, R2 = .59. To display the interaction, we created high and low brooding groups based on scores falling one standard deviation above and below the brooding mean (see Figure 1). Next, we examined possible interactions between gender and reflection; there was a main effect for gender but not for reflection, and there was no interaction between gender and reflection, β = .006, R2 change = .000, p > .10, F(4, 155) = 40.82, p < .001, R2 = .52 (see Table 2). Moderated Partial Mediation We next examined gender as a moderator of the partially mediated model. Among girls, inclusion of brooding into Step 3 of the regression partially attenuated the relation between Time 1 CDI and Time 3 CDI from r = .75 (p < .001) to r = .62 (p < .001; Sobel’s Z = 3.13, p < .001). Among boys, brooding was not significantly related to Time 3 CDI when controlling for Time 1 CDI despite a trend in this direction (r = .21, p = .09), thereby

Table 2. Hierarchical Linear Regression Predicting Time 3 CDI From Time 1 CDI, Gender, Rumination, and Gender by Rumination Interaction Variables Brooding Step 1 Time 1 CDI Step 2 Time 1 CDI Gender Brooding Step 3 Time 1 CDI Gender Brooding Gender × Brooding Reflection Step 1 Time 1 CDI Step 2 Time 1 CDI Gender Reflection Step 3 Time 1 CDI Gender Reflection Gender × Reflection

B

SE B

β

R2 Change

p

1.02

0.08

.710

.500

.000

0.85 1.53 3.09

0.09 0.71 0.73

.590 .120 .250

.070

.000

0.85 1.56 3.12 3.54

0.09 0.69 0.72 1.30

.590 .120 .250 .140

.020

.007

1.02

0.08

.710

.500

.000

1.02 1.58 0.31

0.08 0.76 0.60

.700 .120 .030

.020

.075

1.02 1.60 0.29 0.14

0.08 0.77 0.61 1.28

.70 .120 .030 .006

.000

.913

Note: CDI = Children’s Depression Inventory.

Figure 1.

Gender × Brooding interaction predicting residual Children’s Depression Inventory scores.

61

BURWELL AND SHIRK

violating conditions for mediation. Moreover, its inclusion in the model did not significantly reduce the relation between Time 1 CDI and Time 3 CDI (Sobel’s Z = 1.64, p > .10). The association between Time 1 CDI and brooding was higher for boys than for girls (Fisher’s Z = –2.64, p < .01). Using path analysis, multigroup modeling produced parallel results with the regressions. Specifically, an unconstrained model in which paths were free to vary between girls and boys fit significantly better than a model in which the paths were constrained to be equal (see path model, Figure 2). Although not significant, there was a trend for a stronger correlation between Time 1 and Time 3 CDI for girls than for boys (Fisher’s Z = 1.74, p = .08). Because reflection did not meet criteria for mediation among either girls or boys, it was dropped from subsequent analyses.

Links With Coping The final goal of this study was to examine whether subtypes of rumination are differentially related to coping. There were no significant relations between socioeconomic status or ethnicity and model variables. Girls (MGirl = 2.95, SD = .51) scored higher than boys (MBoys = 2.64, SD = .45) only on primary control engagement coping, t(116) = –5.07, p < .01. Correlations showed that brooding was concurrently positively correlated with voluntary disengagement coping (r = .62, p < .001) and unrelated to primary control (r = –.02) and secondary control coping (r = –.01). Conversely, reflection was positively associated with primary control coping (r = .28, p < .001) and secondary control coping (r = .24, p < .01) and unrelated to voluntary disengagement coping (r = .08). Relations between brooding and coping were virtually identical for boys and girls. Associations between reflection and voluntary disengagement were not significant for both boys and girls. Associations between reflection and primary and secondary control coping were significant for boys (rs = .35 and .33, ps < .01, respectively) but not for girls (rs = .15 and .16, ps < .10, respectively). However, the gender difference between correlations was not significant.

Figure 2.

62

Discussion Previous research with adults has shown that rumination contributes to the maintenance or intensification of depressive symptoms and that women are more likely to ruminate than men (Butler & Nolen-Hoeksema, 1994; Katz & Bertelson, 1993; Nolen-Hoeksema et al., 1999; Treynor et al., 2003). Nolen-Hoeksema proposed that gender differences in rumination contribute to the emergence of gender differences in depression during adolescence; that is, girls enter adolescence with a greater tendency to respond to stress with rumination than are boys, and this response style contributes to the emergent gender difference in depressive symptoms. However, recent studies with adults have shown that rumination may not be a unitary construct and that what has been labeled rumination appears to involve two types of self-focus, reflection and brooding. The results of our study with adolescents are consistent with much of the prior research but also point to some important differences. First, like prior studies we found that rumination is not a unitary construct, as originally proposed, but is in fact multidimensional, consisting of both deleterious, passive brooding and analytical, active self-reflection. These findings are similar to recent studies examining the factor structure of rumination among adults (Bagby & Parker, 2001; Cox et al., 2001; Treynor et al., 2003). In brief, what has been labeled rumination appears to involve both adaptive and maladaptive self-focus, and this distinction is relevant for adolescents as well as adults. Brooding and reflection showed different associations with other coping strategies. Specifically, brooding was associated with voluntary disengagement strategies (e.g., avoidance, denial, fleeing). Although investigators have proposed that rumination is deleterious in part because it interferes with effective problem solving (Lybomirsky & Nolen-Hoeksema, 1995), in our sample brooding was unrelated to adaptive primary and secondary control coping. Instead, brooding was linked with strategies such as denial and avoidance,

Gender differences in the direct effects model between brooding and the development of depressive symptoms.

RUMINATION SUBTYPES IN ADOLESCENCE

suggesting that brooding might reflect a failure to disengage from stress or negative emotions. In contrast, reflection was positively associated with voluntary coping aimed at changing the stressor or one’s attitudes toward the stressor (e.g., problem solving and cognitive restructuring). These findings suggest that reflective awareness of one’s emotional experience is linked with adaptive coping strategies (Martin & Tesser, 1996) and indeed may be a marker of emotional intelligence (Mayer & Salovey, 1997). It is possible that reflection contributes to goal clarification in the problem-solving process. Somewhat surprisingly, among girls reflection was related to brooding. Previous research among adults has similarly found that reflection and brooding are related (Segerstrom et al., 2000; Trapnell & Campbell, 1999; Treynor et al., 2003), an overlap that may be due to both common method variance and to a general tendency toward self-focus. It would be important to examine whether reflection and brooding are related to coping and responses to stress over time. Second, the pattern of findings in this study provides important empirical support for the deleterious process of brooding in adolescence and is consistent with results from adult samples (Roberts et al., 1998; Treynor et al., 2003). Specifically, brooding was associated with interviewer-rated depressive symptoms concurrently and with self-reported and mother-reported depressive symptoms both concurrently and longitudinally. In addition, brooding was associated with the increase in self-reported but not mother-reported depressive symptoms over time. In contrast, reflection was only modestly related to concurrent selfreported and interviewer-rated depressive symptoms and was unrelated to concurrent and longitudinal maternal reports of depressive symptoms and to change in self-reports of depressive symptoms over time. Moreover, once brooding was accounted for, reflection was unrelated to concurrent self-reported depressive symptoms. These results suggest that focus on negative emotions per se does not place adolescents at risk for depressive symptoms over time. Rather, it is when adolescents focus passively on their symptoms and negative consequences of their depression without efforts to address their symptoms that they are at risk for depressive symptoms over time (Treynor et al., 2003). It has been proposed that brooding, like worry, is a means of emotional avoidance (Borkovec, 1994), and associations between brooding and disengagement strategies in this sample are consistent with this view. On the other hand, when a negative emotion is reflected on, it is a means of processing and experiencing, rather than warding off, emotion and can result in both greater self-awareness and emotional clarity (Mayer & Salovey, 1997). Given the positive association between reflection and active problem solving, it

is possible that reflection contributes to goal clarification in the problem-solving process. Although many of our results were consistent with prior findings, some were not. As might be expected, girls reported more reflection than boys (Roberts et al., 1998; Treynor et al., 2003) but surprisingly comparable levels of brooding. It is possible that gender differences in rumination observed in adults are not as pronounced when evaluating broader negative emotions and when brooding is distinguished from reflection. However, the absence of gender differences is consistent with several studies among children and young adolescents (Abela et al., 2004; Broderick & Korteland, 2004) and suggests that gender differences in brooding may emerge gradually over the course of adolescence and into young adulthood. Future research examining the developmental course of rumination would help explicate when the gender difference in rumination emerges. Despite a lack of gender differences in mean levels of brooding, we found support for gender differences in the development of depressive symptoms over time. For girls, brooding partially mediated the increase in depressive symptoms. Although there appeared to be a trend in this direction for boys, it was not of the same magnitude and was not statistically significant, though the lower number of boys in the sample reduced statistical power. In addition, relations between brooding and depressive symptoms differed significantly for boys and girls. Given that there were no mean level gender differences in brooding or in depressive symptoms, these findings suggest that the process of brooding may relate to depressive symptoms differentially for girls versus boys. Research by Rudolph and Hammen (1999) suggested that girls and boys may be differentially reactive to types of stressors. Our findings suggest that similar cognitive processes may have different effects across genders. These findings should be replicated in larger samples to evaluate whether low power influenced the lack of significant finding among boys. Future research might also examine whether brooding relates to different outcomes for boys (e.g., externalizing) than for girls (e.g., depression). Some findings may not generalize to other samples. Adolescent participants were from a low-risk, community sample with low mean levels of depressive symptoms and moderately high stability in depressive symptoms over three time points. The relatively limited variability in depression scores over time may have attenuated findings. Despite this, rumination, and in particular brooding, still predicted variance in depressive symptoms at Time 3, over and above Time 1 depressive symptoms. Significant gender differences in depressive symptoms did not appear throughout the course of the study. However, we did see a trend toward a gender difference in depressive symptoms at Time 3, which is consistent with research indicating that the well-estab63

BURWELL AND SHIRK

lished gender difference in depression emerges around age 15 (Hankin et al., 1998; Petersen et al., 1993). Although minority youth were represented, the study sample was relatively homogeneous, especially with regard to economic status. These findings entail interesting implications for addressing negative emotion in adolescent depression. Although prior studies show that disengagement from negative emotion, for example through distraction, offsets the deepening of dysphoria, such findings should not lead to the conclusion that negative emotions must be avoided or restricted in the treatment of adolescent depression. Instead, our results indicate that it is the manner by which negative emotions are addressed that is critical. Passively experiencing and brooding about negative emotional states may undermine active attempts to cope with or resolve problematic emotional situations. However, active reflection on the causes and consequences of negative emotions appears to be associated with active problem solving and may be the foundation on which effective problem solutions are based. In this connection, active emotional reflection appears to share some properties with other emerging interventions for emotion regulation. Research by Pennebaker and Francis (1996) and Pennebaker and Seagal (1999) on the impact of writing about stressful events has shown that improved health and mental health outcomes are associated with increased narrative organization and use of insight words. These results suggest that actively reflecting on and processing negative emotional experiences results in salubrious effects on both physical and psychological symptoms. In summary, results of this study indicate that rumination in the form of passive brooding but not active reflection about negative emotion predicts the development of depressive symptoms in adolescence. Insofar as active self-reflection is related to other adaptive coping strategies, interventions should target both the reduction of brooding and the development of active self-reflection.

References Abela, J. R., Vanderbilt, E., & Rochon, A. (2004). A test of the integration of the response styles and social support theories of depression in third and seventh grade children. Journal of Social and Clinical Psychology, 23, 653–674. Allison, P. D. (2002). Missing data (Quantitative Applications in the Social Sciences Series). Thousand Oaks, CA: Sage. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Bagby, R. M., & Parker, J. D. A. (2001). Relation of rumination and distraction with neuroticism and extraversion in a sample of patients with major depression. Cognitive Therapy and Research, 25, 91–102. Baron, R. M., & Kenny, D. A. (1986). The mediator–moderator variable distinction in social psychological research: Conceptual,

64

strategic, and statistical considerations. Journal of Personality & Social Psychology, 51, 1173–1182. Borkovec, T. D. (1994). The nature, functions, and origins of worry. In G. C. L. Davey & F. Tallis (Eds.), Worrying: Perspectives on theory, assessment and treatment (pp. 5–33). Chichester, England: Wiley. Broderick, P. C. (1998). Early adolescent gender differences in the use of ruminative and distracting coping strategies. Journal of Early Adolescence, 18, 173–191. Broderick, P. C., & Korteland, C. (2004). A prospective study of rumination and depression in early adolescence. Clinical Child Psychology and Psychiatry, 9, 383–394. Butler, L. D., & Nolen-Hoeksema, S. (1994). Gender differences in responses to depressed mood in a college sample. Sex Roles, 30, 331–346. Connor-Smith, J. K., Compas, B. E., Wadsworth, M. E., Thomsen, A. H., & Saltzman, H. (2000). Responses to stress in adolescence: Measurement of coping and involuntary stress responses. Journal of Consulting & Clinical Psychology, 68, 976–992. Cox, B. J., Enns, M. W., & Taylor, S. (2001). The effect of rumination as a mediator of anxiety sensitivity in major depression. Cognitive Therapy and Research, 25, 525–534. Endicott, J., Wagner, K. D., & Wohlberg, C. (2002, June). Reliability and validity of the Children’s Depression Rating Scale–Revised (CDRS–R) in children and adolescents treated with sertraline. Poster presented at the meeting of the National Institute of Mental Health New Clinical Drug Evaluation Unit meeting, Boca Raton, FL. Ge, X., Lorenz, F. O., Conger, R. D., Elder, G. H., & Simons, R. L.(1994). Trajectories of stressful life events and depressive symptoms during adolescence. Developmental Psychology, 30, 467–483. Grant, K. E., & Compas, B. E.(1995). Stress and anxious–depressed symptoms among adolescents: Searching for mechanisms of risk. Journal of Consulting & Clinical Psychology, 63, 1015–1121. Hankin, B. L., & Abramson, L. Y. (2002). Measuring cognitive vulnerability to depression in adolescence: Reliability, validity, and gender differences. Journal of Clinical Child and Adolescent Psychology, 31, 491–504. Hankin, B. L., Abramson, L. Y., Moffitt, T. E., Silva, P. A., McGee, R., & Angell, K. E. (1998). Development of depression from preadolescence to young adulthood: Emerging sex differences in a 10-year longitudinal study. Journal of Abnormal Psychology, 107, 128–140. Hart, B. I., & Thompson, J. M. (1996). Gender role characteristics and depressive symptomatology among adolescents. Journal of Early Adolescence, 16, 407–426. Hollingshead, A. B.(1976). Four Factor Index of Social Status. New Haven, CT: Yale University Press. Jöreskog, K., & Sörbom, D. (2002). LISREL 8.52 program: Student edition. Chicago: Scientific Software International. Just, N., & Alloy, L. B. (1997). The response styles theory of depression: Tests and an extension of the theory. Journal of Abnormal Psychology, 106, 221–229. Katz, E. J., & Bertelson, A. D. (1993). The effects of gender and response style on depressed mood. Sex Roles, 29, 509–514. Kovacs, M. (1992). Manual for the Childhood Depression Inventory. North Tonawanda, NY: MultiHealth Systems. Kross, E., Ayduk, O., & Mischel, W. (2005). When asking “why” does not hurt. Psychological Science, 16, 709–715. Kuehner, C., & Weber, I. (1999). Responses to depression in unipolar depressed patients: An investigation of Nolen-Hoeksema’s response styles theory. Psychological Medicine, 29, 1323–1333. Lyubomirsky, S., & Nolen-Hoeksema, S. (1995). Effects of selffocused rumination on negative thinking and interpersonal problem solving. Journal of Personality & Social Psychology, 69(1), 176–190.

RUMINATION SUBTYPES IN ADOLESCENCE Martin, L. L., & Tesser, A. (1996). Some ruminative thoughts. In R. S. Wyer Jr. (Ed.), Advances in social cognition (pp. 1–47). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Mayer, J. D., & Salovey, P. (1997). What is emotional intelligence? In P. Salovey & D. J. Sleuter (Eds.), Emotional development and emotional intelligence: Educational implications (pp. 3–34). New York: Basic Books. Nolen-Hoeksema, S. (1991). Responses to depression and their effects on the duration of depressive episodes. Journal of Abnormal Psychology, 100, 569–582. Nolen-Hoeksema, S. (2004). The response styles theory. In C. Papageorgiou & A. Wells (Eds.), Depressive rumination: Nature, theory and treatment (pp. 107–123). West Sussex, England: Wiley. Nolen-Hoeksema, S., & Girgus, J. S. (1994). The emergence of gender differences in depression during adolescence. Psychological Bulletin, 115, 424–443. Nolen-Hoeksema, S., Larson, J., & Grayson, C. (1999). Explaining gender differences in depressive symptoms. Journal of Personality & Social Psychology, 77, 1061–1072. Nolen-Hoeksema. S., & Morrow, J. (1991). A prospective study of depression and posttraumatic stress symptoms after a natural disaster: The 1989 Loma Prieta earthquake. Journal of Personality & Social Psychology, 61, 115–121. Pennebaker, J. W., & Francis, M. E. (1996). Cognitive, emotional, and language processes in disclosure. Cognition and Emotion, 10, 601–626. Pennebaker, J. W., & Seagal, J. D. (1999). Forming a story: The health benefits of narrative. Journal of Clinical Psychology, 55, 1243–1254. Petersen, A. C., Compas, B. E., Brooks-Gunn, J., Stemmler, M., Ey, S., & Grant, K. E. (1993). Depression in adolescence. American Psychologist, 48, 155–168. Poznanski, E. O., & Mokros, H. B. (1999). Children’s Depression Rating Scale–Revised (CDRS–R). Los Angeles: Western Psychological Services.

Roberts, J. E., Gilboa, E., & Gotlib, I. H. (1998). Ruminative response styles and vulnerability to episodes of dysphoria: Gender, neuroticism, and episode duration. Cognitive Therapy and Research, 22, 401–423. Rudolph, K., & Hammen, C. (1999). Age and gender as determinants of stress exposure, generation, and reactions in youngsters: A transactional perspective. Child Development, 70, 660–677. Schwartz, J. A. J., & Koenig, L. J. (1996). Response styles and negative affect among adolescents. Cognitive Therapy and Research, 20, 13–36. Segerstrom, S. C., Tsao, J. C. I., Alden, L. E., & Craske, M. G. (2000). Worry and rumination: Repetitive thought as a concomitant and predictor of negative mood. Cognitive Therapy and Research, 24, 671–688. Trapnell, P. D., & Campbell, J. D. (1999). Private self-consciousness and the five-factor model of personality: Distinguishing rumination from reflection. Journal of Personality & Social Psychology, 76, 284–304. Treynor, W., Gonzalez, R., & Nolen-Hoeksema, S. (2003). Rumination reconsidered: A psychometric analysis. Cognitive Therapy and Research, 27, 247–259. Watkins, E. (2004). Adaptive and maladaptive ruminative self-focus during emotional processing. Behavioral Research and Therapy, 42, 1037–1052. Watkins, E., & Moulds, M. (2005). Distinct modes of ruminative self-focus: Impact of abstract versus concrete rumination on problem solving in depression. Emotion, 5, 319–328. Ziegert, D. I., & Kistner, J. A. (2002). Response styles theory: Downward extension to children. Journal of Clinical Child and Adolescent Psychology, 31, 325–334.

Received January 11, 2006 Accepted August 11, 2006

65