Psychology

Journal of Abnormal Psychology

A Longitudinal Examination of Stress Generation in Depressive and Anxiety Disorders Amanda A. Uliaszek, Richard E. Zinbarg, Susan Mineka, Michelle G. Craske, James W. Griffith, Jonathan M. Sutton, Alyssa Epstein, and Constance Hammen Online First Publication, October 17, 2011. doi: 10.1037/a0025835

CITATION Uliaszek, A. A., Zinbarg, R. E., Mineka, S., Craske, M. G., Griffith, J. W., Sutton, J. M., Epstein, A., & Hammen, C. (2011, October 17). A Longitudinal Examination of Stress Generation in Depressive and Anxiety Disorders. Journal of Abnormal Psychology. Advance online publication. doi: 10.1037/a0025835

A Longitudinal Examination of Stress Generation in Depressive and Anxiety Disorders

Amanda A. Uliaszek Northwestern University

Richard E. Zinbarg Northwestern University and the Family Institute at

Northwestern University

Susan Mineka Northwestern University

Michelle G. Craske University of California, Los Angeles

James W. Griffith and Jonathan M. Sutton Northwestern University

Alyssa Epstein and Constance Hammen University of California, Los Angeles

The current study compared two competing theories of the stress generation model of depression (stress causation vs. stress continuation) using interview-based measures of episodic life stress, as well as interpersonal and noninterpersonal chronic life stress. We also expanded on past research by examining anxiety disorders as well as depressive disorders. In addition, we examined the role of neuroticism and extraversion in these relationships. Participants were 627 adolescents enrolled in a two-site, longitudinal study of risk factors for depressive and anxiety disorders. Baseline and follow-up assessments were approximately one year apart. Results supported the stress causation theory for episodic stress generation for anxiety disorders, with neuroticism partially accounting for this relationship. The stress causation theory was also supported for depression, but only for more moderate to severe stressors; neuroticism partially accounted for this relationship as well. Finally, we found evidence for interpersonal and noninterpersonal chronic life stress continuation in both depressive and anxiety disorders. The present findings have implications regarding the specificity of the stress generation model to depressive disorders, as well as variables involved in the stress generation process.

Keywords: stress generation, depression, anxiety, personality

Research has consistently shown that major stressful life events often precede the onset of an initial depressive episode (e.g., Hammen, 2005; Monroe, Slavich, & Georgiades, 2009). The stress generation model of depression (Hammen, 1991), which states that depression also predicts future stress, has less often been the focus of study. The body of research on stress generation has grown in recent years, with results suggesting that depressed individuals tend to generate primarily dependent interpersonal stress (for a recent review, see Liu & Alloy, 2010). Hypothesized mechanisms of stress generation also have been examined, with a focus on specific cognitive and interpersonal styles, as well as family vari- ables (e.g., Liu & Alloy, 2010). However, there are still many unanswered questions concerning these relationships. For exam-

ple, recent reviews suggest a closer examination of chronic life stress because most research has focused solely on episodic life stress (e.g., Liu & Alloy, 2010). In addition, little is known about whether stress generation is found in other closely related condi- tions (such as anxiety disorders) or whether stress generation is specific to depressive disorders. Moreover, we present two theo- retically distinct interpretations of the stress generation model: the stress continuation theory versus the stress causation theory. This is the first study to explicitly articulate and compare two interpre- tations of stress generation. Finally, additional third variables in the stress generation relationships, such as personality traits, may help inform our understanding of stress generation. The present

Amanda A. Uliaszek, Susan Mineka, James W. Griffith, and Jonathan M. Sutton, Department of Psychology, Northwestern University; Richard E. Zinbarg, Department of Psychology, Northwestern University, Patricia M. Nielson Research chair and Director of Anxiety and Panic Treatment Program, The Family Institute at Northwestern University; Michelle G. Craske, Alyssa Epstein, and Constance Hammen, Department of Psychol- ogy, University of California, Los Angeles.

James W. Griffith is now at the Department of Medical Social Sciences, Northwestern University; Jonathan M. Sutton is now at Edward Hines, Jr. VA

Hospital, Hines, IL; and Amanda A. Uliaszek is now at the University of Toronto, Scarborough.

This research was supported by National Institute of Mental Health Grants R01 MH65651 to Richard Zinbarg and Susan Mineka (Northwestern Univer- sity) and R01 MH65652 to Michelle Craske (University of California, Los Angeles). Richard Zinbarg was also supported by the Patricia M Nielsen Research Chair of the Family Institute at Northwestern University.

Correspondence concerning this article should be addressed to Amanda A. Uliaszek, Department of Psychology, University of Toronto, Scarborough, 1265 Military Trail, Toronto, Ontario, M1C 1A4. E-mail: [email protected]

Journal of Abnormal Psychology © 2011 American Psychological Association 2011, Vol. ●●, No. ●, 000 – 000 0021-843X/11/$12.00 DOI: 10.1037/a0025835

1

study addresses each of these issues in a large sample of adoles- cents.

Dimensions of Life Stress

The majority of stress research has focused on episodic life stress, which refers to events that take place within discrete, limited time periods. In contrast, chronic life stress is defined as ongoing life difficulties (e.g., Wheaton, 1994). Chronic life stress has been shown to be cross-sectionally related to depressive symp- toms in adults (e.g., McGonagle & Kessler, 1990), depressive disorders in children and adolescents (e.g., Rudolph et al., 2000; Uliaszek et al., 2010), and anxiety disorders in adolescents (e.g., Uliaszek et al., 2010). A few prospective studies have also found that chronic life stress predicted the severity of depression and the first onset of depression (e.g., Daley, Hammen, & Rao, 2000; Hammen, Davila, Brown, Ellicott, & Gitlin, 1992).

Life stress can be characterized by two additional dimensions: independent/dependent and interpersonal/noninterpersonal. Inde- pendent stress refers to events that are beyond one’s control; examples include natural disasters, being in a plane crash, or the death of a loved one. Dependent stress occurs, at least in part, as a result of the individual’s own actions (Daley et al., 1997). Examples include financial, marital, or academic difficulties (al- though these examples could be independent stressors under cer- tain circumstances). Interpersonal stress is defined as difficulties with family, peers, or significant others, whereas noninterpersonal stress refers to occupational, educational, and health problems (Hammen, 1991).

Stress Generation

Research has shown that those with a history of depression tend to experience higher levels of interpersonal life stress, even when euthymic, than do those without a history of depression (e.g., Chun, Cronkite, & Moos, 2004; Harkness & Stewart, 2009; Ru- dolph, 2008; Rudolph, Flynn, Abaied, Groot, & Thompson, 2009). Such effects of baseline depression on later stress have been found in clinical, collegiate, and community samples (Chun et al., 2004; Daley et al., 1997; Hammen, 1991; Shih, 2006), child and adoles- cent samples (Cole, Girgus, Paul, & Nolen-Hoeksema, 2004; Ru- dolph, 2008; Rudolph et al., 2000, 2009), and a late middle-aged sample (Holahan, Moos, Holahan, Brennan, & Schutte, 2005).

Rudolph and colleagues (2009) found that stress generated by depression partially accounted for the continuity of depression over time in female youth. These results are consistent with stress generation being a potential mechanism explaining the high rate of relapse in major depression (e.g., Daley et al., 1997). We empha- size that stress generation does not imply that the mean level of stress continues to increase over time. Instead, stress generation implies that a heightened level of stress continues to be maintained for a time period after a depressive episode relative to what would be expected with nondepressed individuals with high levels of stress over the same time period.

Two Contrasting Interpretations of Stress Generation

There is an abundance of evidence supporting stress generation in depression when this phenomenon is construed broadly. How-

ever, there are at least two plausible theoretical interpretations of stress generation that differ in terms of whether intraindividual characteristics (e.g., depression or neuroticism) exert a causal influence on subsequent life stress and there is a paucity of data that would allow us to choose among the two interpretations. The first, which we refer to as the stress causation theory is similar to the way stress generation is typically described (e.g., Hammen, 1991). In stress causation, the characteristics of the depressed person (symptoms of depression or other predictors of depression) are thought to play a causal role in generating stress over time. In contrast, the second interpretation, which is referred to as stress continuation theory, posits that the prospective relationship be- tween depression and stress is actually accounted for by the continuity of stress over time. According to this view, there is no causal influence of depression on subsequent life stress. Instead, there is a cross-sectional relationship between depression and life stress at baseline (e.g., Rudolph et al., 2000; Uliaszek et al., 2010) and that stress demonstrates a large degree of temporal stability. Thus, the partial correlation between depression and subsequent life stress should not be significant when accounting for baseline stress.

A comparison of these two interpretations of stress generation is important from both a theoretical and an applied standpoint. If the stress continuation theory is supported, then we might see evidence of stress continuation in other disorders associated with life stress, as well as in people without a diagnosed disorder who are expe- riencing heightened life stress. However, if the stress causation theory is supported, future studies should focus on depression- specific characteristics as variables of interest in understanding stress generation. These characteristics may be specific to the symptoms of depression (e.g., anhedonia, irritable mood) or to specific correlates of depression. However, it is important to note that if specific correlates of depression (e.g., social withdrawal, high neuroticism) are salient third variables, then other disorders related to these correlates (e.g., anxiety disorders) may be associ- ated with stress causation as well. Regardless of which theory is supported, the applied consequences are the same: those with a history of depression are prone to later stress and this stress might well cause further depression (Rudolph et al., 2009).

Given that the distinction between the stress causation and stress continuation theories has not previously been articulated, it is not surprising that previous studies have adopted analytic strategies that are insufficient for testing both interpretations. That is, many previous studies have only reported the direct relationship between depression and future stress without accounting for baseline stress (e.g., Daley et al., 1997; Hammen, 1991; Shih, 2006), while a small number of studies have only reported results where baseline stress has been taken into account. Thus, the present study reports results relevant to both theories in a single sample.

Neuroticism and Extraversion

Researchers have noted that depressive symptoms alone do not seem to explain the continued elevated stress following the onset of a depressive episode (Chun et al., 2004; Daley et al., 1997; Hammen, 1991). For example, some past studies have found that people who were depressed at a baseline assessment often contin- ued to have elevated levels of stress long after the depressive episode ends (e.g., Chun et al., 2004). Evidence supports negative

2 ULIASZEK ET AL.

cognitive styles, family factors, and interpersonal styles as impor- tant explanatory factors in stress generation (for a review, see Lui & Alloy, 2010). There is also some evidence pointing to neuroti- cism and extraversion as potential explanatory factors in stress generation.

Neuroticism is conceptualized as a core general risk factor for depression (e.g., Clark, Watson, & Mineka, 1994; Klein, Durbin, & Shankman, 2009), with neuroticism prospectively predicting increased risk for depression (e.g., Kendler, Kuhn, & Prescott, 2004; Krueger, 1999). In cross-sectional analyses, neuroticism is associated with both episodic life stress (Kendler, Gardner, & Prescott, 2003; Saudino, Pederson, Lichtenstein, McClearn, & Plomin, 1997) and chronic life stress (Ormel, Oldehinkel, & Bril- man, 2001; Uliaszek et al., 2010). Moreover, one study found that neuroticism prospectively predicted dependent episodic life stress (Magnus, Diener, Fujita, & Pavot, 1993). Research has also ex- amined the interrelationships among neuroticism, life stress, and depression, with cross-sectional findings demonstrating that neu- roticism partially accounts for the relationship between interper- sonal chronic life stress and depression (Uliaszek et al., 2010).

Concerning extraversion, some studies have shown that de- pressed individuals report lower levels of extraversion than con- trols (Reich, Noyes, Hirschfeld, Coryell, & O’Gorman, 1987; Trull & Sher, 1994) and that extraversion is inversely related to risk for a first onset of depression (e.g., Hirschfeld, Klerman, Lavori & Keller, 1989), although the results are mixed (e.g., Kendler, Neale, Kessler, Heath, & Eaves, 1993). At least one study has demon- strated low extraversion to be concurrently associated with chronic life stress in an adolescent sample (Uliaszek et al., 2010).

Based on the above literature, it is possible that these personality variables partially account for the relationship between stress and depression. Specifically, heightened neuroticism and/or low extra- version could be third variables in the relationship between ele- vated stress and depression. Concerning neuroticism, this might be caused by an increase in emotion sensitivity and negative mood states that make interpersonal interactions conflictual. In the case of low extraversion, this might result from withdrawal from friends, family, or work life.

Anxiety Disorders

Depression is the disorder most frequently examined in stress generation research (Hammen, 2005). However, because anxiety disorders are conceptually and empirically related to depression, they are promising candidates for further exploration. First, de- pressive and anxiety disorders are highly comorbid and have many overlapping features (e.g., Kessler, Chiu, Demler, & Walters, 2005; Mineka, Watson, & Clark, 1998). Thus, certain character- istics of depression that have been shown to be important in stress generation may also be present in some anxiety disorders. For example, negative interpersonal characteristics such as an avoidant coping style act as third variables in stress generation in depression (for review, see Liu & Alloy, 2010). There is also evidence that similar negative interpersonal characteristics are present in anxiety disorders (e.g., Heering & Kring, 2007). Second, as with depres- sion, neuroticism is a risk factor for many anxiety disorders (e.g., Clark et al., 1994; Hayward, Killen, Kraemer, & Taylor, 2000). Some research also supports a relationship between low extraver- sion and several anxiety disorders (see Trull & Sher, 1994),

although other research points to a specific relationship between extraversion and social phobia (Uliaszek et al., 2010; Watson et al., 2005). Thus, if neuroticism and/or extraversion are important third variables in stress generation, then we might see stress generation in anxiety disorders because of their relationship to those personality variables. Third, if the temporal stability of stress is an important factor in stress generation, as posited by the stress continuation theory, and anxiety disorders have a cross-sectional relationship with stress, this would constitute a further reason why we would hypothesize stress continuation in anxiety disorders.

Some studies have indeed reported an association between anx- iety and life stress. For example, social phobia was cross- sectionally associated with interpersonal chronic life stress in our sample of adolescents (Uliaszek et al., 2010). This relationship was partially accounted for by neuroticism and extraversion. In a series of prospective studies examining college students, negative events were shown to be risk factors for anxiety symptoms (Hankin, Abramson, Miller, & Haeffel, 2004). However, at least one study has examined stress generation in self-reported anxiety symptoms utilizing a checklist measure of life stress; this study failed to find evidence of stress generation (Joiner, Wingate, Gencoz & Gencoz, 2005). Liu and Alloy (2010) concluded that more research is needed to further explore these relationships.

Present Study

The present study utilized data from the Youth Emotion Proj- ect—a large multiyear, two-site prospective study examining risk factors for psychopathology in late adolescence (see Zinbarg et al., 2010, for more details). The present analyses used data from two time points, collected approximately one year apart. A goal of this study was to expand on previous stress generation research by examining both interpersonal and noninterpersonal chronic life stress, as well as dependent episodic life stress. Only dependent events were examined because, based on the definition of independent life stress, we believe that a person’s direct actions cannot cause these events. These relationships were examined for both depressive and anxiety disorders. This study also examined the competing hypotheses of the stress continuation and stress causation theories for both depression and anxiety disorders. We also sought to test the role of personality traits in stress generation. Given the abundance of research demonstrat- ing the relationships among personality with depressive and anxiety disorders and the relationships between personality and life stress, we hypothesized that neuroticism and extraversion would at least partially account for stress generation in depres- sive and anxiety disorders.

Method

Participants

Participants were from the Youth Emotion Project, a multiyear, two-site prospective study designed to identify risk factors for emotional disorders in a large sample of late adolescents (see Zinbarg et al., 2010 for more details). The present study included two assessment points, collected approximately 1 year apart. A total of 627 participants were recruited for the Time 1 assessment (T1). These participants, all in their junior year of high school,

3STRESS, PERSONALITY, EMOTIONAL DISORDERS

were recruited over three years based on scores from the neurot- icism scale of the revised Eysenck Personality Questionnaire (EPQ-R-N; Eysenck & Eysenck, 1975), which was administered during mass screening sessions (see Zinbarg et al., 2010). Because neuroticism has been shown to be a risk factor for depressive and anxiety disorders (e.g., Clark et al., 1994; Klein et al., 2009), participants scoring in the top third on the EPQ-R-N were over- sampled. This behavioral high-risk design was intended to over- come statistical problems associated with the low base rates of particular disorders in community samples (Hauner, Revelle & Zinbarg, 2011).1 Of the participants who were invited and who participated in the T1 assessment (n � 627), 368 had high, 145 had medium, and 114 had low scores on the EPQ-R-N (58.7, 23.1, and 18.2%, respectively).

Participants were recruited from two large metropolitan areas (suburban Chicago and suburban Los Angeles). There were 305 participants drawn from the Northwestern University site and 322 participants from the University of California, Los Angeles site. The racial makeup of the total sample was as follows: Caucasian, n � 302 (48.2%); Hispanic/Latin American, n � 96 (15.3%); African American, n � 82 (13.1%); more than one ethnicity, n � 82 (13.1%); other, n � 34 (5.4%); Asian American, n � 27 (4.3%); Pacific Islander, n � 4 (0.6%).

At T1, the participants ranged in age from 15 to 18 years, with a mean (M) age of 16.91 (standard deviation [SD] � .39). The participants included 195 males (31.1%) and 432 females. This gender difference in participation was unintentional and occurred for several reasons, such as females being more likely to agree to complete the screening questionnaire and to participate in the study if invited. Because females tend to score higher on neurot- icism (see Costa et al., 2001) and due to our behavioral high risk design, more females were invited to participate.

Measures

Structured Clinical Interview for the Diagnostic and Statis- tical Manual for Mental Disorders–IV (DSM)–IV. (SCID; First, Spitzer, Gibbons, & Williams, 2002). Diagnoses of current Axis I disorders were made using the SCID. SCID interviewers were graduate students, postdoctoral fellows, and bachelor’s level research assistants. Training included approximately 60 hours of didactics, matching gold standard ratings, role-playing, and live observations. Each diagnosis was presented at a supervision and consensus meeting led by doctoral-level supervisors. To maintain consistency across sites, difficult cases were presented at weekly teleconferences that were periodically attended by supervisors from the other site.

Interrater reliability for categorical DSM–IV (American Psychi- atric Association, 2004) diagnoses was assessed by having trained interviewers observe live SCIDs on a subset of 69 cases. Cohen’s (1960) kappa was acceptable to good when aggregated across all disorders (� � .82) and for the individual disorders, including major depressive disorder (� � .83), social phobia (� � .65), generalized anxiety disorder (� � .85), and obsessive– compulsive disorder (� � .85). Kappa estimates are only available for these disorders because they were the only ones rated frequently enough (at least five cases) in this subset of cases.

The depressive and anxiety disorder variables included in the present study required the participant to meet criteria for the

disorder and demonstrate clinically significant distress and/or im- pairment. Participants with a not-otherwise-specified depressive or anxiety diagnosis were labeled as not having the specific disorder (n � 44). Each diagnostic variable was labeled as present or absent for every participant. Participants with no present diagnoses were not excluded from analyses. The depressive disorders variable in the current study consisted of current cases of major depressive disorder (n � 23) and dysthymia (n � 7). The anxiety disorders variable consisted of current cases of the following disorders: social phobia (n � 52); posttraumatic stress disorder (n � 4); obsessive– compulsive disorder (n � 9); generalized anxiety dis- order (n � 17); specific phobia (n � 37); and panic disorder (n � 3). Sixteen participants met criteria for both a current depressive and anxiety disorder.

Life Stress Interview (LSI; Hammen et al., 1987). Life stress was evaluated using the LSI, a semistructured interview that assesses chronic life stress and episodic life stress. The chronic life stress portion assessed the level of ongoing objective stress expe- rienced by the participant in 10 domains over the past year. This version of the LSI contains chronic life stress domains relevant to an adolescent population, including four interpersonal domains (close friendship, social life, romantic relationships, and family) and six noninterpersonal domains (neighborhood, school, work, finances, personal health, and health of close family members). Unlike the episodic stressors, the chronic life stress domains can- not be separated by dependence. To determine chronic life stress scores, the interviewer used suggested general probes to elicit relevant objective information. Subjective impressions offered by the participant were not probed for and were disregarded if offered. Ratings ranged in half-point intervals from 1 (ideal circumstances) to 5 (most stressful circumstances), with specific behavioral an- chors for each point on the scale. Training of interviewers involved approximately 30 hours of didactics, matching gold standard rat- ings, role-plays, and live observations. T1 reliability was assessed by rating 76 intersite and intrasite audio recorded interviews. Intraclass correlation coefficients (ICCs) ranged from .58 for health-other to .92 for neighborhood. Averaged across all domains, the ICC was .70.

Episodic events were probed for within each chronic life stress domain. Interviewers obtained details concerning the description and date of the event, the degree, duration, and impact of its consequences, the participant’s prior experience with the event, and availability of social support. This information was later presented by the interviewer to an independent team of two raters who evaluated the event on its level of contextual threat. Any subjective impressions the participants offered about the stressful- ness of an event, as well as any diagnostic information about the participant, were not presented to the raters. Contextual threat was assessed by objectively rating how much impact a particular epi-

1 To ensure that our results were not biased due to the behavioral high-risk design, all analyses were also completed including sampling weights to adjust for the differing distributions of neuroticism, sex, and ethnicity in our sample compared to the population. There were no differ- ences in statistical significance between results with or without the inclu- sion of sampling weights. Thus, we followed the recommendation of Winship and Radbill (1994) and presented the unweighted results.

4 ULIASZEK ET AL.

sodic event would have for the average person in those exact circumstances.

Ratings for episodic events were made on a 1–5 scale: 1 (min- imal or no threat), 2 (mild threat), 3 (moderate threat), 4 (marked impact with many consequences), and 5 (severe and catastrophic negative impact). T1 reliability was assessed by rating 208 audio recordings of life events across sites. The ICC was .82. In addition to a contextual threat rating for episodic events, raters assessed the dependence of the event, with 1 denoting complete independence and 5 signifying that the event was completely caused by the respondent. Most interpersonal events were given a rating of three with the assumption that the event was at least in part the result of both parties. If raters could not reach consensus, the episode was then presented to a third rater who helped the raters reach consensus. The ICC was .90 for the dependence ratings for the same 208 events used to assess the reliability of the contextual threat ratings. The episodic life stress variable in the present study consisted of the average contextual threat ratings for all dependent events. Dependence ratings greater than 2 were included in the present analyses based on the assumption that independent epi- sodic life stress as defined here could not logically be “generated” by any type of individual difference variable. However, we do acknowledge that there may be some examples where a person might select themselves into an environment where independent events are more likely to happen (i.e., not planning a proper route, ending up in a high crime neighborhood, and getting robbed) or where a stable negative environment might contribute to both a person’s mood state and the likelihood of independent events (Harkness & Stewart, 2009).

Eysenck Personality Questionnaire—Revised, Neuroticism Scale (EPQ-R; Eysenck & Eysenck, 1975). The neuroticism scale of the EPQ-R was the initial screening questionnaire for the present study. It consists of 222 items in a yes–no format, with higher scores indicative of higher levels of neuroticism. Coeffi- cient alpha was .79 and coefficient omegahierarchical (�h; Zinbarg, Revelle, Yovel, & Li, 2005) was .66 (Mor et al., 2006). The extensive construct validity for this instrument is reported in the EPQ-R manual (Eysenck & Eysenck, 1975).

International Personality Item Pool-NEO-PI-R (IPIP-N, 2000). The neuroticism scale from the IPIP-N consists of 60 items rated on a 1–5 Likert Scale. This scale was developed to closely correspond with the neuroticism scale from the NEO-PI-R (Costa & McCrae, 1985). Goldberg (1999) reported that the total scores for the two scales correlate .93. A confirmatory factor analysis completed on one half of the T1 data confirmed six facets and one general factor underlying the IPIP-N (Uliaszek et al., 2009). Because of fit indices below conventional levels of acceptable fit, the model was modified. First, the sample was randomly split and modifications were tested in the second subsample suggesting correlated residuals. This resulted in 21 items being cut from the original measure. This altered model was confirmed in the second half of the data, revealing a satisfactory fit for this revised version and an �h estimate of .86 on the full sample (for details, see Uliaszek et al., 2009). Therefore, the 39 item IPIP-N measure was used in the current analyses.

The Behavioral Inhibition Scale (BIS; Carver & White, 1994). The BIS, which measures concern over and sensitivity to negative outcomes, consists of seven items rated on a 4-point

Likert scale. Carver and White (1994) have demonstrated the convergent and divergent validity of the BIS. The coefficient alpha at T1 in this study was .75.

Big Five Mini-Markers Scale (Saucier, 1994). This 40-item measure consists of eight items assessing each of the Big 5 personality traits: neuroticism, extraversion, agreeableness, con- scientiousness, and openness/intellect. The extraversion and neu- roticism scales were used in the present analyses. Each item is rated on a 9-point Likert scale ranging from extremely inaccurate to extremely accurate. Saucier (1994) reported a coefficient alpha of .76 for the neuroticism scale and .85 for the extraversion scale. In this study, the coefficient alphas at T1 were .80 for both neuroticism and extraversion.

Procedures

A mass screening of potential participants was completed during school hours. For both T1 and Time 2 (T2) assessments, SCID and LSI interviews occurred after regular school hours throughout the entirety of the school year. Participants were interviewed in person for approximately 1.5 to 3 hr at each time point. Questionnaire measures were completed either immediately after the interviews or arrangements were made for the participant to return to com- plete the questionnaires, usually within the next week. All ques- tionnaires were completed at both T1 and T2 except for the EPQ-R-N, which was used only as the screening questionnaire at the beginning of T1.

A total of 497 participants from T1 also participated at T2. Similar to the full T1 sample, T2 included 30.6% males (n � 152) and 69.4% females (n � 345). T1 noninterpersonal chronic life stress was the only variable in the present study to predict attrition at T2 (B � .54, standard error [SE] B � .25, Wald �2 � 4.42, p � .05, odds ratio � 1.71). When each noninterpersonal chronic life stress domain was looked at individually, the only domain to predict attrition was the school domain (B � .38, SE B � .13, Wald �2 � 8.43, p � .01, odds ratio � 1.46). Thus, participants who experienced more school-related stress at T1 were less likely to complete the assessment at T2.

Data Analysis

The present analyses were completed using Mplus structural equation modeling software (Muthén & Muthén, 2007) with full information maximum likelihood estimation. The use of this method allowed us to include all 627 participants in analyses and potentially corrected at least some of the biases that would ensue from including only participants with complete data (see McArdle, 1994). Only neuroticism was modeled as a latent variable because it was the only variable to have multiple indicators. Thus, struc- tural equation modeling was used for all analyses including neu- roticism. All other analyses were conventional regressions con- ducted using Mplus in order to handle missing data using full information maximum likelihood. A description of the models is

2 The original EPQ Neuroticism Scale consists of 24 items. The item referring to suicidality was omitted based on recommendations from the Institutional Review Board. The item, “Do you worry about your health” was also omitted from scoring because it failed to load on any factor in preliminary analyses (Mor et al., 2008).

5STRESS, PERSONALITY, EMOTIONAL DISORDERS

shown in Figure 1. First, we examined the zero-order path from T1 disorder (depressive disorders or anxiety disorders) predicting T2 stress (interpersonal chronic life stress, noninterpersonal chronic life stress, or episodic stress; Figure 1, Panel a). Second, we examined the path from T1 disorder predicting T2 stress with T1 stress as a covariate (Figure 2, Panel b, path a). The covariate was the same type of stress as the dependent variable. A significant zero-order association of T1 disorder with T2 stress with a non- significant path a provided support for stress continuation, while a significant path a provided support for stress causation. If we found evidence for stress generation, we then completed an addi- tional analysis with neuroticism and/or extraversion included in the model to test whether these two personality traits at least partially accounted for the effect of T1 disorder on T2 stress above and beyond T1 stress (Panel c). We also examined the unique relationships of these two T1 personality variables predicting T2 life stress, accounting for T1 life stress, as a means of assessing a stress causation theory for these two personality variables, as well as these two personality traits as third variables partially explain- ing stress continuation (see Figure 2).

Results

Means and standard deviations are displayed in Table 1. Cor- relations among T1 interpersonal chronic life stress, noninterper- sonal chronic life stress, the average contextual threat rating for dependent episodic life stress, neuroticism, and extraversion are displayed in Table 2. All correlations were significant with the exception of the correlation between T1 extraversion and T1 episodic life stress. The T2 correlations among interpersonal chronic life stress, noninterpersonal chronic life stress, and epi-

Figure 1. Panel a. Time 1 (T1) disorder predicting Time 2 (T2) life stress. Panel b. T1 disorder predicting T2 life stress. T1 life stress, the same type of stress as the dependent variable, is entered as a covariate. Panel c. T1 disorder and T1 personality predicting T2 life stress with T1 life stress entered as a covariate. Disorder and stress variables are in square boxes because they indicate observed variables. Personality is represented by a circle because, in the case of neuroticism, it is a latent variable with multiple indicators.

Figure 2. Time 1 (T1) personality predicting Time 2 (T2) life stress. T1 life stress, the same type of stress as the dependent variable, is entered as a covariate. Stress variables are in square boxes because they indicate observed variables. Personality is represented by a circle because, in the case of neuroticism, it is a latent variable with multiple indicators.

Table 1 Means and Standard Deviations for All Interpersonal Chronic Life Stress Domains, Noninterpersonal Chronic Life Stress Domains, the Average Contextual Threat Rating of Dependent Episodic Life Stress, Neuroticism Measures, and Extraversion at Time 1(T1) and Time 2 (T2)

T1 Mean (SD), n � 697

T2 Mean (SD), n � 497

Interpersonal chronic life stress 2.39 (.47) 2.31 (.44) Noninterpersonal chronic life stress 2.20 (.38) 2.14 (.34) Episodic life stress 1.30 (.80) 1.27 (.86) Moderate episodic life stress .90 (1.30) .44 (1.02) Neuroticism

EPQ-R-N 11.89 (4.52) IPIP-N 2.65 (.65) BIS 2.90 (.58) Big Five Mini-Markers Scale-N 4.77 (1.42) Extraversion 5.70 (1.34)

Note. EPQ-R-N � neuroticism scale of the revised Eysenck Personality Questionnaire; IPIP-N � neuroticism scale of the International Personality Item Pool; BIS � the Behavioral Inhibition Scale; SD � standard devia- tion. The range of scores is as follows: interpersonal chronic life stress (1–5), noninterpersonal chronic life stress (1–5), episodic life stress (0 –5), moderate episodic life stress (0 –5), EPQ-R-N (0 –22), IPIP-N (0 –5), BIS (0 – 4), Big Five Mini-Markers Scale-N (0 –9), extraversion (0 –9).

6 ULIASZEK ET AL.

sodic life stress also are shown in Table 2, above the diagonal. All T2 correlations were significant.

Neuroticism Measurement Model

A measurement model was estimated using full information maximum likelihood to evaluate a unidimensional model of the latent structure of T1 neuroticism as measured by the EPQ-R-N, IPIP-N, BIS, and Big Five Mini-Markers N Scale. Standardized factor loadings for each of these measures are as follows: EPQ- R-N � .68, IPIP-N � .90, BIS � .65, and Mini-Markers � .77. This model had excellent fit indices of �2(2) � 1.34, nonsignifi- cant, comparative fit index � 1.00, root mean square error of approximation � .00 (90% confidence interval [CI]: .00 –.07), and standardized root mean square residual � .01.

Depression and Anxiety Predicting Episodic Life Stress

First, we determined that there was a significant direct path between T1 and T2 episodic life stress (standardized regression weight � .18, SE � .04, p � .001). Next, we examined the zero-order path between T1 depressive and anxiety disorders pre- dicting T2 episodic life stress (see Table 3). Both relationships were significant. We then examined T1 depressive and anxiety disorders predicting T2 episodic life stress with T1 episodic life stress as a covariate (see Table 4). Only the relationship between T1 anxiety disorders and T2 episodic life stress remained signifi- cant after accounting for T1 episodic life stress. Together these results supported the stress causation theory for anxiety disorders and the stress continuation theory for depressive disorders.

However, the above analyses included a high frequency of mild stressors. Because research has often supported a specific role of only moderate to severe stress (and not mild stress) in the predic- tion of major depression (see Hammen, 2005; Monroe & Reid, 2009), we thought it was important to also examine the relation- ship between depression and moderate to severe levels of stress. This is because such results would provide stronger evidence that stress generation might play in the role of maintenance and/or recurrence of depression. Thus we completed additional analyses with the focus on the generation of more moderate to severe events (average episodic life stress rating of 2.5 or higher on a 5-point scale). We then repeated the above analyses including only these

stressors. There was a significant direct path between T1 and T2 moderate episodic life stress (standardized regression weight � .16, SE � .04, p � .001). Depressive disorders predicted moderate episodic life stress after accounting for baseline moderate episodic life stress, supporting the stress causation theory (see Table 3). These analyses were repeated for the anxiety disorders, as well as the personality trait analyses (see below). In these cases, there were no differences in results between the full range of episodic stressors and only the moderate episodic stressors. Thus, they are not discussed further and follow-up analyses focus only on the full range of episodic stressors for these variables.

Depression and Anxiety Predicting Chronic Life Stress

For chronic life stress, we first determined that there was a significant zero-order association between T1 and T2 interpersonal

Table 2 Correlations Among Time 1 Interpersonal Chronic Life Stress, Noninterpersonal Chronic Life Stress, the Average Contextual Threat Rating of Dependent Episodic Life Stress, Neuroticism, and Extraversion

Interpersonal chronic life stress

Noninterpersonal chronic life stress

Episodic life stress Neuroticism

Interpersonal chronic life stress — .45��� .15��� — Noninterpersonal chronic life stress .47��� — .17��� — Episodic life stress .17��� .16��� — — Neuroticism .38��� .05��� .14�� — Extraversion �.22��� �.13�� .08 �.47���

Note. The correlations above the diagonal represent the correlations at Time 2, N � 627. Neuroticism is a latent variable measured by the neuroticism scale of the revised Eysenck Personality Questionnaire, neuroticism scale of the International Personality Item Pool, Big Five Minimarkers Scale-N, and the Behavioral Inhibition Scale. � p � .05. �� p � .01. ��� p � .001.

Table 3 Results Examining the Longitudinal Relationships Between Time 1 Disorders and Personality Predicting Time 2 Life Stress, N � 627

Standardized regression weight (SE)

Depressive disorders Episodic life stress (all levels) .10 (.04)�

Moderate episodic life stress .12 (.04)��

Interpersonal chronic life stress .21 (.04)���

Noninterpersonal chronic life stress .20 (.04)���

Anxiety disorders Episodic life stress .16 (.04)���

Moderate episodic life stress 16 (.04)���

Interpersonal chronic life stress .16 (.04)���

Noninterpersonal chronic life stress .16 (.04)���

Neuroticism Episodic life stress .13 (.04)��

Moderate episodic life stress .14 (.05)��

Interpersonal life stress .35 (.04)���

Noninterpersonal life stress .09 (.05) Extraversion

Episodic life stress �.04 (.03) Moderate episodic life stress �.03 (.05) Interpersonal life stress �.20 (.04)���

Noninterpersonal life stress �.14 (.05)��

Note. SE � standard error. � p � .05. �� p � .01. ��� p � .001.

7STRESS, PERSONALITY, EMOTIONAL DISORDERS

chronic life stress (standardized regression weight � .55, SE � .03, p � .001) and noninterpersonal chronic life stress (standard- ized regression weight � .68, SE � .02, p � .001). Next, we examined the zero-order association between depressive and anx- iety disorders and T2 chronic life stress (see Table 3). We found that T1 depressive disorders predicted both interpersonal and non- interpersonal chronic life stress at T2, but these relationships did not remain significant after accounting for T1 chronic life stress (see Table 4). Concerning anxiety disorders, we also found evi- dence for the stress continuation theory but not the stress causation theory in chronic life stress. Specifically, T1 anxiety disorders significantly predicted both interpersonal and noninterpersonal chronic life stress at T2 (see Table 3), but these relationships did not remain significant after accounting for T1 stress (see Table 4).

Effects of Personality Variables

We first examined the zero-order association of T1 personality with T2 life stress. These results are displayed in Table 3. T1 neuroticism significantly predicted both T2 episodic life stress and T2 interpersonal chronic life stress. T1 extraversion also signifi- cantly predicted T2 interpersonal and noninterpersonal chronic life stress. We then examined these relationships while accounting for T1 life stress. The relationships between T1 neuroticism and T2 episodic life stress, and T2 interpersonal chronic life stress, re- mained significant after accounting for T1 life stress (see Table 5), supporting the stress causation theory. T1 extraversion was not a significant predictor of T2 stress after accounting for T1 stress, supporting the stress continuation theory.

Our final set of analyses examined whether neuroticism at least partially accounted for stress causation in the relationships be- tween T1 depressive disorders and T2 moderate episodic life stress (accounting for T1 moderate episodic life stress) and T1 anxiety disorders and T2 episodic life stress (accounting for T1 episodic

life stress). Only neuroticism was examined because we did not find evidence that extraversion was associated with T1 or T2 episodic life stress. We completed the product of coefficients test using the program PRODCLIN (distribution of the PRODuct Con- fidence Limits for Indirect effects; MacKinnon, Fritz, Williams, & Lockwood, 2007). This program converts the regression weights and standard errors from the analyses into z scores and finds critical values based on the product of two random variables. The result is a nonsymmetrical confidence interval around the product of coefficients. If it includes zero, the effect is not significant (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002; MacK- innon, Lockwood, & Williams, 2004). Although this test is often used in meditational analyses, it is also appropriate for tests of confounding or third variables (MacKinnon, Krull, & Lockwood, 2000). In these analyses, the � coefficient is the direct relationship between the disorder of interest (depression or anxiety) and neu- roticism. The � coefficient is the relationship between neuroticism and T2 life stress after accounting for the disorder of interest and T1 life stress. Thus, results (for the product of ��) indicate the role of neuroticism in the relationship between disorder and T2 life stress, after accounting for T1 life stress.

The first analysis examined T1 depressive disorders (standard- ized regression weight � .08, SE � .05, ns), moderate episodic life stress (standardized regression weight � .13, SE � .05, p � .01), and neuroticism (standardized regression weight � .11, SE � .05, p � .05) as predictors of T2 moderate episodic life stress. In this analysis, T1 neuroticism was significantly correlated with T1 depression (r � .18, SE � .04). The product of coefficients test revealed that neuroticism partially accounted for the stress causa- tion effect in depression (�� � .02, confidence interval � .002– .04). Next, we examined T1 anxiety disorders (standardized re- gression weight � .11, SE � .05, p � .05), episodic life stress (standardized regression weight � .15, SE � .04, p � .05), and neuroticism (standardized regression weight � .08, SE � .05, ns.) as predictors of T2 episodic life stress. In this analysis, T1 neu- roticism was significantly correlated with T1 anxiety disorders (r � .37, SE � .04). The product of coefficients test revealed that

Table 5 Results Examining the Longitudinal Relationships Between Time 1 Personality Variable and Time 2 Life Stress, Accounting for T1 Life Stress, N � 627

a b

Standardized regression

weight (SE)

Standardized regression

weight (SE)

Neuroticism Episodic life stress .13 (.05)� .16 (.04)���

Moderate episodic life stress .12 (.05)� .14 (.05)���

Interpersonal chronic life stress .15 (.04)��� .49 (.04)���

Extraversion Interpersonal chronic life stress �.06 (.04) .54 (.03)���

Noninterpersonal chronic life stress �.01 (.03) .69 (.02)���

Note. In the column to the left, a represents the relationship between T1 personality and T2 life stress, while accounting for T1 life stress (see Figure 2, Path a ). b represents the relationship between T1 life stress and T2 life stress within the specified relationship. � p � .05. �� p � .01. ��� p � .001.

Table 4 Results Examining the Longitudinal Relationships Between Time 1 Disorder and Time 2 Life Stress, Accounting for T1 Life Stress, N � 627

a b

Standardized regression

weight (SE)

Standardized regression

weight (SE)

Depressive disorders Episodic life stress .08 (.04) .16 (.04)���

Moderate episodic life stress .10 (.04)� .15 (.04)���

Interpersonal chronic life stress .05 (.04) .54 (.03)���

Noninterpersonal chronic life stress .05 (.03) .68 (.03)���

Anxiety disorders Episodic life stress .14 (.04)��� .16 (.04)���

Moderate episodic life stress .15 (.04)��� .15 (.04)���

Interpersonal chronic life stress �.02 (.04) .54 (.03)���

Noninterpersonal chronic life stress �.004 (.03) .69 (.02)���

Note. In the column to the left, a represents the relationship between T1 disorder and T2 life stress, while accounting for T1 life stress (see Figure 1 Panel b, Path a). b represents the relationship between T1 life stress and T2 life stress within the specified relationship. � p � .05. �� p � .01. ��� p � .001.

8 ULIASZEK ET AL.

neuroticism partially accounted for stress causation in anxiety disorders (�� � .04, confidence interval � .004 –.08).

Discussion

This is the first prospective study to demonstrate stress gener- ation in both depressive and anxiety disorders using DSM–IV diagnoses and an interview-based assessment of stress. In addition, we believe this is the first study to directly examine both the stress continuation and stress causation theories of stress generation in the same study. These results also examined the role of neuroti- cism and extraversion in stress generation, with neuroticism par- tially accounting for episodic stress generation in both depressive and anxiety disorders.

Our analyses supported the stress causation theory when exam- ining stress generation in depressive disorders with moderate to severe episodic life events. Thus, something specific to depression at baseline, apart from elevated concurrent stress, has a predictive relationship to later moderate to severe episodic stressors. One such variable that we found to be important in this relationship is elevated neuroticism. Neuroticism partially accounted for the stress generation relationship between depression and moderate episodic stressors, even after accounting for baseline stress. These results have both theoretical and clinical implications. First, these results show that, relative to the nondepressed, people with depression are at an elevated risk for future dependent episodic stress of moderate to severe intensity. These types of stressful life events may well lead to continued depressive symptoms and future recurrence of depressive episodes (e.g., Rudolph et al., 2009). This provides one possible explanation for the high recurrence rate of depression (e.g., Boland & Keller, 2009) and for subclinical depressive symptoms between episodes of depression (Judd et al., 1998). Second, these findings suggest that clinical interventions should target stress reduction as a means of preventing future depressive episodes.

It is important to note that a few other studies have found a predictive relationship between depression and episodic stress, even after accounting for baseline stress (e.g., Harkness & Stewart, 2009; Holahan et al., 2005; Rudolph et al., 2009), that is, support- ing the stress causation theory. In our sample, this was only replicated with moderate to severe episodic life stress. There are several possible explanations for why we did not replicate these findings for analyses including all levels of episodic life stress. First, our analyses did not separate interpersonal from noninter- personal episodic life stress (although we did for chronic life stress). Several studies have found stress generation for depression to be stronger for interpersonal episodic life stress (e.g., Rudolph et al., 2000; Shih, 2006). Second, we used more rigorous method- ology than many other studies, including clinically significant diagnoses for our disorder variables and an interview-based mea- sure of life stress. It is possible that more studies would have failed to support stress generation in depression if using such stringent criteria.

The findings on episodic stress generation in anxiety disorders are an exciting addition to the stress generation literature, with this evidence demonstrating that the phenomenon is not specific to depression. Because we found evidence of stress generation in anxiety disorders, we are led to believe that there is something about the characteristics of anxiety disorders that lead to elevated levels of episodic stress. We found that high neuroticism is one

such characteristic, partially accounting for the relationship be- tween T1 anxiety disorders and T2 episodic life stress, after accounting for T1 episodic life stress. However, neuroticism did not completely explain stress generation in anxiety. Another im- portant characteristic to examine might be the interpersonal pro- cesses similar to what is seen in depression. Much research points to an interpersonal theory of stress generation in depression, with a focus on interpersonal style and interpersonal stressors (e.g., Eberhart & Hammen, 2009; Flynn, Kecmanovic, & Alloy, 2010), and how this process may lead to future depressive episodes (Rudolph et al., 2009). Future studies examining stress generation in anxiety disorders might examine negative interpersonal charac- teristics as variables that might, at least in part, account for this relationship.

A recent review article has noted the importance of examining chronic life stress in stress generation research (Liu & Alloy, 2010). We found evidence for stress continuation theory, but not for stress causation, for both interpersonal and noninterpersonal chronic life stress in both depressive and anxiety disorders. These results demonstrate that, although characteristics of depressive and anxiety disorders do not appear to be causal agents in the gener- ation of chronic life stress, those with these disorders are prone to future elevated levels of chronic life stress nonetheless. This is clinically and theoretically important when thinking about the developmental course and prognosis of those with depressive and anxiety disorders. Similar to a scar effect, we expect those with a history of a depressive or anxiety disorder would continue to experience elevated levels of chronic life stress relative to those without such a history. It is possible that we did not find support for the stress causation theory with chronic life stress because it is by its nature enduring and so it may be more difficult to detect changes. In fact, results supported the high stability of both inter- personal and noninterpersonal chronic life stress.

As mentioned above, neuroticism was a significant third vari- able in stress generation for depressive and anxiety disorders. Results looking solely at neuroticism also supported the stress causation theory for both episodic life stress and interpersonal chronic life stress. In other words, T1 neuroticism predicted both types of stress at T2, even after accounting for baseline stress. Taken together, these findings demonstrate that being high in neuroticism, at least in late adolescence, is a risk factor for later chronic and episodic stress. We can at least partially attribute this continued elevation in stress to specific characteristics or corre- lated behaviors of a person high in neuroticism. For example, increased emotional sensitivity may lead a person to actions that interfere with social and romantic relationships or school and work performance.

Extraversion was not associated with a significant stress causa- tion effect for any type of stress. However, T1 extraversion did predict both T2 interpersonal and noninterpersonal chronic life stress, supporting the stress continuation theory. Support of this theory does provide evidence that those displaying the withdrawn, nonassertive behavior characteristic of introversion are likely to have continued elevated chronic life stress in the future, if only due to the cross-sectional relationship between chronic life stress and introversion and the temporal stability of chronic life stress.

There are several limitations of the current study. First, because our sample consisted entirely of late adolescents and because we present several new findings, our results need to be replicated in

9STRESS, PERSONALITY, EMOTIONAL DISORDERS

additional samples before firm conclusions can be drawn. This is also pertinent to the results concerning stress generation in anxiety disorders because this area of study is quite new. Second, there was a large difference in the number of participants meeting criteria for a depressive disorder (n � 28) versus an anxiety disorder (n � 92). The resulting difference in power was another possible reason why the stress causation theory was supported for the full range of episodic stressors in anxiety disorders, but not in depressive disorders. In this regard, however, we also note that the effect size estimates for depressive disorders were consistently smaller than those for the anxiety disorders suggesting that power alone cannot entirely explain the differences in results between depressive disorders and anxiety disorders. Third, participants reported an average of two dependent episodic life events in T1 (M � 1.90, SD � 1.63) and 1.5 in T2 (M � 1.54, SD � 1.48). One possible reason for the decrease is the socialization of the partic- ipants to the LSI; they may have learned that reporting fewer episodic events results in a shorter interview. In addition, attrition was predicted by chronic life stress in the school domain suggest- ing that those having higher school stress were also likely to drop out of the study. A fourth limitation is that we only tested a small set of possible third variables. It is possible that there are other third variables with some of them having even larger effects than the ones we examined. A final limitation is related to the approx- imate one-year time lag between assessments. This time lag was a matter of convenience; any shorter period of time would unduly tax the participants, whereas a longer lag would make it too difficult to recall the stress since the previous interview. Thus, it is possible that the results would be different if there were more or less time between assessments.

In conclusion, we believe that this study provides some of the most comprehensive evidence to date on stress generation. We examined the effects of both depression and anxiety disorders on both chronic and episodic life stress. To our knowledge, ours is the first study to distinguish between and test both the stress contin- uation and stress causation theories of stress generation. Moreover, we examined the role of normal personality traits in stress gener- ation, with results showing that neuroticism partially accounted for the predictive relationship between depressive disorders and mod- erate episodic life stress, as well as the relationship between anxiety disorders and episodic life stress. Finally, this study pro- vides new evidence demonstrating that stress generation is not specific to depressive disorders, but also occurs in those with anxiety disorders and elevated neuroticism.

References

American Psychiatric Association. (2004). Diagnostic and statistical man- ual of mental disorders (4th ed.) – Text Revision. Washington, DC: American Psychiatric Association.

Boland, R. J., & Keller, M. B. (2009). Course and outcome of depression. In I. H. Gotlib & C. L. Hammen (Eds.). Handbook of depression (2nd ed., pp. 23– 43). New York, NY: Guilford Press.

Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS scales. Journal of Personality and Social Psychology, 67, 319 –333. doi:10.1037/0022-3514.67.2.319

Chun, C. A., Cronkite, R. C., & Moos, R. H. (2004). Stress generation in depressed patients and community controls. Journal of Social and Clin- ical Psychology, 23, 390 – 412. doi:10.1521/jscp.23.3.390.35453

Clark, L. A., Watson, D., & Mineka, S. (1994). Temperament, personality, and the mood and anxiety disorders. Journal of Abnormal Psychology, 103, 103–116. doi:10.1037/0021-843X.103.1.103

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educa- tional and Psychological Measurement, 20, 37– 46. doi:10.1177/ 001316446002000104

Cole, D. A., Girgus, J., Paul, G., & Nolen–Hoeksema, S. (2004). Stress exposure and stress generation in child and adolescent depression: A latent state-trait-error approach to longitudinal analyses. Journal of Abnormal Psychology, 115, 40 –51. doi:10.1037/0021-843X.115.1.40

Costa, P. T., Jr., & McCrae, R. R. (1985). The NEO Personality Inventory manual. Odessa, FL: Psychological Assessment Resources.

Costa, P. T., Jr., Terracciano, A., & McCrae, R. R. (2001). Gender differences in personality traits across cultures: Robust and surprising findings. Journal of Personality and Social Psychology, 81, 322–331. doi:10.1037/0022-3514.81.2.322

Daley, S. E., Hammen, C., Burge, D., Davila, J., Paley, B., Lindberg, N., & Herzberg, D. S. (1997). Predictors of the generation of episodic stress: A longitudinal study of late adolescent women. Journal of Abnormal Psychology, 106, 251–259. doi:10.1037/0021-843X.106.2.251

Daley, S. E., Hammen, C., & Rao, U. (2000). Predictors of first onset and recurrence of major depression in young women during the 5 years following high school graduation. Journal of Abnormal Psychology, 109, 525–533. doi:10.1037/0021-843X.109.3.525

Eberhart, N. K., & Hammen, C. L. (2009). Interpersonal predictors of stress generation. Personality and Social Psychology Bulletin, 35, 544 – 556. doi:10.1177/0146167208329857

Eysenck, H. J., & Eysenck, S. B. (1975). Eysenck Personality Question- naire manual. San Diego, CA: EdITS.

First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (2002). Structured Clinical Interview for DSM–IV–TR Axis I Disorders, re- search version, patient edition with psychotic screen. New York, NY: Biometrics Research, New York State Psychiatric Institute.

Flynn, M., Kecmanovic, J., & Alloy, L. B. (2010). An examination of integrated cognitive-interpersonal vulnerability to depression: The role of rumination, perceived social support, and interpersonal stress gener- ation. Cognitive Therapy and Research, 34, 456 – 466. doi:10.1007/ s10608-010-9300-8

Goldberg, L. R. (1999). A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models. In I. Mervielde, I. Deary, F. De Fruyt, & F. Ostendorf (Eds.), Personality psychology in Europe (Vol. 7, pp. 7–28). Tilburg, The Netherlands: Tilburg University Press.

Hammen, C. (1991). Generation of stress in the course of unipolar depres- sion. Journal of Abnormal Psychology, 100, 555–561. doi:10.1037/ 0021-843X.100.4.555

Hammen, C. (2005). Stress and depression. Annual Review of Clinical Psychology, 1, 293–319. doi:10.1146/annurev.clinpsy.1.102803.143938

Hammen, C., Adrian, C., Gordon, D., Burge, D., Jaenicke, C., & Hiroto, D. (1987). Children of depressed mothers: Maternal strain and symptom predictors of dysfunction. Journal of Abnormal Psychology, 96, 190 – 198. doi:10.1037/0021-843X.96.3.190

Hammen, C., Davila, J., Brown, G., Ellicott, A., & Gitlin, M. (1992). Psychiatric history and stress: Predictors of severity of unipolar depres- sion. Journal of Abnormal Psychology, 101, 45–52. doi:10.1037/0021- 843X.101.1.45

Hankin, B. L., Abramson, L. Y., Miller, N., & Haeffel, G. J. (2004). Cognitive vulnerability-stress theories of depression: Examining affec- tive specificity in the prediction of depression versus anxiety in three prospective studies. Cognitive Therapy and Research, 28, 309 –345. doi:10.1023/B:COTR.0000031805.60529.0d

Harkness, K. L., & Stewart, J. G. (2009). Symptom specificity and the prospective generation of life events in adolescence. Journal of Abnor- mal Psychology, 118, 278 –287. doi:10.1037/a0015749

10 ULIASZEK ET AL.

Hauner, K., Zinbarg, R., & Revelle, W. (2011). The effects of oversampling on directly and indirectly oversampled predictors. Manuscript in prep- aration.

Hayward, C., Killen, J. D., Kraemer, H. C., & Taylor, C. B. (2000). Predictors of panic attacks in adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 39, 207–214. doi:10.1097/ 00004583-200002000-00021

Heering, E. A., & Kring, A. M. (2007). Interpersonal consequences of social anxiety. Journal of Abnormal Psychology, 116, 125–134. doi: 10.1037/0021-843X.116.1.125

Hirschfeld, R. M., Klerman, G. L., Lavori, P., Keller, M. B., Griffith, P., & Coryell, W. (1989). Premorbid personality assessments of first onset of major depression. Archives of General Psychiatry, 46, 345–350.

Holahan, C. J., Moos, R. H., Holahan, C. K., Brennan, P. L., & Schutte, K. K. (2005). Stress generation, avoidance coping, and depressive symp- toms: A 10-year model. Journal of Consulting and Clinical Psychology, 73, 658 – 666. doi:10.1037/0022-006X.73.4.658

International Personality Item Pool. (2001). A scientific collaboratory for the development of advanced measures of personality traits and other individual differences Retrieved from http://ipip.ori.org/

Joiner, T. E., Jr., Wingate, L. R., Gencoz, T., & Gencoz, F. (2005). Stress generation in depression: Three studies on its resilience, possible mech- anism, and symptom specificity. Journal of Social & Clinical Psychol- ogy, 24, 236 –253. doi:10.1521/jscp.24.2.236.62272

Judd, L. L., Akiskal, H. S., Maser, J. D., Zeller, P. J., Endicott, J., Coryell, W., . . . Keller, M. B. (1998). A prospective 12-year study of subsyn- dromal and syndromal depressive symptoms in unipolar major depres- sive disorders. Archives of General Psychiatry, 55, 694 –700. doi: 10.1001/archpsyc.55.8.694

Kendler, K. S., Gardner, C. O., & Prescott, C. A. (2003). Personality and the experience of environmental adversity. Psychological Medicine, 33, 1193–1202. doi:10.1017/S0033291703008298

Kendler, K. S., Kuhn, J., & Prescott, C. A. (2004). The interrelationship of neuroticism, sex, and stressful life events in the prediction of episodes of major depression. American Journal of Psychiatry, 161, 631– 637. doi: 10.1176/appi.ajp.161.4.631

Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1993). A longitudinal twin study of personality and major depression in women. Archives of General Psychiatry, 50, 853– 862.

Kessler, R. C., Chui, W. T., Demler, O., & Walters, E. E. (2005). Preva- lence, severity, and comorbidity of 12-month DSM–IV disorders in the National Comorbidity Survey Replication. Archives of General Psychi- atry, 62, 617–709. doi:10.1001/archpsyc.62.6.617

Klein, D. N., Durbin, C. E., & Shankman, S. A. (2009). Personality and mood disorders. In I. H. Gotlib & C. L. Hammen (Eds.), Handbook of depression (2nd ed., pp. 93–112). New York, NY: Guilford Press.

Krueger, R. F. (1999). Personality traits in late adolescence predict mental disorders in early adulthood: A prospective-epidemiological study. Jour- nal of Personality, 67, 39 – 65. doi:10.1111/1467-6494.00047

Liu, R. T., & Alloy, L. B. (2010). Stress generation in depression: A systematic review of the empirical literature and recommendations for future study. Clinical Psychology Review, 30, 582–593. doi:10.1016/ j.cpr.2010.04.010

MacKinnon, D. P., Fritz, M. S., Williams, J., & Lockwood, C. M. (2007). Distribution of the product confidence limits for the indirect effect: Program PRODCLIN. Behavior Research Methods, 39, 384 –389. doi: 10.3758/BF03193007

MacKinnon, D. P., Krull, J. L., & Lockwood, C. M. (2000). Equivalence of the mediation, confounding, and suppression effect. Prevention Sci- ence, 1, 173–181. doi:10.1023/A:1026595011371

MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83–104. doi: 10.1037/1082-989X.7.1.83

MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39, 99 –128. doi:10.1207/ s15327906mbr3901_4

Magnus, K., Diener, E., Fujita, F., & Pavot, W. (1993). Extraversion and neuroticism as predictors of objective life events: A longitudinal anal- ysis. Journal of Personality and Social Psychology, 65, 1046 –1053. doi:10.1037/0022-3514.65.5.1046

McArdle, J. J. (1994). Structural factor analysis experiments with incom- plete data. Multivariate Behavioral Research, 29, 409 – 454. doi: 10.1207/s15327906mbr2904_5

McGonagle, K. A., & Kessler, R. C. (1990). Chronic stress, acute stress, and depressive symptoms. American Journal of Community Psychology, 18, 681–706. doi:10.1007/BF00931237

Mineka, S., Watson, D. W., & Clark, L. A. (1998). Psychopathology: Comorbidity of anxiety and unipolar mood disorders. Annual Reviews of Psychology, 49, 377– 412.

Monroe, S. M., & Reid, M. W. (2009). Life stress and major depression. Current Directions in Psychological Science, 18, 68 –72. doi:10.1111/ j.1467-8721.2009.01611.x

Monroe, S. M., Slavich, G. M., & Georgiades, K. (2009). The social environment and life stress in depression. In I. H. Gotlib, & C. L. Hammen, (Eds.), Handbook of depression (2nd ed., pp. 340 –360). New York, NY: Guilford Press.

Mor, N., Zinbarg, R. E., Craske, M., Mineka, S., Uliaszek, A. A., Griffith, J., . . . Waters, A. (2008). Factor structure and factorial invariance of the EPQ-R Neuroticism Scale in high school juniors. Journal of Personality Assessment, 90, 66 –75. doi:10.1080/00223890701693777

Muthén, L. K., & Muthén, B. O. (2007). Mplus user’s guide (5th ed.). Los Angeles, CA: Muthén & Muthén.

Ormel, J., Oldehinkel, A. J., & Brilman, E. I. (2001). The interplay and etiological continuity of neuroticism, difficulties, and life events in the etiology of major and subsyndromal, first and recurrent depressive episodes in later life. American Journal of Psychiatry, 158, 885– 891.

Reich, J., Noyes, R., Hirschfeld, R., Coryell, W., & O’Gorman, T. (1987). State and personality in depressed and panic patients. American Journal of Psychiatry, 144, 181–187.

Rudolph, K. D. (2008). Developmental influences on interpersonal stress generation in depressed youth. Journal of Abnormal Psychology, 117, 673– 679. doi:10.1037/0021-843X.117.3.673

Rudolph, K. D., Flynn, M., Abaied, J. L., Groot, A., & Thompson, R. (2009). Why is past depression the best predictor of future depression? Stress generation as a mechanism of depression continuity in girls. Journal of Clinical Child and Adolescent Psychology, 38, 473– 485. doi:10.1080/15374410902976296

Rudolph, K. D., Hammen, C., Burge, D., Lindberg, N., Herzberg, D., & Daley, S. E. (2000). Toward an interpersonal life-stress model of de- pression: The developmental context of stress generation. Development and Psychopathology, 12, 215–234. doi:10.1017/S0954579400002066

Saucier, G. (1994). Mini-markers: A brief version of Goldberg’s unipolar Big-Five markers. Journal of Personality Assessment, 63, 506 –516. doi:10.1207/s15327752jpa6303_8

Saudino, K. J., Pederson, N. L., Lichtenstein, P., McClearn, G. E., & Plomin, R. (1997). Can personality explain genetic influences on life events? Journal of Personality and Social Psychology, 72, 196 –206. doi:10.1037/0022-3514.72.1.196

Shih, J. (2006). Sex differences in stress generation: An examination of sociotropy/autonomy, stress, and depressive symptoms. Personality and Social Psychology Bulletin, 32, 434 – 446. doi:10.1177/0146167205282739

Trull, T. J., & Sher, K. J. (1994). Relationship between the five-factor model of personality and Axis I disorders in a nonclinical sample. Journal of Abnormal Psychology, 103, 350 –360. doi:10.1037/0021-843X.103.2.350

Uliaszek, A. A., Hauner, K. K. Y., Zinbarg, R. E., Craske, M. G., Mineka, S., Griffith, J., & Epstein, A. (2009). An examination of content overlap

11STRESS, PERSONALITY, EMOTIONAL DISORDERS

and disorder-specific predictions in the associations of neuroticism with anxiety and depression. Journal of Research in Personality, 43, 785– 794. doi:10.1016/j.jrp.2009.05.009

Uliaszek, A. A., Zinbarg, R. E., Mineka, S., Craske, M. E., Sutton, J. M., Griffith, . . . Hammen, C. (2010). The role of personality in the stress- depression and stress-anxiety relationships. Anxiety, Stress, and Coping, 23, 363–381. doi:10.1080/10615800903377264

Watson, D., Gamez, W., & Simms, L. J. (2005). Basic dimensions of temperament and their relations to anxiety and depression: A symptom- based perspective. Journal of Research in Personality, 39, 46 – 66.

Wheaton, B. (1994). Sampling the stress universe. In W. R. Avison, & I. H. Gotlib (Eds.), Stress and mental health: Contemporary issues and pros- pects for the future (pp. 43–70). New York, NY: Plenum Press.

Winship, C., & Radbill, L. (1994). Sampling weights and regression analysis. Sociological Methods and Research, 23, 230 –257. doi: 10.1177/0049124194023002004

Zinbarg, R. E., Mineka, S., Craske, M. G., Griffith, J. W., Sutton, J., Rose, R. D., . . . Waters, A. M. (2010). The Northwestern-UCLA Youth Emo- tion Project: Associations of cognitive vulnerabilities, neuroticism and gender with past diagnoses of emotional disorders in adolescents. Be- haviour Research and Therapy, 48, 347–358. doi:10.1016/ j.brat.2009.12.008

Zinbarg, R. E., Revelle, W., Yovel, I., & Li, W. (2005). Cronbach’s �, Revelle’s �, and McDonald’s �h: Their relations with each other and two alternative conceptualizations of reliability. Psychometrika, 70, 123–133. doi:10.1007/s11336-003-0974-7

Received March 30, 2010 Revision received August 10, 2011

Accepted August 11, 2011 �

12 ULIASZEK ET AL.