Previous research on the cognitive vulnerability model to depression among youth has been based largely on cross-sectional designs with relatively small and often clinically derived samples. This research has yielded inconsistent results with respect to specificity. We used longitudinal data weighted to reflect a public middle school population to examine bidirectional associations and to test for specificity of attributional style, self-worth, and rejection sensitivity as they are associated with depressive versus conduct and anxiety problems during early adolescence. Our findings suggest that in general the temporal pathways between these cognitive features and psychopathology symptoms operate in both directions, although the influences in most cases were not symmetrical. In addition to vulnerabilities, cognitive scars emerged from the experience of elevated symptomatology. Moreover, both specific and general effects emerged for the cognitive features examined.
Contrary to the hopelessness theory, attributional style was not associated with vulnerability to depressive symptoms among adolescents but rather emerged as a scar effect, residual to the experience of elevated symptoms. The experience of depressed mood appears to precede negative attributions among young adolescents. These findings are compatible with other studies that have found that elevated depressive symptoms contribute to the development of negative attributional style in children (Bennett & Bates, 1995
; Gibb et al., 2006
; Nolen-Hoeksema et al., 1992
). However, our results also indicate that the scar effects of depression on attributional style were not unique to depression but were also common to conduct problems. In a recent study, Gibb and colleagues (2006)
found that negative attributional style accounted for some of the relation between two types of conduct problems (delinquency and oppositionality) with depressed mood. Because their study was cross-sectional, the authors were unable to tease apart the nature of these associations. Although our study is limited by the use of only two time points, these results suggest that the pathway may run from behavior to cognitions, with youth who display more conduct problems subsequently demonstrating pessimistic attributions, perhaps because their behavior generates negative outcomes. Thus, negative attributions may result from both depressive and conduct problems. Future research should examine the degree to which such attributions reflect a perceptual bias versus a realistic appraisal.
Although low self-worth created vulnerability to greater depression over the course of a year among adolescents, it was also bidirectionally associated with increased conduct problems. This could indicate that low self-worth is a “core process” shared between these two forms of psychopathology; that is, low self-worth could be a general risk that creates risk toward broad presentations of psychopathology. A recent debate has questioned whether youth with conduct problems have negative or overly positive views of themselves. Our study suggests that there is a negative self-evaluation component for adolescents with elevated conduct problems, largely consistent with a recent series of studies finding robust associations between low self-esteem and problem behavior among adolescents and young adults (Donnellan et al., 2005
). It may be noteworthy that several of the studies that have indicated inflated self-appraisals among youth with externalizing problems were conducted with elementary school age children who are more prone to positive cognitive biases (Hoza, Waschbusch, Pelham, Molina, & Milich, 2000
; Hughes, Cavell, & Grossman, 1997
). Thus, one possible explanation for the inconsistency within the self-esteem literature that warrants further investigation is that there may be developmental shifts in self-appraisals for youth with externalizing behavior occurring between late childhood and early adolescence.
Our findings are also compatible with the notion that a negative self-image could function as a common link between the emergence of conduct and co-occurring depressive symptoms, which typically manifest subsequent to conduct problems (Biederman, Faraone, Mick, & Lelon, 1995
). These results are compelling in suggesting further exploration into cascade models of psychopathology, which posit that poor functioning in one domain of behavior can snowball into problems within other domains (Burke, Loeber, Lahey, & Rathouz, 2005
; Masten et al., 2005
). The “failure model” proposed by Patterson and Capaldi (1990)
, for example, suggested that early conduct problems result in failure experiences that in turn create vulnerability to subsequent depression. Our research suggests that another consequence or “failure” created by conduct problems may be a shift toward a more negative view of the self. Future research should explore the role of both cognitions and consequences of failure experiences in understanding sequential comorbidity between conduct problems and depression.
The third cognitive feature examined, rejection sensitivity, was chosen based on cognitive interpersonal models of depression that suggest that depressed persons have negative expectations and experiences in their interactions with others. Rejection sensitivity created vulnerability only to anxiety symptoms but emerged as a cognitive scar of depression, such that having elevated depression was associated with a subsequent propensity to perceive and expect rejection in ambiguous situations and to react with anxious affect. These results support findings linking rejection sensitivity to loneliness and social anxiety (Downey, Bonica, London, & Paltin, in press
), perhaps because of the nervous affect associated with anticipating rejection. The scar effect found here is also consistent with studies finding that depressive symptoms predict social self-perceptions (Kistner et al., 2006
; Rohde et al., 1990
). The interactional framework for depression suggests that the behaviors of depressed individuals lead them to get rejected by others (Coyne, 1976
). It is possible, then, that youth who have elevated depressive symptoms are more sensitive to rejection due to the scarring effects of actually having experienced rejection more often or due to a social information processing residue that remains following a depressed mood.
Collectively, the findings advanced in this study are consistent with the mutual interplay of cognitive features with youth psychopathology, including particular cognitive features as risks for later psychopathology as well as the emergence of cognitive biases as a result of experiencing psychopathology. In terms of informing our theories of etiology, this research does not suggest that cognitive features uniquely contribute to a pathogenic process for depression. Rather, they suggest that negative evaluations of the self, events, and interpersonal situations emerge in tandem with varied presentations of psychopathology among youth. Several clinical implications of this research warrant further investigation. First, the results show that low self-worth in particular creates vulnerability toward depressive symptoms and conduct problems during the transition to adolescence. Thus, youth who have poor self-esteem may be an appropriate target group for indicated preventive intervention programs. The results also show that rejection sensitivity can result specifically from the experience of depression and would suggest that it would be beneficial to specifically address rejection (both real and perceived) as a component of intervening with depressed youth. Two of the more efficacious genres of treatment for adolescent depression, interpersonal psychotherapy and cognitive–behavioral therapy, can address rejection as an intervention goal, via the interpersonal focus of interpersonal psychotherapy and via the cognitive focus in cognitive–behavioral therapy. Application of a combined cognitive–interpersonal theoretical framework is potentially useful in intervening with depression in adolescents. A third clinical implication is that some focus on cognitions for youth with conduct problems may be warranted, given the associations found in this study. For example, youth with conduct problems may benefit from learning to reassess and reinterpret events in their lives, as well as from learning strategies for exerting control that is appropriate to the situation (Compas, Connor-Smith, Saltzman, Thomsen, & Wadsworth, 2001
), instead of feeling helpless, relinquishing control, and experiencing numerous failures.
Because of several limitations of this study, broad conclusions must be tempered. Although the magnitude of effects was generally small, the findings are enhanced by the use of cross-lagged analyses that control for synchronous relations and stability of measures and that include depression, conduct problems, and anxiety simultaneously. The interval used for the prospective analyses was 1 year during early adolescence. A different interval during a different development period may have yielded different findings. Moreover, although having longitudinal data represents a step forward in understanding the role of cognitive vulnerability and psychopathology, we are limited by the use of cross-lagged panel analyses using only two time points. To facilitate a better understanding of directionality, research examining these constructs over a longer course of time and with more frequent assessments is needed. Likewise, it is unknown at this time to what extent relations between cognitive features and psychopathology are consistent throughout development. We limited our models to cognitive factors, excluding other important risk factors that clearly play a role in the development of youth psychopathology and may interact with cognitive factors, such as biological risk, family environment, and peer interactions. Moreover, our examination of cognitive features was not intended to be fully exhaustive; rather, we strove to include salient constructs that are represented in major cognitive and cognitive-interpersonal theories, but in doing so we have omitted other dimensions of cognition, such as automatic thought content and rumination. Finally, the Children’s Attributional Style Questionnaire–Revised has been criticized for its generally low internal consistency reliability (Hankin & Abramson, 2002a, 2002b
), which may have attenuated the strength of the associations between cognitive style and emotional health conditions. Psychometrically stronger alternatives to assessing attributional style that are appropriate for this age range have only become available since this study began (e.g., Conley, 2001
; Mezulis et al., in press
) and should be considered in future research.
A number of strengths of the study warrant consideration in assessing the contribution of this research. First, we incorporated into our models various dimensions of cognitive style drawn from hopelessness, Beckian, and cognitive interpersonal theories, including attributional style, self-worth, and rejection sensitivity, dimensions not previously examined together. We also modeled and examined three forms of psychopathology to control for co-occurring symptoms. We used a longitudinal design, controlling for baseline levels of earlier symptoms, which allowed us to examine change over time instead of cross-sectional associations and contributed to our ability to disentangle vulnerability from scar effects. Finally, we utilized a school-based sample, weighted to generalize to a large, racially diverse public middle school population. The data support relations in both directions between cognitive features and youth psychopathology and demonstrate that various cognitive features are pertinent not only to depression but also to other forms of psychopathology. To advance the field to a broader understanding of cognitive processes in the development of psychopathology, further research, and an expanded longitudinal view of the interplay between cognitions and symptomatology throughout childhood, adolescence, and adulthood is needed. This understanding will be helpful in developing and refining effective prevention and intervention strategies.