We carried out this study to investigate the mechanisms through which maladaptive cognitions and difficulties in emotion regulation may explain the association between change in life stress and adolescents' emotional and behavioural problems. On the whole, an increase in life stress was associated with negative automatic thoughts, negative cognitive errors, dysfunctional attitudes, expressive suppression and cognitive reappraisal, which were, in turn, associated with emotional and behavioural problems. An examination of the specific indirect effects indicated that only the variable “negative automatic thoughts” was a mediator. No other variable contributed to the indirect effect above and beyond negative automatic thoughts. Examination of all possible pairwise contrasts of the indirect effects showed that the specific indirect effect through negative automatic thoughts was larger than the specific indirect effects through suppression and through dysfunctional attitudes.
The finding that the indirect effect through negative automatic thoughts was larger than that through dysfunctional attitudes broadly reflects the hierarchy of cognitive constructs in the cognitive model of depression. In this model, negative automatic thoughts are reflective of underlying dysfunctional attitudes, and so they are the ‘in the moment’ manifestation of dysfunctional attitudes and the less distal negative cognitive errors. As such, negative automatic thoughts can be expected to be the first cognitive construct to change in response to change in life stress levels. Relatedly, the finding that the indirect effect through negative automatic thoughts was larger than that through suppression probably reflects the temporal ordering of these two responses to stress. Whereas the former is an indication of whether or not a response is triggered, the latter concerns how the response is modulated once it has been triggered (Gross & John, 2003).
Together, these findings have important implications for researchers, policy makers and clinicians working towards identifying youth at risk of emotional and behavioural problems. Our findings, if replicated with experimental data, suggest that an increase in life stress is associated with both emotional and behavioural problems in young people. In addition, our findings suggest that considering specificity at mediator level (McMahon et al., 2003) can improve our understanding of putative psychosocial risk factors, such as life stress, that have associations with a range of common child and adolescent psychiatric symptomatology. Studies, such as ours, testing multiple mediator models can distinguish potential foci for prevention and treatment programmes. If replicated with experimental data, our findings suggest that negative automatic thoughts – rather than maladaptive emotion regulation strategies, dysfunctional attitudes or negative cognitive errors – is the pathway through which an increase in life stress leads to emotional and behavioural problems in adolescents. Therefore, our findings can inform the design and methodology of future targeted prevention efforts.
These conclusions should be evaluated in light of several study limitations. First, our life stress measures, calculated at the sum of adverse life events over a specified period, weighted chronic factors as heavily as one-off adversities. Second, this study, like the well-known Adverse Childhood Experiences Study (Chapman et al., 2004), measured life stress retrospectively. Reporting of adverse events, however, may be related to current state of emotional and behavioural problems due only to colouring of recall by current state (Beck, 2008). Therefore, the relationship between life stress and behavioural/emotional problems is likely stronger when both measures are reported at the same moment in time. Third, although we modelled the effect of maladaptive cognitions in line with theory and evidence (Beck, 2008), we assumed a causal path of associations with cross-sectional data. Prospective longitudinal data are needed to exclude other possibilities, such that an increase in life stress predicts an increase in emotional and behavioural problems, which then predicts maladaptive cognitions, or – in line with findings supporting stress generation models of adolescent psychopathology (Carter et al., 2006; Rudolph & Klein, 2009) – that an increase in life stress is not the contextual risk factor, but the contextual outcome of (an increase in) psychiatric symptomatology, maladaptive cognitions or both. Fourth, our findings may not be generalisable to the UK adolescent population at large. Our data were collected from one state secondary school in Greater London, which was assessed by the Office for Standards in Education (OFSTED) as ‘outstanding.’ According to the OFSTED report, 78% of the students achieved five or more good GCSE grades, which is above the national average. Most students came from the relatively affluent local community, and more than half had an Indian heritage, with Pakistani, white British and African Caribbean students making up significant minority groups. Our study sample broadly reflected the school's student population. Thus, our findings may not be relevant to adolescents from other backgrounds, such as those from less prosperous communities. Future research is needed to determine to what extent our findings can be generalised to the UK adolescent population. Fifth, we acknowledge that our sample size, even after multiple imputation of missing data, was small. However, the bootstrapping method that we used to test for mediation does not impose the assumption of normality of the sampling distribution (Preacher & Hayes, 2008b). Finally, we acknowledge that our findings may benefit from comparison with those from studies on older samples. For example, as the association between emotion regulation and mental health is weaker in children than in adults (Aldao et al., 2010), the nonsignificant mediational path from change in life stress to emotional and behavioural problems via difficulties in emotion regulation may not be replicated in adults. Furthermore, as there is evidence that maladaptive cognitions moderate rather than mediate the association between life stress and psychiatric symptomatology in adults, we may find no evidence for mediation by maladaptive cognitions in adults.