How much stressful life events really matter? Conceptual and methodological difficulties in assessing the impact of self-reported events on adolescents' subjective stress.

Numerous studies have pointed to the strong impact of stressful life events on subjective stress and indirectly on the adolescents' mental health. However, the results of many such studies contain bias caused by the choice of measurement method or an incomplete theoretical framework. To estimate the extent of possible bias when using self-reports, we conducted research on a representative sample of 2201 Croatian adolescents aged 14-18 years. Using manipulation we examined the influence of measurement method and the number of constructs included in the model on the variance of subjective stress explained by stressful events. Structural equation modelling indicated that different measurement methods, occurrence-based and severity-based, provide a marked discrepancy in the impact size estimation. The occurrence-based method provided estimates of a much smaller proportion of the explained variance (6%-25%) compared to the severity-based approach, which estimated the explained variance in the range of 25%-55% depending on the model. The complexity of the tested models additionally contributed to the variation in the estimates. It is recommended that when using self-reports to measure stressful life events, occurrence-based and severity-based methods are used simultaneously and that impact estimates are expressed within intervals.