Normal variation and abnormality: An empirical study of the liability distributions underlying depression and deliquency

Background: Scale scores in studies of emotional and behavioural problems often possess highly skewed distributions. The long upper tails of these distributions place a small proportion of the population at some distance from the main body of the distribution. This invites an interpretation of their forming an abnormal group, one that may be qualitatively distinct. Methods: Item-response models were fitted to data on parent and self-rated depression and delinquency from four large samples of children or adolescents. Results: We found that underlying liability distributions show very little or no evidence of non-normality. Conclusions: The results suggest that (i) the skewed nature of the scale scores may be largely measurement artefacts, (ii) the distributions provide no evidence of a qualitatively distinct process generating abnormality as compared to normal variation and (iii) for characterising the whole distribution, including normality and abnormality, the selection of items in typical current assessments of emotional and behavioural problems is not optimal.