Interpreting data in aetiological studies of affective disorder: some pitfalls and ambiguities.

Although recent research in social psychiatry has produced an encouraging congruence of findings and conclusions, puzzling inconsistencies continue to be reported. One explanation which is often overlooked is that subtle differences in the way seemingly identical variables are grouped can produce sizeable, and sometimes dramatic, differences in the patterning of the same data. Re-examination of existing results can therefore often better clarify confusing inconsistencies than collection of new data. This is illustrated by examples from recent studies of affective disorder where the grouping of variables is discussed in three broad areas: parental loss in childhood, precipitating stress, and social support. Until the aetiology of affective disorder is more fully understood, it will often be clearer if data are analysed more than once, so that several combinations of variables are systematically examined.