“…For example, the inappropriate use of factor analytic techniques with skewed/binary items, a misinterpretation of Cronbach's a to indicate unidimensionality, the use of simple correlations to decide whether constructs are 'moderately' related, or a choice of scales which have no overlapping symptoms. 5,10,17,21 In the present study, internal consistencies were very good for the original depression and vital exhaustion scales, which then correlated at a level which indicated only 18% shared variance. Such correlational results, however, can simply be indicative of different levels of severity of a single trait -as borne out by the subsequent Mokken results presented herein.…”