Problems of internal consistency and scaling in life event schedules

Methodological issues in the selection, scaling and summation of items for life event inventories are discussed and illustrated by reference to previously published factor analyses, cluster analyses and multiple regression analyses of life event schedules. Internal consistency is not an appropriate criterion for the construction of evaluation of life event schedules. Theoretically such schedules should have zero order internal consistency. Five methods of weighting events are reviewed. Group-based values may be affected by experience of item rated, whether the rater is a patient, and a number of demographic variables. The use of each respondent's own scale values does not necessarily result in better production of outcome measures. Unweighted event frequency deserves closer attention. However, the use of total Life Change Unit scores from an intuitively derived, albeit carefully scaled, schedule of events risks masking the effects of critical items by irrelevant ones. These problems set limits to the extent to which the use of life event schedules can be profitably refined.