Extracting more information from behaviour checklists by using components of mean based scores

Sums of responses to behaviour checklist items are commonly used as outcome measures. We argue for the use of mean scores. For sets of responses registering absence and presence at different levels of intensity of behaviours we also show that mean scores may usefully be 'decomposed' into separate measures of the range and the intensity of problematic behaviours. These separate measures are the proportion of items positively endorsed and the 'intensity index' - the proportion of positive scores that are above one. We illustrate their use with primary outcome scores from the Developmental Behaviour Checklist (DBC) in the Australian Child to Adult Development Study. The low mean scores of young people with profound intellectual disability are shown to be a function of the narrow range of behaviours they display rather than of the level of intensity of these behaviours, which is relatively high. Change over time in mean scores is shown to be attributable to change in both the range and the intensity of behaviours as young people age in the study. We show how the technique of measuring these two separate strands contributing to mean scores may be applied to checklists with sets of responses longer than the zero, one, two of the DBC. Copyright (C) 2008 John Wiley & Sons, Ltd.