PACE: D2A Archive

In an earlier version of the ACE-indicator maps, item groupings were based on thematic similarities between items. This procedure resulted in composite indicators which were composed of a small number of items. While the resulting estimates were computed correctly, they appeared to underestimate the prevalence of ACE-indicators across counties. In the updated maps, item groupings were determined empirically using principal components analysis (PCA). PCA is a procedure which identifies groups of items which respondents tended to answer in similar ways (e.g. students who reported they had been "picked on" or "teased" were also more likely to report that they had been "bullied"). Once groupings have been identified, PCA generates composites based on an optimally weighted sum of the individual items. This procedure also identified groupings of items which can be described as protective against the effect of ACEs. Thus, in the updated maps, composites are based on empirically defined groupings of items and include protective factors as well as risk factors.