Identifying dimensions of perceived patient change: some problems in the use of factor analysis.

A study of change in a 'therapeutic community' admission unit is used to illustrate some of the problems associated with the use of factor analysis in evaluating therapeutic outcome. The paper describes the methodological steps involved in an attempt to identify the underlying dimensions of change which reflect how patients see themselves, and are seen by those most closely involved with them. The resulting dimensions, labelled 'pleasantness' and 'illness', are used to illustrate change at three levels: average direction, scatter, and individual cases.