Factorial validity of the EE scales

Abstract

SynopsisHigh ‘expressed emotion’ (EE) has been held to be predictive of a poor course in patients with schizophrenia, a finding that has emphasized the predictive validity of the EE measure. We review appropriate research examining properties of the EE measure to suggest that, in particular, little is known about the construct validity of the EE measure. We report a longitudinal study examining three measures of parental style (EE; Parental Bonding Instrument or PBI; and interviewer ratings). By comparing EE scale scores with ratings derived by an interviewer we find support for the accuracy of the process generating EE scale scores, and by factor analytic techniques we examine the factorial (construct) validity of the EE scales. A principal components analysis suggested two other dimensions of relevance in parents involved with a schizophrenic member (vigilance, and acceptance of illness in their child) which may be of value in predictive studies. We examined the capacity of interviewer-generated scores (as a simple clinical rating) and factor scores (as central constructs) to predict the course, and were unable to discriminate ‘relapsers’ and ‘nonrelapsers’ with either predictor.