Configurations of Therapeutic Change
Analyses of variance and covariance as well as matched t tests were used to evaluate overall group changes in clinical case records and psychological test protocols. Within these overall group comparisons at Time 1 and Time 2, however, different individuals may change to varying degrees and in different directions on separate variables. While analyses of variance and t tests assess group effects over time, they evaluate only the relationship of change in different variables across individuals that are consistent within the total group. To the extent that different individuals may change to varying degrees and in different directions on separate variables, these different effects may cancel each other out and not be evident in various analyses of group effects. To examine possible patterns of change in different spheres of functioning within individuals, we correlated change on psychological test variables with change on the ratings of the clinical case records. In evaluating these correlations (or covariations) between change scores in different sets of variables, we sought to identify patterns of change as individuals progress or regress during the treatment process. Independent of the changes noted for total groups on separate variables, the correlation of change scores among different variables from independent sources permits us to examine whether there are particular configurations of variables that indicate both progressive and regressive change for different types of patients.
KeywordsChange Score Human Response Case Record Interpersonal Relation Separate Variable
Unable to display preview. Download preview PDF.
- 1.The correlation of change scores is a simple and helpful screening device for the preliminary exploration of relationships among variables in a longitudinal data set. Caution needs to be exercised, however, in the interpretation of correlated change scores because they can be misleading in a number of ways. It is possible for the relationship between change scores of two variables to be spurious. If, for example, patients were selected for inclusion in the study (i.e., hospitalized) because of their impaired psychological functioning and behavior, then all measures, because they are likely to be exaggerated at first testing, are subject to regression toward the mean (i.e., to move toward more modulated scores). This statistical artifact could generate spurious correlations among change scores. Similarly, two measures that are both influenced by an unidentified and unmeasured third factor could result in a spurious correlation of change scores. Also, the synchronous correlation between two variables at either Time 1 or Time 2 can generate significance in the correlations of change scores. Despite these cautions, correlation of change scores provides an opportunity to assess patterns of change within individuals as they progress and regress during treatment.Google Scholar
- 2.This relationship between change in OR—scores and change in social behavior in anaclitic patients occurs with both the developmental index and the developmental mean of inaccurately perceived human representations, even though these two variables are only marginally interrelated. The developmental mean and the developmental index of OR+ correlate. 38 and. 22, and these two measures of OR—correlate. 68 and. 29, for anaclitic and introjective patients, respectively.Google Scholar