Response class: a Guttman scale analysis.

Abstract

This study addressed the problem of response class as it relates to coercive child behavior. Guttman scale analyses were used to determine if there were regular, unidimensional progressions across eight coercive response categories; that children who performed high probability coercive behaviors would also tend to perform all of the lower probability coercive behaviors. Rank-order correlations were done to determine if there was consistency in the response class across settings. Two samples of boys were observed in two school settings, classroom and playground. One sample was a derivation sample of 26 boys, the second a replication sample of 27 boys. Reproducibility coefficients for the derivation sample were .92 for the classroom Guttman scale analysis and .93 for the playground analysis. For the cross-validation sample the reproducibility coefficients were .94 in both settings. Rank-order correlations across settings and done in terms of the number of subjects performing each of the coercive responses were .971 (df = 6, p < .001) for the derivation sample and .996 (df = 6, p < .001) for the cross-validation sample. It was concluded that the eight coercive responses observed appeared to constitute a response class, i.e., the ordering the responses was transitive across subjects, and that the coercive response hierarchy was stable across school settings for groups of children.