USING LOG-LINEAR TESTS OF FIT FOR A TRANSACTIONAL SYSTEMS MODEL

Publisher Website
Google Scholar
Abstract
For the past decade the Interact System Model (ISM) of Fisher and Hawes (1971) has been used in conjunction with traditional statistical tests of fit of discrete-time, discrete-state Markov models to examine small group processes of communication. The authors argue that the ISM is an inappropriate model of interpersonal and noninterpersonal communication processes, in part because it takes no account of nonverbal messages. As an alternative, they propose a Transactional Systems Model (TSM) of person-to-person communication, and, using data on police-complainant interaction for illustration, show how log-linear statistical models provide an efficient and flexible means of testing and comparing ISM with TSM models. Log-linear tests may replace traditional tests of such models and open the way to a greater understanding of complex communication processes.

This publication has 19 references indexed in Scilit: