The Sonoma Research Project: Data Analysis Design and Results

Abstract

The major purpose of the Sonoma Research Project was to generate data pertaining to the self-initiated nonlinguistic communication performance of individuals with profound handicaps. After data were collected from the taped segments, they were submitted to various statistical procedures from the Statistical Package for the Social Sciences (Norusis, 1985). For the majority of the statistical analyses, delineations were made between classes of independent and dependent variables. The variable classes of communicative mode, request function, feeling function, and comment function comprised those variable types that were considered independent; need, routine, group, referent, referent proximity, prompt, prompt proximity, and consequence type and action were treated as independent variables. In analyses dealing with prediction (multiple regression), the independent variables were considered predictor variables and the dependent variables considered criterion variables. The data were examined for relationships between independent and dependent variable classes, between independent variable classes and individual types of each dependent variable class, and for the value of independent variable classes for predicting levels of dependent variable types. Data were also analyzed for relationships among dependent and independent variable classes. All data entries were nominal in nature, with each communicative instance treated as an n of 1 for the purposes of these analyses. As a result, 1,954 observations (minus any missing values) were included in the analyses, with these observations generating over 21,400 data entries.