ERIC Number: ED330725
Record Type: RIE
Publication Date: 1991-Apr
Reference Count: N/A
The Importance of Structure Coefficients in Parametric Analyses.
Friedrich, Katherine R.
The recognition that all parametric methods are interrelated, coupled with the notion that structure coefficients are often vital in factor and canonical analyses, suggests that structure coefficients may be important in univariate analysis as well. Using a small, heuristic data set, this paper discusses the importance of structure coefficients for the univariate case and then extends this discussion to all parametric procedures. More specifically, a multiple regression analysis, with three independent variables, was conducted. Examination of the intercorrelations between predictors indicated that there was some degree of collinearity. Beta weights, structure coefficients, and other parametric procedures were used when appropriate. Regression coefficients and structure coefficients represent two different types of information. It is suggested that serious misinterpretations of results can occur if structure coefficients are not used in some analyses. The Statistical Analysis System program used to analyze the heuristic data involved in this study, three tables, and a 19-item list of references are included. (TJH)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Identifiers: Collinearity; Parametric Analysis; Structure Coefficients; Univariate Analysis
Note: Paper presented at the Annual Meeting of the American Educational Research Association (Chicago, IL, April 3-7, 1991).