Abstract
When considering the effect of an independent variable in a statistical analysis, one logical set of statistics to consult would be the regression weights. However, if the independent/predictor variables are correlated, the estimates for the multiplicative weights can no longer be used alone in formulating interpretations. Although many techniques have been suggested to help in these situations, structure coefficients, or the correlations between predictor variables and the synthetic variable, are the method of choice. Yet in multiple regression analysis, the utility of structure coefficients has not been recognized. This thesis examined the interpretation of multiple regression models in published research and the advantage of structure coefficients as an aid in interpretation of these models.
Courville, Troy Gerard (1999). Use of structure coefficients in published reports of regression analysis. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1999 -THESIS -C70.