The standardized effect of a compound of dummy variables or polynomial terms

A method is proposed to obtain standardized regression coefficients for composite variables made up of dummy variables or polynomial terms. The method to be described enables the researcher to compare the effect of the composite variable with the effects of other predictor variables. Forming a composite variable is particularly useful in polynomial regression where individual regression coefficients are hard to interpret. A second application is assessing the impact of a compound of dummy variables. An empirical example dealing with the curvilinear relationship between church involvement and prejudice is used to illustrate the approach.