| Sumario: | The article focuses on qualitative variables that are used in regression analysis. The behavioral sciences such as education, sociology, psychology and political science generate research data that is often qualitative in nature. The usage of qualitative variables in regression analysis has largely been handled by ordering the categories into a pseudo-quantitative variable. This procedure is not satisfactory, either in terms of obtaining a realistic model or in terms of interpreting the results. A general least squares approach provides a sound model and the results of the analysis are easy to interpret. The model used is a factorial model with co-variates. The continuous variables are the co-variates and the qualitative variables are the main effects of the factors. Using design variables and writing the continuous variables first, the model may be written as a multiple regression model. The article author has written a group of multiple regression computer programs which includes the capabilities of the foregoing theory. The generation of a dummy variable is controlled by specifying the original variable to be used, the number of categories of the variable and a vector of the codes used.
|
|---|