SELECTING INDEPENDENT VARIABLES

SELECTING INDEPENDENT VARIABLES

Often you do not know which independent variables together are good predictors of the dependent variable. You want to eliminate variables that are of little use from your equation so you will have a simple, easy-to-interpret model. You can do this with the assistance of what are called variable selection methods . Based on statistical considerations, such as the percent of variance explained by one variable that is not explained by any other variables, some software packages (for example SPSS) select a set of variables for inclusion in a regression model. Although such procedures often result in a useful model, the selected model is not necessarily best in any absolute sense.