Considerations, Issues and Comparisons in Variable Selection and Interpretation in Multiple Regression

Abstract

The selection of independent variables when utilizing multiple linear regression in a study is an involved and complex process. The availability of a variety of computer programs usually referred to as stepwise procedures affords users numerous options about which they often have little understanding. The purpose of this paper, then, is twofold: first, to present the major uses of regression analyses, the advantages and disadvantages of selection procedures and some caveats for researchers and those who teach statistics, and secondly, to present, compare and contrast several variable selection techniques using two data sets.