NotesFAQContact Us
Search Tips
ERIC Number: ED464932
Record Type: Non-Journal
Publication Date: 2002-Apr-2
Pages: 39
Abstractor: N/A
Reference Count: N/A
Stepwise Regression as an Exploratory Data Analysis Procedure.
Thayer, Jerome D.
This paper identifies specific problems with stepwise regression, notes criticisms of stepwise methods by statisticians, suggests appropriate ways in which stepwise procedures can be used, and gives examples of how this can be done. Although the stepwise method has been routinely criticized by statisticians, it is still frequently used in the literature. This paper suggests research situations when stepwise regression may have a valuable function. Stepwise methods can be appropriate for variable evaluation. Since the value of a variable as a predictor is highly specific to the other variables in the prediction model, the use of stepwise methods can provide many reduced models in which the characteristics of the variable can be examined. As variables are found to be good predictors in different models, the different prediction characteristics of the variables in the various models can be used to recognize how the variables function as predictors and can be used to develop a theory or models that can be tested with further research. In order for stepwise methods to be used effectively, they should be used in conjunction with a best subsets procedure and zero-order correlations, default criterion values should be modified, models should not be selected by the computer, and, where possible, models should be generated from multiple subsets of the data. (Contains 21 tables and 17 references.) (SLD)
Publication Type: Numerical/Quantitative Data; Opinion Papers; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A