Use of Stepwise Methodology in Discriminant Analysis.

The use of stepwise methodologies has been sharply criticized by several researchers, yet their popularity, especially in educational and psychological research, continues unabated. Stepwise methods have been considered particularly well suited for use in regression and discriminant analyses, but their use in discriminant analysis (predictive discriminant analysis and descriptive discriminant analysis) has not been the direct focus of as much written commentary. This discussion considers several problems associated with the use of stepwise methods. These include: (1) incorrect degrees of freedom associated calculated by computer packages; (2) sampling error capitalization; and (3) failure to select the best subset of predictors. Although they offer the promise of assisting researchers with tasks such as variable selection and variable ordering, this promise is almost always unfulfilled. Some alternatives to the use of stepwise methods are discussed, including manually correcting degrees of freedom, crossvalidation procedures, and allpossible subsets analyses. (Contains 6 tables and 45 references.) (Author/SLD) *********************************************************************** Reproductions supplied by EDRS are the best that can be made from the original document. *********************************************************************** Stepwise Methodology 1 Running head: STEPWISE METHODOLOGY USE IN DISCRIMINANT ANALYSIS U.S. DEPARTMENT OF EDUCATION Office of Educational Research and Improvement EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC) his document has been reproduced as received from the person or organization originating it. Minor changes have been made to improve reproduction quality. Points of view or opinions stated in this document do not necessarily represent official OERI position or policy. PERMISSION TO REPRODUCE AND DISSEMINATE THIS MATERIAL HAS BEEN GRANTED BY '3-676Ai WHirnk TO THE EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC) Use of Stepwise Methodology in Discriminant Analysis Jean S. Whitaker Texas A&M University 77843-4225

[1]  Bruce Thompson.  Common Methodology Mistakes in Dissertations, Revisited. , 1994 .

[2]  Jacob Cohen.  Multiple regression as a general data-analytic system. , 1968 .

[3]  James D. Moore.  Stepwise Methods Are as Bad in Discriminant Analysis as They Are Anywhere Else , 1996 .

[4]  Bruce Thompson,et al.  Why Multivariate Methods Are Usually Vital in Research: Some Basic Concepts. , 1994 .

[5]  John Bowling.  The Importance of Structure Coefficients as against Beta Weights: Comments with Examples from the Counseling Psychology Literature. , 1993 .

[6]  J. Stevens.  Applied Multivariate Statistics for the Social Sciences , 1986 .

[7]  Bruce Thompson,et al.  The pivotal role of replication in psychological research , 1994 .

[8]  Mike Baxter,et al.  Stepwise Discriminant Analysis in Archaeometry: a Critique , 1994 .

[9]  B. Thompson.  Canonical Correlation Analysis: Uses and Interpretation , 1984 .

[10]  James P. Shaver,et al.  What Statistical Significance Testing Is, and What It Is Not , 1993 .

[11]  Bruce Thompson,et al.  Why Won't Stepwise Methods Die? , 1989 .

[12]  Carl J. Huberty,et al.  An Introduction to Discriminant Analysis. , 1989 .

[13]  R. Fisher.  THE PRECISION OF DISCRIMINANT FUNCTIONS , 1940 .

[14]  L. Cronbach.  Beyond the Two Disciplines of Scientific Psychology. , 1975 .

[15]  Sam Kash Kachigan.  Statistical Analysis: An Interdisciplinary Introduction to Univariate & Multivariate Methods , 1986 .

[16]  Bruce Thompson,et al.  Interpreting Regression Results: beta Weights and Structure Coefficients are Both Important. , 1992 .

[17]  J. Elashoff,et al.  Multiple Regression in Behavioral Research. , 1974 .

[18]  Walter R. Borg,et al.  Educational research: An introduction, 6th ed. , 1996 .

[19]  Bruce Thompson,et al.  Canonical Correlation Analysis: An Explanation with Comments on Correct Practice. , 1988 .

[20]  Jacob Cohen,et al.  THINGS I HAVE LEARNED (SO FAR) , 1990 .

[21]  Bruce Thompson,et al.  The Importance of Structure Coefficients in Regression Research , 1985 .

[22]  R. P. Carver.  The Case Against Statistical Significance Testing , 1978 .

[23]  M. Fireman,et al.  MULTIPLE REGRESSION ANALYSIS OF SOIL DATA , 1954 .

[24]  P. Meehl.  Why Summaries of Research on Psychological Theories are Often Uninterpretable , 1990 .

[25]  T. R. Knapp.  Canonical correlation analysis: A general parametric significance-testing system. , 1978 .

[26]  Bruce Thompson,et al.  The Use of Statistical Significance Tests in Research: Bootstrap and Other Alternatives , 1993 .

[27]  Bruce Thompson,et al.  Stepwise Regression and Stepwise Discriminant Analysis Need Not Apply here: A Guidelines Editorial , 1995 .

[28]  C. J. Huberty,et al.  Applied Discriminant Analysis , 1994 .

[29]  C. J. Huberty,et al.  Issues in the use and interpretation of discriminant analysis , 1984 .

[30]  N. Draper,et al.  Applied Regression Analysis. , 1967 .

[31]  Jacob Cohen.  The earth is round (p < .05) , 1994 .