Communications for Statistical Applications and Methods

The Korean Statistical Society (한국통계학회)
권호 표지

Multicollinarity in Logistic Regression

  • Jong-Han lee (Daewoo Research Institute, Yoido-dong 34-3, Seoul 150-010, KOREA) ;
  • Myung-Hoe Huh (Dept. of Statistic, Korea Unicersity, Anam-dong 5-1, Seoul 136-701, KOREA)
  • Published : 1995.12.01
Download PDF
⟨ Previous Next ⟩

Abstract

Many measures to detect multicollinearity in linear regression have been proposed in statistics and numerical analysis literature. Among them, condition number and variance inflation factor(VIF) are most popular. In this study, we give new interpretations of condition number and VIF in linear regression, using geometry on the explanatory space. In the same line, we derive natural measures of condition number and VIF for logistic regression. These computer intensive measures can be easily extended to evaluate multicollinearity in generalized linear models.

Keywords

References

  1. Regression Diagnostics: Identifying Influencial Data and Sources of Collinearity Belsley,D.A.;Kuh,E.;R.E.Welsh
  2. Computer Programs in Biomedicine A Computer Program for Linear Logistic Regression Analysis Lee,E.T.
  3. Unpublished Ph.D.Thesis Multicollinearity in Logistic Regression Lee,J.H.
  4. Communications in Statistics-Theory and Methods v.18 no.9 Collinearity on Generalized Linear Models Mackinon,M.J.;Puterman,M.L.
  5. Generalized Linear Models(2nd ed.) McCullagh,P.;Nelder,J.A,
  6. SAS User's Guide: Statistics, Version 6(4th ed.) SAS Institute
  7. Journal of Applied Statistics v.19 On the Conditioning Problem in Generalized Linear Models Segerstebt,B.;Nyquist,H.
  8. Statistical Science v.2 Collinearity and Least Squares Regression Stewart,G.W.
  9. Communications in Statistics-Theory and Methods v.20 no.4 A Multicollinearity Diagnostics for Generalized Linear Models Weissfeld,L.A.;Sereika,S.M.