A Graphical Method for the Interpretation of Multinomial Logit Analysis

Abstract

Multinomial logit analysis is a statistical technique for relating a set of continuous or discrete independent variables to a categorical dependent variable. The availability of statistical software and the publication of review articles dealing with logit analysis has resulted in increasing applications of the model. These applications have been limited by a failure to deal adequately with the magnitudes of the parameters, restricting discussion almost entirely to the direction of effects and their statistical significance. This limitation is a consequence of the large number of parameters involved and the nonlinearity of the effects. To deal with this problem, a graphical method is proposed to represent the magnitudes and statistical significance of all effects in a multinomial logit analysis. This allows for a clear interpretation of the relative magnitudes of effects both within and across independent variables. The article concludes with an example to illustrate the technique.

Get full access to this article

View all access options for this article.

References

ALDRICH, J. H. and F. D. NELSON (1984) Linear Probability, Logit and Probit Models. Newbury Park, CA: Sage.

AMEMIYA, T. (1981) “Qualitative response models: a survey.” J. of Econ. Literature 19: 483-536.

AMEMIYA, T. (1985) Advanced Econometrics. Cambridge, MA: Harvard Univ. Press.

BAKER, R. J. and J. A. NELDER (1978) The GLIM System. Release 3. Oxford, England; Numerical Algorithms Group.

DAVIS, J. A. and T. M. SMITH (1984) General Social Survey, 1972-1984. Chicago: National Opinion Research Center.

EDLEFSEN, L. E. and S. D. JONES (1984) Reference Guide to GAUSS. Kent, WA: Applied Technical Systems.

GREENE, W. H. (1986) LIMDEP User's Manual. Department of Business Administration, New York University.

HANUSHEK, E. A. and J. E. JACKSON (1977) Statistical Methods for Social Scientists. New York: Academic Press.

MADALLA, G. S. (1983) Limited-Dependent and Qualitative Variables in Econometrics. New York: Cambridge Univ. Press.

10.

NELSON, F. D. (1978) Qualitative and Limited Dependent Variable Regression Models: Program Documentation. Pasadena, CA: Division of Humanities and Social Sciences, California Institute of Technology.

11.

PINDYCK, R. S. and D. L. RUBINFELD (1981) Econometric Models and Economic Forecasts. New York: McGraw-Hill.

12.

Salford Systems (n.d.) SAS Binary Procedures: PROC MLOGIT. 2214 El Amigo Road, Del Mar, CA 92014.

13.

SAS Institute, Inc. (1982) SAS User's Guide: Statistics. Cary, NC: author.

14.

SCHMIDT, P. and R. P. STRAUSS (1975) “The predictions of occupation using multiple logit models.” Int. Econ. Rev. 16: 471-486.