Interpreting multinomial logistic regression
Abstract
Social and biological scientists widely use logit (logistic) regression to model binary dependent variables such as move/stay or live/die. Techniques for modeling multiple-category dependent variables are a relatively recent development, however. Asking Stata to perform multinomial logistic regression is easy; given a Y with three or more unordered categories, predicted by X1 and X2, you type ‘mlogit Y X1 X2’. If Y has only two categories, mlogit fits the same model as logit or logistic. Otherwise, though, an mlogit model is more complex. This insert, a sort of “beginners guide to multinomial logit” written while stormbound at the Nullagvik Hotel, illustrates several ways to interpret mlogit output.
Department
Sociology
Publication Date
5-1-1993
Journal Title
Stata Technical Bulletin
Publisher
Stata Corporation
Document Type
Article
Recommended Citation
Hamilton, L.C. & C.L. Seyfrit. 1993. “Interpreting multinomial logistic regression.” Stata Technical Bulletin 13(May):24–28.
