Stata macros for multinomial conditional logit models
MCL stands for Multinomial Conditional Logit model. A conditional logit program is used to estimate a multinomial logistic model. This produces the same coefficients and standard errors as a regular multinomial logit program but has the advantage that it provides great flexibility for imposing constraints on the dependent variable. mclgen restructures the data so the model can be estimated by clogit, mclest estimates the model using clogit.
In addition, mclest can estimate two special models: stereotyped ordered regression (SOR) and Goodman's row and columns model 2 (RC2). Both models estimate a scaling metric for the dependent variable; the RC2 model estimates a scaling metric for a categorical independent variable as well.
The depvar argument is required. depvar corresponds with the dependent variable in a multinomial logit model and should indicate a categorical response factor with a maximum of 12 levels.
Note that mclgen will modify the data, and that the data should be saved before running mclgen.
An MCL model uses a conditional logit model to estimate a multinomial logistic model. This provides great flexibility for imposing constraints on the response factor, the dependent variable in a multinomial logistic model. Different constraints can be imposed on the response factor for each independent (dummy) variable. One application is to specify loglinear models for square tables such as quasi-independence, uniform association, symmetric association, into a multinomial logistic model. A further extension provided by mclest is to estimate special nonlinear designs, such as stereotyped ordered regression and Goodman's row and columns model 2.
In order to estimate an MCL model, the data must be transformed into a person/choice file. In a person/choice file, each respondent has a separate record for each category of the response factor (i.e. the dependent variable in a multinomial logit model). The reponse factor indexes the response options for respondents, a stratifying variable indexes the respondents, and a dichotomous dependent variable indicates which record corresponds with response option chosen by the repondent.
So for a response factor with 5 levels, the dataset is expanded 5 times. The response factor specified in mclgen indexes the response options for each respondent. mclgen creates __strata and __didep, the stratifying variable and dichotomous dependent variable for use by clogit or mclest.
In clogit, the dichotomous dependent variable is specified as the dependent variable and the stratifying variable is specified in the strata(varname) option. The main effects of the response factor correspond with the intercept of a multinomial logistic model. Interactions of the response factor with independent variables correspond with the effects of these independent variables.
If the response factor is modelled using a fixed reference category, the log likelihood, estimates and standard errors will be exactly the same as a model estimated with mlogit. However, this procedure followed here allows much more flexibility in imposing restrictions on the response factor.
See mclest for further information
Direct comments to: John Hendrickx
mclest is available at SSC-IDEAS. Use findit mcl to locate the latest version.
Also see On-line: help for mclest, mlogit, clogit, desmat, desrep, xi, xi3, ologit