.-
help for ^fitstat^ - 1.6.2 - 11/27/00
.-

Compute fit statistics for single equation regression models. -------------------------------------------------------------

^fitstat^ [^, s^aving^(name)^ ^u^sing^(name)^ ^b^ic ^f^orce ^save^ ^dif^]

Description -----------

^fitstat^ is a post-estimation command that computes a variety of measures of fit for many kinds of regression models. It works after the following: ^clogit^, ^cnreg^, ^cloglog^, ^intreg^, ^logistic^, ^logit^, ^mlogit^, ^nbreg^, > ^ocratio^, ^ologit^, ^oprobit^, ^poisson^, ^probit^, ^regress^, ^zinb^, and ^zip^. With t > he ^s^aving^()^ and ^u^sing^()^ options, it can also be used to compare fit measures for two different models.

For all models, ^fitstat^ reports the log-likelihoods of the full and intercept-only models, the deviance (D), the likelihood ratio chi-square (G2), Akaike's Information Criterion (AIC), AIC*N, the Bayesian Information Criterion (BIC), and BIC'.

For all models except ^regress^, ^fitstat^ reports McFadden's R2, McFadden's adjusted R2, the maximum likelihood R2, and Cragg & Uhler's R2. These measures all equal R2 for OLS regression. ^fitstat^ reports R2 and the adjusted R2 after ^regress^.

fitstat reports the regular and adjusted count R2 for categorical data models. For ordered or binary logit or probit models, as well as models for censored data (^tobit^, ^cnreg^, or ^intreg^), it also reports McKelvey and Zavonia's R2 > . In addition, it reports Efron's R2 for ^logit^ or ^probit^.

Not all measures are provided for models estimated with ^pweight^s or ^iweight^ > s.

Options -------

^saving(^name^)^ saves the computed measures in a matrix for subsequent comparisons. Name cannot be longer than 4 characters.

^using(^name^)^ compares the fit measures for the current model with those of the model saved as name. Name cannot be longer than 4 characters.

^bic^ only presents information measures.

^force^ will compare two models even when the number of observations or the estimation method varies between the two models.

^save^ and ^dif^ are equivalents of saving(0) and using(0) and do not require the user to provide model names.

Examples --------

To compute fit statistics for a single model:

. ^use mroz,clear^ . ^logit lfp k5 k618 age wc hc lwg inc^ . ^fitstat^

To compute and save fit measures:

. ^logit lfp k5 k618 age wc hc lwg inc^ . ^fitstat, saving(mod1)^

To compare saved model to current model:

. ^logit lfp k5 k618 age age2 wc hc lwg inc^ . ^fitstat, using(mod1)^

.- Authors: J. Scott Long - jslong@@indiana.edu Jeremy Freese - jfreese@@ssc.wisc.edu www.indiana.edu/~jsl650/