New version page

Group comparisons in logit and probit using predicted probabilities

This preview shows page 1-2-24-25 out of 25 pages.

View Full Document
View Full Document

End of preview. Want to read all 25 pages?

Upload your study docs or become a GradeBuddy member to access this document.

View Full Document
Unformatted text preview:

G roup comparisons in logit and probitu sin g p r ed ic te d p r o b a b ilitie s1J. Scott LongIndiana Universit yJune 25, 2009AbstractThe compa rison of groups in regression models for binary outcom es iscomplica ted by an iden tification problem inheren t in these models. Tra-ditional tests of th e equality of coefficients acro ss grou ps co nfoun d themag nitud e of the regressio n coefficients with residual variation. If theamoun t of residual variation differs between groups, the test can leadto in co rrect conclusions (Allison 19 99 ). Allison proposes a test for theequalit y of regression coefficients that remov es the effect of group differ-ences in residual variation by adding the assumption that the regressioncoefficients for some variables are iden tical across groups. In practice, aresearcher is unlike ly to hav e either empirical or theoretical justificationfor this assumption, in whic h case the A llison’s test can also lead to incor-rect conclusions. An altern ative approach, suggested here, uses pr ed ictedproba bilities. Since predicted pro ba bilities are unaffected by residual vari-ation, test s of the e qua lity of predicted p ro babilities across gr oup s can beused for group comparisons without assuming the equality of the regres-sion coefficien ts of some varia bles. Using predicted probabilities requiresresearc her s to think differently about comparing groups. With tests of theequalit y of regression coefficients, a single test lets the researc her concludeeasily whether the effects of a variable are equal a cross groups. Testing theequality of predicte d probab ilities requires multiple tests since group dif-ferences in prediction s vary with the levels of the variables in the model.A researcher m ust examine group differen ces in predictions at multiplelevels of the variables often requiring more complex conclusions on ho wgroups differ in the effect of a variable.1I thank Paul Allison, Ken Bollen, Rafe Stolzenberg, Pravin Trivedi, and Rich Williams for theircomments.1G roup comparisons in logit and probitusing predicted prob abilities1OverviewThe comparison of groups is fundamental to researc h in man y areas and tests com-paring groups have receiv ed a great deal of atten tion . Chow’s (1960) paper, declareda “citation classic” (G arfield 1984), provides a general framework for group com-parisons in the linear regression model. Suppose that we are compa ring the effectof  on  for women and men, where and are the coefficients of in terest.If 0: = is rejected, we conclude that the effect of  differs for men andwomen. Th is approach to testing group differences can be applied to many typesof regression models as shown b y Liao (2002). Allison (1999) poin ts out a criticalproblem when this test is used in models such as logit or probit. For these models,standard tests can lead to incorrect conclusions since they confound the magnitudeof the regression coefficien ts with the amount of residual variation. Allison proposesa test that remo ves the effect of residual variation b y assum ing that the coefficien tsfor at least one independent variable are the same in both groups. Unfortunately, aresearc her might lac k sufficient theoretical or empirical information to justify suc h anassumption. Making an ad hoc decision that some regression coefficients are equalcan lead to incorrect conclusions.2Tests of predicted probabilities pro vid e an al-ternativ e approac h for comparing groups that is unaffected by group differences inresidual variation and does not require assumptions about the equality of regressioncoefficien ts for some variables. Group comparisons are made by testing the equalit yof predicted probabilities at differen t values of the independent variables.This paper begins by reviewing wh y standard tests of the equality of regressioncoefficients across groups are inappropriate in some t ypes of models. I then showwhy predicted probabilities are unaffected b y residual variation and present a testof the equality of predicted probabilities that can be used for group comparisons ofeffects in models such as logit and probit. To illustrate this approach, I begin witha model that includes a single independen t variable, in which case all informationabout group differences can be sho w n in a simple graph. I extend this approach tomodels with m u ltiple independent variables, which requires mo re com plex analysisdue to the nonlinearit y of the models.2Williams (2009) raises other concerns with this tests that are discussed below.22 M odel iden tification in logit and probitDiscussions of the logit and probit model often note that the slope coefficients areonly identified up to a scale factor (Maddala 1983:23). This lac k of iden tification iswhy standard tests of the equalit y of regression coefficients across groups should notbe used. To see how identification causes a problem in group comparison s and whypredicted probabilities are not affected b y this problem, consider ho w these models arederived using an underlying latent variable (see Long 1997: 40-50 for a full derivation).Suppose that the latent ∗is linearly related to an observed  through the structuralmodel∗= 0+ 1 + where I use a single independent variable for simplicity. The laten t ∗is link e d to anobserved, binary  b y the measurement equation =½1 if ∗ 00 if ∗≤ 0 (1)If ∗is greater than 0,  is observed as 1. Otherwise,  is observed as 0. For example,when a person’s propensit y to be in the labor force exceeds 0, she joins the laborforce (i.e.,  =1). If her propensit y is at or below 0, she is not in the labor force (i.e., =0).Accord ing to equation 1, the probability of  =1is the proportion of the distrib-ution of ∗that is above 0 at a given value of :Pr ( =1| )=Pr(∗ 0 | )  (2)Substitu ting ∗= 0+ 1 +  and rearranging terms, the probability can be writtenin terms of the distribution of the errors:Pr( =1| )=Pr( ≤ 0+ 1 | )[ Figures 1 & 2 about her e ]Figure 1 illustrates the relationship between the structural model for ∗and theprobability of  for specific values of the param eters labeled as set :∗= 0+ 1 + (3)= −6+1 + Assum e that is normally distributed with mean 0 and variance 2=1,whicharethe


Loading Unlocking...
Login

Join to view Group comparisons in logit and probit using predicted probabilities and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Group comparisons in logit and probit using predicted probabilities and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?