Estimating The Regression Function In A Semiparametric Logistic Model Using Likelihood Ratio Tests

We consider estimation of the regression function in a semiparametric logistic regression model using likelihood-ratio based inversion. The dichotomous response variable ∆ is influenced by a set of covariates that can be partitioned as (X,Z) where Z (real valued) is the covariate of primary interest and X (vector valued) denotes a set of control variables. For any fixed X, the conditional probability of the event of interest (∆ = 1) is assumed to be a non–decreasing function of Z. The effect of the control variables is captured by a regression parameter β. We show that the baseline conditional probability function (corresponding to X = 0) can be estimated by isotonic regression procedures and develop a likelihood ratio based method for constructing asymptotic confidence intervals for the conditional probability function (the regression function) that avoids the need to estimate nuisance parameters. Interestingly enough, the calibration of the likelihood ratio based confidence sets for the regression function no longer involves the usual χ quantiles, but those of the distribution of a new random variable that can be characterized as a functional of convex minorants of Brownian motion with quadratic drift. Confidence sets for the regression parameter β can however be constructed using asymptotically χ likelihood ratio statistics. The finite sample performance of the methods are assessed via a simulation study. The techniques of the paper are also applied to data sets on primary schoolattendance among children belonging to different socio–economic groups in rural India.