Methods for missing covariates in logistic regression (Q4490143)

Various methods have been suggested in the literature to handle a missing covariate in the presence of surrogate covariates. These methods belong to one of two paradigms. In the imputation paradigm, Pepe and Fleming (1991) and \textit{M. Reilly} and \textit{M. S. Pepe} [Biometrika 82, No. 2, 299--314 (1995; Zbl 0828.62097)] suggested filling in missing covariates using the empirical distribution of the covariate obtained from the observed data. We can proceed one step further by imputing the missing covariate using nonparametric maximum likelihood estimates (NPMLE) of the density of the covariate. Recently \textit{S. A. Murphy} and \textit{A. W. Van der Vaart}, in an unpublished manuscript from 1998, showed that such an approach yields a consistent, asymptotically normal, and semiparametric efficient estimate for the logistic regression coefficient. In the weighting paradigm, \textit{L. Zhao} and \textit{S. Lipsitz} [Stat. Medicine 11, 769--782 (1992)] suggested an estimating function using completely observed records after weighting inversely by the probability of observation. An extension of this weighting approach designed to achieve semiparametric efficient bound is considered by \textit{J. M. Robins}, \textit{F. Hsieh} and \textit{W. Newey} [Semiparametric efficient estimation of a conditional density with missing or mismeasured covariates. J. R. Stat. Soc., Ser. B 47, No. 2, 409--424 (1995)]. The two ends of each paradigm (NPMLE and RHN) attain the efficiency bound and are asymptotically equivalent. However, both require a substantial amount of computation. A question arises whether and when, in practical situations, this extensive computation is worthwhile. In this paper we investigate the performance of single and multiple imputation estimates, weighting estimates, semiparametric efficient estimates, and two new imputation estimates. Simulation studies suggest that the sample size should be substantially large (e.g. \(n=2000\)) for NPMLE and RHN to be more efficient than simpler imputation estimates. When the sample size is moderately large \((n\leq 1500)\), simpler imputation estimates have as small a variance as semiparametric efficient estimates.