In health studies, we often face some variable missing. This missingness can happen in either response or other covariates. In this paper, the discussion focuses on missing covariates. A method is proposed for analysis of logistic regression models in which the response variable is polychotomous and some covariates’ values are missing at random. The maximum likelihood function of the model is derived and the results are compared with the routine methods based on elimination of missing cases. Both the proposed method and the usual method are compared on a real dataset of goiter disease and is shown that the proposed method acts significantly better than usual method.
Key words: Missing at random, logistic regression, polychotomous esponse, goiter disease, likelihood function.
Copyright © 2020 Author(s) retain the copyright of this article.
This article is published under the terms of the Creative Commons Attribution License 4.0