摘要：Missing data is a common problem in statistical analyses. Tomake use of information in data with incomplete observation, missing valuescan be imputed so that standard statistical methods can be used to analyzethe data. Variables with missing values are often categorical and the miss-ing pattern may not be monotone. Currently, commonly used imputationmethods for data with a non-monotone missing pattern do not allow di-rect inclusion of categorical variables. Categorical variables are converted tonumerical variables before imputation. For many applications, the imputednumerical values for those categorical variables must then be converted backto categorical values. However, this conversion introduces bias which canseriously a ect subsequent analyses. In this paper, we propose two directimputation methods for categorical variables with a non-monotone missingpattern: the direct imputation approach incorporated with the expectation-maximization algorithm and the direct imputation approach incorporatedwith a new algorithm: the imputation-maximization algorithm. Simulationstudies show that both methods perform better than the method using vari-able conversion. An application to real data is provided to compare thedirect imputation method and the method using variable conversion.
