Statistical Inference For Bidirectionally Ordinal Square Contingency Tables With Missing Data

Contingency table is an important research filed in statistical analysis. It can be for all kinds of proctical problems.We often meet missing data during analysis of contingency tables.So statistis analysis of missing data are also important aspect of statistics research.For application,they are care for the practice problem how to use statistis theory to analyse missing data,that is, how to reasonably substitute missing data by concrete number that can approximate to the true value of missing data.We discuss the statistics inference for bidirectionally ordinal square contingency tables with missing data.For example,missing data occur for experiment character or the subject absence in medicinal experiment.when the missing data includ sample information and the variables are ordinal variables,the problem is meaningful and challenging.The theory in the article can be used for foregoing problem.taking into account the ordinal character of variables, we introduce a loglinear model under restriction :whereA, and are parameters which satisfyThen we introduce EM algorithm based on before-mentioned model.The algorithm is used to the actual example finely.