Abstract
A method used to compilsate for nonresponse is to impute missing vaiues; that is, to complete the data set by using the respondent data to replace all nonresponses. Due to the re-use of the respondent data, the distribution theory of statistics computed from a completed data set is more complex than that of the sane statistics when there are no missing data. The asymptotic distribution of the goodness-of-fit chi-square statistic computed from a completed data set is derived. The effect of the characteristics of the imputation procedure on this distrihtion is examined for several imputation procedures