hotdeck replaces missing values for the variable indicated by its argument. It should be used within a multiple imputation sequence since missing values are imputed stochastically rather than deterministically. The nmiss missing values in each stratum of the data described by the `by' option are replaced by values sampled from the nobs observed values in the same stratum. The approximate Bayesian bootstrap method of Rubin and Scheker is used; first a bootstrap sample of nobs observations is sampled with replacement from the observed values, and the nmiss missing values are sampled at random (again with replacement) from this bootstrap sample. If a file is specified in a using clause, the modified file is written to disk and the existing data in memory are unchanged. Otherwise the data in memory are modified. This is version 1.65 of the software, requiring Stata v9. hotdeck6 may be used in earlier versions of Stata.