ADJUSTING FOR DATA CONTAMINATION IN STATISTICAL-INFERENCE

The randomized response survey model introduced in 1965 by Warner is r eviewed and applied to the analysis of contaminated data, that is, res ponse or reported data that is truthful with probability less than one . Two generic mechanisms are distinguished: an active mechanism whereb y the contamination is inserted into the process and hence becomes a s tatistical design parameter, and a passive mechanism whereby contamina tion of the response is suspected and hence becomes an analysis parame ter. The impact of contamination on the operating characteristics of s ome common statistical inference procedures is developed for binomial models.