The Validity of the Randomized Response Technique

IN the considerable literature that has appeared on the randomized response technique (RRT), only one study (Lamb and Stem, 1978) has addressed its validity. However, this study utilized a convenience sample of rather small size, thereby limiting generalizability of results. Originally proposed by Warner (1965), RRT is a method designed to obtain reliable information when dealing with sensitive issues on surveys. In instances where the technique was used, it has been assumed that "reliable information" meant valid information. In the technical sense, this, of course, is not necessarily so. Therefore, it is the purpose of this research to investigate the validity of the technique, using a nationwide probability sample of adults. The split-sample version of the RRT developed by Moors (1971) was used in the present study. Parameters for maximizing the effectiveness of this technique in a field situation were established by Zdep and Rhodes (1976). Determining the validity of a method such as RRT is not easy, precisely because RRT offers an alternative to direct questioning in situations where direct questioning is likely to result in obtaining socially desirable responses rather than truthful responses. Validation, then, requires that RRT results be compared with some other estimate of the sensitive characteristic in the population. To be sure, if this estimate were available, there would be no need for using RRT, which is costly and inefficient to administer.