Randomized Response

Profits and Costs of Using Randomized Response Techniques

The advantage of randomized response is that more valid population estimates of sensitive behavior can be derived. The cost is that randomized response is less efficient than a direct question. The randomization procedure increases the sampling variance, which makes it necessary to use larger samples. For example, Warner's method typically needs as many as 10 times the number of respondents to be just as powerful as a direct question design.

Extra costs are also associated with the increased complexity of the randomized response question. When randomized response is used, respondents have to understand and follow the instructions, in addition to understanding and formulating a response to the question itself. This introduces a new source of error, namely, misunderstanding the RRT procedures, cheating on this procedure, or both.

The advantage of using a randomized response technique outweighs the extra costs only when the population estimates become significantly more valid than estimates obtained from direct question-answer designs. Meta-analysis has shown that the advantage of using randomized response outweighs the disadvantage of having to use a larger sample if the social sensitivity of a research topic is large.

Other Randomized Response Techniques

Since Warner's original idea, many adaptations and refinements have been developed. First, methods were developed to improve the power of the design. As previously noted, compared to a direct question, Warner's method typically needs much larger samples. Variations on Warner's original method have been developed that have a larger statistical power. These methods include the unrelated question technique and the forced response technique. In addition, randomized response methods have been developed that are easier for the respondent, such as Kuk's card method and the item count technique.