A procedure designed to reduce the influence of social desirability bias in measuring attitudes in a aggregate group. Subjects or given a pair of questions that are true-false, yes-no, agree-disagree format. One of the questions is the legitimate target question and the other is an ambiguous filter. Each participant then rolls a die to determine which question he or she will answer. They then answer the question, but the interviewer is not aware of which of the pair of questions was answered. Even though the interviewer is unaware with question was answered, the technique has validity because the use of probability theory will provide an estimate to the distribution of responses to the target question in the group.