Modelo de respuesta aleatorizada de Warner para incrementar la probabilidad de obtener respuestas sinceras a preguntas sensibles

Abstract

The veracity of the answers in studies on compromising questions such as drug use is a subject of permanent discussion and analysis, given that the conclusions of any study based on surveys depend on the quality and authenticity of the responses of the interviewees. The randomized response model proposed by Warner was studied in order to reduce the response bias when sensitive and / or delicate questions are asked, the application was carried out at the Faculty of Pure and Natural Sciences of the Universidad Mayor de San Andres. The design of the method is quantitative and not experimental. In the study, sensitive or sensitive questions were considered for the interviewee, such as cheating behavior in exams and drug use, applying the Warner randomized question method and the direct question method. For the calculation of the sample size, a maximum estimation error of 4.07% was considered with a confidence level of 95%, the sample size was 507. For this purpose, a sample survey was conducted with 507 students applying a technique of Randomization regarding the last digit of their identity card in such a way that the interviewee must answer sensitive questions according to the result that the number of their identity card gives, from this the proportion of students who have the sensitive characteristic of indirectly, giving the interviewee anonymity and allowing a sincere response. Therefore, a probabilistic relationship is established between a given answer and the sensitive question. The results showed a significant increase in the number of affirmative responses with Warner’s randomized response method in relation to the direct question regarding drug use and cheating on university exams.