الفهرس يوجد فقط 14 صفحة متاحة للعرض العام
 |  | from | 116 |  |  |
| | | from | 116 | | |
المستخلص
In direct sample surveys about sensitive issues such as illegal use of drugs, tax evasion and bribery, the participants often refuse to reply or reply with false answers. To reduce such response error problems and encourage the respondents to give honest answers on sensitive questions, warner (1965) suggested an indirect questioning technique called 2randomized response3 (RR) technique. Such technique maintains the respondent{u2019}s privacy since the respondents are requested to give a random response in terms of 2Yes3 or 2No3 answers via a random device consisting of two related questions about possessing the sensitive characteristic. Horvitz et al. (1967) suggested that the respondent{u2019}s cooperation might increase if, instead, two unrelated questions are used, one about the sensitive characteristic and the other about a non-sensitive characteristic that is unrelated to the sensitive characteristic. In this study, three unrelated question RR models are proposed to obtain more efficient estimators of the proportion of the population bearing a sensitive characteristic and at the same time increase the respondents{u2019} cooperation. The proposed models are studied under both simple random sampling and stratified random sampling. Two of the proposed models have an additional advantage over many of the existing competitors. These models, in case the population proportion bearing a non-sensitive characteristic is unknown, require selecting only a single simple random sample