Surveys about human behaviors very often involves sensitive questions. Investigating confidential, stigmatizing, threatening topics may lead survey participants to refuse answering or provide untruthful response, particularly when questions are asked directly. Nonresponse and misreporting denote measurement errors that are difficult to treat and may produce unreliable analyses of the surveyed topics. The problem can be mitigated by adopting survey methods that enhance anonymity and, hence, increase respondents' cooperation.

One possibility is to create a trustful and confidential relationship between the interviewer and the survey participants. Alternatively, it is possible to fully protect the privacy of survey participants by adopting procedures that elicit information without any direct sensitive questioning. These nonstandard procedures are encompassed under the term of indirect questioning approach and allow researchers to obtain good estimates of certain population parameters such as the prevalence of a sensitive behavior, or the mean/total of a sensitive quantitative variable.

In the talk we discuss the survey design and the results of a real study which considers both the above said possibilities. The study is conceived as a face-to-face mixed-mode interview which permits to collect information on two sensitive topics, say cannabis use and its legalization, and to estimate the prevalence of cannabis users and the prevalence of people supporting cannabis legalization.

The fieldwork has been carried out by a single interviewer who recruited a voluntary sample of respondents on the basis of personal contacts, and collected data using both the indirect questioning survey mode and the traditional direct questioning approach.

Results highlight the improvements that can be achieved when: (i) the indirect questioning approach is used as an alternative to questions that are asked directly by the interviewer; (ii) a confidential and trusting rapport is established between the interviewer and the respondent.

Further, across the employed survey modes, it is shown the effect on the prevalence estimates exerted by questions of different sensitivity.