Surveying precarious topics
|Convenor||Mr Simon Henke (GESIS - Leiniz Institute for the Social Sciences )|
In an experiment among students (N=6,261), we investigated how respondents assess the anonymity of the crosswise technique, triangular technique, two versions of the randomized response technique, and direct responding as well as their potential for answering trustfully. Preliminary results show that respondents perceived the crosswise and triangular techniques as promoting honest responding compared to the direct response technique, while both randomized response techniques were perceived to impede honesty. Distrust regarding the investigator, fearing negative reactions, assuming a low reliability of scientific research, and high burden of participation decrease honesty perceptions. The perceived anonymity did not vary across treatments.
Comparative evaluation studies for sensitive question techniques relying on the „more is better“ assumption have a serious weakness. If false positives occur, i.e., if respondents incorrectly give the socially undesirable answer, conclusions drawn about a particular question techniques' validity might be wrong. Validation studies where individual self-reports can be compared to some external criterion are obviously superior but are very restricted regarding the items that can be validated. Therefore, we propose a refined comparative approach that includes testing the more-is-better assumption and, hence, allows to assess more reliably the validity of a particular sensitive question technique.
Sexual behaviour is among the most sensitive topics in surveys. Literature shows that men who pay for sex are more likely to engage in risky behaviours than other males. I hypothesize that this correlation might be endogenous – at least partially – and driven by the unwillingness to reveal socially undesirable behaviours. Using Natsal-2 data, I utilize paradata to show how response patterns change depending on the privacy of the interview setting. Moreover, I take the previous research forward by re-estimating the correlation of paying for sex with risky behaviours after imputing item non-response in both sets of variables.
In this presentation I will share our experiences of undertaking a mixed methods study on sex work in Northern Ireland. The study was undertaken in 2014 in order to inform policy making in Northern Ireland on a proposed law that would make it illegal in Northern Ireland to purchase sex. I will specifically talk about the methodological challenges of recruiting both sex workers and clients for an online survey. I will share some key results of the study, but will also allude to implications of the socio-religious context in Northern Ireland for a study of this kind.