Answers are grounded in this paper and cite their sources.
Health behaviour that increases risk is stigmatized in many societies, leading to potential misreporting in surveys designed to capture such behaviour. This sensitivity bias can distort the conclusions drawn from such surveys, thereby influencing policy decisions. Identifying and addressing sensitivity bias in survey questions is crucial to ensure more accurate prevalence estimates and to inform effective policymaking. In this study, we measured sensitivity bias in ten sensitive questions on sexual behaviour, sexual health, drug consumption, and intimate partner violence in a sample of 2270 adults in Dar es Salaam, Tanzania. The data was collected between 19 June 2017 and 24 July 2018. The sensitivity bias was estimated by comparing responses to list experiments with responses to corresponding direct questions. We measured the presence of sensitivity bias in the total sample and analysed heterogeneities by demographic characteristics. We found sensitivity bias in questions on drug use, intimate partner violence, having had sex with multiple partners, and sexual health. Sensitivity bias for the question on transactional sex was negative, implying an overreporting in the direct question. We did not find sensitivity bias for outcomes related to alcohol consumption. Heterogeneities by demographic characteristics were most pronounced between men and women. We conclude that what is considered sensitive among survey respondents is highly context specific. List experiments may help in detecting and minimizing sensitivity bias in sensitive survey questions, but their accuracy is sensitive to several design features and implementation details.
No takes yet. Share an insight, caveat, or question.
Bogler et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: