ABSTRACT
OBJECTIVES: Reliable data on sexual behaviour trends are needed to evaluate HIV interventions in sub-Saharan Africa but are difficult to obtain due inter alia to social desirability bias. The objective of this paper is to assess whether the use of informal confidential voting interviews (ICVI) was associated with greater reporting of socially proscribed behavioural risk factors for HIV infection than were conventional interviewing methods. METHODS: Comparison of changes in reports of risk behaviours for HIV infection in ICVI versus face to face interviews (FTFIs) between the first two rounds of a large scale, longitudinal, population based survey in Manicaland, Zimbabwe. Examination of factors that could contribute to observed changes in the effect of ICVI, including temporal changes in response error and social desirability, and factors affecting statistical power to detect differences between methods-that is, reductions in the prevalence of risk behaviours and sample size. RESULTS: Enhanced reporting of HIV associated risk behaviours in ICVI interviews was not so apparent in the second round as in the first round of the survey, particularly for less frequently reported behaviours. Levels of reported HIV associated risk behaviour and sample sizes both declined between the two survey rounds. The level of response error was higher in ICVI interviews than in FTFI interviews but did not alter over time. CONCLUSION: ICVI interviews can reduce social desirability bias in data on HIV associated risk behaviours. The extent and direction of change in net reduction in bias over time remains uncertain and will depend on local circumstances.
