Background: The accuracy of behavioral data related to risk for HIV and other sexually transmitted infections is prone to misreporting because of social desirability effects. Because computer-assisted approaches are not always feasible, a noncomputerized interview method for reducing social desirability effects is needed. The previous performance of alternative methods has been limited to aggregate data or constrained by the simplicity of dichotomous-only responses. We designed and tested a "polling box" method for case-attributable, multiple-response survey items in a low literacy population.
Methods: A cross-sectional survey was conducted with 812 female sex workers in Andhra Pradesh, India. For a subset of questions embedded in a face-to-face survey questionnaire, every third participant was provided graphical response cards upon which to mark their answer and place in a polling box outside the view of the interviewer. Multiple logistic regression analysis was used to test for response differences to questions about socially undesirable, socially desirable, or sensitivity-neutral behaviors in the 2 interview methods.
Results: Polling box participants demonstrated higher reporting of risky sexual behaviors and lower reporting of condom use, with no conclusive response patterns among sensitivity-neutral items.
Conclusion: Our findings suggest that the polling box approach provides a promising technique for improving the accurate reporting of sensitive behaviors among a low-literacy population in a resource poor setting. Additional research is needed to test logistical adaptations of the polling box approach.