University Cornell and Clark of Atlanta, 2007. 42 p.
Glick, Peter
Sahn, David E.
United States Agency for International Development, USAID
University Cornell and Clark of Atlanta
The authors use repeated rounds of Demographic and Health Survey data from eight african countries to examine changes in and determinants of three HIV risk behaviors: age at first intercourse; number of current sexual partners, and use of condoms. They find some evidence of changes in sample composition, which is easily handled in a multivariate framework, and find evidence as well of changes in how people respond to questions about HIV behavior. Because of the latter, which likely represents an increase in social desirability bias over time, their estimates of risk reduction may be upper bounds on the true effects. Overall the picture is one of reductions in risk behaviors over recent 4-6 year intervals, especially with respect to condom use; in some cases the changes seem large given the short time periods involved. With some exceptions, however, the extent and pervasiveness of these changes seems inadequate in relation to the urgency of the public health crisis represented by AIDS. With respect to the determinants of behaviors, schooling and wealth have contradictory impacts on risk behavior: they both tend to increase the likelihood of using condoms while (for men) also increasing the demand for additional sexual partners.
Record created by: