Abstract
Revised Centers for Disease Control and Prevention recommendations on HIV testing now promote testing of most risk groups. However, positive results for low-risk individuals are more likely to be false positives than for high-risk individuals, making clear communication of test results even more imperative. In a study, we evaluated current counseling of low-risk test recipients via a sample of 29 HIV hotline counselors from U.S. state and national hotlines. 100% of counselors interviewed failed to provide an accurate conditional HIV risk for low-risk women, but were more likely than a 1998 German sample to report that false positives could occur. In a second study, undergraduates read idealized transcripts of interviews with HIV counselors and computed conditional risk for a low-risk individual. The natural frequency format offered a small but significant improvement in conditional reasoning, comparable to the effect of numerical literacy. Applications for ecologically valid numerical presentations of risk and implications for numeracy are discussed.
Keywords: Bayesian reasoning, conditional risk, HIV counseling, HIV testing, numeracy, risk communication.