Multivariable logistic regression models for 1,154 US and 439 South African men who have sex with men (MSM) using scoring in lowest quintile on HIV-KQ-18 knowledge scores as outcome.

*P2).†P2).n.s. = eliminated through backward selection for given model.aVariables considered for inclusion into each model: age, race, education, sexual orientation, HIV testing behavior, employment status, number of gay or bisexual friends known, having and/or had a female sex partner, relationship status, condom and lubrication use, and scores on the gay identity scale. Backward selection procedures (α = .05) were conducted separately for US and South African MSM.bConstruct p-values for HIV testing history were p = .006 (US MSM) and pcConstruct p-values for age group were p = .002 (US MSM) and non-significant (South African MSM).dConstruct p-values for racial/ethnic group were p = .049 (US MSM), non-significant (South African MSM).