Predictors of Sexual-Risk Taking.

Note. Raw regression weights are reported with standard errors (in parentheses). Dashes (−) reflect sets of items that were not included in an analysis due to their negligible contribution to improvement in model fit. * p<.05
aRegression weights represent change in log odds (e.g.,.77 gives e0.77 = 2.16× increase in odds of engaging in sexual behavior for females relative to males, given other covariates in the model.
bAnalyses carried out on data from the subset of individuals who reported previous sexual activity.
cExponentiated coefficients show the multiplicative increase in expected number of lifetime sex partners (e.g.,.34 gives e0.34 = 1.4× increase in number of sexual partners for youth with primary education relative to those without).
dCoefficients represent change in log odds of incremental probability of virginity loss. As examples, holding other variables constant, [A] completion of secondary education reduces the incremental (yearly by age) hazard of virginity loss by a factor of e−0.73 = 0.48 or 52% (1–.48) relative to those who have not completed primary school, and [B] Maasai have an increased yearly hazard of virginity loss equal to e0.35 = 1.42 or 42% relative to non-Maasai.
eAge was not included in the set of basic demographic predictor variables for this analysis.
fNumber of villages (out of 7, excluding reference village) showing a significant positive relationship to outcome.
gMedia consumption is included for consistency, despite having been excluded as a predictor set in each of the best-fitting models in Table 2.