Fig. 3.

Differences in clinical judgment by patient risk behavior. Mean ratings and error bars represent estimated marginal means and standard errors. For each clinical judgment, different letters indicate statistically significant mean differences between two groups according to post-hoc pairwise comparisons (p < 0.050). For example, as indicated by differing letters (a–c), the MSM was judged to be at significantly higher HIV risk than the MSW, and the MID was judged to be at significantly higher HIV risk than both the MSM and the MSW