Figure 2.
Distribution of preferences for enhanced HIV counseling and testing features among 300 female barworkers and 440 male mountain porters, Tanzania, 2017.
Notes: Results of gender-specific mixed logit models with effects-coded correlated random coefficients. Distributions represent kernel densities of individual-level preference estimates for 300 female barworkers (blue) and 440 male mountain porters (orange). The x-axes represent preferences for (to the right of the black line) or against (left of the black line): (a) everyday testing relative to testing on weekdays only, (b) oral swab relative to the estimated more preferred of either venipuncture or finger prick; (c) a complementary health check or (d) a complementary STI examination relative to HIV testing only; (e) confidential assisted partner notification in the event of a positive test result relative to self-disclosure. Effects-coded (mean) coefficient estimates from gender-specific mixed logit models are indicated by ßBarworkers and ßPorters. The statistical significance of differences in the distribution of individual-level preference estimates between risk groups, as evaluated by Student’s t-tests, is indicated by pDifference.