Table 2.
Multivariable Zero-Inflated Negative Binomial Models Showing the Regression of Price Received for the Most Recent Commercial Sex Act Among Female Sex Workers Reporting 1 Commercial Sex Partner at the Most Recent Sex Act (FSW1) or 2 Commercial Sex Partners at the Most Recent and Second Most Recent Sex Acts (FSW2) in the Past Year Against Its Predictors
| Variable | Model 1: FSW1 (n = 127, zeros = 14) |
Model 2: FSW2, Most Recent Partner Included (n = 125, zeros = 9) |
Model 3: FSW2, Both Partners Included, Unclustered (n = 252, zeros = 18) |
Model 4: FSW2, Both Partners Included, Clustered on FSW (n = 252, zeros = 18) |
||||
|---|---|---|---|---|---|---|---|---|
| Excess Zeros | NB Estimate | Excess Zeros | NB Estimate | Excess Zeros | NB Estimate | Excess Zeros | NB Estimate | |
| Logit β Coefficient (95% CI) | Log β Coefficient (95% CI) | Logit β Coefficient (95% CI) | Log β Coefficient (95% CI) | Logit β Coefficient (95% CI) | Log β Coefficient (95% CI) | Logit β Coefficient (95% CI) | Log β Coefficient (95% CI) | |
| Marital history (vs previously married) | … | … | Reference | Reference | … | … | … | … |
| Still in union | … | … | −17.6 (−23 776 to 23 741) | 0.99 (.39 to 1.60)a | … | … | … | … |
| Never married | … | … | 0.59 (−1.44 to 2.63) | −0.10 (−.48 to .28) | … | … | … | … |
| Secondary education (vs primary education) | −0.57 (−2.34 to 1.20) | 0.55 (.20 to .90)a | 1.29 (−2.03 to 4.62) | 0.36 (.06 to .65)b | 0.42 (−1.03 to 1.88) | 0.28 (.08 to .49)a | 0.42 (−1.38 to 2.23) | 0.28 (.00 to .56)b |
| Years of residence | −0.09 (−.28 to .10) | −0.02 (−.03 to −.00)b | … | -… | … | … | … | … |
| Occupation is sex worker (vs not) | … | … | … | … | −21.24 (−46 010 to 45 967)c | 0.37 (.12 to .62)a | −21.2 (−22.6 to −19.9)d | 0.37 (.04 to .69)b |
| Most recent sex partner vs new client | … | … | … | … | Reference | Reference | Reference | Reference |
| Repeat client | … | … | … | … | −1.06 (−3.58 to 1.46) | 0.25 (−.01 to .52) | −1.06 (−4.26 to 2.14) | 0.25 (−.06 to .57) |
| Regular client (pay on credit) | … | … | … | … | 1.07 (−.87 to 3.00) | 0.43 (.15 to .71)a | 1.07 (−1.17 to 3.31) | 0.43 (.05 to .82)b |
| Regular client (pay according to FSW need) | … | … | … | … | −0.47 (−3.40 to 2.46) | −0.23 (−.59 to .13) | −0.47 (−3.74 to 2.80) | −0.23 (−.63 to .17) |
| Dry sex (vs no dry sex) | … | … | … | … | 0.96 (−.50 to 2.43) | 0.44 (.20 to .68)d | 0.96 (−1.15 to 3.08) | 0.44 (.10 to .78)b |
| No. of sex acts >1 | 0.16 (−0.68 to 1.00) | 0.27 (.10 to .43)a | −0.22 (−.97 to .53) | 0.15 (.06 to .25)a | −0.26 (−.89 to .36) | 0.12 (.03 to .20)a | −0.26 (−1.05 to .52) | 0.12 (.03 to .21)a |
| Client does not request condom use (vs client requests) | … | … | 0.13 (−1.63 to 1.89) | 0.45 (.16 to .73)a | … | … | … | … |
| Unprotected sex (vs condom use) | … | … | … | … | 1.11 (−.24 to 2.46) | 0.36 (.14 to .57)d | 1.11 (−.56 to 2.78) | 0.36 (.09 to .62)a |
| Value | Value | Value | Value | |||||
| Vuong test of ZINB vs standard NB | z = 1.49, P = .068 | z = 1.51, P = .065 | z = 2.87, P < .01 | … | ||||
| Log likelihood | −469 | −454 | −913 | −913 | ||||
| AIC | 956.8 | 935.3 | 1864.2 | 1864.2 | ||||
| BIC | 982.4 | 974.9 | 1931.3 | 1931.3 | ||||
All models were zero-inflated negative binomial regressions. Missing payments were omitted. Positive coefficients indicate price is positively associated with the predictor, whereas negative coefficients indicate negative correlations.
Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion; CI, confidence interval; NB, negative binomial; ZINB, zero-inflated negative binomial.
a P < .01.
b P < .05.
c Some cells have zero counts.
d P < .001.