Table 4 Results of mixed rank-ordered logit (MROL) modeld.

From: Using stated preference methods to design gender-affirming long-acting PrEP programs for transgender and nonbinary adults

 

(1)

(2)

(3)

(4)

(5)

MROL—Coeffa

MRO—SDb

MROL—coeff program Ac

MROL—coeff program Bc

MROL—coeff program Cc

Alternative-specific variables

 Site of LA-PrEP injection is arm (= 1)

0.0729**

0.0224

   

(0.0350)

(0.133)

   

 Duration of LA-PrEP protection against HIV is 12 months (= 1)

0.127***

0.531***

   

(0.0478)

(0.0860)

   

 Provider administering LA-PrEP is a gender-affirming provider (= 1)

0.0865**

0.128

   

(0.0399)

(0.219)

   

 Hormone co-prescription given at LA-PrEP visit is for oral and injectable hormones (= 1)

0.0993***

0.144

   

(0.0355)

(0.0948)

   

 Site of LA-PrEP program delivery is HIV clinic (= 1)

− 0.00902

0

   

(0.0370)

(0)

   

 Site of LA-PrEP program delivery is gender clinic (= 1)

0.0522 (0.0368)

0.0958 (0.1041)

   

Case-specific variables

 Age

  

− 0.0699***

− 0.0733***

− 0.0515***

  

(0.0179)

(0.0199)

(0.0154)

 Gender identity

  

0.702**

0.684**

0.747**

  

(0.305)

(0.301)

(0.302)

 Sex assigned at birth

  

− 0.119

− 0.435

− 0.477

  

(0.423)

(0.412)

(0.407)

 Race

  

0.104***

0.0849***

0.0410*

  

(0.0264)

(0.0240)

(0.0247)

 Level of education

  

0.145**

0.160**

0.134**

  

(0.0654)

(0.0645)

(0.0619)

 Annual income

  

− 0.145

− 0.0179

− 0.0360

  

(0.0947)

(0.0900)

(0.0732)

 Ever taken PrEP (= 1)

  

1.299***

1.433***

1.007***

  

(0.325)

(0.349)

(0.326)

 Ever engaged in sex work (= 1)

  

− 0.277

0.111

0.131

  

(0.415)

(0.388)

(0.398)

 Geography of current residential community

  

0.125**

0.0415

0.128***

  

(0.0507)

(0.0448)

(0.0468)

 Constant

  

0.962

1.676**

1.063

  

− 0.0699***

− 0.0733***

− 0.0515***

Observations

24,220

Log-likelihood

− 7769

AIC

15619.61

BIC

15900.14

Parameters

42

  1. aMeans of random coefficients for alternative-specific variables.
  2. bStandard deviation (SD) of the random coefficients for alternative-specific variables.
  3. cUsing No Program/Opt-Out as reference. A positive (negative) sign for an attribute means that level impacted positively (negatively) on utility and thus increased (reduced) the probability of choosing an alternative with that level.
  4. dNumber of participants = 366.
  5. Robust standard errors in (parentheses) ***p < 0.01, **p < 0.05, *p < 0.1.
Back to article page