Skip to main content
. 2018 Oct 15;5(7):431–442. doi: 10.1089/lgbt.2018.0041

Table 4.

Generalized Linear Models of Selected Covariates on Composite Percent of Maximum Possible Outcomes

Covariate HATH-8,aexp(b)e(SE) MHS,bexp(b) (SE) MTS,cexp(b) (SE) HIVS,dexp(b) (SE)
Gender
 Male 1.00 (.) 1.00 (.) 1.00 (.) 1.00 (.)
 Female 0.45*** (0.10) 0.03*** (0.02) 0.53* (0.16) 1.08 (0.29)
 Missing/undisclosed 0.93 (0.54) 0.35 (0.46) 1.30 (1.45) 0.88 (0.72)
Year of graduation
 1942– 1.00 (.) 1.00 (.) 1.00 (.) 1.00 (.)
 1983– 0.95 (0.18) 2.12 (2.12) 0.91 (0.31) 0.40*** (0.10)
 1995– 0.62* (0.14) 0.45 (0.34) 0.52 (0.26) 0.39** (0.12)
 2006– 0.47* (0.15) 0.08*** (0.06) 0.85 (0.35) 0.34** (0.13)
 Missing 1.94* (0.54) 1.25 (0.83) 2.14 (0.98) 0.04** (0.05)
Sexual orientation   Not estimable    
Heterosexual 1.00 (.) 1.00 (.) 1.00 (.)
Homosexual/bisexual/decline 0.35* (0.18) 0.18** (0.12) 0.33** (0.14)
Observations 446 448 444 447

(.) indicates that no confidence interval is presented because the indicated category is the reference category.

Because of marked skewness of composite outcomes, models were fit with gamma family distribution and log link.

a

HATH-8 POMP covariate F tests: (1) gender F (2, 444) = 7.00, p = 0.0010; (2) graduation year F (4, 442) = 4.76, p = 0.0009; (3) sexual orientation F (1, 445) = 4.25, p = 0.0398.

b

MHS POMP covariate F tests: (1) gender F (2, 446) = 10.26, p < 0.00001; (2) graduation year F (4, 444) = 3.64, p = 0.0062. Because all respondents who endorsed being homosexual or bisexual (or declined to state) scored 0 on the MHS POMP outcome, the full generalized linear model failed to converge. Therefore, a reduced model, omitting sexual orientation, was fit for MHS POMP.

c

MTS POMP covariate F tests: (1) gender F (2, 442) = 1.14, p = 0.3197; (2) graduation year F (4, 440) = 0.75, p = 0.5591; (3) sexual orientation F (1, 443) = 3.08, p = 0.0799.

d

HIVS POMP covariate F tests: (1) gender F (2, 445) = 0.05, p = 0.9494; (2) graduation year F (4, 443) = 5.91, p = 0.0001; (3) sexual orientation F (1, 446) = 6.79, p = 0.0095.

e

Exponentiated coefficients [exp(b)] are interpreted as the ratio of the arithmetic means, comparing the index covariate level with the reference level.

*

p < 0.05, **p < 0.01, ***p < 0.001.