Table 3.

Correlates of first use occurring at a nightclub.

	Bivariable tests			Multivariable model
	Non-nightclub venue weighted %	Nightclub weighted %	P	aPR	(95% CI)	P
Age – Mean (SE)	25.2 (0.5)	27.8 (0.8)	.007	1.04	(0.99, 1.09)	.102
Sex			.230
Male	62.3	72.2		1.00
Female	37.7	27.8		1.04	(0.54, 1.99)	.906
Race/Ethnicity			.391
White	48.0	47.9		1.00
Black	3.0	7.5		2.39	(1.05, 5.48)	.039
Hispanic	21.2	28.3		1.21	(0.65, 2.27)	.543
Asian	20.4	12.7		0.95	(0.40, 2.23)	.903
Other/Mixed	7.4	3.6		0.72	(0.24, 2.09)	.540
Sexual Orientation			<.001
Heterosexual	87.0	61.3		1.00
Gay/Lesbian	6.4	31.3		2.81	(1.54, 5.12)	.001
Bisexual	5.9	5.8		1.38	(0.49, 3.86)	.540
Other Sexuality	0.7	1.6		2.56	(1.02, 6.45)	.046

Note. Rao-Scott chi-square was used to determine whether there were bivariable differences, although differences in age, which is a ratio measurement, were determined using linear regression. All variables were then fit as covariates into separate multivariable generalized linear models using Poisson and log link to determine associations with all else being equal. This generated adjusted prevalence ratios (aPRs) and 95% confidence intervals (CIs). A Bonferroni statistical correction was applied to bivariable tests, so results are only deemed statistically significant when p < .0125 (alpha = .05/4 outcomes).