Table 3.

Multilevel models utilizing day-level and individual-level substance use to predict daily sexual engagement and CAS with casual partners.

	Timeline Follow-Back Combined Models
	Model 1			Model 2
	No Sex vs. Sexa			No CAS vs. CASb
	b	AOR	95%CI	b	AOR	95%CI
Intercept	−0.77	0.47***	0.38, 0.57	−1.96	0.14***	0.10, 0.20
Level 1: Event-Level Effects
Day-level heavy drinking	0.97	2.63***	2.28, 3.03	0.13	1.14	0.87, 1.50
Day-level marijuana use	1.30	3.66***	3.06, 4.38	0.46	1.58**	1.17, 2.13
Day-level club drugs use	2.39	10.88***	8.77, 13.50	0.81	2.25***	1.63, 3.09
Level 2: Dispositional Effects
Frequency of heavy drinking: Lower	−0.35	0.70**	0.55, 0.90	−0.01	0.99	0.63, 1.57
Frequency of heavy drinking: Higher	−0.37	0.69**	0.54, 0.88	−0.57	0.57*	0.35, 0.91
Frequency of marijuana use: Lower	−0.18	0.84	0.65, 1.07	−0.04	0.96	0.61, 1.52
Frequency of marijuana use: Higher	−0.78	0.46***	0.35, 0.61	−0.64	0.53*	0.31, 0.89
Frequency of club drug use: Lower	−0.43	0.65**	0.50, 0.85	−0.05	0.96	0.58, 1.59
Frequency of club drug use: Higher	−0.86	0.42***	0.32, 0.56	−0.10	0.91	0.54, 1.53

	Daily Diary Combined Models
	Model 3			Model 4
	No Sex vs. Sexa			No CAS vs. CASb
	b	AOR	95%CI	b	AOR	95%CI
Intercept	−0.79	0.46***	0.36, 0.58	−1.69	0.19***	0.13, 0.27
Level 1: Event-Level Effects
Day-level heavy drinking	0.75	2.12***	1.77, 2.54	0.47	1.60**	1.19, 2.14
Day-level marijuana use	0.85	2.34***	1.93, 2.85	0.31	1.36†	0.98, 1.89
Day-level club drugs use	1.60	4.94***	3.72, 6.56	0.94	2.56***	1.69, 3.86
Level 2: Dispositional Effects
Frequency of heavy drinking: Lower	−0.09	0.91	0.68, 1.22	0.10	1.10	0.71, 1.72
Frequency of heavy drinking: Higher	−0.10	0.90	0.67, 1.21	−0.20	0.82	0.52, 1.28
Frequency of marijuana use: Lower	−0.19	0.83	0.62, 1.11	0.12	1.13	0.72, 1.77
Frequency of marijuana use: Higher	−0.43	0.65**	0.47, 0.90	−0.63	0.53*	0.32, 0.89
Frequency of club drug use: Lower	−0.43	0.65**	0.47, 0.91	−0.48	0.62†	0.37, 1.04
Frequency of club drug use: Higher	−0.50	0.61**	0.43, 0.86	−0.59	0.55*	0.32, 0.95

Note:

^†p ≤ 0.08;

^*p ≤ 0.05;

^**p ≤ 0.01;

^***p ≤ 0.001.

^aBinary logistic regression;

^bMultinomial logistic regression (only one of two comparisons shown). All models were adjusted for HIV-positive status and relationship status as well as day of data collection (i.e., day of TLFB cycle or day of diary cycle). Comparison group for dispositional substance use is the non-use group for each substance.