Table 2.

The Association Between Exposure to Violence, Neighborhood Characteristics, and Any Past Month Tobacco, Alcohol, and Marijuana (TAM) Use and Any Past Year Violence, and Cross-Level Interactions

	Any Past Month TAM Use (N=1661)		Any Past Year Violence (N=1718)
	b	SE	b	SE
Intercept	−2.55**	.11	−1.15**	.07
Exposure to violence	.26**	.04	.46**	.03
Age	.40**	.05	.02	.03
Male	−.07	.19	.20	.13
Hispanica	−.49	.30	−.16	.21
African Americana	−.91**	.28	.51*	.23
Other race/ethnicitya	−.90	.50	−.43	.34
Socioeconomic status	.04	.09	.01	.08
Low self-control	−.01	.08	.20*	.08
Routine activities	.47**	.10	.28**	.08
Curfew	−.12	.13	−.09	.13
Family social support	−.05	.09	−.01	.06
Peer social support	.18	.09	−.06	.07
Peer drug use	.15	.09	--	--
Perceptions of drug harmfulness	−.52**	.08	--	--
Any prior TAM use	1.01**	.22	--	--
Peer delinquency	--	--	.07	.06
Any prior violence	--	--	.92**	.14
χ2	66.25		78.09
Direct Neighborhood Effects
Level-1 intercept	−2.55**	.001	−1.15**	.01
Collective efficacy	.01	.01	.08	.04
Concentrated disadvantage	.001	.001	.004	.01
R2	.03		.05
Cross-Level Interactions
Exposure to violence	.27**	.001	.46**	.001
x Collective efficacy	−.01	.01	.001	.003
R2	.02		.002

^**p ≤ .01

^*p ≤ .05

Notes: Individual-level results are based on Bernoulli models; italicized coefficients indicate that the effect of exposure to violence was allowed to vary randomly across the79 neighborhood clusters; all other variables were fixed across neighborhoods. The level-two and cross-level interaction results are based on Bernoulli models using Empirical Bayes (EB) estimates. Cross-level interactions control for the direct effects of the neighborhood characteristics and all individual-level variables.

^aCaucasian is the reference group