Table 1.

Results ofmultivariate logistic regressions examining the association of social network characteristics with any heavy drinking days during the past 30 days, among Black homeless youth (n = 116) and among White homeless youth (n = 119), controlling for background characteristics

	Black	White
Network characteristicsa	OR [95% CI]	OR [95% CI]
Type of person
Relatives (3 or more)	0.96 [0.26, 3.51]	0.48 [0.19, 1.23]
Students who attend school
regularly (4 or more)	0.29 [0.08, 0.99]*	0.20 [0.05, 0.81]*
Employed persons (4 or more)	0.47 [0.13, 1.68]	1.80 [0.67,4.80]
Homeless persons (2 or more)b	2.53 [0.62, 10.30]	9.82 [2.66, 36.31]*
Drinking behavior
;Relatives who drink to intox. (any)	1.18 [0.32,4.33]	2.02 [0.81, 5.06]
Peers (neither relatives nor adults in positions of responsibility)
who drink to intox.c	1.01 [0.88, 1.15]	1.23 [1.12, 1.34]**

Notes: OR = odds ratio; CI = confidence interval; intoxication.

^aEach social network characteristic was examined in a separate model to avoid multicollinearity problems.

^b”Homeless persons” was significant in the model for White youth but not for Black youth; however, the interaction between “peers who drink” and race was not statistically significant (p < .05) in the model for the combined sample of Black and White youth (n = 335; OR = 4.42, 95% CI [0.73, 26.61], p < .10).

^c”Peers who drink” was significant in the model for White youth but not in the model for Black youth; the interaction between “peers who drink” and race was also significant in the model for the combined sample of Black and White youths (n = 335; OR = 1.12, 95% CI [1.05, 1.41], p < .01).

*p <.05;

**p <.01.