Table 5.

Negative Binomial Regression Results Evaluating Drinking as a Function of Identification and Perceived Norms for the Typical Same-Race Student on Campus

Predictor	B	B SE	IRR	Lower 95% IRR	Upper 95% IRR	t	d
Step 1: Δ-2LL = −88.28; df = 4; p < .001.
Asian	−0.8614	0.0679	0.423	0.370	0.483	−12.69***	0.43
Multiracial	−0.3021	0.0841	0.739	0.627	0.872	−3.59***	0.12
African American	−1.1359	0.1554	0.321	0.237	0.435	−7.31***	0.25
Other	−0.3801	0.0863	0.684	0.577	0.810	−4.40***	0.15
Step 2: Δ-2LL = −125.54; df = 2; p < .001.
Identification	0.043	0.016	1.044	1.012	1.077	2.70**	0.09
Perceived Norm	0.046	0.003	1.047	1.040	1.053	14.68***	0.50
Step 3: Δ-2LL = −12.14, df = 1; p < .001.
Identification × Perceived Norm	0.010	0.002	1.010	1.006	1.013	5.05***	0.17

Note.

^***p < .001.

^**p < .01. Δ-2LL at step 1 represents the change in −2 log likelihood relative to an intercept only model. IRR (incident rate ratio) represents proportional change for each unit increase in the predictor (e.g., an IRR of 1.05 = a 5% increase for each unit change in the predictor). Cohen’s d was calculated using the formula 2t(sqrt[df]). Race categories are coded such that tests of parameter estimates represent the comparison of the given race with Caucasian students.