Table 3.

Results of multivariate logistic regression models comparing the three drug use groups.

		Y3 Self-Reported Past-year Drug Use
		NPA Usea vs. No Drug Usec (n=430)		Drug Use without NPAb vs. No Drug Usec (n=787)		NPA Usea vs. Drug Use without NPAb (n=519)
		AOR (95%CI)	p	AOR (95%CI)	p	AOR (95%CI)	p
Stage 1	Gender = Male	1.80 (1.23, 2.63)	.003	1.32 (0.99, 1.77)	.062	1.40 (0.94, 2.08)	.098
	Race/Ethnicity = White Non-Hispanic	2.70 (1.65, 4.43)	<.001	1.86 (1.34, 2.57)	<.001	1.44 (0.86, 2.41)	.164
	Parental Education
	Some college or less	Reference		Reference		Reference
	Bachelor’s degree	0.97 (0.50, 1.87)	.923	1.09 (0.67, 1.77)	.726	1.07 (0.55, 2.09)	.842
	Graduate degree	0.90 (0.49, 1.66)	.743	1.11 (0.72, 1.72)	.631	0.90 (0.48, 1.67)	.730
	BIC	71.4		92.3		73.6
Stage 2	Affective Dysregulation (DI-A)	1.05 (1.03, 1.07)	<.000	1.01 (0.99, 1.02)	.550	1.05 (1.02, 1.07)	<.001
	Conduct Problems (CECPI)	1.10 (1.03, 1.16)	.003	1.06 (1.02, 1.10)	.003	1.04 (0.98, 1.09)	.181
	Depressive Symptoms (BDI)	1.07 (1.03, 1.12)	.002	1.02 (0.98, 1.05)	.387	1.06 (1.01, 1.11)	.009
	Psychological Distress (GHQ)	1.07 (1.01, 1.13)	.015	1.00 (0.97, 1.04)	.848	1.07 (1.01, 1.12)	.020
Stage 3	Affective Dysregulation (DI-A)	1.05 (1.02, 1.07)	<.001	–	–	1.05 (1.02, 1.07)	<.001
	Conduct Problems (CECPI)	1.07 (1.01, 1.13)	.020	1.06 (1.02, 1.10)	.003	–	–
	BIC	386.6		457.0		336.1

Note. The Stage 1 model for each outcome included the three demographic variables (gender, race/ethnicity, parental education) only. Stage 2 represents a series of four models for each outcome, in which the statistical effect of each of the four predictors on the outcome is evaluated in turn, while adjusting for demographics. The Stage 3 model for each outcome reflects the best-fitting combination of predictor variables for that outcome, while adjusting for demographics and other predictors in that model (if any). For all predictor variables shown, higher scores indicate higher levels of problems or symptoms. BIC=Bayesian information criterion