Table 3.

Tests of Mediation for Path Analysis Models

Model	Pathway	Indirect Effects
		β a	SE
1	1 Marginalization → School Attachment → Depressive Symptoms → Marijuana Use	.001	(.000) *
	2 Marginalization → School Attachment → Marijuana Use	.000	(.000)
	3 Marginalization → Depressive Symptoms → Marijuana Use	−.001	(.001)
2	1 Marginalization → School Attachment → Depressive Symptoms → Marijuana Use Never vs. Initiated	−.010	(.004) *
	2 Marginalization → School Attachment → Marijuana Use Never vs. Initiated	−.017	(.009)
	3 Marginalization → Depressive Symptoms → Marijuana Use Never vs. Initiated	.010	(.009)
	4 Marginalization → School Attachment → Depressive Symptoms → Marijuana Use Both vs. Initiated	.005	(.003)
	5 Marginalization → School Attachment → Marijuana Use Both vs. Initiated	.023	(.011) *
	6 Marginalization → Depressive Symptoms → Marijuana Use Both vs. Initiated	−.005	(.006)
3	1 Marginalization → School Attachment → Depressive Symptoms → Alcohol Use	.000	(.000) *
	2 Marginalization → School Attachment → Alcohol Use	.001	(.001)
	3 Marginalization → Depressive Symptoms → Alcohol Use	.000	(.000)
4	1 Marginalization → School Attachment → Depressive Symptoms → Alcohol Use Never vs. Initiated	−.006	(.003) *
	2 Marginalization → School Attachment → Alcohol Use Never vs. Initiated	−.002	(.006)
	3 Marginalization → Depressive Symptoms → Alcohol Use Never vs. Initiated	.006	(.006)
	4 Marginalization → School Attachment → Depressive Symptoms → Alcohol Use Both vs. Initiated	.009	(.003) *
	5 Marginalization → School Attachment → Alcohol Use Both vs. Initiated	.014	(.009)
	6 Marginalization → Depressive Symptoms → Alcohol Use Both vs. Initiated	−.008	(.008)

^*p < .05.

^acoefficient estimates in models 2 are unstandardized because the endogenous variables are categorical.