Table 3.

Overall Fit Indices and Path Estimates for Structural Models for Each of 4 Adolescent Problem Behaviors


	Model Fit Indices				Path Estimates	Indirect Effects	Total Effects

	χ² (d.f.) sig.	CFI	TLI	RMSEA	β^a	β	β

Running away	19.12 (7)
	.01	.99	.97	.04
Abuse/neglect to prostitution					.10^†	.12^***	.22^***
Abuse/neglect to running away					.30^***
Running away to prostitution					.38^***
Juvenile crime	88.17 (8)
	.00	.91	.84	.09
Abuse/neglect to prostitution					.13^**	.09^***	.22^***
Abuse/neglect to juvenile crime					.19^***
Juvenile crime to prostitution					.44^***
School problems	28.64 (8)
	.00	.96	.93	.05
Abuse/neglect to prostitution					.14^**	.08^***	.21^***
Abuse/neglect to school problems					.19^***
School problems to prostitution					.42^***
Early sexual initiation	.00 (0) ^b
	.00	1.00	1.00	.00
Abuse/neglect to prostitution					.14^**	.07^***	.22^***
Abuse/neglect to early sexual initiation					.18^***
Early sexual initiation to prostitution					.41^***

NOTE: χ² = chi square statistic reflecting difference between specified model and observed model; d.f. = degrees of freedom; CFI = Comparative Fit Index; TLI = Tucker-Lewis Index; RMSEA = Root Mean Square Error of Approximation. Path estimates are direct effects.

β represents the standardized probit coefficient for dichotomous dependent variables (early sexual initiation, prostitution) and the standardized linear regression coefficient for continuous dependent variables (running away, juvenile crime, school problems).

Fit indices could not be calculated because the model is over-identified (i.e., includes all possible parameters).

^†

p < .10,

p ≤ .05,

^**

p ≤ .01,

^***

p ≤ .001.