Table 3.

Predicting alcohol-use problems^a: unstandardized coefficients (b), standardized coefficients (β), and p values from AMOS multiple regression analysis using maximum likelihood estimation (N = 1,242)

	Model 1	Model 2	Model 3	Model 4	Model 5
Religiosity
>Frequency attend church
b	−0.056	−0.030	−0.022	−0.030	−0.010
β	−.118	−.064	−.045	−.063	−.021
p	<.001	.061	.182	.053	.538
Demographics—SES
Gender (f = 1, m = 2)
b		0.338	0.310	0.323	0.327
β		.201	.184	.192	.194
p		<.001	<.001	<.001	<.001
Years of school
b		−0.032	−0.031	−0.020	−0.022
β		−.076	−.073	−.047	−.053
p		.039	.046	.176	.129
Unemployed
b		0.289	0.268	0.216	0.218
β		.165	.153	.123	.124
p		<.001	<.001	<.001	<.001
Social integration
>Frequency seeing mother
b			−0.060	−0.039	−0.032
β			−.076	−.050	−.041
p			.024	.126	.210
Doing less well with friends
b			0.033	0.000	−0.003
β			.053	.001	−.005
p			.113	.987	.887
>Moves in 5 years
b			0.077	0.071	0.072
β			.072	.066	.068
p			.034	.042	.036
Mental health factors
Major depressive disorder
b				0.716	1.026
β				.309	.443
p				<.001	<.001
Mother’s regular use of alcohol
b				0.088	0.089
β				.068	.068
p				.056	.052
Interaction term
Depression × Frequency of Church Attendance
b					−0.119
β					−.172
p					.001
CFIb	.995	.990	.990	.986	.987
RMSEAc	.042	.036	.029	.032	.034
BCCd	46.03	96.86	155.45	216.56	254.42
χ2e	15.88	36.43	46.39	64.84	78.29
dff	5	14	23	29	32
p value for χ2	.007	.001	.003	<.001	<.001
χ2/df	3.18	2.60	2.02	2.24	2.45

Notes: AMOS = analysis of moment structures; SES = socioeconomic status; f = female; m = male; CFI = comparative fit index; RMSEA = root mean square error of approximation; BCC = Browne-Cudeck criterion.

^aAlcohol-use problems is a latent construct created from counts of dependence symptoms, impairment symptoms, health problems, and longer and higher quantity of use;

^bhypothesized model represents an adequate fit to the data if CFI > .95.

^cRMSEA values < .05 indicate good fit between the hypothesized model and the observed data;

^dBCC produces a rank order of models; smaller values represent better fit of the hypothesized model;

^eChi-square (minimum value of the discrepancy) isatest of the fit of the hypothesized model and perfect fit;

^fno. of distinct sample moments: no. of distinct parameters to be estimated (>2 considered adequate fit).