Table 7.

Logistic regression model: Study of human papillomavirus (HPV) vaccine uptake in three rural Georgia counties, 2009–2011

Measure	Block 1			Block 2			[95% CI]

	β	SE β	Odds ratio (p value)	β	SE β	Odds ratio (p value)
Race^a,^b (0 = African American; 1 = other)	.748	.236	2.113 (.002)	.683	.239	1.980 (.004)	[1.239, 3.165]
Religion^a,^b (0 = Baptist; 1 = Other)	1.116	.230	3.054 (<.001)	.979	.236	2.663 (<.001)	[1.677, 4.229]
Perceived benefits (higher % agree Items)				−.012	.005	.988 (.025)	[.978, .999]
Perceived barriers^b (higher % agree Items)				.024	.005	1.024 (< .001)	[1.013, 1.035]
Constant	−1.165	.125	.312 (< .001)	−1.413	.233	.244 (< .001)
Block χ², df (p value)	48.564, 2 (< .001)			20.234, 2 (< .001)
Model χ², df (p value)	48.564, 2 (< .001)			68.798, 4 (< .001)
−2 log likelihood (−2LL)	593.602			573.368; [deviance χ² = 20.234, df = 2, p < .001]
Nagelkerke R	.128			.177
Correctly classified	68.0%			71.4%

Stepwise selection (forward likelihood ratio method) used to select variable to keep within each block.

Block 1: Demographic variables: number of male children, parent’s gender, race, religion, education.

Block 2: Four theoretical constructs: perceived vulnerability, severity, benefits, barriers.

AA = African American; CI = confidence interval; df = degrees of freedom; p = probability; SD = standard deviation; t = Student’s t test.

Religion and race were significantly associated (85% of Baptists were African American; 52.1% of other religions were African American; χ² = 57.032, df = 1, p < .001).

Perceived barrier scores were significantly lower for African American (AA) than for other races: (AA: mean = 31.4 (SD = 23.9); other races: mean = 38.2 (SD = 22.1); t = −2.749, df = 515, p = .005) and for Baptists (Baptists: mean = 31.2 (SD = 3.2); other religions: mean = 39.1 (SD = 24.0); t = −3.262, df = 513, p = .001).