Table 3.

Multivariate Regression. Baseline proportion of “don’t know” responses and ten-month change in proportion of “don’t know” responses regressed on covariates. Ten-month change in proportion of “don’t know” responses considers household randomization. (N=444)

	Baseline “Don’t know” responses [B (SE)]	p value	Change in “Don’t know” responses at follow-up [B (SE)]	p value
Intercept	0.178 (0.057)	0.002	0.012 (.060)	0.845
Gender
Female	−0.031 (0.023)	0.181	0.013 (0.024)	0.610
Male	reference		reference
Place of Birth
U.S.	0.086 (0.028)	0.002	−0.046 (.030)	0.132
Other	reference		reference
Marital Status
Married	0.019 (0.044)	0.665	0.035 (0.039)	0.364
Other	reference		reference
Parental Status
Parent	−0.068 (0.050)	0.181	0.026 (0.046)	0.577
Not a parent	reference		reference
Age Group
18 – 29	0.103 (0.056)	0.065	−0.111 (0.045)	0.014
30 – 39	0.039 (0.040)	0.332	−0.075 (0.046)	0.106
40 – 49	0.022 (0.033)	0.513	−0.069 (0.034)	0.042
50 –59	0.014 (0.034)	0.675	−0.040 (0.035)	0.253
60 +	reference		reference
SES
Owns both house and car	0.006 (0.024)	0.805	−0.021 (0.022)	0.335
Does not own house and car	reference		reference
Educational Level
< HS Education	reference		reference
HS Diploma +	0.011 (0.024)	0.634	−0.018 (0.026)	0.484
Insurance Status
Health Insurance	reference		reference
No Health Insurance	−0.008 (0.025)	0.737	−0.034 (0.024)	0.154
Healthcare System Use
In past year	reference		reference
Not in past year	−0.002 (0.044)	0.965	0.026 (0.039)	0.514
Intervention Effect
One received RAs			0.044 (0.028)	0.117
All received RAs			reference
No Behavioral Recs			0.020 (0.030)	0.513
Behavioral Recs			reference
One received RAs x			−0.104 (0.041)	0.011
No Behavioral Recs
Other (all no recs, all+recs, one+recs)			reference

Note: GEE adjustment with exchangeable covariance structure to account for clustering within the family looking at change from baseline to follow-up for each variable separately (i.e. bivariate associations). RAs = Risk Assessments; Recs = Recommendations