Table 3.

Effect Sizes for Participant Characteristics on Aided Ad Recall Estimated Using the Marginalized Two-Part Model, n = 583.

		95% Confidence Limits
	Estimatea	Lower	Upper	p-valueb
Age (ref: 13–15)				0.796
16–17	1.03	0.80	1.34	0.796
Race (ref: White)				0.054
Black or African American	1.48	1.03	2.12	0.036*
Some other race	0.88	0.63	1.22	0.447
Hispanic (ref: No)				0.066
Yes	1.34	0.98	1.82	0.066
Gender (ref: Male)				0.521
Female	1.09	0.84	1.40	0.521
Vaping Status (ref: Not at-risk of vaping)				0.218
At-risk of vaping	0.87	0.66	1.15	0.331
Current vaper	0.66	0.42	1.06	0.087
Other tobacco use (ref: No)				0.162
Yes	0.73	0.48	1.13	0.162
Home tobacco use (ref: No)				0.822
Yes	1.03	0.78	1.36	0.822
Social media use (ref: Low)				0.003
Medium	1.85	1.29	2.64	0.001*
High	1.74	1.15	2.62	0.008*
Highest parent education level (ref: High school or less)				0.005*
Some college with technical/vocational school	0.72	0.50	1.04	0.081
Bachelor’s Degree	0.98	0.66	1.44	0.903
Graduate Degree	1.38	0.91	2.09	0.126
Household income (ref: Less than $50,000)				0.606
$50,000 to $74,999	0.84	0.57	1.25	0.399
$75,000 or more	0.87	0.64	1.19	0.387
Sexual attraction (ref: All other)				0.212
Only attracted to opposite sex	0.84	0.64	1.11	0.212

^aEstimates and confidence intervals were exponentiated to provide results interpretable in context; ^bP-values corresponding to overall, Type III tests are presented in the header of each row. P-values corresponding to tests of individual levels vs. reference level are presented next to corresponding levels; *Denotes p<0.05; Note. Model also adjusts for reported recall of the decoy ad (ever vs never: 1.89; 95% CI: 1.43, 2.50; p<0.05 and experimental ad condition from the larger study, vaping vs smoking: 1.02; 95% CI: 0.80, 1.29; p = 0.902).