Table 2.

Summary statistics.

	Intent to vaccinate self				Intent to vaccinate child*
	$N$	Mean	Prob. equals	% Highly	$N$	Mean	Prob. equals	% Highly
		intent	no segments	unlikely		intent	no segments	unlikely
Black	675	5.00	0.00	17%	221	4.65	0.04	22%
		(2.25)				(2.38)
Latinx	602	5.47	0.00	10%	103	4.72	0.15	18%
		(2.03)				(2.28)
Conservative	1174	3.66	0.00	33%	449	2.97	0.00	48%
		(2.38)				(2.30)
Religious	719	4.89	0.00	18%	332	4.31	0.00	28%
		(2.31)				(2.48)
Parent	1093	4.63	0.00	23%
		(2.44)
Overall	3668	5.18		15%	1032	4.16		30%
		(2.26)				(2.48)
A member of	2638	4.75	0.00	20%	788	3.87	0.00	34%
$⩾1$ segment		(2.36)				(2.47)
A member of	1030	6.29		4%	244	5.10		16%
no segments		(1.48)				(2.23)

Notes: Standard deviations in parentheses. Intent of 1 corresponds to “highly unlikely” to vaccinate, while 7 is “highly likely.” We intentionally over-sampled our demographics of interest, so our sample is not representative, and the means above are unweighted. Many respondents are in more than one segment (e.g., Latinx and Religious and Parent). Because respondents in different segments received different combinations of message elements, the means are not directly comparable. This table uses the sample for descriptive statistics (see Appendix Fig. 7 for the sampling flowchart).

* For intent to vaccinate child, the sample is restricted to parents; accordingly “a member of $⩾1$ segment” considers only the non-parent segments, as does “a member of no segments.”