Table 4.

Estimated means (95% CIs) of fixed effects from linear mixed models analyses (with random intercepts by the participant) of agreement with misinformation beliefs by study month and estimated mean differences (95% CIs) for pairwise comparisons to Round 1 (April).

Misinformation belief (scale range: 1-7)	April: Round 1 (n=2006)	May: Round 2 (n=1882)			June: Round 3 (n=1369)
	Mean (95% CIs)	Mean (95% CIs)	Mean differencea (95% CIs)	P value		Mean (95% CIs)	Mean differencea (95% CIs)	P value
Data about the effectiveness of vaccines is often made up	2.37 (2.30, 2.44)	N/Ab	N/A	N/A		N/A	N/A	N/A
Herd immunity would be beneficial for COVID-19 and this fact is covered up	2.52 (2.46, 2.59)	2.39 (2.32, 2.46)	−0.13 (−0.19, −0.07)	<.001		N/A	N/A	N/A
The threat of COVID-19 is greatly exaggerated	1.99 (1.93, 2.05)	2.07 (2.01, 2.14)	0.08 (0.03, 0.13)	.002		2.04 (1.98, 2.10)	0.05 (−0.01, 0.10)	.11
Government restrictions are stronger than is needed	2.14 (2.08, 2.21)	2.16 (2.09, 2.22)	0.01 (−0.04, 0.07)	.67		2.19 (2.12, 2.25)	0.04 (−0.02, 0.11)	.19

^aMean difference compared to Round 1 completed in April 2020.

^bN/A: not applicable.