Table 2.

Summary of negative binomial regression model predicting count of shared misinformation tweets^a, negative binomial regression model predicting count of shared control tweets, and linear regression model predicting ratio of shared misinformation and control tweets in study 1.

	Misinformation				Control				Discernment
	Estimate (SE)	IRRb	P value	Estimate (SE)		IRR	P value	Estimate (SE)		P value
Intercept	1.60 (0.56)	4.97	.004 c	1.20 (0.62)		3.32	.05	1.91 (0.72)		.008
Group: social cue	−0.40 (0.21)	0.67	.06	−0.54 (0.23)		0.58	.01	−0.26 (0.27)		.33
Group: misinformation flag	−0.43 (0.21)	0.65	.03	−0.78 (0.23)		0.46	.001	−0.25 (0.27)		.35
Group: combined	−0.66 (0.23)	0.52	.004	−0.32 (0.24)		0.72	.18	−0.75 (0.29)		.01
Age	−0.02 (0.01)	0.98	.01	−0.06 (0.01)		0.94	<.001	−0.01 (0.01)		.63
Education	−0.02 (0.05)	0.98	.70	0.00 (0.06)		1.00	.99	−0.03 (0.07)		.70
Digital literacy	−0.08 (0.08)	0.93	.36	−0.05 (0.09)		0.95	.55	−0.10 (0.11)		.34
Political orientation	0.09 (0.03)	1.09	.009	0.11 (0.04)		1.11	.003	0.06 (0.04)		.15

^aEstimated model 1: θ=0.25; and model 2: θ=0.26.

^bIRR: incidence rate ratio (percentage change in the dependent variable per 1-unit change in the predictor, either >1 or <1).

^cP values are italicized if significant at P<.05.