Table 3.

Negative binomial regression results by themes of tweets (n = 2900).

	Likes				Retweets
Variable/theme	β	SE	Exp(β)	P-value	β	SE	Exp(β)	P-value
User type
Individual (Ref)	—	—	—	—	—	—	—	—
Alcohol industry	−0.41	0.12	0.66	<.001	1.19	0.21	3.29	<.0001
News organization	−3.00	0.21	0.05	<.0001	−1.78	0.37	0.17	<.0001
Public health organization/entity	−2.18	0.22	0.11	<.0001	0.33	0.34	1.40	.332
Other	−1.65	0.09	0.19	<.0001	−0.16	0.17	0.85	.353
Sentiment
Positive/neutral	−0.31	0.08	0.73	<.0001	−0.04	0.14	0.97	.806
Negative	0.26	0.08	1.30	<.001	0.00	0.14	1.00	.988
Tweet themes
Encouragement and support	−0.97	0.10	0.38	<.0001	0.01	0.17	1.01	.964
Experimentation with and promotion of nonalcoholic drinks	−1.15	0.10	0.32	<.0001	−0.29	0.18	0.75	.122
Tweet type
Commercial	−0.33	0.09	0.72	<.001	0.56	0.16	1.75	<.001
News	−2.50	0.15	0.08	<.0001	−1.24	0.25	0.29	<.0001
Humor	0.37	0.10	1.45	<.0001	−0.11	0.19	0.89	.546
Benefits
Any benefits	−0.10	0.12	0.33	<.0001	−0.55	0.21	0.58	.008
Physical	−1.11	0.12	0.33	<.0001	−0.54	0.22	0.58	.014
Psychological and emotional	−1.51	0.18	0.22	<.0001	−0.57	0.31	0.56	.066

β, estimated coefficient; SE, standard error of β; bold values denote statistical significance at the p < 0.05 level.