Table 1 Data sources summary.

News source	Data source	Number of documents	Mean words per document	Mean Vader words per document	Mean document sentiment	Proportion of documents negative	Mean Facebook posts about article	Mean tweets about article	Mean tweets and retweets about article	Mean tweet retweets	Mean tweet replies	Mean tweet likes
Daily mail	News articles	15,881	871.211	59.08	0.047	0.415	5650.936	16.246	67.903
	Facebook posts	89,742,513
	Tweets	258,157	16.672	1.492						3.179	1.06	6.287
	Tweets Vader	180,444	19.191	2.135	-0.114	0.646				3.315	1.208	6.813
Guardian	News articles	12,026	924.36	62.744	0.07	0.29	2671.536	50.242	210.641
	Facebook posts	32,127,893
	Tweets	604,212	19.244	1.6						3.193	0.605	8.431
	Tweets Vader	422,126	22.954	2.29	-0.016	0.519				3.772	0.684	9.83
New York Post	News articles	33,669	482.858	31.084	0.038	0.399	2525.039	14.723	85.498
	Facebook posts	85,015,524
	Tweets	495,777	14.589	1.253						4.807	3.109	13.318
	Tweets Vader	304,700	18.842	2.038	-0.095	0.615				5.489	3.475	15.298
New York Times	News articles	33,706	1216.875	76.29	0.056	0.33	10,867.595	137.514	903.802
	Facebook posts	366,303,163
	Tweets	4,635,405	17.648	1.4						5.572	1.353	15.542
	Tweets Vader	2,985,738	22.026	2.173	-0.016	0.518				7.206	1.729	19.866

Summary statistics for all data employed in the analysis prior to transformation. Information on the “Mean document sentiment” and “Proportion of documents negative” for tweets is provided only for subsets containing one or more words in the Vader dictionary: the “Tweets Vader” data sources. This approach was adopted to prevent these statistics from conveying an inaccurate impression of neutrality, which may have arisen because certain tweets contain minimal text. Summary statistics for data used to produce aggregate findings (reported in “Article Negativity and Sharing to Social Media”, and “Article Negativity, Out-Group Content, and Sharing to Social Media”) are provided in SI 10.

Quick links

Search