Table 5 Summary of the existing domain political leaning scores and their correlations with our audience partisanship scores.
From: DomainDemo: a dataset of domain-sharing activities among different demographic groups on Twitter
Dataset | Description | N | Party | Party reg |
|---|---|---|---|---|
Bakshy et al.35 | Audience-based scores crafted from Facebook data. | 398 | 0.940 | 0.929 |
Eady et al.59 | Media ideology scores based on Twitter data. The authors jointly estimate the ideology of politicians, users and news sources through the news sharing behaviors on Twitter. | 179 | 0.929 | 0.929 |
Buntain et al.55 | Audience-based scores derived from the Facebook URL dataset. | 2,480 | 0.916 | 0.898 |
MBFC (mediabiasfactcheck.com) | MBFC (Media Bias Fact Check) provide rater-based political leaning categories for various news domains. We map their labels “far-left,” “left,” “center-left,” “center,” “center-right,” “right,” and “far-right” to numerical values − 1, − 0.66, − 0.33, 0, 0.33, 0.66, 1 for analysis. | 2,986 | 0.765 | 0.774 |
Allsides (allsides.com/media-bias/media-bias-ratings) | Allsides produces domain bias scores based on their own algorithm. We map their labels “left,” “left-lean,” “center,” “right-lean,” and “right” to numerical values − 1, − 0.5, 0, 0.5, 1 for analysis. | 189 | 0.736 | 0.743 |
Allsides community | Similar to the Allsides algorithmic scores, but based on crowdsourced ratings. | 189 | 0.613 | 0.611 |
Mturk49 | Crowdsourced ratings from Mturk. | 358 | 0.486 | 0.491 |
