Skip to main content
. 2023 Nov 3;9(44):eadg6799. doi: 10.1126/sciadv.adg6799

Fig. 3. Partisanship and preferences for content moderation for respondents who agree headlines are inaccurate.

Fig. 3.

No control variables; 95% confidence intervals are shown. The left panels show the coefficient estimates and confidence intervals from Eq. 1 for respondents who agree headlines are inaccurate and the removal (A), harm (C), and censorship (E) outcome. The right panels show the same estimates for respondents who agree headlines are inaccurate along with the overall gap between partisans for the removal (B), harm (D), and censorship (F) outcome (see section S1.3.1 for details).