From:  Analyzing tweets before and after Meta’s graphic self-harm imagery ban: a content analysis

 Tweet counts, organized by timing, slant, and type. The chi-squared test evaluates whether the proportion of tweets differs pre-ban and post-ban for each slant, category, and theme

InformationTotal n (%)Pre-banPost-banχ² (df), adjusted p
Slant
Anti2,611 (67.89)1,4231,18850.37 (1), < 0.001
Neutral1,128 (29.33) 46965950.85 (1), < 0.001
Pro107 (2.78) 52550.10 (1), 0.796
Category
Advice-seeking120 (3.12)289235.58 (1), < 0.001
Informative939 (24.41)4414986.19 (1), 0.018
Jokes/ridicule72 (1.87)40320.55 (1), 0.511
Personal account/experiences969 (25.20)59237757.09 (1), < 0.001
Personal opinions1,084 (28.19)46462035.76 (1), < 0.001
Other662 (17.21)37928314.06 (1), < 0.001
Theme
Accepting self-harm6 (0.16)511.44 (1), 0.271
Calling out wrongdoings488 (12.69)355133109.18 (1), < 0.001
Comparisons153 (3.98)1084524.78 (1), < 0.001
Feedback on media portrayal133 (3.46)75581.65 (1), 0.249
Political637 (16.56)40023745.23 (1), < 0.001
Reasons for self-harm168 (4.37)1333556.38 (1), < 0.001
Repercussions of self-harm12 (0.31)392.20 (1), 0.184
Self-harm resources106 (2.76)802626.08 (1), < 0.001
Self-harm terms48 (1.25)24240 (1), 1
Support514 (13.36)29821612.76 (1), < 0.001
Understanding self-harm1,581 (41.11)4631,118484.0 (1), < 0.001