Marginal effect comparison of model estimates for content-feature variables of created (original tweets, replies, and quotes) posts. A: Results of the logistic regression model 2 for political and entertainment content, outrage, sentiment (coded negative to positive), the top three emotions, and presence of an external URL. B: Resulting marginal effects of the hurdle lognormal model 3 for subsequent retweets per category of motive. C: Resulting marginal contrasts from the hurdle lognormal model for the full sample of posts containing original content.
Marginal effect comparison of model estimates for content-feature variables of created (original tweets, replies, and quotes) posts. A: Results of the logistic regression model 2 for political and entertainment content, outrage, sentiment (coded negative to positive), the top three emotions, and presence of an external URL. B: Resulting marginal effects of the hurdle lognormal model 3 for subsequent retweets per category of motive. C: Resulting marginal contrasts from the hurdle lognormal model for the full sample of posts containing original content.
