Research indicates that misinformation elicits strong negative sentiments, thereby boosting its traction and engagement on social media. However, data collection methods often sample extreme instances of misinformation, thereby misrepresenting the broader spectrum of news shared online. Studies also commonly overlook the difference between emotions in news and reactions to news, crucial for understanding engagement. Therefore, it’s unclear if observed dynamics are unique to misinformation or reflect common factors like political orientation or emotional content. Here, we studied 9M German Twitter (now X) discussions from October 2020 to March 2022, using machine learning to infer eight emotions and out-group references from the text. Approximating causal inference with a matching approach, we examined the effects of posts citing sources of varying trustworthiness and political orientations on online discussions. The first results show that reactions to misinformation may be driven by emotional and hateful content rather than the lack of truthfulness, suggesting limited effects of misinformation on social media.