Rethinking the Negativity Bias Online: The Heterogeneous Effects of Negative Language on Online News Engagement

Negative language is widely assumed to boost online engagement — but does it always work? Across more than 150,000 randomized headline A/B tests on 398 news platforms and five experiments (N = 8,369), we find that the effect of negative tone and emotion depends critically on readers' goals. Field data reveal no average effect but stark cross-platform heterogeneity, unexplained by content or platform structure. Experiments show this heterogeneity reflects a motivational distinction: under a credibility-seeking goal, negative language triggers perceptions of manipulative intent, suppressing engagement; under an enjoyment-seeking goal, the same language may increase appeal. Moreover, negative emotion heightens affective polarization while negative tone reduces partisan favoritism among credibility-seekers. The match between communication style and reader goals is essential in a fragmented media environment.