Strategies to correct vaccine misinformation on social media for pregnant women and the impact of vaccine skepticism

Abstract

Health-related misinformation on social media may affect vaccination behavior, particularly among (soon-to-be) mothers. Research suggested different strategies to correct misinformation, but it is unclear which strategies work best for what group and in what situation. Addressing the call for more emotion-based debunking strategies, this study examined how text genre (narrative vs. expository) and harm presence (with vs. without harm-stressing messages) interact to affect emotional responses, and debunking efficacy in corrective texts about vaccination and reproductive health, specifically targeting pregnant or planning-to-be pregnant women (N = 432) with varying levels of vaccine skepticism. We further assessed social media engagement, and information-seeking intentions. In particular, harm presence was tested as a moderator in the relationship between text genre and emotional responses, which in turn, mediate outcomes such as engagement with corrective texts and further information-seeking intentions on social media. Results from an online experimental survey showed that, in general, corrective texts about COVID-19 vaccine misinformation were more effective in reducing misconceptions compared to control texts. For women not skeptical towards vaccination, narratives with harm-stressing messages (vs. no harm) induced most negative emotions, reducing debunking efficacy, social media engagement, and information-seeking intention. For women skeptical towards vaccination, narratives (vs. expository) elicited stronger negative emotions, irrespective of harm-stressing messages, leading to decreased debunking efficacy, social media engagement, and the intention to seek information. This study illuminates the importance of tailoring vaccination corrective texts for different vaccine skepticism groups, avoiding one-size-fits-all strategies and being mindful of strong negative emotions and their counter-persuasive impact.

Fig. 1

The estimation graphic shows the comparison of mean differences in debunking efficacy between corrective texts and non-corrective control text condition. The 95% confidence interval of mean differences is illustrated.

Fig. 2

Interaction effect of text genre and harm presence (with harm-stressing messages: read line; without harm-stressing messages: blue line) on emotional responses, along with the emotional responses based on vaccine skepticism (SKEP: pink points; NONSKEP: green points).

Fig. 3

Multiple moderated mediation analysis for NONSKEP. *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 4

Multiple moderated mediation analysis for SKEP. *p < 0.05; **p < 0.01; ***p < 0.001.

Fig. 5

The Facebook post containing the vaccination misinformation.