Designing corrective messages to debunk misinformation online is an important practice toward ending the coronavirus disease (COVID-19) pandemic as health-related misinformation has proliferated on social media misguiding disease prevention measures. Despite research on the use of statistical evidence and message modality in persuasion, the effects of evidence type (assertions with versus without statistical evidence) and presentation mode (text-only versus image-only versus text-plus-image) have been understudied. This study examined the impact of evidence type and presentation mode on individuals' responses to corrective messages about COVID-19 on social media. The results showed that the presence of statistical evidence in assertions reduced message elaboration, which in turn reduced the effects of the message in correcting misperceptions, decreased perceived message believability and lowered social media users' intentions to further engage with and disseminate the corrective message. Compared to the text-only modality and the text-plus-image modality, the image-only modality triggered significantly lower levels of message elaboration, which subsequently heightened message believability and increased user engagement intentions. The theoretical and practical implications are discussed.