Exploring the endorsement effect on scientific crowdfunding performance: Evidence from Experiment.com.

• Scientific crowdfunding campaigns with endorsements have better fundraising and participation performances. • Endorsement genre's analytical framework is constructed, including 2 dimensions and 11 attributes. • The "person-then-project" presentation pattern is superior to the "project-then-person" pattern in the balanced endorsement genre. • A stronger endorsement sentiment induces a better campaign fundraising performance. • The campaigns with a high endorsement sentiment have better fundraising and participation performances. In recent years, endorsers have been involved in many scientific crowdfunding campaigns to improve project progress. While the effectiveness of endorsement has been frequently examined in the marketing field, it lacks in-depth analysis in scientific crowdfunding. To reveal whether and how endorsement influences scientific crowdfunding performance, we employ signaling theory as a theoretical lens and analyze 1,839 campaigns from Experiment.com. First, the result from propensity score matching method indicates that endorsements generally have a positive effect on scientific crowdfunding performance. Using open coding, we refine the endorsement genre's analytical framework, employing 2 dimensions (project-related and person-related) and 11 relevant attributes. The regression results suggest that clearly stating the implications of the project can significantly affect both fundraising and participation performance. The impacts of the endorser's identity and endorser's agreement on the initiator's ability promote fundraising performance, while the research background and condition statement positively contribute to participation performance. Furthermore, we classify scientific endorsements into three genres: project-focused, person-focused, and balanced. Although the performance of the three types of endorsement genres does not differ significantly, we find that the "person-then-project" pattern is superior to the "project-then-person" pattern in the balanced endorsement genre. In addition, we calculate endorsement sentiment, which is also an essential attribute of the endorsement genre, using a deep learning approach. The regression result suggests that a stronger endorsement sentiment induces a better campaign fundraising performance. In this line, we identify two endorsement genres: high- and low-sentiment endorsements. The campaigns with a high endorsement sentiment have better fundraising and participation performances than those with a low endorsement sentiment. We subsequently discuss the theoretical and practical implications and future research directions. [ABSTRACT FROM AUTHOR]