Electronic word of mouth (eWOM) plays a critical role in the contemporary online environment. Past studies mainly focused on the helpfulness of eWOM and explored its antecedents and consequences. However, fake reviews and manipulated reviews highlight a need to investigate the credibility of eWOM messages (information credibility). In addition, as the crawling data approach (secondary data) gets popular, different conclusions may be reach compared with traditional self-report primary data. We designed a meta-analysis that contains both meta-analytic structural equation modeling (MASEM) and bivariate meta-analysis to address these issues. The results first indicate the critical role of information credibility on eWOM induced decisions and then illustrate how the relationships between information helpfulness and its antecedents are contingent on data source.