Meta-analysis in second language research: Choices and challenges

Applied linguists are increasingly conducting meta-analysis in their substantive domains, because as a quantitative approach for averaging effect sizes across studies, it is more systematic and replicable than traditional, qualitative literature reviews. Additional strengths, such as increased statistical power, moderator analyses, and model testing, have also contributed to its appeal. The current review describes typical stages of a meta-analysis in second language acquisition (SLA) research: (a) defining the research domain, (b) developing a reliable coding scheme, (c) analyzing data, and (d) interpreting results. Each stage has a host of equally reasonable decisions that can be made; each decision will influence the conduct of the meta-analysis, the nature of the results, and the substantive implications of findings for SLA. We highlight a number of benefits and challenges that inform these decisions. In general, when a meta-analysis in applied linguistics is well planned, employs sound statistical methods, and is based on a thorough understanding of relevant theory, it can provide critical information that informs theory as well as future research, practice, and policy.