Meta-analysis in second language acquisition research

Meta-analysis has become increasingly popular in second language acquisition research (SLA) and has provided valuable summative information about a number of key areas. There are, however, dangers. This article examines a number of key issues that need to be considered in conducting a meta-analysis – inclusiveness, the heterogeneity of language learners, the definition of the independent and dependent variables, the need to consider alternative explanations of observed effects, the importance of examining the quality of the studies included in the analysis, and the apples and oranges problem. These issues are illustrated in a discussion of number of SLA meta-analyses (e.g. Norris and Ortega, 2000; Plonsky, 2011; Qureshi, 2016; Spada and Tomita, 2010). The article concludes by suggesting a number of factors that need to be considered in deciding whether to conduct a meta-analysis and when carrying one out. I argue the need for systematic reviews but suggest that these can often best present their findings in narrative form rather than statistically. I also suggest that a preliminary narrative account of the findings of s systematic review can indicate whether a meta-analysis is appropriate.