Conceptualization and frequency of sampling (mal)practices in L2 inferential quantitative research

Abstract Poor sampling practices can constitute a questionable research practice when conducting L2 inferential quantitative research. The current study, a methodological synthesis ( N = 433 Scopus/Web of Science (WoS) reports: cluster random sampling) of sampling practices, revealed that L2 inferential quantitative researchers rarely employed randomized and/or effect size-driven sampling processes with only eight (1.8%) and ten (2.3%) of the reports being respectively satisfactory. Furthermore, just 33.9% of the reports featured multisite (convenience) samples. In models assessing what predicted multisite sampling, whether the report was ISLA-focused ( r s = −.33, p < .001) or single-authored ( r s = −.15, p < .001) incurred moderate and weak negative associations. Citation analysis metric values and the Scopus/WoS contrast had no associations. The findings of this study suggest the field’s sampling practices have room to improve and guidance for future improvement is offered.