The theoretical shift known as the “complexity turn” in applied linguistics conceptualises English-medium education (EME) contexts as dynamic and interconnected systems in which educational outcomes function as emergent phenomena. However, empirical research on academic success continues to rely predominantly on variance-based methods, specifically the general linear model (GLM), a statistical approach designed to estimate average relationships across populations. To address this methodological homogeneity, the present study critically evaluates the analytical approaches and embedded causal assumptions underlying current EME research. Building on previous reviews of empirical findings, this methodological synthesis of 28 empirical studies (2014–2025) indicates a high degree of analytical convergence, with 89% of studies utilising variance-based regression approaches. While such variance-based methods are valuable for identifying average population-level trends, their predominant use points to a need for methodological diversification. To encourage such pluralism, this study proposes qualitative comparative analysis (QCA) as a methodologically congruent and complementary approach for questions concerning configurational complexity. Drawing on two empirical exemplars, the paper illustrates how QCA may extend the current analytical toolkit by capturing equifinality (multiple configurations sufficient for academic success) and causal asymmetry (configurations for academic non-success are not necessarily the inverse of those for academic success). Finally, the study outlines best practices for set-theoretic research, concluding that investigating complex learner trajectories benefits from a methodological toolkit incorporating both variance-based and case-oriented analytical logics.
Yu Tang (Thu,) studied this question.