The evolution of quantitative methods in Applied Linguistics: A longitudinal case study

Although improvements in research designs, quality, statistical rigor, and analytic complexity are often assumed in applied linguistics, empirical evidence supporting these claims remains scarce. This paper examines the evolution of quantitative methods in applied linguistics research over a 29-year period, using a corpus of data from the journal Applied Linguistics as a case. It examines longitudinal trends in (a) methodological features, (b) research designs and statistical analytic procedures, and (c) rigor/significance reporting practices. Specifically, it explores changes in sampling, number of variables, and questions/hypotheses, data sources, research design types, analytic procedures, instrument validation, and reporting practices such as effect sizes, p-values, and confidence intervals. Using descriptive summaries, chi-square tests, standardized residuals, Cochran-Armitage trend tests, selected time-series and regression models, Benjamini-Hochberg FDR adjustment, and Cramer's V effect-size summaries, we examine whether coded methodological and reporting practices changed over time. For multi-entry variables, descriptive tables summarize instance-level composition and supplementary figures provide optional graphical views, whereas inferential analyses are based on article-level category-presence indicators so that each article contributes at most one observation per category. The findings show several statistically detectable shifts in quantitative practices within this journal corpus, including changes in sample-size distributions, data-source types, analytic-procedure families, instrument-validation labels, and statistical/effect-size reporting labels. These findings are interpreted as changes within the Applied Linguistics corpus rather than as causal evidence of field-wide methodological improvement.