An exploratory analysis of revision behavior development of L2 writers on an intensive English for academic purposes program using Bayesian methods

Revision is a fundamental part of the writing process and is particularly important in the production of high-quality academic writing. This study is an exploratory examination of changes in revision behavior, as measured by keystroke logging software, at the beginning (T1) and end (T2) of a one-month intensive English for Academic Purposes (EAP) course on n = 39 undergraduate and postgraduate students. Bayes Factors (BFs) are utilized as measures of strength of evidence for changes in behavior. In this paper, we examine the application of a Bayesian Hypothesis Testing (BHT) approach and its implications specifically for exploratory studies, i.e., studies with relatively small samples intended to search data for emergent patterns. The results show that, in most cases, we have moderate evidence against any change in behavior over time. Based on this evidence, we conclude that the experimental parameters of further exploratory work into the development of revisions should be modified to maximize the chance of finding patterns in the data from which to generate any confirmatory hypotheses.