Towards simpler and more transparent quantitative research reports

Abstract

<jats:title>Abstract</jats:title> <jats:p>The average quantitative research report in applied linguistics is needlessly complicated. Articles with over fifty hypothesis tests are no exception, but despite such an onslaught of numbers, the patterns in the data often remain opaque to readers well-versed in quantitative methods, not to mention to colleagues, students, and non-academics without years of experience in navigating results sections. I offer five suggestions for increasing both the transparency and the simplicity of quantitative research reports: (1) round numbers, (2) draw more graphs, (3) run and report fewer significance tests, (4) report simple rather than complex analyses when they yield essentially the same results, and (5) use online appendices liberally to document secondary analyses and share code and data.</jats:p>

Misconduct and Questionable Research Practices: The Ethics of Quantitative Data Handling and Reporting in Applied Linguistics

“…NHST was also connected to what results researchers report, with failing to report nonsignificant findings or effect sizes for nonsignificant comparisons among the most common QRPs. With these results in mind, we whole‐heartedly join the many calls from colleagues across the social sciences to reduce our reliance on NHST and focus instead on aiming for precision through point estimates and their associated confidence intervals (see, e.g., Gass et al., 2021; Norouzian, 2020; Vanhove, 2020).…”

Determining the attributes that influence students’ online learning satisfaction during COVID-19 pandemic

“…A descriptive statistical exploration was also made to profile the respondents for the study. The results of these analyses are presented subsequently after a transparent description of data collection and analysis as well as reliability and content validity checks of the data are completed ( Vanhove, 2021 ).…”

Repetition and Incidental Learning of Multiword Units: A Conceptual Multisite Replication Study of Webb, Newton, and Chang (2013)

“…In order to answer our research questions, we ran two multilevel logistic regressions, one for each test (form recall and form recognition), with item-level gain score as the binary outcome, and repetition, learners' prior vocabulary knowledge (VLT score), education level, L1, and pretest score as predictors. To tap into vocabulary growth, we used a binary gain score as the dependent variable (see also Vanhove, 2021). In cases where the pretest and posttest scores for an item were identical (0-0 or 1-1), the gain score was 0.…”

“…Both congruency and transparency have been found to affect the learning of MWUs (Puimège & Peters, 2020). However, because our main focus was not on item-related variables, we included this secondary analysis in the supplementary online materials and not in the main text (see also Vanhove, 2021). 4 The following two test formats from the initial study were not used in our replication study: the productive form test and the meaning recall test (labelled…”

“…Both congruency and transparency have been found to affect the learning of MWUs (Puimège & Peters, 2020). However, because our main focus was not on item‐related variables, we included this secondary analysis in the supplementary online materials and not in the main text (see also Vanhove, 2021).…”