Paul Collett


No profile


Testing and Evaluation An online application for advancing quantitative data analysis more

Sat, Jul 9, 16:00-16:40 Asia/Tokyo

One aspect of reimagining language learning research involves new approaches to data analysis. For quantitative research into foreign language learning or teaching, the dominant approach is arguably inferential statistics for statistical significance testing. Here, a test statistic (p-value) is calculated from sampled data, and decisions on the variables being tested-whether to accept or reject them as in some way contributing to the processes under study-are made based on the calculated p-value. However, this approach has long been recognised by numerous methodologists and theorists as potentially flawed, possibly holding back much research from contributing to substantive theory creation. Alternative measures are recommended; if not rejecting the approach outright, it is suggested that the results of statistical significance tests are augmented with measures of effect size, confidence intervals, robust variations of inferential statistics, and data-rich graphical plots. This presentation introduces an online application designed to help researchers carry out quantitative analysis focused on these alternatives to significance testing. Aimed particularly at less-experienced researchers, it requires little more than the input of data for the output of a range of useful statistics and plots. The rationale behind and usage of the application will be covered.

Paul Collett