A BAYESIAN CASE STUDY

There’s never been a clear sequence of experiments to show the effect of enthusiasm on student learning. Not without mixed results.

Ben Motz

Bayesian or frequentist, their outcome would have been the same: Spacing techniques, as expected, improved performance, but enthusiastic instruction did not. However, the Bayesian approach enabled the researchers to develop a much more informative analysis that would convey the study’s implications for student learning and future research. The results were published in the July 2017 issue of PLOS ONE.

As Motz explains, “We could have done what we consider traditional stats, where the main goal is to make a binary decision: Is there an effect of enthusiasm? But using Kruschke’s approach to Bayesian estimation, we were able to directly infer how much benefit students get from an enthusiastic instructor, making apples-to-apples comparisons with the benefit of spaced study.” In this case, where frequentist methods would have given limited information about the effect of enthusiasm, Bayesian statistical analyses painted a more detailed picture.

So, do we want to curb our enthusiasm for enthusiastic instruction? Maybe not so fast. But we should perhaps reallocate this enthusiasm, as a Bayesian might say, to Bayesian statistics and the science of learning.