According to Wollack and Schoenig (2018), score differencing is one of six types of statistical methods used to detect test fraud. In this paper, we suggested the use of Bayes factors (e.g., Kass & Raftery, 1995) for score differencing. A simulation study shows that the suggested approach performs slightly better than an existing frequentist approach. We also demonstrate the usefulness of the suggested approach using a real data example. [This is the advance online version of a paper published in "Behaviormetrika" (2020).]