Calculation in a so-called Bayesian situation is known to be challenging. According to a meta-analysis the performance is only 4% if the relevant information is given in form of probabilities. Still, previous research has identified helpful strategies for improving Bayesian reasoning (i.e., so-called natural frequencies and visualizations). Yet, it has hardly been studied, if and how students use these strategies themselves. Thus, in this paper we focus on the process instead of the result of students who work on a Bayesian situation. For that, we present a coding system for analysing the notes of the students and the strategies which they apply for the calculation. We identify that the the students hardly ever use the strategy of natural frequencies. Further, about half of the students use common visualizations such as a probability tree-diagram or a 2×2 table. Yet, completing the 2×2 table was error-prone as students often entered conditional (instead of joint) probabilities into the inner fields.