Due to the recent “reproducibility crisis”, the methodology of psychology has been changing. In this situation, the statistical education for undergraduate students majoring in psychology should also be changed. Statistical hypothesis testing and p-values have been the main topics in conventional statistics classes for psychology students. However, it has been pointed out that the emphasis on p-values has lowered the reproducibility of science, including psychology. In this paper, I propose that a main theme of statistics classes for psychology students should be shifted from statistical hypothesis testing to the concept of statistical modelling. Understanding the concept fitting probability distribution to data is a fundamental concept of inferential statistics. This idea will facilitate students’ understanding of some different statistical approaches. It will also be important to teach the students to refrain from making a dichotomous decision based upon a single index.