In this paper, we use connections between (a) interviews with physics education researchers and (b) methodological literature in education research more broadly to answer questions about research practices in physics education research (PER). In particular, we define two paradigms in PER – case-oriented PER and recurrence-oriented PER – in terms of assumptions about the social world and about what counts as rigorous or trustworthy when it comes to research accounts. Case-oriented PER is predicated on the assumptions that (1) social actions are guided by the meanings that people are making of their local environments and that (2) reality is subjectively constructed. Recurrence-oriented research, on the other hand, is predicated on the assumptions that (1) human behavior is guided by predictable relationships between variables and that (2) real phenomena are reproducible. We will draw connections between (i) these two different sets of assumptions and (ii) differences in selection practices, generalization practices, and causal-claims-making in our field.