Believability and conditional inference in mathematics undergraduates: a replication and extension

This paper replicates a study in which Alcock et al. (2026) found that content influences first-year mathematics undergraduates’ evaluations of conditional inferences . Alcock et al. reported that these students responded more normatively to inferences from highly believable (true) mathematical conditionals than from conditionals with other mathematical, everyday, or abstract content. We replicated this work with a larger and more diverse sample, finding a similar overall believability effect and evidence that some individual differences in the original study were due to recent teaching. We extended the work by relating conditional inference responses to prior performance in mathematics and to students’ cultural backgrounds, finding that better mastery of more mathematics is associated with better rejection of invalid affirmation of the consequent inferences and that students with English as a first language responded more normatively than students with Chinese as a first language, especially for modus tollens inferences. We consider limitations to generalisability, and argue for the importance of replicating effects of all kinds, not only those linked to teaching interventions.