In reaction to misuses and misinterpretations of p-values and confidence
intervals, a social science journal editor banned p-values from its pages. This study
aimed to show that education could address misuse and abuse. This study examines
inference-related learning outcomes for social science students in an introductory
course supplemented with randomization and simulation content. Learning gains
were measured across a suggested taxonomy of inference learning outcomes using
the Reasoning about P-values and Statistical Significance (RPASS-10) scale. Three
graphical comparisons of students’ Pretest and Posttest proportions were encoded by
learning gain or loss, an inference learning outcome taxonomy, or if a correct
concept or misconception was assessed. What students learned and the difficulties
that persisted shape recommendations for teaching and future research.
The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education