Statistical literacy is a complex developmental construct requiring both mathematical skills and contextual understanding. The development of statistical literacy is an important objective of classrooms where the curriculum is approached through considering problems that require the active engagement of learners with relevant social material. Such approaches are often advocated for the middle years of schooling. Little attention has been paid, however, to the effects of these approaches on male and female students. This paper reports on a study that considers Differential Item Functioning (DIF) with respect to gender of questions on a statistical literacy scale derived from archived data. Multi-faceted Rasch models were applied to polytomous data to determine the interactions between gender and item. Three criteria were applied to the results: statistical significance, replicability and substantive explanation of DIF. The results suggested that although there was no overall difference in the average performance of male and female students, items requiring numerical responses or calculations were less difficult for male students and, conversely, items demanding written explanations were less difficult for female students. The implications of these findings for both assessment and teaching are discussed.
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