Which males and which females are favored by differentially functioning items? examining sources of response pattern heterogeneity within gender groups

In gender differential item functioning (DIF) research it is commonly assumed that all members of a gender group have similar item response patterns and therefore generalizations from group level to subgroup and individual level can accurately be made. However DIF items do not necessarily disadvantage every member of a gender group to the same degree, indicating existence of heterogeneity of item response patterns within gender groups. Previous research has examined heterogeneity in the whole sample but not within the manifest comparison groups such as gender. In this dissertation the impact of heterogeneity within gender groups on gender DIF investigations in reading and math contexts was investigated. Specifically, this dissertation examined: whether DIF results varied when comparing males vs. females, gender x socio-economic status (SES) subgroups and latent classes (LCs) of gender; and whether sources of heterogeneity varied across gender groups in reading and math assessments. DIF analyses were conducted on reading and math achievement data from the Canadian sample of the Programme of International Student Assessment (PISA) 2009. Results indicated considerable heterogeneity within males and females in the reading and math contexts and DIF results were found to vary when heterogeneity was taken into account versus when it was not. The strongest source of gender heterogeneity in reading and math response patterns was reading and math achievement, respectively. However other sources of heterogeneity were found, which varied between gender groups and across reading and math contexts. Unique sources of heterogeneity were found for reading which were not identified for math. This research indicates that males and females are heterogeneous groups with diverse educational, linguistic and socio-cultural backgrounds. By taking into account heterogeneity within gender groups information from DIF results is enhanced, that is, information is revealed on which males or which females are disadvantaged by DIF items. Therefore it is essential that the heterogeneity of gender groups is examined before any inferences about fairness or bias is made. Possible implications of the results and future research directions are discussed.Education, Faculty ofEducational and Counselling Psychology, and Special Education (ECPS), Department ofGraduat