English abstract
This study aims to investigate the Differential Item Functioning (DIF) across
gender in items of the mathematics tests in the TIMSS 2019 for fourth and
eighth-grade students using the Transformed Item Difficulty (TID) and the
Mantel-Haenszel (MH) methods. The two tests consist of 157 items for the
fourth grade and 195 items for the eighth grade, comprising multiple-choice and
open-ended items across14 blocks. The mathematics tests included 157 items
for the fourth grade and 195 items for the eighth grade, comprising multiplechoice and open-ended questions across 14 clusters. The study utilized archival
data from a sample of 6,814 fourth-grade students and 6,751 eighth-grade
students from 228 government, private, and international schools in Oman.
Based on the Transformed Item Difficulty (TID) and the Mantel-Haenszel (MH)
methods and in order to achieve the aims of the study to detect the Differential
Item Functioning (DIF) for both the fourth and eighth grades, DIF was
investigated using functions in R studio, and were calculated to determine the
percentage of agreement between the results of the two methods, along with
calculating the Kappa classification stability coefficient using SPSS
Additionally, the relationship between the direction of DIF (for male/for female)
and factors such as item difficulty level, item type, content domain, and
cognitive domain was examined using the Chi-square test in SPSS.. The
findings based on the transformed item difficulty method showed that 7.64% of
test items exhibited DIF in the fourth-grade mathematics test, while the eighthgrade test had a DIF rate of 9.23%. In contrast, the Mantel-Haenszel method
identified 17.2% of test items with DIF for the fourth grade and 26.15% for the
eighth grade. The agreement between the two methods in identifying DIF items
was 87.90% for the fourth grade and 80% for the eighth grade, indicating
moderate and acceptable agreement, respectively. Furthermore, the results
revealed a relationship between the direction of differential item functioning
(for male/for female) and factors such as item difficulty level in the fourth grade,
item type in the eighth grade, and cognitive domain in the eighth grade.
However, no significant relationship was found with the content domain
variable based on the data for both grades.
