Analyzing the Effects of Test, Student, and School Predictors on Science Achievement: An Explanatory IRT Modeling Approach
[ X ]
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Assoc Measurement & Evaluation Education & Psychology
Access Rights
info:eu-repo/semantics/openAccess
Abstract
This study aimed to investigate the impact of item features (i.e., content domain), student characteristics (i.e., gender), and school variables (i.e., school type) on students' responses to a nationwide, large-scale assessment in Turkey. The sample consisted of 7507 students who participated in the 2016 administration of the Transition from Primary to Secondary Education Exam (TPSEE, referred to as IEOG in Turkey). Explanatory item response modeling was used for analyzing the effects of content domain, gender, school type, and their interactions on students' responses to the science items on the exam. Five explanatory models were constructed to examine the effects of the item, student, and school variables sequentially. Results indicated that female students were more likely to answer the items correctly than male students. Also, students from private schools performed better than students from public schools. In terms of content, the biology items appeared to be significantly easier than the physics items. All interactions between the predictors were significant except for the Gender x School Type and Content x Gender x School Type interactions. The interactions between the predictors suggested that test developers, teachers, and stakeholders should be aware of potential item-level bias occurring in the science items due to complex interactions among the items, students, and schools characteristics.
Description
Keywords
Explanatory item response modeling, science, school type, gender, content
Journal or Series
Journal of Measurement and Evaluation in Education and Psychology-Epod
WoS Q Value
N/A
Scopus Q Value
N/A
Volume
13
Issue
1
