Yapay Zeka Hızlı Bakış & Çıkarımlar
AI InsightAs a result of research, it was discovered that there were differences with regard to the number of DIF items identified by three methods and the level of DIF that the items contained; however, methods were consistent in detecting uniform DIF.
📌 3 Maddede Bu Makale
Özet
The purpose of this study is to determine whether English test items of Undergraduate Placement Exam (UPE) in 2016 contain differential item functioning (DIF) and differential bundle functioning (DBF) in terms of gender and school type and examine the possible sources of bias of DIF items. Mantel Haenszel (MH), Simultaneous Item Bias Test (SIBTEST) and Multiple Indicator and Multiple Causes (MIMIC) methods were used for DIF analyses. DBF analyses were conducted by MIMIC and SIBTEST methods. Expert opinions were consulted to determine the sources of bias. Data set of the study consisted of responses of 59818 students to 2016 UPE English test. As a result of the analyses carried out on 60 items, it was seen that one item in translation subtest contained DIF favoring male students. In school type based analyses, it was concluded that there were nine DIF items in vocabulary and grammar knowledge subtest, six DIF items in reading comprehension subtest and four DIF items in translation subtest. Experts stated that one item containing DIF by gender was unbiased, and evidence of bias was found in thirteen of nineteen items that contained DIF by school type. According to DBF analyses, it was found that some item bundles contained DBF with respect to gender and school type. As a result of research, it was discovered that there were differences with regard to the number of DIF items identified by three methods and the level of DIF that the items contained; however, methods were consistent in detecting uniform DIF.
Abstract
The purpose of this study is to determine whether English test items of Undergraduate Placement Exam (UPE) in 2016 contain differential item functioning (DIF) and differential bundle functioning (DBF) in terms of gender and school type and examine the possible sources of bias of DIF items. Mantel Haenszel (MH), Simultaneous Item Bias Test (SIBTEST) and Multiple Indicator and Multiple Causes (MIMIC) methods were used for DIF analyses. DBF analyses were conducted by MIMIC and SIBTEST methods. Expert opinions were consulted to determine the sources of bias. Data set of the study consisted of responses of 59818 students to 2016 UPE English test. As a result of the analyses carried out on 60 items, it was seen that one item in translation subtest contained DIF favoring male students. In school type based analyses, it was concluded that there were nine DIF items in vocabulary and grammar knowledge subtest, six DIF items in reading comprehension subtest and four DIF items in translation subtest. Experts stated that one item containing DIF by gender was unbiased, and evidence of bias was found in thirteen of nineteen items that contained DIF by school type. According to DBF analyses, it was found that some item bundles contained DBF with respect to gender and school type. As a result of research, it was discovered that there were differences with regard to the number of DIF items identified by three methods and the level of DIF that the items contained; however, methods were consistent in detecting uniform DIF.