The present research aims to examine whether the questions in the Program for the International Student Assessment (PISA) 2009 reading literacy instrument display differential item functioning (DIF) among the Turkish, French, and American samples based on univariate and multivariate matching techniques before and after the total score, which is the matching variable, is purified of the items flagged with DIF. The study is a correlational survey model research, and the participants of the study consist of 4459 Turkish, French, and American students who took booklets 1, 3, 4, and 6 in the PISA 2009 reading literacy measure. Univariate and multivariate (bivariate, trivariate, and quadrivariate) DIF analyses were performed through logistic regression before and after purifying the matching variable off the items displaying DIF. Literature was used to detect extra matching variables, and multiple linear regression analysis was carried out. As a result of the analyses, it was discovered that using extra matching variables apart from the total score reduces type I errors. It was also concluded that the exclusion of DIF items (removal of items with DIF) while calculating the total score led to variation in the number of questions detected as DIF and DIF levels of the items, although it did not yield consistent results.
Differential item functioning Multivariate matching Purified matching variable
The present research aims to examine whether the questions in the Program for the International Student Assessment (PISA) 2009 reading literacy instrument display differential item functioning (DIF) among the Turkish, French, and American samples based on univariate and multivariate matching techniques before and after the total score, which is the matching variable, is purified of the items flagged with DIF. The study is a correlational survey model research, and the participants of the study consist of 4459 Turkish, French, and American students who took booklets 1, 3, 4, and 6 in the PISA 2009 reading literacy measure. Univariate and multivariate (bivariate, trivariate, and quadrivariate) DIF analyses were performed through logistic regression before and after purifying the matching variable off the items displaying DIF. Literature was used to detect extra matching variables, and multiple linear regression analysis was carried out. As a result of the analyses, it was discovered that using extra matching variables apart from the total score reduces type I errors. It was also concluded that the exclusion of DIF items (removal of items with DIF) while calculating the total score led to variation in the number of questions detected as DIF and DIF levels of the items, although it did not yield consistent results.
Differential item functioning Multivariate matching Purified matching variable
İlgili araştırma, birinci yazarın ikinci yazar danışmanlığında hazırladığı tezden türetilmiştir. İlgili araştırma hazır veriler ve dokümanlar üzerinden yürütüldüğü ve canlılar ya da insanlar üzerinde araştırma gerçekleştirilmediği için etik kurul onayına ihtiyaç duyulmamaktadır. Bu araştırmada araştırmacılar etik ilkeleri dikkate aldıklarını ve hiçbir etik ihlalde bulunmadıklarını bildirirler.
Birincil Dil | İngilizce |
---|---|
Konular | Ulusal ve Uluslararası Başarı Karşılaştırmaları, Eğitimin Kültürler Arası Karşılaştırmaları:Uluslararası Sınavlar |
Bölüm | Makaleler |
Yazarlar | |
Erken Görünüm Tarihi | 21 Ekim 2024 |
Yayımlanma Tarihi | |
Gönderilme Tarihi | 10 Haziran 2024 |
Kabul Tarihi | 2 Eylül 2024 |
Yayımlandığı Sayı | Yıl 2024 Cilt: 11 Sayı: 4 |