2019 TIMSS ve eTIMSS 4. ve 8. sınıf matematik ve fen bilimleri başarı testlerinin ölçme değişmezliği ve madde yanlılığının incelenmesi
Küçük Resim Yok
Tarih
2023
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Ege Üniversitesi
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Bu araştırmada; TIMSS 2019 4. ve 8. sınıf matematik ve fen bilimleri başarı testlerinden elde edilen puanların çalışmaya kâğıt-kalem ve bilgisayar ortamında katılan ülke grupları için ölçme değişmezliği ve değişen madde fonksiyonu yönünden incelenmesi ve karşılaştırmalı ilişkinin ortaya konması amaçlanmaktadır. Araştırma 1. kitapçığı kullanan 9560 öğrenci verisiyle gerçekleştirilmiştir. Matematik ve fen bilimleri alt testlerinin analizlerinde ölçme değişmezliği için Çoklu Grup doğrulayıcı Faktör Analizi (ÇGDFA) için Mplus 7. programı, DMF analizleri için ise Mantel-Haenszel (MH), Lojistik Regresyon (LR) ve SIBTEST yöntemleri için difR paketini içeren R Yazılımı kullanılmıştır. ÇGDFA analizleri öncesinde IEA'nın IDB analyzer programıyla ülke verileri analize hazır hale getirilmiş ve DFA analizi ile model doğrulanmıştır. Çalışmada ÇGDFA sonucunda tüm alt test karşılaştırmalarında katı değişmezlik sağlanmıştır. DMF analizleriyle çalışmaya kâğıt-kalem ve bilgisayar ortamında katılım şekline göre incelemeler gerçekleştirilmiştir. 4. sınıf matematik alt testinde MH ve SIBTEST yöntemlerinde 3, 7 ve 21 numaralı maddeler DMF'li olarak bulunmuş ve B düzeyinde DMF göstermiştir. 8. sınıf matematik alt testinde MH ve SIBTEST yöntemlerinde 5, 8, 9 ve 29 numaralı maddeler DMF'li olarak bulunmuştur. 4. sınıf fen bilimleri alt testi DMF sonuçları incelendiğinde; MH ve SIBTEST yöntemlerinde 23 numaralı madde B düzeyinde DMF'li olarak gözlemlenmiştir. 8. sınıf fen bilimleri alt testi MH ve LR yönteminde DMF gösteren madde gözlemlenmediği, SIBTEST yönteminde ise dört maddenin (3, 8, 9 ve 29) B düzeyinde DMF'li olduğu tespit edilmiştir. MH ve SIBTEST teknikleri yapılan DMF analizleri açısından birbirleriyle genelde uyum içindeyken, LR tekniği ile aynı sonuçlara ulaşılamamış ve bu iki yöntemle uyumsuz olduğu gözlemlenmiştir.
In this study; It is aimed to examine the scores obtained from the fourth and eighth grade mathematics and science achievement tests of TIMSS 2019 in terms of measurement invariance and DIF for the country groups participating in the study in paper and computer based, and to reveal the comparative relationship The research was carried out with the data of 9560 students using the 1st booklet. In the analysis of mathematics and science subtests, Mplus 7 program was used for Multi-Group Confirmatory Factor Analysis (CGDFA) for measurement invariance, and R Software, which includes the difR package for Mantel-Haenszel (MH), Logistic Regression (LR) and SIBTEST methods, was used for DMF analysis. Before the MGCFA analyzes, the country data was prepared for analysis with the IEA's IDB analyzer program and the model was verified with the CFA analysis. In the study, strict invariance was achieved in all subtest comparisons as a result of MGCFA. In the DIF analyzes, on the other hand, it was examined according to the way of participating in the study in paper and computer based. In the fourth grade mathematics subtest, items 3, 7 and 21 in the MH and SIBTEST methods were found to have DIF and showed DIF at the B level. Items 5, 8, 9 and 29 in the MH and SIBTEST methods in the eighth grade mathematics subtest were found to have DIF. When the fourth grade science subtest DIF results are examined; In the MH and SIBTEST methods, item 23 was observed as having DIF at B level. It was determined that no item showing DIF was observed in the eighth grade science subtest MH and LR method, while four items (3, 8, 9 and 29) were found to have DIF at B level in the SIBTEST method. While the MH and SIBTEST techniques were generally compatible with each other in terms of DIF analysis, the same results could not be achieved with the LR technique and it was observed that these two methods were incompatible.
In this study; It is aimed to examine the scores obtained from the fourth and eighth grade mathematics and science achievement tests of TIMSS 2019 in terms of measurement invariance and DIF for the country groups participating in the study in paper and computer based, and to reveal the comparative relationship The research was carried out with the data of 9560 students using the 1st booklet. In the analysis of mathematics and science subtests, Mplus 7 program was used for Multi-Group Confirmatory Factor Analysis (CGDFA) for measurement invariance, and R Software, which includes the difR package for Mantel-Haenszel (MH), Logistic Regression (LR) and SIBTEST methods, was used for DMF analysis. Before the MGCFA analyzes, the country data was prepared for analysis with the IEA's IDB analyzer program and the model was verified with the CFA analysis. In the study, strict invariance was achieved in all subtest comparisons as a result of MGCFA. In the DIF analyzes, on the other hand, it was examined according to the way of participating in the study in paper and computer based. In the fourth grade mathematics subtest, items 3, 7 and 21 in the MH and SIBTEST methods were found to have DIF and showed DIF at the B level. Items 5, 8, 9 and 29 in the MH and SIBTEST methods in the eighth grade mathematics subtest were found to have DIF. When the fourth grade science subtest DIF results are examined; In the MH and SIBTEST methods, item 23 was observed as having DIF at B level. It was determined that no item showing DIF was observed in the eighth grade science subtest MH and LR method, while four items (3, 8, 9 and 29) were found to have DIF at B level in the SIBTEST method. While the MH and SIBTEST techniques were generally compatible with each other in terms of DIF analysis, the same results could not be achieved with the LR technique and it was observed that these two methods were incompatible.
Açıklama
Anahtar Kelimeler
Eğitim ve Öğretim, Education and Training
