The objective of this research is to examine dimensionality of the data set obtained from the test takers’ responses to the PISA 2012 reading literacy test by comparing item parameters, and item level model data fits estimated based on the two-parameter logistic model and the bifactor model. The PISA 2012 Reading Literacy Test Booklet 12, including fourteen items related to four reading texts, was conducted on a group of 284 students. Model comparisons were done based on item discrimination parameters, S-χ 2 item fit statistics, and the index of explained common variance calculated based on item parameters. Results of the analyses indicate that item discrimination parameters estimated on the general dimension are similar to the two-parameter logistic model item parameters. The bifactor model provided some improvement on the item level fit over the one-dimensional model, however this improvement is not meaningful. Both models produced similar results in terms of the item data fit. Based on these findings, it was concluded that the general dimension representing reading comprehension skill is the dominant dimension underlying the data, and the text effect is small enough to accept that the data holds (essential) the unidimensionality assumption.
The objective of this research is to examine the dimensionality of the data set obtained from the test takers' responses to the PISA 2012 reading literacy test by comparing item parameters, and item level model data fits estimated based on the two-parameter logistic model and the bifactor model. The PISA 2012 Reading Literacy Test Booklet 12, including fourteen items related to four reading texts, was conducted on a group of 284 students. Model comparisons were done based on item discrimination parameters, S-h 2 item fit statistics, and the index of explained common variance calculated based on item parameters. Results of the analyses indicate that item discrimination parameters estimated on the general dimension are similar to the two-parameter logistic model item parameters. The bifactor model provided some improvement on the item level fit over the one-dimensional model, however this improvement is not meaningful. Both models produced similar results in terms of the item data fit. Based on these findings, it was concluded that the general dimension representing reading understanding skill is the dominant dimension underlying the data, and the text effect is small enough to accept that the data holds (essential) the unidimensionality assumption.
Alan : Fen Bilimleri ve Matematik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|