This study is carried out in three stages. First, students' affective characteristics which may possibly explain the difference in the answering a science item correctly or wrong are determined according to the sex of the students who are at the same level of ability. The status of identified affective characteristics' forming latent classes is determined in the second stage. In the last stage, it is aimed to determine whether or not the same items display differential item functioning (DIF) in the emerged latent classes. The study group of this research, which is in a descriptive survey model consists of 875 students. In the first staged of data analysis, latent class analysis is used to determine the latent classes that are formed according to the students' affective characteristics. In the second phase of the analysis, Mantel-Haenszel method is used in order to determine the state of differential functioning of the items for the whole group and in the latent classes emerged according to the students affective characteristics. According to the fact that the DIF analyses are carried out for the entire group, DIF is detected only in one of the items. However, it is found out that when the students are classified into latent classes according to some affective characteristics, this item did not display DIF in two latent groups and in one class, another item also displayed DIF.
Key words: Differential item functioning, affective characteristics, latent class analysis, TIMSS, science
|
