TIMSS 2015 Türkiye Örnekleminde Erken Matematik Aktiviteleri Modelinin Bölgelere ve Cinsiyete Göre İncelenmesi

Abstract

The purpose of the study is to investigate the measurement invariance of the early mathematical activities model, which was used in the TIMSS 2015 study and which was created using the home questionnaire applied to the parents of 4th grade students, by gender and regions. Measurement invariance was investigated at scale and item level among groups. The sample of the study consists of 6456 fourth grade students who participated in TIMSS 2015 research. The data obtained from the Ministry of National Education General Directorate of Measurement, Evaluation and Examination Services were used. SPSS 23.0, Lisrel 8.0 and Latent Gold 5.1 software were used in the analysis of the data. Within the extent of the research, a model for exploratory factor analysis and confirmatory factor analysis was formed for the items related to the early mathematical activities. In the following phase, the number of suitable classes was determined by latent class analysis. Continuously, the measurement invariance by regions and gender was investigated by the multigroup latent class analysis method to the early mathematical activities model. As stated the results of the research, it was decided that the three latent class models are the most suitable model according to the information criteria BIC and CAIC. When the measurement invariance is examined according to regions and gender, three latent class homogeneous models were selected as the most appropriate model according to BIC and CAIC values in both separate analyzes. In the examinations made at the item level, when the homogenous model is taken as a comparative model, it is observed that the homogeneous model is more suitable in all the information criteria values in the items. After the acceptance of the homogeneous model, it was concluded that the model of early math activities is invariable at scale and item level according to regions and gender.