Determination of the Characteristics Predicting Science Achievement through the Classification and Regression Tree (CART) Method: The Case of TIMSS 2015 Turkey

İzzettin Aydoğan, Selahattin Gelbal

Abstract

In the current study, it was aimed to determine student, teacher and school characteristics that predict science achievement of eight grade students in Turkey. In the study, the data of TIMMS 2015 were used and the study group was comprised of a total of 6079 students and 220 teachers from 218 different schools. As the data collection tools, the eighth grade science achievement test used in TIMMS 2015 and the scales administered to students and teachers and reflecting student, teacher and school characteristics were used. Since there was a multi-level data structure where students were nested in schools, the created model was analyzed by using the RE-EM algorithm, which enables multi-level data structures to be analyzed through the classification and regression tree (CART) method. The predicted variable of the model was students’ science achievement scores and the predictor variables were the seventeen student, teacher and school characteristics expressed in the scales. According to the results obtained, it was determined that five of the seventeen predictor variables predicted the students’ science achievement, which are students confident in science, student bullying, teaching limited by student needs, school discipline problems and school emphasis on academic success. It has been observed that students who have students confidence in science, level of bullying they are exposed to, emphasis on academic success in their schools, school discipline problems in their schools are more, and whose teachers stated that they have more teaching limited by student needs, are more successful.

Keywords

Science achievement, Predictor variables, Student, teacher and school characteristics, Classification and regression tree (CART), RE-EM algorithm


DOI: http://dx.doi.org/10.15390/EB.2022.9368

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.