Investigating the role of computerized assessment and other correlates on students' science performance in PISA 2015

Abstract

The aim of this study is to explore possible reasons for the decline in the Program for International Student Assessment (PISA) 2015 average science scores. Until 2015, the test was paper-based, but in 2015, for the rst time, students took the test via computer. The decline was investigated based on data from Qatar, Peru, Turkey and Ireland as these countries represented various science scores and information and communication technologies (ICT) resources. By controlling student- and school-level variables, two-level regression was conducted for the analysis, taking into account plausible values and sample weights. The results indicated that there was a statistically signi cant positive relationship between ICT resources and science performance for Turkey and Peru, but not for Qatar and Ireland by controlling student- and schoollevel variables. On the other hand, almost all of the variables related to student characteristics were statistically signi cant predictors of science performance in these four countries, but signi cant variables related to school characteristics were di erent for each country. The science performance di erence among schools was larger in Turkey compared to other countries. Overall, it could be concluded that the decrease in the PISA 2015 average science scores could be attributed to the computer based version of PISA as ICT resources were signi cant predictors for countries with low ICT resources when controlling major student and school level variables.