Multilevel Analysis of Factors that Determine the Science Achievement of Fourth-grade Students in TIMSS 2019

Abstract

This study examined the factors that determined the science achievement of fourth-grade students on the Trends in International Mathematics and Science Study (TIMSS) 2019 in the USA. The data were retrieved from the TIMSS international database and imported to the R program for manipulation. The EdSurvey package was used to conduct multilevel analysis, taking plausible values of outcome variables, and weighting into account. Level 1 predictors included gender, sense of school belonging, no student bullying, instructional clarity, and attitude toward science. Level 2 predictors included instruction being affected by science resource shortages, school emphasis on academic success, a lack of school discipline problems, and students’ preexisting with literacy and numeracy skills. The findings include that the variance between schools equals 33% and boys outperformed girls in science. Students’ attitudes toward science, their confidence, and the school’s emphasis on academic success were the two strongest positive predictors of both individual student’s achievement and the school’s mean science achievement. The other predictive variables were found to have only trivially positive or no effect. The utilized models could explain the variance within and between schools by 6% and 39%, respectively. The implications for policymakers and educational practitioners include boosting up confidence in learning science and by actively engaging students in inquiry-based activities and providing instant and constructive feedback, encouraging, and treating all students regardless of gender identity equally in teaching and learning science and schools should emphasize academic success.