LEE- Economics-Master Degree
Bu koleksiyon için kalıcı URI
Gözat
Başlık ile LEE- Economics-Master Degree'a göz atma
Sayfa başına sonuç
Sıralama Seçenekleri
-
ÖgeThe effect of instruction time on student outcome by using first difference approach with TIMMS data(Graduate School, 2023) Özgen, Eylül Berke ; Aygün Hızıroğlu, Aysun ; 787719 ; Economics M.Sc ProgrammeHuman capital is one of the key elements of economic development. Upon the importance of human capital, some detailed assessments on student achievement enable researchers to measure subject-specific abilities of students over countries. From this perspective, the International Association for the Evaluation of Educational Achievement (IEA) conducts detailed large-scale assessments on 8th and 4th grade students, named TIMSS, over countries. While some countries' pupils and schools have high scores, the rest obviously falls behind. These academic achievement gaps of pupils between countries and within countries have different and reasonable explanations in the overall literature. Our study's purpose is to enlighten one of the possible reasons for score variation and propose policy advice based on empirical outcomes. Through this study, instructional time has been taken as one of the significant inputs in the education production function by including other control variables which were extracted from TIMSS 2019 dataset. Four different educational systems and countries were investigated and to avoid foregoing biases and obtain a causal relationship, first differences was used in a within-pupil setting. The focused dataset includes over 19,000 observations of pupils from four countries namely Canada, Azerbaijan, Slovak Republic, and Spain. The overall results which are based on the whole dataset show a significant and positive point estimate of the effect of instructional time. In this setting, the effect of mathematics instruction time on mathematics score was estimated as 0.050 point increase per minute/week and the effect of science instruction time on science score was estimated as 0.094 point increase per minute/week. In the country-specific analysis, three of four countries showed insignificant responses to the explanatory variable. Upon that outcome, in the second phase, only Canada which showed a significant and positive response to instructional time has been studied. In Canada, the effect of mathematics instruction time on mathematics scores was 0.061 per minute/week and the effect of science instruction time on science scores was 0.133 per minute/week. Afterwards, heterogeneity tests on gender and immigration status were examined. Outcomes showed that the group of girls and immigrants were affected more by instructional time additions than the boys and natives. Then, to check the correctness of the identification strategy, one placebo test was employed by using a uniformly random number generation method to produce the counterfeit instruction times and reallocate them with the original ones. The test resulted in insignificant outcomes in 98 out of 100 random variations in regressions as expected, meaning that significance of instructional time effect was not obtained based on chance or coincidence. Besides, each country's educational system was investigated as an interpretation by focusing on several key metrics including educational autonomy, government expenditure on education, teacher-pupil ratio, teacher certification, academic output in the educational research, curriculum comprehensiveness and teacher salary. Within the boundaries of dataset, as a quantitative approach, logit and probit regressions were run to obtain the explanation of productivity differences among educational systems. Correspondingly, adequate time on each of the topics during class hours was found foremost for instructional time productivity of achievement while the teacher-pupil ratio does not have any effect. As an inference, approaching a self-deterministic structure of governance of schools and increased research outcomes for education sciences in the academic area were found to be other possible important metrics for instructional time productivity. Overall, the importance of this study comes from contributions to the literature by complementing the previous findings in several ways. Firstly, the response to the increased instructional time was investigated in four diverse countries and educational systems by quantitative methods including General Linear Regression, First Difference regression and Logit and Probit regressions. Secondly, several variables for productivity of achievement to instructional time were proposed. Thirdly, detailed research for heterogeneous effects on the productive country, Canada, was accomplished. Moreover, the law of diminishing returns for instructional time was examined and possible evidence for an optimum time of instruction was presented. Fourthly, a placebo test using random number generation to check the robustness of the significance of the instructional time effect was applied. In conclusion, additional instructional time may not always result always with increased student achievement. Rather, the subject country should be investigated carefully by considering its status quo education system characteristics (autonomy, curriculum intensity on the given time, demographics and research on education). On the other hand, as a specific analysis for the productive country Canada, science and mathematics instructional time effects are significantly positive for all pupils but higher for girls and immigrants. Also, science instruction's effect is higher than mathematics in all subgroups. In the literature, a considerable number of studies focus on the same subject. However, as student achievement scores are the final products of piles of diverse and complex inputs, the results of these studies may include biases. That includes selection bias when pupils are sent to schools/classes depending on their ability on a specific subject. Moreover, as the education production function includes family, student, school, class, teacher and even environment-based inputs, empirical outcomes from other studies may suffer from omitted variable bias and measurement errors.