Düşük Sosyoekonomik Statüye Sahip Öğrencilerin Başarısını Etkileyen Faktörlerin Belirlenmesi: Bayesyen Model Ortalama Yaklaşımı
Derya Topdağ, Ebru Çağlayan AkayAkademik başarı ve sosyoekonomik arka plan arasındaki ilişkinin analizi, eğitim araştırmalarında önemli konulardan biridir. Türkiye’de düşük sosyoekonomik statüye sahip öğrenci yüzdesinin uluslararası ortalamanın üstünde olmasına rağmen, bu öğrencilerin özellikle ortalama matematik başarı puanlarının uluslararası ortalama puanına göre nispeten yüksek olduğu görülmektedir. Bu makalenin amacı, Türkiye’de düşük sosyoekonomik statüye sahip öğrencilerin matematik başarısını etkileyen değişkenleri Bayesyen yaklaşımın sunduğu küçük örneklem boyutu ve modelleme esnekliğinden yararlanarak belirlemektir. Çalışmanın verileri Uluslararası Matematik ve Fen Eğilimleri Araştırması (TIMSS) 2019 sekizinci sınıf matematik değerlendirmesinden elde edilmiştir. Çalışmada, çok sayıda bağımsız değişken içeren büyük ölçekli eğitim verileriyle çalışırken hangi değişkenlerin modele dahil edilmesi gerektiğini belirlemek için Bayesyen model ortalama (BMA) yaklaşımı kullanılmıştır. Bayesyen model ortalama sonuçlarına göre, evdeki kitap sayısı, öğrencinin akademik beklentisi, okula ait hissetme, matematiğe karşı tutum, devamsızlık ve zorbalığa maruz kalma, matematik performansının en önemli açıklayıcıları olarak tespit edilmiştir. Düşük sosyoekonomik statüye sahip öğrencilerin matematik başarısızlığının okula ve matematik dersine karşı olumsuz tutumlar, zorbalığa maruz kalma ve artan ödev sıklığı ile yakından ilişkili olduğu belirlenmiştir. Ayrıca düşük sosyoekonomik statüye sahip öğrencilerin matematik başarısında annenin eğitim seviyesi ve cinsiyetin etkisi olmadığı tespit edilmiştir. Sonuçlar düşük sosyoekonomik statüye sahip öğrencilerin okul içinde ve okul dışındaki eşitsizlik unsurlarından etkilendiğini göstermektedir. Sonuç olarak, eğitim politikalarının sosyoekonomik eşitsizlikleri dikkate alarak düşük sosyoekonomik statüye sahip öğrenciler için fırsat eşitliği sunması beklenmektedir.
The Bayesian Model Averaging (BMA) Approach for Determining the Factors Affecting the Achievement of Students with Low Socioeconomic Status
Derya Topdağ, Ebru Çağlayan AkayAnalyzing the relationship between academic achievement and socioeconomic background is an important subject in educational research. Even though the percentage of students with low socioeconomic status in Türkiye is higher than the international average, these students’ average mathematics achievement scores can be shown to relatively higher than international average scores. This study aims to identify the variables that influence the mathematics achievement of students with low socioeconomic status in Türkiye using the small sample size and modeling flexibility provided by the Bayesian approach. Data were employed for this purpose from the 2019 International Survey of Mathematics and Science Trends (TIMSS) 8th-grade mathematics assessment. The study uses the Bayesian model averaging (BMA) approach to determine which variables should be included in the model when working with large-scale educational data and a large number of independent variables. According to the Bayesian model averaging results, the number of books at home, students’ academic expectations, sense of belonging to school, attitudes toward mathematics, absenteeism, and exposure to bullying are the strongest predictors of mathematics achievement. The findings from this study show the mathematics failure of students with low socioeconomic status to be closely associated with negative attitudes toward school and mathematics courses, exposure to bullying, and greater frequency of homework. Furthermore, the study has determined mother’s educational level to have no influence on the mathematics achievement of students with low socioeconomic status, while gender does have an effect in terms of father’s education level. The results show students with low socioeconomic status to be impacted by the components of inequalities inside and outside of school. Consequently, education policies are expected to provide equitable opportunities for students with low socioeconomic status by taking socioeconomic inequalities into account.
Ensuring the academic achievement of all students has now become necessary to satisfy the increasing demands of the global economy and improve people’s welfare. According to long-standing research, family is the most reliable predictor of a student’s academic and future achievement. Researchers commonly focus on socioeconomic inequalities regarding academic achievement. Socioeconomic inequalities generally explain differences in students’ reading and mathematics achievement across Organisation for Economic Co-operation and Development (OECD) countries. Many empirical studies have demonstrated that students with low socioeconomic status perform worse academically. Based on the findings, students with high socioeconomic status tend to outperform students with low socioeconomic status. In this framework, education systems are supposed to provide an equitable opportunity to students with low socioeconomic status by developing policies that deal with socioeconomic inequalities. The percentage of students in Türkiye with low socioeconomic status has been observed to exceed the international average. For these reasons, identifying the factors that influence the mathematical achievement of students with low socioeconomic status in Türkiye is critical for the education system and policymakers. This study aims to identify the variables that influence the mathematics achievement of students from low socioeconomic status in Türkiye using the modeling and small sample size flexibility provided by the Bayesian approach. For this purpose, data have been employed from the 2019 International Survey of Mathematics and Science Trends (TIMSS) eighth-grade mathematics assessment.
The Bayesian model averaging (BMA) approach is used to determine what variables should be included in a model when employing large-scale educational data. This approach takes into account model uncertainty in addition to parameter uncertainty. BMA considers the uncertainties in the model format and assumptions and incorporates them into inferences about the unknown parameter being studied. The BMA approach solves the problem of how to choose a model by incorporating multiple competing models into the estimating procedure. This study considers the BMA approach as a solution to the model uncertainty problem.
According to the analysis results, the strongest predictors of mathematics achievement are the number of books at home, students’ academic expectations, sense of belonging at school, attitudes toward mathematics, absenteeism, and exposure to bullying. The variable that most increases student performance is observed as the number of books, which as an indicator of home education resources. As expected, the frequency with which the test language is spoken at home is also found to have a positive and significant effect on mathematics achievement. The findings indicate the variables of academic expectation and positive teacher expectation to also be important in explaining mathematics achievement. In accordance with previous studies, these variables have been identified as the most influential factors increasing the performance of students with low socioeconomic status. Furthermore, having a positive attitude toward mathematics has been reported to positively influence student achievement, whereas having a negative attitude negatively influences student achievement. The findings show that having a negative attitude toward mathematics is the most harmful factor affecting mathematics achievement. On the other hand, the study suggests a student’s sense of belonging to school to decrease mathematics performance because of negative school-related thoughts. Increases in the frequency of homework and exposure to bullying (e.g., name calling) are shown to have a negative effect on students’ mathematics performance. Another important finding involves the mother’s level of education not being a significant factor in students’ mathematics performance, while having a father with a higher level of education is shown to increase those student’s mathematics performance. The fact that only the father’s education level has an impact on student performance demonstrates the presence of a gender effect. Overall, the results indicate that students with low socioeconomic status are impacted by the components of inequalities inside and outside of school.
