Inequalities in science education: Using PISA data to investigate the role of family and school factors

Motivation
Recent literature focuses on the relationship between and the impact of various education production function, such as school input and household, on academic achievement (Fryer, 2017). Existing studies found that teacher qualifications, teacher-student ratio, and class size may play important roles in student learning (Hammond, 2000; Hanushek, 1999; Hanushek, 2007). However, the evidence is mixed. Study in Kenya found that teacher hiring and contract system affects student test scores (Bold, et al., 2018). In Tanzania, school grant combined with teacher incentives show promising impact on student test scores (Mbiti, et al., 2019). Yet, the evidence on large fund for school is mixed (Glewwe and Muralidharan, 2016). Furthermore, the positive impact of increasing in expenditure per student on academic performance is limited (Hanushek, 2007).
In terms of curriculum and learning methods, Banerjee, et al. (2016) shown that "Teaching at the Right Level" was promising in increasing learning outcomes in India. The use of computer-aided learning then was proven to be effective to improve productivity in delivering education (Muralidharan, Singh, and Ganimian, 2019). Moreover, some studies found that cooperative and collaborative learning practices improve student achievement (Nichols, 1996; Vercellotti, 2017; Slavin, 1983).
However, little is known on how much learning activities at school affects student achievement in developed country setting and whether female student is affected differently by learning activities at school, compared to male student.

Expected Outcomes
This study aims to understand the effect of learning activities in science class and student's grouping on performance in science, math, and reading. Moreover, we will answer whether the effects are different among several groups of students: female and male student; state schools and independent schools; rural and urban, dense and less dense school community. Lastly, this study serves as an exploratory study to find some possible interventions to improve learning activities and supporting disadvantaged students in Scotland.

Estimation Method
This study will use rich Programme for International Student Assessment (PISA) scores and survey data. For the teacher's characteristics, we will collect an additional information on teacher education and certification, teacher training, and teaching experiences. We will also collect information on curriculum used by the teachers and the mode of assessment to ensure the representativeness of PISA score on student's performance.
The study will estimate the effect of learning activities on student achievement using Two-Stage Least Squares (2SLS) method, controlling for household characteristics. The first stage will determine how teacher characteristics affect learning activities in class, while the second stage aims to understand the relationship between learning activities and student achievement.

Additional Analysis
Using the similar PISA 2015 survey data, we will analyze whether student achievement is affected by student's behavior at school (e.g., skipping class, arriving late, students do not focus during the class); and peer effect of school with female-dominated student and male-dominated student.