I am currently working on a paper that uses within student between subject approach. I never used this approach before, and I am struggling with commands to estimate it.
I have a standardized test survey that evaluate students' knowledge in mathematics and language through a combination of multiple-choice and open-response questions. My sample uses 10 countries with several control variables and two variables of interest: student test scores ( mathematics and reading), and another variable that I called x1 (dummy variables). I have one wave (2019).
From what I understood, the first step: two estimations have to be done, one using the fixed effect and a pooled old estimation. So, I estimated the two after setting my data panel.
- The fixed effect one: xtreg y x1 x2 x3.... fe, (x2, x3,...) are set of control variables that I use
- Ols: reg y x1 x2 x3....
The second step: for this step, I have to estimate the within student between subject approach, where the dependent variables is the test score difference between the two subjects, and the independent variables are the differences between my control variables, that is teacher characteristics (experience, teacher's age, teacher holds a postgraduate degree....).
within student between subject approach:
test score math - test score reading= coefficient1 (teacher experience math - teacher experience reading) + coefficient2 (teacher age math- teacher's age reading) + coefficient3...
I have to compare the within student between subject approach estimates with fixed effect estimates that I estimated in the first step (the estimates will be the same or almost the same)
My problem is in my survey, the teacher that teaches math is the same that teaches reading, so teachers characteristics will be the same. I don't know how to compute the difference of teacher characteristics. or should I go ahead and estimate the within student between subject approach without differentiating teachers' control variables?
Does anyone had the same issue? How have you overcomed it?
I am doing it right? are my commands look good?
Thank you in advance for your help.
I have a standardized test survey that evaluate students' knowledge in mathematics and language through a combination of multiple-choice and open-response questions. My sample uses 10 countries with several control variables and two variables of interest: student test scores ( mathematics and reading), and another variable that I called x1 (dummy variables). I have one wave (2019).
From what I understood, the first step: two estimations have to be done, one using the fixed effect and a pooled old estimation. So, I estimated the two after setting my data panel.
- The fixed effect one: xtreg y x1 x2 x3.... fe, (x2, x3,...) are set of control variables that I use
- Ols: reg y x1 x2 x3....
The second step: for this step, I have to estimate the within student between subject approach, where the dependent variables is the test score difference between the two subjects, and the independent variables are the differences between my control variables, that is teacher characteristics (experience, teacher's age, teacher holds a postgraduate degree....).
within student between subject approach:
test score math - test score reading= coefficient1 (teacher experience math - teacher experience reading) + coefficient2 (teacher age math- teacher's age reading) + coefficient3...
I have to compare the within student between subject approach estimates with fixed effect estimates that I estimated in the first step (the estimates will be the same or almost the same)
My problem is in my survey, the teacher that teaches math is the same that teaches reading, so teachers characteristics will be the same. I don't know how to compute the difference of teacher characteristics. or should I go ahead and estimate the within student between subject approach without differentiating teachers' control variables?
Does anyone had the same issue? How have you overcomed it?
I am doing it right? are my commands look good?
Thank you in advance for your help.
Comment