Consider the following baseline model: yijk = beta * xijk + errorijk
where i indicates a child, j indicates a family, and k indicates a school.
y is a measure of human capital and x is the quality of the school that the child attends.
I want to decompose beta into a within family component and an across family component. I haven't found a very straight-forward way of doing this in stata (the extended model contains controls and weights). I was hoping to be able to use some of Stata's panel tools but haven't found a way to do so.
Related Posts with Decomposing regression into within and between effects
Is it possible to match families/groups and not individual observations?- Tabulating the variable's highest value and show the highest frequency
- Probit marginal effects: correct interpretation?
- Would boot strapping be used as an alternative to confidence interval along with Chi-Square test statistic?
- Constraining Discrimination Parameters in Three-parameter Logistic Item Response Theory Model
Subscribe to:
Post Comments (Atom)
0 Response to Decomposing regression into within and between effects
Post a Comment