A while ago, I tried to replicate tables from a textbook on panel data analysis to learn more about fixed effects, random effects and hybrid models. While code and files for the book were available on a webpage for the book, I still had some questions. I received a detailed reply only three days after I wrote to one of the authors.
I had asked for guidance through the STATA .do file to match the tables in the book with the models in the code; variable transformation questions; and I wanted to see the balanced version of the data set that went into the model to check if I created the exact same version in R. Here are parts of the correspondence:
here is the code for the hybrid model 1 (model 1 in table 4.12)
here is the code for the hybrid model 2 (model 2 in table 4.12)
Yes, dichotomous “divorced” is treated as a continuous variable, its mean is computed and a new demeaned variable generated.
I include the balanced data as a stata file.
More replication correspondence
I collect more responses from original authors here.