Hi,

I have a dataset with COVID-19 mortality rates categorized by occupation and industry, consisting of around 1,000 cells. Please see a snapshot of the data below. The full reshaped dataset has approximately 100,000 workers * 1000 cells. I would like to perform an occupational choice model using the mortality rate * post2021 variable as an independent variable and the probability of employment as the dependent variable. However, I also want to control for occupation by industry fixed effects and year fixed effects, and I am curious if it is feasible to cluster standard errors at the occupation by industry level. Although I was able to run a linear probability model, I prefer to use a non-linear estimation method that takes into account the fixed effects and clustering to make accurate predictions.

My question is, is it possible to perform this estimation using clogit or some other Stata commands? I have only seen examples where independent variables vary across both case and choice, but not for choice only where there is only significant variation when there are multiple choices. I would greatly appreciate any advice on this matter.



Array