What were the primary data limitations you faced in your academic projects, and how did you address potential endogeneity or selection bias?
