sergiocorreia/reghdfe

Is reghdfe capable of producing survey adjusted standard errors?

bob111111111111111111111111111110 opened this issue · 0 comments

Is reghdfe capable of producing survey adjusted standard errors?

The issue is that survey adjustments to standard errors come it two flavors.

  1. They inflate the standard errors due to clustering within survey clusters (clusters are called primary sampling units or PSUs in the typical survey language).
  2. Then they also deflate the standard errors to take advantage of information about survey strata (survey strata are independent of one another and using the knowledge that they are independent means the standard errors should be smaller).

Survey regressions make both adjustments (see the “Stata survey data reference manual” pages 3-4 (https://www.stata.com/manuals/svy.pdf) for an explanation of this). From what I can gather, reghdfe can only make the first type of adjustment. Maybe there’s some way to make it make the second type of adjustment, but I can’t figure out how. Your help would be greatly appreciated.