Can you get in trouble for that?
Asking for a friend named Claudine.
Changing signs and magnitudes of regression coefficients in your dissertation
-
If you really believed what you said was true, and could factually back that up, you would step out from your anonymity and put forth your evidence and argument.
But alas...a seething White Grieving InceI who spends frantically squeaks "Claudine! Claudine!" on an obscure phony econ forum. What a man!
-
Are we talking log(NAICS) here?
I think it's a Harvard insider joke. It alludes to the fact that Claudine Gay changed some White turnout coefficient signs and magnitudes in her dissertation (in her mind, the faked results "show" that White people are "racist")
The other issue is in fact the opposite of log(NAICS). A large coefficient value is driven by a coding "error" (likely not accidental): instead of campaign spending ratio, for one (critical) observation she uses actual amount spent. Remove that error and the coefficient shrinks close to zero.