p<.01. The first model (panel ordinary least squares, with fixed effect (FE), first column of Table 2) is our benchmark model and is based on the hypothesis that the unobserved individual factors (for instance
Or maybe it is that there are 10 kinds of people (those who understand binary and those who don't). Obviously there is no perfect solution to this sort of unreal situation, but I think the stacking behavior may be much cooler for a whole bunch of normal cases. Now what these weird ca...