borusyak/did_imputation

hetby option error: The hetby variable takes too many (over 30) values

SaiChrisZHANG opened this issue · 2 comments

SaiChrisZHANG commented

Hi Kirill,

Thank you for the awesome package! I'm having a problem estimating treatment effect heterogeneity, I intend to use the hetby() option, but get the error:
The hetby variable takes too many (over 30) values
too many categories
r(149);

When I use project() option, the estimation can be done. I was wondering if you have any insights on this error? The heterogeneity variable is a 0/1 dummy.

Also, Is there a way to plot the pre-treatment differences between the 2 groups for each period, like an event study for the heterogeneity?

Thank you in advance!

borusyak commented

Hi,
Please check the number of unique values of the heterogeneity variable via "levelsof". If this doesn't solve the mystery, please email me the minimum working example of the problem.

borusyak commented

Regarding plotting, you can always do estimation separately by group and combine the graphs.