search

setzler / eventStudy

R package and guide for performing event studies with heterogeneous dynamic effects.
MIT License
71 stars 18 forks source link

linearize_pretrends() and residualize_covariates() does not respect anticipation #8

Closed davidnov closed 5 years ago

davidnov commented 5 years ago

Currently, both use supplied long_data and restrict to the pre-treatment period, i.e., cal_time_var < onset_time_var. For the case with anticipation > 0, we'd want this to only use the pre-anticipation period.

davidnov commented 5 years ago

This is fixed. Both now look at the cal_time_var < onset_time_var - anticipation period, as they should.