damiancclarke
commented
1 year ago Hi Emil, By definition synthetic DID finds weight to minimise discrepancies between trends in the pre-treatment periods (see eg equation (4) from Arkhangelsky et al.) So the resulting trends will already be the best match, in the sense of minimising (4). It may be worth looking to see if the time weights are important in these periods. For example, if there is some discrepancy but time weights are very low for these periods, perhaps it is not such a concern. Best wishes, Damian
Hello, Running the sdid command, I match the pretreatment evolution, however for two time periods out of approx 10, there is some discrepancy. In the synth command, you can place weight on some pretreatment periods by adding e.g. outcome(2015) outcomevar(2013) if those periods are not matched properly. Is there a way to do the same for the sdid command? I tried adding them in the same manner in my covarlist but without any change in the output.
Best, Emil