Open bquistorff opened 7 years ago
Question about synth_runner's generated CIs: Are these from one-sided pvalues? (i.e. Pr(alphahat^PL_it>=alphahat_it) in your paper at http://econweb.umd.edu/~galiani/files/synth_runner.pdf? For some results they looked quite symmetric, but I just got a set where the LB is highly skewed relative to the UB.
Thanks for the reply and for the nice package.
They're two-sided p-values.
I would love to see this enhancement.
Provide mechanism to graph confidence intervals. Have option to put around counterfactual or around real (depending on how treatment is thought of).