search

bfifield / hettx

hettx: Detecting and Measuring Treatment Effect Variation
9 stars 0 forks source link

whether binary Z is required #18

Open kai0511 opened 4 years ago

kai0511 commented 4 years ago

Dear Authors,

I am thinking whether the package can handle ordinal or continuous treatment variable Z. Could you please clear this issue for me?

Thanks!

Kai

lmiratrix commented 4 years ago

The package currently can handle only binary treatment. Continuous treatment would require some sort of dose-response model, I believe, and thus would rely on a differ kind of approach to impact estimation. Ordinal (or multiple treatments) would likely be doable as an extension.

kai0511 commented 4 years ago

Imiratrix, Thanks for your detailed reply!