Open Mbjoerkh opened 5 years ago
@Mbjoerkh This is a good point. We will include an option so that researchers can specify their preferred function as needed.
Great.
Came to think of a related point..
It would be helpful to also include an option to plot the absolute values of the outcome variable for both the treated and control units (in addition to the resulting coefficients), to the functions to plot your estimates. As the researcher in many use-cases will want to show that the effect is coming from the treated units improving, as opposed to the control unit worsening.
I'm matching on a variable where the standard deviation of this variable doesn't make much sense, could you allow for the plotting of the covariate balance to be done with absolute values?
I think this would be useful regardless of my specific application which I doubt is common (I'm matching on a state specific field rank, where the # of firms varies by state, thus making the SD of the rank state-specific). Not sure exactly how I should proceed, but absolute values would be more meaningful than SD's.
Thank you again for a great package and promising framework.
Markus