if i apply a smoother to variable=1 do I need to add the same smoother for variable=0

[Rose Matthews]

Hello,
I'm plotting kdensity graphs as perhaps seem in my previous post.

I'm using the following code:

twoway (kdensity score if treatment==0) (kdensity score if treatment==1, bwidth(10))

This produces a graph where treatment 0 (blue line) has a higher peak curve than treatment 1 (red line)
However, then I thought, I should be applying the -halfwidth- (bwidth) to both treatments. When I do, the red line peaks higher than blue line.
Should I be applying the same halfwidth's to both options of the treatment variable 0 and 1. ?

[Nick Cox]

My answer: Yes, without a really good reason to do differently.

If you smooth differently, you have to disentangle side-effects of doing that from what you infer to be real contrasts.

The principle is no different from comparing two histograms of similar data with different bin widths, namely don't do that if you can avoid it.