Closed jkcshea closed 2 years ago
Hi @jkcshea , the equality constraint example is a bit odd, since it implies that the treatment effect is constrained to be identically 0 for everyone. I think imposing it on a covariate makes more sense, and would be the more common use case. Otherwise that looks good to me. We can always make small adjustments or correct typos later.
Ah right, thanks.
The example has been updated so that equal.coef = 0 + x
.
Hi @a-torgovitsky,
I've updated the vignette to include examples of the regression approach and use of the
equal.coef
parameter (#204) So below are the changes, in case you want to take a look.But a quick overview of what has changed:
lsei
is required for constrained least squares.equal.coef
parameter.