I agree with Liz. It doesn't matter much how you calculate it. The
particular calculation is a heuristic only.
There is also the separate issue that with propensity score matching, most
calipers in most applications will increase imbalance. In fact, the
stricter the caliper, the worst the imbalance. This is a problem with
pscore; the same issue does not usually arise with any other method of
matching so far as I know.
Gary King
http://gking.harvard.edu
617-500-7570
(from my phone)
On Aug 12, 2011 2:24 PM, "Stuart, Elizabeth A." <estuart(a)jhsph.edu>
wrote:
Hi Ana,
I think this is more of a "choice" than a "mistake." I have seen a
number
of different SD's used for this purpose; in fact Rosenbaum and Rubin
(1985)
use something different from Rubin and Thomas (2000). But perhaps I am
missing something in the literature (In particular I am not sure what you
mean by "caliper theory.")
This also shouldn't matter much in practice, unless the two groups have
very
different SD's, which I don't think is that common. But in any case you
could always use the SD you want by multiplying the caliper size constant
(e.g., .2) by the appropriate value. i.e., if you want 0.2*SD1 but we use
SD2, you can specify that you want the caliper size to be 0.2*(SD1)/(SD2)
(or something like that; I haven't fully thought through the details).
I hope this helps.
Liz
On 8/12/11 1:01 PM, "Ana Kolar" <annakolar(a)yahoo.com> wrote:
Hi there,
I've been using matchit function together with the caliper argument and
I'm afraid that this argument is not programmed according to the "caliper
theory". To my knowledge if caliper=0.2 this number has to be multiplied by
the sqrt((var(psT) + var(psC))/2), while the results of matchit function are
that 0.2 is multiplied only to the sd(psALL).
psT - propensity scores of treated group
psC - propensity scores of control group
psALL - propensity scores of treated and control group
Please correct me if I'm wrong otherwise I'm wondering if its possible
that you correct this mistake.
I'm looking forward to hearing from you soon.
Best regards,
Ana