Closed felixthoemmes closed 1 year ago
Hi Felix,
Shook-Sa and Hudgens don't actually describe the ESS; they describe the design effect (deff), which is
deff = \frac{n \sum w^2}{(\sum{w})^2}
The ESS is
ESS = \frac{(\sum{w})^2}{\sum w^2}
So, we have that
deff = \frac{n}{ESS}
The expression in the documentation for the ESS is correct.
Argh, my mistake. Yes, indeed that's correct. Apologies for the pre-mature and unnecessary pull request.
On Sat, Jan 21, 2023, 2:39 PM Noah Greifer @.***> wrote:
Hi Felix,
Shook-Sa and Hudgens don't actually describe the ESS; they describe the design effect (deff), which is $$deff = \frac{n \sum w^2}{(\sum{w})^2}$$
The ESS is $$ESS = \frac{(\sum{w})^2}{\sum w^2}$$
So, we have that $$deff = \frac{n}{ESS}$$
The expression in the documentation for the ESS is correct.
— Reply to this email directly, view it on GitHub https://github.com/ngreifer/WeightIt/pull/44#issuecomment-1399319125, or unsubscribe https://github.com/notifications/unsubscribe-auth/AECIYYQT6ZEEXTPRTS2E6OTWTQ3QVANCNFSM6AAAAAAUCEBHJY . You are receiving this because you authored the thread.Message ID: @.***>
Hello Noah, I may have found a very small typo in the help file on the ESS function.
Formula was missing an "n" in denominator, as per reference Shook-Sa et al. (2020)