Timeline for Bias correction in weighted variance
Current License: CC BY-SA 3.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 11, 2019 at 12:29 | comment | added | gaborous | @Confounded See the discussion in the comments here, basically it works with reliability weights if they are not standardized/normalized, but if your reliability weights are (since that's kind of the purpose usually), then you lose the base frequency (the total count) and thus the Bessel correction cannot be applied (because the necessary total count information is lost!) | |
| Jan 4, 2019 at 10:31 | comment | added | Confounded | @gaborous Where in the above derivation do you see any restrictions on weights being placed that makes it valid only for "reliability"-type weights? As far as I can tell, the derivation is general an is valid for any weight > 0. | |
| Nov 23, 2013 at 16:58 | comment | added | gaborous | @rajatkhanduja I was not talking about the proof but the final derived equation (the top one in this answer). But indeed it's correct, the numerator is just squared because we multipy by V, thus the numerator ends up being unsquared. Anyway, this estimator remains biased as I explain in my answer below since it relies on "reliability"-type weights. | |
| Nov 22, 2013 at 6:44 | comment | added | rajatkhanduja | @user1121352, the variant (C) and this equation are the same. Observe the meaning of [\lambda] in this equation, it is not just the weight, but normalized weight. Plugging in the value of [\lambda], we get the same result as variant (C). | |
| Jun 13, 2013 at 10:38 | comment | added | gaborous | This equation is wrong according to Wikipedia, Matlab, R, and others which are implementing this equation. The numerator here is squared, but it should NOT, it should be just like the (C) proposed by the OP. See en.wikipedia.org/wiki/… | |
| Jun 9, 2013 at 10:19 | vote | accept | Has QUIT--Anony-Mousse | ||
| Jun 8, 2013 at 23:08 | comment | added | gaborous | I have checked this solution empirically and it does NOT work... The only one that does is solution (A) that I have also implemented in the past by myself, but it only works with weights being integer numbers and >= 0 | |
| May 29, 2013 at 14:12 | history | notice removed | whuber♦ | ||
| Jan 13, 2013 at 0:39 | history | edited | ThePawn | CC BY-SA 3.0 |
added 602 characters in body
|
| Jan 12, 2013 at 17:29 | history | notice added | whuber♦ | Needs detailed answers | |
| Jan 10, 2013 at 7:42 | history | edited | ThePawn | CC BY-SA 3.0 |
Corrected typo ")" in formula
|
| Jan 10, 2013 at 4:17 | history | edited | ThePawn | CC BY-SA 3.0 |
added 2 characters in body
|
| Jan 10, 2013 at 3:36 | history | answered | ThePawn | CC BY-SA 3.0 |