Talk:ANOVA on ranks

Keep your shirt on!
This is just a stub, removed from another page at another editor's request.. It will take some time to clean it up.Edstat (talk) 03:51, 15 November 2010 (UTC)


 * I've made a start with the header. I'll work on the reference section when I get some time.Edstat (talk) 23:40, 17 November 2010 (UTC)

Marginal ranks
Thanks for your efforts, EdStat.

Most authors refer to Puri-Sen's book, which uses marginal ranks (whose obsolescence I thought was universally acknowledged since the 1980s or certainly the 1990s). Maybe an article on "marginal ranks" would be better than an article on "Anova on ranks"? I think that "marginal ranks" is the most widely used term for the obscolescent approach.

Best regards, Kiefer.Wolfowitz (talk) 23:48, 17 November 2010 (UTC)

Randomized assignment of treatments
EdStat, please randomly assign the treatments to the subjects. Your description confounds group membership (block) and treatment. Thanks, Kiefer.Wolfowitz (talk) 14:52, 18 November 2010 (UTC)

Reset
Lets clarify. Anova on ranks is NOT for violations on variability! Rank-based statistics do not work when homoscedasticity has been violated! This is not a procedure to test any hypothesis about variances - it is restricted to a test of difference in means.

Marginal ranks is not the topic here. See the famed Iman and Conover "Bridge" article in JASA for further clarity.Edstat (talk) 18:47, 18 November 2010 (UTC)

1st draft finished
I am calling it a draft because I'm sure others will now have something further developed to edit, not that the text is just sandbox quality. I'm assuming the two tags have been fulfilled and have thus removed them.Edstat (talk) 19:36, 18 November 2010 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on ANOVA on ranks. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20110302030459/http://is.ba.ttu.edu/conover/Dr.Conover.htm to http://is.ba.ttu.edu/conover/Dr.Conover.htm

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 03:13, 24 June 2017 (UTC)

Proposed merge of Scheirer–Ray–Hare test
I suggest merging new article Scheirer–Ray–Hare test into ANOVA on ranks after greatly editing the former article down, possibly to only a couple of sentences. Having read https://www.real-statistics.com/two-way-anova/scheirer-ray-hare-test/ and the mathematical description in the original Scheirer–Ray–Hare paper, it's equivalent to ANOVA on ranks but referring the test statistic to a chi-squared distribution instead of an F-distribution, i.e. using the asymptotic limit listed as the 6th property of F-distribution), so assuming a large number of residual degrees of freedom. I can't find a reference for this though from a search of Google Scholar, however. Maybe someone watching this page knows of one? Or  ? --Qwfp (talk) 11:14, 17 February 2019 (UTC)
 * Oppose: many statistical test have similarities, and have similar goals, but your description implies that they are different tests. I suggest improving the page rather than merging. Klbrain (talk) 17:39, 19 February 2020 (UTC)
 * Closing, given the uncontested objection and no support. Klbrain (talk) 08:32, 17 April 2020 (UTC)

Aligned-Rank Transform (ART), ATS/WTS, robust ANOVA, permutation AN(C)OVA is missing
The discussion about ANOVA on ranks is OK, yet very incomplete. The RT ANOVA (indeed - poor) is currently not the only option. ART is widely accepted method and doesn't suffer from the RT issue. ART: http://depts.washington.edu/ilab/proj/art/index.html

ANOVA-Type Statistics and Wald-Type Statistics are another proposals, becoming popular especially in clinical biostatistics (. http://www.ncs-conference.org/2010/3B_07.pdf https://pdfs.semanticscholar.org/88cb/15520b2f84fd2a5a09e0341e791f40ab4118.pdf https://arxiv.org/pdf/1606.03973.pdf https://www.sciencedirect.com/science/article/pii/S0047259X08000201 https://www.jstatsoft.org/article/view/v050i12

Also the robust estimation of the variance (e.g. sandwich) is worth mentioning for sure. Last but not least, permutation (or finite resampling) AN(C)OVA (in general - General Linear Model) is definitely a thing that should be discussed at least briefly. All of the methods are available in R, the first two - also in SAS.

It would be good to develop the ANOVA part. — Preceding unsigned comment added by 85.14.85.194 (talk) 09:10, 19 August 2019 (UTC)