Add as a feature the Cochran-Mantel-Haenszel test

[Sal-Ramos]

Hello all,

First of all I would like to thank you for the awesome initiative (JASP). I use it regularly not only to teach but also to do the great majority of my statistical analyses. It easy to install, it has Windows, MacOs and Linux versions, it is easy to use and easy to report the results. Really amazing software!

I would like to suggest a new feature: the Cochran-Mantel-Haenszel test.

It is used to compare proportions for nominal matched-pairs responses (Agresti).

It compares repeated measures (Before-After) categorical variables that contain more than two levels. For two levels it simplifies to the McNemar test (the request #1014).

I believe it could be placed in Frequencies>Contingency Tables menu below the Vovk-Sellke maximum p-ratio statistic, along side with the future McNemar test.

The report could be just the usual table with the values of the statistics, df and p-values.

Sincerely,

Sal Ramos

[EJWagenmakers]

Indeed! @FBartos, can you add this to "more love for contingency tables"?

[TarandeepKang]

Hi Team,

I would just like to add some new literature to this discussion.

I have recently seen a new framework for the analysis of proportions: Laurencelle, L., & Cousineau, D. (2023). Analysis of proportions using arcsine transform with any experimental design. Frontiers in Psychology, 13. https://www.frontiersin.org/articles/10.3389/fpsyg.2022.1045436

And also this suggesting a superior version of the McNemar:

(But my French is too bad to understand much of this)!

Laurencelle, L. (2023). Pour la différence entre deux proportions jumelées, un nouveau test, plus valide et plus puissant [A new standard normal-based test for the difference between paired proportions to supersede obsolete McNemar-like and other indirect procedures]. The Quantitative Methods for Psychology, 19(3), 254–264. https://doi.org/10.20982/tqmp.19.3.p254

2322 should probably also be considered part of the discussion, but no one has attended to it yet.

[aboyle57]

Briefly adding voice to the calls for McNemar's test to be included in JASP! Thanks for all the efforts, it's a great programme.

[EJWagenmakers]

Frantisek please oh please :-)

[TarandeepKang]

Hi EJ, I don't suppose anybody on your team reads French? In particular in regard to the 2023 paper I posted above, which suggests an improved alternative to the McNemar which is more accurate and powerful indeed the author goes so far in the English version of the title to call it "obsolete". My French is too bad to know if that's grandstanding, but it does sound relevant if you are considering adding the McNemar test?