My view is that a student can only be judged on what they did get and only be awarded in addition some of what they didn't get. The ratio of what you did to what you didn't get is an odds so awarding some constant increment on the logit scale seems reasonable.

Agreed but observed variation >> true variation >> actionable variation, so we need to think and analyse carefully.

pmc.ncbi.nlm.nih.gov/articles/PMC...

Blatant misrepresentation. You can’t even call it a bear faced lie.

Does it depend on having binary outcomes?

Remember Cromwell’s Principle

www.senns.uk/wpoetry.html...

I was thinking more along the lines of “up to a point Lord Copper”.

I think what they have is a pre-programmed announcement that say what should happen if everything goes to plan but which runs on regardless of reality, and another one that has real information, and has to contradict it when necessary.
It reminds me of something; what can it be?

Interesting! What role does measurement error play? In the field in which I work, randomised clinical trials, if the trials are double blind, measurement errors are randomly distributed between treatment and control. GWAS studies may have problems.

www.goldenhelix.com/blog/wp-cont...

Sympathy. There are a lot off aura bores around.

There was never a trewser word spoken.

“Enjoy the result you have found by exploratory subgroup analysis, for you will not find it again.” Sayings of Confuseus

However, I believe that John Tukey proposed “a quarrel of statisticians” as the collective noun. Other possibilities are “a variance” , “a disagreement” or even a “disagreableness” .

I don’t see AI as a replacement for statisticians unless it comes in the shape of Marvin en.wikipedia.org/wiki/Marvin_...