Публикация

"A p-value is the thing that determines if you get published."

Relatedly, I can also define alpha in one sentence. But that would be showing off.

Nope, because for a traditional frequentist the data can't have a probability

Be careful what you wish for or we’ll get stream of consciousness stats papers

Alive or undead?

In the German language it can be done in a single world, Nullhypothesenabweichungsextremitätskoeffizient.

For the audiences in mind eg beginner students or New Yorker readers, this means more sentences explaining 'the null', and possibly 'data'.

Sorry, but is this bait? (Yes, it's an incorrect definition ... would need at least an "or something more extreme" to be correct, unless I'm missing something ...)

In brief: "probability of observing your data" isn't a well-defined statement. For example: the probability of flipping exactly 500 heads out of 1000 from a fair coin is ~2%, but that would not be a useful p-value. Likewise, the probability of observing any specific values from a Normal dist is 0

I think my biggest issue with definitions like these is the it neglects that it is an estimation. Part of the mass confusion around stats, from my pov, is that people treat stats as ground truth and not models. You don't do stats on a ground truth, you don't make estimates on ground truths.

A standard deviation is the second most widespread statistical concept that is not definable in a single sentence.

I’m afraid this flips the statistical conditional — “given these data, how likely just chance?” — when the logic is closer to “given just chance, how likely these data?” Now how bad that is probably depends on the context and the extent to which it’s taken seriously in the first place 😅