Talk:Regression toward the mean

Confusion about what is being implied

It says in this article, in the "Other statistical phenomena" section, "For example, following a run of 10 heads on a flip of a fair coin (a rare, extreme event), regression to the mean states that the next run of heads will likely be less than 10..." but what does that actually mean? Because, the odds of throwing 10 heads in a row haven't change at all due to that rare first event. The odds of throwing 10 heads in a row is still as it always was 1/1024 and those odds don't change over time, previous experience is completely irrelevant to any future probability. There is no God evening-out things, trying to make things fairer. It is true that the chance of throwing a run of 10 heads on a flip of a fair coin is a rare event and so if you keep throwing the coin it is highly probable that you will not throw another 10 heads in a row, but the probability of that same rare event happening again has not changed at all, the odds for the second time of a run of 10 heads on a flip of a fair coin are exactly the same as they were on the first. When you look at all the throws of the coins, say a couple hundred times later, the ratio of head to tail is highly likely, but only highly likely, to be close to 50/50 but that definitely does not mean that the probability in the second run of throwing 10 coins was less likely to end up with 10 heads than the first, which seems to be what is being subtly implied in this Wikipedia article. It is true that following a run of 10 heads on a flip of a fair coin (a rare, extreme event), the next run of heads will likely be less than 10, but your chance of throwing 10 heads on a flip of a fair coin (a rare, extreme event) was always less than 10, even before the first run, so what is significantly being said there?! The probability for the second run of throwing the coins hasn't been changed, the odds are still the same as they were for the first time, 1/1024 of them all turning out to be heads. Even if you did by chance throw another 10 heads the second time, then if you were to consider throwing the coin a third 10 times your calculated odds for the third time are still exactly the same as they were for the very first run, 1/1024 chance of getting 10 heads in a row! Once an extreme pure chance event happens it does not lessen the likelihood of another extreme pure chance event happening, it is perhaps less likely that you'll get two extreme pure chance events happening rather than just one but once an extreme pure chance event has happened it will have absolutely no effect on the future probability of another extreme pure chance event happening. The "Other statistical phenomena" section confuses me, what on Earth is it implying!? It almost sounds like a confidence trick, a scam is being marketed.

Change of Lede – old was false, misleading or irritating

It states “In statistics, regression toward (or to) the mean is the phenomenon that arises if a random variable is extreme on its first measurement but closer to the mean or average on its second measurement and if it is extreme on its second measurement but closer to the average on its first.” That doesn’t make sense at all. Everything beginning from “and” can be cut off. Also there’s nothing inherently special about the very first measurement only being extreme. It should say “In statistics, regression toward (or to) the mean is the phenomenon that arises if a random variable is extreme on its first or first few measurements but closer to the mean or average on further measurements.” I edited like so. Correct me if I’m wrong, though. Zyzzyvy (talk) 17:53, 15 June 2020 (UTC)

When did "reversion to the mean" become historical?

I was under the impression that "reversion to the mean" was a common name for this, almost as common if not just as common as "regression." For instance, Wolfram lists reversion to the mean as the preferred term. (https://mathworld.wolfram.com/ReversiontotheMean.html) However, in this article, it is presented as a former name. I think this should be changed, but don't feel confident enough to do it myself; do others agree?

Why is there no mention of the Central Limit Theorem in the entire page?

I am not sure if these two concepts are mathematically related but to me they seem highly similar. Should the Central Limit Theorem not at least be a "See Also"?

Nickfury95 (talk) 18:31, 6 October 2021 (UTC)

Confusing unsourced statement in the linear regression section

This statement "If −1 < rxy < 1, then we say that the data points exhibit regression toward the mean." is not sourced and contradicts at least one primary source "Before we discuss the RM effect further, let's point out a common confusion in its interpretation. T...) he term “imperfect correlation” is used to indicate that 0 < rxy < 1. Regression to the mean on the other hand is equivalent to 0<ˆβ<1. Confusion arises because the two conditions are not the same. When rxy = 1 and sy < sx, then ˆβ<1 so that there is regression to the mean in spite of the fact that rxy = 1. Vice versa, when ˆβ=1 - A Statistical Explanation of the Dunning–Kruger Effect Jan R. Magnus1Anatoly A. Peresetsky2

(also the wikipedia editor on phones seems very buggy, large chunks of my comment seem to disappear at random, apologies for the formating and missing elements. I fear editing further for fear of the consequences.) ~~ Xelote (talk) 01:14, 21 September 2024 (UTC)