David:
I wanted to reply to your original message, but the system won't let me. Says it is being moderated.
So I'm replying to a reply!
I think that the you are reversing conditional probabilities, thus making the same kind of error that people make when they misinterpret a p value.
It's identical to saying that if a coin comes up heads 10 times in a row that there is a 0.001 probability that it is fair. That won't fly.
Even if we agree that there was, at some point, a 50% chance that Biden actually won each of the 10 states he was initially behind in, the 0.001 probability you calculated is NOT the probability that he won those 10 states. It's only the conditional probability that he would have won all 10 given the 50-50 initial probability.
It's easier to understand with a single state. Let's assume that Biden has a 50% chance of truly winning Pennsylvania based on early results. He might truly have a majority or he might not.
He then wins the state by a substantial majority, as he did, and as was predicted. Does that mean that there is only a 50% probability that he truly won? No!
There are two possible states of reality: (1) PA really does support him, with a majority of voters voting for him or (2) PA does not.
So, Bayes theorum requires the a priori probability of winning truly (which we have set at 50%), the probability he would win if the voters really want him (for simplicity call that 100%) and the probability he would win by massive fraud if the voters did NOT truly choose him.
It's that last probability that is the kicker. If you think voter fraud large enough as to overturn a Trump victory in PA is likely, then Biden has only a modest chance of having truly won. I think that is very unlikely, and would assign it a conservative 1% probability.
So, by Bayes theorem p (Biden truly won PA) = 1 x 0.5 / (1 x 0.5 + 0.01 x 0.5) = 99%.
If we say that since his victory was predicted, there was a 90% chance he truly won (a priori) the state (PA) then the calculation becomes 1 x 0.9 / (1 x 0.9 + .01 x 0.1) = 0.9989
The same kind of reasoning applies to the 10 state example. In arguing that Biden didn't win all 10 truly, you are replacing a 50% flip in some of the 10 with a 1% (or less!) vote fraud explanation which is far less plausible than a flip.
Our elections are done carefully, with many doublechecks. The actual result is the gold standard in the absence of solid evidence to the contrary.
Ed
------------------------------
Edward Gracely
Drexel University
------------------------------
Original Message:
Sent: 01-25-2021 17:28
From: David Wagner
Subject: Election results
Hi Jim,
Thanks for replying. What sort of debunking did you have in mind? I was hoping you could explain in detail why the probability calculation was incorrect, if so. Actually, the advertiser-supported press floated the ideas that Republicans voted in person yet Democrats were told to vote by mail and did so by mail. Typically, absentee votes and military votes come in late, so that would make the late votes skew Republican. In any event, President Trump had an early lead in the swing states, so the question using probability theory, is how likely would it be for Former Vice President Joe Biden to come back and win the state given some reasonable assumptions of odds. Even with a 90 to 10 split for Biden, the probability of Biden flipping the 10 states is only 34%. Of course, it is possible that Biden won, but is it probable? No.
Thanks,
David A. Wagner, Ph.D.
------------------------------
David Wagner
Trident University International
Original Message:
Sent: 01-25-2021 16:40
From: James Knaub
Subject: Election results
<span;>Why has "Is it probable that Biden won the
<span;>2020 election" been posted without giving anyone a chance to debunk this?
<span;>David Wagner's calculations are not valid because most Democrats voted by mail, and many more Republicans voted on election day, as they were encouraged to unsafely do. Many States count the election day votes first. Therefore it was expected that Trump would have more votes counted first no matter how badly he lost. (Also, the "...<span;>hundreds of affidavits of fraudulent activities..." were repeatedly debunked in court. Let's not keep repeating falsehoods.)
------------------------------
James Knaub (Jim)
Retired Lead Mathematical Statistician
------------------------------