Data Blindness Is An Epidemic Among Those With A Religious Belief in Polling

I DO LIKE a number of those members of Congress who are freaking out over the polling on Joe Biden's chances of winning against the two-time loser of the popular vote, criminally convicted, multiply successfully sued and prosecuted Donald Trump.   But I agree with AOC that these otherwise serious Democrats aren't presenting us with any real viable alternative to Joe Biden or Kamala Harris being the Democratic nominee.   Basing any such a call for the most successful Democratic President in a half a century to step aside on polling has a history of such conclusions to consider, something I don't think those expressing such faith in "data" have really understood.   The polls that showed that Alf Landon would beat FDR or Dewey would Truman were as based on data as the polls being pushed by the anti-Biden media now.   If there's one thing we know about polling today, it shares a lot of the same problems with polling done back then.   Here's some relevant history of that from the conclusion that Truman's chances of winning were hopeless, from the the department of Mathematics and Computer Science at the Oxford School of Emory University

In the 1948 presidential election, Thomas Dewey, then governor of New York, ran against the incumbent, Harry S. Truman. Three major polls (Gallup included) predicted Dewey would win. He did not. The Gallup poll came closest with 49.5% Dewey, 44.5% Truman, 5.5% other, and 0.5% undecided.

There were several problems:

First, they stopped polling too soon, and Truman was notably successful at energizing people in the last days before the election. The prediction came two weeks in advance of the election with 15% announcing undecided. It was assumed that the 15% undecided would split in the same proportions as those who had decided, leaving only 0.5% truly undecided. (The 0.5% undecided was Gallup's error factor.)

Second, the telephone polls tended to favor Dewey because in 1948 telephones were generally limited to more well-to-do households, and Truman was less popular among elite voters.

Third, one of the methods used sent interviewers into an area and told them to interview a certain number of people who met some given demographic criteria. For example, a pollster might be told to choose and interview 10 men, 11 women, 8 african americans, 2 asians, and 11 caucasians, etc... where the numbers chosen for each demographic is representative of the overall population.

Unfortunately for the pollsters, just because a sample is representative of the population demographically, doesn't mean that it will be representative with regard to the issue at hand.

Even worse, as long as the pollsters matched the quotas for their demographics, they were free to choose whom they like -- which can lead to substantial bias in the sample when pollsters avoid certain people that for some reason are hard to approach.

One of the things that analysis leaves out is the problem of people who choose not to talk to pollsters, ESPECIALLY THOSE WHO USE A PHONE TO REACH PEOPLE.   It's an absolute fact of polling in 2024 that only a small fraction of those called will willingly subject themselves to answering the questions asked.   I stopped talking to pollsters because I became convinced that virtually ALL POLLS ARE PUSH-POLLS DONE TO ACHIEVE RESULTS THAT THE PERSON PAYING FOR THE POLLS WANT.   I don't trust any of them, especially when they won't say who is paying for the poll. Virtually every poll done today, when so many must be as sick and tired of polls as I am, has an unrepresentative sample of the population WHICH IS THEIR ONLY SOURCE OF DATA,  that "data" that so many People with little to no actual critical knowledge of statistical analysis think means the same things as certain fact.   I have no idea who is going to win the election in November AND NEITHER DOES ANYONE ELSE, NO MATTER HOW DEVOTED TO "DATA" THEY LIKE TO BELIEVE THEMSELVES TO BEING.

I do know that no election has ever been won by throwing out the person who even the polls show has the best chance OF ANY NAMED ALTERNATIVE candidate to beat the worst person to have ever gotten the presidency BY LOSING THE POPULAR VOTE.   I certainly don't think that those who aren't willing to go on the record or those who have a track record of making wrong predictions are helping to defeat Trump by trying to reset the match this late in the game.   That has a record of proven disaster, despite what idiotic Hollywood bull shit magical thinking would have such idiots believe.   I think frequently consulted "Democratic pundits" as seen on TV and read in the NYT are as fully able to cut their babblage to the preferences of those who put them on the screen or in cheap ink on cheap paper.