November 02, 2024
Polling Follies and Mittens for Tuesday
Timothy Birdnow
I have been arguing that people are afraid to declare their support for Trump in much the same way they didn't declare for Reagan in 1980. I appear to be correct.
Here is an article about incorrect polling since '80.
FTA:
The UCLA profs failed to address WHY Republicans were unwilling to declare for Reagan. They also believed the "undecideds" made up their minds at the last moment, something entirely unprovable. More likely they had been afraid to vote Reagan but thought about how bad things had been under Carter. In other words, it was inevitable.
The author of this piece disputes that this Wilder Effect is in play:
This is a flawed argument; online polls are utterly worthless as they have no controls and are primarily taken by Millenials and others who tend to be liberals and would skew the results. And it also presupposed polls - and pollsters - are honest brokers of information, something I deny.
The article continues:
That only buttresses my claim about online polls, and egven about traditional polling. You have to be home and have the time to waste on a rather lengthy survey. Democrats have lots of time on their hands; not so much Republicans or those who would be likely to actually vote Trump (they are too busy earning a living and taking care of their families).
This article was written before the Trump/Biden debate and here we see how myopic the thinking:
Biden would be finis after the first debate as polls showed it was impossible for him to win. The "recent polls" mentioned were undoubtedly incorrect even then; the public had absolutely no reason to support the baby biter in chief, none whatsoever. These polls were trying to lead the public, not reflect it.
I strongly suspect the polls in this election will be way off, as they were in 1980.
Anyone remember the iconic photo of Harry Truman holding up the paper that said "Dewey Wins!"? Polls showed Truman would be out. Anyone remember polls showing Nixon would beat Kennedy? How about the claims the GOP stole the election from John "I Served in Vietnam" Kerry because the election did not go the way polls suggested it would? The polls have been wrong more then they have been right.
Polls should be taken with a huge grain of salt. And given the overall trends right now I think we will be very happy next week. But of course none of this can take theft into account, and what the Democrats can't win they can manufacture.
It's going to be a nail-biter; you may want to buy some mittens for Tuesday.
I have been arguing that people are afraid to declare their support for Trump in much the same way they didn't declare for Reagan in 1980. I appear to be correct.
Here is an article about incorrect polling since '80.
FTA:
Speaking to The Bruin in that article, UCLA professors attempted to explain "Where the pollsters goofed.”
"The polls immediately preceding the election did not accurately predict the election results either because poll respondents included a large number of Reagan supporters who were unwilling to declare their support for him, or because a very large portion of ‘undecideds’ and Anderson supporters decided to support Reagan over the weekend,” the article said.
End
"The polls immediately preceding the election did not accurately predict the election results either because poll respondents included a large number of Reagan supporters who were unwilling to declare their support for him, or because a very large portion of ‘undecideds’ and Anderson supporters decided to support Reagan over the weekend,” the article said.
End
The UCLA profs failed to address WHY Republicans were unwilling to declare for Reagan. They also believed the "undecideds" made up their minds at the last moment, something entirely unprovable. More likely they had been afraid to vote Reagan but thought about how bad things had been under Carter. In other words, it was inevitable.
The author of this piece disputes that this Wilder Effect is in play:
"Some pundits similarly argue that undecided voters, as well as those that may have been embarrassed to share their preference for Trump with pollsters, disrupted the polls in 2016. This "shy Trumper” theory has been largely disproven by the fact that Trump did not perform significantly better in online polls, where voters would not have had to reveal their views to real people."
This is a flawed argument; online polls are utterly worthless as they have no controls and are primarily taken by Millenials and others who tend to be liberals and would skew the results. And it also presupposed polls - and pollsters - are honest brokers of information, something I deny.
The article continues:
In the four years since the 2016 election, Pew Research Center researchers have examined the polling methods that resulted in erroneous predictions in 2016. These researchers have determined that nonresponse bias, which occurs when certain groups within a population tend to not respond to polls, and incorrect assumptions about who would turn up to vote are largely to blame.
Many pollsters relied on data from past elections to predict who would vote in 2016. However, significant changes in voter demographics invalidated many of these assumptions, resulting in inaccurate polls. Notably, nonwhite voters did not turnout for Clinton at the same rates as they did for former President Barack Obama in 2012.
End
Many pollsters relied on data from past elections to predict who would vote in 2016. However, significant changes in voter demographics invalidated many of these assumptions, resulting in inaccurate polls. Notably, nonwhite voters did not turnout for Clinton at the same rates as they did for former President Barack Obama in 2012.
End
That only buttresses my claim about online polls, and egven about traditional polling. You have to be home and have the time to waste on a rather lengthy survey. Democrats have lots of time on their hands; not so much Republicans or those who would be likely to actually vote Trump (they are too busy earning a living and taking care of their families).
This article was written before the Trump/Biden debate and here we see how myopic the thinking:
"Some of the most recent polls show Democratic candidate Joe Biden ahead of Trump in most states, including key swing states such as Wisconsin, Pennsylvania, Arizona and Florida. Early ballot counts also indicate that 2020 will witness historic levels of voter turnout, unlike the 20-year low seen in 2016. "
Biden would be finis after the first debate as polls showed it was impossible for him to win. The "recent polls" mentioned were undoubtedly incorrect even then; the public had absolutely no reason to support the baby biter in chief, none whatsoever. These polls were trying to lead the public, not reflect it.
I strongly suspect the polls in this election will be way off, as they were in 1980.
Anyone remember the iconic photo of Harry Truman holding up the paper that said "Dewey Wins!"? Polls showed Truman would be out. Anyone remember polls showing Nixon would beat Kennedy? How about the claims the GOP stole the election from John "I Served in Vietnam" Kerry because the election did not go the way polls suggested it would? The polls have been wrong more then they have been right.
Polls should be taken with a huge grain of salt. And given the overall trends right now I think we will be very happy next week. But of course none of this can take theft into account, and what the Democrats can't win they can manufacture.
It's going to be a nail-biter; you may want to buy some mittens for Tuesday.
Posted by: Timothy Birdnow at
10:36 AM
| Comments (2)
| Add Comment
Post contains 746 words, total size 5 kb.
1
You may want to buy a few other provisions, such as a few jugs of your favorite adult beverage(s), plenty of popcorn, and if the other side wins, lots of canned soup.
Posted by: Dana Mathewson at November 02, 2024 11:38 PM (wQ5RW)
2
Great idea Dana!
Posted by: Timothy Birdnow at November 03, 2024 10:09 AM (vQnzY)
24kb generated in CPU 0.0691, elapsed 0.5692 seconds.
37 queries taking 0.563 seconds, 169 records returned.
Powered by Minx 1.1.6c-pink.
37 queries taking 0.563 seconds, 169 records returned.
Powered by Minx 1.1.6c-pink.
