Ignore opinion polls.
Posted on 05/21/2019 10:39:31 AM PDT by 2ndDivisionVet
Professional pollsters say the old way of conducting surveys is fading fast, but new methods might not be trusted and ready for next year.
TORONTO — When veteran pollster Scott Keeter appeared at a recent gathering of industry professionals, he began his presentation with a somber caveat about the methods at the center of his life’s work.
Telephone polling — for decades the backbone of efforts to measure public opinion and the subject of his new study — are in “wheezing condition,” Keeter told a roomful of colleagues. And experiments to prolong their use are akin to putting on “a great party for the deck of the Titanic.”
The impending death of the telephone poll comes just as the 2020 presidential election is approaching — and without enough time for a tested and trusted alternative to replace it. That raises serious concerns about the reliability of polling results heading into the election, other survey researchers told POLITICO on the sidelines of their conference, with scrutiny of the industry set to be heavier than ever after President Donald Trump’s surprise victory in 2016.
Fewer Americans than ever are willing to pick up the phone and talk to pollsters, sending costs skyrocketing to roughly double what they were four years ago. Despite enjoying a largely successful 2018 election, pollsters are furiously experimenting to fill the void left by the slow failure of the telephone poll, looking at everything from internet-based solutions to snail mail.
But the possibility of another polling miss in the 2020 presidential race looms, and the next election could present new, unforeseen challenges that polls may struggle to address as they test new methodologies under exacting scrutiny. Both parties are framing the race as an existential contest over the country's future.
(Excerpt) Read more at politico.com ...
-PJ
If pollsters were paid on the basis of their accuracy, many would be out of business.
I am of the opinion that pollsters should reflect the local voter registration in their sampling. If 30% are Dem then 30% of their sample size should be Dem. If it is higher then use that percentage.
Stop making sense
Today, polls are about reinforcing narratives, not predicting outcomes.
-PJ
In ‘16, they designed their polls to reflect their dem desires.
Trump supporters simply won't admit to a pollster that they support Trump.The only time they'll acknowledge such support is on Election Day...IN THE POLLING BOOTH...not to any pollster standing outside.
“Today, polls are about reinforcing narratives, not predicting outcomes. “
Sadly, you are correct. That is why they get paid.
Trump has good instincts. He needs to trust himself.
The GOP-e blocked the true teeth in the “Do Not Call” list so they could use it to contact donors and voters.
People told them they were making a mistake.
Now no one answers cell phones unless its a known caller.
Thus polling data is worthless.
If the public loses confidence in polling, it’s futile regardless of accuracy. It’s the same as confidence in the reporting of news. If it’s not believed, it doesn’t matter if it’s true or not. This is a problem modern journalists are incapable of understanding.
"Shaming" stops citizens from telling pollsters - but IT DOES NOT STOP THEM FROM PULLING THE LEVER FOR PRESIDENT TRUMP.
That's what threw them off last election... citizens said undo' undecided or 'hillary' - but they voted Trump.
I do not have a land line. I have an area code from 5 years ago in another state and I do not take any unsolicited calls. So right there their model is out of date as I am not the only one in that boat
"Shaming" stopped citizens from telling pollsters the truth - but IT DID NOT STOP THEM FROM PULLING THE LEVER FOR TRUMP.
That's what threw them off last election... citizens said 'undecided' or 'hillary' - but voted Trump.
One factor was that polls became weaponized, no longer as snapshots of the public mind but tools to influence it. It's pretty obvious that you can't have it both ways. That led to two sorts of polling: one weaponized one for public consumption, another internal sort for the private eyes of political campaigns, who make tactical decisions on that basis and so are unlikely to wish to share that information as a result. Some surprisingly experienced people appear to have confused the two.
Another factor was that random operator-moderated calls are expensive and slow. To be accurate these need large sample sizes, meaning large calling centers and lots of time. It obviously takes longer to call 10,000 people than to call 100. You can get away from that by carefully crafting sample to consist of key demographics but error rates rise inexorably as sample size goes down.
Enter data analytics. This was, according to the authors of Shattered: Inside Hillary Clinton's Doomed Campaign, a principal factor in her loss; according to them, her personal intervention led to turning off traditional polling in favor of the sort of database mining that depends on past voting performance. It's faster and much cheaper, and time was of the essence because the vote loomed. But there is an inherent lag in that system in the reflection of numerous small-scale deltas that added up to a defeat. (I'm surprised Robby Mook is still drawing breath).
What numerous market-research companies do - this is hinted at in the article - is to develop large private sample pools of people who actually agree to be called. You lose randomness in this but you gain in call completion percentage. Where candidates demand daily updates, there simply isn't time to use large, random databases and human call centers anymore, at least in theory, although frankly no campaign on the record has actually made tactical decisions that rapidly. Largely, unreasonable expectations on the part of the customer are driving this into unknown territory, not necessarily the technology.
The sad upshot is that if you're hearing poll results in the media, they're at least suspect and probably tainted. 2016 in America and 2019 in Australia are the result, and it's useless for polling companies to complain that they didn't see it coming because even if they did they wouldn't tell you.
They’re all going to quit and enjoy exciting & secure careers in Climatology.
Meet the new lies. Same as the old lies.
It’s not simply Trump voters won’t talk to pollsters it is the fact Trump attracted people to the pollls who had abandoned politics so they were filtered out by pollsters as unlikely because they hadn’t voted in several cycles before.
Many of those same type of voters who didn’t vote for Trump (or anyone) last time will be coming to vote for him this time, and pollsters will once again consider them unlikely
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.