Must-read political junkie stuff
Thursday, Dec 19, 2019 - Posted by Rich Miller
* From pollster Dave Fako’s latest client newsletter…
Many of our recent messages have discussed the changes being make in the public opinion research profession, adapting to the new communication habits of potential respondents to surveys and how to assess the quality, reliability and accuracy of polls. In these messages we have shared information from Pew about changes in their methods and the status of polling.
Well, Pew has put out another great article, which details some topics we have written about too, particularly methodological disclosure, cell phones and the use of online / Internet methods. We strongly recommend it to anybody who is a consumer of public opinion research or has an interest in the profession.
* From that Pew story…
Is polling broken?
Here’s a myth that we can set aside right at the start: Polling is not “broken.” Well-designed and carefully administered surveys still work, and there’s plenty of empirical evidence to back this up.
Even outside the U.S., there is substantial evidence that polling hasn’t witnessed a substantial decline in accuracy. A comprehensive review of polling accuracy published in 2018 found that “relying on vote intention polls from more than 200 elections in 32 countries over a period of more than 70 years, there is no evidence that poll errors have increased over time….”
But there are developing issues, so go read the rest.
- very old soil - Thursday, Dec 19, 19 @ 11:59 am:
Great article
- Lester Holt’s Mustache - Thursday, Dec 19, 19 @ 12:20 pm:
Fako writes a good story, and his sourcing is very well done. He’s correct that polling isn’t broken, but that argument has been used (in my personal opinion) as cover for laziness within Clintons campaign and among dem voters generally in 2016. Summer polling showed Hillary winning big, and I think this caused dem voters and Clinton campaign folks to take their foot off the gas. I heard it myself in the days before the election - “oh there’s no way trump can win” and “eh, I’m not too worried about it, Hillary will win”. Sure, undecideds broke late to trump like Fako says. But a whole lot of dems didn’t bother to vote because the early polling that showed a cakewalk for Hillary got stuck in theirs heads (that and Hillary not being particularly well-liked among some traditional dem voters). Dems corrected for that overconfidence in 2018, and I assume will do so again in 2020.
- SSL - Thursday, Dec 19, 19 @ 12:29 pm:
There’s a lot in the article. Polling was done a huge disservice in 2016, and many won’t ever believe a poll they don’t like again. It’s unfortunate.
- Downstate - Thursday, Dec 19, 19 @ 2:26 pm:
It seems that last week’s polling in the UK election was off considerably as well.
- Dave Fako - Thursday, Dec 19, 19 @ 3:27 pm:
In response to Lester…
My main point of distributing this article from Pew is to help put information out there about how the public opinion profession is changing and adapting, and to provide information that can be used to assess the quality and reliability of polling. These aspects have been a focus of our business for over twenty years, with our firm constantly assessing, adapting and innovating in the public opinion research profession to ensure quality, reliability and accuracy. We feel its an important topic to promote. There is a significant lack of information within the various networks of consumers of polling on how to properly interpret and assess the quality of polling, so its somewhat of a guide. My message and the Pew article has nothing directly related to the 2016 election, although its referenced in the Pew article in the simple context that it was accurate on a national level.
- Been There - Thursday, Dec 19, 19 @ 4:27 pm:
Fako and his colleagues have their work cut out for them. 25 years ago I wouldn’t answer polls honestly because I didn’t trust who was actually on the other line. Then after being in the political business and knowing that most pollsters are reputable I would answer them. Now unless the call is says it’s a pollster I probably never answer. Either way it’s a tougher business than
- low level - Thursday, Dec 19, 19 @ 7:21 pm:
== It seems that last week’s polling in the UK election was off considerably as well.==
To a certain extent but pretty much every poll had the Conservatives winning by +10 or more which is what happened.