Over the past sixty hours, there’s been a lot of discussion concerning how political polling this election cycle missed the mark, especially in states like Maine, where the available body of public opinion research incorrectly suggested that entrenched incumbent Republican Susan Collins was toast.
Fire the pollsters has now become something of a rallying cry on Twitter and other social media platforms, in varying flavors.
“Fire pollsters, hire local reporters,” tweeted Caroline Darya Framke.
“Fire pollsters and replace them with film programmers,” tweeted Adam Piron.
“Fire all pollsters. Let’s not bother with them in 2024,” tweeted Dale Maharidge.
“Fire all pollsters and go back to reading animal entrails the way God intended us to predict the future,” joked Kyle Smith.
“Polling seems to be irrevocably broken, or at least our understanding of how seriously to take it is,” Margaret Sullivan wrote in a Washington Post op-ed.
“The polling industry is a wreck, and should be blown up,” Politico declared.
While our team at NPI can certainly understand the frustration with important results not turning out the way the research indicated they would, not every pollster and not every poll got it wrong this year.
Consider our poll from three weeks ago.
Public Policy Polling of North Carolina surveyed six hundred and ten likely Washington State voters for us on October 14th and 15th. 45% participated via landline, and 55% via text; yielding a margin of error of +/- 4.0% and a confidence interval of 95%. We asked about every single contested statewide race on the ballot except for ESJR 8212, a proposed constitutional amendment, and found:
In each race, the candidate that we found to be ahead in our October 2020 polling is winning, with no exceptions, not even in the closer races, like Secretary of State or Superintendent of Public Instruction.
The races that were closest in our polling have all been called, either by the Associated Press or by observers (including Secretary of State, the closest, which pitted Kim Wyman against NPI’s senior boardmember Gael Tarleton.)
In races where we had a low number of undecided voters, even the margins in our polling are pretty close to the current election results.
For example, in the presidential race, we found Biden with 60% support, Trump with 37% support, and 2% not sure. In the actual election, as of the third day of counting, Biden had 59.13%, Trump had 37.79%, and other candidates had less than 2%. That’s an almost perfect correlation with our poll finding.
In the gubernatorial race, meanwhile, Democratic Governor Jay Inslee currently has 57.74%; our poll had him at 56%. Republican challenger Loren Culp achieved 40% in our poll and presently has 41.95%% in the election; it appears that he and Inslee split the small number of voters who were undecided.
It’s not unusual for there to be some deviation between poll findings and actual results; the fact that there is barely any in this case shows that our sample truly was representative of the 2020 Washington State electorate.
Just like it has been for several election cycles now.
NPI’s partnership with Public Policy Polling has been going strong for over half a decade, consistently producing data that has foreshadowed election results.
Take the 2018 election.
In a poll conducted six months before the election, NPI’s research found Maria Cantwell with a sixteen point lead over Susan Hutchison.
This was the very first statewide poll pitting Cantwell against Hutchison; it was conducted within one hundred hours of the close of filing by PPP for NPI.
Though half a year elapsed between the poll and the certification of the midterms, Cantwell went on to win by.… wait for it… 16.86 points.
The numbers:
NPI’s research (May 22nd-23rd, 2018)
QUESTION: If the election for U.S. Senate were held today and the candidates were Democrat Maria Cantwell and Republican Susan Hutchison, who would you vote for?
ANSWERS:
- Maria Cantwell: 52%
- Susan Hutchison: 36%
- Not sure: 12%
Election (November 6th, 2018)
VOTES:
Maria Cantwell: 58.43%
Susan Hutchison: 41.57%
Total Votes: 3,086,168
At NPI, we care about the quality of data.
If the inputs are bad, then the outputs will be bad (garbage in, garbage out).
We are emphatically subjective in our views as an advocacy organization — we’re very transparent about what our values, principles, and policy directions are — but as a research organization, we aggressively strive for objectivity when we field a survey or another research instrument. A scientifically accurate poll needs to have a sample that is representative of the electorate.
And neutral wording is essential. Essential.
We cannot know what people’s views really are on the issues if we were to ask loaded questions that suggest their own answers.
Tim Eyman’s push polls — those falsely-labeled “advisory votes” that keep showing up on Washingtonians’ ballots — are a perfect example of how not to take the pulse of the electorate. Those are pieces of propaganda… utterly useless for ascertaining what voters really think.
And by the way… most Washington voters who have an opinion about Eyman’s push polls want them gone. We know because we’ve asked. Repeatedly.
Each time we have asked, it has been in the form of a neutrally worded question that pitted Tim Eyman’s best honed arguments against our own. Our arguments have beaten his every single time, as this presentation explains.
During the past few weeks, we’ve had more than one right wing commenter show up and insist that our research was not credible. For example, Marty Wentz wrote: “your [sic] so full of it. Insley [sic] has about as much a chance to win as I do.” Nick Kellogg sneered: “I think this if [sic] a fake news and a fake poll.”
I imagine that November 3rd was a rough night for Marty and Nick.
Like other Trump and Culp fans, they sadly dwell in their own “Upside Down” alternative universe, where anything that does not fit with the party line is immediately assailed as fake and fabricated, and conspiracy theories are rampant.
What this election has actually proven that our research is producing useful, credible data that can help anyone interested in Washington State politics understand how voters perceive issues and candidates.
Even in a year of polling misses, our polling here in Washington State was on the mark. If you’d like to help fund our next survey, we invite you to become a member of the Northwest Progressive Institute or make a contribution.
Friday, November 6th, 2020
NPI’s 2020 polling correctly foreshadowed the result in every Washington statewide race
Over the past sixty hours, there’s been a lot of discussion concerning how political polling this election cycle missed the mark, especially in states like Maine, where the available body of public opinion research incorrectly suggested that entrenched incumbent Republican Susan Collins was toast.
Fire the pollsters has now become something of a rallying cry on Twitter and other social media platforms, in varying flavors.
“Fire pollsters, hire local reporters,” tweeted Caroline Darya Framke.
“Fire pollsters and replace them with film programmers,” tweeted Adam Piron.
“Fire all pollsters. Let’s not bother with them in 2024,” tweeted Dale Maharidge.
“Fire all pollsters and go back to reading animal entrails the way God intended us to predict the future,” joked Kyle Smith.
“Polling seems to be irrevocably broken, or at least our understanding of how seriously to take it is,” Margaret Sullivan wrote in a Washington Post op-ed.
“The polling industry is a wreck, and should be blown up,” Politico declared.
While our team at NPI can certainly understand the frustration with important results not turning out the way the research indicated they would, not every pollster and not every poll got it wrong this year.
Consider our poll from three weeks ago.
Public Policy Polling of North Carolina surveyed six hundred and ten likely Washington State voters for us on October 14th and 15th. 45% participated via landline, and 55% via text; yielding a margin of error of +/- 4.0% and a confidence interval of 95%. We asked about every single contested statewide race on the ballot except for ESJR 8212, a proposed constitutional amendment, and found:
In each race, the candidate that we found to be ahead in our October 2020 polling is winning, with no exceptions, not even in the closer races, like Secretary of State or Superintendent of Public Instruction.
The races that were closest in our polling have all been called, either by the Associated Press or by observers (including Secretary of State, the closest, which pitted Kim Wyman against NPI’s senior boardmember Gael Tarleton.)
In races where we had a low number of undecided voters, even the margins in our polling are pretty close to the current election results.
For example, in the presidential race, we found Biden with 60% support, Trump with 37% support, and 2% not sure. In the actual election, as of the third day of counting, Biden had 59.13%, Trump had 37.79%, and other candidates had less than 2%. That’s an almost perfect correlation with our poll finding.
In the gubernatorial race, meanwhile, Democratic Governor Jay Inslee currently has 57.74%; our poll had him at 56%. Republican challenger Loren Culp achieved 40% in our poll and presently has 41.95%% in the election; it appears that he and Inslee split the small number of voters who were undecided.
It’s not unusual for there to be some deviation between poll findings and actual results; the fact that there is barely any in this case shows that our sample truly was representative of the 2020 Washington State electorate.
Just like it has been for several election cycles now.
NPI’s partnership with Public Policy Polling has been going strong for over half a decade, consistently producing data that has foreshadowed election results.
Take the 2018 election.
In a poll conducted six months before the election, NPI’s research found Maria Cantwell with a sixteen point lead over Susan Hutchison.
This was the very first statewide poll pitting Cantwell against Hutchison; it was conducted within one hundred hours of the close of filing by PPP for NPI.
Though half a year elapsed between the poll and the certification of the midterms, Cantwell went on to win by.… wait for it… 16.86 points.
The numbers:
NPI’s research (May 22nd-23rd, 2018)
Election (November 6th, 2018)
At NPI, we care about the quality of data.
If the inputs are bad, then the outputs will be bad (garbage in, garbage out).
We are emphatically subjective in our views as an advocacy organization — we’re very transparent about what our values, principles, and policy directions are — but as a research organization, we aggressively strive for objectivity when we field a survey or another research instrument. A scientifically accurate poll needs to have a sample that is representative of the electorate.
And neutral wording is essential. Essential.
We cannot know what people’s views really are on the issues if we were to ask loaded questions that suggest their own answers.
Tim Eyman’s push polls — those falsely-labeled “advisory votes” that keep showing up on Washingtonians’ ballots — are a perfect example of how not to take the pulse of the electorate. Those are pieces of propaganda… utterly useless for ascertaining what voters really think.
And by the way… most Washington voters who have an opinion about Eyman’s push polls want them gone. We know because we’ve asked. Repeatedly.
Each time we have asked, it has been in the form of a neutrally worded question that pitted Tim Eyman’s best honed arguments against our own. Our arguments have beaten his every single time, as this presentation explains.
During the past few weeks, we’ve had more than one right wing commenter show up and insist that our research was not credible. For example, Marty Wentz wrote: “your [sic] so full of it. Insley [sic] has about as much a chance to win as I do.” Nick Kellogg sneered: “I think this if [sic] a fake news and a fake poll.”
I imagine that November 3rd was a rough night for Marty and Nick.
Like other Trump and Culp fans, they sadly dwell in their own “Upside Down” alternative universe, where anything that does not fit with the party line is immediately assailed as fake and fabricated, and conspiracy theories are rampant.
What this election has actually proven that our research is producing useful, credible data that can help anyone interested in Washington State politics understand how voters perceive issues and candidates.
Even in a year of polling misses, our polling here in Washington State was on the mark. If you’d like to help fund our next survey, we invite you to become a member of the Northwest Progressive Institute or make a contribution.
# Written by Andrew Villeneuve :: 1:04 PM
Categories: Elections
Tags: Research Polling Retrospectives, WA-Ballot, WA-Exec, WA-Gov
Comments and pings are currently closed.