Support the Timberjay by making a donation.

Serving Northern St. Louis County, Minnesota

Did new NYT poll of the Eighth get it wrong?

The poll's author admits it could be off by as much as ten points

Marshall Helmberger
Posted 10/15/18

REGIONAL— Is the race to replace Congressman Rick Nolan a nail-biter or a blowout in the making? It could be either one, depending on what one makes of recent polling on the race. Back in …

This item is available in full to subscribers.

Please log in to continue

Log in

Did new NYT poll of the Eighth get it wrong?

The poll's author admits it could be off by as much as ten points

Posted

REGIONAL— Is the race to replace Congressman Rick Nolan a nail-biter or a blowout in the making? It could be either one, depending on what one makes of recent polling on the race. Back in September, the New York Times published a poll showing a near dead-heat, with DFLer Joe Radinovich clinging to a 44-43 percent lead, which suggested it would be the same kind of down-to-the-wire race experienced in previous election cycles. Internal polling by the Radinovich campaign has also shown the race neck-and-neck, which lent credence to the narrative that race is very tight.

That narrative was upset, however, when the New York Times published a new poll this past weekend, which put Stauber up by 15 points, 49-34 percent. For the Radinovich campaign, it was not a happy Monday. While it’s just one poll, such results can have a major impact on a race, particularly where one side, in this case the GOP, is pouring millions of dollars into attack ads against Radinovich. As Democratic-leaning funders consider whether to invest their own resources to defend Radinovich— by running ads attacking Stauber, for example— a poll result like this can be the kiss of death. If they decide the race is lost, the money people can quickly close the wallets and leave the candidate twisting in the wind. It can also affect how the public views the race, along with their willingness to donate money or volunteer for get-out-the-vote efforts. Indeed, the Radinovich campaign has seen an immediate slowing of the small dollar individual donations that the campaign has, up until now, been able to raise very successfully. The campaign, just last week, reported it had raised a remarkable $1.25 million in the third quarter, almost all of it in small donations. Like it or not, polls can have a very significant impact on political races.

So, does the latest poll reflect the state of the race, or something else? The Timberjay put that question to Don Levy, the director of the Siena Research Center, who has been conducting the polling for the Times, as well as the Time’s polling analyst Nate Cohn, who has led the newspaper’s polling efforts around the country.

Their answers were surprising.

First, keep in mind that the numbers that appear in most polls do not reflect the actual raw data. Instead, the results are typically weighted based on an expectation by the pollsters of who is likely to turn out in any given election. While the public might believe that election results reflect the views of the public, they actually reflect the views of those who vote, and that electorate changes with each election. Turnout is key, so pollsters make educated guesses in their polling about who will turn out. Making this call for the Times is Nate Cohn’s job.

In the Times’ September poll, Cohn assumed that the 2018 electorate in the Eighth District would be comprised of 34 percent Democrats, 32 percent Republicans, and 29 percent independents. And based on the 500-600 people they actually sampled, they concluded that such a hypothetical electorate would favor Radinovich by one point.

The raw results in the Times’ October poll actually weren’t that different, as Cohn explained on Monday in an analysis he posted on the Times website after getting questions from the Timberjay and others.

Assuming the same electorate as the Times used in September, Stauber would have led by just four points, which is within the poll’s margin of error. But the Times assumed a very different electorate in its October poll, one comprised of 37 percent Republicans, 27 percent Democrats, and 29 percent independents, which would be a remarkable change from previous elections in the Eighth. They also assumed a turnout of 305,00 voters in the district, which would vastly surpass the turnout in previous mid-term elections, and that most of that increase would be due to greater GOP turnout.

“We have no evidence to show that’s going to be the case,” said Bennett Smith, campaign chair for the Radinovich campaign, who says the Times poll isn’t anywhere close to their own internal polling.

Donna Victoria, of Victoria Research, who has been polling for the Radinovich campaign since the primary, agreed. “I’m just floored to read what Nate Cohn wrote,” she said. “They [the Times] sold this poll for four days essentially as entertainment. Now Cohn’s written an article saying it could be off by as much as ten points?”

Cohn never really explained why he changed the methodology as much as he did, although he said it reflects, at least to some degree, the different responses pollsters received from the people they surveyed in October, versus those they spoke to the previous month. “Last time, voters disapproved of Mr. Trump by one point,” wrote Cohn, defending the poll. “Now they approve by 18. Last time, Democrats outnumbered Republicans by two percentage points; now Republicans outnumber by Democrats by 10.”

Victoria said such an electorate “doesn’t exist to my knowledge,” in the Eighth.

Could this poll be one that simply missed the mark? Victoria said even the best polls have a confidence margin of only 95 percent. “That means one-in-twenty is just going to be awry,” she said. Cohn, in his assessment of the Eighth District poll, acknowledges that sometimes poll results turn out “weird.”

Still, Victoria acknowledged that Radinovich could really be down by a few points, in part reflecting the impact of more than $6 million worth of attack ads directed at him by GOP-leaning Super PACs over the past four weeks. Those ads have run heavily in the very expensive Twin Cities media market and appear to have had an impact in that portion of the district, which tends to lean more Republican than other parts of the district. In the September poll, Radinovich trailed in that portion of the district by nine points. In the October poll, he was down by 34 points, a swing of 25 points in a month. While the poll showed Radinovich underperforming his September results across the district, the Twin Cities exurbs showed his biggest decline, and by a wide margin. The Radinovich campaign, which has not seen much Super PAC backing in the general election, has not been able to respond to the relentless attack ads that have aired heavily on television and online in the Twin Cities area.

Cohn points out that polling can be difficult in Minnesota because the state does not currently have party registration. In most other states, people have to declare a political affiliation when the register. That will soon be the case in Minnesota as well, but that change has yet to take effect. That means pollsters have to rely on the self-identification of an individual who is being surveyed, and that isn’t always an accurate reflection and it makes it difficult to accurately weight the results.

Victoria doesn’t buy that explanation. She said she polls in plenty of states without party registration and because of that she doesn’t weight her results by party affiliation, since those affiliations often change.

In any case, the Times poll likely won’t be the last word on the race. Victoria said her firm will be re-visiting the district before the Nov. 6 election. And the Times is considering taking another look at the race, according to Levy, of Siena Research.