Trump's lawsuit over Selzer poll is wrong on the law and the facts

a weak case by any measure

Screenshot from President-elect Donald Trump’s December 16 news conference. His remarks about Selzer and the lawsuit begin around the 55:00 mark of this video.

This original reporting first appeared at Bleeding Heartland and is shared here as part of the Iowa Writers Collaborative. For regular emails linking to all recent Bleeding Heartland articles and commentary, subscribe to the free Evening Heartland newsletter. If your email provider truncates this post, you can read the whole article without interruption at this link. Ed Tibbetts’ take on Trump’s new lawsuit is also worth your time.

President-elect Donald Trump followed through this week with his threat to sue pollster Ann Selzer and the Des Moines Register over the final pre-election Iowa Poll, which showed Vice President Kamala Harris leading Trump by 47 percent to 44 percent.

Many others have pointed out that Trump’s lawsuit is part of his broader “revenge tour” and “war on journalism.” In Greg Sargent’s words, the case is “putting people in the media and polling on notice that they will face real legal harassment if they anger or criticize Trump.” The president-elect admitted during a December 16 press conference that he will use lawsuits to influence news coverage: “I think you have to do it, because they’re very dishonest. We need a great media, we need a fair media.”

This post will focus on the legal, factual, and logical problems with the court filing.

THE STATE OF PLAY: CASE REMOVED TO FEDERAL COURT

Here’s a link to the full text of the petition filed in Polk County District Court on December 16. The defendants are Selzer, her polling firm Selzer & Co., the Des Moines Register and Tribune Company, and the Gannett Corporation, which owns the Register and hundreds of other newspapers across the country.

Trump brought the suit under Iowa Code Chapter 714H, which provides a private cause of action for consumer fraud. He characterizes the poll as an “unfair act or practice” that deceived Iowa consumers and Iowans who contributed to the Trump campaign. The court filing further asserts that “Defendants engaged in this misconduct to improperly influence the outcome of the 2024 Presidential Election.”

Even though Trump won Iowa by a 13-point margin—indicating the faulty poll didn’t suppress his support here—he is seeking damages to compensate him for having to “expend extensive time and resources, including direct federal campaign expenditures, to mitigate and counteract” the poll’s impact. He also asked the court to issue an order blocking defendants “from publishing or releasing any further deceptive polls designed to influence the outcome of an election” and requiring them to release “all data and information” used to produce and publish this year’s final Iowa Poll.

Trump’s attorneys in this case are Edward Paltzik of the New York-based Bochner law firm and Alan Ostergren, who has represented various Iowa Republican politicians and conservative causes. (Disclosure: In an unrelated case, Ostergren is representing a plaintiff in a libel lawsuit filed against me, Bleeding Heartland, and the guest author of a 2022 post.)

Lark-Marie Anton, a spokesperson for The Des Moines Register, said in a statement,

We believe this lawsuit is without merit. We have acknowledged that the Selzer/Des Moines Register pre-election poll did not reflect the ultimate margin of President Trump’s Election Day victory in Iowa by releasing the poll’s full demographics, crosstabs, weighted and unweighted data, as well as a technical explanation from pollster Ann Selzer. We stand by our reporting on the matter and will vigorously defend our First Amendment rights.

It’s encouraging that Gannett and the Register will fight, in contrast to ABC, which recently settled a defamation claim on terms very favorable to Trump.

Attorneys for Gannett landed the first punch on December 17 by removing the case from state court to the U.S. District Court for the Southern District of Iowa. The Notice of Removal pointed out that the case involves citizens of different states, that the plaintiff is seeking damages exceeding $75,000, and that three defendants (Selzer, her polling company, and the Des Moines Register) had not yet been served.

The legal maneuver (known as “snap removal”) means the Polk County District Court will not be able to take any further action on this matter unless and until a federal judge sends the case back to state court.

Whether Trump’s lawsuit proceeds in state or federal court, his case has many weaknesses.

SUIT MISAPPLIES CONSUMER FRAUD LAW

For a legal analysis of the claims, I turned to Bill Brauch. He currently chairs the Polk County Democrats. But from 1995 to 2015, he was the assistant attorney general leading the Consumer Protection Division in the Iowa Attorney General’s office. In that role, he lobbied state lawmakers for years to approve a private cause of action for consumer fraud, finally succeeding in 2009 (the third year of a Democratic trifecta). Iowa Code Chapter 714H ostensibly provides the legal basis for Trump’s lawsuit.

Iowa was the last state in the country to enact such a law. Before 2009, only the Iowa Attorney General’s office could bring consumer fraud claims. The office lacked the staff and resources to try every case, which meant many Iowans who had been defrauded had no recourse.

Brauch told me via email, “While the lawsuit has clearly been crafted by someone who has read Chapter 714H, Iowa’s private remedy for consumer fraud, it appears to misapply the law in several respects.”

For one thing, the lawsuit “purports to bring the action not only on behalf of Donald Trump, but on behalf of other Republicans and other voters. There is no such authority for an individual to bring a lawsuit on behalf of anyone other than the individual alleging a violation of chapter 714H, except in the context of a class action lawsuit. This lawsuit was not filed as a class action lawsuit.”

Another problem: Iowa’s statute requires a plaintiff to show there was fraud related to the sale, lease, or advertisement of “merchandise.” Brauch told me,

In this case, the suit correctly identifies the merchandise as the newspaper. But the suit does not allege any deception, misrepresentation or unfair practices in the Register’s advertisement or sales of its newspapers. The purchasers here got what they paid for—a newspaper. Rather, the suit alleges that a story within the newspaper was misleading, therefore making the sale or advertisement of the newspaper misleading. In other words, they are attacking the content of the newspaper, not the sale or advertisement of the newspaper itself. The content of a media source, other than an advertisement for merchandise it might contain, is subject to strong First Amendment protection.

Having lobbied Iowa legislators to approve a private remedy for consumer fraud, “I can tell you that it certainly was not created for situations such as this,” Brauch added. “It was created to address the need for consumers who purchased faulty used cars or who were victims of home-improvement, sweepstakes or other scams to be able sue the perpetrators of these scams to receive compensation and have their attorney fees paid by the defendant if they prevailed.”

Brauch said Iowa’s consumer fraud law “was not created to give a cause of action to individuals who feel harmed by the content of a news story. If that were the case, our courts would be so inundated with private lawsuits they wouldn’t have the resources to handle anything else.”

Eugene Volokh, a longtime law professor and author of a textbook on the First Amendment, wrote on December 18 that Trump v Selzer is “likely going nowhere.” He explained the relevant case law, which has held “the First Amendment generally bars states from imposing liability for misleading or even outright false political speech, including in commercially distributed newspapers—and especially for predictive and evaluative judgments of the sort inherent in estimating public sentiment about a candidate.”

So even if Iowa’s consumer fraud statute enabled this kind of lawsuit against a news organization, the case would fail in court on other grounds.

The facts asserted in Trump’s lawsuit are equally problematic.

A SELECTIVE HISTORY OF SELZER POLLS

The court filing is sloppy, misspelling the surname of Republican Party of Iowa state chair Jeff Kaufmann throughout as “Kaufman.”

It also (bizarrely) uses “the Harris Poll” as a shorthand reference for Selzer’s final Iowa Poll for the Des Moines Register and Mediacom. The Harris Poll is a different company with no connection to Selzer. Many state polls overestimated support for Kamala Harris. That doesn’t make them “Harris polls.”

More important, the section on “factual allegations” cites three examples to demonstrate that “Selzer has quietly used her polls to try and influence recent races in favor of Democrats.” The supposed evidence:

• A poll that was in the field four weeks before election day 2022, and published two weeks before that election, showed Attorney General Tom Miller leading Republican challenger Brenna Bird by 16 points. Bird ended up winning by 50.8 percent to 49.1 percent.

• Selzer’s final poll of the 2018 governor’s race showed Democrat Fred Hubbell 2 points ahead. He lost to Kim Reynolds by 50.3 percent to 47.5 percent.

• Iowa polls from June and September 2020 showed Democratic challenger Theresa Greenfield ahead of Senator Joni Ernst, after which Selzer “belatedly put Ernst ahead by four points” in her final poll that year.

The lawsuit asserts, “These races were not a matter of being wrong in the end—they exposed that Selzer and the rest of the Defendants were manufacturing fake support for Democrat candidates.”

That narrative is wrong in several ways. First, it omits the fact that while an October 2022 poll did show Miller 16 points ahead of Bird, Selzer’s final pre-election survey found the GOP challenger gaining fast and trailing Miller by only 2 points. That’s consistent with the result: a narrow victory for Bird.

Second, it doesn’t acknowledge that polls conducted by several other firms also showed Hubbell ahead in the 2018 governor’s race and Greenfield leading Ernst during the 2020 campaign (click here and here for examples). Selzer’s final 2020 poll was closer than most to Ernst’s 6.6-point margin of victory.

Third, Trump’s lawsuit doesn’t mention other Selzer polls that showed Republican candidates ahead, sometimes by larger margins than in contemporaneous surveys. As Bleeding Heartland discussed here, Selzer was the first pollster to find a significant lead for Ernst in the 2014 U.S. Senate race. Her final pre-election poll that year looked like an outlier, but was very close to Ernst’s eventual winning margin.

Similarly, Selzer’s final pre-election polls in 2016 and 2020 showed a larger lead for Trump in Iowa than any other public pollster at the time (see here and here), and were close to the mark.

Trump alluded to the defendant’s reputation for accuracy during his December 16 press conference, when he described Selzer as “a very, very good pollster who got me right all the time, and then just before the election, she said I was going to lose by 3 or 4 points.” Which is it? Was she “very, very good” at her job, or did she have a long history of misrepresenting public opinion in order to help Democrats?

Ironically, one of Selzer’s bigger misses—which this lawsuit doesn’t mention—came in January 2016, when her final poll before the Iowa caucuses showed Trump leading the GOP field with 28 percent to 23 percent for Senator Ted Cruz. In fairness, most other pollsters also found Trump ahead among Iowa Republicans. However, Cruz won the caucuses with 27.6 percent to 24.3 percent for Trump and 23.1 percent for Marco Rubio.

MISREPRESENTING SOURCES OF POLLING ERRORS

Trump’s lawsuit cites a post by James Piereson for The New Criterion, which stated in part,

The Selzer poll, with a margin of error of 3.4, missed the real outcome by 16 points, or by as many as five standard deviations from the true result as revealed on election day.

What are the odds of drawing such a sample by legitimate means? Answer: roughly one time in 3.5 million trials. In other words, given these odds, the results in the Iowa poll likely did not come about by “honest error.”

A few days before Trump filed suit, Kaufmann amplified that scientific-sounding claim on X/Twitter.

Assuming Selzer polled a perfectly representative sample of Iowa voters and used flawless methodology, the odds of being off by five standard deviations may be one in 3.5 million. Putting “honest error” in quotes implies the pollster was dishonest. But other factors can easily explain a big polling miss.

Selzer never adjusts her findings to match the recalled vote from last election or the partisan makeup of turnout from previous elections. She has valid reasons for not wanting to assume Iowa’s future electorate will have the same composition as past elections. But this year, the methodology served her poorly. If she had weighted the findings using recalled 2020 presidential vote, her final 2024 poll would have found Trump 6 points ahead of Harris.

Selzer also has avoided weighting by education level, even though partisan preferences are increasingly correlated with whether a voter has a college degree. During an appearance on the Iowa PBS program “Iowa Press” last week, she explained that in previous election cycles her company looked at weighting by education, but found “it made some things worse,” throwing off “the proportion of people who were married or not married.”

Research has shown Democrats are more likely than Republicans to agree to participate in a telephone survey. That “response bias” alone could have produced a sample with too many Harris supporters and not enough voters leaning to Trump.

Finally, the Utah-based company that conducts telephone surveys for Selzer finds participants through random digit dialing. Many other political pollsters draw respondents from lists of registered voters. Selzer prefers her method because in theory, it gives every Iowa adult an equal chance of landing in her poll sample. Natalie Jackson, vice president of the polling firm GQR, recently posted that random digit dialing “is no longer viable for election polling.” She observed that the sampling method had “always worked best in populations that look like Iowa’s”—older and whiter than most of the country—which could explain why Selzer’s previous polls usually panned out.

NO EVIDENCE TO BACK UP ALLEGED MOTIVE

Piereson’s article acknowledged other “possible sources of errors,” such as “some difficulty in finding truly random or representative samples in the age of cell phones and the internet, and when many voters do not speak to pollsters.” Nevertheless, he jumped to the following conclusion:

It is more likely that someone deliberately manipulated the sample so that it included too many Democrats, or simply made up the numbers as they came in for the purpose of giving confidence to Harris voters and worry for Trump supporters, or to bring national attention to a poll taken in a state not regarded as competitive. The poll did receive national attention and was widely discussed. Selzer appeared on television interviews to talk about the poll and its implications. If the goal was to promote the poll, then the gambit succeeded—at least until election day, when it was revealed to be ridiculously far off the mark.

Trump’s lawsuit quotes that paragraph in full, and asserts repeatedly that the defendants “manipulated” or “doctored” the poll numbers “for the improper purpose of deceptively influencing the outcome of the 2024 Presidential Election.” No evidence is offered to back up that claim: no internal memos or emails, no affidavits from current or former employees of Selzer & Co or the Register. Rather, the claim rests on public statements of opinion by Trump (“She knew exactly what she was doing”; the poll “caused great distrust and uncertainty at a very critical time”).

Selzer’s final Iowa Poll sampled 808 “likely voters” who turned out not to be representative of the 1.67 million Iowans who cast ballots in the 2024 general election. That points to problems with the sampling method, likely voter screen, and/or weighting decisions, but it’s not proof of “concealment, suppression, and omission of material facts,” as Trump’s lawsuit claims.

The court filing highlights that someone may have leaked the findings hours before the Register published the poll, “giving rise to a clear inference that the Harris Poll was intended to aid Harris and harm President Trump.” What? All that suggests is someone outside the organization gained access to the numbers early. We don’t know for sure there was a leak; someone may have found a way to access the data on the Des Moines Register or USA Today websites before the story was live.

Either way, that doesn’t prove Selzer or anyone at the Des Moines Register or Gannett was trying to help Harris. The poll would have generated national media coverage regardless.

The reality is that an accurate survey is incredibly valuable to nonpartisan pollsters and news organizations.

In addition, Selzer’s track record suggests she has not manipulated numbers to reach any desired result for the Des Moines Register. Remember: her polls found comfortable leads for Trump in Iowa shortly before the 2016 and 2020 elections, even though the Register’s editorial board had endorsed Hillary Clinton and Joe Biden for president. Why would she put her thumb on the scale for Harris in 2024, when the Register (like other Gannett properties) did not endorse a candidate in the presidential race?

NO EVIDENCE OF HARM DONE TO TRUMP

One of the biggest logical errors in Trump’s lawsuit is the unsubstantiated claim that the poll hurt his campaign and therefore cost him money:

Due to Defendants’ actions, the public could not discern who was truly leading in the Iowa Presidential race and, as a result, were, or could have been, badly deceived into thinking that Harris was leading the race. […]

Because of Defendants’ false, misleading, and deceptive conduct, President Trump has sustained actual damages due to the need to expend extensive time and resources, including direct federal campaign expenditures, to mitigate and counteract the harms of the Defendants’ conduct.

Trump won Iowa for the third time, by his largest margin yet. There is no evidence the Selzer poll changed voter preferences or behavior in any meaningful way. More than 600,000 Iowans had already cast ballots before the survey results became public. You could even argue (as Selzer later suggested) that the final Iowa Poll motivated more Republicans to turn out on election day. I am skeptical, since the Iowa GOP’s well-funded turnout operation had been engaged for months.

The Trump campaign was already spending tens of millions of dollars on advertising and GOTV in the home stretch, and had deployed those resources around the country well before November 2, when Selzer’s poll became public. He didn’t schedule any big rally in Iowa to counteract the poll.

Again and again, Trump’s court filing purports to read Selzer’s mind: “In truth, the Harris Poll, was just a piece of political theater concocted by an individual—Selzer—who, as a supposedly legendary pollster with the power to shape public perception of elections, should have known better than to poison the electorate with a poll that was nothing more than a work of fantasy.” Who was poisoned? And how did that incur “actual damages” to Trump?

SOME IOWA REPUBLICANS ENDORSE TRUMP’S LAWSUIT

Iowa GOP state chair Kaufmann has repeatedly accused Selzer of using her polls to boost Democratic candidates. So naturally, he cheered news of the lawsuit, posting on X/Twitter, “There needs to be accountability.”

U.S. Representative Mariannette Miller-Meeks, a Republican who just won a third term in Iowa’s first Congressional district, wrongly claimed in a December 17 post, “Nov 2nd, Selzer predicted I’d lose by 16 points. In a costly battleground House race, what negative impact did the poll have in a close race? This action by President Trump will uncover how Selzer got President Trump’s & my race so incredibly wrong.”

I have long criticized the Des Moines Register for publishing Congressional level findings from Selzer polls. Those numbers have little value, since the sample sizes are small (producing a large margin of error) and the candidates are not named. But Selzer has never characterized the district-level data as predictions for a Congressional race. The write-up of those findings on November 3 made clear the survey tested a generic ballot, not Miller-Meeks against her Democratic challenger, Christina Bohannan.

Attorney General Brenna Bird, who has frequently decried what she calls “lawfare” waged against Trump, has not posted about this case on her social media feeds. A spokesperson for the Iowa Attorney General’s office did not respond to Bleeding Heartland’s questions. Among other things, I wondered whether Bird thinks publishing an erroneous poll constitutes consumer fraud, and whether she believes Selzer and the Des Moines Register deliberately manipulated polling data to help Democratic candidates, either this year or during the 2022 attorney general race.

Ed Tibbetts commented in his latest Substack column, “I believe the saddest thing about this suit is that top Republicans in Iowa know in their hearts that Trump’s claim is not true, yet out of fear or dishonor they remain silent.”

Another sad reality: even if Trump loses in court, he may win by discouraging other media from publishing unflattering news coverage or poll numbers. Yet it would be far worse for the defendants to fold. Doing so would validate what Josh Marshall has called “a performance art of power enabled by a shameless abuse of the legal system.”

Whatever your views on the Register’s editorial policy or Gannett’s management of its newsrooms, it’s important for both companies to fight this case alongside Selzer and not agree to a settlement as the cost of doing business.