The following story was written by The Utah Investigative Journalism Project in partnership with The Salt Lake Tribune.
For someone being cold-called by a stranger for a “scientific survey,” the man was glad to talk about the upcoming election.
He thoughtfully weighed his responses about his pick for governor in the four-way Utah Republican primary on June 30. After about half a dozen typical questions, the survey reader swallowed and grimaced, then plunged ahead with a different kind of query. Sticking to his provided script, he asked the man how he felt about Lt. Gov. Spencer Cox having received donations from a radical left group that supported open borders, gun control, “abortions on demand” and sex education in elementary school.
“Ninety-seven percent of the group’s donations go to Democrats and socialists like Bernie Sanders and Elizabeth Warren and only 3% go to the few radical liberal Republicans who will support their agenda like Spencer Cox,” the surveyor continued. “After hearing this message, if the Republican primary election for governor were held today for whom would you vote …”
“I was unaware this was an anti-Spencer Cox advertisement, so, thank you, I’m through with this poll,” the man said simply and hung up. The poll reader scarcely had time to mouth “whoa” before the dialing software was automatically connecting him to the next Utah voter.
I know exactly what the survey reader felt, sitting there in his basement, awash in embarrassment while dialing random Utahns in his sweatpants and a dirty shirt as he tried to keep his cats off his keyboard. I know because it was me. Since May, I had been working as an independent contractor for Utah-based G1 Research.
I didn’t tell the respondents who was paying for the Utah primary poll — that wasn’t part of the script, and in fact I didn’t know. The company never told us who the clients were.
Piecing it together, though, it was pretty clear that the poll was being done by the campaign of former state House speaker and then-gubernatorial candidate Greg Hughes. What made this activity even more questionable than it was on its face was that Hughes had been key in enacting a Utah law aimed at dragging such secretive tactics — which he denounced as sleazy — into the light of day by requiring pollsters to disclose the identity of their clients.
The Hughes campaign on Wednesday confirmed it was responsible for the survey but denied it was a “push poll,” saying it was a legitimate message-testing survey that informed its strategy in the final days leading up to the primary.
The campaign and G1 Research denied knowingly violating the disclosure law, blaming each other for the oversight.
Push me, poll you
Over the summer, I dialed surveys in support of Republican candidates across the nation, usually those most closely aligned with President Donald Trump.
In Kansas, it became clear, we were working on behalf of, if not for, U.S. Senate candidate and former Kansas Secretary of State Kris Kobach. Kobach, once under consideration as Trump’s “immigration czar,” had made his name working with immigration hard-liner Joe Arpaio. Trump had endorsed Kobach in his unsuccessful 2018 run for Kansas governor but not in this year’s Kansas Senate primary.
The script for this poll essentially accused rival Senate candidate Roger Marshall, a GOP member of Congress, of helping fund the Wuhan Virology lab in China, telling respondents that was “where many experts agree the COVID-19 coronavirus originated.”
The surveys hopscotched the country. We dialed to test messages against expanding Medicaid in Oklahoma and legalizing medical marijuana in Nebraska.
In Georgia’s 7th Congressional District, where Democrat Carolyn Bourdeaux lost in 2018 by only 419 votes, we asked voters how they felt knowing that Bourdeaux was supported by a group that loans money to the Chinese government, “even though the money could be spent on Chinese spying operations and the stealing of intellectual property and technology from the United States.”
Political polling experts are quick to point out that it’s not illegal or uncommon for campaigns to conduct “message testing” where they gauge voter reaction to different possible lines of attack — even down and dirty ones. They concede that biased surveys can be unethical but that doesn’t make them “push polls” — a term referring to surveys that are meant to sway opinions rather than gather them.
What is clear, however, is that these surveys are in violation of a Utah law requiring pollsters to disclose to the survey respondent who’s paying for the poll. In its training, G1 Research repeatedly tells contractors to tell survey respondents that we don’t know that information and that even if we did know, releasing it would bias the survey.
Utah’s law was enacted in 2013 amid corruption investigations into former Attorneys General Mark Shurtleff and John Swallow, when it was revealed that Swallow campaign operative Jason Powers conducted push polls on behalf of the payday loan industry.
The law hits close to home for G1 Research because its founder is Greg Powers, a brother to Jason.
‘Do NOT ask for permission’
When I decided to try out the G1 Research position in May, I figured it would either be an interesting reporting project or a way to make some extra cash. I decided I couldn’t go into it misrepresenting myself. I would not lie if asked about my background or omit anything from my resume or, if requested, my complete work history.
As an independent contractor rather than an employee, however, the company didn’t ask for a complete work history or resume, and the interview itself was done virtually. I simply filmed myself answering questions about relevant work experience and submitted them through a website. I included what I deemed the most relevant job history, including past call center work and as a contractor investigating, primarily over the phone, workplace harassment complaints for a government agency. I didn’t list my credentials as an investigative journalist because I didn’t see it as particularly relevant to the position. Upon being accepted as a G1 Research contractor, there was no nondisclosure agreement to sign or any written prohibition about use of information that would come to me through my work with the company.
I later decided that there was a news story here, and The Salt Lake Tribune expressed interest in publishing it, depending on what my reporting turned up. The story then became more important to me than the part-time paycheck and, after discussing it with The Utah Investigative Journalism Project board, we have decided to donate my G1 Research earnings to a deserving nonprofit.
A lot has changed in polling since the early 2000s, when I did call center work for a Spanish Fork company while I was in high school. Voice over Internet phone technology has become so cheap and efficient, survey companies don’t need to rent out a calling floor. Instead, they can recruit dozens of home-based contractors who just need internet and a headset to start dialing. (It’s also the reason robocalls about your car’s extended warranty bombard you nowadays.)
The job training consisted of a number of YouTube videos posted to a Slack channel that helped explain how to set up the software, log in to the system and conduct surveys.
“You will GREATLY help your response rate if you just jump into the 1st question right after the intro,” the post says. “Our best callers do NOT ask for permission to do the survey — they just start into it.”
My first survey was on the Senate Republican primary in Kansas, and the pace was frantic.
I was soon asking voters how they felt about Marshall, the U.S. House member and physician running against Kobach, having allegedly shut down a Catholic hospital because “its Catholic rules limited the types of procedures he could perform on women.” I’m saying “allegedly” here as a reporter after the fact. It was not part of the script.
We read statements about funding from Marshall that helped spread COVID-19, about him joining House Speaker Nancy Pelosi, D-Calif., in seeking Trump’s tax returns and about him opposing spending cuts.
“Marshall wouldn’t even cut spending on things like the National Endowment for the Arts, which spent tax dollars funding alleged art like a crucifix in a jar of urine titled ‘Piss Christ’ and a play celebrating transgender people changing genders.”
Not all surveys we did were negative, some were generic, but the negative ones were shocking even to us script readers.
G1 Research′s work-from-home contractors have a Slack channel that takes the place of an office watercooler for chatting and gossip, and there was plenty of venting about the Marshall survey.
“We made him out to be the Republican anti-Christ!” one remote colleague posted.
“'Piss Christ' never rolls off the tongue,” commented another.
Howard Fienberg is the vice president of the Insights Association, a Washington, D.C.-based group advocating for the market research and data analytics industry. He says that people often mistake “push polling” with “message testing.”
There are thousands of campaign-consulting firms out there that conduct surveys to test a message — some negative, some not — to see how effective it might be. Then a candidate can decide if he or she wants to incorporate it into a wider campaign.
He says “push polling” is virtually extinct in large statewide and federal races because providing negative surveys to a few hundred potential voters wouldn’t make an impact.
“The idea behind a push poll," Fienberg says, “is to hit as many people as possible.”
It’s a point echoed by Kevin Banda, a professor of public policy at Texas Tech University who studies public opinion and elections.
“Industrywide [push polling] has died out,” he says. “But maybe there are smaller shops that are willing to do it — if they are paid well enough.”
At G1 Research, we have dialed on statewide and congressional races, but we’ve also done smaller races, including a Jefferson Parish race in Louisiana, where our survey included statements about a candidate for a judge’s seat having gambled “away over a million dollars of his elderly parents' retirement savings and money.”
Jae-Hee Jung is a postdoctoral fellow at the Blavatnik School of Government at Oxford and has written on and researched negative campaigning. She notes that the most current research shows that it’s unclear if negative campaigning actually works despite its embrace by U.S. political consultants.
“The bottom line is it’s a very attractive strategy from the perspective of campaign consultants," she says, “and is more likely to be used when you, the candidate, are desperate.”
While negative “message testing” is considered legitimate, the American Association for Public Opinion Research’s code of ethics states that surveyors should avoid misleading participants, and not “conduct other activities (such as sales, fundraising, or political campaigning) under the guise of conducting research.”
It also says survey takers should abide by all laws and regulations.
Greg Powers is the founder of G1 Consulting and G2 Consulting, which do business under the name G1 Research. He says that while his company is not a member of the industry association, its surveys match what that organization considers legitimate, and that he does not engage in “push polling.”
He points out how his surveys target scientifically random samples of between 400 and 1,500 voters compared to push polls that dial thousands of respondents with negative messages.
“If your goal is just to deliver a message," Powers says, “then you have as few questions as possible and often just start with the nasty message.”
In the week leading up to Utah’s June 30 Republican primary, I signed up for a shift dialing on a survey titled “UT Governors.” The poll was fairly generic in asking about primary candidates. But then it veered to a more pointed question about Lt. Gov. Spencer Cox, one of the two front-runners to replace outgoing Gov. Gary Herbert. It focused on Cox’s receipt of a $75,000 donation from the National Education Association, which the poll script described as “national education union bosses.”
A couple of days later, I signed up for a new shift and this time the survey was titled “Hughes Brushfire.” In scanning the list of surveys that had been done in the recent past, I came across numerous jobs with “Hughes” in the title.
Hughes Brushfire was the survey we dialed in the waning days of the primary when we read a statement implying Cox’s support of abortions, open borders and other causes anathema to most conservative Utahns.
While I was never given the identity of the client, a review of campaign records showed the Greg Hughes for Governor campaign paying $88,576 to G1 and G2 Consulting.
It fit with what the experts say about push polling and/or negative message testing — that, generally, it comes from candidates who are behind and likely to resort to such tactics out of desperation. Hughes, at that time, was polling a distant third in the race, with about half the support of leading contenders Cox and former Gov. Jon Huntsman.
Still, it’s an ironic twist that the negative survey was conducted on behalf of Hughes, when the former lawmaker was the sponsor of HB44, the 2013 legislation requiring pollsters to disclose the client to those surveyed.
Hughes, in persuading colleagues to support the measure, said he personally had been the victim of such secretive negative polling in an election and called it part of “the uglier side of campaigns.”
HB44, he added, was “an important transparency bill” and good public policy.
Greg Hartley, former campaign manager for Hughes, who ended up finishing third in the primary with about 21% of the vote, readily acknowledged that the campaign ran the surveys in question. The results were useful, he says, and informed the campaign’s decision to send mailers, texts and run a radio ad on the teachers' union donation.
“Some folks thought we were attacking,” Hartley says. “But we were contrasting. And it was a serious issue based on research.” He says the campaign disclosed those ads, unlike attacks from mysterious third parties that also appeared in the campaign.
When I asked about G1 Research violating the law by not disclosing the client’s identity, Hartley responded he was unaware that was the case.
“They should be following the law,” Hartley says. “As far as we know, they were disclosing who was paying for the polls.”
That was not true, however, based on my own experience conducting the polls.
Powers says he, too, was surprised that the disclosure was not made, adding that was the responsibility of the client.
“We’re not writing the surveys," Powers says, “we’re just conducting them.” He says Utah clients should know to put in the disclosure.
“Now if the client forgets, I would hope that we would catch it,” he says, adding that one of his managers must have just missed it.
Powers also says it is standard for the company to not disclose who pays for surveys of voters outside of Utah, arguing that the 2013 law applies only to in-state surveys.
The language of the law, however, does not make that distinction, stating only that: “A person who conducts a poll shall disclose to the person being surveyed who paid for the poll before or at the conclusion of the poll.” A violator is subject to a $100 fine.
I informed Rep. Brad Daw, R-Orem, about the polls I had been conducting for the company of Greg Powers, brother to Jason Powers, who had orchestrated what Daw described as the shadowy push polls against him in 2012 on behalf of the payday loan industry.
“It’s kind of sickening,” Daw said in response. “But it’s not that surprising.”
Daw recalls as brutal the attacks against him by payday lenders for his support of regulations on the industry, which last year in Utah charged an average annual interest rate of 523%. His conservative Utah County constituents were bombarded by “polls” and coordinated anonymous political mailers falsely asserting Daw was a champion of illegal immigration and “Obamacare.”
The tactics seemed to work, and Daw was defeated in that election, although he bounced back two years later to reclaim his legislative seat.
When the House took up Hughes' poll transparency bill in 2013, Daw sat shoulder to shoulder with Hughes at the committee witness table to press for “this very needed ray of sunshine.”
“The whole notion behind the disclosure is simple,” Daw says now. “Say whatever you want to say — but put your name on it.”