When Public Policy Polling came to Portland, Maine

The inside story of how polling was weaponized in a small-town city council election.

This is the last thing I will write about the 2017 election.

Before memories began to fade, I wanted to record a complete account of how polling was used, abused and possibly manipulated during the campaign. This was the first time that polling was used as a political weapon in this way in a Portland city council race, and I hope that this does not become a permanent part of our local politics.

May this account serve as a cautionary tale for small-town candidates and politicians everywhere.

Polling can be a very useful tool. By surveying likely voters, campaigns can learn about their weaknesses and blind-spots and adjust their strategies accordingly.

But polling can also be used as a weapon against other campaigns. By carefully timing a poll, cleverly designing it, and selectively releasing its results, a campaign can create a “narrative” that works in their favor and against their opponents: favorable stories in newspapers, useful blurbs on campaign materials, and positive social media posts.

In a small-town political environment like Portland, where campaigns rarely have the funds to conduct their own polling, manipulating and weaponizing a poll can be an extremely effective tactic.

In the 2017 campaign for city council, polling from Public Policy Polling — a well-known, national organization — was weaponized in this way.

In January of 2017, I announced my candidacy for an at large seat on the Portland City Council. Councilor Jill Duson, the five-term incumbent, announced her candidacy shortly thereafter.

Months later — in June — Bree LaCasse announced that she would be joining the race. LaCasse and I were both to the left of the incumbent politically, and some were concerned that we would split the “left” vote, letting Councilor Duson easily win re-election. But neither LaCasse nor I would be deterred, and we both conducted intense campaigns during the summer.

On August 25, about ten weeks before election day, I received an email from Steven Biel with the subject line, “Polling — please respond.”

Biel is the founder and director of Progressive Portland, a left-leaning, Moveon.org-type political action group. He was also a self-described “super-volunteer” for LaCasse. Biel had spent months canvasing for LaCasse’s campaign and was known to be in her inner circle of advisors.

Biel’s email read, “We’re developing our September poll and need to decide some questions that we need to discuss with you. Please give me a call to discuss.”

Based on previous conversations I knew that Progressive Portland would eventually poll our race, but I did not expect to be consulted about it beforehand.

(By the way, it is common practice by political operatives to prefer phone conversations to email, so that there is no paper trail of what was discussed.)

Instead of calling Biel, I chose to contact an intermediary: another member of the Progressive Portland steering committee who was supporting my candidacy. Through this person, I learned that Progressive Portland intended to hire Public Policy Polling (PPP) to conduct a poll.

In addition to questions about local issues and the Mayor’s job performance, Progressive Portland wanted to ask three questions about the at large city council race:

  1. Who would you vote for in a three-way matchup with Jill Duson, Joey Brunelle, and Bree LaCasse?
  2. Who would you vote for in a two-way matchup with Jill Duson and Bree LaCasse?
  3. Who would you vote for in a two-way matchup with Jill Duson and Joey Brunelle?

I was told that the two-way matchup questions were designed to determine whether LaCasse or I was better positioned to challenge incumbent Councilor Duson.

The intermediary then relayed that Biel proposed that I join LaCasse in a pledge to drop out and support the one of us that scored better against Duson in the Progressive Portland poll.

I had no intention of joining this pact, but I wanted to speak to Bree LaCasse to see if her campaign was sanctioning this proposal. So I gave her a call.

Bree was indeed aware of the proposal. She said that Biel had informed her of the poll and the pact the night before. “I would be open if you’re open,” she said.

I declined to participate in the pact, and hoped that this would all just go away. And it did — for a about a week.

One of the biggest issues in the 2017 election was whether to pass a bond question that would repair four local elementary schools. Both LaCasse and I supported the bond, and we were invited to be on the steering committee of Protect Our Neighborhood Schools (PONS) — a political organization formed to support it.

On September 6 all members of the PONS steering committee (as well as Portland Mayor Ethan Strimling) received an email from Emily Figdor, the organization’s director. The email read,

PONS Advisory Committee,
I have some good news: The Maine Education Association is interested in helping with our campaign if polling results show that they can make a difference. In that case, they’d essentially pay for (and possibly help run) our digital campaign, which is about $9k. They’re also going to pay for the poll, which wasn’t in our budget.
We included the three questions below in a larger poll that Progressive Portland has been working on. Public Policy Polling is the vendor running the poll, and they’re getting started right away. I think we may have results by the weekend.

Figdor is married to Steven Biel, president of Progressive Portland. Figdor had also publicly endorsed Bree LaCasse in June, and together Biel and Figdor had given $1,550 in contributions to the LaCasse campaign.

That same day, I began to receive reports that the poll was “in the field” — i.e. Portland residents were receiving automated phone calls from PPP. Portland Press Herald reporter Randy Billings soon heard the same thing.

Respondents were asked about a wide array of local issues — the school bond, other ballot questions, and whether an issue or candidate endorsed by Mayor Strimling would make them more or less likely to vote for that issue or candidate.

They also asked about the at large city council race, but only two questions (paraphrased below):

  1. Who would you vote for: Brunelle, Duson or LaCasse?
  2. Are you fully committed to your vote or a is there a chance that you would change?

My contact at Progressive Portland told me that since neither LaCasse or I had agreed to the pact, that there was no point in asking the two-way matchup questions as they had originally planned.

On September 20 Press Herald reporter Randy Billings tweeted, “[Progressive Portland] tells me they won’t be releasing the results of their recent poll.”

And indeed, they never publicly released the full results of their poll. Why not? It’s impossible to say for certain.

Biel wanted to release the results of at least some survey questions. In a September 11 email to PPP’s Anniken Williams, Biel gave the pollster instructions for their official press release:

“We will release questions 1–10 and want you to emphasize the at large race, bond, and rent stabilization. Our take on the at large race is that Jill Duson, the incumbent, is incredibly weak with only 13.5% of the electorate strongly for her. Jill Duson is in SERIOUS danger.”

In draft versions that were sent to Biel and forwarded to Mayor Strimling, the poll showed LaCasse and myself in a virtual tie for second: LaCasse with 18% and myself with 16%, with a nearly 5% margin of error. Jill Duson led the pack with 27%.

Nevertheless the subject line of PPP’s draft press release read, “New Poll Could Mean Trouble for Incumbent Jill Duson.”

But these drafts never saw the light of day. In the end, Progressive Portland released none of the poll’s data.

A member of Progressive Portland’s steering committee told me at the time that they decided to not release the poll because they did not want to influence the city council race.

I was relieved to hear that. My fear all along was that, since it was developed at least partially by allies of my opponent, that it could be manipulated. If the poll showed LaCasse with better numbers, it could be deployed as a weapon against my campaign.

(In the end, my concerns would be justified when a second poll was secretly conducted in October, under more suspicious circumstances.)

The results may not have been released publicly, but my contact at Progressive Portland later offered to give me strategy advice based on the results. He legally could not provide any numbers, since that would have to be reported as a campaign contribution. But, he said, he could provide his subjective advice based on the results — for example, which demographics he thought I should target more, or which neighborhoods he thought I should send mail to.

He also indicated that he had no doubt that the LaCasse campaign would be receiving similar information — if not more — via Biel.

I chose to accept this offer, in part because I did not want the LaCasse campaign to have this unfair advantage.

In our subsequent conversations, my contact revealed that not only had the questions been developed by Progressive Portland, but that they had designed the “weighting” of the poll as well.

What’s weighting? It’s the subjective judgement by the pollster of who will likely vote in the election, and the filtering of poll responses based on those subjective judgements. So if the pollster expects more women to vote in the upcoming election, or more young people, or more people of color, they will include more of those responses in the final poll results.

Weighting is one of the most critical steps in conducting an accurate poll, and to leave this to the people paying for the poll increases the potential for inaccuracy or deliberate manipulation.

Typically pollsters like PPP design their own weighting, but in this case Progressive Portland had been allowed to weight the poll as they saw fit. Apparently there had been internal debate at Progressive Portland about how to weight the results, including a discussion about whether the chosen weighting was likely to favor LaCasse or myself.

I was assured that the final weighting was not chosen to favor any candidate.

Weeks passed, and Progressive Portland continued to sit on the poll results.

Then, on October 2 Protect Our Neighborhood Schools held a press conference in which they announced the results of a “new poll” that showed widespread support for the school bond.

In reality, this “new poll” had been conducted a month before, from September 6–10. This was the Progressive Portland poll.

But you wouldn’t have known that from the official statements from PONS. In their press conference and press release, as well as the reporting in the Portland Press Herald, the poll was described as “commissioned by the Maine Education Association”.

The MEA paid for the poll — that is true. But it was designed and weighted by Progressive Portland. And it had included questions about other issues, the existence of which was never disclosed by Protect Our Neighborhood Schools. Nor was the complete methodology of the poll ever made public.

Without the full text of the questions or the methodology, it is difficult to determine the poll’s validity.

The National Council on Public Polls, an association of polling organizations that sets standards for the practice, has defined what it calls “Principles of Disclosure” for its members. It declares that all reports of survey findings must include:

  • Who sponsored the survey
  • Who conducted the survey and when
  • Sampling method employed (for example, randomly dialed telephone sample or list-based telephone sample, and use of any oversampling)
  • Population that was sampled (for example, general population, registered voters or likely voters)
  • Size of the sample
  • Margin of error
  • Survey mode (for example, automated telephone or online)
  • Complete wording and ordering of questions mentioned in or upon which the release is based
  • Percentage results of all questions reported

Public Policy Polling is not listed as a member of the National Council on Public Polls on their website.

Neither PONS nor Progressive Portland has ever publicly disclosed the sampling method it employed, nor the complete wording of any of the questions asked.

With just a few weeks left until election day, many of us assumed that this season’s polling drama was over. But that was not to be.

On the afternoon of Sunday October 29, Emily Figdor and Protect Our Neighborhood Schools held a press conference to announce the results of a another poll. This one was truly “new.”

Exactly like the first, this second poll had been been conducted by PPP and paid for by the MEA. It had surveyed Portland residents between October 24–26, and had — like the first poll — asked about a number of local issues, including the at large city council race. Once again, PONS never disclosed the complete wording of any of the questions, nor the complete methodology.

Did Progressive Portland design the questions and the weighting, like they had with the first poll? Probably not. By this point Progressive Portland was in disarray: Steven Biel had just temporarily resigned after sharing part of the organization’s email list with the LaCasse campaign without authorization, and other members of the steering committee had resigned or temporarily stepped aside over the matter.

But given Biel’s connection to Figdor, PONS and the LaCasse campaign, and given that he had been the primary point-of-contact with the pollster, it is safe to assume that Biel and/or Figdor devised the questions and the weighting on their own.

No longer constrained by a steering committee, they were free to write the questions and weight the poll however they wished, and release whatever results they wanted.

They released more results this time — not only about the school bond issue, but about two other ballot questions (regarding rent stabilization and zoning) and the at large city council race. They did so with a single-page press release on PPP letterhead, much like the drafts exchanged in September. They contained no information regarding the wording of the questions, and incomplete information about the methodology employed.

Randy Billings at the Press Herald expressed his surprise on Twitter in a series of tweets: “Maine Education Association is polling citywide issues (i.e. council races?) in #portlandme? Releasing on a Sunday?”

And what did this poll show? Regarding the at large race, PPP wrote in their press release:

The Portland City Council race is still tight between incumbent Jill Duson and Bree LaCasse, with no one holding a sizable lead. Duson has 30% support, but is trailed closely by LaCasse, with 25%. Joseph Brunelle sits in third place with 13%. 32% of respondents did not know who they would vote for.

Within hours of PONS releasing these results, the LaCasse campaign and its allies picked it up on social media to promote the idea that LaCasse was the only viable challenger to the incumbent.

One day later, The Maine State Building and Trades Council shared the PONS announcement on their Facebook page, adding, “In the Portland At-Large contest, it has clearly become a two-person race.” The Building Trades Council had endorsed LaCasse a few weeks before. (It should be noted that the Executive Director of the Building Trades Council is Jason Shedlock, the former assistant to Mayor Ethan Strimling.)

The following morning, LaCasse ally Mayor Strimling went on local talk radio [10:35] to discuss the new poll.

“There’s some polling I know you wanted to briefly talk about,” host Matt Gagnon said to Strimling. Strimling mentioned the poll’s results about the school bond issue, then added,

“And one little interesting tidbit is the at large race — the at large race for city council. It seems to be wide open. There’s about 30% of the people who are undecided. Jill Duson, who is the incumbent, Bree LaCasse is challenging her. Bree is about five points behind in that, with about 30% undecided. Joey Brunelle is also in that race — he seems to have faded a bit, into the background.”

Then, on the afternoon of November 6 — the day before the election — the LaCasse campaign deployed a citywide, automated “robocall” with a pre-recorded message from LaCasse supporter, state senator and gubernatorial candidate Mark Dion:

“This is state senator and former Cumberland County Sheriff Mark Dion. A new poll shows Bree LaCasse within a few points of unseating 16-year incumbent Jill Duson, with Joey Brunelle too far behind at just 13%.”

In the days leading up to the election, this new poll — paid for by the MEA, designed by Figdor and/or Biel, and released by PONS — was used by the LaCasse campaign and its allies in an attempt to build the narrative that she was the only viable challenger to Duson.

But was she?

It turned out that the late October poll had been quite off the mark.

In the end, Jill Duson held onto her lead and received 44% of votes cast. However, my campaign won second place with 30% of the vote, and the LaCasse campaign was third with 26%.

If the late October poll is to be believed, LaCasse’s campaign picked up effectively none of the outstanding undecided voters, while my total vote share more than doubled — surging 17 points and leaping into second place. Councilor Duson’s vote share increased by 14 points.

Yes, there were a few negative news stories about the LaCasse campaign between the poll and election day, including one particularly explosive op-ed in the local newspaper two days prior to the election. Even so, it seems improbable that this bad press could have precipitated a 17:1 difference in the change in our support.

Here’s another way to think about it: if you assume that “undecided” voters are not that different from “decided” voters, you would expect in the last two weeks of a campaign that the “undecideds” will eventually vote in roughly similar proportions to how the “decideds” responded in the poll — or at least in proportions that are not wildly different.

But in this case, they were wildly different. The relative proportions radically changed between the October 24–26 poll and the November 7 election — almost completely inverting LaCasse and myself. Curiously, Duson’s proportion of the vote hardly changed at all:

Neither I nor my campaign’s ground team detected a last-minute surge of this magnitude. We find it difficult to believe that we had been that far behind to begin with.

Since PONS never released the questions or the methodology of the late October poll, we are forced to speculate that it may have been deliberately designed and weighted by Biel, Figdor and potentially other allies of the LaCasse campaign to produce results that they then used in a coordinated effort to mislead about the state of the race.

On their website, PPP says, “Our team will work with you every step of the way to help you confidently leverage public opinion for your cause.”

“Leverage public opinion” can mean a few different things, I guess.

It could mean measuring public opinion to improve your tactics and strategies.

It could also mean using the perception of public opinion to influence actual public opinion.

When local politics are driven by these kinds of tactics instead of substance, we all lose. It benefits only those most skilled at manipulation, and makes it harder for the public to know what is true and what is spin.

Unfortunately, I doubt we’ve seen the last of these tactics in Portland. It’s imperative that candidates, elected officials, and citizens be vigilant and vocal in our opposition to this sort of behavior.