A News Growl Special Report uncovers evidence of bias against third party candidates among some of America’s most respected polling organizations.
By Steve Goodale
June 22, 2018
On November 6th New Yorkers will choose their governor using ballots with at least four options on them. Two of those options, Green Party candidate Howie Hawkins and Libertarian Larry Sharpe, are running unusually strong third party campaigns and have been doing so for months.
Yet reputable polling firms like Siena, Marist, and Quinnipiac routinely restrict the choices of their respondents to just the two major parties – ignoring the disconnect between the question they ask in the polls and the questions voters will ultimately answer in the voting booth.
We wanted to know why, so News Growl contacted all three organizations. Two of them replied, but not with the answers we were expecting.
Hawkins & Sharpe: two unusually strong third party candidates
Much of the media attention for the governor’s race has been focused on Sex & the City star Cynthia Nixon entering the fray, and Governor Andrew Cuomo’s reaction to having a celebrity competitor. Far away from the glare of the cameras, however, the Libertarians and Greens have been quietly mounting what may be one of New York’s best ever election results for third party candidates.
Green Party candidate Hawkins received 4.8% of the vote when he ran for Governor of New York in 2014. He is running again in 2018 (his third attempt) and thanks to increased name recognition, may do even better this time.
Larry Sharpe is true rock star for Libertarians, and is widely considered the strongest candidate the party is fielding in any race across the country this year. He stunned the New York political establishment in January when his reported contributions were over six times that of any declared Republican currently in the race. Charismatic and tireless, the former Marine is proving to be a formidable campaigner.
And yet, major polling orgnanizations repeatedly ignore both Hawkins and Sharpe in every poll they produce. Media tend to report on candidates doing well in the polls, so being consistently frozen out is a huge frustration for both campaigns.
The Sharpe campaign recently collected $5,000 from donors to commission its own poll from Gravis Marketing. Although not a household name, Gravis is a reputable polling firm and is one of 19 organizations tracked by fivethirtyeight.com. The poll found that in a five-way race that included Sharpe and Hawkins (and Cuomo, Nixon, and Republican Marc Molinaro) Sharpe polled 5.5% and Hawkins 4.2%.
The Gravis poll had a 3.8% margin of error, which means the Libertarian and Green votes could be significantly smaller. But the poll also found that among voters who knew who Sharpe was his vote jumped to an impressive 24% (and Hawkins’s share among the same group rose to 7%). To the Sharpe campaign the message is clear: the more voters who know about Larry the better he will do.
But getting the word out is difficult when the media refuse to focus on anyone not doing well in the polls. And the polls, in turn, consistently refuse to include anyone not regularly covered in the media.
Three respected pollsters – Siena, Marist, & Quinnipiac
Between Sharpe and Hawkins, a 10% share of the overall vote is (conservatively) a real possibility. This could have an enormous impact on the final result, and ultimately determine who is sworn in as governor in Albany.
News Growl contacted three widely-respected pollsters who cover the New York gubernatorial race – Marist, Siena, and Quinnipiac – and asked them the same three questions:
1) Why are you leaving Larry Sharpe and Howie Hawkins out of your polls?
2) As a respected polling organization, what would you say to the allegation that by leaving Sharpe and Hawkins out of your polls you are weakening their legitimacy to voters and therefore influencing the outcome of the election?
3) Why do you think polling organizations of similar reputations in other countries (Canada, the UK, Germany) include parties of similar size to the New York Green and Libertarian parties in their polls?
We were not entirely sure what to expect. We were not even sure if we would get any replies. We assumed if we did they would include technical justifications for excluding third party candidates, referred to polling samples, margins of error, and the like.
We could not have been more wrong.
Both Marist and Siena sent responses to our questions, and (as far as we can interpret them) it appears the two groups generally ignore third party candidates just because they want to.
[Regarding Quinnipiac we should point out, all three organizations received our emails, but only Siena and Marist replied. We sent repeated emails to various officials at Quinnipiac over a two week period but never received a direct response. One of our journalists was inadvertently included in an internal email exchange about our enquiry, however, so we can confirm that Quinnipiac did receive, read, and discuss our questions. Three days have passed since that incident, however, and, as of publication, we have had no (intentional) response from Quinnipiac.]
What Siena said:
Replying on the record, Siena College Research Institute spokesperson Steven Greenberg told us Hawkins and Sharpe had been left out of Siena’s recent June 4-7 poll because… well… because they are not Republicans or Democrats.
“We only did head-to-head, Democrat vs. Republican, matchups as the rest of the ballot is still in flux,” he said in an email.
We questioned Greenberg about his use of the word “flux.” In fact, the two ballot lines arguably least in flux are the Libertarian and Green Party lines. Both Sharpe and Hawkins are already official nominees of their parties, unlike Cuomo, Nixon, or Molinaro (all of whom Siena did include).
Thanks to Hawkins’ strong showings in 2010 and 2014 the Green Party has automatic ballot access, so his place on the ballot is rock solid. The Libertarians still need to petition their way onto the ballot, but they have managed this successfully in every New York gubernatorial race since 1990.
Neither the Democratic nor Republican nominees will be formally chosen until the September 13th primary (and considering the discontent among many GOP members with front-runner Marc Molinaro there may still be a surprises in store yet). Cynthia Nixon is struggling to even qualify for the ballot in the Democratic primary, and may or may not be on the Working Families Party ballot line as well.
Describing the ballot status of the candidates not running as Democrats or Republicans as being in “flux” seemed wildly inaccurate and unfair to us. We suggested to Greenberg it might even indicate a serious misunderstanding of ballot access laws, or even evidence of prejudice against third party candidates at Siena. When we asked him to clarify his response in light of our concerns Greenberg declined, however. He said he would stick by the answers provided.
Greenberg did send us a copy of a poll from the 2014 race that included Hawkins, and pointed out that, “not atypical for a minor party candidate in pre-election polling – he polled at nearly twice what he wound up receiving at the ballot box.” But poll results being off by 4-5% is hardly atypical for any candidate from any party. That same Siena poll underestimated the Republican candidate’s ultimate vote share by 7%. If anything, this would indicate a shortcoming of polling generally and not a reason to question the legitimacy of any particular candidate.
Greenberg answered our remaining two questions succinctly and without elaboration. To the question of whether leaving third party candidates out of polls weakens their legitimacy to voters, Greenberg said, “Absolutely not.”
Regarding pollsters in other countries including smaller parties in their polls he said, “That’s their decision to make.”
What Marist said
So far this cycle, Marist has only polled the Democratic primary race and not the general election. Although they have a track record of ignoring third party candidates in the past, it would be inaccurate to say Marist have left Larry Sharpe and Howie Hawkins out of their 2018 polls (yet). Still, they were kind enough to send us a general response to our enquiry without addressing the individual questions separately:
“When The Marist Poll asks a general election question, we include ballot status candidates. In instances when we poll with a media partner, candidate inclusion is determined at the discretion of the media partner.”
If Marist Poll policy is to include candidates with ballot status this means they should be including Hawkins and Sharpe in any future polls of the general election. We asked for confirmation, and we were assured that this was official Marist Poll policy (although they noted that Larry Sharpe would not be included in polls until his petitioning was successfully certified – just being overwhelmingly likely to appear on the ballot is not sufficient).
But there is one major loophole in Marist’s policy. As noted above: “candidate inclusion is determined at the discretion of the media partner.”
In other words, Marist includes all third party candidates with ballot access when their media partners allow them to. In practice, their media partners tend to prefer sticking with just Democrats and Republicans.
For example, in this March 2014 Marist Poll of the New York gubernatorial general electiononly Democrats and Republicans are mentioned. This is despite the fact that in March 2014 the Green Party had guaranteed ballot access for the general election. At the time of this poll, Hawkins had been a declared candidate for two months, and was a shoo-in for the official nomination at the party convention in May.
But Marist had media partners (NBC 4 New York and the Wall Street Journal) and was forced (presumably kicking and screaming) to drop any mention of Hawkins by name from the poll. Respondents were not even given an option for “other” – the only choices were Democrat Andrew Cuomo, several possible Republican candidates (Rob Astorino, Carl Paladino, or Donald Trump – yes, that’s right), or undecided.
This means instead of asking respondents a question that closely resembled the general election ballot, Marist asked a question with an artificially low number of party options. In total, 206,349 New Yorkers cast ballots for candidates representing parties that Marist Poll ignored. That is a seriously large number of voters for a polling organization to simply write off.
We also spotted what at first appeared to be a contradiction in Marist’s policy of only including candidates with confirmed ballot access. In their recent poll of the Democratic primary race, released on April 12th, Marist asked respondents if they would vote for Cuomo, Nixon, or were undecided. Nixon’s place on the Democratic primary ballot is not confirmed, however. A month after this poll was published she actually failed to get enough delegate votes at the state Democratic convention to appear on the ballot automatically and is being forced to petition her way on instead.
If Marist policy is to only include candidates with ballot status, why did they include Nixon in their latest poll? Or Donald Trump in their 2014 general election poll? To us it seemed like a massive contradiction.
Marist provided News Growl with a clarification, saying, “Regarding primaries, this would be a news judgement. A primary tossup may be measured without all petitions being in if there is thought to be news value at that time. There is no contradiction.”
Perhaps no contradiction, but quite literally a double standard. Candidates running for the Democratic or Republican nominations are included in polls if they are considered newsworthy. Third party candidates, no matter how strong the candidate, are only included if their ballot status is official (and then, in practice, they are generally left out at the discretion of Marist’s media partner).
Sharpe, Hawkins, and New York voters ignored
Marist, Siena, and Quinnipiac are not public institutions – they have a right to exclude anyone they want from their polls. But, as is increasingly the case for polling organizations, Marist, Siena, and Quinnipiac are all attached to institutions of higher learning and benefit reputationally from the association with academic rigor. Following this investigation, we have to question if these organizations are as dedicated to academic rigor as many people assume.
This is not because including third party candidates in polls makes the results more accurate (although it might). But presenting a current snapshot of voter intention to the public without allowing respondents the full range of expression they will ultimately have as voters simply seems wrong. It is like playing poker with only 51 cards – the result might not be affected, but the result will seem illegitimate to most players.
Historically, it has not been uncommon for third party candidate votes to amount to little more than a rounding error. But in this instance there is strong evidence to believe the Hawkins and Sharpe campaigns will each have a significant impact on who is elected for Governor of New York.
The influence of non-major-party candidates on the race grew again on Wednesday, when former Syracuse mayor Stephanie Miller announced an independent bid for governor in an increasingly crowded field.
Simply pretending everyone but the Republicans and Democrats does not exist results in a simpler story-line for the media, but smacks of an intellectual dishonesty that is difficult to explain away.
It is hard for us to conclude that the organizations we spoke to are not (at some level) biased against third party candidates. In the case of Siena it appears they are simply prejudiced against third party candidates generally until late in the race, no matter how strong their campaigns. In the case of Marist it appears they let whoever is paying them determine who gets included, even if that means going public with an incomplete set of factors.
And yet there will be many frustrated New Yorkers (and candidates) who will be thankful to Siena and Marist (but not Quinnipiac) for at least being open with us about their decision-making process. We would like to thank them for taking part, and hope a public discussion will someday result in polling that all voters, candidates, and parties consider fair.