Recourse in Polling
Evolving from small groups gathering on a village greens to heavily scrutinized voting models, the American polling process now more resembles market research than individual political expressions. The American public is bisected in dozens of ways to isolate potential voter preferences in a manner as if they were choosing a new product rather than a representative. Equating personal political preferences however, may not be fully synonymous with launching new merchandise. As the 2016 Presidential Election has shown, viewing the American voting public as “political consumers” may not paint the complete picture. What then, is the recourse for improperly or possibly recklessly conducted polling? Is this a matter of the general public or the poll’s sponsor alone?
The critical determination is that polling boils down to sampling assumptions of what an electorate may “look like.”3 Conventional wisdom reduces this to certain targeted demographics, college educated vs. non-college educated, male vs. female, and urban vs. rural, to name a few.4 One can effectively shape a poll to be heavily Democratic by oversampling from traditionally liberal urban centers while inversely constructing a heavily Republican poll by oversampling traditionally conservative rural areas.5 On first glance, there may seem to be no deficiency in polling conclusions, however, the truth may be buried in the details.
In 2016, state-level polling was materially off from the final marks.6 Democratic candidate Hillary Clinton was shown to be well ahead in several states which eventually went for the then Republican candidate, Donald Trump.7 Polling agencies took a long hard look at exactly how they got it so wrong.8 Hundreds of hours and thousands of spreadsheets later, most centered on polls not discerning enough between college and non-college educated voters during their sampling.9 As they believed, they focused too much on racial composition rather than educational factors.10 However, this may not be the entire story.
2016 Presidential Election Results:
-Wisconsin Polling Average: Clinton +6.5. Actual: Trump +.7.11
-Michigan Polling Average: Clinton +3.4. Actual: Trump +.3.12
-Minnesota Polling Average: Clinton +6.2. Actual: Clinton +1.5.13
Polling agencies are often hired by institutions to present results that are amenable to the purchaser, frequently to generate potential market research contracts.14. The CNN/SSRS poll data shows the Democratic candidates up roughly 12 points compared to the Republican candidates.15 Alternatively, Rasmussen Reports, with a likely right of center audience, has the Democratic candidate up only 3 points on the Republican.16 Rasmussen is unique in that it does not subcontract its polling, making its figure likely susceptible to internal selection biases. 17
In a possible repeat of the 2016 presidential election results, would a poll sponsor potentially have a legal recourse against their agent if they were wide of the mark? The short answer likely is “no.” While there is limited case law in the area, any claim would likely have to be presented under an in-the-market negligence theory.
Under a theory of in-the-market negligence, any plaintiff would also find it incredibly difficult to prevail. In an area with highly subjective and often fluid metrics, it would be nearly impossible to prove a negligence claim where reasonable minds may widely differ.18 Further, the damage any sponsor suffers is likely too attenuated to their pollster’s actions.19. A media brand likely has the departmental capabilities to do extensive vetting before selecting their pollster and also is not under obligation to publish material. Absent fabrication by the polling agency, the sponsor will likely not have a cause of action.
The old saying, “the only poll that counts is election day” may ring especially true this year.20 The commoditization of the American voter may once again prove to be incorrectly ascertained. The enormous weight polling has on the American psyche likely will not change once all the votes are counted in 2020. However, the nature of the industry and the means of sampling may come under more scrutiny than ever.
See, e.g., Bill Snyder, How Polls Influence Behavior, Stan. Graduate Sch. of Bus. (Oct. 20, 2012), https://www.gsb.stanford.edu/insights/how-polls-influence-behavior [https://perma.cc/3CZU-82B2].↩
See Stephen Mills, Public Polling’s Reputation Damaged by US Failure, RZN (Nov. 16, 2016, 12:09 PM), https://www.rnz.co.nz/news/on-the-inside/318181/political-polling’s-reputation-damaged-by-us-failure[https://perma.cc/XCN6-XPMC] (last visited Oct. 30, 2020).↩
See John Gramlich, What the 2020 Electorate Looks Like by Party, Race and Ethnicity, Age, Education and Religion, Pew Research Center: FactTank (Oct. 26, 2020), https://www.pewresearch.org/fact-tank/2020/10/26/what-the-2020-electorate-looks-like-by-party-race-and-ethnicity-age-education-and-religion/[https://perma.cc/7WJD-ZXJ5].↩
See Sara Savat, The Divide Between Us: Urban-Rural Political Differences Rooted in Deography, Wash. Univ. In St. Louis: The Source, https://source.wustl.edu/2020/02/the-divide-between-us-urban-rural-political-differences-rooted-in-geography/[ https://perma.cc/H52Q-LCGS].↩
See Andrew Mercer, Claudia Deane, and Kyley McGeeney, Why 2016 Election Polls Missed Their Mark, Pew Research Center: FactTank (Nov. 9, 2016), https://www.pewresearch.org/fact-tank/2016/11/09/why-2016-election-polls-missed-their-mark/[https://perma.cc/N4U6-2NCN].↩
Geoffrey Skelley and Nathaniel Rakich, What Pollsters Have Changed Since 2016 — And What Still Worries Them About 2020, FiveThirtyEight (Oct. 13, 2020, 6:00 AM), https://fivethirtyeight.com/features/what-pollsters-have-changed-since-2016-and-what-still-worries-them-about-2020/[https://perma.cc/8BTB-6P2A].↩
Nate Silver, The Real Story of 2016, FiveThirtyEight (Jan. 19, 2017), https://fivethirtyeight.com/features/the-real-story-of-2016/ [https://perma.cc/79UX-4D84] (last visited Oct. 30, 2020).↩
2016 United States Presidential Election, Wikipedia, https://en.wikipedia.org/wiki/2016_United_States_presidential_election#cite_note-524 (last visited Oct. 30, 2020).↩
See Eric Chemi, Most Election Pollsters Aren’t Really in it for the Money, CNBC: The big Crunch with Eric Chemi (Oct 16, 2016, 1:07 PM), https://www.cnbc.com/2016/10/18/most-election-pollsters-are-not-really-in-it-for-the-money.html[https://perma.cc/S3T9-XDY4].[/footnote] Whether this is express or implied, the nature of the relationship likely creates pressure to produce results that correspond with the sponsor’s preference. For example, CNN, with a likely left of center audience, contracts with SSRS Research to conduct their polls.[footnote]See Amy Mitchell et. al., Section 1: Media Sources: Distinct Favorites Emerge on the Left and Right, Pew Research Center: Journalism and Media (Oct. 21, 2014), https://www.journalism.org/2014/10/21/section-1-media-sources-distinct-favorites-emerge-on-the-left-and-right/[ https://perma.cc/L688-PME4](discussing political lean of CNN); see also CNN/SSRS Poll Oct. 23-26 2020, CNN, http://cdn.cnn.com/cnn/2020/images/10/28/rel15.pdf (last visited Oct. 30. 2020).↩
Id. at 2.↩
See Rasmussen Reports Poll Oct. 27-29, Rasmussen Reports, https://www.rasmussenreports.com/public_content/politics/elections/election_2020/white_house_watch_oct30 (last visited Oct. 30. 2020).↩
See generally Id.↩
See Rosenblatt v. Baer, 383 U.S. 75, 84 (1966) (finding media publications are held to a reckless disregard for the truth in order to open themselves up to liability).↩
See Time, Inc. v. Hill, 385 U.S. 374, 395 (1967) (noting failure to make reasonable investigations is a component of a negligent misstatement).↩
See The Only Poll that Counts Is Election Day, Reimagine America, https://reimagineamerica.org/poll-counts-election-day/[https://perma.cc/47V6-WC6J] (last visited Oct. 30, 2020).↩