American Voter 2016: What the Polls Do – and Don't – Say

*     *     *     *     *

“The principle
[of representative government] … does not require an unqualified complaisance to every sudden breeze of passion, or to every transient impulse which the people may receive from the arts of men, who flatter their prejudices to betray their interests.”
     — Alexander Hamilton, Federalist Paper No. 71 (1788) 1

“If the voter cannot grasp the details of the problems of the day because he has not the time, the interest or the knowledge, he will not have a better opinion because he is asked to express his opinion more often.”
    — Walter Lippmann, The Phantom Public (1925) 2

“This damned pollster approach means taking the pulse of the people and if they get a fever you stand up and say, ‘Fever is great.’ I’m for fever and motherhood and against men who sweat on TV screens.”
    — Eugene Burdick, The 480 (1964) 3

“Leaders must know
what their people are thinking … The classic demagogue swayed the crowd through oratory. Polls sometimes suggest a kind of demagogy in reverse: the crowd seems to sway the politicians through the polls.”
    — "Do Polls Help Democracy?", Time (1968) 4


DOES ANYONE RECALL at what point during the current election cycle the mainstream media shifted its emphasis from reporting on election issues to covering election polling?

Or has the media “crowd sourced” itself into a relentless feedback loop, calculating that tens of millions of viewers, readers and listeners are of the opinion that their opinions will be shaped by their opinion about the opinions of others?

Is there anyone reading this who hasn’t been contacted by a pollster in the last year? Or by someone pretending to be a pollster in an effort to indirectly propagate a favorable/unfavorable view about a product, candidate, civic issue?

No need to answer those questions … this is not a Stealth Poll posing as a common thought piece.

*     *     *

Some analysts believe voters’ mercurial shifts in candidate preference are triggered and sustained by excessive media coverage of transitory polls.

“There’s a general mood that’s created,” commented Georgetown University political science professor Michele Swers in a recent report. “Right now there’s more coverage of these polls regarding the debate than the September jobs numbers.” 5

Polls have come to be considered an indispensable tool for elected officials wishing to discern what their constituents think – and possibly why they think that way – not only in the political realm but in regard to a host of commercial and social topics that may affect voter choices at the ballot box.

And yet the science of extrapolating a likely future occurrence from a miniscule sample of respondents is still, in the public mind, subject to a statistical margin of error plus-or-minus 50/50. Kind of like predicting the weather.

“If you were able to poll voters on what they think about polls they’d think they're crazy,” says Vanderbilt University political science professor Marc Hetherington. “The idea that 1,000-1500 people can make a prediction on what people think in the country is crazy.” 6

*     *     *

The most famous epic fail of polling in U.S. political lore came in 1948 when all major polls predicted a big win for Republican Presidential nominee Thomas Dewey over incumbent Harry Truman.

A Roper Poll executive offered this Zen-like analysis: “I don’t think the polls were wrong in terms of measuring national sentiment. Clearly they were wrong in determining the election. I think the 1948 polls were more accurate than the 1948 election.” 7

More accurate for two reasons – because of who was polled (and who wasn’t) and when the polls were taken. The 1948 polls ceased active polling weeks before the election and relied on lists of voters who were easily accessible via private telephone … which turned out to access more Republicans than Democrats.

Today’s polltakers rely on two basic methods for obtaining what they hope is a reliable cross-section of opinion holders. One is to use lists of registered voters, which plugs into a populace of individuals who (theoretically) have an opinion to offer.

The second method in general use is RDD (Random Digit Dial), which can access a much wider range of respondents – registered and non-registered voters alike including those who have cell phones as well as those with land lines, people who live in houses as well as those who may not, etc.

“You typically end up using a blend of both approaches,” says Prof. Patrick Murray, Director of the Monmouth University Polling Institute in West Long Branch, NJ. “What we’re finding out is that people who aren’t registered voters tend not to answer telephone surveys.” 8

Murray believes polls can provide a close reading of the electoral pulse, depending upon when they are taken. “Polls have an ‘iffy’ history of predicting the eventual outcome a few months away from an election. That’s not what they do. They take a snapshot of exactly what the electorate looks like now.” 9

More importantly, adds Murray, “a good poll should tell you more than just who’s ahead; it will tell you why. It will tell you what’s going on underneath the headlines, what are the dynamics affecting voters. In a presidential race, for example, there are very few truly undecided voters. You want a useful poll that tells you why they are undecided.” 10

Trying to explain the Present Why and how it might lead to the Future Where.

“People follow the smallest movement in polls as if they mean something large,” Murray says. Each poll is a piece of a larger puzzle that is changing over time.” 11

*     *     *

Polling professionals have labored to get a firmer grasp of that puzzle.

In The Rational Public: 50 Years of Trends in Americans’ Policy Preferences, authors Benjamin Page and Robert Shapiro assert that collective polling measurements will tend to accurately reflect true underlying or long-term opinions.

“Parallel opinion change by many individuals at once is especially likely in an advanced industrial society like that of the United States, in which the media of communications are pervasive and highly centralized. Sooner or later, directly or secondhand from friends and acquaintances, most Americans are exposed to the same news – at least the same big news.” 12

Even with the vastly expanded news sources provided by the internet, some scholars see political public opinion as possessing “emergent properties” – a theory deriving from physics that defines an emergent property as a characteristic of a system that results from the interaction of its components, not from their individual properties.

Influenced by seminal works such as Micromotives and Macrobehavior by economist Thomas C. Schelling and Flocks, Herds and Schools: A Distributed Behavioral Model by computer scientist Craig Reynolds, pollsters are seeking to explain their survey results in terms of “rational” behavior models. 13

Or perhaps, less rational models as examined by journalist Bill Bishop in The Big Sort: Why the Clustering of Like-Minded America is Tearing Us Apart, which details Americans’ tendency to self-segregate into increasingly homogeneous communities. 14

*     *     *

And now – polling by tweet!

Researchers at Carnegie-Mellon University believe they have found a way to drill into the abundance of Twitter messages and come up with an accurate gauge of public opinion able to “capture important large-scale trends”. Their initial study examined 1 billion Twitter messages posted over a two-year period and claimed a correlation rate as high as 80%. 15

Note the authors: “The results highlight the potential of text streams as a substitute and supplement for traditional polling. Expensive and time-intensive polling can be supplemented or supplanted with the simple-to-gather text data that is generated from online social networking.” 16

For more specialized, face-to-face polling, the voter focus group is a potent tool. This is where, says Patrick Murray, “you actively recruit people who are like-minded or share a set of characteristics and have them sit around a table and talk to each other. If the group is too diverse in opinions or experiences, in ethnicity or income and so on, people will sense that and hold back from expressing more forthright opinions.” 17

The purpose of the voter focus group isn’t to engender argument or debate, cautions Murray. “You want them to agree or disagree but start a free-flowing conversation rolling. A conversation where words and phrases pop up from their subconscious that tell us what they’re really interested in and what may really be driving their decision-making. You’re looking to discover key words, phrases, images that resonate.” 18

It’s the interaction that can potentially help politicians and policy makers figure out what’s going on in the deeper recesses of their constituents’ minds. “It’s getting people to talk in their own language,” Murray says. “A poll can tell you, ‘This message is more useful than that message.’ But it can’t tell you why. A good voter focus group can tell you what kind of emotions are setting off the bells.” 19

*     *     *

But what about the mainstream media obsession with polls and polls about polls?

According to Patrick Murray, there’s a curious sidebar to that main story.

“There seems to be a much more concerted effort to discredit polls in this 2012 election cycle. With every poll release in the media, there is a concerted letter-writing email campaign to say it’s not valid because the sampling is wrong.

"This is happening from the Republican side, which is interesting, because some of the polls have the Republican candidate ahead. The emails all have the same language, and other pollsters I’ve spoken to are getting the same form-emails. This is the first year I’ve seen what appears to be a concerted effort on the part of one party to undermine the credibility of the entire polling industry.” 20

It’s fine to be skeptical of individual polls, he says. “But once trends start to build and you start seeing underlying dynamics – such as this group is more enthusiastic than that group, this issue is more important than that issue – those are things that politicians should be paying attention to. This is what the poll has learned is important to the voter.” 21


“If you wish to see the truth,
then hold no opinions for or against anything.”
     – Chien-chih Seng-ts’an, Third Zen Patriarch, 606 CE



· Public Opinion Surveys – Cornell Institute for Social and Economic Research

· – political and commercial opinion poll digest

· USA Election Polls – current U.S. election-related polls

· National Council on Public Polls (NCPP) – defines professional standards for polling


1 “The Duration in Office of the Executive”, by Alexander Hamilton. Federalist Paper No. 71.

2 The Phantom Public by Walter Lippmann. Transaction Publishers, 1925. Pp. 36-37.

3 The 480 by Eugene Burdick. McGraw-Hill, 1964. P. 57.

4 “Do Polls Help Democracy?”. Time, May 31, 1968.

5 Quoted in “Numbers Overload: Polling Data Hype Sways Voters” by Halimah Abdullah,, Oct. 9, 2012.

6 Ibid.

7 “‘Dewey Defeats Truman’ Disaster Haunts Pollsters” by Will Lester. Associated Press,
   Nov. 1, 1998.

8 Video interview at Hall Institute of Public Policy-NJ, Aug. 24, 2012.

9 Ibid.

10 Ibid.

11 Ibid.

12 The Rational Public: 50 Years of Trends in Americans' Policy Preferences by Benjamin Page and
     Robert Shapiro. University of Chicago Press, 1992. P. 33.

13 Micromotives and Macrobehavior by Thomas C. Schelling. W. W. Norton & Company, 1978
     (revised 2006); “Flocks, Herds, and Schools: A Distributed Behavioral Model” by Craig W.
     Reynolds”. Computer Graphics, Volume 21, Number 4, July, 1987.

14 The Big Sort: Why the Clustering of Like-Minded America is Tearing Us Apart by Bill Bishop.
     Mariner Books, 2009.

15 “From Tweets to Polls: Linking Text Sentiment to Public Opinion Time Series” by Brendan
    O’Connor, Ramnath Balasubramanyan, Bryan R. Routledge, Noah A. Smith. Proceedings of the
    International AAAI Conference on Weblogs and Social Media, May 2010. P. 1.

16 Ibid.

17 Video interview at Hall Institute of Public Policy-NJ, Aug. 24, 2012.

18 Ibid.

19 Ibid.

20 Ibid.

21 Ibid.