In 2012, U.S. Gallup polls predicted that Mitt Romney would beat Obama in the presidential election with a slight edge in public support. More recently, in 2015, public opinion surveys in Turkey predicted only trivial gains in vote for the ruling Justice and Development Party (AKP) during the country’s November general election.
In both cases, the polls missed the mark. President Obama blindsided Romney, winning a second term by five percentage points—a result even Romney’s own polling experts did not see coming. Turkey’s AKP won back its parliamentary majority with 49.5% of the vote and an unexpected 8.5% rise in public support, a rebound even the best polling companies in the country had barely foreseen.
Inaccurate poll results are not rare nowadays. An increasing number of disproven poll predictions, particularly in the context of elections, fuels the growing scrutiny over political polls. Cliff Zukin, Professor of Public Policy and Political Science at Rutgers and past president of American Association for Public Opinion Research (AAPOR) said in his article for the New York Times, “election polling is in near crisis.”
Is polling facing some major challenges? And what are they?
Some trends, among others, are likely driving the public concerns about the quality of polling.
- The dwindling response rate threatens the fundamental premise of opinion surveys. As detailed by The New Yorker’s Jill Lepore, the promise of polling is that the sample represents the population. However, as distrust of polls increases among the public, the number of people who are willing to participate in surveys goes down. To ensure an adequate response rate and minimize non-response bias, pollsters have to call many more people, which can be costly, and rely on demographic weighting (read more on Pew). Pollsters now also have to compete for time and attention in a hyper-speed, social media-fueled environment, in which individuals are becoming too busy for a ten-minute survey.
- The growth of mobile technology and Internet use has transformed the polling landscape. Surveys reliant on traditional landlines are likely to omit the growing population of those who only use mobile phones. Similarly, web-based surveys can omit those on the lower rungs of the digital divide. Digital stratification has a significant impact on polls’ ability to represent the full range of voices.
The truth is that polls of varying levels of quality and rigor are wielding enormous influence over the public agenda.
Polls do not simply take the pulse of the public. In many cases, they drive the public opinion.
On a small scale, for instance, Fox channels have relied on poll results to determine the lineup of its Republican primary debates in the United States. Last August, for example, Fox News Channel selected participants for its Facebook co-hosted debate based on Republican candidates’ standing in the five most recent national polls. Later in November, Fox Business Network’s long-waited primetime debate included only those candidates who had scored 2.5% or higher in the four most recent national polls. Those who did not make the cut appeared only in a “lower-tier” debate, broadcast two hours earlier.
While it is questionable whether polls are an accurate barometer of the public’s response to the field, it is clear that they have had an influence on which candidates got the most media exposure. Therefore, polls can influence how much the public thinks about a candidate and their platform.
On a broader scale, psychologists have long observed that people conform to majority opinion (a phenomenon called “bandwagon effect”). With the “majority opinion” consistently presented as polling data, we are all, to some degree, susceptible to the tendency to shift our preferences in favor of the “popular will.” In a recent study conducted by Stanford and Microsoft (Are public opinion polls self-fulfilling prophecies?), researchers confirmed that polls indicating public support for three public policies affected individual attitudes toward those policies. In this manner, pollsters can unwittingly set, rather than reflect, the public agenda and shape the “will of the people.”
The discussion about polling raises more questions than answers. To some, the real problem—as Jill Lepore put it—is “neither methodological nor technological.” Rather, it is political. Amid the fuss and buzz around the prevailing opinion and the forerunners of major polls, minority or dissenting voices are often diminished or silenced. This is a disheartening prospect for those who value democracy.
As the media continue to deluge us with polling results and influence our judgment, the real solution here perhaps is not blaming journalists or pollsters. Instead, the public in most democracies, where polls have played a prominent role in politics, needs a more sophisticated understanding and realistic expectation of polling. Some important reminders include: 1) Not all polls are created equally. We need to pay as much attention to their methodology as to their findings; 2) Poll results are more of an indication of attitudes than actions. Likely voters can still end up not voting at the ballot box (read more in Can Likely Voter Models Be Improved?). Therefore, stay calm and don’t get too sucked into the media hype over the “leading candidates;” 3) Polling is never the complete picture of public opinion. It is essential to hear the citizens in their own words through other sources and value minority voices that are not at the top of the polls.
Follow PublicSphereWB on Twitter!
Photograph of Proud Iraqi Women Vote in Nasiriyah by DVIDSHUB via Flickr, some rights reserved
Photograph of Open Data installation by Sebastiaan ter Burg via Flickr, some rights reserved