Skip Navigation

Primary Polling Problems: Can Horse Race Coverage Alter Results?

During the primary season, news cycles are dominated by a plethora of poll numbers telling us who’s up, who’s down, and who’s crushing whom in what state. Candidates cite their standings in the polls as an indication of their ability to attract votes and dollars. As a result, these polls become more than just a depiction of current voter opinion. Our obsession with them causes the polls to shape our perceptions of who can win the nomination and whom we should support. In this way, surveys are self-fulfilling prophecies: They provide a sense of opinion that influences the outcome of the election.  But polls are by no means always accurate. In fact, polls are heavily influenced by their methodology, their sample size, and the number of candidates running. While it’s tempting to fixate on polling numbers, our willingness to let potentially incorrect results shape the outcome of primaries is concerning.

Over the years, the accuracy of polls has decreased due to new regulations and changing habits.  For example, fewer people have landlines now — the primary method for executing surveys. In 2014, 43 percent of Americans did not use a landline, which means that just under half of the public couldn’t participate in polling that helps to predict the way the public as a whole will act.  Furthermore, the 1991 Telephone Consumer Protection Act prohibits automatic dialers on cellphones — another crucial tool used by pollsters to lower costs and reach more voters. The effects of these changes are only exacerbated by the fact that response rates to polls have decreased from 80 percent in 1970 to 8 percent in 2014. As a result, it is now more difficult and more costly than ever before to reach a representative population that accurately reflects public opinion.

“Live polls” are not much better. During the presidential debates, for example, a web tool created by Microsoft called Microsoft Pulse asks participants their opinions and broadcasts them in real time.  Participants respond to the question, “Do you agree with what the candidate is saying?” and a color-coded graph displays the results. However, the sample of these polls is naturally skewed: Internet users who are actively watching the debate and agree to participate are not representative of the voting public. Intuitively, one would expect these polls to be skewed towards younger voters. While viewers may instinctively take these results with a grain of salt, live “polling,” as it is often picked up by news media in debate recaps, is another outlet through which public opinion is shaped with perhaps biased information.

Additionally, the abundance of candidates in the Republican field complicates polling and hinders polling accuracy.  A Suffolk University poll transcript, for example, lists twelve candidates in a random order, asking the participant to choose their first and second choices among them.  With such a long list of candidates, voters may become overwhelmed and choose the candidate whose name they have heard most frequently, instead of having the ability to vet three or four candidates logically.

At the end, the poll includes head-to-head comparisons.  This is a clear attempt to simplify the field by only comparing the frontrunners of the Republican and Democratic parties (Donald Trump vs. Bernie Sanders and Donald Trump vs. Hillary Clinton) in order to predict the ultimate outcome. By listing Donald Trump as the only choice for Republicans, pollsters are reinforcing the idea in voter’s minds that Donald Trump will be the nominee, which might lead them to a greater acceptance of this idea.

All together, these challenges make polling less precise. However, the media tout polls as if they are news. Frank Bruni of the New York Times equates today’s polling condition with “gorging ourselves into a state of morbid obesity.”  American media are rampant with polls that Bruni suggests might not provide quality information; rather, he equates it with over indulgence that taints the voter’s ability to form an opinion. And, with the media’s flare for the dramatic, the news has a tendency to emphasize polls that contradict beliefs and shock the audience.  “Desperation makes a better story,” Bruni reminds us, as polls highlighting massive disparities between front-runners gain public attention.

The effects of the polls on public opinion are furthered through policies like the GOP primary debates’ eligibility criteria. Opinion polls were often used to determine which candidates were allowed to participate in the primetime debate (rather than the debate earlier in the evening). For example, the front runners in the ten latest polls decided who participated in the first primetime Fox News debate of the primary season.  Those who didn’t have high enough percentage didn’t get the opportunity to be prominently featured and perhaps win over more voters. In this way, the mobility of candidates in the field was limited by the results of potentially imprecise polls, especially for those candidates who had polling numbers clustered around the cutoff. In fact, all four candidates who were relegated to the undercard debate on January 14 have since dropped out of the race.  Carly Fiorina, who dropped out of the race on February 10, was excluded from the February 6 ABC News debate despite finishing ahead of Christie and tying with Kasich in Iowa, both of whom participated in the primetime debate.  The criteria for inclusion in that debate was a top three finish in Iowa, or top six in New Hampshire or nationally; the latter two based on polls.   While it is logical that the number of candidates had to be narrowed in order to conduct an informative and civilized debate, the ranking of candidates from the beginning based on opinion polls, which have questionable accuracy, limited all of the candidates’ ability to connect with the public.

Consequently, people’s perceptions are altered, and the bandwagon effect is not to be ignored. Donald Trump employs this strategy, as he touts winning poll numbers like achievements during rallies and debates. “People love me” is a phrase he proclaims.  This strategy is peculiar and has come under fire from the other candidates in the race.  However, he is still at the top of the polls.  Is the notion that more citizens claim to prefer Trump than any other candidate reassuring to voters who might be hesitant to support such an outspoken, politically incorrect candidate?  Is it a seal of approval that symbolizes to other Republicans that this is the most legitimate candidate?  Nick Gass of Politico suggests that,  “Strategically, it’s made his candidacy look as if it were feasible to primary voters,” in an attempt to legitimize candidate that people might otherwise be hesitant to support.

If this is the case, polls must be conducted with utmost accuracy, as the results of these polls effect further results through Trump’s strategy. Voters who may be less informed or interested may trust the decisions of the majority, which would only increase a candidate’s lead. If a majority of Americans are said to support a certain candidate, that candidate becomes a more legitimate force. This would be a healthy occurrence of democracy, if it weren’t for the inaccuracy of polls and their effect on voter’s ability to access all of the candidates.

This nature of a self-fulfilling prophecy in polls creates a dangerous outcome in American politics.  Poll numbers are so publicized and frequent that they attract a lot of voters’ attention.  Polls are viewed as measures of success, influencing the news coverage candidates receive and voters’ attitudes towards campaigns. This perception gives polls are great influence, but there are legitimate reasons to be concerned about the accuracy of many of the results. It is unclear to what degree polls alter voter opinion, but their prevalence is concerning.

About the Author

Kelly Conway '18 is a Political Science concentrator and a Staff Writer for the Brown Political Review

SUGGESTED ARTICLES