Polls unreliable for predicting elections

Diane Sun

Editor's Note: This column appears in a point-counterpoint regarding whether polls are useful in predicting elections. To find its counterpoint, click here.

The democratic value of public opinion polling is hard to deny: how could popular sovereignty exist without a tangible way to capture public sentiment on the subject at hand? Yet readers should be skeptical of attempts to predict political outcomes based on opinion surveys.

In recent years, the predictive reliability of polling has been humbled. Polls inaccurately predicted the results of multiple 2014 Senate race and the 2016 state primaries. They failed to gauge referendums on Scottish independence and Brexit.

The traditional polling methodology is clearly faltering. Traditional assumptions of representative random sampling and error quantifications no longer hold. Even Gallup, Inc., which has ascendancy that is grounded in a dark-horse prediction of the 1936 election, would forgo making any election predictions this cycle than further risk its reputation.   

Polls struggle to capture a sample that is representative of the broader population. Determining likely voters is more difficult than ever; past voting history as an indicator is becoming less reliable. Tech-savvy candidates, like Trump and Sanders, leverage social media to mobilize groups with historically low participation. This year, youth and minority voters defied the polls and led to primary upsets in Michigan and New Hampshire.    

Response rates to surveys fell to a paltry 9 percent in 2012, and older, more conservative voters are more likely to respond, creating another source of bias. Statistical methods exist to increase the reliability of a non-representative sample, but then polling becomes more art than science. The predictive value, the argument of objectivity as truth, doesn’t hold. It is dependent on who is conducting the analyses.   

Still, other issues with polls are inherent to the surveys themselves. The spiral of silence, a term coined by sociologist Elisabeth Noelle-Neumann, states that individuals would rather not respond than express an unpopular opinion when surveyed. This effect is now amplified by the news media’s power to frame issues, as well as the vitriol of political discourse on social media, both of which tend to be more populated by liberals.

This corroborates the unexpected outcomes of both the Brexit and Colombia peace referendums. “Stay” and “Si,” respectively, were seen as the socially desirable responses, discouraging respondents who opposed those options. The Conservative victory in the 2015 U.K. elections invokes the spiral of silence on their supporters’ aversion to being labeled as “heartless.”

We have little to lose from weaning our faith in polls as predictors and treating them as they are — useful but fallible descriptive measures. Better to put faith in our own political judgment than rely entirely on our trust in numbers.

Sun is a business, honors, accounting and government junior from Sugarland. Follow her on twitter @sun_diane