The last few years have seen opinion polls take something of a beating, with skeptics arguing that they have failed to accurately predict recent political events such as the Brexit referendum and the election of Donald Trump.
New research from the University of Michigan finds another reason to dispute the reliability of polls, but this time, the problem is more to do with us than the pollsters. The study finds that we’re much more likely to think poll results are credible when they show our preferred candidate or party in a strong position.
“On a number of fronts, it is clear that people believe what they want to believe,” the researchers explain. “It’s depressing, but not really surprising, that they are willing to cherry pick which polls to trust in ways that support the narrative they want to hear.”
Polarized society
The results do nonetheless confirm some of the challenges our increasingly polarized society faces however. For instance, when people have such differing expectations based upon their political persuasion, it can easily foster feelings of fraud when the actual outcomes are so markedly different.
What’s more, it seems clear that people are not evaluating the reliability and validity of polls via the methodology they use or even the reputability of the polling company, but rather by the results and whether they confirm their existing biases.
“The process of motivated reasoning, especially in our currently polarized environment, is complicating civil discourse about politics,” the researchers say. “The evidence available through well-conducted polls is not subject to evaluations based on their methodological quality. Accuracy and credibility are assessed in terms of whether the results confirm preexisting attitudes and beliefs.”
This was clear as the researchers presented polls to several hundred volunteers via a newspaper article reporting the results. The polls concerned Donald Trump and Hillary Clinton, but the results were manipulated to ensure one or the other was leading. They also manipulated the methodology to present the polls as either high or low quality.
The volunteers were then asked to rate the reliability and accuracy of the polls, before making a prediction for the outcome of the election if it were held the next day.
Distorted view
While the findings were broadly consistent across the participants, there was nonetheless a trend whereby more educated volunteers were better able to spot the high-quality polls.
It’s an outcome that the researchers believe have some significant implications for the robustness of our democracy.
“Biased perceptions of polls can affect election turnout and voting preferences,” the authors conclude, noting that highlighting methodological quality can lessen voters’ biases.
