After Big Data Failed: The Enduring Allure of Numbers in the Wake of the 2016 US Election

  • Yanni Loukissas Georgia Tech
  • Anne Pollock Georgia Tech
Keywords: big data, prediction, polling, elections


When widespread polling failed to accurately predict the 2016 US presidential election, producers and consumers of data didn’t abandon faith in numbers. Instead, they have reconfigured their relationships with big data. Producers are formulating redemption narratives, blaming specific datasets or poor interpretation, and the broader reception looks similar. Seeking an explanation for Trump’s unexpected victory, news audiences are calling out failed pre-election polling numbers, while at the same time embracing empirically dubious exit polls. This Critical Engagement piece argues that Science and Technology Studies scholarship has prepared us to see that polling errors would not undo the prestige and power of quantitative methods, but rather reveal the intensity of our attachment to data as a readily available arbiter. We show that data’s ambivalent qualities make it a durable ground for claims-making, with the capacity to be mobilized to do different kinds of work: blame, exoneration, and broader sense-making.

Author Biographies

Yanni Loukissas, Georgia Tech
Assistant Professor of Digital Media
Anne Pollock, Georgia Tech
Associate Professor of Science, Technology and Society
17 Feb 2017
Critical Engagements