In the wake of the 2016 election, researchers nationwide have been working to determine exactly why the results were so shocking—that is, why weren't we able to predict that Donald Trump would be our next president?
According to a new comprehensive report by the American Association for Public Opinion Research (AAPOR), higher education may have had something to do with it.
The report looked closely at past and present polling practices both at the federal and state level. After extensive analysis, AAPOR researchers were able to deduce a number of factors that may have led to skewed predictions, most of which occurred at the state level. One such factor was the failure to survey enough voters without college degrees.
As it turned out, voters without higher education were much more likely to support Donald Trump for president than Hillary Clinton. But they were also more difficult for pollsters to reach and less likely to respond to the poll requests that did reach them.
Not all predictions were wrong about Trump. These two college students got it right
Pollsters are generally careful to weight their polls; they try to gather a fair representation of voters from different demographic and socioeconomic groups and mathematically account for any gaps in their sample. But they weighted them inaccurately in this case because level of education has not historically had as much as an impact on votes as it did this past November.
In the 2012 election, for instance, voters without college degrees were similarly likely to vote for Barack Obama as the most educated voters—so even with polls that disproportionately considered voters with higher education, the predictions still held relatively true to the final results.
This past election, though, the skewed representation had a more significant effect, particularly in the most divisive Upper Midwest states. The AAPOR report reveals that in these states, "highly educated voters were terrible proxies for the voters at the lowest education level."
According to Courtney Kennedy, the director of survey research at the Pew Research Center, "If you did a poll and you had proportionally too many college graduates... you almost certainly overstated support for Hillary Clinton."
Other factors the report found to sway popular predictions past November include:
- A failure to account for last-minute voter decisions;
- An overreliance on voter turnout from previous elections;
- An inconsistency in poll quality at the state level compared with the national level; and
- Mistakes by poll aggregators in the media, which combined polling data with other data streams.
(AAPOR report, accessed 5/9; AP/New York Times, 5/4; Shepard, Politico, 5/4).
What Trump’s social microtargeting can teach enrollment leaders
Next in Today's Briefing
US opioid epidemic spreads to college campuses