A YEAR AFTER the votes were tallied in the 2016 presidential election, many questions about the election’s outcome continue to vex even close observers of US politics. Though broad-brush portraits of certain voter groups have circulated widely, many consequential facts about the composition of the electorate, and voting patterns within it, remain far less well known. Further, some prevalent narratives that emerged in the immediate aftermath of the election persist despite the appearance of evidence refuting, reducing, or otherwise complicating them. And this is not even to speak of issues for which research has only scratched the surface, and that may yet prove just as important as the dynamics laid out in this report.1 All our gaps in knowledge about the election make it not only difficult to comprehend how we got where we are, they also leave us ill-equipped to prepare for upcoming elections.
Few US presidential election outcomes can be explained by one single overarching factor, and we should be skeptical of any such explanations proffered for the 2016 election. Numerous analysts have argued for one or another dynamic as the key determinant of the election’s unforeseen result. Such arguments generally take the form of identifying a voter demographic that gave too little support to Secretary Hillary Clinton, flocked overwhelmingly to the GOP’s Donald Trump, or failed to show up in sufficient numbers to the polls.2 When one closely evaluates these arguments, what is astonishing is how many of them, in fact, make a convincing case. Many are the scenarios that, had they gone only a bit differently, might have tipped the election in Clinton’s favor. In this sense, the only short answer to the ubiquitous question, “What happened?” is: Everything.
Our What Didn't Happen? report provides a more thorough and detailed elaboration of what that “everything” included. This report is the first in a set of publications from a collaborative research and analysis venture between the Haas Institute for a Fair and Inclusive Society at UC Berkeley, and Tides to engage the evolving post-2016 US political landscape.3 Any such undertaking must be rooted in a nuanced and accurate diagnosis of the current moment. This report offers such a diagnosis.
It is organized around six broad sub-groups of voters—the young, women, people of color, “working class” voters, Rust Belt voters, and whites—for which prominent narratives developed during the 2016 presidential race, and since. Each section of the report addresses at least one of these narratives, investigating its accuracy through various breakdowns of election data.
Beyond responding to popular “takes” on the election, we also use available data to reveal less widely discussed patterns of voting behavior, and to put forth our own conclusions.
We analyze voter data disaggregated by age, sex, race and ethnicity, and other social and economic characteristics, as well as geographic units. For any given group of voters classified in terms of one or more of these variables, several types of figures may prove important to understanding election results. A first and most obvious class of data are those that show which candidates received how many votes from which demographic group. This may be rendered as the share, or percentage, of the group that voted for a candidate, or as the margin by which a candidate won or lost the group. A second class of figures describes voter turnout. It includes both participation rate—the percentage of a group’s voting-age citizen population that actually cast ballots—and the related measure of vote share. The vote share of a given group refers to the percentage of the total population of participating voters that belongs to that group. It is an inter-group, relational way of talking about turnout. In some cases we also consider a demographic group’s potential share of voters, that is, its share of the total voting-age citizen population.
Arriving at reliable breakdowns of voter behavior requires considerable care and triangulation across multiple data sources, each with its own strengths and weaknesses. The obstacles to direct knowledge of these breakdowns begin with the longstanding norm of ballot secrecy, which is meant to shield voters from undue influence or intimidation by ensuring that their vote choices will never be disclosed compulsorily. Any analysis of voting patterns must thus begin with geographically-defined counts of anonymous ballots, and voters’ self-reports of who they supported. The complexities multiply from there. The conclusions drawn in this report are based on in-depth analysis of state-certified voting results, exit polling figures, data from major post-election surveys, and the US Census Bureau’s Voting and Registration supplement. We mediate inconsistencies among these sources to present a clear and detailed portrait of the electorate—the total of eligible voters who actually cast ballots—grounded in the best available evidence. A brief addendum following the report’s main text describes each of the principal data sources we used and explains their respective strengths and weaknesses for understanding the election.
Finally, to identify the important patterns in any voting cycle also requires contextualizing current data within longer historical trends. At the same time, it is important to remain mindful that every election involves certain historically particular factors that must be taken into account if our comparisons are to yield valid lessons. Much of this report compares results from the 2016 presidential election to those of its most recent precursor in 2012. As we draw attention to shifts between these elections, we make every effort to acknowledge when the simple comparison of data points requires qualification due to particularities of the 2012 election. To historically contextualize the 2016 results is also about more than just explaining them, of course. It also means pointing us to trend lines that will be important as we shift from grappling with the recent past to embarking upon strategy for the near future.
- 1. Given multiple ongoing investigations into Russia’s role in the election, recent revelations about the manipulation of social media during the campaign, and the release of Hillary Clinton’s memoir, What Happened, we may still see new issues claiming pride of place in prevailing narratives about the election.
- 2. This is leaving aside arguments emphasizing the role of foreign meddling and contrived scandals.
- 3. Future publications will address, for example, what recent research tells us about how we can fight voter suppression; how different forms of racial, cultural, and economic anxiety influenced 2016 voting choices; and what role anti-government and anti-elite sentiments are playing in ongoing reconfigurations of the electorate.