The 2020 election will come down to two important questions: Who actually votes, and who do they vote for? Starting with the 2016 electorate, explore how shifts in turnout and voting patterns for key demographic groups could affect the race. Either start with the example below, or design your own scenario.
CLINTON
232
TRUMP
306
Click on a preset scenario or try it yourself with our interactive below.
Choose a group and see how it could impact the election. The gray zoneon the controls represents shifts from 2016 to 2020 that are most plausible, based on historic results.
How
voters could swing the 2020 election
BIDEN
232
TRUMP
306
This analysis helps illustrate a long-running argument in political science: Is it a better strategy to try to win over swing voters or to persuade people who will probably support you to actually vote?
Both are important, but you may have noticed that in this graphic, it’s a bit easier to flip states by changing the vote margin than by changing turnout. (We’re defining turnout as the share of voting-aged citizens – not registered voters – who cast a ballot.)
That’s because persuading a swing voter to change sides nets two votes: plus one for you, minus one for your opponent. Getting a new voter to cast a ballot is worth just one vote. Vote share also tends to shift more than turnout from election to election.
This analysis is a simplification of the actual 2020 elections, for several reasons.
First, we’re only allowing you to shift one demographic group at a time. Let’s say turnout among women increased by 10 percentage points in 2020. That could be caused by enthusiasm, but also by changes to state laws that made voting easier. Those law changes would affect many groups of voters, not just women.
[Vote by mail: Which states allow absentee voting]
Secondly, demographic groups don’t actually shift en masse. In this analysis, if you change Latinx voters to become five points more Republican nationwide, that shift is applied equally to Latinx voters in every state. Cuban Americans in Florida are a more conservative voting bloc than Mexican Americans in Texas. In reality, you wouldn’t necessarily expect to see the voters in these two states shift equally.
This analysis is based on the demographics of the 2016 electorate. It doesn’t account for how the voting population in these states has changed since 2016. As a result, some states that are changing more rapidly — such as Arizona and Texas — are considered in play for Democrats in 2020 but are relatively difficult to flip blue in this graphic.
Our own understanding of whether and how people voted in 2016 is inexact, based on big surveys such as the Census Bureau’s Current Population Survey, which measures turnout, and a variety of surveys that measure how different groups voted. These estimates of the 2016 electorate also do not account for voting-age citizens that have lost their right to vote, such as felons in many states.
The voting simulations rely on estimates that lean on the highest-quality source available for turnout and vote preference of key groups at the national and state levels, including the American National Election Studies validated voter survey, network exit polls and the Cooperative Congressional Election Study (CCES). Statistical modeling of the CCES was used to estimate support by educational attainment as well as other groups in states where stand-alone exit polls were not conducted.
Because of the sample sizes in these states, this graphic does not provide estimates for the smaller racial groups contained within the “Other” category, such as Indigenous Americans. Nor does it provide estimates for most crosstabs, such as combinations of age and gender.
Additional design work by Madison Walls and Lucio Villa. Ted Mellnik, Scott Clement, Emily Guskin and Lenny Bronner contributed to this analysis.
