Our Data-Driven Approach to Making Sense of the 2020 Presidential Election

Written by FrederikBussler | Published 2020/06/16
Tech Story Tags: data-science | data-analysis | data | 2020-presidential-election | politics | hackernoon-top-story | politic-data | coronavirus | web-monetization

TLDR A data-driven dashboard analyzes how unemployment figures, GDP growth, restaurant activity, and more across swing states may impact the outcome of the 2020 election. The dashboard aggregates data from sources including FiveThirtyEight, RealClearPolitics, and PredictIt. Florida, Pennsylvania, Michigan, North Carolina, Arizona, and Wisconsin are key swing states that will largely determine the election's outcome. Biden would likely become President if he won the vote in Florida and Arizona, while Trump would likely stay President. The data is updated weekly with Apteo's data connectors.via the TL;DR App

In less than five months, the world’s attention will be drawn to the outcome of the US Presidential election.

What's At Stake?

Regardless of whether you're on team "No Malarkey" or team "Keep America Great," you'll recognize that America is facing unprecedented challenges, and the 2020 election is of utmost importance.
Issues like COVID-19, police brutality, and mass unemployment won't just "go away" in the next few months, which means we'll have a pandemic, protests, and widespread fear looming overhead in November.
To understand what role these variables will play, we'll need to take off the blinders and look at the data.

Swing States: The Key to The Election

Predicting the outcome of the 2020 election largely comes down to the swing states, so we don't need to analyze all 50.
California's electoral votes, for example, almost always go to a Democrat, while Texas' electoral votes almost always go to a Republican. Most states consistently vote for the same party, election over election.
Swing states, on the other hand, don't have this affiliation.
There are currently 6 key swing states that will largely determine the outcome of the election: Florida, Pennsylvania, Michigan, North Carolina, Arizona, and Wisconsin.
Between these states, there are 102 electoral votes. Beyond typically blue states, Biden would need 38 additional electoral votes to win, while Trump would need 66 additional electoral votes.
To give an example, Biden would likely become President if he won the vote in Florida and Arizona, while Trump would likely stay President if he won Florida, Arizona, Wisconsin, and Michigan.

The Data

This data-driven dashboard analyzes how unemployment figures, GDP growth, restaurant activity, and more across these swing states may impact the outcome of the 2020 election.
The dashboard aggregates data from sources including FiveThirtyEight, RealClearPolitics, and PredictIt.
Among other insights, this shows that the aforementioned swings states lost over 8 million jobs in 12 weeks.
As CNN summarizes, "the economy has gone from President Donald Trump's greatest political asset to perhaps his biggest weakness." A poor economy and high joblessness has resulted in lower approval ratings as well.
Record unemployment in those six states hardly bodes well for President Trump. As Ed Mills, Washington policy analyst at Raymond James, writes:
"Every incumbent president facing a recession in the lead-up to reelection has lost."
With 29 electoral votes, Florida in particular will be crucial in the upcoming election. A graph of Florida's initial unemployment claims + unemployment rate is looking bleak, with similar graphics for the other swing states.
That being said, the future is looking brighter than the past. Using OpenTable restaurant activity as a proxy for the idea of an economic "return to normal," we see that things are moving up for Florida.

Some Caveats

While high unemployment in the swing states increase Biden's odds of winning, there are some (major) caveats.
First off, you shouldn't use the static graphs above to try to predict the outcome in November. Fortunately, the actual workspace is updated weekly with Apteo's data connectors.
Secondly, there are many more variables at play. Data like unemployment and restaurant activity, while informative, are not conclusive indicators of the voting population's psyche.
You can connect other data sources you see relevant (as shown in the video below), and make more accurate predictions based off a greater set of information.
Finally, 2016 should serve as a reminder that elections are incredibly difficult to predict. Virtually every major news outlet predicted a landslide win for Clinton, and we know how that turned out.
By collaborating on the dashboard we've just analyzed, you can be part of the solution in forming better insights!

Written by FrederikBussler | Published author and writer.
Published by HackerNoon on 2020/06/16