Using Data to Decode an Unprecedented Election Cycle
“Working with numbers is a way through all the noise.” - Bloomberg Politics editor Kelly Bare
Originally posted on Bloomberg Media Forum.
Political reporting focuses on human behavior: both the messages candidates choose and how people respond. But in this year’s unconventional election cycle, traditional coverage alone only tells part of the story. Enter Bloomberg Politics, which is forging a new, data-driven journalistic approach.
“Working with numbers, as long as you have discipline about doing it, is a way through all the noise to present insights into what the campaigns are doing, the impact it’s having on voters, and what might actually happen,” says Bloomberg Politics editor Kelly Bare. Read on for three ways Bloomberg Politics is mining unmatched data sets to impart a unique view of how elections are won and lost.
1. Radical Transparency: the Bloomberg Politics Poll Decoder
New technologies have driven increasing sophistication in the way campaigns conduct their own polling – and in the way journalists can crunch poll numbers. But what’s behind the polls themselves, and what does it mean for the big picture? That question is at the heart of the Bloomberg Politics Poll Decoder.
“Usually a poll comes out and there is reporting on the numbers, and oftentimes it just looks at the horserace – what presidential candidate is ahead and by how much,” explains Bare. “But as we know this time around, there are various demographic constituencies that are getting targeted like crazy. So we said, ‘How do we show people what polling looks like for these slices?’ ”
Seeing those details – what percentage of each polling sample was made up of Republicans, Democrats and Independents, for example – provides deeper insight than a single poll would, or even simply aggregating poll data.
The Bloomberg Politics Poll Decoder is a dynamic visualization that makes nuanced insight easily accessible. It plots individual national polls – credible phone surveys by major media institutions such as Bloomberg, YouGov, Morning Consult, NBC, and ABC, in some cases including otherwise unpublished data – to show how subsets are voting, and what the polling samples looked like demographically and by party affiliation.
