Thread by @gelliottmorris

#NEW We have matched up county-level preferences for music genres with 2016 election results and have found the👏 key👏 data 👏 that capture the cultural currents that made Obama supporters flip to Donald Trump:

liking hard rock music.

THREAD.

economist.com/graphic-detail…

The data we have (from @VividSeats) is truly incredible. In every county, we have a record of the share of concert/etc ticket sales that fall into each genre in 2015/19. Here's that map (with each county scaled for population). Cities like rap/hip hop. Rural areas like country.

The music genre that most correlates with Republican vote share is, as you might expect, country music. A preference for "New Age" music (whatever that is) and Rock are also "red". Rap, Hip Hop & R&B, as mentioned, are "blue" music—as is liking alternative artists or Latin music.

And, if you zoom in on the Midwest, you'll see that Obama-Trump areas also REALLY like hard rock. In fact, a county's share of music sales that went to hard rock is THE SINGLE BEST predictor of voting for Barack Obama in 2012 and then Donald Trump in 2016, as far as genres go.

Perhaps, you might say, we are just tapping into demographic characteristics. Rural whites like country. They also like Republicans. Nothing to see here, right?

Wrong!

Incredibly, music is a good predictor of voting EVEN AFTER controlling for demographics!

We created models to predict 2016 Democratic vote share AND vote swings from 2012 to 2016 with every county's race, ethnicity, educational attainment, age and population density. Music preferences STILL MATTERED. Here's a plot showing that:

In 2016, Trump’s vote share in places where country out-sold hip-hop was 22 %age points higher than in hip-hop-heavy counties. Race, age, education & pop density explain just a 18-point gap. The remaining four points consist of factors reflected BY MUSIC but not by demography.

Under the hood, we also find that our ability to predict election outcomes with music preferences increases from 2012 to 2016. This suggests there's also something about which artists we like that taps into polarization and vote-switching!

One's appetite for different genres clearly has predictive power above standard demographic and political variables. This means there's SOMETHING about being a Chance the Rapper/Luke Bryan/Metallica fan that makes you more Democratic/Republican/Trumpy.

Shall we speculate?

Perhaps liking Latin music signals an openness to immigrants among white voters who are otherwise demographically the same? Liking Hip Hop could be a proxy for being involved in African America political communities. (Liking EDM also correlates with Dem vote share.)

On the other side of the aisle, perhaps liking country signals a familiarity with themes of rural living and "family values". Being a fan of AC/DC is perhaps connected to whites that are more "nostalgic", something that social science has identified as part of Trump's appeal?

Whatever they represent, music preferences clearly tap into cultural/psychological divides that demographics do not.

We think we've uncovered a particularly telling way to measure and convey them.

Please read the piece, and share if you liked it!

economist.com/graphic-detail…

A followup tweet. People keep asking me what I mean by "hard rock". @VividSeats designates Metallica, Guns N Roses, AC/DC, Tool, Iron Maiden, etc as hard rock

Here are some more examples of the "hard rock" artists included in the data...

Black Sabbath
Five Finger Death Punch
Disturbed
Korn
Rammstein
King Diamond
Avenged Sevenfold
Dream Theater