​

I'm no Steve Kornacki, but here's a map

Actually, it's three maps!

For fun, last week I bootstrapped a small but super-current social media corpus to bring you three fun maps. I looked at LinkedIn posts last week from 100 political superposters in each US state. State by state, I mapped out positive and negative sentiment in social posts about Joe Biden, Donald Trump, and Kamala Harris, respectively.

Naturally, all three politicians are the subjects of several positive and negative posts in all 50 states and Washington, DC. I wanted a simple way to map the most prevalent statewide sentiment about each one, so this is how I labeled things:

• Very positive: >70% of the collected posts from the state are positive
• Somewhat positive: 55-70% of the the collected posts from the state are positive
• Neutral: Sentiment in the collected posts from the state is a balanced mix of positive and negative (within the 55/45 threshold in either direction)
• Somewhat negative: 55-70% of the the collected posts from the state are negative
• Very negative: >70% of the collected posts from the state are negative

I find election season stressful. Mathing on some data is my way of coping with stress. Welcome to a very special episode of nerd processor that stands outside our usual content arc! (But if you like simple, accessible data stories, you should totally subscribe because I write this kind of thing often.)

All the caveats

My methodology is probably flawed. Are LinkedIn users more likely be to red or blue? What about political superposters? What makes a superposter, anyway? (In this case, it's someone posting about politics at least weekly for the last 12 weeks.) Did I sample correctly? Are most superposters in cities, which tend to vote more Democrat? How will it look a month from now, when people are discussing something other than the Trump assassination attempt and the Democratic nomination process? All totally reasonable questions.

Then again, this methodology may not be significantly worse than what your favorite news outlet is showing you. So let's take a look!

Joe Biden

If I had sampled posts from three weeks ago, this map might have looked even redder. At that time, most posts about Biden commented on his viability as a candidate or on the specific decisions he was making as President. Last week, though, most posts about Biden were about his overall tenure as a public servant and/or his decision to step off the ticket. These tended to be more positive than the posts a month back.

Donald Trump

Here too, the map might have looked slightly redder if I'd built it a month ago. Last week, many of the Trump posts talked about the assassination attempt. Posts on this theme have mainly been sympathetic.

Kamala Harris

As you might expect, the volume of posts about Kamala rose sharply last week compared to a few weeks back. This is common sense, but I'm mentioning it because the map visualization does not make this clear on its own, and it feels relevant. A sitting Vice President gets less attention on social than the presumed nominee in the national Presidential election.

The bottom line

Taken in aggregate, Kamala posts are significantly more positive than Biden posts. In fact, Kamala post sentiment is neutral or positive in several states that have generally been Trump strongholds such as Texas and Florida. Plus, a few traditional swing states (Arizona, Nevada, Ohio) that are negative on Biden are positive on both Trump and Kamala.

These patterns are especially notable because circumstances from this past week ought to make both Biden and Trump as sympathetic as possible to their respective bases. But the sentiment in these maps makes it look like both Southern states and several swing states that looked like automatic Republican winners are very much in play.

That said, I'm no election prognosticator, and 14 weeks is an infinity of time in election season. November will show how well these maps hold up.

What do you think?

Thanks for reading!

Kieran

​Catch up on nerd processor case studies | Subscribe | nerdprocessor.com​

​

kieran@nerdprocessor.com
​Unsubscribe · Preferences​