Why cover the horse race? 📊 May 24, 2020
Elections have consequences. How do we properly calibrate our coverage of them?
Welcome! I’m G. Elliott Morris, a data journalist and political analyst who mostly covers polls, elections, and political science. Happy Sunday! This is my weekly email where I write about politics using data and share links to what I’ve been reading and writing.
Thoughts? Drop me a line (or just respond to this email). Like what you’re reading? Tap the ❤️ below the title and share with your friends! If you want more content, I publish subscriber-only posts 1-2x a week.
Dear reader,
Many of you will know that I’ve been spending much of my time recently working on a pretty complex forecasting model for this year’s presidential election. That’s for my day job, so I won’t blab about the technical details here on my blog, but one thing it has got me thinking about is the role that quantitative forecasting plays in our election coverage. Some of you have been simultaneously reaching out with questions on how I view the impacts of horse-race coverage at a time when the state of democracy in America is… precarious, to say the least. So that’s what I want to talk through today.
Why cover the horse race?
Elections have consequences. How do we properly calibrate our coverage of them?
David, a reader, emailed a very good question to me the other day. He gave me permission to excerpt the exchange here so that I could discuss the answer with everyone. He sends:
I was wondering how you are feeling about your work at a moment when it feels like the stakes of the election are increasingly consequential. Personally, I’m frustrated and frankly bewildered by the nihilism of this administration and their political allies and I suspect you are in a similar boat as me (though I know your work stands apart from your personal political feelings which I don’t need to know about if you feel it’s not what you want to reveal). Astead Herndon used to say this great thing: “Democrats tell themselves that 2020 is the most important election of their lifetimes, but really it was 2016 and you can’t just undo the damage.” I always begrudgingly agreed with that but now I think 2020 might really be that most important election. We are facing a near certain depression or close to depression economy due to a barely controllable pandemic runs a significant tail risk of wreaking much further havoc in the face of the administration’s disinterest in competence layered on top of deeply dogamatic conservatism. That’s a lot of anxious words; my apologies.
Anyway, was wondering how you are feeling writing about the electoral data given the policy and human stakes of the November election and if that makes you feel more inclined to do the work or makes you trepidatious.
It will surprise none of you to read that this election cycle has prompted me to spend a lot of time reconsidering the role of my work. I am biased and think it’s good and worthwhile, obviously, but I have found myself in need of a well-thought-out justification for why I feel that way.
The impetus for this subject is not hard to spot. Recent months have hammered home that elections can have massive consequences. Voters don’t just elect legislators and a president—they are putting in power policymaking, bureaucratic and judicial regimes that could very well be responsible for their very survival. As David points out, with America facing the threat of depression and a pandemic the stakes today are higher than ever.
When people ask me why we should cover the horse race, I assume they are raising the question because of one (or both) of two premises. The first has become more popular in recent years as there has been an increase in the research into the consequences of probabilistic forecasting. There are some obvious ways they might impact peoples’ behaviors; if you don’t think an election is close, it stands to reason that you might not “waste” your time actually making the trip to your local polling place. There is some empirical evidence for this; researchers Sean Westwood, Solomon Messing and Yphtach Lelkes find support from the ANES that people who report believing one candidate is very likely to win are less likely to turnout than those who believe the contest will be close:

With that being true, the operative questions become (1) whether seeing an election forecast (with a lopsided probability) will cause you to change your certainty about the contest; and (2) whether that will make you less likely to vote. (The second point is not directly supported by the above chart as there is possibly an intervening variable causing both high certainty in the outcome of an election and lower likelihood to vote, such as the strength of one’s partisan identity.) Westwood, Messing and Lelkes find that presenting readers with a probabilistic election forecast produces expectations that are actually better calibrated than if people don’t receive the information at all. When shown a low win probability they are less certain about the outcome; when shown a high one, they are more.

On the subject of whether being exposed to (over)confident election forecasts decreases turnout, the authors present some evidence from a behavioral study that I do not find totally convincing. Read their paper for more details on how it works, but I think it’s likely that people think about political participation in different ways than how the study manipulates the participants. So I have some reservations about whether or not the data actually translate to findings about the real world. But to be clear: I still think this is a big potential problem where we need to do more research. If my work is causing people not to vote, it’s probably not worth it.
On to the second premise. When people ask me “why cover the horse race?” there is often a subtle implication that there are more important subjects to write about. I agree with this to a point. In the context of political journalism, it is obviously often more impactful to cover government wrongdoing or corruption. Stories about politicians or civil servants misusing their power often have much larger downstream effects than who is up or who is down in the recent polls. This is especially true when an election is not on the horizon. That’s why my official job title is not election handicapper, but data journalist; I cover other topics when they are more important than this.
But this is not often the case. I think high stakes, politically-charged elections are where data-driven coverage can shine. Empiricists can cut through a lot of the partisan noise and biases that others don’t, or can’t, parse. And we are less reactionary, able to avoid the whiplashes of campaign coverage and articles driven by single datum by relying on properly-calibrated models of reality. And that’s valuable! If elections are as high-stakes as we think, it’s important to have the best journalism available about how the campaign is affecting voters. Knowing if a candidate is far ahead of another can help activists direct resources or other journalists direct their coverage.
It is not lost on me that determining the worth of data-driven political journalism is a comparative practice. Whether or not what I do is worth it depends on what alternative is available. Today, if election forecasters like me disappeared, the people would be left with their choice of pundits, most of whom have decidedly worse instincts about the horse race.
But there are still questions about how our work is affecting democracy and democratic participation. And I don’t know the answers to them.
Posts for subscribers
May 22: The US House majority is not up for grabs. The Democrats are very likely to hold the chamber after November’s elections.
What I'm Reading and Working On
I just got a copy of Eugene Burdick’s The 480 in the mail. It’s an old work of political fiction from 1964 about a future where elections are decided by computers and algorithms that tell candidates the strategies they need to use to win. (Sound familiar??) Burdick, a political scientist by training, thinks this is a bad future. I disagree, but it’s an interesting piece of narrative fiction, and it does prove to be remarkably prescient and a good read.
Thanks for reading!
Thanks for reading. I’ll be back in your inbox next Sunday. In the meantime, follow me online or reach out via email if you’d like to engage. I’d love to hear from you!
If you want more content, I publish subscribers-only posts on Substack 1-3 times each week. Sign up today for $5/month (or $50/year) by clicking on the following button. Even if you don't want the extra posts, the funds go toward supporting the time spent writing this free, weekly letter. Your support makes this all possible!