Pulling Back the Curtain on Race Ratings

by Nathan L. Gonzales February 5, 2021 · 2:26 PM EST

Each election cycle, race ratings are an integral part of the discussion about the fight for the House and Senate majorities. But where did they come from and how do they work?

While a few folks such as Kevin Phillips and Alan Baron dabbled with ratings going back to the late 1970s, the first person to make ratings a regular feature was Charlie Cook back in April of 1984, even before his newsletter was called The Cook Political Report. Stuart Rothenberg followed suit in The Rothenberg Political Report (now Inside Elections) a few years later, and the rest is history.

Now it seems like everyone does ratings, putting races in Toss-up, Lean Democratic, Likely Republican, etc., but I can only speak to how we do it. 

There are many ingredients to race ratings including past election results, district demographics (including partisanship, race, and education), open seat vs. incumbent race, challenger quality, incumbent strength, fundraising and outside spending, district-specific polling, the national political environment, and any unique dynamics, such as someone getting indicted...hypothetically

When it comes to race ratings, there are a few key points that might be helpful to understand how they come to be.

We don’t have a specific formula, and we weigh factors differently throughout the cycle. For example, early on, ratings reflect past election results, demographics and incumbent strength. But as the cycle progresses, candidate quality and fundraising matters more. And, at the end of the race, ratings are more heavily weighted toward district or state-specific polling. 

Ratings are meant to evaluate a party’s likelihood of winning and not an attempt to predict the margin. For example, in 2018, we never moved the Texas Senate race any more competitive than Likely Republican- not because we didn’t think Democrat Beto O’Rourke could get close, but that he ultimately couldn’t get enough votes to win against GOP Sen. Ted Cruz. O’Rourke lost by 3 points.

Ratings are a combination of where a race stands and where it’s headed. If ratings were merely a reflection of current polling, there wouldn’t be any need for ratings. We try to look at the important factors and project where it is most likely to end up on Election Day. For example, in Tennessee, Democrat Phil Bredesen was leading Republican Marsha Blackburn in polls into mid-September of 2018. But we never moved our rating to anymore more competitive than Lean Republican. And Blackburn won by more than 10 points. 

Sometimes looking ahead means not changing ratings with every burp and hiccup of a race. I’d rather wait and identify the trend of the race than bounce the rating back and forth a bunch of times.

This cycle, we’ve made a slight change to our ratings process. Instead of using specific rating categories from the get-go, we’re using broader terminology. We’re identifying which races are part of the Battleground —  in other words, which races are competitive and will decide the majority — and which races start as Solid for one party or the other. Later on, as the cycle and the races develop, we will shift to our more traditional categories.

Finally, ratings aren’t set in stone. One of the worst things we can do is to put a rating on the race a year or more from an election and be stubborn and not move it. As the cycle evolves, so do our race ratings.