So Long, RFK Jr! Don’t Let the Door Hit You On the Way Out
Good golly, what a week, huh? Honestly, when was the last time you couldn’t say that?
With the Democratic National Convention behind us and Labor Day fast approaching, I can confidently say that the Solid Purple forecast is at its most uncertain. From here on out, based on historical state and national polling error as well as an expected escalation in the number of surveys from reputable pollsters across both battleground and non-battleground states, we will soon be out of this weird little period where the tail ends of the distribution are so wide that Iowa is rated Lean R in our forecast despite the model finding “only” a three-point lead for Kamala Harris nationwide.
What does that mean, exactly? Those of you who have been paying close attention to the forecast since it launched earlier in the summer will start noticing, especially in the House and Senate races, that the projected margins will start looking a lot more like the polling averages. For example, in the Ohio Senate race, Sherrod Brown has not trailed in a single poll this cycle against his Republican challenger, Bernie Moreno, but the fundamentals and the generic ballot have weighed so heavily that the race had been Lean R until recently. On the other hand, Jon Tester had been reliably ahead in early polling of the Montana Senate race, but recent surveys have been more in line with where the forecast expected them to be all along.
After all, we are trying to forecast what will happen in November, not what would happen if the election were held today. Though, if you’re curious about what a “now-cast” would say about the race, stay tuned.
Before I get to the headline, a quick update on where things stand just before the point in the calendar where everyone agrees that normal people (i.e. not you and me) start paying attention to the election. In the Presidential race, Harris has continued her steady rise from a 34% chance of winning the election on the day Joe Biden stepped aside on her behalf, to a 55% chance today. For the first time, the forecast projects her getting over 50% of the vote, a feat neither Donald Trump nor Joe Biden has been able to achieve since Solid Purple launched.
Democrats are starting to look like they have a real shot of holding the Senate these days too, which to be honest, I had previously felt was impossible, and still have a hard time picturing in practice. It’s still a substantial lift, as they only have a 35% chance and almost all simulations where they do hold on are by the skin of their teeth, but the path is starting to appear. Presidential swing states where Democrats are playing defense are solidifying in their favor because Republicans picked candidates who were, frankly, dogshit, and pickup opportunities in states like Texas and Florida, while unlikely, no longer appear farfetched. And who knows? There’s an independent candidate in Nebraska who is threatening to make things interesting. Watch this space.
We finally see Democrats taking the narrowest lead in the House now, too, in no small part due to the model reorienting slightly around Mary Peltola getting over 50% in the Alaska jungle primary, bringing her odds up from around 40% to 60%. Even that’s too low, I’d argue, since as I stated before, the model is baking more uncertainty into the simulations right now than it will at any other point the rest of the cycle.
One thing not adding uncertainty into the forecast is the exit of anti-vax nepo baby Robert Kennedy Jr. from the Presidential race. In a juiceless attempt to try to recreate the paradigm shift introduced by Biden’s timed exit almost immediately after the Republican convention, Trump held a rally together in Phoenix to announce the withdrawal and subsequent endorsement by a guy whose family name is synonymous with the Democratic Party.
Some have worried that RFK Jr., who once had been getting anywhere from 10-15% in the polls but had nosedived into the low-mid single digits, will be able to successfully carry his support over to Trump. Nate Silver, FiveThirtyEight, and other models are now having to make changes to their models to account for Kennedy exiting the race. Here at Solid Purple, no adjustment is necessary. Why? You can argue whether it’s good or bad on the merits, but I never took him seriously as anything other than a vessel for third-party voters to protest against the duopoly and baked that assumption into the methodology.
The way Solid Purple handles third-party candidates is somewhat rudimentary and I actually considered not factoring them in at all. I looked at how third parties almost always substantially underperform their polling, tracked how their standing in polls depreciates over time as the election draws nearer, pooled the results for all third-party candidates (RFK Jr., Jill Stein, Cornel West, Chase Oliver, etc.), and grafted that onto estimates of how “Other” tends to perform in each state. There are some slight adjustments made based on the number of undecideds in each demographic group in national surveys based on race and education, but frankly, I think I was being a bit pretentious when I decided to factor that in.
If my theory of the case is correct, then the RFK vote might drift off to Trump or Harris a little, but just as much of it might pivot toward the Libertarian and Green tickets, and there will end up being roughly the same amount of protest voting as there would have been otherwise. As of right now, that’s about 3%, significantly lower than 2016 but substantially more than 2020.
In that case, what benefit does Trump get from having the support of the weirdest fucking man in the world? I guess we’ll have to see.
