Deconstructing WAR — and is it really stacking the deck for moderates?
Here’s a very detailed breakdown of our WAR metric, which also responds to questions from the last few days.
Post
Conversation
![David Watson 🥑](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
Post your reply
The reason progressives do poorly in WAR has to do with some elementary controls applied. Informing them (incumbency, lagged presidential vote) would get you laughed out of the room.
And yes, we fit on the real 2024 topline. That’s different from Grumbach/Bonica and Morris.
As for what’s better or not, you decide. Modeling is hard and you can make a lot of choices. I do like ours more — it’s much more explainable. I can’t explain Miler-Meeks being +1 while Sanford Bishop, Adam Gray and Shomari Figures are net negatives.
One last note — modeling is very difficult and it took us three years to arrive at a good, stable model. I fundamentally disagree with a lot of what Morris/Bonica did, but this is really hard stuff.
Happy to talk about whatever.
I updated this piece to add one important explanation for why we disagree with the Strength in Numbers approach of not using the actual presidential result in WAR.
You simply have to consider it, IMO, because outrunning Harris by 2 is very different from running behind her by 2.
The same things that cause Harris to get more votes than expected in the district generally *also* cause the House Democrats to get a few more votes, because the model missed on how Democratic the district was. IMO, this is why you have to control for the real presidential vote.
Is there an easy to way to run this just for competitive districts? Jayapal winning by 68 when she "should" have won by 77 doesn't seem very informative as a strategy to winning more seats. Whatever MGP is doing, potentially very useful.
You run it for everything and then can look at the top performers if you so wish :)
This is great, and FAr more logical than the complaints. But - aren't all winners in purple districts going to score well here? And winners in purple districts are PROBABLY moderate? And like, that's sort of obvious and intuitive, but also perhaps the crux of the complaints?
Not really. Why would all winners in purple seats score well? Baldwin is mediocre in our model, for instance.
the baseball analogy they make seems funny in light of rise of three true outcomes and all that in modern mlb
I have quibbles with WAR, but there's a lot of big WTF stuff in Morris' model. Since I care about progressives mainly, I looked up a bunch of the big ones and I can't find any explanation for Jayapal being +2. There's zero question like Omar she massively underperforms every year
I do think comparisons to “The Squad” or Freedom Caucus for moderates is a bit misleading (it doesn’t particularly make sense to me why a candidate would care about winning by 50 vs 60) but this is otherwise a strong rebuttal and I appreciate the extra info :)
What about selection into treatment? If there is some moderation effect, you would expect primary voters in swing districts to pick moderates who also have good unobserved candidate quality. This would mean the moderation effect is positive but overstated.
Good piece! I find it funny though that you repeatedly use Ritchie Torres as an example of a moderate given that he was a progressive on pretty much everything except palestine and crypto until recently. I guess his immigration pivot and leaving the CPC last year make it fair
Couple methodological issues.
1. Run a logistic regression. Margin of victory isn't the question -- it's victory.
2. By calling all the residual variance WAR, you end up inserting massive variance into the figure. Lack of correlation to outcomes implies this is an issue.
FWIW, as a leftist, I'm convinced you are correct on the merits of this discussion. That said, I resent the cynicism of the politics that Yggy et. al. advocate on this basis and will absolutely not vote for anyone fence sitting (or worse) on genocide.
ty for your detailed explanation. Your examples and writing style elucidate each point you make. As you got more into it, it was also fun to see your restraint as you were essentially calling out some of the idiocy of the critique against your WAR modeling.