Wednesday, March 06, 2024

The Uncommitted Story, Part II


I've been trying my best to give a dispassionate account of how the "uncommitted" campaign is doing. Obviously, supporters have an incentive to pump up its successes; opponents have a perhaps more mixed set of incentives (you don't want to give the impression that they represent the true majority, but there can be benefit in promoting a scary monster lurking in the woods). 

But for my part, I'd like to think we do ourselves no favors when we delude ourselves about the state of the world. If "uncommitted" is doing exceptionally well and demonstrating a genuine groundswell of opposition to Joe Biden's policies, there's no sense denying that just because one wishes it weren't so. If "uncommitted" is not performing especially impressively and doesn't stand out from always-present grousing at a coronated incumbent, then there's no sense denying that just because one wishes that weren't so.

So to actually figure out how "uncommitted" really is performing, it's important to establish our comparator. JTA put a piece up last night breathlessly comparing "uncommitted's" Super Tuesday support against how "uncommitted" fared in 2020. Framed that way, "uncommitted" had an outstanding night:

The uncommitted percentages barely dented Biden’s overwhelming win in each state, but far outdid 2020 percentages for uncommitted voters. In Minnesota, with 74% of votes counted at 10 p.m. Central Time, uncommitted was getting 20% of the vote; it garnered less than a half percent in 2020.

[...]

In Colorado, with 74% of the vote counted at 9 p.m. Mountain Time, uncommitted was getting 7.5% of the vote. It did not register at all in 2020.

In North Carolina,  at 11 p.m. Eastern Time, with 93% of the vote counted, uncommitted voters were 12.5% of those voting in the Democratic primary. In 2020, it was 1.64%.

In Tennessee, at 11 p.m. Eastern Time with 80% of the vote counted, uncommitted garnered 8% of the vote. It got less than a quarter of a percent in 2020.

In Massachusetts, with 51% of the count recorded at 11 p.m. Eastern Time, uncommitted was getting 9% of the vote. It got less than a half percent in 2020.

The problem is that 2020 is obviously not the right year of comparison -- an open Democratic primary with a sprawling field of candidates to choose from is very different from a reigning incumbent running for reelection (virtually) unopposed (if in 2020 you couldn't find a single Democrat of the approximately 531 running for president to "commit" to, I don't what your problem is).

So in terms of trying to give an objective assessment of "uncommitted's" performance, the actual comparison is to the last analogous presidential primary -- Obama 2012, since that was the last time we had an incumbent Democratic president running for re-election. 

In such cases, we would expect that there will always be some baseline number of people dissatisfied with the incumbent and looking to cast a protest vote. The question for "uncommitted" in 2024 is whether it is exceeding that baseline. Generously, we can assume that any overperformance compared to the 2012 figures is attributable to the "uncommitted" campaign vis-a-vis Gaza (though obviously, that might not be true). By contrast, if "uncommitted" isn't performing any differently (or worse!) than it did in 2012, then it seems unlikely that the "uncommitted" campaign is actually making much of a mark. So, for example, in Michigan "uncommitted" got 2.5% more in 2024 than it did in 2012, and then we have to decide what that level of improvement says about the strength of the underlying sentiment -- my conclusion was that this was a modest impact, but ultimately not too impressive save for the fact that Michigan's narrow margin makes anything meaningful.

With that in mind, how did "uncommitted" do compared to baseline expectations on Super Tuesday?

Unfortunately, Colorado and Minnesota didn't hold primaries in 2012, so we can't do a direct comparison. I will nonetheless eyeball agree that the 19% uncommitted took in Minnesota looks relatively impressive (though it actually isn't necessarily an outlier figure, as we'll see below). In the other three states, by contrast, things look very different for "uncommitted":

Massachusetts: 9% (2024) compared to 11% (2012)

Tennessee: 8% (2024) vs. 11.5% (2012)

North Carolina: 12.5% (2024) vs. 21% (2012)

These are all substantial underperformances compared to what we saw in 2012. Again, I understand why "uncommitted" backers are trying to juice them up, but these are not good showings! And these are the highlighted state where uncommitted did best! Except for Minnesota and Oklahoma (which seems to have a disproportionate share of randos on the ballot), Biden's broke 80% in every state he ran in on Super Tuesday. By contrast, back in 2012, Politico was running stories about Obama's primary weaknesses by pointing to states where he wasn't even cracking 60% of the vote (uncommitted got over 40% in Kentucky that year!).

So why is the media making a mountain out of this molehill? Certainly, "uncommitted" can give us some interesting microdata (the frustration among Michigan Arab and Muslim voters seems real, for instance, and notable). And in close states, any type of discontent can make a difference (though that proves too much -- any type can make a difference, meaning that any potential grouse or grumble is equally problematic). But I also think that we're seeing the effects of some relatively online journalists who are attuned to a relatively online campaign and so think there must be a "there" there. That, coupled with a deep-seated desire for anything that makes the horse-race story more interesting, and of course this is a tempting morsel.

But the reality seems to be that Biden actually is doing fine, compared to Democrats in analogous situations, of consolidating support. If anything, we've been seeing pretty persistent underestimation of his electoral appeal (itself perhaps a worthy topic for a post). "Uncommitted" right not seems to be mostly (not entirely, but mostly) sizzle rather than steak.

No comments: