Yves here. I am deeply prejudiced against popularism because that the chief tool Marcie Frost is using to run CalPERS into the ground. Poll like crazy to find out what people want to hear, tell them what they want to hear, and rely on them not paying attention to what you are doing (which at CalPERS is staff rather than beneficiary-serving).
In the details of the discussion below, Matt Bruenig correctly points out how Daniel Shor tries to make out that the difference between 49% and 53% in a poll response is a big deal in deciding on policy. Aside from the fact that this was a “down the menu” choice (you’ve decided on a burger, now you are deciding on whether to have it with bacon or cheese), yet another reason for questioning it is that poll results are very much dependent on how questions are phrased and ordered. Experts in the dark arts of surveys can easily move results by ten points by the structuring of a poll.
By Matt Bruenig, a lawyer, policy analyst, and founder of the People’s Policy Project. Originally published at his website
Ezra Klein did a piece last week about David Shor and so-called “popularism,” which was ultimately defined this way:
Democrats should do a lot of polling to figure out which of their views are popular and which are not popular, and then they should talk about the popular stuff and shut up about the unpopular stuff.
I thought Ezra did a good job with the piece and I left the piece mostly unshaken in my views about what messaging wins elections, which are ¯\_(ツ)_/¯.
It’s an important topic that reminds me of a similar debate about what unions need to do to reinvigorate the labor movement. Winning elections and reinvigorating the labor movement are critical to social democracy. Both are hard to do. Both have a bunch of self-styled gurus claiming to know the secret sauce. And yet, after reading a ton of words about both, I really don’t know the answer and I am pretty skeptical of those that say they do.
Not too long ago, we were being told that Barack Obama had revolutionized the election messaging game by using “big data” to microtarget voters. Under the microtargeting theory, you don’t choose between message one and message two. Rather, you deliver message one to people who like to hear message one and deliver message two to people who like to hear message two. In a world with tons of data about every individual, this is something you can conceivably do, or so the story goes.
The microtargeting stuff seemed to fall out of fashion after the 2016 election, perhaps because Hillary Clinton used it and lost. Or perhaps because the nature of messaging talk is that it waxes and wanes over time. Whatever the reason, there seems to be a market for new things now, and so “popularism” has entered the game.
Although I don’t have much to say about the theory of “popularism,” I have recently had a run-in with its theoretician, the man himself, David Shor. In general, I am fond of Shor even if I am not so sure about the whole genre of election messaging theories. He got fired for some bullshit and he’s good with numbers. What’s not to like? Well, his opinion on the Child Tax Credit for one.
At Yglesias’s Slow Boring newsletter, Shor argued that the Democrats should create a stricter income test for the monthly Child Tax Credit in order to make the benefit permanent. If we put aside certain policy confusions contained in the piece, the argument is fairly simple: if we have a fixed pot of money to work with, it would be better to use that money to make the CTC permanent (rather than just extending it a few years) instead of using that money to keep the benefit nearly universal, which it currently is.
From a popularism perspective, the whole piece is very confusing.
For starters, popularism is supposed to be about what you say not what you do. Popularism does not tell us to do popular things and not do unpopular things. It tells us to talk about popular things and not talk about unpopular things, all while doing whatever we want. In this sense, issue polling, no matter how scrupulously done, is simply irrelevant to the question at hand since the question is about policy and not about messaging.
But even if you allow popularism to inform policy decisions, in this case, it seems like it would have reached the opposite conclusion than the one Shor did.
In the piece, Shor says that a fully universal CTC polls at 49% while a CTC that phases out at $50k of income polls at 53% (4 points higher!). Normal people might look at these numbers and conclude that it doesn’t seem to matter very much whether you make the benefit universal or have a strict income test. But Shor knows better:
The CTC with no income threshold — meaning it’s not means-tested at all — ranks in the 21st percentile of the nearly 200 issues Blue Rose Research has polled. Targeting the expanded CTC only to households making less than $50,000, however, ranks in the 60th percentile. That’s the difference between an issue Democrats will face attacks on and one they can feel comfortable talking about.
With all due respect, this is one of the dumbest paragraphs that I’ve ever read in my life. It sounds sophisticated, but it makes absolutely no sense.
All Shor is saying here is that, of the 200 or so issues he polled, about 78 (or 39%) of them polled somewhere between 49% and 53% support. That’s seriously it. No more data-wizardry is going on than that.
He makes this out to be a big deal, but the claim “the difference between 49% and 53% is pretty small” is absolutely not refuted by saying “39% of all the issues I polled came in between those two numbers.” These two things have nothing to do with one another!
Furthermore, the percentage of policies that fall between 49% and 53% support is entirely dependent on what policies you choose to poll. I can think of hundreds of policies that would poll outside of that range (e.g. legalizing murder), which were not polled. In the total universe of all possible policies, the difference between 49% and 53% is definitely not 39 percentiles, and 49% support is definitely not sitting at the 21st percentile.
If we included way more policies into the data set, the percentage of policies polling between 49% and 53% would shrink dramatically, which by Shor’s own logic would make this 4 point gap fairly insignificant (but also remember here that the logic is trash).
Put simply, the observation that a lot of the issues Shor chose to poll hover around 50% support does not generate the conclusion that the difference between 49% and 53% is big when we are talking about the effect that a single detail of a single policy will have on elections and election messaging.
Believe it or not, it actually gets even crazier than all this. Remember from above that Shor is saying we should give up making the CTC nearly universal in favor of making it permanent. Yet, in his piece, he makes no mention of how making the program permanent polls. I don’t pretend to know what the deep-down real number is for that question, but plenty of polls have it way underwater, such as this recent Politico/Morning Consult poll, which says that only 35 percent of Americans favor a permanent CTC benefit.
So near-universality polls at 49 percent while permanence polls at 35 percent and thus popularism tells us to … sacrifice the more popular thing in order to achieve the less popular thing? Huh?
Of course, pointing out that Shor seems to have gone way off the script on this issue doesn’t disprove “popularism,” which is a theory that stands alone from its chief theoretician. But Shor should perhaps take a lesson from his own book and be careful about what kinds of things he associates with the term. If the term becomes associated with shoddy policy prescriptions that overstep the bounds of what the theory is even about (election messaging) and don’t even really follow from the data available, then popularism may become unpopular among the audience it is trying to reach.