Dynamic Constraint and Other Beautiful Lies
The concept of “dynamic constraint”1 has been hanging around political science and sociology quarters like an old mythical creature: it’s always cited, often hoped for, regularly tested, and rarely found. I know at least a dozen incredibly clever people who have gone looking for it in the wild with usually “mixed” or negative findings, or at least, results that are hard to interpret.
1 The notion that when people change their mind about one belief, they also change other related beliefs to stay logically or ideologically consistent.
2 OK, this is not a verb.
Almost all of us file-draw2 our trials, since most of us think that even though dynamic constraint works in theory quite well, the empirics are just too hard to pin down: (a) large and durable changes in political preferences are rare, (b) there are numerous political influences which may affect an individual’s propensity to update, and (c) even though experimental interventions can induce effects, they are small, short-lived, and reliant on questionable assumptions on “change.”
Well, against this background, naturally, I decided to give it a go myself.
I wrote up a tidy little theory, preregistered an experiment, ran it with over 2,400 respondents, and found exactly what everyone else seems to find: strong updates on one’s focal belief, and nothing for everything else. I decided not even trying to publish it. So here it is—the article, the results, and the wasted grad student money. As a responsible citizen, of course, I am sharing the data and code here. Do with it what you will. I am hoping that someone out there will prove or kill it one day.
Below is the full theory as originally written (well, slightly original), for those curious enough to read. The write-up for experiment and findings is sketchy. But—at least—I have pretty figures!
Update in August 2025.
After several comments and private talks, I’ve found myself turning this into a “working paper,” so I’m killing all the write-up in this blog post in favor of this pre-print in SocArXiv:
![]() |
---|
While all my comments and fancy looks about not even trying to publish this is now mute, I’m happily keeping this blog post for posterity.