Read on Twitter
I’m excited to share some newly-published research!
In this paper, Ben Johnson and I argue that there is actually no good evidence that the Supreme Court responds to public opinion. https://twitter.com/PRQjournal/status/1308440974213906433">https://twitter.com/PRQjourna...
In this paper, Ben Johnson and I argue that there is actually no good evidence that the Supreme Court responds to public opinion. https://twitter.com/PRQjournal/status/1308440974213906433">https://twitter.com/PRQjourna...
Numerous studies over the last few decades have argued the changes in public preferences cause the Supreme Court to alter (some of) its decisions to comport with the “public will.”
People have offered lots of different theoretical explanations for this, but the consistent empirical finding seems to have led to something of a consensus that the Supreme Court responds to the public even if we aren’t sure why it does so (h/t to Epstein and @WashUChancellor)
Ben and I argue here that the apparent empirical regularity at the heart of this literature—the finding of a consistent correlation between public preference and Supreme Court outputs—is itself misleading.
This figure presents the basic correlation that drives this whole body of research: a positive, ‘significant’ correlation between public mood and Court output:
But we show that if you look a little closer, the story doesn’t really hold up.
For example, there is no consistent relationship over time—and as (f) indicates, there’s no evidence of a significant correlation in the post-Warren Era (i.e. since 1968)!
For example, there is no consistent relationship over time—and as (f) indicates, there’s no evidence of a significant correlation in the post-Warren Era (i.e. since 1968)!
And this is also true when the outcome is the individual Court decision, and we account for the amount of public attention a case receives (pre-decision attention, or salience):
In further analyses in the manuscript and supplement, we show that the null relationship is robust to many different model specifications, measurement strategies, subsets of the data, etc.
We also replicate 6 prior studies, and show that the key finding in each empirically suspect. In each, the finding is timebound, based on an endogenous measure of salience, based on an empirically unsupportable interaction term, or some combination of these things.
In sum, as yet we don’t think there is any good evidence that the Supreme Court responds to public opinion. This isn’t to say that it never does! Only that we actually don’t know—even though political scientists have treated this as more-or-less settled for quite some time.
Ben and I are working on a couple of other papers in this vein, so if you have questions/comments/suggestions, please pass them along so we can incorporate them into future work.
