starting to think I must just be really good at finding null effects

Going to #MPSA2026?

I have 3 papers on the program:

1. "Do virtual museums highlighting the experiences of minorities persuade visitors?"
April 24, 3:20-4:50pm CDT

2. "The China Penalty"
April 25, 8:00-9:30am CDT

3. "The Electoral Effects of Protests in the Trump Era"
April 24, 8:00-9:30am CDT

"The Electoral Effects of Social Protests in the Trump Era"

papers.ssrn.com/sol3/papers....

yep! let's set something up over email

Good idea.

excellent; thanks!

It's a fair point--one that prior work, which leverages these types of geographic proximity codings that we use tends to ignore.

Our plan, after MPSA, is to test for effects in neighboring counties. Hans and I did some of this in our work on gun violence and elections.

"The Electoral Effects of Social Protests in the Trump Era"

with:
@hjghassell.bsky.social
@reuning.bsky.social
Cindy West (a very talented FSU PhD candidate!)

papers.ssrn.com/sol3/papers....

This is very much a work in progress, so we welcome feedback.

We'll be presenting this paper at #MPSA2026 on Fri, April 24 from 8:00 to 9:30am CDT.

Why? It may be that in a polarized era, co-partisan protest preaches to the choir.

Opposition protest — especially when it turns confrontational — activates the other side.

It's not your marches that move elections. It's theirs.

When there rarely are effects, the mechanism appears to be one of backlash, not mobilization.

In particular, conservative-aligned protests appear to spark people in the surrounding area to vote more for Democrats.

Why do we get such different results from prior historical work?

The easy answer: the current moment is clearly more polarized.

The slightly more nuanced answer: prior work has studied protest at its most exceptional. We study it as a system-wide fact of political life.

To be clear: our estimates for protests' effect on vote shares are 1–2 pp, in 2% of counties, only for the most confrontational events.

Prior work using historical protests and a similar design frequently yields estimates that range from 5-12 pp.

Does that donation surge translate to votes? Modestly. Confrontational protests — and especially conservative ones — are associated with roughly a 1–2 percentage point increase in Democratic vote share in the following election.

Small effects, rarely realized.

The exceptions are revealing — and rare.

Fewer than 2% of protest county-months involve confrontational events (arrests, injuries, property damage). In those counties, Democratic donations surge. Effects accumulate for months. Republican donations: zero response.

So, do protests change elections?

Usually not.

Protests had no detectable effect on registration & turnout: regardless of the ideological roots of the protest, the protest size size, the scope of the protest.

The null holds for young voters, Black voters, female voters: all null.

We leverage difference-in-differences and event study techniques (including recent staggered estimators) to examine the effect of protest incidence.

Our paper uses the full universe of American protests from 2017-2024.

We examine ~200,000 protest events

Our protest data comes from the Crowd Counting Consortium.

We merged protest data with nationwide voter files, turnout returns, House election results, and FEC donation records.

Do protests change elections?

Prior research: "yes, and dramatically so."

But is that always true? What about protests in the last few years?

Our latest working paper challenges prevailing logic.

Our finding: most recent protests have failed to do anything to influence elections.

My favorite new diff-in-diffs estimator is whichever one doesn’t take three days to run.

New paper with Martha Johnson in APSR (@apsrjournal.bsky.social)😀 — Open Access 🎉📄

lnkd.in/eEPUxdi8

1) Terrorism ≠ rally effects.

2) Do women leaders face unique crisis penalties? No evidence.

3) Event-during-survey → trends matter ⚠️

Let’s go Mets!

A social media platform--but, like, purely for funsies.

April Fools' Day is dumb

There's one trait where economists clearly outperform the other social sciences:

Hubris.

Whoa.

The Republican advantage among Mormons has dropped nearly 20 points over the past two decades.

Nothing quite like the uniquely demoralizing experience of a reviewer recommending rejection for a paper they clearly didn’t read very closely.

Great new paper in The Review of Economic Studies using randomized incentives to detect non-response bias, using administrative data to provide ground truth for comparison. While incentives increased participation, they didn't reliably reduce NR bias

academic.oup.com/restud/advan...