Descriptive findings (Wave 1): Disapproval of real‐world violations was much lower (47.5%) than for hypothetical ones (63.7%). Familiarity, concreteness, and context really amplify partisan biases, which are heavily underestimated with hypothetical scenarios, *even when keeping party constant*.

🛎️New WP with @morganlcj.bsky.social @timallinger.bsky.social and @danbischof.bsky.social

Against the surge of conjoints and other hypothetical experiments in relation to democratic backsliding, we study the consequences of using hypotheticals versus real-world scenarios.

osf.io/preprints/os...

Abstract below:

We also quantify the anticipated differences in breaking with Trump or Biden using various close-ended measures - the figure below shows that elites anticipate a severe vote loss after breaking with Trump, but not Biden. This was when Biden was a certain as Trump to become nominee (March 2024).

Leveraging textual responses from U.S. state legislators, challengers for Congress, and local officials (during the primaries in 2024) to scenarios with leader-opposing candidates, we compare the perceived narratives of breaking with Trump to Biden or Harris - which is seen as way less costly.

4/ BUT: Experiments show that the pro-democratic alliance risks wasting this benevolence by using an aggressive restoration strategy, pushing laws and norms to the limit. This strategy increases support for anti-democratic means even though the strategy isn't strictly undemocratic.

3/ Good news: Using panel data, we show that public opinion responded benevolently to the election. PiS voters spiked in anti-system but not outright anti-democratic attitudes, and if anything, such attitudes cooled off among voters of Tusk's pro-democratic coalition. "A way back" presented itself.

Thank you! Posting the link here
www.cambridge.org/core/journal...

There is a list of all undem behaviors across studies in the SI (screenshot below). Think you're right there; perhaps a specific one tied to partisanship? The ones below are violations of core principles more detached from that.

Worry that the findings are due to some idiosyncratic feature in the way I conduct experiments? I replicate the findings in eight related published studies with assignment of undemocratic behavior, measurement of age, and an evaluation of the undemocratic (or democratic) actor.

For example, young people punish anti-minority/nativist behavior more than older people. Comparing Figure 2 above and Figure C9 below, we see that young people punish both behavior types somewhat, whereas older people punish only undemocratic behavior.

I also illustrate the results plotting the regular linear model against a kernel. The negative effects are quite linear and increase substantially as age increases.

The main results are posted below. The negative effects of undemocratic behavior increase with age. And the pooled estimates provide quite precise estimates for that.

I employ data from five experimental studies covering 10 countries, 17 country-year samples, and almost 400k observations in total. That yields quite some statistical power gains. Studies include undemocratic candidate behavior (binary), age, and voting intentions (coded 0-1).

Delighted that my paper Young People Punish Undemocratic Behavior Less Than Older People is now conditionally accepted at @bjpols.bsky.social.

I show what the title suggests using a large amount of survey experimental evidence. Thread below.

Numbers are also high for expectations of harsh attacks on opponents (read: Tusk) or spreading conspiracies or lies. Completely unpredictable/erratic behavior (including severe anger) is less expected from Kaczyński. But importantly, numbers are generally lower for PiS voters.

I hand-coded a random sample (N = 435) of the entire sample; planning to code or use ML to classify the rest. Results are classified according to different democratic violations. About 50% expect a norm violation by Kaczynski if losing and 25% if winning (not visible).

I measured Polish voters' sentiment to each of the eight thinkable - some more than others - coalition possibilities (7-point scale) by party. The results suggest that PO, NL, and TD should not fear cooperating - their voters just want anything else than another PiS-government.