Thanks for following along, and we look forward to hearing your thoughts on the paper.

... question, but because their support for the war is weak enough to be affected by the way the question is asked on the list experiment.

For example, a respondent might say they support the war when asked directly, but if asked in a list experiment alongside other issues they feel more strongly about (e.g., an increase in pensions or gay marriage) then they may withhold their support -- not because they were lying on the direct ...

In essence, the threshold effect (which is related to but distinct from a traditional design effect) is the idea that the *control* items included in a list experiment format induce comparisons that can influence the threshold for a respondent to indicate support for a particular issue/policy.

Applying our estimator to data from Russia, we find that the share of respondents who are outright lying is fairly small, but a large fraction of people who say they support the war when asked directly actually have quite weak preferences that are vulnerable to what we call the "threshold effect."

In the paper, we argue that list experiments as they're traditionally conducted aren't very good at distinguishing preference falsification (e.g., lying about war support) from *weak* preferences that can shift depending on how survey questions are asked. We propose a way to separate the two.

Does this mean that Russians are increasingly misrepresenting their attitudes about the war on surveys, and that secret opposition has been increasing over time?

... But, at the same time, estimates of war support derived from list experiments (a common tool for identifying preference falsification on surveys) suggest a growing gap between nominal support for the "SMO" and support measured indirectly.

Broadly, the paper asks to what extent we can trust data on sensitive political issues that come from surveys conducted in authoritarian regimes. More specifically, we wanted to understand the nature of Russian public support for the war against Ukraine, and whether we can use surveys to measure it.

The figures in this post are sobering!

❗️ One person killed and others injured after Russian strike on National Academy of Sciences building in Kyiv NV journalist showed the aftermath of the attack in central Kyiv. Read our live updates for everything known about Russia’s Sept. 28 attack on Kyiv 👇 🔗https://pm9dfw.s.gy/4sgJ3K