atlaspl.pt
via @pcmagalhaes.bsky.social

Oh, was there in July. Wonderful.

🤔 Do surveys exaggerate democratic support due to social desirability bias (SDB)?

➡️ Using survey-mode variation & list experiments in 24 countries, @pcmagalhaes.bsky.social & @aarslew.bsky.social find no evidence that SDB inflates democratic attitudes www.cambridge.org/core/journal... #FirstView

PS- This is the product of a visit by @aarslew.bsky.social to @ics-ulisboa.bsky.social. It was a super fun project to work on, and I'm ever so grateful to him. Also to @ess-survey.bsky.social central team — and the UK and FI teams — for help with data access and to the PT team for help with Cronos.

All this suggests we should probably stop assuming respondents are insincere when answering questions about "democracy" and, instead, continue caring deeply about the many other challenges involved in measuring these attitudes. (end)

Finally, in a list experiment in Portugal, we find, like @dielea.bsky.social and Kiewiet de Jonge before us, that there are no relevant differences to write home about. (7)

Second, in two countries where two similar samples were simultaneously interviewed using different modes, self-completion is associated, at most, with both lower democratic AND autocratic support, suggesting that mode effects here have no connection with social desirability. (6)

First, in a battery of questions about democracy and how it is conceived in the
@ess-survey.bsky.social Democracy module, modules, we show that the move from face-to-face to self-completion caused by the pandemic in several countries is not associated with discernible changes in responses. (5)

Here, @aarslew.bsky.social and I use the well-demonstrated fact that, in issues where people have incentives to withhold socially undesirable behaviours or opinions, the absence of an interviewer facilitates their expression. (4) www.cambridge.org/core/journal...

However, evidence of this “social desirability bias” is surprisingly scarce. For example, list experiments — allowing people to respond indirectly to interviewers— showed very small and/or insignificant differences between direct & indirect questioning about democracy. (3)

The concern with people's untruthfulness about democracy is so prevalent that it has almost become an assumption, justifying alternative measurement strategies and even entirely new research designs, as seen in Inglehart & Welzel and Svolik, among others. (2)

Long ago, in a study of the democratic attitudes in Germany, Dalton raised the possibility of “Fragebogendemokraten” (questionnaire democrats): people who hesitate to express their sincere dislike for “democracy” in surveys, providing instead the “socially desirable” response (1)

4/6 🧵Using survey experiment, Magalhães et al (2025) in @ispp-pops.bsky.social show voters (in 🇵🇹) support weaker due process rights for corruption suspect from opposing party vs their own party. Out-group derogation prevails in-group favoritism in corruption evaluation. https://doi.org/p4qt

With David Pimenta, a piece about the latest legislative elections in Portugal. www.maspoderlocal.com/index.php/mp...

Perhaps one case where external validity issues are less concerning? ;-)

We already knew how party preferences bias electoral/vertical accountability for alleged corruption. Now we know that, in horizontal accountability, we must always protect people’s due process rights but maybe not so much of those nasty politicians I dislike. (End)

These and other findings in an article just out in @ispp-pops.bsky.social with Nuno Garoupa, Luís de Sousa and Rui Costa doi.org/10.1111/pops... (5)

The main result is easy to describe: the importance of the rights of investigated politicians depends on our relationship with their party. It’s very important to protect those rights when we like the party, less so when we don’t. (4)

However, the scenarios we presented varied randomly. In some cases, the politician under investigation was the leader of the respondent’s preferred party, in others just a regular party member, and in others, the same roles but from the parties the subjects liked the least. (3)

Then we asked them to tell us how important the right to privacy, to a good name and reputation, and to the presumption of innocence should be in that investigation (using more concrete and accessible language). (2)

In a survey experiment, we asked a sample of Portuguese voters to imagine that a politician was being investigated for a corruption case—money in exchange for favoritism in a public tender. (1)

Hi Ben. This is going to take a long time, would you say? How fast is vote counting normally?