13/05/2026
New paper. Most people in Europe think there are far more immigrants in their country than there actually are. That belief, accurate or not, shapes how people vote and what policies they support — so it matters who tends to get it wrong, and why.
Researchers have been studying this for twenty years and have converged on a familiar list of suspects: education, political views, personal contact with immigrants, what people watch and read. Almost all of these studies pick the variables to look at in advance.
In this paper I do something different: I let a machine-learning method (Lasso) pick from 35 candidates, and see which ones survive.
Most of them do — misperception isn't reducible to one cause. But three less-discussed factors stand out: depressive symptoms, low social trust, and how much television people watch (specifically TV in general, not political news). Each predicts believing the immigrant share is larger than it is.
Working paper:https://www.ucd.ie/economics/t4media/WP2026_09.pdf
University College Dublin UCD College of Social Sciences and Law Kevin Denny