A death at the hands of police raises turnout in the commune where it happens. Does it also move the communes next door? Scroll through four decades of events and three urban units to see where the effect travels and where it does not.
Between 1980 and 2022, Basta! records 689 deaths in police interventions that can be placed in a commune of mainland France. Keep scrolling to run the clock.
210 of the 689 deaths (30%) happened in the Paris urban unit, which holds 17% of the population of mainland France.
INSEE's unité urbaine groups communes into one continuous built-up area. The Paris unit has 411 of them; 215 are in the analysis sample (shaded).
The question: if a death happens in one commune, does turnout move in the others?
When the death happens in Paris itself, turnout in the other communes of the urban unit rises by +1.08 points at the next election (95% CI 0.62 to 1.53), even though nothing happened there.
Exposure to a death in any other commune of the same urban unit shows no effect: −0.04 points (SE 0.16, all urban units pooled). What travels is a death in the centre.
Lyon is the second-largest urban unit. Outside Paris, a death in the central commune moves turnout in its neighbours by −0.10 points (95% CI −0.45 to 0.25). No detectable spillover.
Direct effects look similar inside and outside Paris. Spillover appears only when the death happens in Paris and the neighbour belongs to its urban unit. Adding a 20 km distance ring barely changes the centrality estimate (0.58 to 0.51), so this is not proximity alone.
One reading is that national media treat Paris events as national news. The design cannot separate this from everything else that makes Paris unique.