Europe's Autumn of Unrest: How to Test the Claims With Data

In mid-October 2026, mass demonstrations over living costs, housing and public services turned violent in parts of France, Italy and Albania, while housing protests continued in Spain and Portugal. Explanations arrived quickly, and they conflict. This article takes no side. It shows how to test the kinds of statistical claims that circulate in such debates.

What has been reported

According to the Associated Press, rising fuel costs, unhappiness with schools, declining public services and housing pressures lie behind an autumn of protests. The AP reported that the weekend's unrest in France led to 2,400 arrests even though hundreds of thousands marched peacefully, and that Milan police arrested 271 people, most of them minors. Interpretations diverge. President Macron accused far-left activists; right-wing politicians, including US President Donald Trump, portrayed immigrants as responsible, while the AP noted that French authorities have not blamed immigrants. AllSides found that left- and right-leaning outlets split on the motives.

Counts are not rates

2,400 arrests sounds large or small depending on the denominator. If, hypothetically, 300,000 people had marched, 2,400 arrests would be 0.8% of them; at 600,000 it would be 0.4% (the marcher figures here are illustrative, not reported data). Arrests also measure policing, not only behavior: they depend on where officers were deployed and whom they detained. Always ask who is in the data, and what the total is.

A worked example with two regions

Suppose an analyst collected 200 protest events in each of two regions and found clashes at 40 events in region A and 20 in region B (invented numbers for illustration). The rates are 20% against 10%. The difference of two proportions calculator gives z ≈ 2.80 and p ≈ 0.005. Entered as a 2×2 table, the chi-square test of independence gives χ² ≈ 7.84, the same result, and the odds ratio calculator gives 2.25 with a 95% interval of about 1.26 to 4.01. The gap is unlikely to be chance alone. But Cramér's V is only about 0.14: a statistically reliable but weak association that says nothing about why.

Significance tells you the gap is probably not chance. It does not tell you the cause.

Correlation across regions

A popular claim links fuel-price rises to protest activity. With one figure per region for each, the Spearman rank correlation calculator checks whether regions with bigger price increases also rank higher on protest counts. Even a strong result would describe regions, not individuals (the ecological fallacy), and factors such as population size, unemployment or differences in reporting could drive both numbers.

Claims about groups need extra care

Some commentators attribute disorder to particular groups, others to political organizers. Whatever the claim, a fair test needs the group's rate among all participants, a comparison group, and reliable, consistently collected data on both. Arrest records alone usually cannot supply that. Where such data exist, the same tools apply: compare proportions, report an odds ratio with its interval, and state how strong the association is, not only whether it is significant.

Where to find data

Eurostat publishes comparable EU statistics on prices, housing costs and employment, national statistics offices publish local figures, and ACLED records protest and riot events with dates and locations. Check each source's definitions before comparing countries.