Even with the caveats about limited data and untangling causation and correlation, the statistics are striking: the first year of a scheme in Wales where the speed limit on urban roads was lowered to 20mph resulted in about 100 fewer people killed or seriously injured.
The 2023 figures (the latest ones and which had the 20mph change) also changed the way the statistics were gathered.
They state
(KSI is “Killed or seriously injured”. Crash is the new collection method)
100 people is about a 10% variance. To call that “striking” when there’s also a change in record keeping is bullshit. You need more data.