I didn't think it would affect the actual race data, because hey, that was still just 22 drivers.
Dutch GP race data itself:
Russell was the most scrutinized compared to expected, followed by Hulkenberg and Lindblad, with Stroll, Ocon, Bottas and Bearman the least scrutinized.
What I don't understand is why Verstappen isn't on the under-scrutinized list given he was out before the end of the first lap, before Bearman who is on that list.
Most peculiar.
Any differences weren't statistically significant, with a χ² of 1.38.
The thing I thought the Hadjar swap out-Tsunoda swap in would affect was the season-to-date ranking. But what I forgot was that, because it's observed vs expected, Tsunoda will still get a value for the one (or more depending on Hadjar's wrist) grand prix he participated in.
The thing it changes is the χ² critical value, because Tsunoda's appearance increases the degrees of freedom in the data from 21 to 22. The critical χ² value goes up from 32.671 to 33.924.
The season to date chart looks like this:
At the bottom, it's still the cars that don't finish a lot, with Stroll and the two Cadillacs down there.
Tsunoda's car has been tested exactly as much as expected after 1 race, and so has Piastri's after 12 races.
The χ² value continues to go up, it's now 4.03, still nowhere near significant but rising. I think the cars being tested less than expected are the stronger contributors because their standardised residuals have larger values.
We also didn't have a random check of a top 10 finisher, so the evidence is really piling up that that is a truly random check that only occurs sometimes and the early-season pattern was just a quirk of the numbers.
