The excerpt above is from page 185 of Traffic: Why We Drive the Way We Do (and What It Says About Us) by Tom Vanderbilt.
It comes at the end of a chapter on risk perception, ie. roads that seem safer can be more dangerous than we think because they encourage us to drive more dangerously, while roads that seem dangerous can actually be safer than we think since they make us slow down and pay more attention. The dangerous-looking roads might still be more dangerous than the safe-looking roads in the absolute, but both of them might not be respectively as safe or dangerous as drivers tend to think…
Anyway, what annoyed me is the last sentence of the excerpt. I think it’s a good real-world example of misleading statistics.
While it might be literally true that most crashes “happen on dry road, on clear, sunny days, to sober drivers” (I wouldn’t swear to it, I haven’t seen the stats), it doesn’t take into account the difference in sample sizes. In most places, the roads are dry more often than not, and most days are sunny, and most drivers are sober.
These conditions might produce a higher total number of crashes, but what really matters is how many crashes they produce per driver. If you look at it this way, it’s probably pretty obvious that wet roads, at night, with drunk drivers cause a lot more crashes.
See also: Rationality Articles