3 ms·
Measuring automobile deaths per population is highly uninformative, because of the massive difference in exposure over time. People were driving way more in 196
by mikeash 11y ago
Measuring automobile deaths per population is highly uninformative, because of the massive difference in exposure over time. People were driving way more in 1960 than 1937. If safety were constant, you'd expect the fatalities per population to also go way up. Fatalities per vehicle mile or passenger mile, while still not telling the whole story, is a much better measure.
- Someone1234 11y ago> Fatalities per vehicle mile or passenger mile, while still not telling the whole story, is a much better measure. And says the same thing, that safety has continued to improve dramatically since 1960. But fundamentally if population bound statistics are unfair in one direction, miles driven statistics are unfair in the opposite. I picked the statistic which actually makes it hardest to prove my point, and still proved my point. If I used miles driven I'd prove my point even harder.
- mikeash 11y agoI know. I am objecting to your reasoning, not your conclusion. Edit: wait, how are mileage based statistics unfair in the opposite direction? The inaccuracy with those comes from inconsistent exposure to danger (i.e. city roads versus divided highways versus narrow rural roads) and inconsistent numbers of people per vehicle. It's really hard to say which way that inaccuracy goes, only that it exists.