> measuring crash rates by “number per vehicle miles traveled” and instead by “number per vehicle hours traveled.” there's another issue, which is plane risk

> measuring crash rates by “number per vehicle miles traveled” and instead by “number per vehicle hours traveled.” there's another issue, which is plane risk is uneven during their journey, so a type that does half the flights but twice the distance on average will end up almost 1.5 times as safe all other things being equal. this source claim 45% crashes happening on takeoff and landing which are a minimal percentage of travel if you measure risk by mile or hour: https://eu.usatoday.com/story/travel/columnist/cox/2013/08/26/ask-the-captain-danger-takeoffs-landings/2693891/ https://eu.usatoday.com/story/travel/columnist/cox/2013/08/2... a way to normalize that would have to use "number per flights" but then you'd get trains, with their multi-stop journeys, skewing the numbers the other way.