Hacker News Reader

It’s percentage of owners. What other normalization are you looking for?

This study is not measuring the number of DUI per car model. It's measuring the amount of drivers that have had a DUI in their life, per car model.

So this study could start by normalising against the age of drivers, because the older the drivers, the more likely they've have had a DUI strike in their life.

It is normalised.

“For each model, the share of drivers with a prior DUI violation was compared against the total population of motorists that owned that model. The 10 models with the highest proportion of drivers with a DUI were then selected”

The key word here is "prior".

A car released in 2022 that is popular with old people will have more drivers that have had a DUI violation than a car released in 2022 that is popular with 16 year olds, because the older folks would have had more chance to get DUI in their life.

> The key word here is "prior".

Yeah, it would be better if that keyword was in the title.

It is nowhere near a list of the most popular cars in the US.

It is absolutely normalized

I wish I lived in a country where WRXs were that popular!

It's not saying that the WRX is popular. It's saying that, of people that own a WRX, a disproportionately high percentage of them have had a DUI.