I’d like to see this data against ownership numbers. Like, what percentage of Dodge Ram owners have a DUI.
I don’t think that’s what this data is showing us. From what I see, it looks like, from a random set of 1000 DUI’s, these were the most common vehicles, and/or, from 1000 vehicle drivers, those that got DUIs were driving these cars most often… Sort of thing.
I want to know, how many were purchased, and how many individual drivers were cited for DUI while driving that vehicle.
I’m betting that all the very rare and expensive vehicles would have a huge percentage of DUI drivers. Someone driving around in a hurrican, probably has enough money to afford whatever ticket the law can throw at them, and there’s probably so few of them that actually obey the laws that a large portion have been caught DUI.
I like how you explained how the dataset was biased, then instead of describing how to eliminate/reduce that bias, you applied the same underlying tactic to make someone else look bad instead.
Oh, I’m just curious about alternatives. There’s a lot of ways to measure this, but only a few that are mostly unbiased.
The results will likely always be skewed one way or another, since this is about punishment, not the number of people committing the crime. It’s impossible to know how many people drive while intoxicated, we only know how many get caught doing it. The data is flawed from the start.
