Hacker News Reader

Comment by tshaddox | Hacker News Reader

I'd love to see studies that attempt to control for attributes related to "driver skill" or "driver responsibility" that you could objectively test yourself on, so that you could more accurately predict your own risk and compare it against the risk of alternatives like public transit. It's not really good enough to say "I don't drive drunk, and I'm pretty sure I'm a more skilled driver than average, therefore I can discard all studies on the risks of injury of driving versus public transit."

View on HN

I don't think you can completely discard all comparisons, but people are intuitively right to recognize that if they don't engage in risky behavior (like drunk driving) they're less at risk of being in an accident. To argue otherwise is basically saying driving drunk is no more risky than driving sober.

> but people are intuitively right to recognize that if they don't engage in risky behavior (like drunk driving) they're less at risk of being in an accident. To argue otherwise is basically saying driving drunk is no more risky than driving sober.

Sure, that’s why it’s important to know how risky automobiles are if you discard the cases of drunk drivers injuring themselves. I suspect drunk drivers injuring themselves accounts for a very small portion of automobile injuries, but we need to see the data.

It’s not enough to say “I don’t drive drunk, and driving drunk is very dangerous, therefore automobile risk estimates don’t apply to me.” You could make the same argument about driving blindfolded.

A quick google says that 26.8% of drivers killed or severely injured in a car accident had alcohol in their system. Not drinking and driving significantly reduces your risk.

That’s “drivers (excluding motorcyclists) seriously injured or killed in crashes and whose blood was collected at one of the par- ticipating trauma centers or by MEs” in 2020. There were 474 such drivers, 127 of which had alcohol in their system. But there were 38,000 automobile fatalities in 2020.

The fact that we rarely if ever see such data sets is a strong data point by itself.

Insurance cost (data is widely available) goes down with age and you can extrapolate from that to safety but it only falls fast initially and the rest is largely a reflection of how driving habits change with age.

Occasionally someone like Tesla or Volvo will trot out some cherry picked data that boils down to "yes the very safe demographics who buy our very safe cars die less than the peasants".

Other than those two there's not much.

Can you? How about controlling for mileage driven? Older drivers are e.g. more likely to be retired and drive substantially fewer miles due to not commuting.

Reply on news.ycombinator.com