Yup, there's a lawsuit trying to get the raw data behind the 40 percent number [1]. The original report [2] gives two averages, the crash rate per million miles driven before/after autosteer, 1.3 and 0.8. It doesn't say how many miles were driven, how many crashes, or when any of these occurred, so it's impossible to verify that result.

Further, the same NHTSA report notes that AEB reduces rear-end collisions by 40 percent. Supposedly, AEB and autosteer were enabled around the same time.

[1] http://www.safetyresearch.net/blog/articles/quality-control-...

[2] https://static.nhtsa.gov/odi/inv/2016/INCLA-PE16007-7876.PDF

I would argue that the people who can afford a Tesla are rich people and thus likely well educated and are safer drivers.

Poor people with bad cars, no education and unsafe driving practices probably have more accidents.

I don't think these stats mean much.

I keep seeing statements to this effect, and they are incorrect. The comparison is between Tesla cars with and without the AutoPilot feature installed.

The statistic has nothing to do with people not driving a Tesla.

“...airbag deployment crashes in the subject Tesla vehicles before and after Autosteer installation. The data show that the Tesla vehicles crash rate dropped by almost 40 percent after Autosteer installation.”

The bigger problem is that their study seems to have revealed that while crashes went down, the severity of the crashes went up. People weren't dying in their Teslas before Autosteer, but now we're looking at 4 deaths and counting in the space of 2 years, all directly attributable to the use of Autopilot.

Okay lets move the goalposts.

> People weren’t dying in their Tesla’s before Autosteer

This is false. Although Tesla fatalities tend to be a bit over the top like driving off cliffs, or stolen car escape gone wrong type of situations, there have certainly been non-Autopilot related fatalities in Teslas.

> but now we're looking at 4 deaths and counting in the space of 2 years

This is false. There have been 2 fatalities in Tesla’s which are known to have been operating with Autopilot at the time. There was also one fatality in China where the family claims autopilot was on at the time but Wikipedia states this was not confirmed.

> all directly attributable to the use of Autopilot.

I take issue with “directly attributable” simply based on NTSB (a neutral 3rd party if ever there was one) disagreeing with that. They found Autopilot was not at fault in the Florida crash. I think a neutral fact-finder would defer to the NTSB and rank your claim as false.

But finally, you disregard the obvious flaw in your logic, which is you can’t possibly know how many potentially fatal crashes or maimings have been avoided due to AutoPilot, except the one thing we do know is that airbags in Teslas are deploying 40% less with the feature than without.

It's not as bad as that. It looks their methods could take that into account, they just don't detail anything. I copied the analysis from the report into my original comment, so it's clearer what they did.

There's still no proper control, because people driving habits (and many other things) change with time and a before/after analysis has selection bias with time. Maybe they get used to the car as time goes on and crash less, for example. A nice way to get around this would be to look at people who didn't get autopilot and see how the autopilot reduction compares to the non-autopilot reduction (difference in differences). Only then can you start to say it may be due to autopilot, and it's still a fuzzy statement.

This was a before/after comparison, so it's like vs. like.