> The data is suspect at best and should require a third party evaluation. And the NHTSA doesn't count as a third party? > I'm speculating that the reason for

> The data is suspect at best and should require a third party evaluation. And the NHTSA doesn't count as a third party? > I'm speculating that the reason for that 40% safety improvement is more a result of the difference between where autopilot can and cannot be used. To say that tesla owners without autopilot tend to drive in situations with 40% more likelihood of accidents than those with autopilot, sounds like a massive stretch without data of your own.