Elon repeatedly tries to misdirect from the issue at hand. Teslas do not catch fire more often than the average U.S. car. But that is not in question. The question is, "after a collision, what happens?" And so far, the answer is, "the Model S has a 25x increased chance of catching fire." That is the focus of the investigation. Fires post-collision are a bad thing, even if the never reach the occupant: they can ignite other cars gas tanks (or other objects), and impede rescue work.
It's not reasonable to say that the Model S has a 25x increased change of catching fire. The sample size is orders of magnitude too small (there were 3 instances so far?).
Put simply, there is no statistically significant difference whatsoever between the Model S and the broader population in terms of fires post-collision, and Musk is understandably frustrated about the bogus press claiming there is.
Engineering lets you model things without having to do experiments. It's possible to know that a car design is more likely to catch fire just by analyzing it. You don't have to wait for thousands of cars to catch fire.
Where did you get the quote which you use to answer your question? It was not written in the article linked here, and I can't seem to find any similar quotes when searching except for a comment in a slashdot thread.
I would like to see how they define collision and what exactly is 25x more likely but am finding myself unable to learn more about what exactly was considered with that calculation.
Using the binomial model, you can at least be pretty sure that the Model S has a higher chance of catching fire after a collision than your average car. We don't have enough data to determine the degree, though. So claiming 25x is premature.
