Technology

Well, the problem is, even if I take the single case where this one guy exploded himself with his truck and compare it to the Pinto data, the poisson distribution difference will probably be statistically significant, yet the measure would be absolutely useless from a real-world perspective, because it has nothing to do with the vehicle's design.

I'd also argue that many of these events might not even be entirely occurring independently from each other (i.e., some of the key assumptions of Poisson are incorrect here) when people do all sorts of stupid shit with these rolling garbage cans like shooting at them, submerging them, etc. in a meme-like fashion for Tiktok views. So 4 events might very well be influenced by non-design-based, non-random human factors, which applied to other vehicles could generate similar results, and if the analysis were serious, they would have individually reviewed how these whopping 4 events happened, accounted for reporting bias towards EV fires (especially Tesla) and compared it to the F150 or the Ford Lightning as an analogous vehicle.

And I know the internet tends to conflate condescension with competence, but seriously, you should understand the above-listed things as a stats teacher.

edits for clarity

edit 2: also, in the times of the prussian army they did not have to account for stuff like people suddenly starting to pull the horses' tails for social media views.