I don’t know why you are mentioning coldcar, he hasn’t even weighed in. I am still not sure I believe the old data is appropriate to the original article. The newer data may be appropriate, but certainly only in the non rural cases, and West/Dunn doesn’t describle the road data they were using. I wasn’t trying to arrive at the conclusion that slow drivers are at fault, per my quotes and post, but that the difference in speed is, and that according to the original article, his test car was forcing a difference in speed on the jersey turnpike. I don’t think more than 30mph between two cars is safe, but I also don’t believe many speed deltas between two cars lower than 30mph between two cars is safe in todays traffic situations. A large percent of cars tailgate (by traditional standards) in modern commutes, visibility is impaired by large vehicle size differentials, and a disrespect for lane courtesy/manners only increases the danger in having a speed difference between vehicles.
And it is your conclusion that is wrong, that “If the speed limit is 55 and the average speed is 65, then anyone between 50 and 80 is equally likely to be involved in an accident”. That is only supported by that one plot, (figure 3 I believe), and not the data. Delta’s on both ends of speed differentials of any speed result in an increased likely-hood of accidents, it is not a flat curve (the other 4 figures). And I fail to see how what I am doing is bad science, and if it is, how is what you are doing is any different? I think the crux of your and I’s problem is that we agree and don’t know it, that is if I am reading your double negative correctly.
But I do disagree with you on this: “the author tried to factor out the turning accidents which included vehicles moving very slowly. This skewed the data to show that going only a bit below the average speed caused accidents. That is a false conclusion.” How is slowing for a short exit ramp, disabled car, ear-to-phone idiot, or not building speed for an on-ramp any different than slowing for a cross traffic merge. Most people brake before exiting, and are not up to speed before merging. Same problem, different situation. And the data does show overwhelmingly that those slower vehicles are likely to be involved in accidents. If nothing, the author was wrong to exclude all of the data from those crashes, as many of those situations may still apply.
So, in summary, very high speed/deceleration kills but millions are willing to take that chance, speed delta’s around flow of traffic cause a greater amount of accidents both above and below the prevailing speed, coldcar hasn’t posted here, and 40 year old rural data is not a good representation of the New Jersey Turnpike. Can we agree on that?