It's just an abstract that we're discussing here, I know, but the abstract absolutely, categorically does not say that. At all. You're guessing that's what it means.
> it is saying that speed differential is more dangerous than speed.
No it's not, the article is saying that a delta in speed is a factor in crash frequency. (And no, I didn't copy paste the whole abstract, because I was posting the part of the abstract germane to the quote I cited from the original article.)
> This is a horrible starting model. The severity of crashes is physics, but the frequency can't be looked at in isolation.
No, it's still a good starting model. If you read all the way to the end of that abstract, I think you'll see my point:
> Without exception, a vehicle that moved (much) faster than other traffic around it, had a higher crash rate. With regard to the rate of a (much) slower moving vehicle, the evidence is inconclusive.
So, much faster moving vehicles always had higher crash rates, while much slower vehicles... inconclusive.
There's a saying in statistics (George Box) that all models are wrong, but some models are useful. I think you'll find examples where speed isn't a problem, but those aren't the norm. If you're looking for a simple univariate model that goes a good long ways toward explaining vehicle safety, I'd challenge you to come up with a better one than what I've proposed here.
There was little evidence presented that justified the assumption that most people won't drive any faster with a raised speed limit, but if we take that as true, then its conclusion is consistent with the research you linked. It is NOT making the claim that driving faster is safer. In fact, it again clearly states that in an ideal world, everybody would just travel at slower and safer speeds, but that we can't effectively enforce such a low speed limit.
Emphasis mine. Your slower moving vehicle could relate to vehicles moving slower than other cars. Which is what gp claimed.
What about the Autobahn, me passing traffic with a difference in speed of 100km/h at times (it's entirely possible to reach higher values w/ faster cars, reckless driving, unusually slow traffic on the slow lane, like trucks).
I think it can work. Obviously it cannot work without education and experience, but pointing at speed differences as the one stat that leads to chaos and mayhem seems empirically wrong.
>Other studies looked at speed dispersion and found evidence that this is also an important factor in determining crash rate. Larger differences in speed between vehicles are related to a higher crash rate. Without exception, a vehicle that moved (much) faster than other traffic around it, had a higher crash rate.
Which is the whole point of this article. The article isn't saying that isn't dangerous, it is saying that speed differential is more dangerous than speed. It is saying that speed limits are generally ignored by most people (most people drive some combination of the speed of traffic and the speed they feel comfortable with), however the few people that DO listen to them cause accidents because it makes the differential speed much higher.
> A good starting mental model is that vehicle speeds increase both the frequency and severity of crashes.
This is a horrible starting model. The severity of crashes is physics, but the frequency can't be looked at in isolation. Even the article you cite hints at the fact that the relationship isn't that simple: In figure 1, rural roads with higher speed limits have a shallow linear relationship between crashes and speed, while urban roads with lower speed limits have a much steeper, exponential relationship between crashes and speed. Which counteracts a very simple "speed increases the frequency of accidents".