I took the time to read the study, the whole thing. It's 9 years old
Yes, it is. That, however, does not necessarily invalidate it.
Not necessarily. As previously mentioned, automakers are using the increased efficiency of LEDs to make tail lights (stop lights, turn signals, headlamps) smaller and smaller for styling reasons. As a result, photometric performance is not improving, and in a significant number of cases it's actually decreasing.
What is it you imagine has improved in the nine years since that study was done?
and the study admits limitations to the data analysis due to lacking the ideal samples (mid year changeover).
Well, yes. The study doesn't say LED stop lights aren't better, it says no such conclusion was reachable given the data set studied.
Am sure more data and further studies will come out.
I'm not so sure. Incandescent stop lights, as a class, are disappearing from new car designs (and facelifts/refreshes, etc). And -- again -- manufacturers are aggressively restyling lights so they are fast moving away from the sizes and shapes of traditional (incandescent) lights. Seems to me like every day that passes is that much harder to compile a data pool of incandescent versus LED stop lights that are otherwise closely similar. That 2013 study might very well have been the last, best shot at that kind of study.
UMTRI did another study just last year, which, given the difficulty in finding "all other factors equal" vehicles to compare crash data for, is probably the best technique to find answers to this question. They modified cars with LED stop lights so that the stop lights behaved either normally (instant on/off) or like incandescents (250 ms rise/drop time). The stop light behavior was randomized beyond the driver's knowledge or control, and the instrumented vehicles (data loggers, rear-looking cameras and radars) were driven at the posted speed limit (= people gonna tailgate) in actual traffic for six weeks. Result: no benefit to instant-on versus incandescent-type rise time.
As far as the ramping down to off... incan allows for a more gradual dimming... giving an indication that something's changing.
Last year UMTRI did quality research on sequential turn signals of various types; they compared the effective conspicuity (speed and accuracy of grokking the message) of sequential-on, sequential-off, LED and simulated-incandescent conventional on/off, and two kinds of "moving spot" turn signals (whole lamp lights up, and a brighter spot moves across the width of the lamp). Results: the instant on/off LED turn signals were better than the fade-on/fade-off incandescent signals; the sequential-off signals were better than the static LEDs, the sequential-on signals were better than the sequential-offs, and the moving-spot signals were better than the sequential-ons.
That, plus the stop light study with instrumented cars in traffic, pretty well puts to bed the idea that ramping down the stop lights is helpful (or harmful).
Leds just 'slam the door'... and if you miss/blink in the instant they turn off, you've missed the cue.
I don't think so; there's a car in front of you without its stop lights lit -- that's what you need to know; it's less important (maybe not important at all) to know whether the driver took his foot off the brake half a second ago or half an hour ago.
I used to think some fade-time on LED turn signals might be a good idea (instant-on, fade-to-off), for pretty much exactly the same reason you describe: increase the effective conspicuity of the turn signal by increasing the likelihood of a momentary glance being enough to detect there's a turn signal operating. Then along came Mazda's CX-30 doing exactly that (see video above), and yuck, its turn signals seem a whole lot
less conspicuous because of the fade-to-off dilutes the state-change. It looks a lot less like a turn signal, and a lot more like a steady-burning amber light that intermittently briefly flickers off. I think the problem is too long a fade-to-off time. If it were cut in half, to be more like an incandescent bulb, I think it would probably be fine. But as per the recent UMTRI research, it likely wouldn't help anything versus an instant-on/instant-off LED turn signal.