"But an as-yet-unreleased NHTSA study of accidents in 12 states compares accident rates for some hybrid vehicles and their internal combustion engine counterparts."

How convenient. That way no one can debate the study and we'll all just have to assume that the study has merit. I'm confident that the nanny state will save us all from death yet through their indefatigable dedication to studies that show them what they already "know".

In that regard, Bob, maybe you could help them out bypreparing one of those graphs that shows how many lives are saved per federal government study? And then we can draw conclusions as to how many studies are needed to put an end to death entirely. Kind of sounds like something that would have appeared in Augustine's Laws back in the 80's; a simply priceless book (e.g., LAW NUMBER XI: If the Earth could be made to rotate twice as fast, managers would get twice as much done. If the Earth could be made to rotate twenty times as fast, everyone else would get twice as much done since all the managers would fly off. The graphs in the book are as good as the Laws and quotes.) I'm sure Dilbert would be proud.

Pardon my cynicism, but the feds just never stop do more good than the people can bear!

Greg

...And the growing problem of pedestrians bumping into each other will be solved by mandating the wearing of cowbells. The size and volume of cowbell will vary according to the body mass of the pedestrian and the risk they represent.

So let's add up what little has been made available:

• 8,000 hybrid electric vehicles - this is probably the count of 'injuries' reported in the 12 state set of data. Now 12 states is an interesting number because that corresponds to the traffic hospitalization and injury reporting states. As Christopher Hogan pointed out, this is a 'for fee' database versus the public domain FARS data.
• 600,000 gasoline-fueled cars - so we're looking at a total of 608,000 incidents counting both or 8/608 = 1.3%
• 50% more likely pedestrian accidents - so if they were equal, then an equal rate as far as numbers are concerned would be ~5,280 'hybrid electric vehicles' in the incidents or 8,000 - 5,280 = 2,720 potential incidents avoided out of 608,000.

So what would a skeptic of this unpublished report want to ask?

• Is the 50% based upon a count of vehicles in the fleet or one based upon miles per year per vehicle? - If they made a simple mistake of claiming as a percentage of fleet vehicles and not vehicle miles, then we would volunteer that "Quality Planning" claims hybrid owners drive 25% more. This means the per mile risk of a hybrid is less than 50% ... ~38%. But this is just a nibble and the professional statisticians would not make that mistake ... but an administrator might.
• How many of the 8,000 hybrid-electric accidents were fatal versus fatalities in the 600,000 gas-only accidents? - The reason is we already know from the FARs data that the fatalities per 100 million vehicle miles is statistically the same Prius to non-Prius vehicles. They have had to go into costly, non-fatality databases of 12 states to make this claim ... hummmm.
• Why is the NHTSA not addressing the 4,700 fatal pedestrian accidents? - When we ask about deaths, suddenly, we're not finding any concern by the NHTSA about pedestrian deaths. Is this report just eye-wash so they don't have to address the 4,700 pedestrian deaths that happen each year?
• Is the risk of a pedestrian accidents across the gas-only models uniform? - I suspect that looking at the rate of gas-only accidents by models will reveal that there are gas-only models equal to and even more hazardous than the different hybrid electric models. In fact, I suspect a mapping of gas-only models will reflect what Christopher Hogan discovered, pickup trucks and SUVs are the most hazardous vehicles to pedestrians ... because we see that in the FARS accident data.
• How was "the potential problem arises at speeds less than 15 mph" found? - This "15 mph" has my 'spider sense' tingling because last summer, Rossenblum, the NFB researcher reported in an early claim that it wasn't until a speed just over 25 mph that Prius and his selected reference vehicles had noise levels at the same level. Over the past year, I've seen this speed threshold decrease and 15 mph is the lowest I've seen yet. This suggests 'sand bagging.'

When I posted the note about the CBS report, it was because it follows a familiar pattern. One report is followed by a lot of "me too reports" and it is easy to think there is some sudden, new risk. In reality, it is just the 'echo machine' reverberating.

So the first thing is to "not panic" and have your facts and data ready. The truth stated simply, with dispassion is the most effective Malleus Maleficarum.

Bob Wilson

Whether you like it or not this is already over. The feds will pass some kind of "add noise" bill.

I agree with you but think it should be implemented as part of the rear light assembly. A piezoelectric electric buzzer that sounds whenever the backup lights are lit makes a lot of sense. It would fairly low cost to implement. But I've also looked through the back-over accident report.

I have a backup camera kit for our RV and I'm thinking about getting a couple for our Prius.

Bob Wilson

So ... you've SEEN my COW-orkers



Wait! 1,000s of pedestrians are being killed each year, from NON-hybrids ... so WHERE's THEIR noise, and it's needing to be LOUDER than their engines ... since the ICE's noise isn't keeping people from being run down.

But wait! What about noise pollution / abatement? We have to repeal all those laws, or we'll all be even more schizophrenic than we already are!

.

I'd like to see an NHTSA study on the probability of fatality when the vehicle that backs into the pedestrian is a Prius versus a Suburban or Hummer. With the Prius, you can turn around and look and see the pedestrian and if you are backing slowly, you will hear a "thud." But with a Suburban or Hummer, you don't see the pedestrian, and you can't hear the "thud." This is exactly why SUVs should be required to have backup beepers, like work trucks have.

I'm the guy who did the study of deaths and hospitalizations. I realized when I did it that it wasn't what needed to be done -- deaths were not the issue here. I only did it out of a sense of frustration that this was being sold to the Congress as a life-and-death issue, and the Congress was acting without any information whatsoever. So I tracked deaths and hospitalizations for blind pedestrians and found zero evidence to suggest that this was a life-and-death issue.

The real issue is low-speed collisions. That said, it sounds like researchers at NHTSA are now doing what needs to be done. They are studying accident reports involving low-speed collisions, pairing hybrid and non-hybrid versions of the same vehicle, using the only data they can get their hands on. Of course, I would like to see the details of the report, and of course any epidemiological study of this type is subject to uncertainty. But I'm a scientist, of sorts, and if the facts change, I'll change my mind.

If the best estimate is a 50% excess accident risk, and that is cheaply correctable, then the NHTSA is pretty much burdened to move forward, possibly subject to some reasonable cost-benefit analysis, possibly not. This is particularly true when the excess risk is not borne by the car driver, but by strangers. At that point, it's a public health issue, not a personal safety issue.

That said, I also believe that the absolute additional risk associated with this is trivial compared to other identifiable excess risks. It takes all of about 30 seconds to see scholarly studies showing that, per travel mile, SUVs have about a 50% excess risk of killing pedestrians, relative to cars, and that the largest of the SUVs on truck-type frames have a much higher risk of killing the occupants of other vehicles in the event of a crash. And, as noted, plenty of scholarly articles documenting the roughly two-fold excess risk of backover deaths from SUVs as opposed to cars, which might or might not also be relatively cheaply corrected.

So it would be rational to see some cost-benefit analysis of fixing this problem (which I see as primarily one of non-lethal injuries) versus legislation to address other identified excess risk issues in the vehicle fleet.

But I'm not holding my breath on that.

It's possible that the NHTSA administrators have some sense that this is worth doing, even relative to other known issues. But it's more possible that they are proceeding here because they can, because there's a warm-n-fuzzy advocacy group pushing this and because it affects foreign-badge vehicles far more than domestic.

That said, I'd still want to see an estimate of cost per injury avoided to see whether we are talking about $100 per broken bone or $1,000,000 per broken bone. I'd settle for knowing what the excess risk is in terms of hospital admissions, or even hospital ER visits. In other works, OK, they leaked news of 50% excess risk, for reasons that, as a former Fed, I don't quite grasp. But it really would be nice to see the details before the Congress votes on it.


Bob, are you going to ask them for a copy of the study? Understand, they know you'll give it a hostile review so that might be a little reluctant to allow you to review a draft. But if you get one, I'd like to see it, please.

EDIT: And there is at least one more thing that the research should have addressed. Full hybrids differ from their standard models in two ways: they are quieter at low speed, and they have a more actively distracting dashboard. Ideally, the research would split full hybrids (capable of moving on electricity only) from others, and would, in my opinion, also include the quietest non-hybrid vehicles, perhaps as measured by those European tests posted here last year.

In other words, we have a hypothesis that noise is the issue, but we have tested that with hybrid as the control variable. You would get a stronger study if a) you could include noisy and quiet hybrids (meaning, hybrids that require the engine to run in order to move, like the HCH), and you would demonstrate a dose-response relationship by including quiet non-hybrid cars (fully realizing that you lose your natural controls in having hybrid/nonhybrid pairs).

Let me state that more affirmatively: if the study finds the effect for the HCH, or even if the bulk of vehicles in the study are HCHs, then the study is bullshit, plain and simple, if it is interpreted as demonstrating a need for noisemakers on hybrids. If, by contrast, they find no effect for the HCH but do find an effect for other hybrids, then that strengthens the study.

So, Bob, that's the question to ask: How'd they treat the HCH? Since they didn't use the Prius (they said matched pairs of hybrid/non-hybrid vehicles), then, golly, I'd have thought that the HCH would constitute the bulk of the rest of the market. If the study leans on the HCH, it can't be interpreted as a study showing that lack of ICE noise is the issue. I'm guessing NHTSA staff were plenty smart enough to avoid the HCH for their main results, what I think I'm mostly asking if they were smart enough to include an HCH comparison as a control. So, here's the stronger statement: I not only want to be assured that they dropped the HCH from their main comparison showing that noise is the issue, I'd want them separately to demonstrate a null result (no excess risk) for the HCH, to rule out the possibility that some other aspect of hybrids (such as the distracting dash display) is at issue.

It's NOT just a 'low speed' noise issue. It's a "people not paying attention" issue. No matter HOW loud you make noise ... some people simply WILL NOT pay attention.

[ame=http://www.youtube.com/watch?v=qgf1yd-zwtY]YouTube - person being hit by a train[/ame]

.

My attitude is, let's see what the data say. There can be no doubt driver or pedestrian or both were being inattentive. But if inattention is the only issue, then there should be no reason to expect excess accident rates from the hybrid version of the vehicle, which is what they're testing. The question is whether inattention plus lack of noise adds up to more accidents than inattention alone.

Subtle effects like this are always hard to infer from non-experimental data. I mean, the rate of pedestrian accidents is pretty low, so you're talking some thin data. You're using "hybrid" as a proxy for "noise", so anything correlated with "hybrid", that you can't independently measure off the accident report form, may confound the results. Could be the distraction of the MFD, could be that individuals attracted to hybrids are systematically different from others. Could be they are driven in more urbanized areas (though I'm sure NHTSA would adjust for that.) So it's a tough slog to make this a credible study. I'd like to be sure that they are doing the slogging, and I'd like to see them apply the same method in situations where they expect no result (the HCH) to make sure that they aren't just looking at a bunch of false positives.

Thin indeed! Even their own 'lay report' cites only 8,000 instances versus 600,000 engine cases.

I note that "Quality Planning" in their July 15, 2009, "Hybrids: Is a Little of the Green Rubbing Off?" claimed:

The traffic citations number is interesting:

(.38-.23) /.23 = 65% higher citation risk
NHTSA reports 50% higher risk of pedestrian accidents

Are we generally more dangerous drivers? NOTE: "Quality Planning" went out of their way to emphasize comparison of hybrid and non-hybrid, same models. This means our Prius and the Honda Insights would have been excluded. In fact, it reduces the sample set to:

• Highlander and hybrid
• Escape and hybrid
• Camry and hybrid
• Civic and hybrid
• GM and hybrid-lites
• Saturn and hybrid-lites

Yet we know from sales figures tracking in GreenCarCongress that the Prius is ~2/3ds of all hybrids yet by this methodology, the elephant would not be looked at. ... hummmm.

I am data driven and can change my mind but it needs to be credible. So far, my eyes are open.

This weekend I plan to put my collected Prius and accident records up on a web page ... some place where we can share the numbers, the facts and data. When I get it up, I sure would appreciate any and all comments. With luck, we'll get something up that may be useful to cross check whatever is in the pending report.

I'll also go through the PriusChat discussions about "Quality Planning." If I remember correctly, someone had a source that broke out the distribution of traffic citations.

Bob Wilson

Bob, my recollection was they said they were going to do a regression-based analysis. So anything that can be identified from the accident reports (age, sex, time of day, urbanicity, etc.), that's going to get factored in, to see if there is an unexplained residual.

However, your point about the tickets is really well taken. By me anyway, because it suggests one more way they could test their findings. They need to take accidents where noise arguably was NOT a factor, and do the same regression analysis, and observe that they do NOT get a positive result for hybrids.

In other words, what they are going to do is take accidents where they think noise might have been a factor (low speed pedestrian collisions) and look for some residual, unexplained difference between hybrid and non-hybrid vehicles. If they find such a difference, they will attribute it to the quiet nature of hybrids.

They should also take some similar accident types where noise level should not be a problem -- presumably that would be low-speed vehicle-to-vehicle collisions, where the hybrid is the striking vehicle -- and test for a hybrid effect.

So, that's my second firm suggestion when you get to talk to NHTSA: Did they test their methodology for the potential for false positives by rerunning the pedestrian-accident regressions on vehicle-to-vehicle accidents? If not, it would strengthen the research if they could not only demonstrate a hybrid effect for pedestrian accidents, but demonstrate a lack of hybrid effect in vehicle-to-vehicle accidents.

But really, Bob, as these things go, if their initial results confirm the direction the Congress wants the policy to go, ... well, it's tough to get them to take that back by pointing to potential problems. And since, as you noted, we can't get the data (or at least not for free), there's no way to do the additional analysis ourselves.

I think we need to add this horn to our O-SO-QUIET-CARs
[ame=http://www.youtube.com/watch?v=24DFiLfzjfE]YouTube - Loud Air horn prank[/ame]