Miata Mailing List: May 1998, Message #1124

sponsored by

This started out as a public post, but since the conversation on the list
has died down, I thought I'd just finish it privately. (I didn't get back
to it for a week while I was on the road.)

At 03:30 -0400 980502, AkiraRdstr wrote:

>Most advised to let the matter drop as the thread was becoming
>tiresome.

Interesting that the "This thread is tired" meta-thread seemed to generate
far more traffic than the original thread ever did. ;)

>One told me that I should make a quick statement of what I
>disagreed with and sign off ("say your piece and get out," is actually how he
>put it).  Guess whose advice I took.  ;-)

I'm glad you did. While I will openly admit that I've always found your
position on this issue rather difficult to understand, it's your opinion,
and you're entitled to it. In general though, I've always found these
discussions to be far more civil and intelligent than your average flame
war. (Is "intellectual flame war" an oyxmoron? ;)

>So after Zandr's answer to my counterpoint, rather than respond immediately, I
>thought I'd wait until the reviews came in.  They're in:

<reviews snipped>

I'll admit that the reviews were encouraging, especially in light of the
flaming of one individual who was so tired of deleting the same old thread
that he failed to notice that is was no longer the same old thread. That
said, I wasn't exactly fishing for fan mail. I just wanted to get some real
statistics out in the open.

>Did everyone visit the website and read the source material, or did everyone
>rely on Zandr's statements about the study?

Private comments indicated that those who care about the issue at all had
either read the study cover to cover, or had bookmarked it and intended to
do so.

>>Akira, I'm beginning to think that if you were killed by an airbag, you'd
>>deny that you were dead.
>
>All right, I probably deserved that.  I re-read my post and the tone did seem
>a bit peevish.  In the pursuit of brevity, I failed to convey my honest
>skepticism.  Let me back up and take another stab at it, this time without the
>chiding or cute song lyrics.

I'll admit that my response wasn't as level-headed as I'd have liked. Yes,
let's try this again.

>I believe you've drawn incorrect conclusions from the information in the
>report, as I'll explain.

I don't think were all that far apart on the conclusions, really. We don't
see eye-to-eye in a couple of places I'll note, but quite a bit of our
differences lie in the "is it good enough" domain.

>>Get past the executive summary and actually READ THE REPORT, AKIRA! >
>
>I did.  Zandr, YOU drew the conclusions you stated in your last post.

I tried not to present any conclusions that weren't in the report.
Apparently, I failed in that, as my opinions on the matter may well have
seeped in.

>I
>wanted to know what conclusions were presented by the study itself.  Besides,
>this excerpt is not from the executive summary but is from the body of the
>study.  The summary tells the findings of the study;  pulling isolated quotes
>from the work and adding one's own conclusions can skew the balance.

As you proceeded to do in other places, to refute my statements and support
yours. Yes indeed, the information from NHTSA is a very mixed bag; there's
something there for everyone. ;)

>>  Ten percent? Yes, sure, the 34% number is wonderful, but it's the best case
>>scenario for an airbag, and the best it can do is 34%.
>
>Perhaps the explanation of that figure from the NHTSA report will clear up
>your confusion:

I'm not confused in the least. Read on...

>" Belted drivers in cars equipped with air bags experienced a
>statistically significant 21 percent fatality reduction in purely
>frontal crashes, relative to belted drivers in comparable cars without
>air bags.

So, in the best case scenario for an airbag, there is a 21 percent
reduction in fatalities due to adding the airbag to the seatbelts.
Understand what we're saying here: If you smash into (or are smashed into
by) something (by their definition) within 15 degrees of head-on, the
airbag will increase your odds of survival by 20%. In isolation, that's an
admirable result. But there's no such thing as a free lunch. This comes at
a cost of, as clearly stated in the study, increased risk of some forms of
injury. This is also the vast minority of collisions. While the stats are
readily available on the percentage of collisions that result in
fatalities, the stats are not available on the percentage of collisions
that involve airbag-induced injury (at any level). This is partly because
it is very difficult to locate "equivalent" crashes

The report doesn't explicitly state the percentage of crashes that are
purely frontal. However, other reports from NHTSA do point out that the
number of fatal accidents are in the VAST minority. Of the 6.8 million
crashes in 1996, 2.2 million involved injury, but only 37,351 involved
fatality. In my opinion, this makes the fatality statistics for airbags FAR
less significant than the injury statistics. In addition, the report states
that the airbags are ineffective in reducing fatality in other than purely
frontal crashes. Another NHTSA report, "Traffic Safety Facts, 1996"
indicated that crashes in which the "front" (which, based on the phrasing
of the stats, would include partially frontal crashes)  is the intial point
of impact comprise 45% of all crashes, and 62% of fatal crashes.

>Unbelted drivers with air bags experienced a statistically
>significant 34 percent fatality reduction in purely frontal crashes,
>relative to unbelted drivers without air bags....

I don't put a lot of weight in the "vs. unbelted" stats, I've always worn
belts and that won't change.

>" The two preceding estimates need to be carefully interpreted. The 21
>percent reduction for the belted driver with an air bag is measured
>relative to the belted driver without an air bag; it does not include
>the very substantial effect of belts,  ****but represents the increment of
>air bags plus belts over belts alone.**** "
>
>Bottom line:  airbags have a measurable benefit in reducing fatalities.

Agreed. (Really!) The benefit is indeed measurable. But in all crashes,
that improvement is only 10%. The statement you've emphasized is actually
deceptive. That 21%, (like the 10% for all crashes) is _relative_to_ the
effect of seat belts alone.

>Correct---but this report also illustrates the substantial benefits of airbags
>in preventing fatalities.  Besides, at the end of your first post, you
>mentioned the 2---er, ONE airbag-related fatality of a Miata driver---in well
>over two million airbag deployments.  Might be why I strayed from the injury
>topic.

I guess I should have prefaced or separated that information more from my
post. I included it on a "I found this, too" basis, not as a critical point
in my argument. Apparently, I didn't make that clear at all.

>Perhaps I wasn't clear---I was referring to the NUMBER of injuries.  You did a
>great job noting the exhibits in the NHTSA Third Report to Congress, but this
>report says nothing about the NUMBER of injuries;

Errrm.
(Number of collisions) X (reduction of injury rate) = (Number of injuries
prevented), no?

While that report doesn't comment on the absolute number of injuries, it
does comment on the injury reduction offered by belts, airbags, etc. Unless
I'm grossly misreading the report, all it offers estimates of the reduction
in specific types of injury. In fact, claiming that airbags reduce the
severity of injury is only an inference based on the relative statistics on
reduction of injuries of a particular severity. (i.e. greater reduction of
severe injury than moderate injury implies that the severity is reduced by
the device.)

I can't see how you can say:

>My
>statement cannot be proven or disproven using the information in the report.

and still say:

>However, based on THIS report, airbags DO reduce the severity of injuries AND
>fatalities, over seatbelts alone.

By the way, the data doesn't really point to a reduction in the severity of
injury. It does point to a small reduction in the rate of moderate injury,
but the severe injury statistics are basically identical for bag+belt vs.
bag. Even if you were to accept the logic of my comments above, the stats
aren't there to support them.

>Well, let's see what the report says about the data from which you've drawn
>your conclusion:

[snip]

>****It should be
>emphasized that these analyses required further subsetting of the
>database, which considerably reduced the available number of cases for
>analysis. As a result, many of these estimates are not statistically
>significant.****

Actually, most of the data belted drivers and belted driver sitting in
front of airbags was indeed significant. There were a couple of holes, but
for the most part that data was deemed significant by the statisticians.
Most of the holes were in the airbag-only rows, with the exception of a
couple of numbers in the height and weight groups.

>Perhaps that's why the info seemed "squirrelly"---it wasn't meant to be
>definitive, and much of it is statistically insignificant.  How can one base
>valid conclusions on such evidence?

I don't agree that whether it was "meant to be definitive" is relavent. The
stats are what they are. As for significance, there are gaps, but the
trends are clear. Yes, the stats do get a little weird, and I agree that
the small sample size is a factor there. But I don't like even the
significant stats I see.

>Still with me?  Let's take a look at one more:
>
>>In support of my observation that, bottom line, airbags are statistically
>>proven to reduce injuries and fatalities.
>
>NO. They are not statistically proven to reduce injuries. I invite you to
>open your mind, cast aside your preconceived opinions, and re-read the
>NHTSA report.
>
>They do appear change the nature of the injuries, trading injuries to the
>extremities for head injury, and to some that may be an acceptable
>tradeoff. It isn't to me. >
>
>To the report once more:

[snipped section of the report reinforcing the tradeoff]

>Still don't mind trading a serious head injury for a serious extremity injury?
>Then consider:  if the damage is that serious, you can always >amputate< your
>arm.

Well, I've stated my opinion. (and that's what it is, an opinion) While I
can't  cite the statistics, it seems to me that I'm far more likely to be
in a crash where the bag goes off and breaks my hands (and being able to
use my hands is cruical to my livelihood) than a crash where the airbag
would prevent head injury. The only statistic I can easily cite to (weakly)
support this is the fact that roughly 30% of *reported* crashes involve
injury, but only 0.5% involve fatality.

BTW- The rate of fatality per vehicle mile has been declining steadly since
the 70's, but levelled off in 1990. There isn't the huge decline in the
post-airbag years that some would like you to believe. (Yes, I know there
are other factors, I don't put much weight in that statistic.)

>I could go on, but I believe I have made my point.  Referring people to a
>website to check information for themselves is terrific---if the debate has
>inspired several of the community to do that, then it's worthwhile.  But keep
>in mind that one often looks for (and finds) just the information that one
>wants when perusing a detailed scientific study.   Don't take someone else's
>word for what's there---don't accept their conclusions as gospel.  Go to the
>source and find out for yourself.

True enough. Even though I tried to present the statistics fairly, it's
obvious that my opinion on the matter leaked through.

Thanks for keeping this going, and keeping it civil, Akira. I do rather
enjoy this conversation, and consider is "debate among friends."

-Zandr

---------------------------------------------------------------------
Aleksandr Milewski                                              N6MOD
n6mod@amt.org                               http://www.amt.org/n6mod/