Re: Bad driving [Was: Re: Cupholders & Americans] [long]

[Elliott Potter <epotter@abraxis.com>]

Orin Eman wrote:
> 
> > (nn percent of crashes involved the driver at fault being in age group
> > xx-yy)
> > Age   Percent
> > 16-24 25%
> > 25-34 24%
> > 35-44 21%
> > 45-54 14%
> 
> > Wait, what's that?  There's not much of a difference until you hit
> > 45...the oh-so-mature 25-34 segment just barely edges out the
> > serial-killer-on-wheels teenagers.  That's 1%...and we're only 4% over
> > the 35-44 wise-and-tired group.
> 
> You are being just as guilty of trying to use statistics to prove
> your point.  Those numbers are meaningless without the number of
> miles driven by the age group in question.  Ie, if the 16-24 age group
> is driving half as many miles as the others, think again about
> your conclusion.  You don't have enough data here...
 
First off, let's remember that this is a *rate*, and a rate doesn't
illustrate how many checks the insurance company writes.  For actual
numbers read on.

Second, I'm just presenting another view of the same numbers; I exactly
which columns they use, but all of the ones I see agree pretty well -
drivers under 25 do get in slightly more accidents than drivers 25-35,
but only slightly fewer, and the real drop doesn't come until drivers
hit 45 or 55.  If you're interested in actual numbers of crashes,
instead of rates (this might be a little more relevant, as it's a better
indicator of how many checks an insurance company might have to write),
I can post those numbers as well, but again they agree with the rates
here - for fatal, injury, and property damage, young drivers get in 1%,
3%, and 4% more than drivers 25-34.  Still not a whole lot, even when
you're talking about 4% of 7,982,000 crashes involving only property
damage...just around 300,000 more.  Not a lot when you consider how many
insured drivers they're bilking to pay for it (around 9,126,000 licensed
drivers *age 19 and under* in the US in 1995).

Having said that, I can't find any information regarding miles traveled
vs. age (checked NHTSA, NPTS, US Census, USDOT, and Statistical Abstract
of the United States).
But your point here is valid; without knowing how many miles were
traveled, it's hard to make a good assumption.  FWIW, the closest I
found was from NPTS, "number of vehicle trips by driver's age and
purpose," for October 1997; the age were grouped by 10's but starting
20-29, 30-39, etc.  In that table the 30-39 age group drove about 8
million more miles, about 12% more.

> Furthermore, if you are trying to relate this to insurance rates,
> then you have to look at the severity of the crash and the insurance
> payout.  Find total insurance payout due to driver at fault by age group
> and if that shows the same trend as above, you would have a point.

This included 3 ranges of severity; crashes involving a fatality,
crashes involving only injuries, and crashes involving only property
damage.  There's going to be a range of costs associated with each one,
and it would be beyond the scope of this argument to try to break it
down even more ... besides, if you look at the data, there are definite
trends going on, so it's probably a safe assumption that on average,
insurance payouts in these 3 groupings will be fairly similar in the age
groups.  

Here's the breakdown looking at the 3 ranges of crashes:

Drivers in fatal crashes:
Age	Percent
16-24	24%
25-34	22%
35-44	19%
45-54	13%

Drivers in injury crashes:
Age	Percent
16-24	27%
25-34	24%
35-44	20%
45-54	13%

Drivers in property-damage-only crashes:
Age	Percent
16-24	27%
25-34	23%
35-44	20%
45-54	15%

So the biggest discrepancy here, 4%, occurs in the (generally) least
expensive category, one where there are no hospital bills or expensive
lawsuits - on the low end you would have little Tommy, who backed into
Ms Helm's mailbox, and on the high end, perhaps little Joey hit a
telephone pole or drove through a storefront.  Generally,
property-damage-only accidents tend to be fairly inexpensive compared to
other types.

> > But wait, let's look at some more measures of "maturity."  How 'bout
> > alcohol involvement [2]?
> 
> > (Of all crashes in age group, what percent of them involved alcohol).
> 
> > Age   Alcohol inv.
> > 16-24 27.2%
> > 25-34 31.1%
> > 35-44 27.3%
> > 45-54 18.9%
> 
> > "All of those drunk teenagers, with no respect for the law, endangering
> > the lives of others."  Oh, wait.....
> 
> Look at that 16-24 another way.  Only 21-24 are actually allowed to
> drink... how do you think that 27% would split between 16-20 and 21-24?
> 50% for each?  Mostly 21-24?  It wouldn't look too good in the latter
> case would it... 27% for a 4 year range vs 31 for a 10 range.
 

Let's not forget two things here.  First of all, only age 21-24 are
*legally* allowed to drink, but kids of all ages drink anyway.  Also,
we're looking at a *rate* here, not a total number of cases.  The rates
broken down are:

Age	Alcohol inv.
16-20	35.0%
21-24	21.5%
25-34	31.1%
35-44	27.3%
45-54	18.9%

The thing to note with this statistic is that the numbers are reported
by police, and since many states have a "zero tolorance" law for minors,
or give them a much lower BAC threshold, a crash involving a BAC that
wouldn't normally be considered a factor will be considered one when
dealing with a minor, hence the seemingly high number in the 16-20
range.  (this is from fatal crashes, which is where the percentages came
from...injury-only and property-only are similar).  Using numbers might
be a little more useful here for ages 21-24, alcohol involvement shows
1,650 accidents, from a total of 7,670 ... for ages 25-34, alcohol
involvement shows 3,852 accidents from a total of 12,378.  So by the
numbers, insurance companies are still writing about 1/3 as many big
checks for younger drivers.

> "lies, damn lies, and statistics..."  Be careful with pulling out
> numbers like this.  They can and do get twisted to show the
> opposite of what you intended...
 
I had that thought in mind the whole time.  The problem is, the
statistics are always shown one way, but not shown the other way - I'm
presenting the other side here.  Which side is "right?"  Well, probably
neither.  My argument is that the stereotype that exists here is created
by people who are trying to make reasons for young people to pay more -
of course they're going to massage the data so that their argument looks
better.  Well, now I'm massaging the data the other way, and when I do
I'm met with a bunch of self-righteous 30 year olds (not you Orin; most
of the really self-righteous replies were private) telling me to shut
up, bend over, and take it like a good little boy because I'm definitely
wrong.  I still think my argument stands; there's not enough difference
there to warrant a sizeable difference in insurance rates between 24 and
26 year olds, or 20 year olds and 30 year olds.

> Orin.

--
Elliott