Updated January 12, 2010:
Many times in science, a model is constructed to analyze something. In this case, I decided to create a model of the average car to see the probability of death due to an accident and compare this model to the number of deaths found with the Smart Car (smart fortwo). Does the smart fortwo show a higher or lower death rate than the average car?
The government keeps track of the number of cars and the number of deaths, so generating a simplified model was easy.
According to:
FARS Encyclopedia
Looking at the data fro 2007, a total of 30,527 people died in their car in 2007. The number of registered cars in 2007 was 255,748,000. 255,748,000 Cars / 30,527 deaths = 8,378 (8377.76395) cars per death average in 2007. In a one month period, the number would be 12 times higher, 100,533 (100,533.167) cars per death average.
I first went with this figure, but later decided to use a more conservative approach. The first step was to count only the driver's deaths. The second step was to take into account the fact that the number of deaths per 100,000 cars seems to be dropping each year. While the data for 2009 has not been published yet, I used a linear interpretation to come up with an estimate of 1 death out of 10,000 cars per year. This would come to 1 death out of 120,000 per month.
By taking the number of Smart Cars sold each month and doing a running total for each month, I can calculate the probability (or fractional amount) of a death. The total for each month is divided by 120,000 and the fractional number represents the potential death for that month. Now, we all know there is no such a thing as a fractional death, but the number is more of an average estimate.
As seen from the chart below, the estimate says that at the start of this month the should have been around 4.3 deaths and assuming zero car sales, 4.6 deaths by the end of the month.
In all my internet and news searches, I can only find 3 Smart Car Deaths listed for the USA from January 2008 to today. Thus, the real world data says that the actual number of deaths is a bit more than 1/2 the average expected deaths.
This strongly says that in the real world, the Smart Car's safety systems do exactly as advertised; the car is safer than the average car.
On the other hand, if the real world data had shown 8 or 9 deaths, twice as many as expected, the logical conclusion would be that those who call the car a "Death Trap" would have support from the results.
The only know deaths in the USA to date are:
Union County, NC: July 2009; 81-year-old Donald Dixon ran a stop sign, hit by minivan
Smart car driver killed in Union County, NC - WIS News 10 - Columbia, South Carolina |
MIAMI, FL: March 28, 2009; 62-year old Raonel Tapanes collided with a Honda Civic; the Civic's driver also died in the crash.
Crash Victim's Family Shares Grief As 'Mini' Results Released - Automotive News Story - WPLG Miami
Arielle Krause, passed away July 2, 2009 in Houston, Texas after a tragic (Smart Car) accident.
http://www.hcnonline.com/articles/20...ccg1516701.txt
http://nationaltransportllc.com/blog...p/#comment-786
Even with more conservative numbers, the current count of 3 deaths is still less than the estimate of 4 or 5 deaths. The FARS data for 2009 does not come out until next year and the count of the current deaths is based on finding any news accounts of Smart Car deaths. (Something the news media loves to cover.) Once we have the updated data from FARS, we'll be able to see if my count was correct.
We do know from the FARS data that the number of Smart Car deaths for 2008 was ZERO.
When dealing with single digit results, there is some question as to if this is really normal or shifted due to the random nature of events. Thus, the result of 3 deaths is likely to be +/- 1. It's not impossible to be +/- 2, but it's less likely.
If the car is a "Death Trap" one could argue that in 2008, due to dumb luck we had zero deaths. However, as 2009 is in the final months, 2009 is showing a low number of deaths too. The Smart Car is just not showing the pattern of a "Death Trap".
With each month, I'll update the numbers. This message will show the updates.
While I have chosen the more conservative approach to estimating the average death rate, it is possible to conclude that the number should be higher. In another thread where people report their millage per year, the average is coming to around 15,000+ miles per year. This is at least 1.25x higher mileage than the national average of 12,000 miles per year. One could argue that increased miles equals increased risk. If I take that into account, 4.3 x 1.25 = 5.4 (5 or 6) expected deaths so far. The end of month number of 4.6 x 1.25 = 5.75
What if my linear interpretation was too conservative? The old system gave a number 1.19x higher. If I factor in the old system
and the increased miles, the risk increases by 1.49x or 4.3 x 1.49 = about 6.4 (6 or 7) deaths so far. The end of the month number would be 4.6 x 1.49 = 6.9
While I can't provide a perfect model, one thing is clear, the real number of deaths is far lower than what we would expect from the average car.
Bob Diaz
ADDITION: 11/7/2009
Special thanks to forestacademy for finding this information:
In looking through the main link:
http://www.iihs.org/research/hldi/fa...s/default.html
It appears that under #4
Quote:
|
* Personal injury protection coverage: comparison of losses by vehicle class and size/weight group, 2006-08 models
|
is the key link to look at (the link below).
http://www.iihs.org/research/hldi/fa...erage_0909.pdf
This link points to the key data we are interested in:
Quote:
|
Personal injury protection coverage insures against medical, hospital, and other expenses for injuries sustained in crashes to insured drivers and other people in their vehicles, regardless of who is at fault in the collision.
|
There is only one car that fits in the 2 door micro category in the IIHS's data, the Smart Car; better know as the smart fortwo. Thus any reference to a 2 door micro car has to be the smart.
Under "
Relative claim frequency" the Smart shows 82; where average car is 100 and lower numbers represent less claims. In comparison, a 2 door mini is 107 and a 2 door small is 147; both higher than normal. A 4 door mini is 184 and a 4 door small is 144.
Let's consider three possible explanations for the reductions in claims:
(1)
The ESP (Electronic Stability Program), Brake Assist, and Anti-Lock Brakes help to avoid the number of accidents: This is the most believable, because if you search through the IIHS web pages, their tests show that ESP helps drivers help to avoid accidents. If you avoid an accident, you don't have to submit a claim.
(2)
smart fortwo drivers are more careful drivers: I can't rule this out as a factor, but then again, I don't have any data to show that Smart drivers are more careful. So, this remains an unknown.
(3)
smart fortwo drivers drive less miles per year than other drivers: There's nothing to show this, but there is a simple non-scientific survey on this forum that suggests that smart drivers drive MORE miles per year than the average driver. Roughly 1.25 times more miles per year and increased miles equals increased risk.
Number 1 seems the most probable explanation, but I can't fully rule out number 2 playing a part in the reduction of claims. No matter what, the driver in the Smart fortwo is less likely to have accident than the driver in the average car.
In "
RELATIVE AVG LOSS PAYMENTS PER CLAIM", the smart was 94 and the average car was 100. One would expect that a car that was a "Death Trap" would have a higher level of injury than below average and greater injuries translate into higher medical costs. Clearly, the smart fortwo does NOT fit the pattern of unsafe and higher risk of injury in an accident.
Unfortunately this number is a bit distorted by the fact that larger cars tend to hold more people, resulting in higher costs when there are multiple people on board. However, the majotity of solo drivers is very high, roughly 75.5% (USA average to work). Thus, the impact of extra people in the car is reduced by the high number of solo drivers.
http://factfinder.census.gov/servlet...0&-CONTEXT=grt
The IIHS's data is consistent with my findings that the Smart Car is a safe car and NOT a "Death Trap".
Added January 12, 2010:
It will be a bit of a wait for the FARS Numbers to come out. The numbers will prove or disprove my system for estimating deaths before the FARS Data is published. What if 2 or 3 more deaths occur that I missed? That would show that my system has an under count problem, but add 2 or 3 to the 3 known deaths and the total falls within the expected 5 or 6 deaths for the average car. In order to be double the average expected, my system would have to miss 7 to 9 deaths.
Bob Diaz
The Beast
The Chick Magnet
Today
Linear Mode
