RV Crash Analysis - Safety Statistics

RV Roamer said:

I think some of the data assumptions are arguable.

Click to expand...

Gary,

I agree, which is why I pointed out a couple of potential data problems.  They only relate, however, to the normalization factor, not to the absolute fatality statistics.  Those can be read with a good deal of certainty given the absolution nature of the reporting system.

I am interested if anyone has another option for mileage data given the scarcity of it at the federal level?

Also of note, the reported average mileage for a motorhome in the NHTS was 4238 miles per year.  This is low relative to the value that FMCA reports of their membership, which is 9000 per year according to their survey.  FMCA does not release their survey so it is impossible to determine how their numbers correlate to the Bureau of Transportation Statistics, but you could argue that the average mileage should be higher than 4238.  This would lower the overall rate of fatality in the motorhome class compared to other classes.  Unfortunately, without a better/transparent source for that information, this is the only defensible number that can be derived.

More later this evening on accident causation, driver response, vehicle age, trailer presence, etc. :-X

About the only data that's "Assumable" in those figures are miles driven.. that is an estimate

However most everything else is somebody going 1,2,3,4,5 as it were... However even there there is an issue with Non compliance

Police agencies are required to investigate ALL fatal accidents and to report all fatalities to the NHTSC.

They are required to report on alcohol use, blood alcohol level, seat belt usage, air bag deployment and a host of other things

(I used to have to type the prelim reports)

However some agencies are less than compliant in making those reports.

The RVIA estimates that a typical RV (all types, not just motorhomes) travels about 6000 miles/year, but I suspect that is somebody's conjecture.

About the only data that's "Assumable" in those figures are miles driven.. that is an estimate

Click to expand...

Yes, but there are a couple big assumptions in there related to mileage and mileage is a critical component of the calculation. It's the divisor in the equation and you don't have to tweak the divisor much to substantially alter the calculated fatality index

RV Roamer said:

Yes, but there are a couple big assumptions in there related to mileage and mileage is a critical component of the calculation. It's the divisor in the equation and you don't have to tweak the divisor much to substantially alter the calculated fatality index

Click to expand...

Gary,

I am comfortable that the denominator derived from the National Household Transportation Survey data is in fact accurate for the survey year.

The 2001 survey had 139,000 responses.  Of those responses 958 reported driving or owning a motorhome or "motorized" RV.  The NHTS corrects for certain demographic factors in the response pool and generates an "escalation factor", which is just a multiplier used to normalize the number of survey responses for the entire U.S. population (see table below).  When that was done, the estimated number of motorhomes in the U.S. comes out to 1,400,452 in 2001. The RVIA reports total RV (non-motorized included) ownership in the U.S. in 2001 at around 7,000,000.  RVIA Sales reports show that the typical ratio is 20-25% motorhomes sold to 75-80% non-motorized RV's (rough average).  This ratio supports the NHTS number since 20-25% of 7,000,000 would be 1.4 to 1.75 Million, consistent with the reported NHTS data.

Each household in the survey also reported the number of miles driven by each vehicle.  Motorhomes were reported as being driven an average of 4,268 miles per year, and that includes a large number of responses that listed 0 miles.

Taking the product of the average miles driven and the number of motorhomes results in a total number of miles driven by motorhomes in the U.S. in 2001 of 5,977,129,136, which is the denominator in our fatality rate equation.

Given the strength of the NHTS data, I am highly confident of the 2001 mileage information.  Where the soundness of the data becomes less certain is when you extrapolate out for years 2000 and 2002-2006.  I assumed that motorhome miles driven would correlate with the number of miles driven for large, single unit, heavy trucks with 6 wheels or more since that is a category tracked by the Bureau of Transportation Statistics for each year.  I believe that assumption is reasonable, but you could argue that linking motorhome miles to that category would UNDER-report miles driven since motorhome have become more popular in the last 5 years.  Regardless, the impact on the result will be very small (less than 5% at most).

Anyway, sorry for the long post, I just want to be transparent in the assumptions.  I have significant revisions to discuss in my next post since I discovered some problems with the actual fatality listings last night.  More soon.

Christian

Ok,

Sorry for the retraction above.  I ran into a significant problem with the FARS data and needed to refresh all of the analysis with the revised figures.  I had been using data from the "PERSON" file, which lists every victim separately.  The FARS Coding Information Guidebook gives specific guidance that if there is an entry in the database, there was a fatality.  This turns out NOT to be true.  I have since discovered that all persons in a crash where "A" fatality occurred will be listed, regardless of whether they themselves died.  I reanalyzed the data using only persons with a listed "date of death" (macabre I know, but no way around it).  I cross referenced this against the "VEHICLE" files for each year 200-2007.  The VEHICLE file lists only the vehicles involved in crashes and how many individuals perished in the listed vehicle.  I now have an exact match across the two data files AND a third confirmation using the FARS summary reports for total fatalities in a given year.  Lastly, I had my sister, who has an MS in statistics review my work this time, and she agreed with the revised findings.

So what does that mean? ???

It means the data I submitted yesterday grossly overestimated the rate of fatalities :-\ I sincerely apologize for that.  The following constitutes a revised and verified summary findings:

Summary:
The revised FARS analysis shows that a total of 212 individuals perished in motorhome accidents in the years 2000-2007 for an average of just over 26 fatalities per year.  This represents an average rate of fatality of 0.44 per 100 Million Vehicle Miles vs. 1.48 for all vehicles in the United States, or roughly one third the average rate of all motor vehicles.  (See Chart "Fatality Mileage Normalization Chart")

Observations:
1) The "Initial Harmful Event" which is the event deemed to have caused the crash, was overwhelmingly due to striking another vehicle in "your" roadway (45%).  That can mean a vehicle traveling the same direction of a divided highway, or a vehicle traveling in either direction of an undivided road.  Vehicle Roll-over, Striking a Guard Rail, and Striking a Tree, each represented approximately 9% of total crashes respectively. (See Chart "Initial Harmful Event")
2) The majority of fatalities occurred in the front seats of the motorhome, with 80% being either the driver or passenger.  Of the 26 rear compartment fatalities, only 2 persons died while using a restraint (seat belt).  No children using child safety seats died during the analysis period. (See Chart "Fatalities by Seating Position and Restraint Usage).
3) Alcohol did not appear to be a significant contributor to motorhome fatalities. (Less than 1% reported drinking as a factor)
4) The majority of fatalities occurred on rural interstates and/or major rural highways (54%). (See Chart "Road Type")
5) Trailers of any type were only reported in 26 of 212 fatalities during the study period.  Of those, only 3 were reported as towed vehicles, however that statistic was only added to the database in 2005 and is therefore statistically irrelevant. (See Chart "Fatality by Reported Trailer Use")

Conclusion:
It is difficult to come to any supportable conclusions about accident causation, however, what it is clear that motorhomes are statistically very safe relative to the overall vehicular population.  What limited fatal crashes do occur appear to be largely caused by striking other vehicles and fixed objects near the roadway and the resulting fatalities seem to most often occur in the front seat.  Given that there were only 26 reported deaths of individuals in the rear area and only 2 of those were belted, one can draw their own conclusion about the merits of using a belt.  This author draws solace in the fact that not a single child in a child safety seat perished during the study period.

I will continue to review the information and attempt to draw out any possible nuance in the information that the readers here might find useful, but I am comfortable myself in the knowledge that the activity is statistically quite safe.  I am also pleased to conclude that some exaggerated media claims of motorhomes being "death traps" that I have read are simply and astoundingly false.

Christian,

If you ever get around to doing a survey of motorhomers on this RV Forum, it would be interesting to know the average number of miles driven per year by our members.  I should think that we have a large enough population that you could get some statistically significant numbers with which to extrapolate what's needed.

ArdraF

RE: Most fatalities are in the front seat

I think most motor home miles are driven by a a husband/wife or person/significant other team. (you can take that several ways and all of them are right).  Many RVers are retirees and do not travel with "kids" in the ride

This would load the percentages toward front seat, since that's all that's occupied in many, if not most,cases

Also, traditionally the right front seat is called the "Suicide seat"in the US. (And other left hand drive countries)

As the driver, in my rig, I am fairly protected against "Stuff" that comes flying up to the front if I stand on the brakes or hit sometihng  Wife, is not so protected

What protects me....... Main slide out

I really have problems with 'Average' statistics. To me they don't mean anything. 10 fulltimers will skew the results of 1000 part-timers to a great degree. So what does it prove.

  Maybe one should go back in time and use the "bell curve".

Just an idle thought... A 'median' approach may be better.

What do I know..I ain't no statistician..

carson  Fl


 

carson said:

I really have problems with 'Average' statistics. To me they don't mean anything. 10 fulltimers will skew the results of 1000 part-timers to a great degree. So what does it prove.

   Maybe one should go back in time and use the "bell curve".

Just an idle thought... A 'median' approach may be better.

What do I know..I ain't no statistician..

carson  Fl



 

Click to expand...

I agree to a certain extent.  There is no such thing as an average RV'er.  Everyone has discrete risk factors that will increase or decrease their risk parameters.  The exercise was intended to determine the probabilistic occurrence of death across a large spectrum to determine a "relevant" level of risk as compared to a national average.  Feel free to discount that if you like.

Using a median only makes sense when you have a large data set are are interested in what's in the middle.  It has little relevance to discrete event statistics with multiple outcome potentials.  There also isn't a large enough data set of fatalities for median to be relevant to anything other than perhaps annual miles driven. 

If there are any specific parameters you are interested in, I can try and isolate those and get back to you, however.  Just let me know.

Christian

To the O/P.  in another place, and in other threads here, the issue of braking systems on towed vehicles has come up time and time again

Would it be possible to glean an idea of crashes where there was no towed braking system, and where such a system might have made a difference? (Or better yet, WOULD have)

Thanks in advance

John In Detroit said:

RE: Most fatalities are in the front seat

I think most motor home miles are driven by a a husband/wife or person/significant other team. (you can take that several ways and all of them are right).   Many RVers are retirees and do not travel with "kids" in the ride

This would load the percentages toward front seat, since that's all that's occupied in many, if not most,cases

Click to expand...

Yup, no doubt that the rate of occupancy up front is much higher which would naturally lead to a higher fatality occurance, which is why no real conclusions can be drawn from the number.

Also, I had meant to post the age distibution, but neglected to do so in my latest post.  The modal fatality age was 65 (that one is for you Carson ), with a clear concentration around the 60's, which would support the retiree's theory.  See Chart "Fatalities by Age" below.

John In Detroit said:

To the O/P.  in another place, and in other threads here, the issue of braking systems on towed vehicles has come up time and time again

Would it be possible to glean an idea of crashes where there was no towed braking system, and where such a system might have made a difference? (Or better yet, WOULD have)

Thanks in advance

Click to expand...

John in Detroit,

Unfortunately, no data was collected on braking systems of towed vehicles.  There were only 22 cases out of 212 where trailers where present.  I'll see how the cause and driver response correlated to those incidents and get back to you with anything I find.

Christian

Nice work, Christian.  Thanks!  ...  Steve

Christian, I will give you my thinking

IN a panic stop situtation with all wheels locked the formula for calculating skid distance has but two variables
Road condition (Coefficient of friction)  and speed.  The first (CoF) is constant for all vehicles on a given road at a given time, so if you and the car in front of you hit the brakes at EXACTLY the same time,  You will stop the same distance apart you started.  Size is not a factor in that calculation

But if the towed is not breaking now you have say 22,000 pounds times CoF trying to stop 26,000 pounds times speed squared

Now that's a lot more math... LOTS more.. And it is a longer skid too  Causes what I call (in cars) Leaky radiator syndrom

(That's when your radiator springs a leak as you stop)

Now in DP.ss Well,  it is often the Generator's radiator that leaks.... If that is it has one

Where the soundness of the data becomes less certain is when you extrapolate out for years 2000 and 2002-2006.

Click to expand...

Those are the assumptions I was alluding to. I don't have a better way to extrapolate/estimate mileage , but it's not much more than a SWAG in my opinion.  [IMO] There is no good reason to believe that motorhome miles correlate with heavy truck miles any more than with passenger car miles. We know that RV ownership increased and we know that miles driven increased in those years, but don't have any data to indicate the amounts.

I'm not trying to pick apart your work - you did a remarkable job of finding some real data and massaging it into something we can understand.   I merely caution people to avoid leaping to any major conclusions based on it.  Your subsequent revision based on a clearer understanding of the actual data shows what a seemingly small change can do. Sometimes a detail can change the whole picture.

RV Roamer said:

Those are the assumptions I was alluding to. I don't have a better way to extrapolate/estimate mileage , but it's not much more than a SWAG in my opinion.  [IMO] There is no good reason to believe that motorhome miles correlate with heavy truck miles any more than with passenger car miles. We know that RV ownership increased and we know that miles driven increased in those years, but don't have any data to indicate the amounts.

I'm not trying to pick apart your work - you did a remarkable job of finding some real data and massaging it into something we can understand.   I merely caution people to avoid leaping to any major conclusions based on it.  Your subsequent revision based on a clearer understanding of the actual data shows what a seemingly small change can do. Sometimes a detail can change the whole picture.

Click to expand...

Fair enough.  If anything, the extrapolations provided are conservative and would "overestimate" the actual fatality rate for motorhomes since we all seem to agree that the actual mileage driven has likely increased faster  for motorhomes than for the rest of the population.  I would rather be conservative and err on the side of overstating the risk than vice versa.

Hopefully someday, NHTSA, RVIA, or someone will invest some money into a more comprehensive study so we can further validate the indicative research we have thus far.  :-\

Maybe this is why it is difficult to get good data: Rv'ers are a convoluted group of people, from Pup trailers to mega-yachts fortresses  motorhomes. Convoluted data creates strange data.
  I enjoy this discussion.
carson FL

John From Detroit said:

RE: Most fatalities are in the front seat

I think most motor home miles are driven by a a husband/wife or person/significant other team. (you can take that several ways and all of them are right).  Many RVers are retirees and do not travel with "kids" in the ride

This would load the percentages toward front seat, since that's all that's occupied in many, if not most,cases

Click to expand...

What I think would be interesting and relevant to the car seat question that referred me here, although probably not reported in the raw data, would be how many fatal accidents had people in the back seats (whether killed or not), and of those how many back seat passengers did perish.  In other words, is the percentage of fatalities among back seat passengers in crashes higher, lower, or the same as among front seat riders.

-- ML

twokids said:

What I think would be interesting and relevant to the car seat question that referred me here, although probably not reported in the raw data, would be how many fatal accidents had people in the back seats (whether killed or not), and of those how many back seat passengers did perish.  In other words, is the percentage of fatalities among back seat passengers in crashes higher, lower, or the same as among front seat riders.

-- ML

Click to expand...

Unfortunately, the FARS data base, which is derived from crash reports filled out by responding officers, does not collect data on number of occupants in the crash and their position in the vehicle.  It reports only the fatality data and where the victim was seated and their restraint usage.

The only data I can offer is (as you can see from the above chart), over 160 of the 211 reported fatalities were in the front two seats.  This doesn't offer you any normalization for occupancy, but I strongly believe the rear of the vehicle is safer due to the proximity risk for where crash incursions generally occur.