| Member |
 Discussion Topic  |
|
Axiom2000
 Moderator
968 Posts
[Mentor]
Georgetown, Delaware
USA
BMW
R1200RT/ R1200C
|
 Posted - 10/15/2009 : 5:19 AM
|
Recently Released.
http://www.fhwa.dot.gov/pressroom/fhwa0928.htm
I understand the decision to go ahead with this research has been made. Unfortunately, I also understand this research will go forward with a limited sample size of about 300 crashes versus the original 900 called for. That is disappointing.
Apparently, the cost of this study came in between 8 and 9 million dollars (Whew!) which far exceeded the 2 million pledged by the government and the 3 million put up by the motorcycle industry. I am certainly not any kind of expert on research of this nature but it seems to me the sample size will greatly impact the validity of the results. Further I understand the motorcycle industry decided not to participate with it 3 million pledged funding due to the small sample size. So the way I am reading this, is they can do the research using 300 crashes for 2 million dollars, my math tells me then 900 would cost 6 million. So my first question is, where does the 8 to 9 million comes from then?
Further, If I got this right the original study came in between 8 and 9 million. Between the 2 million in tax dollars and the 3 million put up by the industry we came up 3 million short. Pardon me if I seem cynical about this but I doubt the 3 million more needed would hardly be missed given we are spending trillions of unaccounted for dollars to bail out mismanaged and corrupt financial firms. Seems to me 3 million spent to perhaps reduce the number people killed (5300) or injured (96000)in 2008 while riding is a better use of fund than the funding with tax payer dollars for corporate bonuses.
Enough I guess, just not happy about any of this, I had high hopes that people in responsible positions were going to do the right thing. Naive I am I guess.
|
|
ananga73
Ex-Member
|
Posted - 10/15/2009 : 2:13 PM
|
I feel your pain however I just wanted to also chip in that the relation between data points and complexity of analysis is not a linear one. Meaning one cannot say if using 300 data points costs $2 Million, 900 data points should cost in the region of $6 Million = $2M*900/300. Analysis becomes exponentially more complex with added data points.
I agree more is better however I still hope they can draw some meaningful insights from the sample size they have chosen.
Regards |
 |
|
|
DataDan
Standard Member
198 Posts
[Mentor]
Central Coast, CA
USA
Honda
|
Posted - 10/15/2009 : 8:36 PM
|
FYI, Hurt did his study of 900 crashes (investigations started in 1975, results were published in 1981) for a half-million. It's worth noting that the manuscript (400 pages and at least 100 tables) was produced by typewriter, and the analysis was done on computers that would be recognized as such only by us deteriorating relics of that bygone era. Post-investigation costs were undoubtedly a big part of the budget.
My impression from reading about this new study for the past few years (a half-dozen or so archived articles) is that the guy to whom it was assigned at OSU really didn't want to do it; it was a pork project courtesy of some OK congresscritter, not something they were pursuing. Further, I get the idea that he's making a meal out of it by highballing the cost estimates. If you're saddled with a project you don't want, at least get a lot of money to spread around the lab. Thus, the projected cost--corrected for inflation--more than 4 times that of Hurt's study.
I don't see cost of analysis changing much, whether it's 300 cases or 900. The time it takes for Excel to compute the median of a column of data is insignificant. The major cost would be in the investigations. Hurt recruited his team, all with motorcycling experience, and trained them for 6 months. Then, for more than a year, they investigated crashes on-site throughout LA County at all hours, reconstructing them, interviewing witnesses and involved parties, and following up. I'd guess Hurt benefited from the use of slave labor devoted graduate students at USC. |
|
|
|
gymnast
Moderator
2805 Posts
[Mentor]
Meridian, Idaho
USA
Harley-Davidson
Sportster Sport
|
Posted - 10/15/2009 : 9:15 PM
|
DataDan, that is a good analysis of the sum and substance of the whole "OSU fiasco" which has been suggested by some to be an intentional attempt to delay any meaningful large scale study of motorcycle safety in the US for the foreseeable future. Your analysis is short, to the point, and unfortunately, pretty well describes the status quo.
I was over at Goldiron this morning where I found he reported on the status of the NHTSA Study and also had a citation of an MSF press release. See www.goldiron.wordpress.com |
|
|
|
Axiom2000
Moderator
968 Posts
[Mentor]
Georgetown, Delaware
USA
BMW
R1200RT/ R1200C
|
Posted - 10/16/2009 : 5:46 PM
|
| I say the heck with them all. My sense is there is enough experitse among the member of this site to do the job and do it for a lot less than $8,000,000. I will contribute the first $5.00, who's in? |
|
|
|
scottrnelson
Advanced Member
5247 Posts
[Mentor]
Pleasanton, CA
USA
KTM
990 Adv, ST2, XR650L
|
Posted - 10/16/2009 : 6:54 PM
|
quote:
Originally posted by Axiom2000
I say the heck with them all. My sense is there is enough expertise among the member of this site to do the job and do it for a lot less than $8,000,000.
I don't know. I would be very interested in taking six months or a year to go study motorcycle accidents, but I can't afford to take a pay cut to do it. I'm pretty expensive.
For a study like that, an area like the San Francisco Bay Area would actually be pretty good, because we often get posts on the local forums about motorcycle accidents within minutes of when they happen. Except that a study here would be heavily biased in the direction of riding too fast for conditions on the twisty mountain roads, whereas a study in Texas or Kansas would have a very different bias for the cause of the accidents. |
|
|
|
rayg50
Advanced Member
851 Posts
[Mentor]
NYC, NY
USA
Honda
Shadow Spirit 750DC
|
Posted - 10/16/2009 : 8:03 PM
|
quote:
Originally posted by scottrnelson
[quote]Originally posted by Axiom2000
..Except that a study here would be heavily biased in ... whereas a study in Texas or Kansas would have a very different bias ...
+1 excellent point.
Seems to me that it might make some sense to seek out individuals across the country with what was considered to be an appropriate background. They could train those individuals in gathering the needed information at the crash site, photographing or video taping the site, conducting follow up interviews, surveys, victim history etc. These individuals would need to be willing to be available "as needed". LEOs would need to know that they must be contacted immediately, given full cooperation and access to whatever local enforcement gathered and determined. In many instances a local "accident investigation" conducted by LEOs would provide enough time for the individual to get to the scene if they are local.
Were this possible the regional bias would become a valid point to study. If the gatherer of the information were paid on a per accident basis you would get people that were not in it for the money. Spread across the 50 states a study of 300 would come out to 6 per state. Spread that out over the year to be able to get seasonal weather into the study and that is 1 ever other month. I think any number of people could squeeze that into their full time job schedule and if you had a few in each state it would be even less of a drain.
If the information gatherers were paid 1k per accident that is only 300k out of the budget. The analysis could be done centrally by full time researchers / analysts. There are a lot of holes in this scenario but some of it could be plugged by using retired accident investigators.
Sorry for the mind drift.
|
|
|
|
Axiom2000
Moderator
968 Posts
[Mentor]
Georgetown, Delaware
USA
BMW
R1200RT/ R1200C
|
Posted - 10/16/2009 : 8:56 PM
|
quote:
I don't know. I would be very interested in taking six months or a year to go study motorcycle accidents, but I can't afford to take a pay cut to do it. I'm pretty expensive.
For a study like that, an area like the San Francisco Bay Area would actually be pretty good, because we often get posts on the local forums about motorcycle accidents within minutes of when they happen. Except that a study here would be heavily biased in the direction of riding too fast for conditions on the twisty mountain roads, whereas a study in Texas or Kansas would have a very different bias for the cause of the accidents.
quote:
+1 excellent point.
Seems to me that it might make some sense to seek out individuals across the country with what was considered to be an appropriate background. They could train those individuals in gathering the needed information at the crash site, photographing or video taping the site, conducting follow up interviews, surveys, victim history etc. These individuals would need to be willing to be available "as needed". LEOs would need to know that they must be contacted immediately, given full cooperation and access to whatever local enforcement gathered and determined. In many instances a local "accident investigation" conducted by LEOs would provide enough time for the individual to get to the scene if they are local.
Were this possible the regional bias would become a valid point to study. If the gatherer of the information were paid on a per accident basis you would get people that were not in it for the money. Spread across the 50 states a study of 300 would come out to 6 per state. Spread that out over the year to be able to get seasonal weather into the study and that is 1 ever other month. I think any number of people could squeeze that into their full time job schedule and if you had a few in each state it would be even less of a drain.
If the information gatherers were paid 1k per accident that is only 300k out of the budget. The analysis could be done centrally by full time researchers / analysts. There are a lot of holes in this scenario but some of it could be plugged by using retired accident investigators.
Well there ya go we got ideas flowing already. Although my comment was made mostly in jest there is a part of me that believes it is possible.
|
|
|
|
RickRussellTX
Standard Member
155 Posts
Hawthorne, CA
USA
Honda
CN250
|
Posted - 10/16/2009 : 10:21 PM
|
"Analysis becomes exponentially more complex with added data points."
I would say the opposite is true -- more data usually means that less advanced forms of analysis are required to identify a statistically significant signal. Small data sets require techniques that are not well-mastered by many scientists and engineers, and the results are less reliable (or in statistical parlance, "the confidence interval is larger").
Ultimately it depends on how much effort they want to spend on accident reconstruction, post-crash interviews, post-crash skill tests, etc. I think that there is a legitimate argument to be made that we might learn a lot more from 300 thoroughly-studied accidents than 900 less-studied accidents. |
|
|
|
RickRussellTX
Standard Member
155 Posts
Hawthorne, CA
USA
Honda
CN250
|
Posted - 10/16/2009 : 10:28 PM
|
quote:
Originally posted by rayg50
Seems to me that it might make some sense to seek out individuals across the country with what was considered to be an appropriate background.
I like this idea -- there would be difficulties (e.g., the guy in a particular region who skips all the non-serious accidents) that would need to be accounted for. There would need to be some way to make sure the accidents were truly randomly selected from the region.
But even if it was volunteer work (in fact ESPECIALLY if it was volunteer work), I could probably convince my employer that I had to duck out for a couple of hours once or twice per week.
RR
|
|
|
|
gymnast
Moderator
2805 Posts
[Mentor]
Meridian, Idaho
USA
Harley-Davidson
Sportster Sport
|
Posted - 10/16/2009 : 11:41 PM
|
| Some day when I have a year or so I would like to write a book on how the opportunities to conduct research on motorcycle crashes, crash prevention and safety/training countermeasures in the US has been largely (and intentionally?) squandered for the past 35 years. Motorcycle safety research with few exceptions, notably the Hurt study, some work by Oulett (sp?) and a few others who worked primarily in the area of head injury countermeasures has been notably (and shamefully) lacking or actively discouraged in my opinion. |
|
|
galileo
Ex-Member
|
Posted - 10/17/2009 : 6:04 AM
|
In my opinion, any crash research that doesn't look at those who don't crash is flawed as it doesn't get at the source of the crashes. It's also very unlikely that a study that only looks at crashes will find anything significantly different than the Hurt study. The numbers may change some but the causes will be about the same.
The biggest problem with the Hurt study is the denominator of the causes was not known. N% of the riders who crashed not having licenses may simply mean N% of the riders who rode didn't have licenses.
Recently Virginia Tech did a study on 100 drivers using a data recorder for all the drivers over a one year period. Out of this 100 drivers there were 67 crashes. (They did some preselection to eliminate very safe drivers but that number is still astoundingly high.)
Each driver was paid $125 a month for the study. So, 1,200 apiece. A 1,000 drivers would be about 1.25 million plus analysis costs. The biggest thing a person can do to reduce crashes is to drive in a way so braking forces are kept under .3g's.
They have outfitted their first motorcycle for a study and I expect more to come.
Here is a link to a summary of the study.
|
|
|
|
Axiom2000
Moderator
968 Posts
[Mentor]
Georgetown, Delaware
USA
BMW
R1200RT/ R1200C
|
Posted - 10/17/2009 : 6:32 AM
|
quote:
Recently Virginia Tech did a study on 100 drivers using a data recorder for all the drivers over a one year period. Out of this 100 drivers there were 67 crashes. (They did some preselection to eliminate very safe drivers but that number is still astoundingly high.)
Normally I would think when a description like "astoundingly high" were used to describe a reaction to a surprising result it would grab your attention. In this case, it seems inadequate. That many results just seem unreal to me. Unless of course they really picked the cream of the crop of accident prone, reckless drivers. I did not read the entire study so I may be all wet but yep that number is astoundingly high |
|
|
galileo
Ex-Member
|
Posted - 10/17/2009 : 7:06 AM
|
quote:
Originally posted by Axiom2000
quote:
Recently Virginia Tech did a study on 100 drivers using a data recorder for all the drivers over a one year period. Out of this 100 drivers there were 67 crashes. (They did some preselection to eliminate very safe drivers but that number is still astoundingly high.)
Normally I would think when a description like "astoundingly high" were used to describe a reaction to a surprising result it would grab your attention. In this case, it seems inadequate. That many results just seem unreal to me. Unless of course they really picked the cream of the crop of accident prone, reckless drivers. I did not read the entire study so I may be all wet but yep that number is astoundingly high
The word "Astoundingly" was my description. There were about 67 crashes, but some had several crashes. Some were eliminated from the study for crashing too often. The definition of a crash was quite mild as I understood it.
I think you have to download the PDF to get the entire study. The step that hasn't been done publically is to tell people the results of the study and how to drive in a way that reduces the risk of crashes. This HAS been done in the Israeli military and has worked very well.
I look at this particular study as a pilot study. As a pilot, it seemed to work very well. And there is a good reason for this. By reducing the number of variables looked at in driving habits, the type of crash didn't matter. Something like 73% of the crashes were a result of an identifiable lack of attention as confirmed on video. When one isn't paying attention, the type of crash doesn't matter except to the participants.
Rather than taking my generalizations as gospel on the study, read the whole thing before making a decision. My generalizations are lacking in accuracy because that's how generalizations are. |
|
|
|
Axiom2000
Moderator
968 Posts
[Mentor]
Georgetown, Delaware
USA
BMW
R1200RT/ R1200C
|
Posted - 10/17/2009 : 7:25 AM
|
| I understand, and I was not taking any shot at you for your use of the description ,"Autoundngly High". It is just hard to believe that they picked 100 people for a study on crashes and 67 of them randomly crashed. I guess I would need to read about the selection process to gain a better understanding. |
|
|
|
haithabu
Standard Member
232 Posts
Canmore, Alberta
Canada
Honda
Varadero
|
Posted - 10/17/2009 : 8:42 AM
|
| I wish I could see a longitudinal study of a set of riders extending over a number of years. I.e. of my cohort of those who began riding in 1974, what proportion have had accidents at what stage in their experience? Does cumulative lifetime mileage significantly affect the chances of an accident at any given point past the first six months? What is the risk curve upon reentry into the pool of riders after an absence of some years? Is it significantly different from that of a new rider? etc. |
|
|
|
DataDan
Standard Member
198 Posts
[Mentor]
Central Coast, CA
USA
Honda
|
Posted - 10/17/2009 : 11:00 AM
|
quote:
Originally posted by galileo
In my opinion, any crash research that doesn't look at those who don't crash is flawed as it doesn't get at the source of the crashes. It's also very unlikely that a study that only looks at crashes will find anything significantly different than the Hurt study. The numbers may change some but the causes will be about the same.
The biggest problem with the Hurt study is the denominator of the causes was not known. N% of the riders who crashed not having licenses may simply mean N% of the riders who rode didn't have licenses.
Hurt did look at non-crashers, as did MAIDS, and I can't imagine that professionals conducting a similar study would omit that--even if budget-constrained.
To collect their "exposure sample", Hurt's team interviewed motorcyclists at gas stations and other spots near the crash locations, on the same day of the week and time of day. A total of 2310 riders were interviewed, compared to 900 crash-involved riders. This enabled them to draw conclusions such as this: "Motorcyclists between the ages of 16 and 24 are significantly overrepresented in accidents; motorcyclists between the ages of 30 and 50 are significantly underrepresented." |
|
|
Niebor
Ex-Member
|
Posted - 10/17/2009 : 8:13 PM
|
quote:
Originally posted by DataDan
This enabled them to draw conclusions such as this: "Motorcyclists between the ages of 16 and 24 are significantly overrepresented in accidents; motorcyclists between the ages of 30 and 50 are significantly underrepresented."
Dare I ask the obvious question, relating to statistical significance of those of us, Over 50?  |
|
|
|
DataDan
Standard Member
198 Posts
[Mentor]
Central Coast, CA
USA
Honda
Peer Review:
1
|
Posted - 10/17/2009 : 9:58 PM
|
quote:
Originally posted by Niebor
Dare I ask the obvious question, relating to statistical significance of those of us, Over 50?
We are, of course, the elite. 
Somehow, we managed to survive those deadly "overrepresented" years 16-24 at the time of Hurt's study and lived to achieve what we might expect to be the coveted "underrepresented" ranking this time around.
Your post raises a serious question, though. In Hurt's study, the 50+ age group comprised only about 3% of the both the crashers and the exposure sample, far too few instances to have any statistical power. We have many more 50+ riders now, and some have many years of continuous experience, which should tend to reduce crash involvement. Others, though, are relative newbies, or have been away from riding for so long that that haven't retained any useful skills, but they would be expected to generally have lower risk tolerance due to their age. What should we expect to see in the next study?
And here's another question about experience that I sometimes ponder when I find myself wide awake at 3:00am: To what extent is the well-demonstrated risk reduction benefit of experience due to skill and judgment gained from years of riding? And to what extent is experience simply a weeding-out process, through which people who don't possess some unknown innate abilities and attitudes find that the perceived risk exceeds a level they find tolerable? |
|
|
|
rayg50
Advanced Member
851 Posts
[Mentor]
NYC, NY
USA
Honda
Shadow Spirit 750DC
|
Posted - 10/18/2009 : 8:14 AM
|
quote:
Originally posted by DataDan
And here's another question about experience that I sometimes ponder when I find myself wide awake at 3:00am: To what extent is the well-demonstrated risk reduction benefit of experience due to skill and judgment gained from years of riding? And to what extent is experience simply a weeding-out process, through which people who don't possess some unknown innate abilities and attitudes find that the perceived risk exceeds a level they find tolerable?
I wonder about what I have put in boldface, not infrequently. I tend to refer to it as ones personality. I mentioned it in another thread where I wondered if a particular personality was drawn to Iron Butt riding. To me it is a critical piece to the puzzle.
IMO, rider "personalities" differ from non riders. We find the risk level tolerable, they do not. Within riders I believe the risk tolerance varies greatly. I think some clues could be gathered in the post crash interviews of family and friends but those 2 groups may paint very different pictures of a personality.
How could something like that be quantified in a study?
|
|
|
|
gymnast
Moderator
2805 Posts
[Mentor]
Meridian, Idaho
USA
Harley-Davidson
Sportster Sport
|
Posted - 10/18/2009 : 11:28 AM
|
| Ray, Google "Eysenck personality Profiler". This psychological measurement instrument has often been used in studies predicting the probability of an individuals predisposition to personality factors likely to lead to increased incidence of trauma.. |
|
|
|
Discussion Topic |
|
|
Most Recent Posts
|
Activity History
|
Safety Tips
OSU Motorcycle Crash Research