Crash prone 'modern roundabouts'

[jeffandnicole]

This is a hilariously incomplete list of 2x2 or greater roundabouts in the country.  Apparently the 'average' takes into account just high-crash roundabouts, and weeds out anything without a large crash rate.  That's like saying I was an A student in high school if you don't count any grades of B and below.

You were an “F” grade student because you didn’t show your work.  You could easily prove I ignore low-crash rate complex roundabouts by simply citing them.  Let’s say any roundabout with a crash rate below 0.8 MEV is considered low (that is about the average crash rate for a signalized intersection).  You yourself said my list was hilariously incomplete.  Then there must be a hilariously high number of complex roundabouts with low-crash rates i ignored.  Humor me.

We'll start with this one, which is your own posting: https://www.aaroads.com/forum/index.php?topic=15546.msg2223079#msg2223079

Don't like that one?  Then choose any of the other dozen or so 2x2 roundabouts that exist around The Villages, FL.

[tradephoric]

^I fail to see how a post about a driver crashing through a building in the middle of a roundabout in The Villages is proof that the roundabout has a low-crash rate.  I asked you a specific request -  to cite a complex roundabout that has a crash rate below 0.8 MEV.  You failed to do that.

[Rothman]

Shouldn't that be .8 per MEV?

[jeffandnicole]

And I said your list is incomplete. You provided a minimal list and then claimed the average of ALL complex roundabouts is computed based on that list.

I don't need to show a single one...not that it matters because I don't know where you get your data from. I quickly and easily proved You didn't list all complex roundabouts. Since we all know you have the data, just provide the list.

And I noticed you failed to address the fact you included Canadian roundabouts on your American list.

[Rothman]

I'm also wondering about more recent data on the NY 85/140 roundabout in Slingerlands, NY.  Crash rates must have plummeted there.

[Scott5114]

To begin with, we would need a number of the number of complex roundabouts in the US, in order to determine what a representative sample size would be.

[jakeroot]

To begin with, we would need a number of the number of complex roundabouts in the US, in order to determine what a representative sample size would be.

I think the bigger issue is that none of us have access to the data required to do a comprehensive study on the matter.

[US 89]

This is a hilariously incomplete list of 2x2 or greater roundabouts in the country.  Apparently the 'average' takes into account just high-crash roundabouts, and weeds out anything without a large crash rate.  That's like saying I was an A student in high school if you don't count any grades of B and below.

You were an “F” grade student because you didn’t show your work.  You could easily prove I ignore low-crash rate complex roundabouts by simply citing them.  Let’s say any roundabout with a crash rate below 0.8 MEV is considered low (that is about the average crash rate for a signalized intersection).  You yourself said my list was hilariously incomplete.  Then there must be a hilariously high number of complex roundabouts with low-crash rates i ignored.  Humor me.

Uh, you’re the one who didn’t show your work. You cherry picked 40 or so roundabouts that you could find that had very high crash rates, and found the average crash rate for only those, and you’re saying that applies to all multilane roundabouts. That makes your stats invalid. It’s like a poll where they tell you only negative things about a politician and then ask what your opinion on that politician is now.

Also, it looks like you ran out of roundabouts in the US with high enough crash rates, so you decided to pull a few out of Canada to help prove your point.

To begin with, we would need a number of the number of complex roundabouts in the US, in order to determine what a representative sample size would be.

I think the bigger issue is that none of us have access to the data required to do a comprehensive study on the matter.

I’m sure tradephoric could find that data somewhere, but he wouldn’t want to dig into it because of fear that it wouldn’t prove his point.

[Scott5114]

To begin with, we would need a number of the number of complex roundabouts in the US, in order to determine what a representative sample size would be.

I think the bigger issue is that none of us have access to the data required to do a comprehensive study on the matter.

It could probably be acquired through creative FOIA requests, but doing so would require a lot of time and effort.

[jakeroot]

To begin with, we would need a number of the number of complex roundabouts in the US, in order to determine what a representative sample size would be.

I think the bigger issue is that none of us have access to the data required to do a comprehensive study on the matter.

It could probably be acquired through creative FOIA requests, but doing so would require a lot of time and effort.

And Trade has come out and admitted before that he piecemeal's his data, but only because so many transport departments don't publish crash numbers. He's lucky to find any numbers to work with. Further in his defence, it's rather interesting that the only intersections he can find information for, almost universally perform poorly.

FOIA requests are a sure-fire way of getting data, but would indeed by time consuming. Our best hope would be for the FHWA to perform a comprehensive study, since (AFAIK) they don't need to go down the FOIA trail.

[Rothman]

The fact that FOIA or state-equivalent requests (e.g., NY's FOIL) are required does not negate the fact that Tradephoric's sample of data is probably biased and unrepresentative of the population.  Count me with those that consider him a data cherry picker.  Coming to conclusions based upon biased data is still inappropriate when proper data is just harder to get than a Google search.

[kalvado]

The fact that FOIA or state-equivalent requests (e.g., NY's FOIL) are required does not negate the fact that Tradephoric's sample of data is probably biased and unrepresentative of the population.  Count me with those that consider him a data cherry picker.  Coming to conclusions based upon biased data is still inappropriate when proper data is just harder to get than a Google search.

Everyone has an agenda today, but very few have an idea about statistics - so good dataset is not expected until FHWA gets a good shock (like a major school bus accident on a roundabout)
Within this forum, other side didn't present any dataset from past 10 years, so there is no logical choice other than taking trade's data at face value. Any other opinion is uneducated guesswork at best. But of course, First Amendment protects religious believes - so feel free...

[tradephoric]

Uh, you're the one who didn't show your work. You cherry picked 40 or so roundabouts that you could find that had very high crash rates, and found the average crash rate for only those, and you're saying that applies to all multilane roundabouts. That makes your stats invalid. It's like a poll where they tell you only negative things about a politician and then ask what your opinion on that politician is now.

Or like how the IIHS only tells you positive things about roundabouts and then ask you what your opinion on that roundabout is.  The IIHS national study analyzed 9 multi-lane roundabouts in Colorado.  What about the other 49 states?  Cherry picking!  Of the 9 multi-lane roundabouts, 6 of them serviced interchanges along I-70.  Interchange roundabouts historically have lower crash rates when compared to non-interchange roundabouts.  Cherry picking!  Of the 3 non-interchange roundabouts analyzed, they were all along Avon Road which is signed for 25 mph.  The safety statistics of non-interchange multi-lane roundabouts are based on a 1500 foot section of Avon Road where drivers are traveling at subdivision street speeds.  Cherry picking! 

More recently the IIHS acknowledged that people are having difficulty navigating two-lane roundabouts.  Even then they only analyzed two multi-lane roundabouts near Bellingham, Washington (and only one of them was a true balanced 2x2 roundabout).  Cherry picking!  In addition, the 2x2 roundabout picked for the analysis has a very low AADT compared to other multi-lane roundabouts in the country.  At a low volume 2x2 roundabout there are more gaps in traffic and it's less likely that a vehicle entering the roundabout will even encounter another vehicle.  Cherry picking!  Even with the favorable analysis, the IIHS acknowledged that noninjury crashes were 6 times as high at the Wiser Lake intersection as they would have been without the roundabout conversions.  Based on a single 2x2 roundabout analyzed, the Insurance Institute's senior vice president Anne McCartt states the following:

"We don't know why noninjury crashes are so much higher than expected, but it may be related to confusion about right-of-way rules and other issues that drivers reported to us," McCartt says. "Nevertheless, these roundabouts are making travelers safer by reducing injury crashes."
http://www.iihs.org/iihs/sr/statusreport/article/48/2/3

Good thing the IIHS cherry picked one the most favorable 2x2 roundabout to come to the conclusion that injury crashes are being reduced (not to mention a 2x1 roundabout was thrown in with their analysis for good measure).   The IIHS doesn't seem to understand the problem if they think it's appropriate to lump the crash data of unbalanced 2x1 and balanced 2x2 roundabouts together.  They are just looking for a result they want.

[jeffandnicole]

Uh, you’re the one who didn’t show your work. You cherry picked 40 or so roundabouts that you could find that had very high crash rates, and found the average crash rate for only those, and you’re saying that applies to all multilane roundabouts. That makes your stats invalid. It’s like a poll where they tell you only negative things about a politician and then ask what your opinion on that politician is now.

Or like how the IIHS only tells you positive things about roundabouts and then ask you what your opinion on that roundabout is.  The IIHS national study analyzed 9 multi-lane roundabouts in Colorado.  What about the other 49 states?  Cherry picking!  Of the 9 multi-lane roundabouts, 6 of them serviced interchanges along I-70.  Interchange roundabouts historically have lower crash rates when compared to non-interchange roundabouts.  Cherry picking!  Of the 3 non-interchange roundabouts analyzed, they were all along Avon Road which is signed for 25 mph.  The safety statistics of non-interchange multi-lane roundabouts are based on a 1500 foot section of Avon Road where drivers are traveling at subdivision street speeds.  Cherry picking! 

More recently the IIHS acknowledged that people are having difficulty navigating two-lane roundabouts.  Even then they only analyzed two multi-lane roundabouts near Bellingham, Washington (and only one of them was a true balanced 2x2 roundabout).  Cherry picking!  In addition, the 2x2 roundabout picked for the analysis has a very low AADT compared to other multi-lane roundabouts in the country.  At a low volume 2x2 roundabout there are more gaps in traffic and it’s less likely that a vehicle entering the roundabout will even encounter another vehicle.  Cherry picking!  Even with the favorable analysis, the IIHS acknowledged that noninjury crashes were 6 times as high at the Wiser Lake intersection as they would have been without the roundabout conversions.  Based on a single 2x2 roundabout analyzed, the Insurance Institute’s senior vice president Anne McCartt states the following:

"We don't know why noninjury crashes are so much higher than expected, but it may be related to confusion about right-of-way rules and other issues that drivers reported to us," McCartt says. "Nevertheless, these roundabouts are making travelers safer by reducing injury crashes."
http://www.iihs.org/iihs/sr/statusreport/article/48/2/3

Good thing the IIHS cherry picked one the most favorable 2x2 roundabout to come to the conclusion that injury crashes are being reduced (not to mention a 2x1 roundabout was thrown in with their analysis for good measure).   The IIHS doesn’t seem to understand the problem if they think it’s appropriate to lump the crash data of unbalanced 2x1 and balanced 2x2 roundabouts together.  They are just looking for a result they want.

If you're aware of these roundabouts and their MEV rates, why aren't they included in your table?

If you are looking at American roundabouts, why did you lump poorly performing Canadian roundabouts in your table?

[kalvado]

If you're aware of these roundabouts and their MEV rates, why aren't they included in your table?

If you are looking at American roundabouts, why did you lump poorly performing Canadian roundabouts in your table?

Well, maybe because IIHS has their agenda and reported injury crush rate only? Everything else is in relative terms, which is a common way to obscure data presenter wants to withhold. I can understand when that happens to proprietary data, but for IIHS it is what it is.

[tradephoric]

If you're aware of these roundabouts and their MEV rates, why aren't they included in your table?

If you are looking at American roundabouts, why did you lump poorly performing Canadian roundabouts in your table?

The crash rate is the number of crashes per million entering vehicle.  That is listed on my table (the very last column).  OTOH, injury crashes are a lot harder to come by and those are not listed.  I know that leads to "well roundabouts are safer.. who cares about total crashes.. injuries are what matters". But there has to be some understanding that there is a percentage of injury crashes that occur at roundabouts.  When total crashes spike, that increases the opportunity for injury crashes as well.  In the case of the Minnesota research, an over 200% increase in total crashes at complex roundabouts led to a 6% increase in injury crashes as well.



Would ignoring the few Canadian roundabouts listed on this chart really change anything?  There is still a crash problem at these complex roundabouts.  Canadian drivers don't drive on the opposite side of the road or drive much differently than American drivers.  But if you think there is an abundance of complex Canadian roundabouts that have a low crash rate that I'm simply ignoring, you are welcome to cite them.  If you do find them, let me know because i would be interested in reviewing their design.

[kphoger]

Would ignoring the few Canadian roundabouts listed on this chart really change anything?  There is still a crash problem at these complex roundabouts.  Canadian drivers don't drive on the opposite side of the road or drive much differently than American drivers.  But if you think there is an abundance of complex Canadian roundabouts that have a low crash rate that I'm simply ignoring, you are welcome to cite them.  If you do find them, let me know because i would be interested in reviewing their design.

What does driving on the other side of the road matter, except allow you to ignore 2x2 roundabouts in England that have low crash rates?

[jakeroot]

Canada does have a lower deaths-per-100k-inhabitants rate than the US:

https://en.m.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

Canada: 6.0
USA: 10.6

As far as I can tell, Canadian provinces do have more thorough driving exams than most US states (BC even uses rear decals like the UK and Australia).

[kalvado]

Canada does have a lower deaths-per-100k-inhabitants rate than the US:

https://en.m.wikipedia.org/wiki/List_of_countries_by_traffic-related_death_rate

Canada: 6.0
USA: 10.6

As far as I can tell, Canadian provinces do have more thorough driving exams than most US states (BC even uses rear decals like the UK and Australia).

US range of numbers if from 4.0 (DC) to 23.1 (MS) vs 6.0 in Canada.
Normalized to 1B km travel, Canada is 6.2, US 7.1 (range 10.6 KY to 4.1 in MA, MN).
US (and Canada, for that matter) is big and diverse. I would say comparing US as a whole to Canada as a whole is better than comparing Northeast to mid-south. Is it valid to put US and Canada into same statistical bucket? Probably better that US and China... 

[tradephoric]

I don't need to show a single one...not that it matters because I don't know where you get your data from. I quickly and easily proved You didn't list all complex roundabouts. Since we all know you have the data, just provide the list.

You believe I have a secret list of crash rates for hundreds of complex roundabouts and I'm only listing the worst ones.  J&N, let me help you out.  I'm sure there is a developer out there that has built a grandiose 2x2 roundabout at the entrance of their subdivision.  Since a complex roundabout in the middle of a subdivision where drivers are already traveling at about 25 mph isn't going to do much to enhance safety, it would be mainly for aesthetics.  Of course, don't tell the IIHS that - the only non-interchange multi-lane roundabouts analyzed in their "national"  study were 3 multi-lane roundabouts along a 25 mph section of Avon Road in Colorado.  The point is if you found the grandiose subdivision entrance has a low crash rate, you could "prove" that all my data is biased.  To a rational person you wouldn't have proven much, but at least you could convince yourself.  Come to think of it you could convince the IIHS pretty easily too.  They love focusing on the safety of complex roundabouts along 25 mph roads.

[Rothman]

The IIHS' bias, if any, does not justify your own.

[jeffandnicole]

I don't need to show a single one...not that it matters because I don't know where you get your data from. I quickly and easily proved You didn't list all complex roundabouts. Since we all know you have the data, just provide the list.

You believe I have a secret list of crash rates for hundreds of complex roundabouts and I'm only listing the worst ones.  J&N, let me help you out.  I'm sure there is a developer out there that has built a grandiose 2x2 roundabout at the entrance of their subdivision.  Since a complex roundabout in the middle of a subdivision where drivers are already traveling at about 25 mph isn't going to do much to enhance safety, it would be mainly for aesthetics.  Of course, don't tell the IIHS that - the only non-interchange multi-lane roundabouts analyzed in their "national"  study were 3 multi-lane roundabouts along a 25 mph section of Avon Road in Colorado.  The point is if you found the grandiose subdivision entrance has a low crash rate, you could "prove" that all my data is biased.  To a rational person you wouldn't have proven much, but at least you could convince yourself.  Come to think of it you could convince the IIHS pretty easily too.  They love focusing on the safety of complex roundabouts along 25 mph roads.

I already proved your data is biased by referring you to your own posting.  But since you don't have data for that particular roundabout (or so you claim), it was magically left off your list of ALL complex roundabouts from of all the states and provinces in America.

[tradephoric]

If you remember my original query of roundabouts to analyze was based on a comment you had made.  This crash analysis of complex roundabouts is largely thanks to you J&N.

It would probably be fair to see a longer list, which will show both 100' diameter roundabouts that don't have high crash frequencies, and roundabouts with larger diameters and how they rank. 

Here is a list of 40 multi-lane roundabouts that I think could be meaningful to this conversation.  This was queried from a database of over 5,100 modern roundabouts.  This was the criteria used to query out the list:

-roundabouts constructed within the last 10 years
-all approaches have 2 entry lanes
-roundabouts have 4-legs
-main & side streets are major state or county routes (ie. likely high AADT roundabouts)
-interchange roundabouts (and frontage road roundabouts) not included



Now the hard part is finding accurate before/after crash data to make any type of analysis.  Here is a google KMZ file that includes the 40 roundabouts in the chart above:
http://www.mediafire.com/download/cd570rarros2c3g/Multi-Lane+Roundabouts+%28with+high+AADT%29.kmz

I queried out 40 roundabouts from a comprehensive database of over 5,100 roundabouts.  At the time, I readily acknowledged that the "hard part"  would be to find accurate before/after crash data for the list of roundabouts.  Now that original list has evolved to the Critical Crash Roundabouts posted above.  There are things you can criticize me on, but the list includes a large sample size of complex roundabouts and it pretty clearly shows they experience high crash rates.

But don't worry, even if there is a massive increase in crashes there will still be a reduction in injury crashes.  But what about the research that found a 6% increase in injury crashes at complex roundabouts in Minnesota?  And ignore the fact that the national IIHS study so often cited to make the claim that roundabouts reduce injury crashes by 76% is based on analyzing just 3 non-interchange multi-lane roundabouts in Colorado... built in 1997 or earlier... along a road signed for 25 mph... talk about a comprehensive "national" study.

[kalvado]

The IIHS' bias, if any, does not justify your own.

Too bad those agencies living on my tax dollars for supporting the road system -  FHWA and NYSDOT - don't bother presenting unbiased and complete dataset to clarify the issue.
While I am on the fence regarding FHWA capability of doing so, our state DOT - with at least one major roundabout being on independently compiled "most dangerous local intersections" list - would probably be unable to understand what we're talking about...

[Hurricane Rex]

If the IIHS was smart, they'd start a new study with different variables and at different speed limits.