[Sia]

Thanks all, appreciate the comments.

Beyond the rollover test, are there other tests or safety features that Volvo puts into the cars that the rating agencies don't account for? Am sure there are lots but haven't been able to find 'the full list' of why Volvo's are so safe. (e.g., I read somewhere that they design their gas tanks differently)

(Farmer and Grecian, I'm still relatively new so trying to understand what's in the '12 ounces')

 

[31355]

Volvo probably crash tests one model more than other companies crash test their entire lineup.....

Volvo's Safety Research Center is the most advanced in the world, able to recreate virtually ANY accident in a controlled lab environment, something no other manufacturer can do. Their swinging launch tunnel allows any angle of impact to be tested and is unique to Volvo. Volvo's WHIPS seats are another example of something no other manufacturer has developed, and now the XC90 has road run-off spine protection in the seat, another exclusive feature only Volvo is testing for.

Volvos are so safe because it's their #1 design mission, and they have been doing it since 1927. Many other manufacturers have copied Volvo's structural design now in many ways. Plus Volvo gathers real life data with a team on EVERY accident that occurs around Gothenburg, having a special agreement with the government (as in, no crashed Volvo is moved until Volvo has looked at the cars, and many crashed crash go back to Volvo from insurance companies to anylasis). These are unique insights only Volvo has. Google Volvo's safety center and you can read a lot about it.

 

[gweempose]

Volvo probably crash tests one model more than other companies crash test their entire lineup.....

Volvo's Safety Research Center is the most advanced in the world, able to recreate virtually ANY accident in a controlled lab environment, something no other manufacturer can do. Their swinging launch tunnel allows any angle of impact to be tested and is unique to Volvo. Volvo's WHIPS seats are another example of something no other manufacturer has developed, and now the XC90 has road run-off spine protection in the seat, another exclusive feature only Volvo is testing for.

Volvos are so safe because it's their #1 design mission, and they have been doing it since 1927. Many other manufacturers have copied Volvo's structural design now in many ways. Plus Volvo gathers real life data with a team on EVERY accident that occurs around Gothenburg, having a special agreement with the government (as in, no crashed Volvo is moved until Volvo has looked at the cars, and many crashed crash go back to Volvo from insurance companies to anylasis). These are unique insights only Volvo has. Google Volvo's safety center and you can read a lot about it.

Given all these facts, I would think the numbers should speak for themselves. I would expect Volvo to have a lower frequency of serious injuries and fatalities than other brands. Is there a site where one can lookup these kind of statistics.

 

[Linqid]

Given all these facts, I would think the numbers should speak for themselves. I would expect Volvo to have a lower frequency of serious injuries and fatalities than other brands. Is there a site where one can lookup these kind of statistics.

One of the larger Swedish insurance institutes, Folksam, has since 1980 been collecting data from thousands of real world accidents. Their latest study shows that the latest generation of Volvo cars are all 50 % safer than the average car. The first gen XC90 is 48 % safer than the average car. This gives all of the current gen Volvo cars a (shared) number one place on the list, with the XC90 becoming second. In third place you find a Honda.

As a side note, in 2013 years study the best ranking Volvo was 59 (!) % better than the average car. Luckily, it is because of the average car getting safer and not because Volvo being worse than before. The main factor to this is older cars (without airbags, ESP, etc.) slowly disappearing from our roads.

Press release in English explaining the study

The results. Scroll down and you will se the complete list. For English, press the flag on the top, right side of the list.

 

[IIHS]

Given all these facts, I would think the numbers should speak for themselves. I would expect Volvo to have a lower frequency of serious injuries and fatalities than other brands. Is there a site where one can lookup these kind of statistics.

You'd think so, wouldn't you? However, that's not the case, at least when it comes to driver fatalities. Here's the most recent survey of driver death rates for '11 model vehicles sold in the US. All of the data is compiled from actual fatal crashes in the US as documented by the NHTSA:

http://www.iihs.org/iihs/sr/statusreport/article/50/1/1

The XC90 was part of the "zero" list. But so were about a dozen other vehicles from manufacturers including Toyota, Honda, Audi, Subaru, MB, Kia, etc.

Here's the driver death rate survey before that (for '08 model year vehicles):

http://www.iihs.org/iihs/sr/statusreport/article/46/5/1

There were at least 7 vehicles on the "zero" list. The XC90 was not one of them. It actually did quite a bit worse (28), although better than the average luxury SUV.

Here's the DDR survey before that (for '04 model year vehicles):

http://www.iihs.org/iihs/sr/statusreport/article/42/4/1

Two Volvos were present: the S60, which had a DDR of 34, that was again...not near the top of the list, and the S40 with a DDR of 89, which was...worse.

 

[chicagoxc90]

You'd think so, wouldn't you? However, that's not the case, at least when it comes to driver fatalities. Here's the most recent survey of driver death rates for '11 model vehicles sold in the US. All of the data is compiled from actual fatal crashes in the US as documented by the NHTSA:

http://www.iihs.org/iihs/sr/statusreport/article/50/1/1
.

The IIHS driver death reports have to be interpreted very cautiously. These are not actual death rates but adjusted death rates, and the adjustments have injected a huge amount of noise in the data, especially in the 2011 report. Back in 2011, I noticed a few anomalies in the report (especially in Subaru and Audi data) and then downloaded and analysed the raw data from NHTSA (I do a lot of modeling and statistical analysis in my profession, so this is a hobby for me), and reached these conclusions (this is just for 2011, I did not look at 2015):

-The 2011 report used modeling that was very different from previous years, and therefore a direct comparison to previous reports cannot be made. I believe the modeling was changed again in 2015. Therefore you can see strange results. e.g. The Audi A4 going from 14 (2007) upto 28 (2011). The XC90 goes from 28 (2011) to 0 (2015) for the same model. These are not real (statistically valid) differences.

-There is a huge amount of uncertainty introduced by the adjustments, making direct comparisons between models basically useless, unless the death rate difference was greater than about 60 (depending no the vehicle-years and strength of adjustments).

-Differences between classes (e.g. small sedans versus mid-sized sedans) were valid; the much larger numbers allowed for more reliable comparisons.

- The automaker that had the most unfavorable distortion was Subaru; the adjustments made their cars appear worse than they actually were.

 

[31355]

You'd think so, wouldn't you? However, that's not the case, at least when it comes to driver fatalities. Here's the most recent survey of driver death rates for '11 model vehicles sold in the US. All of the data is compiled from actual fatal crashes in the US as documented by the NHTSA:

http://www.iihs.org/iihs/sr/statusreport/article/50/1/1

The XC90 was part of the "zero" list. But so were about a dozen other vehicles from manufacturers including Toyota, Honda, Audi, Subaru, MB, Kia, etc.

Here's the driver death rate survey before that (for '08 model year vehicles):

http://www.iihs.org/iihs/sr/statusreport/article/46/5/1

There were at least 7 vehicles on the "zero" list. The XC90 was not one of them. It actually did quite a bit worse (28), although better than the average luxury SUV.

Here's the DDR survey before that (for '04 model year vehicles):

http://www.iihs.org/iihs/sr/statusreport/article/42/4/1

Two Volvos were present: the S60, which had a DDR of 34, that was again...not near the top of the list, and the S40 with a DDR of 89, which was...worse.

I've seen these "numbers" before....if you delve into them that are basically useless for comparisons sake and only goof for some entertainment. Others here can tell you why, technically, these numbers are difficult to use at best. There is no great way to compare accident rates like this, just generalizations (and even then they can be misleading). Your numbers here suggest about the exact opposite of many, many, many other data results collected by insurance companies and others.

If 20 Audis are hit by a car at 25 mph and 0% die, and 20 Volvos are hit by semis at 50 mph and 100% die, is the Volvo 100% more unsafe than the Audi? No...... these numbers can be spun in any way possible to say just about anything. Use extreme caution when interpreting results like these. They are not worthless, but at the same time not really helpful.

 

[WillW]

I'm a stat guy too. I did not think one could verify the IIHS death rates independently, because they have their own data that is not publicly available?? I think, then, IIHS should be more upfront with their data and analysis method changes. There seems to be, right now, a lot of people doing things that are not above board -- e.g., VW, Hillary Clinton, the Pope, Turing Pharmaceuticals, and now IIHS, possibly? If you want me to believe your analysis, provide the raw data and analysis methods.

 

[chicagoxc90]

I think, then, IIHS should be more upfront with their data and analysis method changes.

To be fair to the IIHS, they HAVE mentioned the changes in their modeling methods both in 2011 and 2015. The actually state that when they used their 2011 adjustments in their 2015 analysis, "the results were unstable", and they went back to the 2007 model. No shenanigans here. Unfortunately, by the time the data are reported in the media, all these details are lost.

The raw data is from NHTSA's Fatality Analysis Reporting System (FARS). This is freely downloadable from the NHTSA's ftp site.

 

[IIHS]

The IIHS driver death reports have to be interpreted very cautiously.

I understand you disagree with their results, but I'll have to go with the IIHS' work, since they've presumably got a number of folks in-house who do this for a living.

I've seen these "numbers" before....if you delve into them that are basically useless for comparisons sake and only goof for some entertainment.

I'll respectfully disagree here as well. By analysis from a 3rd party industry that makes a living off crashing and comparing vehicles for safety to reduce insurance costs (and does so using NHTSA data), Volvos don't stand out in reducing driver fatalities compared to other makes. They've been gradually catching up to other manufacturers, but there are plenty of other manufacturers who do as well or better in keeping their drivers alive.

 

[chicagoxc90]

I understand you disagree with their results, but I'll have to go with the IIHS' work, since they've presumably got a number of folks in-house who do this for a living.

This is not a matter of disagreeing with the IIHS, since the following are in the IIHS reports themselves:

1) The death rates are not actual driver death rates but "adjusted driver death rates"
2) A new adjustment model was used in the 2011 report. In 2015 the IIHS reverted to the old model because when the new model was used "the results were unstable"
3) Due to the change in the 2011 adjustments it is "impossible to do an apples-to-apples comparison" to previous reports
4) The confidence limits (uncertainty) for individual model death rates is wide

The point is to understand the limitations of the IIHS reports.The IIHS is not trying to hide these limitations. The problem is that when the reports make it to a press release or the media, these important details are lost, often resulting in significant misinterpretations. One does not have to download and reanalyze FARS data to understand the above points. A careful reading of the IIHS reports is enough. The reanalysis did give me additional insights into the size of the adjustments, and the distortions of the 2011 model.

 

[sergjut14]

This is not a matter of disagreeing with the IIHS, since the following are in the IIHS reports themselves:

1) The death rates are not actual driver death rates but "adjusted driver death rates"
2) A new adjustment model was used in the 2011 report. In 2015 the IIHS reverted to the old model because when the new model was used "the results were unstable"
3) Due to the change in the 2011 adjustments it is "impossible to do an apples-to-apples comparison" to previous reports
4) The confidence limits (uncertainty) for individual model death rates is wide

The point is to understand the limitations of the IIHS reports.The IIHS is not trying to hide these limitations. The problem is that when the reports make it to a press release or the media, these important details are lost, often resulting in significant misinterpretations. One does not have to download and reanalyze FARS data to understand the above points. A careful reading of the IIHS reports is enough. The reanalysis did give me additional insights into the size of the adjustments, and the distortions of the 2011 model.

Thanks Chicago, for this, your other explanations and insight!!! eace:

 

[Gary-16-Xc90]

Thanks as well ChicagoXC90! Footnotes and endnotes are absolutely critical to fully understand an IIHS report, a financial statement, a court decision and even advertisements. Thanks for bringing the footnotes and qualifications to light.

 

[IIHS]

3) Due to the change in the 2011 adjustments it is "impossible to do an apples-to-apples comparison" to previous reports

Can you point out exactly where the IIHS states this? Because here's what I see in the status report for 2011 vehicles:

"The rates are adjusted for driver age and
gender, but not all the demographic factors
that can influence results are accounted
for. Four years ago when IIHS released
death rates for 2008 models, researchers
found that they needed to include an ad-
justment for calendar year in order to ac-
count for the effects of the recession. For
this reason, researchers developed another
model that included the calendar year ad-
justment, as well as adjustments for vehi-
cle age and vehicle density at the garaging
location, in addition to driver age and
gender. That more-complex model worked
well at the time, but when researchers
used it to calculate the 2011 death rates,
the results were unstable. Since the U.S.
economy didn't see such large fluctua-
tions during the new time period, the
researchers went back to the previ-
ous model."

Is there an additional status report you've read that states that this report cannot be compared to any other, or are these your personal interpretations of the report?

 

[chicagoxc90]

Can you point out exactly where the IIHS states this? Because here's what I see in the status report for 2011 vehicles:

Sure. It is in the 2011 report: http://www.iihs.org/iihs/sr/statusreport/article/46/5/1

Because the latest round of driver death rates has been fine-tuned in a way that previous ones weren't, it's impossible to do an apples-to-apples comparison, but some broad trends are clear.

 

[IIHS]

Sure. It is in the 2011 report: http://www.iihs.org/iihs/sr/statusreport/article/46/5/1

You might want to keep up with the reports, because they don't seem to have any trouble comparing reports to each other as of 2015:

"The list of models with the lowest death rates illustrates
just how much vehicles have improved. Eight years ago,
there were no models with driver death rates of zero (see
Status Report, April 19, 2007). Now there are nine. These
vehicles - which include several luxury models but also
some less expensive ones such as the Kia Sorento midsize
SUV and the Subaru Legacy sedan - had no driver deaths
during the calendar years studied."

http://www.iihs.org/iihs/sr/statusreport/article/50/1/1

Thanks.

You're welcome!

 

[chicagoxc90]

You might want to keep up with the reports, because they don't seem to have any trouble comparing reports to each other as of 2015:

As the IIHS has acknowledged comparisons between different adjustment models is very difficult. This is especially true for comparisons between models and within models. However, as they have stated you can look at broad trends, since the additional 2011 adjustments mostly affect intra and inter model comparisons. Now since in 2015 they reverted back to the 2007 model (they abandoned the 2011 model), the 2007 to 2015 comparisons have far more validity than the 2011 to 2015 comparisons (where you have strange results like the XC90 going from 28 to 0 for the same model).

 

[IIHS]

As the IIHS has acknowledged comparisons between different adjustment models is very difficult. This is especially true for comparisons between models and within models. However, as they have stated you can look at broad trends, since the additional 2011 adjustments mostly affect intra and inter model comparisons. Now since in 2015 they reverted back to the 2007 model (they abandoned the 2011 model), the 2007 to 2015 comparisons have far more validity than the 2011 to 2015 comparisons (where you have strange results like the XC90 going from 28 to 0 for the same model).

I agree with you here. :beer:

However, (and this isn't toward you, but toward gweempose) whether we look at the 2000 study, the 2005 study, 2007 study, the 2011 study by itself, or the 2015 study, returning to this question...

Given all these facts, I would think the numbers should speak for themselves. I would expect Volvo to have a lower frequency of serious injuries and fatalities than other brands. Is there a site where one can lookup these kind of statistics.

If you look at the stats the IIHS provides, the best Volvo vehicles, when present, either do much worse than, slightly worse than, or very rarely, equal to the best vehicles in their classes by other manufacturers.

In 2015, when '08-'11 models were studied, the XC90 was one of 4 mid-sized SUVs (including the Toyota Highlander Hybrid, Kia Sorento, and Lexus RX) to have had 0 extrapolated driver deaths.

In 2011, when '05-'08 models were studied, the XC90 had a DDR of 28, which was worse than that of the best mid-sized SUV: the Ford Edge, which had a DDR of 0.

In 2007, when '01-'04 models were studied, the S60 had a DDR of 39, worse than that of the best mid-sized car, the Infiniti G35, which had a DDR of 11, and the S80 had a DDR of 89, worse than that of the best large car, the MB E-Class, which had a DDR of 14.

In 2005, when '00-'03 models were studied, the S80 had a DDR of 45, worse than that of the best large car, the MB E-Class, which had a DDR of 10. The S40 had a DDR of 88, worse than that of the best small car, the Infiniti G20, which had a DDR of 46.

In 2000, when '94-'97 models were studied, the 850 had a DDR of 39, worse than that of the best mid-sized car, the Infiniti J30, which had a DDR of 20.

 

[Red Leather]

I'm not very familiar in either the details on how IIHS performs it's test crashes or how IIHS calculates these DDR results, but I find it quite strange that car models that they rank as Top Safety Pick and Top Safety Pick Plus (such as the mentioned Volvo models) scores such bad DDR results as mentioned in the previous post above.

What should we believe in, the test crashes that IIHS performs or the DDR calculations that IIHS publishes?

Or perhaps non of them?

Perhaps it is better to believe in Euro-NCAP test crash results and real world crash data from Folksam. Those seem more in line, at least for the Volvo models mentioned.

 

[chicagoxc90]

I'll reiterate what I said earlier in the thread:There is a huge amount of noise in the individual model death rates making direct comparisons between individual models basically useless, unless the death rate difference is greater than about 60 (depending on the vehicle-years and strength of adjustments). In other words, the margin of error is very large. If one takes the report literally, you'll get absurd results like the XC90 miraculously improving its ranking in its class from 12th (in 2011) to a tie for 1st four years later. As we know, this is the exact same XC90 while other models have changed and improved. This is nothing but noise, and not understanding this point is tantamount to being "Fooled by Randomness" [hat-tip to Taleb].

That said, these IIHS reports do have some use. As I had mention before, the data are useful for discerning differences between classes (e.g. small sedans versus mid-sized sedans). The much larger numbers allowed for more reliable comparisons, and showed the strong effect of vehicle size (mass).