This topic should be in a separate thread.
Your assumption that Mazda didn't change anything on 1st-gen CX-5 since 2013 is simply not true. Every car manufacture is trying to make some changes to improve the ratings in ANY crash tests even though the model change didn't happen. Here is an example from
IIHS Small Overlap Front rating on 2014 CX-5 which got improved from 2013's "Marginal" to "Good" all because Mazda made some changes:
The A6 Special Program for 2016(.5) CX-5 is also an attempt to make frontal crash on passenger side safer by re-adjusting the timing of the airbag explosion which definitely has something to do with poor 3-star ratings on NHTSA frontal crash on passenger side IMO. But Mazda won't tell us the truth though.
I'm not a fan of big government. But at least NHTSA opens its crash data and let you access. You think Mazda will let you access all the "secret changes", either for getting better crash ratings, or for saving a penny but backfired on them and got a poorer crash ratings? Between the two, and from more than 2 years of owning the CX-5, I definitely trust more toward NHTSA than Mazda! You also admitted NHTSA won't play any favoritism against any car manufacture. Why couldn't Mazda, like Honda on CR-V or many competitors, simply hold up the structure design of previous gen for NHTSA's side crash test so that the 2017 CX-5 could get 5-star overall safery rating like 2017 Honda CR-V? Blaming NHTSA for inconsistency but not blaming Mazda for screwing something up is simply a cop-out! As long as everybody plays a fare game, NHTSA 5-star safety rating is a useful info for consumers to compare the safety on cars.
More stars mean safer cars still stands!
Again, I was amazed how many people here now trashing the NHTSA safety ratings, but 2 years ago with 2015 CX-5 having a perfect score and nobody had any issues? We also forgot it's NHTSA who caught the safety issue on fuel filler pipe of CX-5 during a rollover test and forced Mazda to stop the sale immediately until an acceptable resolution given and a recall was initiated. If Mazda didn't make some "secret" changes, why that recall doesn't include 2013 CX-5 which has exactly the same fuel filler pipe?
On this forum, like most car forums, fans will hold up positive news about their rides and put down negative news. That doesn't change my point.
SSP A6 was a software change to address airbag tearing, which as far as I can tell was associated with the infamous Takata airbag inflators. SSP A6 was 1.5 years after the crash test that determined the 2016 CX-5's ratings. The airbag didn't tear during the crash test. So there is no connection.
You think that Mazda made some material change to the 2016 model, in secret, that made it less safe, and NHTSA NCAP testing revealed it. I don't believe that. Publicly, the 2016.5 has some new bells and whistles and the allocation of features to trim levels got reshuffled a bit, but it's the same car. It also looks the same: from the outside, inside the cabin, engine bay, underneath. I know that looks aren't everything, and it's theoretically possible they made some structural or material change that's not visible and not publicly disclosed. But why would they do that for half a model year? And why risk re-testing it when it was rated 5 stars across the board based on the previous test? That makes no sense. More importantly, I downloaded the frontal barrier crash test reports and they reinforce my belief that it's the same car. And like I said, this isn't a one-off occurrence with the CX-5. It happens to other makes and models too. Like the CR-V I mentioned.
Here's a couple things I learned from my wife who (in a former career) was an airbag test engineer for Delphi (GM). First, full scale crash testing of production vehicles is not statistically valid. Manufacturers don't spend the money to destructively test a statistically significant number of identical vehicles. For NCAP, they test one vehicle, get one result, and have no idea whether that result is typical of their fleet or a statistical outlier. Most of the testing done during vehicle development utilizes small-scale test rigs which are relatively inexpensive to operate and produce repeatable results. That level of testing can be statistically valid, but it characterizes variables such as airbag ignition delay and inflation time, not variables that are meaningful to a consumer. Then once you integrate things together (such as airbag, steering wheel, and dash) you'll validate the design by doing sled testing. Sled tests are a more controlled environment than a full-scale vehicle test, but they're still too expensive to perform repeatedly such that you can characterize the statistical distribution of results.
Another thing I learned, which sounds pretty obvious, is that positioning of the ATD (dummy) can have a large effect on the results, so exact positioning is critical to getting repeatable results. The technicians try to be millimeter accurate. In full scale vehicle testing, the vehicle is rapidly accelerated to a constant velocity before it impacts the barrier. During that acceleration, which is much greater than what the car could manage under its own power in real life, the ATD moves, so at impact it's not quite in the same place you put it. You can see the movement in high speed video, but it's not quantified, and neither is the effect on the results. That's one of the reasons why, in sled testing, the ATD and items under test are kept stationary prior to the simulated impact, e.g. in a frontal impact sled test, the sled is accelerated backward with an impulse. It also means that how you sit in the car and how you adjust the seat can have a big impact on how you fare in a collision.
Nobody can say for sure whether the results of the CX-5 crash tests in 2013 and 2015 are just two random draws from the same statistical distribution or whether they represent different distributions due to a design change. Because nobody is going to crash a hundred new CX-5's to find out. All I can say is that I've looked at the test reports and based on the pre- and post-crash pictures and the accelerometer data it looks like the same car.
Here's a summary of my POV on this topic:
1. The transparency and general competence of NHTSA is questionable
- Their web site is skin deep, and when I try to dig in, I get a lot of broken links and some database queries that don't work
- NHTSA has been audited, investigated, and publicly criticized many times in recent years for a lack of transparency and accountability
- NHTSA helped GM conceal their ignition problems for a decade
- NHTSA failed to recognize and get on top of the Takata airbag problems
2. I can't translate the NHTSA NCAP star ratings into anything meaningful
- I looked for but couldn't find a document or summary explaining how NHTSA star ratings are determined from crash test data
- When the same model of car is tested multiple times, the star ratings are often different, even when the data looks similar
- The only study I found that tried to correlate NCAP ratings to real world outcomes wasn't conclusive (too small sample size I think)
3. Full-scale crash test results are not appropriate for splitting hairs
- Too many variables, and important variables such as the position of the dummy are hard to control
- It's too expensive to crash test a sufficient number of identical vehicles to make the results statistically valid
- Because of that, nobody really knows how repeatable the results are, even the engineers conducting the test
I'm not saying the ratings are useless, just take them with a grain of salt. It is very common for NHTSA to alternate between 4- and 5-star ratings for the same vehicle in different model years, with no known changes to the vehicle. In my layperson's opinion, it looks to me like the difference between 4-star and 5-star ratings is probably within the statistical variability/margin error of the test, and may not indicate any meaningful difference in the safety of two vehicles. If a vehicle has a 2-star or 3-star rating, that suggests to me that there might have been an obviously bad outcome in the test so I'll dig deeper if I can. But a 2-star or 3-star rating is uncommon, despite the fact the 3 stars is supposed to be average. The last time I saw a 3-star rating was when my sister-in-law was looking at a Rogue, and I went to the NHTSA website to download the test report and got a frigging error (too common). She ended up getting the bigger Pathfinder anyway.
Since you mentioned the IIHS, I'll say that I hold them in higher regard than the NHTSA. But their testing methodology still suffers from the same problem of undersampling. You never know whether their test result is representative of the vehicle you're going to purchase. Your car could have a critical design defect that affects 1/100 customers, which is small percentage but a huge number, that never appeared in safety tests. A large number of vehicles with Takata airbags went through NHTSA NCAP and IIHS testing with no issues, because the problems with Takata airbag inflators occur infrequently and under conditions that full-scale crash tests wouldn't reveal. I think most real passenger vehicle safety issues are like that, failing tires or failing brake or failing safety systems or whatever, stuff that doesn't show up in crash tests of new vehicles. So to me, obsessing over 4-star vs. 5-star ratings is not the best way to keep your family safe.