TrueDelta

Last month members provided 17,861 responses to the Car Reliability Survey, a response rate of 36.2%. The updated car reliability stats include full results for 458 cars, and partial results for another 351. This compares to 404 full results and 320 partial results three months ago.

The updated results cover owner experiences through the end of June 2010. Other sources of reliability information won’t cover the most recent months until the summer or fall of next year.

With this update we have more solid results for a number of 2010 models and an initial look at some early 2011s.

Audi’s newest A4 and S4 (now reported separately) continue to be better than average, though they might be drifting upward to join the 2009 in “about average.” More repairs are being reported for the related Q5, and while still “about average” any further increase will make it significantly worse than the average.

Few repairs have been reported for the 2010 BMW 3-Series. But we had far fewer responses than for earlier model years, so it’s too soon to tell whether the 335i’s high pressure fuel pump problems have been resolved. Pumps continue to fail in the 2007s through 2009s, though it does seem that if a pump lasts for 10,000 miles it’s no longer likely to fail.

The 2010 GM products we had enough responses for–the compact and large crossovers and the Camaro–are all about average, and not far from “better than average.” The Camaro would be “better than average” if not for a common problem with a loose screw in the spoiler.

Fords seem to be similarly about average and not far from “better than average,” with one exception. The new Taurus is close to “worse than average” thanks to a common problem with chrome trim peeling off the tail lights. The PTU seal might still be prone to failure in the Edge, but we don’t yet have enough responses to be sure.

Hondas remain generally better than average, though problems with the tire pressure monitor in the Fit bumped it to “average.” Earlier years have been excellent, so the 2010 is expected to improve in future updates.

Hyundai’s record is mixed. We do not yet have enough data on the new 2011 Sonata and 2010 Tucson to provide full results for these models. The data we do have suggests that in their first model year they’re about average. The Elantra Touring is similarly about average. The Genesis sedan has improved to better than average in its second model year. 2009s with the Technology Package continue to have more problems. The reported repair frequency for the Genesis Coupe continues to increase, and it’s now quite a bit worse than average. Some owners are very dissatisfied with their cars and with Hyundai’s response. Hyundai appears to have launched this one too soon.

The Jaguar XF has improved little if any in its second model year, and remains much worse than average. Minor electrical problems are common.

Kia’s record is much the same as Hyundai’s. We don’t have quite enough data for a full result for the 2011 Sorento, but what we do have suggests that it’s about average–just like the new Sonata and Sorento. The Forte, new for 2010, is similarly about average. The Soul has improved to better than average now that some intial bugs have been resolved.

We don’t have enough data yet for a full result for the new Land Rover LR4. What we do have suggests that it is considerably more reliable than the LR3 it replaced, and most likely about average.

Very few repairs continue to be reported for the 2010 Mazda3–excellent for a car in its first year. The 2009 Mazda6 has settled in not far from the average.

The 2010 Mercedes-Benz E-Class and GLK are both about average. As is the 2009 C-Class, the first model year for that car. All of these models are doing far better than the GL-Class in its first two model years.

The 2010 Nissans are consistently better than average. However, we only had enough responses for full results for the cube and Rogue. A number of models started the 2010 model year relatively late, so we’ll have full results for them with the next update, in November. Looking at the 2009s, the Altima, Maxima, Murano, and Rogue are all better than average, while the 370Z and Versa are about average but not far from “better than average.”

A number of owners of the redesigned 2010 Subaru Legacy and Outback have reported problems with steering vibrations. A few cars have even been bought back when these problems could not be fixed. Even so, the reported repair frequency is better than average. The Forester is also better than average.

Despite the recent media circus and claims that Toyota’s quality has declined, the 2010s continue to be consistently much better than average in TrueDelta’s Car Reliability Survey. We don’t have full results for many 2010 models yet, but the data we do have strongly suggest that these cars rarely require repairs. Looking across all recent model years, Toyotas fare better than any other make.

The 2010 Volkswagen CC appear to have fewer problems than the 2009, and is about average. The MkVI Golf, GTI, and Jetta SportWagen continue to lag the MkV they replaced. The gas models are about average. The TDI remains worse than average, but not by nearly the same degree as three month ago. An early common problem with the O2 sensor has likely been resolved, so this result should continue to converge with that for the gas models. The reported repair frequency for the Tiguan has jumped quite a bit with this update, and while “about average” is not far from “worse than average.”

Finally, the 2010 Volvo XC60 is about average so far.

For detailed results for all model years:

Car Reliability Survey results

If German cars had a stellar reputation for reliability, Lexus would not be where it is today. TrueDelta’s latest Car Reliability Survey results, based on owner experiences through the end of March 2010, provide some evidence that a corner has been turned, but other evidence that work remains to be done.

Two years ago the current Mercedes C-Class had a relatively trouble-free launch, and in the latest results the 2008 is better than average. Nearly three-quarters of owners haven’t had a single repair in the past year.

The redesigned-for-2010 Mercedes E-Class appears poised to go down the same path. TrueDelta’s first reliability stat for the car, 56 repair trips per 100 cars per year, is very close to the average for all cars. For an all-new car with above-average complexity this is quite good. It’s also far better than the record compiled by the make’s SUVs.

Initial stats for the new MkVI Volkswagen Golf, GTI, and Jetta SportWagen are mixed. Gas-powered 2010s barely managed an “about average” score (74, lower is better) while the diesel-powered TDIs, plagued by faulty O2 sensors, scored considerably worse than the average (143). Both scores are considerably worse than those for the 2008 and 2009 model year cars.

All of the updated results:

Car Reliability Survey results

Every year Ford hires RDA Group, a market research firm, to conduct a reliability survey for it. Mostly it uses the data internally, but it also usually issues a press release touting any Ford wins.

This year Ford announced that, based on this survey, they have the auto industry’s highest satisfaction with quality.

This survey also measures things gone wrong, which I can relate to. Satisfaction with quality or reliability is a different animal. While it will track with reliability, it also includes a subjective component and permits owners’ feelings about other aspects of the vehicle to affect responses. Like how the car drives? Then a small problem or two might seem okay, and you’re still “satisfied.”

Expectations also play a large role. If you expected many problems, you’ll be quite satisfied if you only have two. Expect zero, and you’ll be dissatisfied with the same two.

Also, read the text and you’ll find that they’re not actually claiming to be the highest of any manufacturer, but of any “full line manufacturer.” Who does this include, only GM, Chrysler, Toyota, and Nissan? No one else offers a large pickup.

Since TrueDelta promptly updates its Car Reliability Survey results four times a year, we can report on new models ahead of anyone else. Last year, we announced that the 2009 Jaguar XF was faring poorly. This provoked a blistering backlash from owners at a particular Jaguar forum. In the end, threads on reliability were deleted and future ones all but banned in the interest of preserving what remained of the UK auto industry.

The outraged owners argued that TrueDelta’s results could not be correct, since Jaguar had just been declared the most dependable make by J.D. Power. I pointed out that the VDS covers the third year of ownership, 2006 in that case, and that Jaguar had discontinued, redesigned, or replaced every model in its line save the XJ in the interim. So the results did not apply to the XF, or the current XK for that matter.

Well, J.D. Power has now released the 2010 Vehicle Dependability Survey (VDS), which covers 2007s in their third year of ownership, and, as predicted, the redesigned XK has, all by its lonesome, sunk Jaguar’s ranking from 1st to 23rd. And it’ll only get uglier once the XF is reflected in these stats in another two years.

#1 this year: Porsche. Many people will wonder how Porsche fared so well. One likely factor: Porsches are often weekend cars that aren’t driven much. J.D. Power might consider doing what TrueDelta does, and post average odometer readings. A larger factor: THERE WAS NO 2007 CAYENNE—Porsche skipped straight from 2006 to 2008. The Cayenne is likely more troublesome than the sports cars, and is certainly driven more. So don’t expect a top VDS score for Porsche next year, when the Cayenne is again part of the mix.

“Long term” for J.D. Power continues to mean “the third year of ownership.” It used to mean the fifth year, but manufacturers have little use for fifth-year data, and this survey primarily exists to serve manufacturers willing to pay large sums for detailed results.

Many car buyers, though, are much more interested in how cars fare after the 3/36 warranty ends. J.D. Power has no information for them, hoping that car buyers will accept third-year problem frequencies as a sufficient indicator of how a car will perform over the long haul. Unfortunately, in many cases it is not. TrueDelta’s data suggest that all too often cars take a turn for the worse either soon after the warranty ends or after 100,000 miles.

As usual, the public gets brand-level scores rather than model-level scores from J.D. Power. Brand-level scores are of limited use for a car buyer, and can actually misinform as much as they inform. After all, people don’t buy the entire line. They buy a particular model. And the scores of models can vary widely within a brand.

Much is made of which brands did better this year (Porsche, Lincoln), and which did worse (Jaguar). Well, as noted above, the brand averages can be heavily influenced by the introduction of a single new design or the absence of a single old design.

For these and other reasons a focus on model-level scores would be much more valid and useful.

Also worth noting: as in the past most makes are tightly bunched around the average, 155 problems per 100 cars this year. Consumer Reports considers any score within 20 percent of the average in its own survey to be “about average.” Applying this metric to J.D. Power’s results, 21 of the 36 brands are “about average.”

J.D. Power notes that for Cadillac, Ford, Hyundai, Lincoln, and Mercury perceptions of reliability lag reality. No surprise, since (as I’ve found all too often) people often judge (and more often than not reject) data based on how these data fit their perceptions rather than judging their perceptions based on how they fit the data.

J.D. Power’s explicit solution: convince consumers of gains in reliability. The implicit solution: pay to include VDS results in your ads. But are perceptions based on the VDS any more likely to be correct? Or, as seen in the Porsche and Jaguar cases, are they just as often part of the problem?

When looking over the results of our Car Reliability Survey, I check for consistency across model years. As the sample sizes grow, they’re consistent more often than not. When one result is out of line with the others, the question is then whether we’re looking at an inaccurate result, or whether that particular model year might actually require more repairs. The latter is most often the case when a design was in its first model year.

For example, number of repair trips per 100 cars per year for the Honda Accord:

2009: 34

2008: 26

2007: 36

2006: 18

2005: 28

2004: 34

2003: 82

With the possible exception of the 2006, the results from 2004 through 2009 fall within a tight range. This is generally what we see with sample sizes of 50+, as we have for all of these.

But what happened with the 2003? Unlike the others, it’s only about average. Well, this was the first model year of the 2003-2007 Accord, and it has consistently had a higher reported problem rate in our survey.

The 2008 was also a first model year car, but has done better. Honda apparently got most of the bugs out before launching this car.

Another example, the Audi A4:

2010: 30

2009: 64

2008: 48

2007: 48

2005.5-
2006: 78

2005: 154

2004: 135

2003: 119

2002: 159

The 2009 was the first model year of the latest A4. The 2010 is faring better, as is often the case with the second model year.

But the most notable difference is between the 2002-2005 and the 2005.5-2008. The reported repair rates for the more recent group are far lower than those for the older group. Well, the A4 was revised halfway through the 2005 model year, hence the 2005.5 designation, and the revised car has required far fewer repairs. Even with its “first model year” disadvantage, the 2005.5-2006 has been doing much better than the 2005 car it replaced.

Our latest set of results:

Car Reliability Survey results

With the 2010 Chevrolet Equinox and GMC Terrain General Motors finally figured out what car buyers are looking for in a compact crossover. Dealers have had trouble keeping them on lots. But did GM work out all of the bugs prior to launching the vehicle?

With prompt quarterly updates, TrueDelta’s Car Reliability Survey can answer this question well ahead of other sources. The latest update, which includes owner experiences through the end of 2009, reports that 2010 Equinox and Terrain owners have been experiencing 94 repair trips per 100 cars per year. This is worse than average. This said, the reported problems have generally been minor and fixed quickly on the first attempt.

Other recent GM crossover launches, including the 2007 GMC Acadia and Saturn Outlook and the 2008 Saturn VUE, had similar stats at this point. Both later improved to average, and the same will likely happen with the new Equinox and Terrain.

While most recent GM launches have been worse than average initially, not all have been. The new Chevrolet Camaro has fared better: with 41 repair trips per 100 cars it’s about average, and the most common repair has involved loose bolts for the rear spoiler.

With prompt quarterly updates, we’ll track all of these cars closely as they get more miles on them. The stats for the Equinox and Terrain could improve as soon as the next update, in May.

Car Reliability Survey results

In its flash drive of the new Jaguar XJ, Autoweek suggests, “Promising results from the 2009 J.D. Power Vehicle Dependability and Sales Satisfaction surveys should help reassure buyers concerned with dependability.”

Unfortunately, the J.D. Power Sales Satisfaction Survey has absolutely nothing to do with dependability. And the 2009 Vehicle Dependability Survey covered 2006 cars, and so none of the models Jaguar currently offers in the U.S. With the new XJ, they’ve now all been redesigned.

In contrast, the Jaguar XF has done poorly in all reliability surveys, beginning with TrueDelta’s Car Reliability Survey, which runs as much as a year ahead of the others.

In TrueDelta’s latest update the XF has improved to the point where it’s “only” about 50% worse than the average car, but it had a very rough first year.

How will the XJ fare? We hope to have results first, as we did with the XF. Just a matter of getting enough owners involved, which can be a challenge with a low-volume luxury sedan. Know someone who has bought one? Please send them here:

About the Car Reliability Survey

We’ve updated the Car Reliability Survey results to include owner experiences through December 31, 2009.

The new email process, which reduced the number of emails to members whose responses were up-to-date, was a big success. For the past year the response rate has been about 36 percent. This time around it was 38 percent. In absolute terms, we received responses for over 15,400 cars, up from 13,140 three months ago.

This was enough to provide full reliability stats for 355 car models, compared to 295 three month ago, and partial results for another 291. Coverage for the “odds” stats increased from 99 to 125.

I’ll cover some notable results in future blog entries.

A big thank you to everyone who helped make these results possible. The more owners participate, the more information we can provide.

Over the last few months, the media have become increasingly critical of Toyota and its handling of the unintended acceleration problem. Recently, Ralph Nader joined the fray, charging that Toyota has lost control of its quality control process. Has it? Or is this problem indicative of a broader, deeper problem in how cars are currently developed, everywhere? Should this problem have been caught during development? The more serious concern could be how well Toyota and other auto makers learn of and fix the problems that customers experience.

Consider a few facts. First off, if the quality control lapse were remotely obvious, Toyota would have quickly pinpointed it once the reports of unintended acceleration started rolling in. They haven’t. First, back in October 2009, they blamed–and recalled–the floormats. Then, in January 2010, they blamed–and recalled–the gas pedal mechanism. Even now some instances and historical overviews indicate that neither of these is the real problem, and that said real problem might be in the software.

So the source of the problem hasn’t been easy to find, even once Toyota was aware of the problem. Discovering it during development, when not aware of the problem and so not looking for its source, would have been highly unlikely.

Second, the problem is rare. About 5,400,000 cars have been recalled for a problem that has been reported about 2,000 times. Even assuming that the problem has occurred ten times for every time that has been reported, we have something that happens in one out of every 250 cars.

To discover this problem, you would need to test several hundred cars, perhaps even several thousand cars. Well, these days all car manufacturers build far fewer physical prototypes than they used to. Much testing that used to occur in the real world now occurs in computer simulations. I doubt anyone still builds even one hundred pre-production “alpha” prototypes. A few dozen, perhaps.

Once production starts a few hundred “pilot” cars are distributed to employees to drive and note any problems. This brings up the third piece of the puzzle: the amount of time the car must be tested before the problem occurs. This clearly isnt a problem that happens every time the car is driven, or even during the first 10,000 miles.

During development, only a small number of prototypes are driven more than a few thousand miles. And hardly any of the pilot build cars accumulate more than a few thousand miles. Compressed development schedules play a role. Led by Toyota, auto makers spend far less time developing a car than they used to. This translates to less time for a problem to appear in a prototype.

Put all of the pieces together, and any problem that strikes a very small percentage of cars after these cars have been on the road for a while is not likely to be discovered during the car’s development.

Of course, there could yet be a smoking gun: it could turn out that someone did notice the unintended acceleration problem within Toyota, and they either decided not to pursue it or tried to pursue it and were blocked by others within the company. But there’s no hint of this yet.

Move beyond product development and Toyota becomes more culpable. The pedal recall includes one five-year-old model, the 2005 Avalon. Some reported cases include 2005 Camrys (though these aren’t included in the pedal recall). Even if a problem that affects a small percentage of cars didn’t pop up during development, it clearly started popping up once hundreds of thousands of cars were in customers’ hands. Dealers must have been aware of multiple cases of unintended acceleration by 2007 or 2008, and perhaps even back in 2005.

What system does Toyota have in place to learn of the problems car owners are experiencing and rapidly develop engineering fixes for them? Judging from responses to TrueDelta’s Car Reliability Survey, Toyota generally does a good job identifying and fixing common problems early in a model’s run. Which is why Toyotas generally continue to perform well in reliability surveys. Common problems are caught and fixed. In this sense, Toyota has not lost control of its quality control.

But the system failed in this case—which notably does not involve a common problem. Why? Does Toyota’s system focus much less attention on rare problems, even if they can result in fatal accidents? Does it track cars less closely after the first year or two of ownership? Either of these could be a contributing factor. But is anyone asking these questions?

These aren’t only questions for Toyota. All car makers should take Toyota’s current predicament as a wakeup call to improve their systems for learning of the rare but potentially fatal problems car owners are experiencing, thoroughly researching these problems, then fixing them.

Living in the U.S., I feel I’ve been immersed in discussion of car reliability my entire life. It’s a key factor in many car purchases here, and I assumed it was the much the same elsewhere.

So a Jaguar XF forum thread served as a real eye opener. Specifically, this post:

I’m a Brit by birth but have lived in USA for the last 15 years. The reason I stayed here is because the attitude towards business and service is uncompromising. You either deliver or you go under. I couldn’t stand the “Can’t complain, mustn’t grumble” attitude back in England any longer. So it frustrates me doubly because I really want Jaguar to succeed and the product has many of the right qualities to succeed, but they have to focus on fixing reliability before they start offering frivolities like the rotary gear shifters. Poor reliability will kill resale value which will kill the market here.

Do attitudes towards reliability differ so much in the US and UK? Are UK customers truly much more forgiving? Stiff upper lip and all that?