By on March 5, 2012

A FIAT is available in the United States for the first time in decades. It’s manufactured in Mexico. Volkswagen has an all-new Passat built in an all-new American plant. One of them appears to be defying expectations of unreliability. Which one would you bet on?

We now have an early indicator. TrueDelta has updated the results of its Car Reliability Survey (on our totally redesigned site) to include owner experiences through the end of 2011. Others won’t cover the months since last April until next October.

Our sample sizes are small for both car models—twenty of each. But if the results are at one extreme or the other than a larger sample size would likely yield the same conclusion. And these are at the extremes.

A big surprise: the 2012 FIAT 500. With only a single repair reported for those twenty cars during 2011, and that one back in April, the 500 seems to be much more reliable than anyone expected, at least so far. The calculated stat: 16 repair trips per 100 cars per year. If this keeps up Tony’s going to be about as busy as the Maytag repairman.

Then there’s the new 2012 Volkswagen Passat. Its calculated statistic of 147 repair trips per 100 cars per year is about three times the average. If we had responses for another twenty cars, and somehow none of them required a repair, the stat still wouldn’t be pretty. So Chattanooga has a problem. Or perhaps that’s had a problem? Sometimes manufacturers catch and quash bugs quickly. Other times they don’t.

With the last update we provided such an early indicator for the Nissan LEAF. In the fourth quarter of 2011 the LEAF continued to be virtually fault-free, with no non-software repairs for 56 cars. We don’t count software updates as long as they’re free.

This has helped the 2012 Ford Focus. It remains about average in the updated stats (52 repair trips per 100 cars per year), but would fare considerably worse if SYNC updates counted.

And the redesigned 2012 Honda Civic? About as reliable as the FIAT!

We’ll see how Volkswagen has been doing with those bugs, and whether the 500 continues to have few problems, with the next update, in May. The more people participate, the more models we can cover and the more precise these results will be.

To view the updated repair trips per year stats for over 600 model / model year combinations:

Car Reliability Survey results

Come across something interesting? Have a question? Post it in the comments.

Get the latest TTAC e-Newsletter!

153 Comments on “New FIAT 500. New VW Passat. Which Is More Reliable?...”


  • avatar
    NotFast

    Are 20 replies truly statistically reliable? I’d suspect ballot-stuffing as much as I’d trust unreliability.

    Similar to my shopping on Amazon: I don’t trust any review score with a limited number of replies.

    • 0 avatar

      I checked every single car with the FIAT. Nearly all were owned by members who joined with a previous car, often over a year before buying the 500. So unless quite a few people were playing a very long con the results are real.

      As for the 20–if you need a larger sample to detect a difference, then that difference won’t be meaningful to anyone buying a single car. I wouldn’t use such a small sample to make a small distinction. But, as noted in the review, when the stat is at an extreme then it’s a different story.

      • 0 avatar
        dmw

        “But if the results are at one extreme or the other than a larger sample size would likely yield the same conclusion.”

        Really? Are you saying that if I walked into a subway car and asked 5 people what their salaries are, and they each said $500K/year, then the estimate is necessarily likely even higher than if I, say, polled a actual random sample of a size yielding a reasonable confidence level? The suggestion that the more extreme a result in a small sample the more likely it would be to appear in the population turns the entire theory of sampling on its head. Heard of the law of large numbers? That cited statement would not be a reasonable concludion even if there was an unbiased/random sample, and I see no evidence that the sample here is unbiased or corrected.

      • 0 avatar
        CJinSD

        Have you ever known 20 people that all had the same car? How about five? If all five people you know that have a particular car say it is flawless or a disaster, do you have the confidence to draw a conclusion about the car? Suppose you really like a particular make of car. No other car has exactly the characteristics that you like. Eventually, you get one that is a complete lemon. Based on your previous experience and nobody else offering exactly what you want, you buy another one. It is a lemon too. At what point do you stop trying the same thing and expecting a different outcome?

      • 0 avatar
        etrnlrvr

        “As for the 20–if you need a larger sample to detect a difference, then that difference won’t be meaningful to anyone buying a single car. I wouldn’t use such a small sample to make a small distinction.”

        Huh? I don’t have a PhD or anything but if 10,000 of those cars have a faulty part and the 20 on your site happened to have been made before or after that part was used then I’d say a larger sample would be meaningful to a single car buyer.

        20 cars seems statistically insignificant for any meaningful analysis to me. Fun to look at sure, but not enough to make a serious blog post about IMHO.

    • 0 avatar
      tresmonos

      I like the idea of True Delta, but I’ll never accept it as a statistically significant representation of reliability. It’s a niche survey that is championed by a negatively charged auto blog. It’s a bias machine.
      There’s not way to check accuracy of data collected.
      No way to contact people who are hard to reach.
      Relies 100% on word of mouth and/or people who actively seek reliability data (more source for bias). It’s online thus inaccessible or hard to find for a large % of your population. Throw your sampling method out the window.

      It’s a great way to see the general color of your picture. But making any statistical inference from it is impossible.

      • 0 avatar

        One aspect of the research design that you might not be aware of is that the data are collected going forward beginning with the month someone signs up. Unless people are able to see into the future, they don’t know what they’ll be reporting. This is the next best thing to a truly random sample. And since there’s no way to get a random sample without a high non-response rate, which is also a major potential source of bias, it’s arguably superior to retrospectively surveying a random sample.

        We also word the survey in such a way to limit under-reporting. Other surveys ask people to report only problems they “considered serious.” With this wording, people can bias the results without lying. We ask respondents to report all repairs, even the most minor ones.

        There are other ways we minimize the chances of biased results,ut these are the big ones. Overall the results have been of much higher quality than I expected when I started. The methods have worked as intended.

      • 0 avatar
        jmo2

        “There’s not way to check accuracy of data collected.”

        I’m not aware of any survey that does independent confirmation – they are all based on consumer surveys.

      • 0 avatar

        We post all reported repairs to the site. If these were being handled incorrectly, participants would notice. No one else does this.

        Also, when I’m checking the data for errors I usually only see the vehicle code–I don’t know the model. Though when I see something like “HPFP replaced for the third time” I know what the make is…

      • 0 avatar
        tresmonos

        jmo: follow up with the survey respondent. Generally done via the contact info you leave at the dealer or on questonaire the manufacturer sends following your purchase of a new car.

        Michael: it’s great your survey is designed to eliminate bias via design. What would be more interesting is to see your sampling method. Where you get your data (links, advertisements, response rate, etc). I have filled out True Delta data before, so I’m familiar with the survey. It’s nice. I have filled out my info and have neglected the site since. There may have been follow up, but I haven’t been motivated to fill out additional info on my vehicle. I haven’t checked my junk mail fitler, either.

      • 0 avatar
        jmo2

        “questonaire the manufacturer sends following your purchase of a new car.”

        IIRC that info is strictly proprietary and is never released to the public.

      • 0 avatar

        We could analyze the sources of respondents, but I haven’t taken the time to do this. It’s pretty widely distributed among TTAC, forums, organic search (the source of over 3/4 of our traffic) and word of mouth. But I haven’t checked the actual percentages.

        Many one-time responses are discarded by design. It’s quite possible that your single responses wasn’t included in the analysis.

        The design does continue to evolve. I haven’t sent the email to members yet this month because we’re finalizing another round of tweaks. Another advantage of quarterly updates is more rapid learning. The latest update is the 22nd. With an annual survey this would have taken 22 years.

      • 0 avatar
        tresmonos

        jmo2: Never said that info was public domain. It was just a counterpoint to your thoughts that follow up conversations do not happen with surveys.

        Michael: I see a lot of value in tracing the source of your respondents. It’s also good that you track log-ins (or whatever you track) to determine data integrity. Good stuff.

    • 0 avatar
      DeadWeight

      I wouldn’t be surprised if VW is actually has the worst reliability, regardless as to where any particular model is built, of any make that is from a high volume manufacturer, in the United States.

      Land Rover & Jaguar would not count by the high volume manufacturer criteria, but probably only they would be as bad or worse than VW.

      I know that some claim Consumer Reports is not a sound source in terms of reliability particulars, but I think that they are, given that they literally survey hundreds of thousands of vehicle owners over the short and long term, and specifically break down problem/failure areas by component (e.g. transmission, cooling system, motor, electrical, suspension, etc.), and if you look at the VW reliability as graphed, they really begin to suffer major problems in about year 3, and then the problems skyrocket into the stratosphere between years 4 and 10.

      Anectdotally, this is perfectly consistent with feedback on VW reliability from an approximate dozen or so family members and friends who own or have owned VWs, and track my own experience in owning a 2006 Passat 2.0T.

      • 0 avatar
        Ubermensch

        This mirrors my experience as well with a 2001 Passat 1.8T. First 3 years were fine, then all hell broke loose. I suspect this is why many of the VW fan boys claim that reliability has improved always saying to look at the last couple of years.

        I don’t learn my lesson though and am still close to pulling the trigger on a GTI. Why does VW have to make their crappy cars so damned desirable!?

      • 0 avatar
        DeadWeight

        Often times, even pretty (nice interior/exterior), fit (knobs feel right and work well) and seemingly pure women (virtuous & loyal) have undisclosed genital flare ups that don’t make their presence known until well after the test drive.

      • 0 avatar

        According to my experience with cars I own(ed) Consumer Reports reports problem areas correctly. E.g. I bought slightly used Ford Focus and month or two later noticed that it had small leak from torque converter seal. It was repaired under warranty (Ford had extended warranty for that part). First they replaced seal and it did not help and then replaced torque converter. It was fault in design or something like that. Focus was first year model with 2.3L engine. I viola – some time passed and in new issue of CR I saw black mark against transmission problems for that particular year, everything else being marked white or red, and only for that particular year (they probably changed torque converter design next to fix the problem since I got one with problem fixed).

  • avatar
    threeer

    Interesting, to say the least. Although as you’ve said, 20 data points doesn’t statistically mean much. But hopefully the early trend continues for Fiat. As for VW, I sorely wish they’d show the same trend. Time will tell with more data…

    • 0 avatar
      carlisimo

      It just means the margin of error is pretty big. If the results are at one extreme, then that margin of error can’t change the results much.

    • 0 avatar
      Robert.Walter

      Thought Michael said that such a restricted sample population is questionable when making fine comparisons, but if the sample consistently biases in one direction, then the result – for that point in time and those conditions – is valid.

      When we tested our product, a key chassis safety part, we used sample sizes of (depending on the feature being validated) of about 25 – 30 pcs; if we needed a quick indicator, 6 pc was considered statistically relevant provided the results were consistent with a smallish deviation.

      • 0 avatar
        PenguinBoy

        I’m not in the automotive industry, but we use smallish samples for testing design changes. Half a dozen to a couple of dozen in many cases, depending on the nature of the change.

        I think Michael’s sample size is enough to draw some conclusions from.

        I also wouldn’t be surprised if the slow start to Fiat 500 sales has helped quality – it’s often easier to get quality right when production ramps up slowly.

    • 0 avatar
      d002

      I don’t see why its unexpected.

      The Polish made Fiat 500 in Europe has proven to be very reliable.

  • avatar
    Hildy Johnson

    Good on the Fiat. However, it is not really new – it had been around for several years in Europe before it was shipped here, and it shares its underpinnings with the Ford Ka.

    Apples and oranges.

    • 0 avatar

      They made a lot of changes for North America, including a different engine and different automatic transmission. And they’re making it in a different plant, likely using parts from different suppliers. Not nearly as simple as shipping the same car here from the same plant that’s been making it.

      Another thing to keep in mind: a single widespread problem will bury a car in surveys like this one these days.

      This happened with the Nissan Versa when it launched here. It was based on a car that had been sold for a while elsewhere, but two of the new parts often required repairs (remote release cables, tire pressure monitors).

      • 0 avatar
        Bowler300

        I was thinking about this and I have a possible explanation. By tooling up a plant to make an existing design, but by statute having to make some changes, you are given the opportunity of a “do over”. You can’t say “no changes to keep costs down” because you are required to meet the regulations of the new market. This gives you the chance to make subtle improvements without adding significant cost.
        I base this on the fact that Fiat made a very slight change from the European to the American headlight (or possibly tail light, I can’t remember which) size that is virtually unnoticeable. The article I read said that this was not due to regulations but did not say why. Perhaps an improvement for reliability or manufacturing ease? Anyway, this is the type of change that a new plant in a new market can give you.

      • 0 avatar
        Hildy Johnson

        Good reply, Michael. So I guess it’s lemons and oranges, then ;)

      • 0 avatar
        ciddyguy

        Bowler3000,

        I can’t recall specifically but one thing I DO know is that the standard halogen headlight on the 500 in Europe uses a parabolic optical reflector design, whilst the US version uses a bi-halogen projector unit instead and uses the infrared coated halogen bulb, known as the H1R2 bulb (also known as 9012)and no HID option available but as to the size, there maybe something to that as well, I forget now myself.

        The taillights did change, a little tiny bit. In Europe, they I think used a separate brake/tail light assembly as one of the lights were for side marking or similar function and the Euro versions had either a rear fog light or a reflector in the right taillight cluster where a backup light would go, the right hand drive units had this reversed and the other cluster had the backup light, so in essence a single backup light is used in Europe. The N.A. version has a backup light in each cluster and a single 2 filament bulb for brake/taillight, otherwise, from the outside, they look identical.

    • 0 avatar
      Toad

      One car appears to reliable, one car does not.

      Apples and apples.

    • 0 avatar
      Carrera

      I kind of agree with Hildy on this one. Yes the European Fiat is a bit different than the Mexican Made one, but let’s not forget, everything about the Passat is new : the factory, the people working there, the model of the car, everything. In Mexico, before making the FIAT for the North American market, they were probably making FIATS for other markets ( South American, etc). The Chatanooga plants is brand new. I am aware of all the problems that VW has experienced in not too distant past but I was hoping that they are learning from their mistakes. I don’t own a VW and I would’ve never considered one due to their spotty reputation. I drove them a few times ( for 2 months, a 115HP 2000 VW Jetta). My friend had a lot of problems with his ( power windows failed in all four doors, A/C, all within the warranty 2/24K at that time), but he loved that car so much that now he drives a brand new CC.
      As for me, I really like the new Passat TDI. Unless mazda brings the sky diesel in a year or two, I will buy the Passat TDI. I did assume that the first year will have problems but I don’t expect the same problems in two years.
      I have been on the TDI forum often although not a member and I don’t see too much complaining about the new Passat…just small rattles and cars with bad alignment from the factory ( which should be caught by the PDI at the dealership). Now I would never take a VW to the dealerships for maintenance…they are notorious for bad service (most of them at least) but I would seek one of the mechanics found on the VW TDI forums. It seems that those guys know a lot more about the cars than the dealers do.

      • 0 avatar
        seatiger

        The Toluca, Mexico plant that makes the Fiat 500 made the Chrysler PT Cruiser for the North America market and now it makes the Dodge Journey for the world market.

  • avatar
    twotone

    Long-term (5+ years) reliability statistics are more important to me than what happens during the first year. I have yet to buy a car newer than five years old.

    • 0 avatar

      As soon as these are possible we’ll have them. A key advantage of promptly updating these statistics four times a year is that we can track these cars much more closely as they age. No need to wait a year until the next update.

      TrueDelta participants have even less of a wait–we’ll be previewing the next update in just a little over a month.

  • avatar
    carguy

    A sample of only 20 is modest but the magnitudes are significant. The Fiat 500 results are surprising. The VW results, I’m sorry to say, are well in line with what they have been doing for the last 15 years.

    • 0 avatar
      Acd

      15 years? VW has been plagued with reliabilty issues since the 1970′s. Every new generation promises improved reliability but so far none has delivered.

      • 0 avatar
        benzaholic

        I’ve owned a couple of VWs, both purchased new.

        The 1984 Jetta, last of the Mk I years in the US, assembled in Germany, was an outstandingly reliable and eminently tossable little beast for the 130K or so miles I owned it before selling it.

        The second one I’ve blocked out of my memory a little, but I think it was a 1990 Mk 2 Golf, assembled in Mexico during the month of May. My guess was that it was assembled on May sixth, while everybody was still hungover from Cinco de Mayo. Other than the immensely useful form factor of the Golf, this one was a turd.

        So my personal data set includes a VW among the most reliable and enjoyable cars I’ve owned. Again, that one was assembled in Germany.

      • 0 avatar
        windnsea00

        Except Mexico doesn’t really celebrate Cinco de Mayo…

  • avatar
    jmo2

    “In the fourth quarter of 2011 the LEAF continued to be virtually fault-free”

    As I’ve said before I think that is going to be the biggest selling point of EVs – they will be orders of magnitude more reliable than ICE vehicles.

  • avatar
    tikki50

    I have to agree I care little about repairs while the car is new, I get it something wasnt installed correctly and that SHOULD be caught before the customer gets take delivery. Plus all these cars are under warrenty, who the heck cares all most costs are covered. However Im more concerned about 5 years to 10 years out. Who the heck drops 30K-50K every 3 years on a car/truck? Oh wait the 1% does I forgot. I think thats what made Honda well Honda, and Toyota well, they never really hit the numbers like Honda but are much better off than others over time. And I’ll toss in some reality for GM, they aren’t there yet and until they can build realtively problem free car that goes 200K, Im not buying one because in my eyes THAT’s the benchmark for quality.

    • 0 avatar
      Toad

      Between 12 and 15 million people will buy new cars this year in the US, and to most of them new car reliability is very important.

      Initial reliability is usually a pretty strong indicator of future reliability, and poor initial reliability can cripple a car model, and even the automaker itself. It also makes a difference in the long run since reliability affects resale value as well as the operating costs for subsequent owners.

      Obviously, even if new car repairs are covered under warranty nobody wants to have their new car in the shop, and since most dealer service departments don’t offer loaner cars the hassle factor is multiplied exponentially.

      So who the heck cares? Virtually everybody buying a new car.

      BTW I hate the initial news about the new Passat. I had high hopes for the new VW and their new assembly plant. I hope that VW’s US assembly history does not repeat itself.

  • avatar
    tankinbeans

    I just checked the updated numbers for my car and they are pretty good, although I forgot what the * means on the site and can’t find the explanation.

    My friend has an ’11 Jetta, since March – now at 8700 miles – and hasn’t had any issues. Of course the 2.5 is supposedly the most solid engine they offer; I could be wrong though about this last point though.

    • 0 avatar

      The * means that the sample size is small.

      The new Jetta has been about average so far, which means that over half of owners will have no repairs at all over the course of a year. With the most reliable cars the percentage with no repairs is over 90 percent. We have stats like these on the site where the sample size is large enough.

  • avatar
    morbo

    I would include software updates as dings against the manufacturer, unless the update is applied while the car is not in use; kind of how my Sirius can do a quiet update on gas prices/weather/sports while my car is parked. Cars are as much computers nowadays as they are mechanical and chemical devices.

    I just hope the type of reliability testing historically performed (A0, Ai, MTBCF, etc.) normally performed on manufacturing is being performed on these touchsccreens and software packages.

    • 0 avatar

      My logic here is that software updates are usually somewhat voluntary–the system works without the update, which is how it got out the door in the first place. The line between a repair and an enhancement gets so thin that I wouldn’t like to have to split it. Also we wouldn’t want to discourage manufacturers from offering them. As is, they are getting so common that for many cars they’d be the great majority of repairs if they did count.

      • 0 avatar
        Robert.Walter

        Michael, I was going to offer similar commentary about the inadvisability of excluding software issues. I would think that an OEM isn’t going to take fewer customer enhancement actions because he might have been called out for getting it wrong first time thru.

        Sync’s poor functionality, or similar, is a significant TGW and items like this are arguably more significant in terms of satisfaction and safety than an NVH or other CSI issue.

        As s/w plays a bigger role in UI and vehicle control, excluding it, will make the TD results less representative of the actual experience and consequently less relevant overall.

        If I have to first lose patience with the product, and then time going to the dealer for a remedy, be it h/w- or s/w-related, then I’d like to know it.

      • 0 avatar
        srogers

        I think that not counting software updates is correct. In the world of software, updates are a routine and expected thing. Counting these as a defect would be like counting each time you top up your tire pressure.

      • 0 avatar
        FromaBuick6

        “The line between a repair and an enhancement gets so thin that I wouldn’t like to have to split it.”

        “Enhancement” is BS marketing speak used to downplay a previous defect or substandard part. Frankly, I can’t believe you’d even use the term, Michael.

        Sorry, but no, a trip to the dealer is a trip to the dealer. And that’s an irritating waste of my time. To dismiss software updates because they “work without the update” is a cop out. By that logic, squeaks, rattles, paint blemishes, along with every blanket recall for potential hazards should be left out, because the cars still “work.” After all, they “got out the door in the first place,” right?

        “Also, we wouldn’t want to discourage manufacturers from offering them.”

        Right, because manufacturers only offer updates out of the pure goodness of their hearts. On the contrary, my experience with Ford makes me believe they wouldn’t have lifted a finger to address the SYNC/MFT glitches if the Consumer Reports crowd didn’t beat them over the head with it. That software’s obviously not fully baked. Therefore, it’s a defect. Period.

        Michael, given how notoriously picky and thorough you are, I’m disappointed you’re letting manufacturers slide on this.

      • 0 avatar
        NulloModo

        There are different levels of software glitches and updates. If a particular vehicle has a software problem that prohibits it from working as advertised, I’d consider than update to be a repair. If the software update just adds extra functionality on top of what was promised, that shouldn’t count.

        Some of the Sync updates have been to fix stability problems with the system, some have been to update compatibility with new phones and devices that weren’t out when the previous software version was launched. In 2010 there was an update to activate the functionality of turn by turn directions and realtime traffic/weather. Early 2010 models had the hardware to support the features, but the software wasn’t ready so they weren’t advertised as having the capability. When the software was available those owners got bonus functionality through a (self installable) software update.

        It becomes tricky when talking about phone/media-player integration to track down where the fault is if there’s a problem. Our dealership had a couple buy an Edge recently and the husband’s iPhone connected perfectly with Sync, doing everything it was supposed to to, while the wife’s off-brand older phone would connect via bluetooth but not allow a phonebook download. She felt as if there was a problem with the car, but the car was working exactly as it was supposed to, her phone just didn’t support downloading the phonebook data.

        On the plus side, the MyFord touch related complaints are nearing an end – the USB drives and new map cards for the major software revision started shipping this week, and most dealerships should soon have a stock of available drives and cards to update the software for those who don’t want to wait for their personalized packet in the mail. Over the next month or so all dealer stock on lots should be updated, and all 2013 models are shipping from the factory with the new software pre-installed.

      • 0 avatar
        tallnikita

        Guess you are too young to have suffered through early MS Windows “updates”

      • 0 avatar

        I understand where you’re coming from, FromaBuick6. And if there are many more cases like SYNC, I might have to revisit this policy.

        I initially decided not to count software updates before SYNC and such were even around. Instead, I wondered whether transmission updates should count. A number of manufacturers were frequently updating the software of their new transmissions, and since these were quick and free I decided that every time the transmission software was updated should not count as a repair.

        This is a difference with software fixes: there is often a series of them. I do think that counting each one as a repair would discourage manufacturers from offering interim updates as they evolve their software. Instead, they’d delay offering software fixes until they what they felt was the final fix. Good in theory, but in practice it would mean a much longer delay before the fix was offered.

        Aside from the issues I’ve already noted, in practice counting software updates that fixed a clear problem would require hair splitting, and hair splitting requires more detailed instructions than many people are willing to read and more detailed responses than many people are able to provide.

        We do still post these software updates to the site, where people researching the problems a model is having will see them, we just don’t include them in the count.

      • 0 avatar
        redav

        In my experience – companies do not offer free ‘upgrades’ to improve performance. They only offer stuff for free if there’s something wrong with it and the alternative to not fixing it would cost more.

        A few key indicators stand out for me:
        - Do things work the way they are supposed to? If the software does not, it is a defect.
        - Does it involve a trip to the dealership so they can make a change? If it does, it’s a defect not unlike a recall.
        - Is the update part of a known, scheduled maintenance plan? If yes, like map updates for nav, then it’s fine. If no, like the MFT update being put on a USB & mailed to owners, then it’s a defect.

        A huge problem is the fact that microsoft has so damaged our expectations of computer reliability that we do not expect or demand computers & software to actually do their jobs. Somehow MFT crashing and rebooting has become par for the course, but if the same thing happened with any mechanical system in the car, it would be pulled from the road.

        Now for my personal rant: Reliability studies that do not include these defects perpetuate the problem and give software writers a pass when they would fail in any other field.

  • avatar
    brettc

    If you check out the B7/NMS forums on tdiclub.com there are posts from people saying that VW has been flying engineers around the U.S. trying to fix major problems that have cropped up. So while it’s good that VW is apparently taking ownership of the problems, it’s also crappy that the cars are buggy to begin with.

    But it’s good to see that early indications are showing reliable Fiats at least. Maybe they’ll lose the “Fix it again, Tony” association in the U.S. and maybe it’ll even result in some higher sales eventually.

    As for Chattanooga and the Passats, that’s why it’s not a good idea to buy a brand new model from a brand new factory. Especially when the manufacturer is VW. Give them a couple years to work the bugs out, then we’ll see how they fare long-term.

    • 0 avatar

      I fully expect the Passat to improve to the point where it’s in the “about average” range for the first 3-4 years of ownership. The question is how quickly they’ll get it there. No doubt some of their people are feeling the heat quite intensely, given the importance of this product to their North American ambitions.

  • avatar
    dima

    Can someone please tell me why in Europe people regard VW as a quality cars? I have family in Germany, and they love theirs VW’s. I know that some of the things have to do with regular maintenance of the car. I do know people in US who are religious about maintaining their VW’s but the final result is, they also have to go to garage to pay for the repairs most people do not incur on other car brands with similar year/mileage. So, what is it?

    • 0 avatar

      It’s not a matter of the amount of maintenance until cars are 5+ years old. And even after that point most repairs involve parts that cannot be maintained.

      Instead, people in Europe seem to have a much higher tolerance for problems that don’t keep the car from being driven. You’ll find more people there who don’t count minor repairs against the car.

      How often the car is driven is likely a factor. Fewer people there use their car to commute to work, so taking it to the shop is less of an inconvenience.

      It wasn’t too long ago in the U.S. that everyone expected to have to take the car to the dealer at least once to fix problems at delivery. These days most cars are delivered with absolutely no problems, and expectations have changed accordingly. The same shift doesn’t appear to have happened in Europe, at least not yet.

      Expectations of how long a car should last have also been increasing, probably in the U.S. more than elsewhere. My gut sense is that the average American car owner now expects no major problems for the first eight years and 120,000 miles. When I was growing up, many people considered 100,000 miles the expected lifespan of a car. Now that’s probably around 160,000.

      • 0 avatar
        dima

        Thanks Michael for validating my guess.

      • 0 avatar
        George B

        In addition to the points Michael makes, I get the impression that the major Japanese manufacturers like Toyota make engineering decisions favoring reliability over incremental improvements in performance. For example, the 2012 Toyota Camry doesn’t have direct injection and the automatic transmission is a conventional torque converter slush box while Volkswagen uses direct injection and an automated manual on the higher-end Passats. Based on past reliability of the 2010 and 2011 Camry engine and transmission, I would expect the same units to also be very reliable in the 2012. Volkswagen, on the other hand, is the Lindsey Lohan of auto manufacturers with somewhat lower expectations for reliability that they need to overcome. I have seen many 2012 Toyota Camrys on the road in the Dallas area compared to maybe two 2012 Passats.

      • 0 avatar
        Robert.Walter

        I recall seeing a sign about 20 years ago, above the main door leading into the chassis engineering department, in the Ford light-truck engineering building, also known as building one, which said: “life begins at eight years, or 120,000 miles”.

        PS I no longer recall clearly, but it could be that the sign said 10 years instead of eight years.

    • 0 avatar
      johnhowington

      the 2012 USA VW passat is manufactured in Tennessee.

  • avatar
    AMC_CJ

    The wife liked the new Jetta when we saw it at the recent auto show. But after of being under the hood way too much on her older model, and looking that car over at the show, there was no way in hell I was going to put another VW product in our driveway.

    Got a new Mustang, and we both love it.

  • avatar
    ajla

    I’m surprised some people are asking for reliability data on 10-year old, high mileage stuff.

    • 0 avatar

      I go back and forth on whether to provide stats on cars over 10 years old. Maintenance is a huge factor at that point. Also, many of the parts are probably no longer OEM. So these data are inherently problematic.

      But many people requested that I cover older cars, so I’ve been providing the stats. I can see where they’re coming from. My Mazda is nearly nine years old, and I’m still interested in tracking the model.

      A possible solution: different stats for older cars that have more exclusions, since at that point many people live with minor problems.

      • 0 avatar
        srogers

        I think that beyond a certain mileage/age, reliability stats would become self-fulfilling prophesy. Owners of Accents and Cobalts let them degrade while owners of more desirable and/or known reliable cars (ie. Accord, Camry, BMW 3 series) will happily do what is necessary to maintain their investment. As Steve Lang says – “it’s the owner, not the car” – at least as the years and miles rack up.

      • 0 avatar
        ajla

        It sounds like a nightmarish job to me. On an out-of-warranty vehicle, there is a ton of stuff people would want to know.

        You’ve got number of repairs, cost of repairs at a shop, cost of DIY repairs, time spent on DIY repairs, days out of service, cost of maintence, effect of driving environment, quality of replacement parts, availablity of replacement parts, and I’m sure a bunch of other stuff.

        I can’t think of any metric you could use that would satisfy many people.

      • 0 avatar
        Sinistermisterman

        @srogers
        Whilst I would normally agree with you, subjectively my Cobalt has already had 2 warranty repairs in it’s first year on the road. One minor (door seal), one relatively major (steering column – no, not the recall).
        In my experience of new cars (my own and family members), my Cobalt has been the problem child. Despite all this, I’ve babied the thing, it is what Steve would call a ‘Dealer Queen’.

      • 0 avatar
        ciddyguy

        Michael, I agree about the tracking of older cars as I have the same car you do, even the same year (2003) and it WOULD be good to know how cars of this age fair but what I would also do is ask owners if they are long term owners, ie, had it when the car was new or MUCH newer in age in relation to its age now and mileage and perhaps how they maintained it/driven it to get a base feel fore the car itself.

        Then from there begin to track it and see how it fairs, noting larger issues as they crop up, say, a tie rod goes south at 130K miles, a wheel bearing at 140K, the timing belt, where present (though that is a maintenance item and needs to be done at a set period of time, say, 60-90K range for most makes), things that force the car off the road for at least a day while the repair commences.

        I bought the Mazda because my nearly 20 YO Ford Ranger developed too many issues near the end and I had to give it up. But despite regular oil/filter changes at roughly 5K mile intervals, the vehicle remained very reliable with two major repairs in 6 years, the clutch master and slave cylinders had to be replaced, the master in late 06, the slave in early 09. Outside of that, new tires and exhaust were it outside normal routine maintenance. I took that truck from 189K to just shy of 237K in that time and it always started and ran without complaint.

        Even at the end with an adjusted throttle, I was able get the truck to the dealer to trade it in just fine even with a bad idle air controller valve reliably too as the truck started up without issues, never mind the shifter being all wonky as it just went out, again (had repaired it the first time in 2009). The truck even still passed emissions on its last check in 2010 when I renewed the tabs for that year.

  • avatar
    replica

    A VW unreliable?

    Impossibru!

    • 0 avatar

      In VW’s defense, most of their 2011s are average or better. No repairs were reported through the end of 2011 for the 18 GTIs in the survey at that time. The new Jetta has been about average. So far the new Passat is much worse than the recent VW average.

      • 0 avatar
        DeadWeight

        Michael, you are way too hung up on reliability claims based on such short time frames, IMO.

        1 to 3 years of even zero-problems mechanically does not a reliable engine make.

        When the same can be said about that mill after 5,6 or 8 years, with 70K miles plus on it, now we’re talking.

        I would also bet good money on the following (I actually betting good money on it by not buying one, until it’s proven to be no longer true): The wave of small displacement motors that are turbocharged tied to relatively heavy vehicles (the Cruze is a good example with a 1.4 Liters turbocharged mill) are going to be problematic and expensive to fix. There’s no getting around the laws of thermodynamics, and the claims that advanced materials such as ceramics and more robustly constructed motor components will prove that such engines will be reliable over the long term are unproven claims at this point.

      • 0 avatar
        psarhjinian

        “1 to 3 years of even zero-problems mechanically does not a reliable engine make.”

        The thing is, for most people, it’s not about the engine. Heck, for Volkswagen it’s often about any one of a number of things that are only tangentially related to the engine.

        Car enthusiasts don’t seem to understand this: consumers don’t see or care about the theoretical advantages of a Volvo 240′s overbuilt engine block or that of a W123 Benz, and how either can go a million miles **if you keep up with the maintenance**.

        They see the yearly dollars associated with keeping a car running, and for the money you can’t beat, eg, a Toyota Corolla, which might not manage a million miles given an owner’s care, but can easily do a quarter-million miles of relative neglect and abuse on the cheap without a regular supply of window regulators, ignitions, front suspension components and so forth.

        If, and this is a big if, Fiat can deliver something approximating the Corolla experience, then they stand a reasonable chance. On the other hand, if VW can’t (and it’s dicey; they’re not consistent) then all this effort to build cars for the American market, at the risk of their semi-premium status, is for naught.

  • avatar
    geeber

    What is it with 2010 Hondas and Acuras? EVERY 2010 model reports a high number of repairs, but the other model years – including the 2009 and 2011 model years – are all either average or better.

    • 0 avatar

      Thanks for the heads-up. It looks like the programmer just broke something–will get it fixed ASAP. This bug only affects the page where all stats are displayed.

      Until then, look at the actual numbers, not the faces, or at the pages for individual models.

  • avatar
    Secret Hi5

    My personal experience based on multiple rentals of Fiat 500s and the latest VWs: The Fiat interior felt flimsy, but nothing was actually broken. The VWs seems to be higher quality, but there were actual problems.

    My unscientific conclusion: VW is actually bad, and Fiat is “to be determined.”

    • 0 avatar
      JMII

      This mirrors my experience with a previous Passat. It looked and felt great (for the first four years), but had tons of random (yet common to VW) problems. End result: VeeDub off the list forever! The wife refuses to buy another one regardless of where its made or what any survey says, she hated what became of that purchase. Ironically if we had sold the car after 3 years she would have gladly bought another VW since at that point all was fine with it. However 7 years later she wanted to donate it just to get rid of the constant problem. Prime example: on the way to trade it (for my 350Z) the cruise control failed and panel that houses the sunroof controls broke for the second time.

      Its hard to escape the feelings (both good and bad) from a given manufacturing when you’ve dumped so much money into their product. After all these aren’t $100 DVD players we are talking about.

      • 0 avatar
        DeadWeight

        I can’t speak for any Italian car, having never owned one, but after owning or leasing approximately 15 vehicles, Japanese, Domestic and German, all I have to say is the most reliable cars were Japanese (and that would be 5 of those cars). On all but one of those vehicles (an early 1990s Nissan that had an automatic transmission eat itself with only 23k miles on it), from 1993 through my present one, routine maintenance was all that was required.

        The two German vehicles (a bimmer and VW) were nightmares that, unless a true revolutionary step forward in reliability is achieved by German manufacturers, has permanently sworn me off any future German cars no matter how sexy or dynamic any make/model may be.

        The domestics were a mixed bag. I had a Ford that had exactly one problem in about 40k miles, a Jeep that had one minor problem in 50k miles, a Dodge pickup that was flawless over 4 years of ownership, and the rest were relatively unreliable, with two General Motors vehicles from the late 80s that were abominations of the highest order.

        The Ford and VW dealers were the worst customer experiences I’ve ever had, by far, with any product or company, in my lifetime.

  • avatar
    kurtamaxxguy

    Granted, sample size is small. But given the “usa” Fiat 500 is doing well despite being made in a new plant suggests whoever set up that plant was very thorough, possibly doing an extended pilot run to make sure production kinks were worked out.

    WRT VW and AUDI, new products seem buggy (the Audi Q5 had a rash of water pump failures for the first year, along other things). For VW and AUDI in general, have their first year product reliability proved much worse than average?

    • 0 avatar

      See my reply about four comments up. This is NOT typical of VW and Audi lately.

      The bad water pumps were in all Audis with the 3.2, though.

    • 0 avatar
      threeer

      If I recall correctly, the 500 is not made in a greenfield (ie, new) plant, but one that was being used to produce PT Cruisers. the VW plant in TN is a brand-new facility. I’d assume there are some economies of scale associated with using the existing plant versus a new start up…

  • avatar
    hifi

    From the sounds of it, this makes me question the process and samples used to measure reliability more than the actual reliability of these two cars. It also makes me question TrueDelta’s decision to publish information that is likely grossly inaccurate. A sample size of twenty? Really? That’s just not done.

    I’m inclined to ignore all of this until there is a reasonable sample size.

    • 0 avatar
      gslippy

      Twenty is a small sample, but I’d definitely call it an early indicator.

    • 0 avatar

      I don’t claim that these stats are precise. But they’re also unlikely to be grossly inaccurate. We do post the sample sizes and even all o the repair descriptions. People are free to disregard results as they see fit.

      • 0 avatar
        hifi

        Michael,

        Done. No statistician would consider a sample of twenty as a good indicator of anything. There’s likely to be some very pronounced swings in any direction with a larger sample. Looking forward to an update when you have a larger data set.

      • 0 avatar
        redav

        hifi,

        You are not entirely correct. For example, let’s say that a coin is flipped 20 times and comes up heads 19 times. A statistician would definitely have enough data to know that’s not a normal coin. What he does not have enough data for is to determine just how biased the coin is.

        Here’s a way of thinking about it: Suppose that the true reliability rating is x. Just like the coin above, a sample size of 20 with 19 ‘passes’ is enough to say with high confidence that x > 50%. It may even be enough to say that x > 75%. Is it enough to say that x is 92%? No.

        Thus, it is not true that a statistician would not consider it a good indicator, rather, it is a not a good indicator of a precise estimate, but it is perfectly fine for a broad estimate. The real question then should be how precise does the estimate need to be? If all you need to know is whether it’s better/worse than something else with a moderate level of confidence, it may be sufficient.

    • 0 avatar
      Ben

      Granted that it is a statistically small sample, however ask yourself if you bought 20 FIATs and there was only one problem among them would you not be impressed?

      Everyone expects the passion and fun that goes with an Italian badge, to have scored one with only oen problem, fantastico!

    • 0 avatar
      Lokki

      Hifi -

      DaymonRunyon once famously said,

      “The race isn’t always to the swift, but that’s the way to bet.”

      So, while I grasp your point, are you willing to bet $25,000 on your argument?

  • avatar
    Dimwit

    Michael, is there a commonality to the Passat’s problems? VW is notorious for having “a” problem, but it’s pervasive and hard to fix. Like the wiring harness on the gasser motors particularily the 2.slow.

  • avatar
    CrapBox

    My 2007 VW Rabbit has been faultless… at least until now.

  • avatar
    hubcap

    I’m glad to hear the Jetta is at average reliability. I’m looking to pick up a 2012 or 2013 GLI. I just want to wait to see what the Dart SRT4 will offer.

    FWIW sitting in my garage right now is a Jetta III GLX with 224,000+ miles. It still runs very strong and I’ve had people offer to buy it. I’ve owned since new (it was the car I purchased after completing Navy OCS) and don’t plan to ever sell it.

  • avatar
    Marko

    So, in non-USA built VWs, does the plant seem to make a statistically significant difference in reliability?

    • 0 avatar

      I don’t have the data to analyze this. But I don’t think problems are often the assembly plant’s fault. Most would appear to be due to flawed designs or due to a mistake at the supplier level. Consider the water pumps that failed in many Audis last year or the fuel pumps in many BMWs–they fail no matter which plant installed them in the car.

      Having said this, some of the problems with the new Passat have been rattles and wind noise, and these might be due to assembly. I say “might” because some designs are easier to assemble well than others.

      • 0 avatar
        Ubermensch

        That was my experience as well. VW problems are usually problems with the parts quality and in engineering. I had a few minor squeaks and rattles in the interior, but what were the worst problems were parts just failing outright for no reason. Bosch has been getting away with murder.

  • avatar

    I’ll be watching for stats on the Toyota FT86 and its rebadge siblings

  • avatar
    gslippy

    VW’s plans for world domination don’t look promising on the US front, anyway.

    • 0 avatar
      PintoFan

      How do you figure that? The Jetta and to a lesser extent the Passat are both major sales successes already. VW won’t manage a complete turnaround overnight in terms of market share, but they seem to have made all the right moves thus far.

      • 0 avatar
        gslippy

        The US domestics were sales successes, too, until their quality issues caught up with them.

        The myth of ‘German engineering’ is dispelled by most reliability data, and seems like it will always hold back VW.

        There are lots of former VW owners out there who won’t be back, and I’m one of them.

  • avatar
    Brian P

    I have a 2006 Jetta TDI, and it’s been in the shop the odd time for unscheduled maintenance. Fuel temp sensor went bad under warranty ($18 part), it has needed wheel bearings at all four corners and an ABS wheel speed sensor, and the occasional burned bulb … and that’s about it! Oh yes, the leather covering for the gearshift boot is starting to come off, and the driver’s floor mat has worn through.

    But … 364,6xx km and counting. No plans to sell it, because I like this one more than the new ones. At this point I can safely say that it’s the best car I’ve owned. Another 2000 km and it’ll be beyond the most that I’ve personally put on any car. Another 98,000 km and it’ll pass the total that was on my previous VW TDI when I sold it (I had bought that car used originally).

    Regular maintenance is by the book and then some.

    • 0 avatar
      28-cars-later

      Just amazes me how much a rarity TDIs are in this country when you hear about one who has clocked roughly 364K km/226K mi in maybe seven years (assuming your ’06 was built in summer ’05). I’ve heard the argument TDIs are niche cars and in the era of peak oil I just no longer believe it. These things should be flooded into the market instead of gas hybrids. The Volt is a cute toy but its baffling to think a fraction of its billion dollar R&D could have been spent adapting an Opel diesel for US use, even before the bailout. Just shows how backward the market is when this proven technology is almost forbidden by both the industry and the EPA but you can drive an ugly SMART deathtrap which runs what 35MPG combined? The up-and-coming 5spd Dodge Dart is shooting for 40MPG combined, and it looks like somewhat of a real car, and heck you may even survive a collision with one!

      • 0 avatar
        NulloModo

        Diesel engines can last nearly forever, but from what I understand it’s usually not the engines that give VWs problems as much as the electrical stuff.

        Still, assuming the car is perfectly reliable, most drivers won’t keep their car past 100,000 miles if they buy it new, many not even close to that. You also have to deal with being able to find diesel to fill it up. It’s readily available in almost any town, sure, but not at every station. Of the four gas stations I regularly use not a single one sells diesel, so I’d have to go out of my way to fill up.

        I’m not saying diesels are a bad idea, but comparing the take rate in the US to the take rate in Europe without taking into account the European diesel-first infrastructure, lower European diesel prices compared to gasoline, and in some cases lower taxes on diesel vehicles, doesn’t allow for an apples to apples comparison.

  • avatar
    richmich

    Personally, I would not bet on either. Volkswagen has a reputation recently for making cars that are not reliable and expensive to repair. I like the Fiat 500, but the three star crash rating is enough to look elsewhere. Reliability is not even a concern here.

  • avatar

    All the haters will keep saying it’s a chick car or something ‘worse’. They’ll never accept conclusions by reputed pros like you, Michael, since it doesn’t fit into their limited worldview.

    I try to ignore those mindless idiots. But sometimes I’d like to shove this kind of info their faces and say F you. You know what, won’t waste my time. Some of us will just enjoy this car.

    Keep up the good work.

  • avatar
    340-4

    I owned a VW for less than a year; I’ll never go back.

    2001.5 Passat.

    Yeah.

    I have no interest in the 500.

    However, the new Dart R/T… now that’s another story.

    • 0 avatar
      Strippo

      Ha! I had the same Passat. The only car I’ve ever owned (wife’s car) that spewed black smoke from under the hood (turbo seal). I would have pulled the trigger on a GTI last year (my car) but for that experience. On paper the GTI is the car for me, but no. Never.

      Never.

      • 0 avatar
        CJinSD

        A VW hater is just a VW lover that’s bought one.

      • 0 avatar
        mikey

        @CJinSD…..Best, and most honest comment so far.

      • 0 avatar
        mike978

        Mikey – really? I owned a MkIV Golf (99-02) and a SEAT Ibiza (02-05 and essentially a Polo) and both gave me no issues. So while I understand CJ’s comment it is only accurate for some owners. Not all.
        I assume his statement stands for those who bought other makes and had a bad experiences because n=1 seems to be the rule here!

      • 0 avatar
        hubcap

        Mikey,

        Didn’t you work for GM? Over the last few decades, if there was ever a company known for building pure unadulterated crap. For having clueless management and workers who couldn’t put together a jig saw puzzle much less a car it would be GM.

        Allow me to fix your statement.

        ” A GM hater is a former GM lover whose been burned numerous times in the past by their derivative engineering and slip shod build quality.”

        There ya go. Fixed it for ya.

      • 0 avatar
        mikey

        Hold IT! It was CJ that made the statement. I just agree with his view.

        IMHO VW makes a good low priced German car. That superior German enginering translates into a lot of shop time. On the other hand a BMW or a Merc spends just as much time in the shop,and costs twice as much money.

        So if I wanted the German car experience, I’d go with the V dub.

  • avatar
    ciddyguy

    The early stats on the Fiat doesn’t totally surprise me. I’ve been reading up on this car since 2009 and its European version and how it and the Panda that spawned it were one of Fiats more reliable models and things like switchgear don’t seem to break and seem to hold up quite well too.

    I also was keen on the developments of Fiat and Chrysler and became really interested in their development after Fiat announced it was bringing the Fiat and Alfa brands back to NA, including the US.

    What got me interested was out of curiosity seeing what/how Fiat was doing with current models and this was just before the bankruptcies took place too.

    One thing to remember with the 500 is that not only is it built in Toluca Mexico, but many of its parts are now sourced right here in the US and/or Mexico. The motors are built in Dundee Michigan, just for starters. While the cars may LOOK the same (with a few minor details like actual side marker lights and the standard Bi-HIR projector headlamps), it is very similar to the one in Europe, federalized changes not withstanding.

    Some parts got changes of their own, such as the radio used in the NA looks identical to the Euro version, but includes Sat radio, whilst the Euro one doesn’t. The US uses Bose for the source of its upgraded speaker system, though it looks just like the upgrade system, called Interscope used in Europe as they fit behind the exact same speaker grill design found in both cars,that kind of thing.

    So if these statistical results now hold up, this could bode very well for Fiat in the long run.

  • avatar
    toxicroach

    This is the kind of thing that brings up my old concerns.

    20 cars is not a statistically significant number of cars. You can’t derive useful information from 20 cars.

    • 0 avatar

      If you can’t derive useful information from 20 cars, then for the individual car buyer there’s no point in having any reliability information, regardless of the sample size. Because the implication is that you’d have to buy more than 20 of the same car to notice a meaningful difference. And most people only buy one.

      Statistical analysis is often used to determine if very small measured differences are significant. Seven percent of the people who took the placebo died. Five percent of those who took the new drug died. Did the new drug help? For this purpose, you need a large sample size.

      Another example: one reliability survey has a minimum sample size of 100. With this sample size, they use dots to indicate differences as small as one problem per 100 cars, and routinely make a big deal out of differences of 3-4 per 100–without ever clearly stating that the differences are this small. To split such a fine hair a sample size of 100 isn’t nearly sufficient.

      To report whether a car has few problems or a lot of problems when your data indicate one of these extremes, 20 will serve far better than 100 will in the above case. Of course, larger is always better. But information from a sample of 20 in this instance is much more useful than no information at all.

      • 0 avatar
        Robert.Walter

        This is the point I was making in my reply to threeer far above.

      • 0 avatar
        Pch101

        With a sample size of 20, the margin of error is about 22%. That sort of spread is large enough that the results could be completely wrong.

        I’ll take whatever data that I can get, but the error rate is inherently high, and it should be used with caution.

      • 0 avatar
        supersleuth

        Did you miss the part where the difference between the 500 and the Passat is a hell of a lot bigger than 22%?

      • 0 avatar
        toxicroach

        Well, you’re kind of doing a reducto ad absurdum of my argument, but you have quite a bit of truth to it.

        People on this website all the time show up with the “I have a Land Rover and the electrical system is made of awesome.” and “I bought a Corolla and I had to go through 3 transmissions in the first year.”

        So you are kind of right; buying a car is always a gamble. You might get the bad one of a good car or the good one of a bad car. But when you are buying a car, statistically significant reliability information helps you place your bet to maximize your chances of getting a reliable car.

        Placing your bets on information that may well be totally wrong you still might win, but you might also increase the odds of losing. The fact that its a bet doesn’t change the fact that you can’t derive mathematically sound statistics from 40 cars.

        To me this article took two cars that were having the opposite effect of a statistically insignificant number of cars, in that one looks freakishly reliable while the other one looks terrible, and compared the two for the sake of getting some hits. Can’t say it I think it was good journalism.

      • 0 avatar
        Pch101

        “Did you miss the part where the difference between the 500 and the Passat is a hell of a lot bigger than 22%?”

        I read the article, so I understand the results. But of the two of us, it’s pretty obvious that you are the one who missed the point.

        A 22% margin of error for a survey is extremely high. There’s just no getting around that. If it was a medical trial and the results were heavily skewed (i.e. a treatment was very successful among a small group), then that would call for additional research with a larger sample, with the understanding that the results are promising but could be wrong.

      • 0 avatar
        supersleuth

        You’re changing the subject. It is simply not true, as repeatedly claimed on this thread, that a large difference in the mean values of a statistic between two different populations can’t be statistically significant because the population sizes are only 20. It is in fact very clear that the difference Michael is reporting, given its very large magnitude, IS significant. Now, the quality of the underlying data may be an issue but it is is a completely separate issue. Even then, I see little reason to doubt that the 500 has in fact had a cleaner launch than the Passat.

      • 0 avatar
        Pch101

        “You’re changing the subject.”

        I correctly pointed out that the margin of error for a sample size of 20 (with an acceptable confidence interval) is about 22%.

        That’s just a fact. Unless you plan on reinventing statistics, nothing that you or anyone else can say here can possibly change that fact.

        Small data pools are something close to anecdotal in value. They can be heavily skewed. Get over it.

      • 0 avatar
        supersleuth

        You’re still missing the point. Confidence intervals can be as wide as all get out, but if the means in the two populations are so different that the intervals don’t even overlap at all, the difference clearly is statistically significant. If you want to calm that Karesh’s underlying samples are biased in a way that invalidates the comparison, you need to make a totally different kind of argument from the one you’re making. And one that involves more than just pointing at the sample size.

      • 0 avatar
        Pch101

        “If you want to calm that Karesh’s underlying samples are biased in a way that invalidates the comparison, you need to make a totally different kind of argument from the one you’re making.”

        I didn’t say anything about Mr. Karesh’s samples. I just pointed out — accurately — that the margin of error for a sample of 20 is 22%. Period.

        Again, there is nothing that you can say that changes this fact. The odds that his sample pool does not accurately represent the population are high, because the pool is small. When sample sizes are small, that’s just how it is.

      • 0 avatar
        supersleuth

        So you are in fact claiming that the samples are biased simply based on the population sizes and nothing else. That’s not a good argument.

      • 0 avatar
        Pch101

        “So you are in fact claiming that the samples are biased simply based on the population sizes and nothing else.”

        As I keep repeating (and this is going to be the last time that I say it), the margin of error for a sample of 20 is inherently high.

        I don’t know whether or not this sample of 20 is representative of the population. You don’t, either. Neither does Mr. Karesh. There is no way to tell. If we want greater certainty of the accuracy of the data, then a larger sample is needed.

      • 0 avatar
        supersleuth

        And as I keep trying to explain to you, that margin of error is actually quite small compared to the very large difference in the sample means, so your conclusion is not supported in the absence of actual reasons to question the representativeness of the samples. You do realize that sample size is taken into account in statistical tests, no?

      • 0 avatar
        Pch101

        “You do realize that sample size is taken into account ins statistical tests, no?”

        That comment is astoundingly obtuse, given what I’ve said above.

        You really don’t get it.

      • 0 avatar
        supersleuth

        It’s impossible to discuss this with someone so innocent of any understanding of statistics that he can’t grasp that citing a margin of error by itself, without any consideration of the size of the the difference whose significance you’re trying to determine, is completely meaningless. So I’ll give up now, and simply repeat that on its face Karesh’s result is clearly statistically significant, and in the absence of actual reasons to question the representativeness of the samples, is much more useful than no information at all.

      • 0 avatar
        Pch101

        “in the absence of actual reasons to question the representativeness of the samples, is much more useful than no information at all.”

        Maybe you need to look up with a “margin of error” is.

        In the meantime, you could reflect on my first comment, which you apparently failed to comprehend: “I’ll take whatever data that I can get, but the error rate is inherently high, and it should be used with caution.”

    • 0 avatar

      PCH101 is correct that the margin of error is large. But, as supersleuth points out, because the results are at opposite extremes they are meaningful even with this large margin of error. If the two stats had both been somewhere about the average then I wouldn’t have reached any conclusions based on them.

      I’ve put together a system that can provide initial reliability results on new models FAR ahead of any other publicly available source. It would scale just fine with more participants, so size of the samples does not reflect on the system itself.

      toxicroach:

      I designed this research to provide reliability information on new models far ahead of any other source. Thanks to this system, we now have some early indicators for two models that many people have been wondering about. I fail to see how working very hard to provide people with the information they want, and that won’t be available from anyone else for months, is a cheap shot simply for the sake of hits.

      • 0 avatar
        Pch101

        “because the results are at opposite extremes they are meaningful even with this large margin of error.”

        Your mistake is that you are arguing this on the basis of the “repair trip” count reported by each respondent.

        The issue here is that the experience of your pool may or may not accurately reflect the population. The specific number of repair trip incidents that each of the individual respondents reports isn’t the point; the issue is whether those twenty individual respondents each serve as a good representation of the total population of cars in the market.

        And you know that the odds of them not being representative are high, by their very nature. You could inadvertently cherry pick to one extreme or the other, and never know it. There’s nothing that you can do to get around that, aside from increasing the number of respondents.

      • 0 avatar
        supersleuth

        For someone who claims to have taken stats classes (I happen to be a scientist and you can bet your butt I have), you clearly didn’t understand the most basic concepts in them. The margin of error is ALREADY telling you the chance that *due to random sampling error* you picked samples that are so unrepresentative as the make the apparent difference in mean # of repair trips an illusion that is merely due to random chance. So you’ve already accounted for sample size via the margin of error; you don’t get to drag it into your argument a second time. A difference which is well outside the margin of error is statistically significant, period. Now, it could still be misleading; either a very rare fluke (REALLY rare in this case since, just from eyeballing the numbers, I would think the test would yield a pretty tiny p value)or there is *systematic* bias in the samples- which, unlike random sampling error, is not directly related to sample size and which you keep disclaiming any intent to invoke.

      • 0 avatar
        Pch101

        Wow. Now you’re making me fear for the state of modern science.

        One last time — What I said was this: small sample size = high margin of error. You can’t possibly dispute that.

      • 0 avatar
        supersleuth

        Yes, that’s the one and only thing you’ve said that’s true. The problem is that your conclusions don’t follow from it, at all, because the difference in #s of trips in Michael’s comparison is FAR larger than the margin of error (and that is basically the *definition* of statistical significance). The fact that you’re STILL not grasping this tells me that you simply have no idea what you’re talking about. You may have taken cookbook stats classes that didn’t convey any actual understanding of concepts (sadly there are plenty of those around), but you have clearly demonstrated that you don’t understand the first thing about statistical inference. If you flipped a coin 20 times and got 19 heads, would you claim there’s no reason to suspect that it is an unfair coin just because of the small number of tosses?

        Please Google “first rule of holes”.

      • 0 avatar
        Pch101

        The sample size is determined by the number of respondents, not by the number of repair trips reported by each respondent. This should be very obvious.

        Again, I am starting to fear for the state of science.

      • 0 avatar
        supersleuth

        Also true and also irrelevant. The sample size, thus computed, is exactly what went into your margin of error. In that way it has already been taken into account. If you don’t understand that a difference way outside the margin of error is statistically significant (and that that indeed is what statistical significance MEANS), then you understand nothing at all and you’re just slinging words around with no comprehension of what they mean.

        First rule of holes.

  • avatar
    SherbornSean

    True Delta’s statistics are all wrong, because they fail to confirm the racial stereotypes I believe about the people in the sites of manufacture and nationalities of the carmakers.

    I therefore question his methodology. How can MK possibly say the 2012 Fiat 500 is reliable without 10 years of data from at least 50,000 owners, independently verified by every mechanic in America, living or dead?

    • 0 avatar
      Robert.Walter

      Because he’s a sharp-eyed statistical magic mo-fo!

    • 0 avatar
      toxicroach

      Oh come on.

      20 car sample, and the Fiat is coming in at 10% of the normal # of repair trips, and you think people are questioning the numbers because of stereotypes?

      • 0 avatar

        Seems you’ve misread something. The FIAT isn’t coming in at 10% of normal. A little under half, but with the margin of error it could end up near the average with a larger sample. All I’m saying here is that it’s not worse than average, which is what many people expected or at least thought might happen.

      • 0 avatar
        Pch101

        Judging from some of the exchanges on this thread, it would seem that some of the readers haven’t taken a statistics class.

        Unfortunately, I’ve taken a few of them. For something like this, a sample of 20 will include an inherently high margin of error. There isn’t anything that Mr. Karesh or anyone else can do about that. The origin of the car doesn’t change that, either. A sample of about 300 would be preferable.

      • 0 avatar

        Of course a sample size of 300 would be preferable. So would world peace.

        Even well-funded outfits with decades of experience don’t have a minimum sample of 300+. There are good reasons for this.

        You’re working from a general heuristic, with no regard for:

        –the variance

        –the precision with which the averages must be measured

        –the tolerable level of uncertainty

        This isn’t a drug test where being off by a few percent could have fatal consequences. If the results here were actually off by 22% (or even 22 points), they’d still be useful.

      • 0 avatar
        Pch101

        “Of course a sample size of 300 would be preferable. So would world peace.”

        I never claimed that a sample size of 300 would be easy for you to get. I just said that it would produce a margin of error that is acceptably low.

        In the past, you’ve attacked Consumer Reports for its failure to report on certain cars. But CR excludes cars for which a sample size of 100 is not available. Even a sample based upon 100 respondents would produce a pretty high margin of error compared to something like a Gallup poll, but it’s not as high as it would be for a sample of 20.

        “You’re working from a general heuristic, with no regard for:

        –the variance

        –the precision with which the averages must be measured

        –the tolerable level of uncertainty”

        I did? I just pointed out that the margin of error is high when sample sizes are small. As a result, the conclusions of the findings should be used with caution. They could be right, but the odds of them being completely wrong are quite high.

        In other words, I would look at more than just one survey. If possible, I would use a few of the better ones, and compare them to each other. And if there was a conflicting result from a survey with a very small sample, versus two other credible surveys that each have large samples, then I would defer to the findings of the surveys with the larger samples.

  • avatar
    28-cars-later

    I can’t wait to see one of these 500s on 22s… eventually it will happen if they take off.

  • avatar
    highdesertcat

    I am actually surprised that the 500 hasn’t caught on yet in the US, especially in America’s large metropolitan areas. During my last visit to Europe in December 2010 I saw them by the hundreds in every country I drove through during my visit.

    As far as reliability in relation to where they are made, that may very well be a function of the training that the assemblers and suppliers are given by the respective auto makers. Remember the rusting frames and faulty CTS gas pedals on Toyotas. Someone cut corners there. American suppliers!

    There is documented evidence from all the domestic manufacturers who have plants in Mexico that their vehicles made in Mexico are much better than those made in the US. Fusion, MKZ, Avalanche, RAM, etc.

    But Hyundai, Kia and Subaru, because of their extensive training programs for their employees, have also managed to crank out vehicles in the US that are on par with those made in foreign plants.

    It should not be overlooked however that all foreign manufacturers had some serious quality programs when they started making them in the US. Toyota, Honda and Nissan all went through the phase of sub-par assembly when they first started making them in the US.

    Once VW gets their training up to speed we may well see VW products in the US on par with those assembled in Germany and other European countries. Mercedes and BMW were able to pull it off and are now exporting US-made vehicles all over the world. It can be done.

    • 0 avatar
      redav

      Every person IRL I’ve talked to about the 500 thinks it’s as ugly as homemade sin. And most Americans don’t like subcompacts.

      I’m not surprised in the least that the 500 sales are struggling.

    • 0 avatar
      vwgolf420

      I’m in Birmingham, Alabama and these cars are all over the place here–in the city and out in the suburbs. I saw tons of them in Europe too.

  • avatar
    oldyak

    I remember when the dipstick handle melted on my 96 Jetta.
    No overheating problems…It just melted!
    And they had the part in stock..what a surprise.ha..ha
    Ive owned two Fiats in my lifetime and the problems with both(other than rusting away) would NOT equal the problems with my 1996 base model Jetta!
    German engineering my ass!
    More like German pick pockets!
    Glad to see that the Fiat 500..a taste of modern Italian engineering ..just a taste..Blows VW out of the water!!
    VW cars and dealers have just been bad..bad..bad..but people keep buying them.
    Go figure?

    • 0 avatar
      highdesertcat

      I owned a VW Quantum and that soured me on VW for eternity.

      OTOH, I also owned several Beetles of different vintages, and they ran good. I even converted several of my old Beetles into dune buggies and they are still being flogged by the mudders I sold them to.

      Maybe the new VW stuff just isn’t as good.

      • 0 avatar
        Robert.Walter

        Is it fair to judge an OEM, for good or bad, for eternity based on experiences from more than 30 years ago? Gawd lard, everybody who designed those vehicles is probably retired or dead by now!

      • 0 avatar
        Lokki

        Robert-

        Is it fair? Well companies have a culture, and new engineers train at the hand of their bosses who are the old engineers who are responsible for the fiascos of yore. Methods of thinking are passed on.

        An easy example is the GM “Hail Mary Pass” approach to introducing new technologies. They have been leapfroging their competition with new ideas since at least the 50′s. However thir last successful use of that approach may have been the mass introduction of the small block chevy engine.

  • avatar
    APaGttH

    No, no, no, you have it all wrong.

    FIAT stands for Fix it Again Tony, my dad told me and he had a friend in college who owned a Fiat 128, and that was a steaming pile of shit. It is for that reason, the 500 also must be a steaming pile of shit.

    On the other hand, I love my Volkswagen and any reasonable person will tell you replacing ignition coils every 30,000 miles is regular maintenance and everyone has to add a quart of oil every two-thousand miles to their cars. I don’t know what people whine about with VW having quality problems. Lies I tell you, LIES!

  • avatar
    Morea

    1) Small-sample statistics is a highly developed field, especially in the sciences where experiments can be very expensive, e.g. clinical trials. Karesh needn’t go it alone here (and likely doesn’t) because there is a vast literature on the subject. Sure more is better (up to the limit of getting info on every car sold!) but small doesn’t mean wholly non-representative.

    2) Regarding older cars and reliability being largely based on the owner’s maintenance habits and on non-OEM replacement parts: so what? Stats are stats: if a brand generally has more conscientious owners, and the replacement parts are better, then such brands will have older cars that will be in better shape. As a prospective buyer interested in only purchasing one car this is still valuable information to me.

  • avatar
    Philosophil

    The usefulness of statistics is generally relative to the uses to which they are put. It seems to me that statistics such as these can be useful, for example, as a measure of early trends for these specific vehicles, but they shouldn’t be overextended (at least not in isolation) to try and justify any broader, more sweeping claims about the long-term reliability of those vehicles or about the brand or company in general.


Back to TopLeave a Reply

You must be logged in to post a comment.

Subscribe without commenting

Recent Comments

New Car Research

Get a Free Dealer Quote

Staff

  • Contributing Writers

  • Jack Baruth, United States
  • Brendan McAleer, Canada
  • Marcelo De Vasconcellos, Brazil
  • Vojta Dobes, Czech Republic
  • Matthias Gasnier, Australia
  • W. Christian 'Mental' Ward, Abu Dhabi
  • Mark Stevenson, Canada
  • Cameron Aubernon, United States
  • J Emerson, United States