|
It's actually only four. I'd love to have more data, and we will in August. Past experience suggests that, despite the small sample size and extrapolation, this result is more likely to stay about the same than to change much.
One reason for this is that I have intentionally selected a metric with relatively low variance, to make the most of small sample sizes. Introduce other variables, such as repair shop competence, and the variance would increase and a larger sample size would be needed to maintain the same level of accuracy. Hence the focus on successful repair trips. I would love to have a separate stat for the success rate of repairs, as this can be a major source of dissatisfaction, but this must wait for a future time when more people are involved. Correct on the last point--the analysis collects data going forward to prevent whatever past experience might have motivated someone to participate from potentially distorting the results. |
for Michael Karesh-Reliability
Out of 26 vehicles, only four have gone in for repairs and this gets a 56 out of 100 and a yellow (caution) light?
The more I look at this, the less I understand it. While I have concerns re: Consumer Reports, they require a minimum of 100 vehicles before they reach any conclusions. Their stats are also much easier to understand. The person who participates because he has has unresolved problems is the one that really needs to be in the sample. Also, the information is much more useful and valid if all information re: the vehicle since day one is included. What about the people that had four trips for a problem before they found your site? They have to ignore what has happened already. Reliability is a function of history - all of it, not part of it. Appears to me that you have way too little information to establish any kind of rating. |
Four in a little over three months. Over an entire year this calculates to about 14.5.
You might think you understand more about CR because they hide almost everything. It's like saying you better understand what's going on with your car from the idiot lights than the instruments. Instruments are harder to read than a light that is either on or off. If a car has an average rating in CR, about how many problems have been reported per 100 cars? Pretty basic information, but can you get it from their ratings? You're concerned about the time period covered by my results. What time period is covered by their results? CR has a minimum sample size of 100 cars, but they also end up splitting much finer hairs. To do what they do they should have a much larger minimum sample size, at least 500. Partial histories work just fine. The results stay pretty stable for most models quarter after quarter even though the time period keeps changing. You've got to remember that the data are being pulled from multiple cars, not just one car. You might miss something with one car because it fell outside the window, but you'll get the same with another. It evens out. |
What about people who bring their new car in for service because they think something is wrong but in actuallity it's fine.
|
Doesn't count. Only successfully completed repairs count for this analysis. This way we know that not only did the owner perceive a problem, but the repair shop confirmed it and was able to do something to make the problem go away.
|
Bogus?
Originally Posted by mkaresh
(Post 4094854)
Four in a little over three months. Over an entire year this calculates to about 14.5.
You might think you understand more about CR because they hide almost everything. It's like saying you better understand what's going on with your car from the idiot lights than the instruments. Instruments are harder to read than a light that is either on or off. If a car has an average rating in CR, about how many problems have been reported per 100 cars? Pretty basic information, but can you get it from their ratings? You're concerned about the time period covered by my results. What time period is covered by their results? CR has a minimum sample size of 100 cars, but they also end up splitting much finer hairs. To do what they do they should have a much larger minimum sample size, at least 500. Partial histories work just fine. The results stay pretty stable for most models quarter after quarter even though the time period keeps changing. You've got to remember that the data are being pulled from multiple cars, not just one car. You might miss something with one car because it fell outside the window, but you'll get the same with another. It evens out. Time period for CR is from date of delivery till survey completed-ALL information is considered. For example, they did not report on 2009 E Class because not enough data available. You report on as few as 18 cars and then say it is a small sample and infer it may not be valid. They do not split hairs anymore than you do. I know because I just completed my fourth survey in April. Did yours on a previous car a couple of years ago. You are just rationalizing away the deficiencies in your system. Too few cars, too little information - simple as that. |
I would certainly like to have larger sample sizes. The best way I know to achieve this is to get more people involved. Saying "I'm not participating because the sample sizes are too small" contributes to the problem you're complaining about. Not to mention your earlier attempts to discourage owners from participating. You're the last person who should be complaining about the sample size.
I label anything under 25 cars as a "partial result" that is asterisked on the site and only visible to signed-in members. I report partial results because many people feel it's better to know how a fairly small number of cars have been doing than to know nothing at all. Incorrect on CR's time period. Their survey includes the past year. Anything before that period should not be reported. I've also learned that memories fade fast, within weeks. So many participants no doubt forget to report repairs that occurred earlier in this time period. This makes CR's information far from complete. Also incorrect on that CR doesn't split hairs any finer than I do. Look more closely at what they're reporting, though it isn't easy since they don't divulge much. They report on 16 or so subsystems. This effectively splits the sample 16 ways. What's 100 divided by 16? What's the minimum difference between the different dots in their results? Do you know? The answer: one problem per 100 cars. At their minimum sample size this means that a single response can easily mean the difference between one dot and another. This is what I mean by splitting hairs far finer than I do, and far finer than their minimum sample size permits. And where is CR's result for the 2010 E-Class? We'll updated our stats again in August, and then again after that in November. They'll have their first results, covering through roughly April, around November. |
???????????
Originally Posted by mkaresh
(Post 4095693)
I would certainly like to have larger sample sizes. The best way I know to achieve this is to get more people involved. Saying "I'm not participating because the sample sizes are too small" contributes to the problem you're complaining about. Not to mention your earlier attempts to discourage owners from participating. You're the last person who should be complaining about the sample size.
I label anything under 25 cars as a "partial result" that is asterisked on the site and only visible to signed-in members. I report partial results because many people feel it's better to know how a fairly small number of cars have been doing than to know nothing at all. Incorrect on CR's time period. Their survey includes the past year. Anything before that period should not be reported. I've also learned that memories fade fast, within weeks. So many participants no doubt forget to report repairs that occurred earlier in this time period. This makes CR's information far from complete. Also incorrect on that CR doesn't split hairs any finer than I do. Look more closely at what they're reporting, though it isn't easy since they don't divulge much. They report on 16 or so subsystems. This effectively splits the sample 16 ways. What's 100 divided by 16? What's the minimum difference between the different dots in their results? Do you know? The answer: one problem per 100 cars. At their minimum sample size this means that a single response can easily mean the difference between one dot and another. This is what I mean by splitting hairs far finer than I do, and far finer than their minimum sample size permits. And where is CR's result for the 2010 E-Class? We'll updated our stats again in August, and then again after that in November. They'll have their first results, covering through roughly April, around November. Furthermore, CR has never "extrapolated" results based on an inadequate sample, nor have they said "above average [repair trips], but typical for a new car." as you have in the past. BTW, I take what CR reports with a grain of salt, especially their car reviews. |
Originally Posted by El Cid
(Post 4096589)
I have completed at least four of their surveys and read the explanations they have for their ratings, how many problems per car, etc., etc., etc. None of what you said is accurate re: CR. Maybe you need to subscribe to CR or if you do - read the stuff they send you.
Furthermore, CR has never "extrapolated" results based on an inadequate sample, nor have they said "above average [repair trips], but typical for a new car." as you have in the past. BTW, I take what CR reports with a grain of salt, especially their car reviews. You've misquoted me. Check what I actually wrote. But, yes, I say what I say. I don't simply repeat what someone else says. I don't base what I do on what CR does. Why should I? Here are CR's actual instructions, from a copy and paste, not based on memory: "If you had any problems with your car in the past 12 months that you considered SERIOUS because of cost, failure, safety, or downtime, select the appropriate box(es) for each car." Some relevant critiques I've written on CR's methods based on a very careful reading of their actual words and actual results: Consumer Reports' new rating system Consumer Reports survey critique Thousands of people have read these, including people at CR. No one has every factually refuted anything in them. Simply saying "not correct" proves nothing. |
MKaresh-Let's make a deal?
You have your opinion and I have mine. You know what you know and I know what I know.
So, let's let it go and stop this silly arguing. |
Works for me. It has become quite silly when the meaning of "in the past 12 months" is a matter of opinion.
|
We'll have updated results in August, with a preview for participants next month. As has been noted, a larger sample size would make the results more precise. So I hope more owners can sign up and participate.
Car Reliability Survey |
Why? JD powers has already given an initial report and ranked MB #2. You are wasting your time.
|
Originally Posted by petee1997
(Post 4127811)
Why? JD powers has already given an initial report and ranked MB #2. You are wasting your time.
|
Originally Posted by mkaresh
(Post 4127982)
Some people want to know what happens (or doesn't happen) after the first 90 days.
I find the qualitative ratings provided by Truedelta to be much more informative than CR, whose information base is IMO much more subjective and therefore prone to greater inaccuracy. But, let's get back to the OP topic. It is very encouraging to see that the W212 has done so well in this IQS and that Mercedes as a brand is back to being a high quality manufacturer. In contrast, if you look at the brand rankings, BMW has fallen dramatically. My guess is the now 4 year long problem of HPFP failures in the 3.0 L gas twin turbo engine has severely dented BMW's reliability ratings. |
I don't see why so many people are so pissy about these Reliability Ratings.
Who cares if JD/CR/BS, etc. does it, the more information the better. No reason to deter someone from gathering info. About BMW's TT motors, this problem could even get worse once they're forced to use TT's more and more. As well, M-B is about to start using TT's more, so I wonder if they'll get hit with some problems. If seems Turbo motors are usually prone to more problems, more expensive repairs, and less confidence in longetivity (unless you're Porsche, lol). |
To me I usually dismiss much of these stats as most are taken very early in the ownership even before someone has much experience with what ever vehicle is in question. Powers surveys usually come to me within a couple of weeks after taking delivery. Most owners at this early time are still enamoured with their purchase and really do not even know if something is not quite correct. A survey after say 2 years of ownership and maybe 25K miles would tell lots more.
|
Originally Posted by K-A
(Post 4129195)
.
About BMW's TT motors, this problem could even get worse once they're forced to use TT's more and more. As well, M-B is about to start using TT's more, so I wonder if they'll get hit with some problems. If seems Turbo motors are usually prone to more problems, more expensive repairs, and less confidence in longetivity (unless you're Porsche, lol). DI requires high pressure fuel pumps. BMW's have witnessed HPFP failures. Unfortunately so have the Porsche 997.2 models with DI. Also Audi have had increased carbon issues with DI now as well. A lot of design will be tested in the real world and so we'll hear about all sorts of issues for a while. Mercedes is going DI now, too. Diesels motors are DI and their pumps are much higher pressure than gasoline DI pumps. But the difference is that diesel fuel acts as a lubricant. They have a solid history. Manufacturers are rushing to develop more than adequate power and torque under stringent fuel consumption and emission regulations. A lot of it is relatively new engineering (Remember when fuel injection came out? It had years of headaches at first.) So, yeah, expect all kinds of new issues. Although eventually things will get fixed as the issues start showing up. Consumers will be the real world guinea pigs as usual. Anyway, it's nice to know that the factory decided to build the car correctly so that it works for the first 90 days. But internal design failures rarely happen in 3 months. The following 5-10 years is the real question. |
logged mine
|
Thanks, Lance.
Interesting info on DI, 220S. And definitely true that most mechanical problems appear after the first 100k miles. The BMW HPFP problem is an exception. Most DI systems aren't having this problem, at least not yet. This coming week we'll start previewing the next set of results to participants. The preview results will be updated as responses come in. To help provide the best possible information on the W212: Car reliability research |
Originally Posted by 220S
(Post 4129476)
The problem is with the direct injection.
DI requires high pressure fuel pumps. BMW's have witnessed HPFP failures. Unfortunately so have the Porsche 997.2 models with DI. Diesels motors are DI and their pumps are much higher pressure than gasoline DI pumps. But the difference is that diesel fuel acts as a lubricant. They have a solid history. Manufacturers are rushing to develop more than adequate power and torque under stringent fuel consumption and emission regulations. A lot of it is relatively new engineering (Remember when fuel injection came out? It had years of headaches at first.) So, yeah, expect all kinds of new issues. Although eventually things will get fixed as the issues start showing up. Consumers will be the real world guinea pigs as usual. Anyway, it's nice to know that the factory decided to build the car correctly so that it works for the first 90 days. But internal design failures rarely happen in 3 months. The following 5-10 years is the real question. |
Originally Posted by DerekACS
(Post 4155616)
The high rate of HPFP failures with the twin turbo 3.0L BMW engine seems to have occurred primarily in the USA, where ethanol/gas blends are common. In Canada, this fuel mix is not widespread and the HPFP failures I'm told have been of a far lower percentage of engines than in the US. Ethanol/gas blends have also played havoc with marine outboard engines.
Under Bush, gas companies were mandated to sell an ever-increasing amount of ethanol. E85 hasn't exactly taken off, so where's all of that ethanol going to end up? Thank you, ethanol lobbyists. |
Originally Posted by DerekACS
(Post 4155616)
The high rate of HPFP failures with the twin turbo 3.0L BMW engine seems to have occurred primarily in the USA, where ethanol/gas blends are common. In Canada, this fuel mix is not widespread and the HPFP failures I'm told have been of a far lower percentage of engines than in the US. Ethanol/gas blends have also played havoc with marine outboard engines.
A fix will be found esp in respect to any fuel lubrication issues, but there may be compromises, we'll have to wait and see. I'm sure BMW has kept detailed records on failures and where they occur. |
Solving the HPFP issue certainly should be a very high priority within BMW. It's costing them a lot of money, and probably quite a few customers as well. Certainly not good for the brand's image.
Updated stats next month. For the W212 E-Class, we especially need more participants for the 2011. Though more 2010s would also be helpful. To help provide better information on these cars: Car reliability research |
We have an initial reliability stat for the new W212 E-Class based on owner experiences through June 30, 2010.
Other sources of car reliability information won't cover the more recent months until the summer or even fall of next year. With a reported repair frequency of 59 repair trips per 100 cars per year, the new E-Class remains about average, which is quite good for an all-new German luxury car. Thank you, once again, to everyone who has been helping. We'll have further updates in November and February. To see how competitors compare, and to sign up to help: Mercedes-Benz E-Class reliability comparisons |
| All times are GMT -4. The time now is 08:48 AM. |
© 2024 MH Sub I, LLC dba Internet Brands