Consumer Reports is not a very good ratings website.
What's a better one for car reliability?
When is the last time anyone has ever used consumer reports? Are you also under 65?
I do. I subscribe to their print magazine and web site. I read some of it. I most definitely look at their car reliability ratings as a screener for cars I'm considering. I'm under 65.
I've found their reliability results to be reasonably accurate for cars that my parents and I have owned. I also have found them to often correlate w/known problem areas on other cars that I've owned vs. popular web sites for them (e.g. my350z.com, priuschat.com, maxima.org, mynissanleaf.com). Are their results perfect? No.
Sometimes, we can't really tell what's behind the rating. For example, when I last looked, I'd seen a trend of Priuses once hitting a certain age (usually when their 12 volt goes bad and requires replacement) getting dinged w/electrical system bad ratings. Should be it electrical? Unfortunately, the symptoms on a Prius of a dying/dead 12 volt are NOT the same as a regular ICEV. On a Prius, you will NEVER hear a starter cranking too slowly or not being able to turn the ICE because the ICE isn't started by the 12 volt and it has no dedicated starter motor.
Instead, weird stuff happens usually accompanied by intermittent no start conditions (unable to put car in to READY mode). So, if the Prius owner doesn't know this (they would if they're on Priuschat paying attention), they might think they have an electrical prob and take it to a repair shop/dealer.
IIRC, once the wave of dying 12 volts passes (usually ~4 to 6 year mark), electrical system reliability looks fine again.
I have no issues with CR's testing methodology since they have experts testing the vehicles on a broad spectrum of metrics that are beneficial to the consumer. But their annual reliability survey is NOT a statistically accurate sampling. When they say they surveyed 1400 Model S owners, those owners are CR subscribers who chose to participate in the survey. For it to be statistically valid, it must be a random sampling; therefore, the reliability survey is really most accurate to the type of individuals who would subscribe to CR and like to take the survey.
Ok, so what other reliability survey of cars is statistically valid in your mind and has much more than 1400 responses?
And that is my point, it is poor analysis. It is barely analysis at all. They do not break down reliability by model year, instead lumping them all together.
They do give reliability ratings for each model year.
These guys did another reliability test and found exactly what I have been saying: While Teslas are less reliable than typical new cars
Full stop! You've just stated it right there.
So instead of like CR is extrapolating, reliability is getting better as Tesla matures as a manufacturer, and as Teslas technology matures.
You're extrapolating as well. How do you know this? Reliability does NOT always improve for a given model or automaker. CR has shown this time and time again.
Common sense for most people, but something the people at CR can't seem to grasp.
No, you're trying to interject something else into the results that were given.
All 3 model years are early generation products. Even 2016 models are early generation, Tesla has JUST started making cars.
Yes, Model S was the first car they built from the ground up. But, they did have some help from the Mercedes parts catalog.
There was the Roadster which first shipped in early 08.
The question is has the reliability and build quality been improving?
I'm not sure I can answer that. Their survey is to readers of what happened in the past 12 months for THEIR car. The averages (already pointed to earlier) get worse as cars get older. They're comparing the problem rates to the averages for those years.
Unless this survey can break down its data by model years, it does nothing to answer that question.
They do.
CR's brain dead prediction of the same 40% below average forever.
They don't do that. They had a different prediction last year. Whatever that number was (don't have it handy), it worked out to average.
And, if there were reliability ratings the year before, they had another one. In a year from now, you'll see a different one.