Most of my work and research is in statistics so this is coming from an engineering/mathematics point of view.
Statistics are just numbers. The numbers by themselves do not tell the whole story. For example, Ford is rated above BMW for reliability on the 90 day survey. So then how come the Fords depreciate really fast?
Interpreting the numbers, you'd be correct to say that people who buy expensive cars are very picky. This factor alone will skew the results. Ford F150's are probably one of the most commonly leased vehicles for construction. I don't think construction workers driving a company truck will really take care of the truck as much as the owner of a BMW. What we're talking about is several different types of buyers and populations, then drawing conclusions between populations.
Also, there is no indication on the severity of the problems. Sure you can have 100 small problems where 99 are cosmetic...vs one problem where the transmission dies on you. Which car then is more "reliable"?
There's also the method of study itself. They said they sent questionaires to 81,500 people. I'm pretty sure this is the total sample since I doubt there would be 81,500 2008 Porsche owners (and the fact that JD Power has to get ALL of them to participate). From this figure, I would suspect that for each car brand, they had an insignificant sample size. Assuming that they have even sample sizes, that'll mean 514 cars per sample. I would suspect that more people own Toyotas and Hondas over Porsches so that probably would be a sample size of 100 for the Porsches or less. BMW would be slightly more  maybe 200 or 300. This means that if one person ends up with a lemon, it would artificially skew the results as well. The larger the population size, the more accurate the number. A proper study would say something like "this study is accurate 95% of the time with an error of +/ 3.5%". JD Power doesn't specify the error (because if they did it would be very very high).
Using the same logic that most people use when it comes to numbers, you could take statistics on the number of sales and number of people in a shopping mall starting December 1 and do your survey until December 25. You'll notice that there will be more people in the malls and sales are higher. You'll notice on average that people will be spending more. The numbers will peak at December 24 and suddenly Christmas happens on December 25 and all numbers drop off. Therefore, from the statistics, it is clearly obvious that SHOPPING CAUSES CHRISTMAS.
Really, if you look at the survey they did, it's full of problems and the conclusions they're drawing are very very thin. JD Power makes their money by selling their information to companies to find out what their customers think and also by selling the licensing fee to use its logo and company name on marketing products. So it's NOT a scientific study but more of a marketing study.
If you put numbers on something and say that it's credible, people will believe you. I know this because I'm right 95.57% of the time.
