J.D. Power released the results of its 27th annual Initial Quality Survey (IQS) for cars today at the Detroit-area Automotive Press Association (APA). The survey was significantly revised this year to include new systems such as lane departure warning among the 233 covered and to use an online rather than a paper form. But the output and implications remains much as they have been since the third-generation IQS was introduced in 2007. So the previous strengths and weaknesses continue to apply.

Strengths include a rigorous process that drills down to a fine level of detail–the form lists 233 specific problem areas for car owners to select from. Thanks to the additional flexibility of the new online format, the potential level of detail is much higher than before. The data is also quite fresh, with a lag of just 2.5 weeks since the survey period ended in May. In all, four waves of surveys were conducted, with the first in February. The target sample size for each model was 450, with a minimum of 100. Over 83,000 surveys were completed, a response rate of 17 percent. Even with this number of responses, the scores for nearly identical models often differ by ten or more points.

Largely thanks to excellent results for its large pickups and SUVs, GM was the top-ranked manufacturer for the first time. Porsche, with an intense focus on initial quality (and likely far more inspectors per 1,000 cars produced than the average plant), was the top-ranked nameplate. Ford continued to score lower than in the past, with much of the blame going to its MyFord Touch infotainment interface. Scion ranked dead last thanks to a few common problems with the FR-S sports car (lumpy idle, chirping fuel pump, tail light condensation). As in the past, though, most makes clustered tightly around the average of 113 problems per 100 cars in the first 90 days of ownership. Much will be made of who “won,” but the differences are often too small to be meaningful unless the entire enterprise is treated like a basketball game, where losing by a single point is losing nonetheless.

We’re often asked how J.D. Power’s stats compare to TrueDelta’s. The short answer: we ask different questions and measure different things, so the results can differ, especially if you don’t know what to look for. To begin with, J.D. Power asks about both “design problems” (a difficult to use navigation system or hard to reach mirror control switch) and “manufacturing defects.” Not only this, but the former are now about two-thirds of the total. TrueDelta only asks about manufacturing defects. We feel that if you’re measuring two different things, you should have two different stats. If people only look at a car’s subpar IQS score, they could conclude that it will require many repairs when the real issue is a voice recognition system that doesn’t work well.

Across all cars, the most common problems involved poorly functioning voice recognition, poorly functioning Bluetooth connectivity, wind noise, too easily soiled or scuffed interior parts, and difficult to use nav systems. Of these five, only the third is likely to count as a problem in TrueDelta’s survey, though the fourth would count if the part had to be replaced and not just cleaned.

On jdpower.com you can look up “circle dot” ratings for individual models, such as the Ford Focus. The lowest score, representing the bottom 30 percent, is two dots. Going up from there, a car in the 30-60th percentile range gets three dots, one in the 60-90th gets four, and one in the top ten percent gets five. As in the past, no car gets a single dot–which would be the most useful thing to know. Consequently, the lowest rating covers all the way from the worst cars to some just below the average. The Ford Focus is worse than the average, but how much worse? There’s no way to tell from the publicly released information. Maybe you should avoid a car with such a rating, but (given a bell-shaped curve) it’s likely near the average and still okay for all but the most repair-averse car buyers.

Helpfully, J.D. Power doesn’t just provide an overall rating. For each model they have both a “design quality” and a “mechanical quality” score, as well as subscores for each of these. The “mechanical quality” score is the closest match to TrueDelta’s car reliability stats. The Focus, like many recent Fords, gets two-dot scores almost across the board–but somehow a four dot score for “predicted reliability” down the road.

Yet it’s still not all that close. TrueDelta only counts problems verified and successfully repaired by a mechanic (which can be the car owner), to rule out things that are really design defects and to exclude the variable of repair shop competence. J.D. Power counts just about anything owners report, even if the car hasn’t been to the dealer yet. With design problems, the car had been taken to the dealer only 9 percent of the time, and within this low percentage entirely fixed only 13 percent of the time. Though design issues are now the majority, only about one percent of cars had a successfully repaired design issue.

Even with mechanical problems, only 28 percent of owners had taken the car to the dealership to get them fixed, and only 42 percent of these times was the dealership able to fix the problem. These numbers both strike me as shockingly low. I asked the J.D. Power executive why he thought mechanical problems were successfully repaired only 42 percent of the time. He had no further detail. Multiply both percentages to find that only about 12 percent of the reported mechanical problems had been entirely fixed at the time of the survey. Granted, the survey was conducted about 90 days after delivery, and many people were probably waiting until the first required service to report minor problems (and the great majority of problems are minor). But with the average score 113, and mechanical problems only a bit over one-third of the total, and it seems that less than one car in twenty had a successfully completed repair.

Want to know the specific issues behind these subscores? Unfortunately, you can’t without checking other sources such as car forums and TrueDelta. While J.D. Power has the ability to provide a high level of detail, they continue to only provide this detail to car manufacturers willing and able to pay substantial sums for it.

With the average so low, and most makes close to the average, should car buyers worry about differences in initial mechanical quality? Even those who are concerned should ignore the overall IQS rating, since it combines two different things, and instead check the separate design quality and mechanical quality subscores.