8 December 2006
Reconnecting the dots
Consumer Reports' auto-reliability ratings are known nationwide, and while some swear by them, others swear at them. (An example of the latter is here.) While their data for cars I have owned have tracked fairly well with my own experience and yes, I do fill out the questionnaire every year obviously anything I could report is too small a sample for any kind of meaningful statement.
I have noticed, though, that they've changed some of the methodology. Used to be, there was a definite range for each colored dot: the "white" ("average") dot meant a failure 5.0 to 9.3 percent of the time, and that was that; half-red and full-red dots were better, half-black and full-black dots were worse. To make this work, you had to compare it to their statistical Average Model, which had dots of various colors in each of the problem areas surveyed.
The new system, detailed in the 2007 Buying Guide, is on a relative scale, and all vehicles of a given model year are considered as a group before the dot is assigned. They're not giving out the actual percentage ranges anymore, and maybe that's just as well, since I never found them especially useful. They did state, however, that black dots, full or half, will not be issued unless the actual problem rate is 3 percent or higher, which seems reasonable: if the average for such-and-such a subsystem overall is 1 percent and the same subsystem on Brand X fails 2 percent of the time, you're still looking at a fairly-negligible risk, even though it's by definition worse than average.
Under the new system, Gwendolyn and her sisters draw 11 red or half-red dots, three white ones, and one of the dreaded black ones, under the heading "Ignition". (This is consistent with at least one other survey I've seen.) Still, no survey can tell you for sure the one thing you really want to know, which is "Is this going to happen to my car?"