The May results have now been posted. As with the earlier two sets of results, some results are official, while others are asterisked because the number of responses was low.

The official May results include 33 models, up from 26 three months ago. This understates the increase, as this time I required 25 responses for the results to be official. If the cutoff had been 20 like last time, the number of models would have been 46. (Note: I’m using “model” as shorthand for “model / model year.” For example, the 2005 Accord and the 2006 Accord count as two separate models.)

I have also raised the cutoff for the asterisked results only members see from 10 to 15. Even so, the number of models has increased, for a total of 65. With the old cutoff, the number would have been 87.

The response rate, 43% again, was excellent compared to other such surveys. But if everyone responded to the surveys when asked to, the official results would have included more than 85 models.

The confidence interval charts have been moved to a separate page, because many people found them confusing. In their place, I have added a ribbon graph for each model showing how that model compares to the best and worst for its model year. Currently the best and worst are estimated. In the future, when more models are included in the results they will be actual.