The fine print and grading on the curve

I was reading the July 2009 Consumer Reports (as usual, I'm about a month behind on magazines) and reached a set of ratings for chain restaurants. Read the commentary and the neat little sidebar where trained tasters compared oversized New York strip steaks at Morton's with slightly less oversized New York strip steaks at Outback, Applebee's and Friday's. (Conclusion: The steak's best at Morton's--duh--but Outback's probably the best value.)

But then got to the actual ratings--and noted a lot of chains with black dots (the worst rating) for taste, some also for service, mood and choice.

And thought, "wait a minute: Why do people go to restaurants if they think the food stinks?" These aren't fast-food chains; these are all sit-down, table-service restaurants. Starting the guide notes, I see that the chain with the most visits reported is one with the worst rating for taste. What? Are people all masochists? As it happens, it's a chain I've eaten at sometimes, at conferences or other situations where I'm not sure of the local choices--and while I'd never call the food first-rate, it's solidly in the "not great, not terrible" category. (And, to be sure, I'm likely to eat at airport outposts of two of the other chains, one with worst taste rating, one with next-worst, because they're there. And the food's usually adequate.)

The key is midway through the second paragraph of the guide, which isn't tiny print but is still more than you might read:

Scores for taste, value, service, mood and choice are based on the percentage of readers who judged the chain excellent for each. Those scores are relative, reflecting how each chain differed from the overall average.

The second sentence certainly clarifies the presence of so many black dots, particularly in the Family category (as compared to red dots, the highest rating, in Traditional American, Seafood, and Italian categories): Grading on the curve.

The first sentence makes me wonder, though. If I responded to the survey (which I may have), assuming there are the usual five Likert-scale choices, it's unlikely any "Family" or "Pub style"--or, indeed, any--chain restaurant would get an "Excellent" for taste. (Haven't been to Morton's; it's out of my price range.) I find it odd that, apparently, it makes no difference whether you think meals are very good or poor--if they're not excellent, it doesn't count. I guess I would have assumed the use of weighted ratings--e.g., 5 for Excellent, 4 for Very Good, and so on, with the total divided by the number of responses.

No great meaning here. Although, frankly, as a sometimes respondent to CR surveys, I'm a little peeved at the idea that there's no point in distinguishing between levels of quality other than excellence.

(Typo corrected in quote from CR. My typo, not their copy-editing, which is consistently excellent.)

More like this

Wow, that is some serious curve grading from the sound of it. When you look at it, the ratings do not seem that meaningful. And yet, so many people who don't read closely will likely miss it.

No Morton's here, but if there were, out of my price range as well. We do have some of the basic chains, including Outback, which in the sticks here, is considered a big deal (i.e., any major holiday like, say, Mother's Day, it's the default place to take someone out).

Best, and keep on blogging.