The Bauer Curve
I got curious. I wondered about the distribution of Bauer stars. The engineer lurking deep inside of me came out. I ran the numbers.
As of yesterday morning, there were 1432 restaurants in the San Francisco Chronicle's database that had received the full star rating treatment. The distribution of Bauer stars is…
4 stars – 7 restaurants
3 ½ stars – 18 restaurants
3 stars – 115 restaurants
2 ½ stars – 388 restaurants
2 stars – 695 restaurants
1 ½ stars – 181 restaurants
1 star – 27 restaurants
½ star – no restaurants
no stars (aka, the dreaded box) – 1 restaurant
So, what does this all mean?
According to the Chron, four stars means extraordinary; three, excellent; two, very good; one, good; and the no-star box, fair.
At first glance, it’s obvious that the Chron is hedging. The lowest rating is fair? Are you going to tell me that, out of 1432 restaurants reviewed, none were poor??? Gimme a break! Smells like the Chron’s lawyers wrote the definitions.
In fact, as much as restaurateurs like to kvetch about Bauer and Company’s reviews, a quick look at the distribution of stars shows that they’re biased toward the good. With 1400+ samples, you’d expect them to fit a bell curve. As the graph below shows, there are way more 3 and 2 ½ stars than 1 ½ and 1 stars. For a bell curve, you'd expect the 3's and 1's to about balance. Ditto for the 2.5's and 1.5's. (Granted, there’s probably some sample bias as the Chron isn’t likely to visit greasy spoons but I’d be surprised if that accounts for all of the abnormality. And some greasy spoons are good!) Bay Area restaurateurs should thank their lucky stars (pun intended) that Bauer isn’t as tough a grader as your average college professor.
So here’s the Grumpy interpretation of the Bauer galaxy…
4 stars – top ½ of 1 percent, superstars in the truest sense of the word.
3 ½ stars – top 2%, great, incredibly great.
3 stars – top 10%, excellent.
2 ½ stars – top 40% to top 10%, are they just above average or borderline excellent? The range is so wide you can't tell.
2 stars – bottom 15% to top 40%, way below average to average – is the Bay Area’s restaurant scene really so good that our below average joints are “very good”??? I think not.
1 ½ stars – bottom 2% to bottom 15%, way below average, hope they’re cheap.
1 star – bottom 2%, lousy but the Chron says they’re good – looking at one-star reviews such as Celia's, Daily Grill and Little Thai, I don’t understand what’s good about ‘em.
½ star – not used, when some sucks so bad, half-star distinctions don’t matter.
no stars (aka, the dreaded box) – appallingly sucktastic but our lawyers won’t let us say so.
If I could magically change one thing about the Bauer curve, it would be to reduce the number of restaurants with 2 ½ stars. I’d find it interesting and useful to know who’s in the top 10% to 20% of reviewed restaurants. The range of the top 10% to top 40% includes a lot of average restaurants in what could and should be a group of outstanding performers who don’t quite crack the upper ranks.
Oh, and who’s the poor sucker with the only no-star box? That would be Spenger's Fresh Fish Grotto in Berkeley. I’ve eaten at Spenger’s several times and, while it’s not great, it's not THAT bad. Maybe Bauer isn’t the only one who grades on an easy curve.
Why, pray tell, should the distribution of scores follow the bell curve? Me thinks someone needs to read The Black Swan.
Reply to this
Methinks that the quality of restaurants, and critics' representation of said quality, belongs to Mediocristan, not Extremistan. Sorry for the delay in responding but I had to buy and read the book first. Thanks for inspiring me to do so.
If you haven't read the book, you won't understand my response. I recommend that you check out The Black Swan by Nassim Nicholas Taleb
Reply to this
Oh, the people, they are a-talking. I heard not one, but two people mention your engineering feat.
Excellent.
Reply to this
The funny thing is, it wasn’t THAT much work, less than an hour. Far less difficult than a post with a lot of links. The Chron’s search and sort functions made it easy. The biggest headache was figuring out how to use the charting function in the new version of Excel, but I’d have had to do that sooner or later anyway.
I just posted a follow up to my original post on the subject.
Reply to this