It's Zagat time again in old L.A. Town. A new edition (the second) of the Zagat Los Angeles Restaurant Survey is in the works. Questionnaires are being collected even as we speak; restaurants are being ranked numerically by hundreds upon hundreds of amateur (and occasionally professional) restaurant critics throughout the Southland; snide or admiring comments are being salt-and-peppered throughout the Zagat forms. The common diner, at last, is having his or her own say.
Meanwhile, though, I am in receipt of a letter about the first Zagat guide from David Holzgang, a Chatsworth-based computer scientist with a special interest in statistics (and, incidentally, an old eating and drinking buddy of mine). I have expressed my own qualms about the Zagat guide in this column already--about the fact, for instance, that the survey's methods permit such obvious nonsense as the rating of Lyon, a pleasant little 15-seat, counter-only place on the edge of downtown L.A., as one of the three best French restaurants in town (and one of the five best of any kind on the basis of food alone). I'd like to quote and/or paraphrase some of Holzgang's remarks here, too, since they question the very premise of the survey from a professional's point of view:
The big problem, Holzgang writes, is that neither the restaurants that are ranked nor the roughly 1,400 citizens who did the ranking (the latter of whom volunteered for the duty, and hence form a kind of random selection) constituted what a statistician would consider a satisfactory "population." Those responding to the survey, that is, almost certainly lacked any sort of common background or history of judging restaurants; and though the original catalogue of restaurants to be rated was suggested by Zagat editors, participants were free to write in additional places or to ignore as many of the listed restaurants as they wished--and thus in effect to choose the places they would judge themselves. This probably means, writes Holzgang, that most people ranked primarily the restaurants they either loved or hated--inherently biasing the process. "To establish the validity of a survey", Holzgang continues, "you must be able to classify (the statistician's word is stratify ) the population into some sort of common groupings that are more or less homogenous; and the selection of items to be tested must be unbiased by any consideration that may affect the survey results." Clearly, Zagat fails on both counts.
Too, Holzgang points out, restaurants that have a small but enthusiastic following might well score higher by the Zagat method than their "actual" ranking, due to block voting by their admirers--and obscure restaurants might do better still under these circumstances, since they would be unlikely to be known by patrons who would give them low or unfavorable rankings to balance out their higher scores. "The numeric ratings that the survey publishes," he continues, "have no common basis and there is no argument I can think of to torture them into a semblance of uniformity. They are essentially meaningless."
Holzgang is quick to add, though--and in this I agree with him--that the Zagat guide is very useful for providing addresses, hours of operation and other nuts-and-bolts information about a wide variety of local eating places, and that the abbreviated comments listed for most restaurants are of both interest and practical use.
If you want the chance to add your own (statistically questionable) scores to the Zagat tote board, send a stamped, self-addressed envelope to Zagat Survey, 2226 Moreno Drive, Los Angeles 90039. But hurry. Deadline for the receipt of completed forms is June 15.
PORK QUOI?: The May edition of France's popular Gault-Millau magazine recommends to Parisians "Un 'deli' comme a New York"-- "a deli just like in New York"--called Sami's, on the rue d'Edimbourg. Among the dishes the publication is particularly taken with at Sami's are "the best pork spareribs in Paris" and "a very good meat loaf of ground beef, veal and pork." Uh, guys. . . . There's something we ought to tell you before your next eating trip to Gotham.