Essentializing food fashions

« previous post | next post »

English is a fine language in most respects, but its morphosyntactic resources for talking about sampled properties of groups are remarkably poor.  (Not that other languages are any better, as far as I know.)  In particular, English speakers have no simple alternative to the use of generic statements about whole groups or typical members, when discussing cases where between-group differences are significant, but small relative to within-group variation — Xs are P-er than Ys, or Xs prefer A to B, or an X prefers A while a Y prefers B.

The result is especially unfortunate when it involves invidious comparisons in socially sensitive areas, but the problem comes up whenever anyone discusses properties of groups.  This morning, Stephen Jones brought a particularly striking example to my attention: an article in today's Independent, "Oodles of noodles: Britain prefers Chinese to curry".

Here's the crucial paragraph:

In a poll, 83 per cent of adults liked eating tangy Chinese, ahead of the 71 per cent who favoured highly-spiced Indian food. When eating out, Britons also prefer Peking duck to a lamb balti – almost a third of people have visited a Chinese restaurant in the past 12 months compared with 30 per cent who have been to a curry house.

Of course, the language does offer us some choices, even without resorting to talk about quantiles and standard deviations. A better way to describe these numbers would probably be to say that Chinese and Indian restaurants are now about equally popular in the U.K. But the reporter (or the flack who wrote the press release) decided that framing the story in terms of a preference for Chinese food  would attract more readers.

At least in this case, the reporter does give us the numbers (if not the confidence intervals) right next to  the egregious over-interpretation — and "When eating out, Britons … prefer Peking duck to a lamb balti" turns out to involve a comparison between 32% and 30%, which is only slightly disguised as "almost a third" vs. "30 per cent".

Compare the CNN/health article on innate differences between boys and girls, discussed here and here,  which gives no numbers with its generic statements ("Girls' hearing is more sensitive in the frequency range critical to speech discrimination"; "From birth, a girl baby tends to be more interested in looking at colors and textures, like those on the human face, while a boy baby is drawn more to movement, like a whirling mobile"), although the numbers turn out to be even less meaningful than those behind the "preference" of Britons for Chinese over Indian food.


  1. Stephen Jones said,

    April 2, 2009 @ 8:33 am

    The claim 'Britain prefers Chinese to Indian' is particularly ridiculous when you read that against the small preference in restaurants, the figures on supermarket spends are £556m to £367m in favour of Indian food.

  2. Faldone said,

    April 2, 2009 @ 9:19 am

    My favorite line from the story so far is:
    almost a third of people have visited a Chinese restaurant in the past 12 months compared with 30 per cent who have been to a curry house.

    How significant is that?!

  3. Mark P said,

    April 2, 2009 @ 9:31 am

    "… almost a third of people have visited a Chinese restaurant in the past 12 months compared with 30 per cent who have been to a curry house."

    I'm going to leave a bruise on my forehead from smacking it if I see much more like that.

    I didn't see any reference in the Independent to the raw numbers of the poll. I did see that the original report is pretty expensive.

  4. Andrew West said,

    April 2, 2009 @ 9:32 am

    almost a third of people have visited a Chinese restaurant in the past 12 months compared with 30 per cent who have been to a curry house

    Wouldn't "almost a third" be pretty much the same as "30 per cent" ?

    I wonder if the use of "almost a third" (expressed as words) contrasting with "30 per cent" (expressed as numbers) is a deliberate journalistic trick, hoping to mislead mathematically-challenged readers into thinking that there is a significant difference in numbers between Chinese-eaters and Indian-eaters.

  5. Morten Jonsson said,

    April 2, 2009 @ 9:41 am

    "Almost a third . . compared with 30 per cent" reminds me of W. C. Fields, in You Can't Cheat an Honest Man, as the barker whose sideshow boasts The World's Smallest Giant and The World's Tallest Midget. They're identical twins. As Fields says, "They baffle science."

  6. Bobbie said,

    April 2, 2009 @ 9:57 am

    "There are three types of lies – lies, damn lies, and statistics."

    [(myl) In this case — and in the case of the cited CNN sex story — a more accurate version of the maxim would be "… lies, damned lies, and public relations". (For all my friends in PR, of course, I mean that in the best possible way.)

    The history of the quotation is discussed here. It seems to have begun in the early 1890s, probably as a variation on earlier sayings along the lines of "… liars, damned liars, and experts". ]

  7. Picky said,

    April 2, 2009 @ 10:46 am

    And somewhere behind this is a definition of "ethnic" which it would be interesting to see them define. It doesn't include Italian, evidently, unless we've at last given up on the dreaded pizza and lasagne.

  8. Mr Punch said,

    April 2, 2009 @ 11:58 am

    All the objections seem valid, but still … There is a widespread impression that Indian food has become the de facto standard restaurant/carryout cuisine for Britons, so if Chinese is, or better yet has recently become, more popular — even by a little — that's news. As supermarket sales are presumed to be heavily influenced by the numbers of customers of the respective ethnicities, they are largely irrelevant in this regard.

  9. Stephen Jones said,

    April 2, 2009 @ 12:26 pm

    The Guardian presumes to echo The Independent without any text at all. It claims more prefer Chinese takeouts to Indian takeouts.

    The Independent does give a figure of 83% to 71% for Chinese food, so that may well be the takeaway figure.

    There is of course no historical reference for the takeaway figure; I suspect there has always been a higher proportion of people who like Chinese food to Indian food, because the figure is basically one for people who don't like their food highly spiced.

  10. Adam said,

    April 2, 2009 @ 12:41 pm

    That reminds me of Darrell Huff's book _How to Lie with Statistics_ (1954).

  11. Jonathan said,

    April 2, 2009 @ 1:28 pm

    I rarely eat Chinese food in a buffet, and rarely take out (or "carry away") Indian food. On the other hand, often carry out Chinese and eat Indian buffets. Chinese food doesn't withstand the buffet table, and Indian food doesn't quite do as well on the ride home.

  12. mae said,

    April 2, 2009 @ 1:39 pm

    Monty Python said it better in "I like Chinese" to quote …

    I like Chinese.
    I like Chinese.
    They only come up to your knees,
    Yet they're always friendly, and they're ready to please.

    I like Chinese food.
    The waiters never are rude.
    Think of the many things they've done to impress.
    There's Maoism, Taoism, I Ching, and Chess.

    So I like Chinese.
    I like Chinese.
    I like their tiny little trees,
    Their Zen, their ping-pong, their yin, and yang-ese.

  13. Benjamin Zimmer said,

    April 2, 2009 @ 2:34 pm

    Re "lies, damn lies, and statistics" discussed above: Stephen Goranson has been on the trail of this one. The current date to beat is June 13, 1891, in a letter from Thomas Mackay appearing in the (UK) National Observer. (See Stephen's posts to ADS-L here and here.)

  14. Dan T. said,

    April 2, 2009 @ 8:47 pm

    I've eaten at Chinese buffets much more often than Indian buffets, though both exist near where I live and work.

  15. joseph palmer said,

    April 2, 2009 @ 9:42 pm

    I don't see how any case is made in this article that there is a problem with the language rather than the journalists. What am I missing?

  16. joseph palmer said,

    April 2, 2009 @ 9:47 pm

    Oh, the link, my apologies. I'm not convinced though. I think anyone could do better if they actually wished to.

    [(myl) If you mean that the people who wrote the press releases, and the journalists who transcribed them, could have used the English language in a less misleading way, I agree entirely. To the extent that I'm making a serious point rather than a joke, it's that the vagueness of generic references (a group essence? a dominant propensity? a small but statistically significant difference in mean values?) makes it easy to turn a near-meaningless effect into an attention-grabbing headline. ]

  17. Feminist Tweets! « This is What a Feminist Blogs Like said,

    April 2, 2009 @ 11:09 pm

    […] Essentializing food fashions: […]

  18. Sili said,

    April 3, 2009 @ 6:39 am

    (For all my friends in PR, of course, I mean that in the best possible way.)

    I'm mildly surprised that didn't include a link. I thought it'd've fitted the Liberman wit.

  19. Wordoch said,

    April 4, 2009 @ 4:13 am

    This is journo-bilge at its British best. Formula: take stupid, unfounded truism about British people, in this case something like "curry is the other national dish", compile useless uncomprehensive pseudo-sociological study, twist unsympathetic statistics till they refute said truism, and voila! An easy fill, but ultimately unsatisfying and in the end probably quite harmful to one's health. Bit like Peking duck and lamb balti actually.

    Anyone remember that twaddle about "the nation of dog lovers" buying more cats?

    What happened to the Independent? Does tabloid sizing lead to tabloid writing in the end?

RSS feed for comments on this post