The fallacy of composition in brownstone Brooklyn

The following book review appeared in the New Yorker recently:

osman.png

I don’t know much about Brooklyn and I haven’t seen the book in question but I think the reviewer is making a classic statistical error in the last sentence above. Various people bought houses in these Brooklyn neighborhoods: some of these people were white, some were not, some were liberal, some were not, maybe some styled themselves as romantics but surely many did not, some fought the construction of low-income housing projects and probably most were apolitical. That various people involved had different goals does not mean that many individuals were torn in the way described above. The fallacy of composition is to attribute a common view to each member of a diverse population. Better to think about variation: imagine the different attitudes as circles on a complicated Venn diagram.

4 Responses to The fallacy of composition in brownstone Brooklyn

  1. Barry June 29, 2011 at 3:53 am #

    It seems to me that the subject of the last sentence — those who were torn — were self-styled romantic white, middle-class, liberal Brooklyn brownstone owners. To me, that seems like a pretty small group that was torn. The author isn’t saying that all brownstone owners in Brooklyn were torn — just that the self-styled romantic white, middle-class, liberal ones were. I guess I don’t see a fallacy of composition here so much as an attribute of a very small subset of terribly pretentious people who go around calling themselves romantics.

  2. Daniel Lippman June 29, 2011 at 9:14 pm #

    Osman is one of the best thinkers I know on American cities; he really knows what he’s talking about. He’s also a great professor in American studies at GWU.

  3. Andrew Gelman June 29, 2011 at 9:15 pm #

    Daniel:

    Just to emphasize: I haven’t seen the book, and my criticism applies only to the above review.

  4. Vic July 16, 2011 at 4:22 pm #

    I think a more charitable reading of the review suggests it’s not a statistical fallacy but a tautology.

    “These self-styled romantics…found themselves torn between A and B.”

    You could interpret “these self-styled romantics” as referring to the set of all of the “various people [who] bought houses in these Brooklyn neighborhoods”, but I think it’s fairer to interpret it as the set of “white middle-class liberals” who held certain attitudes such that they (1) envisioned Brooklyn as a ‘village’, (2) bought and renovated brownstones with the aim of restoring some sort of aristocratic grandeur, (3) evicted tenants along the way, and (4) found themselves torn between A and B above.

    You could take (4) to be a condition for the definition of this set of people, or, loosely speaking, as a kind of plausible inference from (1) to (3). Either seems like a better reading to me.