Demolition Man

Ed Kilgore offers a mixed review of Mark Penn's ludicrous Microtrends that takes a whack at Penn detractors:

A sprawling book like Microtrends, which purports to identify seventy-five distinct subcategories (sixty-four American, and eleven international) of people who have yet to get noticed by corporate and political marketers, provides plenty of targets for Penn detractors. In a review for In These Times, Ezra Klein cherry-picked some of the sillier and sloppier sections of the book, and constructed a demolition not just of Penn, but of political pollsters generally.

Ed doesn't approve of deploying the book in this way, but it seems like a telling concession to me. Penn, after all, is a political pollster. A highly respected and highly successful political pollster. And yet it's possible to "cherry-pick" several instances of sloppy or mishandled interpretation of survey data from Mark Penn's book. That seems important to me. Sometimes cherry picking is bad. But a person who purports to be a professional collector and analyst of survey data should rarely if ever make elementary analytical mistakes. Certainly he shouldn't make them in a book. It tells you something about the state of the profession that one of its leading practitioners can routinely make mistakes like this:

Not surprisingly, men are more flirtation at work than women (66 versus 52 percent, according to one survey); and substantially more men (45 percent) have had an interoffice romance than women (35 percent). That latter discrepency either means that men are serial Office Romancers; that men are more honest about this; that women more often leave the workplace after having an affair; or that some of those men's affairs are homosexual. I think the first theory is most likely -- the office has become the twenty-first century singles bar. Water is the new gin and tonic, and Muzak the new club beat.

That's dumb. Obviously, of those explanations the one that's "most likely" to be true is that some of those men's affairs are homosexual since we're absolutely sure that some men are gay men and that, therefore, some affairs are gay affairs. What Penn means is that he doesn't believe gay men fully explain the gap, and he's using the wrong words because he doesn't understand what he's talking about.

Another possibility he rejects as unlikely is that this has to do with women leaving the workforce. But, again, we know it's true that women are more likely to drop out of the workforce (after childbirth, for example) and that, therefore, the men in the workforce are older on average than the women and therefore likely to have more experiences in general than are women.

He also rejects the possibility of an honesty gap. And, indeed, anyone who knew anything about survey data (i.e., not professional survey data analyzer Mark Penn) would tell you that a gap between male and female self-reports of sexual activity is typical.

Last, after rejecting as unlikely three things that are definitely true, he homes in on the "most likely" explanation for why a higher proportion of men than women have office affairs -- "men are serial Office Romancers." The only problem here is that Penn's preferred explanation is inconsistent with the data! If (as Penn falsely believes) no gay people have jobs, men and women drop out of the workforce at the same rate, and men and women both accurately self-report their data, then Penn's result could only be explained by women being serial Office Romancers.

And though Ezra stands accusing of "cherry-picking" he didn't fit this into his review, nor did he mention the bizarre snipers poll that Penn uncritically accepts. There's a serious question here about Penn's field that someone who's this sloppy at the handling of his subject matter can be a leading member of the profession.