Opening the Curtain on Emily Glassberg Sands, Part II

Last week I took apart the final chapter of Emily Glassberg Sands's (at left) much-discussed study on sexism in the theatre, and her (illusory) conclusion that Broadway producers had closed female-written shows "early." Indeed, I demonstrated that her entire argument there was a charade of averages, dummy variables, and proxies for data she actually lacked. Next, I mean to look more closely at her "script experiment," in which a survey of regional theatre artistic directors and literary managers revealed that women rated female-written scripts lower than men did.

But before I do, let's take a time out to consider the press response to Sands's work for a moment. I find it striking that no one else in the mainstream media or the blogosphere has questioned Sands's final chapter, even though only a moderately educated person like myself could quickly see the huge holes in her argument. And several heavy hitters have reported on the study - the New York Times did, twice - once even in its "Economix" blog! New York magazine also wrote about Sands's work, the Guardian posted the thoughts of two different writers on her conclusions, NPR interviewed Sands on the air, and the LA Times published a hilariously uninformed piece on its "Culture Monster" blog. Yet none of these writers questioned the study seriously; Sands's somewhat-sophisticated statistical snow job sailed right over all their heads.

Why did that happen? Part of the reason is the technical illiteracy of most journalists, and, of course, most writers on theatre. But another reason is the backing Sands had from three heavy hitters - Steven Levitt, the co-writer of Freakonomics (who seemingly suggested the study to Sands, at the behest of playwright Julia Jordan), Christina H. Paxson, the chairwoman of Princeton’s economics department and the newly named dean of Princeton’s Woodrow Wilson School of Public and International Affairs, and Cecilia Rouse, a member of the White House Council of Economic Advisors, and co-writer of a famous paper that inspired Sands. All these people "vouched" for Sands (according to the Times), and all have been somewhat derelict in their academic duty in this case, I'd say - not because of any intellectual dishonesty within Sands's study itself (she's upfront about the gaps in her data in print, although less so in public), but rather in giving her work their imprimatur as unassailable in its methodology, and allowing it to be floated as a kind of cultural touchstone. It's obvious that either they hadn't actually read the study carefully, or were simply sure that sexism must exist on Broadway, so the gaps in methodology didn't matter. And undoubtedly they have their defenses for their behavior already formulated; they've all toed their "professional" lines (the wordings in their statements have been careful); they've merely allowed uneducated journalists to draw their own conclusions. But nevertheless, a false impression has been created, and they've been behind it.

But why are the gaps in Sands's last chapter so important, anyway?

In a word, because her final, flawed conclusions are the only time she "finds" evidence of sexism in the theatre.

But how is that possible, you may ask, when there are so few plays by women being produced?

Hints regarding the answer to that question are easily found within the study itself. But first, I want to say upfront that yes, very few plays by women have reached Broadway - less than 8% of Broadway productions are by females, in fact. And it's undeniable that in the not-so-recent past, the theatrical world, like the world in general, was just as sexist as it was racist. I'd even argue that sexism remains a more powerful force than racism in the country at large; or at the very least it seems true that sexist statements and actions can more easily be culturally cloaked than racial bigotry can.

But is that true of the theatre? I'm not so sure. The stage is its own political world, with a decidedly left-liberal tilt, and is chock-a-block with aging radicals and former hippies. In several cities (such as mine, Boston), it's entwined with the academic left. Looking around my local burg further, I can't help but note that all the major print critics are female - and all of them are middlebrow feminists. (In New York, it's true, almost all the important print critics are gay men - which leads to a certain catfight atmosphere, but little, I think, overt sexism.) And when you add into the mix the fact that the majority of theatre tickets are bought by women, you realize you're looking at an industry where, at least in my hometown, women are the critics, and the customers - and increasingly the artistic directors, too (we've got two new women A.D.'s in Boston).

Thus, unsurprisingly, in Boston (at least) one often hears that "we need more plays by women." (While I've never heard anyone ever say, "We need more plays by men!" much less "We need more plays by white men!") And for what it's worth, this is almost always a blatantly political, not aesthetic, statement; it's always made by, yes, a woman, yet without reference to any particular text, or theory, or even author. It's just an un-self-conscious political assertion: we need more plays by people like me! I suppose there's nothing wrong with that - perhaps it's how everyone feels inside - but I do want to point out that it indicates the political currents in the theatre actually run counter to the assumptions of Emily Glassberg Sands. (Indeed, it's no surprise to read in her study that there are some thirty women-only theatre companies, such as New York's New George Theatre, in existence.)

So why the gap in play production? Well, first of all, outside of Broadway, Sands's study indicates that there isn't much of a gap. This conclusion must, of course, be taken with a grain of salt, because the source of her data is www.doollee.com, a site with a gigantic, but largely self-reporting, playwright database. Still, that grain of salt is much smaller than the one we had to swallow regarding her conclusions about Broadway, because this time she at least has the type of data she needs, and she deploys it in straightforward multiple regressions, or in even simpler cross-tabulations.

And Sands admits flat-out at the top of her analysis that "I find that scripts on Doollee written by women are equally likely as those written by men to be produced" (p. 41). The gap in play production, it seems, is largely tied to the fact that there are more than twice as many male playwrights as female playwrights, and what's more, those male playwrights write more plays. Indeed, in the more restricted sample Sands must use for her multiple regressions, she even discovers that "female-written are slightly more likely to reach production." (Emphasis added, p. 44.)

Hmmm. Undeterred, however, Sands keeps digging - and we quickly realize that once again we're on a freakin' Freakonomics fishing expedition. Sands tells us that "it has been hypothesized that, in order to get their works produced, female playwrights feel compelled to write plays with fewer total parts." But this oppression scenario doesn't get her very far - it turns out that on average, female-written plays have 6.8 parts, while male-written plays have 7.7 parts - a difference which she's forced to admit "reduces the likelihood of production by about 0.7 percentage points" (p. 44). And indeed, later she discovers that women are "1 percentage point less likely to have one work produced." That's right: one percentage point (the difference is slightly higher - 4 percentage points - within a restricted American sample).

You get the picture: Sands has got nothing, or at least not much. But she keeps scrambling: she notes that female playwrights are somewhat less likely to have an agent (but this doesn't seem to matter much in gaining actual production). They also write scripts with predominantly female roles - and such scripts are six percentage points less likely to achieve production (again, not that much, but intriguing given the results of her "script experiment;" more on that later). To be fair, she all but admits defeat in the conclusion of her chapter about Doollee: "I find ample occupational differences between men and women, but no evidence of employment differences between male and female playwrights," she sighs.

But when Sands doesn't like what comes out of her equations, she is quick to doubt her data: "These results, however, must be considered with an eye to the incomprehensive nature and likely sample selection problems of the data" (p. 53). It's ironic, therefore, that she wraps her study with a "conclusion" based on faultier, or even non-existent, data, don't you think? Still, her search for sexism does inspire Sands to implement what's by far the most interesting, and certainly the most statistically valid, part of her study - her by-now-notorious "script experiment," in which she sent play excerpts by female playwrights - but attributed to males - to a large sample of theatres. More on that intriguing gambit in the third part of this ongoing series.