The Year of Living Statistically, Part II


By Thomas Garvey
Special to The Clyde Fitch Report

Story continues below.

The response to my first post on the Clyde Fitch Report was, it turns out, record-breaking, and almost entirely negative — perhaps because people still like their sexism, and still like their statistics, and don’t like being weaned from either.

But few of those who “wrote in” — or threw tantrums on their own blogs — seemed ready to deal with the contradictions in their own political and statistical assumptions. Feminists clung to older, over-arching numbers, but resolutely ignored the latest information (that men actually aren’t biased against women’s scripts, and that female authors are finding plenty of productions). Meanwhile outraged “critical thinkers” insisted that there was somehow a contradiction in citing some statistics as valid while critiquing others as misleading. And all this confused rhetoric was of course administered in a peremptory moral tone — always an indication that those crying for “dialogue” were only
interested in enforcing monologue.

Story continues below.

The welter of commentary was perhaps at least partly traceable to a widespread naivete about both statistical techniques and how applicable they might be to the world of theater. Speaking more generally, there’s an odd contradiction at the center of the worldview of many of today’s artsy “critical thinkers” — these folks dote on books like Freakonomics and writers like Malcolm Gladwell because they seem to show a way to justify progressive action without relying on controversial political ideas. But, at the same time, not too many liberal-arts graduates have taken any courses in statistics; I doubt few of my attackers could derive a regression from a data set, much less attempt marginal analysis. In a word, they don’t actually understand the techniques they’re trusting. So in a way, their positions are just as faith-based as those of the Southern Baptists.

Indeed, when Emily Glassberg Sands (I know, I’m sick of her too) got up in front of her New York audience with her Powerpoint slides, she was trading not only on her audience’s valid sentiment against sexism, but also on its ignorance and its faith in statistical analysis — a faith that in the case of theater is completely misguided if not flat-out wrong. That’s why, in so much of her paper, Sands flailed so obviously. But no one in the theater community wanted to call her out on her mistakes; instead, they seized on her fudged conclusions as validating their own assumptions, while ignoring her more-valid audit study because it led somewhere new and politically troubling. Or, if they didn’t ignore it, they twisted it into some weird assertion that, yes, women still faced sexism, anyway: this was the Word that Emily had brought down from the mountain. “Praise the Goddess and pass the Excel spreadsheet!” became their battle-cry.

For many reasons — some subtle, others obvious — the theater resists the techniques of statistics. Valid statistical analysis requires large, reliable data sets — for mathematical reasons I won’t go into here, you really need at least 500 data points, and preferably a whole lot more, before you can talk in a meaningful way about “margins of error” or much of anything else. And Sands had nothing like that. When she did have the numbers (and sometimes she had access to gigantic databases) the data was self-reported, thus unreliable. Or the required information simply wasn’t available at all. She had to dream up a kind of Ptolemaic system of dummy variables to get her equations to work. And even then, with phony numbers going in one end, only mild statistical effects came out the other.

But pay no attention to that woman behind the curtain!, her advocates reply. Just look at the big numbers, Thomas Garvey! There are still so few plays by women on Broadway! How do you explain that, Mr. Smarty-pants?

Well, certainly there are more and more, but still relatively few, plays by women on Broadway. But that fact actually ties into the deeper way theatrical practice is a completely inappropriate target for Sands’ statistical stalking-horse.

Indeed, when I glanced back into her thesis before I began writing this article, I had to laugh out loud: Sands chatters on about “human capital differences” and Chicago price theory as if she were preparing to analyze a large-scale industry that produced widgets, from which she could tease out smooth curves on coordinate axes.

But the theatrical “industry,” if you want to call it that, isn’t so much an industry as an interlocking set of guilds in which unions and oligopolies have learned to coexist. What’s more, theater doesn’t really “scale up” the way mass production does and it hardly produces “widgets” — or any other product that can be broken down into aggregate components of profit and cost. Ticket prices are “sticky” and stratified, for instance, while the production costs behind them can vary wildly — there’s little reliable relationship between cost and price point. And, of course, if a show proves successful, it takes a great deal of time for it to “roll out” — a national tour might follow a major production, which might then lead to regional productions, over a span of something like several years. And at the same time, the theater is under constant economic pressure: It’s not an expanding economic enterprise but practically under assault, a situation widely known to be hostile to newcomers. Many, if not most, of the “widgets” that eventually make it to the theater’s single, clearly for-profit sector — Broadway — have actually been “developed” under nonprofit conditions. They’re charity chases, promoted by cultural or political aims rather than economic ones.

To be fair, Sands gave lip service to many of these caveats in her thesis. But essentially she went ahead and ignored them anyway; she didn’t seem fully aware that the essential correspondence between statistical techniques and what one might call observable economic reality didn’t hold up when it came to theater.

And there were other problems in her latent assumptions that Sands didn’t seem to even dream of. In the kind of economic world best suited to statistical analysis, widgets are never “revived” — older widgets never drive newer widgets off the market, and there’s no such thing as a widget that cannot be reproduced. There’s no Shakespeare widget, no Sondheim widget, so no resulting vogues for Sondheim-like or Shakespeare-like widgets. There are just widgets.

Which leads to an even deeper problem: in theater, there really aren’t any widgets. Great plays, and even good plays, are often unique, and, in strict terms, “incomparable.” The audience for one may not accept another. And, of course, great (or even good) playwrights hardly arrive on a steady, predictable basis; we count ourselves lucky if the whole
country generates more than a half-dozen interesting dramatists in a decade, and, in historical terms, it’s unusual for more than a few playwrights to last from a given century. Then there’s that odd problem of “development” and the fact that a great production requires not only a great script but a great director and great designer and great cast — in an arrangement that’s less like an assembly line with a roster of suppliers than a constellation of unique sympathies and talents. This means that relationships count more in theater than they do on a factory floor. It also means that, in a cold economic environment, a talent with a track record gets precedence over an untried one.

But hold on a minute. Does this, as many claim, describe a form of “structural” sexism? That is, even if we toss Emily Glassberg Sands’ study and admit that the current faith in statistics is often misplaced, doesn’t the fact that her techniques were misguided actually reveal a hidebound structure in the theater anyway — one that is hostile to women and minorities and resistant to change?

To hear my answer to that cliffhanger, you’ll have to wait for the third part of this ongoing series.

Thomas Garvey has at various points in his checkered career impersonated a director, screenwriter, architect, strategic analyst, and Boston Globe theater critic. He’s still impersonating a critic at, where you can read his “cantankerous, but brilliant” reviews of theater, music, art, film and dance.

  • RLewis

    Ya know, I’ve followed this Women’s Plays Stat’s thread for months, from blog to blog to New Dramatists’ town hall, and what I can’t figure out is why anyone is comparing “plays on Broadway” to anything. There are so few of them these days that I don’t know how anyone can say that there is a quorum sample that is statistically sufficient to be a control.

    What there is on Broadway is a ton of musicals, but I see no other place in NYC or the US where musical creators can be accurately compared. And no one seems to want to analyze musicals anyway. (Is that because musicals on Broadway are a gay man’s domain to which neither str’ men nor women of any preference are dominate?). Playwrights are everywhere except Broadway where composers and lyrists rule. So, shouldn’t we be drumming up statistics of plays by women compared to plays on LORT contracts, $250K productions, or some other criteria?

    I can see working up accurate statistics of red apples from one tree to green apples in an orchard; but, from plays by women to plays on Broadway really seems like apples and oranges. What am I missing?

  • Thomas Garvey

    Not much.

  • As a professor myself, what I find most objectionable is that an undergraduate student was put by her professor in the public eye. I don’t care what the data seems to show, she is an undergraduate whose work is likely to crumble under intense examination. That is the nature of being an undergraduate. Had Cecilia Rouse not jumped out of academia to work for Obama, she should have been reprimanded. You just don’t do this to young people.

  • Thomas Garvey

    But Emily was so eager for it, Scott. And she worked her fifteen minutes so hard, even though she knew most of her conclusions were pure bull. You should have heard her chattering away about how her paper was about to be accepted for publication by a major journal and how she’d landed a spot on the Colbert Report. (Did either thing ever happen?) For awhile she maintained a blog about all her press mentions (she may still be doing that). I don’t think anyone “put her in the public eye” but herself.

  • John F.

    The misuse of statistics continues in theater. The recently published Outrageous Fortune, a study of the development (or lack of it) of new plays falls prey to the same problem. Any cursory examination of the list of persons and organizations on whose interviews the report is based will reveal that the participants were far from a representative sample of the groups they were intended to represent. That is not to say that the interviews didn’t reveal interesting qualitative data. However, giving these findings the gloss of statistical validity is a real problem, because they are being treated as a new gospel by theater professionals. As a theater professional who did learn statistics, I find all of these studies too self-serving to take seriously.

  • Pingback: The Year of Living Statistically, Part III: The Quality Question « Clyde Fitch Report()