What Happens When a Theater Critic Tries to Use Statistics?

By Ed Sobel, Associate Artistic Director

To put it simply, bad math and a questionable conclusion.

In a recent column Terry Teachout, the generally respected arts journalist at the Wall Street Journal wrote: Facts, it’s said, are stubborn things. Anyone curious about the state of American theater will find plenty of stubborn facts to chew on – some of which are tastier than others – in American Theatre’s annual list of the 10 plays and musicals, not counting Shakespeare revivals and seasonal shows, that are produced most frequently in the U.S… I recently spent a couple of hours poring over American Theatre’s lists and came up with this meta-list of the 11 plays produced most often between 2000-01 and 2009-10…

Mr. Teachout counts a lack of “classic plays” on his meta-list, and concludes:

American theaters have a pronounced bias in favor of new and newish plays by American authors, especially ones that have high public profiles… But it also appears that far too many of those same companies may be steering clear of the classical revivals that are no less central to the continuing health of a theatrical culture – and that is very bad news indeed.

I’m not a statistician, but you don’t have to be one to see the serious flaw in reasoning. He’s mixing individual plays and genre. Say 20 TCG theaters across the country are doing the same new play in a given year. That title will show up on the most frequently produced list for that year. But if those same theaters, in an average four-play season, are all also producing three different “classic” plays, then those titles would not show up on the most frequently produced list. But the proper score at a given theater would be one new play for every three classic plays. And across the field nationally it would be one new play for sixty classic plays. This would represent not a lopsided predisposition toward new work, but rather the opposite.

A quick hypothetical illustration:

Steppenwolf does Endgame, All My Sons, Mother Courage and Intimate Apparel.

Arena does Waiting for Godot, Tartuffe,, Guys and Dolls and Intimate Apparel.

McCarter does Oedipus, The Children’s Hour, The Most Happy Fella and Intimate Apparel.

Tally: 9 “classic” plays, none of which make the list of most frequent, but the one new play, Intimate Apparel does.

A related problem: The TCG lists are cross-sectional and not longitudinal. In other words, if Death of a Salesman receives three productions every year for 10 years it would not make the most frequently produced list in any year, despite receiving 30 productions in the decade, while Intimate Apparel, receiving 9 in one year and 16 in a second for a total of (a lesser) 25, would.

To do this accurately, Mr. Teachout should look at every play from every season of the decade at each theater (not the “most frequently produced list”), categorize them as “classic” or “new”, and then look at the numbers. I don’t know what he’d find, but his conclusions would be drawn from actual stubborn facts, not poor statistical work.

The danger here (and why I’ve bothered to post a public response) is that an article like this, already going viral, becomes received wisdom very quickly, especially when appearing in a publication as influential as the WSJ. It begins to have impact on corporate and government funding attitudes and policies and within artistic institutions.

Frankly, I suspect real research would show at least one different conclusion. Namely, most theaters are doing more “classic” plays than new work. They are just doing the same new plays, and different “classic” ones.

But where was the editorial oversight to catch the statistical blunders at this predominantly business-oriented publication?

Update: This blog post was picked up by The Wicked Stage, which is edited by Rob Weinert-Kendt, an Associate Editor at American Theatre Magazine. Terry Teachout and Ed, among others, have continued this conversation there. Click here for the full post and comments.