As a grant writer, I fill out more than a dozen final reports for funders each season. Theatre companies are often asked in these reports to provide data on the age, gender, and ethnic composition of our audiences. But the hard truth is, the data we’re providing are almost certainly unreliable.

Let’s be realistic: theatres can’t force self-identification on our audience members. Most people want to come to the theatre, see a show, then leave with their friends. They aren’t interested in meeting our nonprofit funders’ reporting requirements. Still, arts administrators have to come up with estimates, because we have to complete those final reports. And unfortunately, any estimation method we select has problems.

The simplest approach to estimating your theatre’s audience demographics is:

  1. Google the most recent census data for your local area.
  2. Write down the percentages of people by age, gender, and ethnicity.
  3. Break up your total annual attendance number to mirror the percentages of people in your area.

Done, right? Except that accurate audience research means not assuming we’re already mirroring our local population. This approach, despite its advantages of being inexpensive and fairly quick, obviously pre-empts learning anything about the people we’re actually serving in the seats. And it may hide populations that we are not reaching at all, hindering inclusivity work.

The second most common approach is to measure audience demographics through surveys. However, most surveys just don't get responses from the vast majority of theatre attendees. People are bombarded with survey requests everywhere from the grocery store to the mall to their online banker, so the result is survey exhaustion. Whether offered on-site, over the phone, by email, or through social media post-show, typically only a small portion of any overall audience responds to surveys, regardless of the incentives offered to participate. The survey respondents, therefore, may or may not be a statistically representative sample of the overall audience.

Social science research demonstrates that more educated, more affluent people are more likely to complete surveys than less educated, less affluent people. Younger people are more likely to participate than older people.  Women are more likely to participate than men. And white people are more likely to participate than non-white people. The survey results we collect are only as accurate as the people who choose to respond. And as a result, survey responses tend to mask a population’s income, age, and ethnic diversity. This problem is known as non-response bias, when the population that takes a survey skews the results compared to what they would be if the survey included non-respondents. Thus, audience survey approaches, though more precise and resource-intensive than census extrapolation, still don’t guarantee accuracy in demographic reporting, because of the large number of audience members who opt out.

The third category of alternate sampling methods, while the most resource-intensive and potentially the most accurate, introduces new problems. Like people who research crowds at protests and political events, do we really want to photograph or videotape every night’s audience, and count and classify attendees who might be people of color? Or use expensive facial recognition software at our box office to guess audience members’ ethnicity, gender, or age by their appearance, without their consent? Such invasive, expensive approaches seem the opposite of the kind of authentic personal relationships theatres want to build with our audience members.

Meanwhile, our marketing departments just can’t get enough patron data. Take Opera Philadelphia, recently held up as an exemplar in Opera News:

With the financial support of several blue-chip foundations, Opera Philadelphia has engaged top-tier consumer product marketing firms to measure all manner of audience behavior—retention, churn, demographics, psychographics, probabilities, preferences and (the holy grail) the Net Promoter Score, a measure of a consumer’s willingness to recommend the product to another… Drilling down by age, gender, education level, potential of attendance, consumption patterns and likelihood of return, one encounters increasingly exotic species of operagoers who may or may not be fruitful and multiply.

Analyzing audience composition in these ways can also be invasive of privacy. In the United Kingdom, regulations have been passed that require arts organizations to get explicit consent for any patron data that they hold. These regulations ensure a “right to be forgotten” by companies with whom people do business. In the US, audiences don’t have such privacy protections. Gathering patron information for future audience “conversion” and donor “targeting” is a theatre industry standard. The marketing jargon evokes religious fervor and military violence at the same time: is our mission really to be converting people to theatrical attendance? or acquiring them like targets?

In quantifying and reducing our audiences in these ways, without their permission, we’re avoiding signs that many people are increasingly concerned about revealing personal information or having it tracked in patron databases. Repeated requests for demographic information can drive some privacy-seekers away from the theatre altogether. Need evidence? Look at how many people unsubscribe from your email list within forty-eight hours of the next time you send out a patron survey.

One of two things is true. Perhaps theatres need more ethical approaches for accurately measuring audience composition along various demographic lines. Such approaches would require audience consent up front, address known problems with survey bias, and include sufficient funding and training for the staff needed to conduct audience research ethically and transparently.

Or, a more radical proposal: we should stop attempting to measure audience composition altogether, since it cannot be done accurately when there’s no mandate for audiences to self-report, without significant resource investment and invasion of privacy.

Theatres can and should focus our diversity and equity efforts on hiring practices onstage and backstage, where self-reporting can be made a condition of employment. We can feature stories by diverse playwrights, cast inclusively, and select design teams with an eye to equity, because these are areas over which artists and arts administrators have direct responsibility and control over decision-making.

When it comes to audiences, however, the best way to be inclusive may actually be to stop trying to gather so much information about who’s coming through our doors. Instead of tracking, or rather imprecisely guessing at, our audience’s identity data, we should respect the privacy rights and personal identities of our audience members. It’s their right to choose when and how to self-identify, not our funders’ right to know. And we must challenge our funders to stop demanding patron data that requires invasive, unethical, and inaccurate data collection.