A couple of years ago I came across Paul DiMaggio and Michael Useem’s “Cultural Democracy in a Period of Cultural Expansion: The Social Composition of Arts Audiences in the United States” (1978) which is a condensed version of their NEA Report (published with Paula Brown), “Audience studies of the performing arts and museums: a critical review” (1978). The pieces are studies of published and unpublished reports as well as surveys which had no formal write-up summarizing results of surveys.

In all, some 268 (NEA lists 270) audience surveys from various arts and humanities organizations were analyzed. For the purposes of this post I’m only going to focus on the Orchestra Audience Surveys in a general sense. As the authors state, “[o]f the studies, more than 80 percent dated from 1972 or later and almost none were earlier than 1966,” which mean that for all intents and purposes we’re looking at studies from 1966 (which I’m assuming is the Baumol and Bowen work) to the publication of the article in 1978–some twelve years.

The piece discusses the problems with using data with audience surveys such as the fact that many may exclude data by young respondents which would obviously skew median ages upwards

Most of the surveys analyzed had studied only people over a certain age. Surprisingly, however, there were no systematic differences between the findings of studies that restricted their sample populations and those that did not. We suspect that many studies either formally or in practice excluded young respondents without reporting that they did so. Furthermore, it would appear that, except for science and history museums, individuals under the age of sixteen usually represent a negligible percentage of arts audiences.

**Median Age of Audiences, by Art Form**

I think one of the more interesting findings, given the short time span, is the range of Median Ages we see for Orchestra Audiences. Here’s Table 2 from the paper which lists the range and frequency distribution of the Median Ages.

A range of Median Ages between 24 to 49 (with a Median of Medians of 40). Note that practically all of these studies are by individual organizations (i.e. individual Orchestras) so that would imply that at least one Orchestra can report an audience with a Median Age of 24–which is well below the Median Age of the US for any of the years within the span of DiMaggio and Useem’s paper (1966-1978). This also means that at least one Orchestra has reported an Audience with a Median Age of 49 which is nearly twice the median age of the US population in 1970 (MA = 28.1) when the latter had its lowest point since the earlier part of the 20th century.

This Median Age of 49 is well above the US Median age during any of the years in question within that range–and is well above the Median Age for Classical Music in most of the NEA SPPAs (1982 MA = 40) which hadn’t yet been conducted as of publication of the paper. In fact, a Median Age of 49 wouldn’t be reached in the NEA SPPAs until 2008–which is 30 years after the publication of the paper. It must be noted that the SPPAs only give Median Ages for “Classical Music” which includes Orchestral as well as Chamber Music concerts (Opera has its own reported Median Age) so we don’t know if Orchestral audiences have a higher or lower Median Age than the Classical Music Audience as a whole.

What we can see is when we’re talking about individual Orchestra surveys, we’re going to get a pretty wide level of dispersion between the Median Ages. Many of those values don’t line up very well with the Crisis hypothesis of a “Graying Classical Music Audience” (i.e. that the Classical Music Audience is aging at a faster rate than the population as a whole). If we were to just look at the Median Age data from the largest studies Audience Studies (Baumol, McKinsey, NEA SPPA) we see an Audience to Total Population ratio which is perfectly consistent with an aging population. If we take studies from smaller and unrepresentative samples and add them into he mix, as some have done, we can chart an emerging and widening gap between the Median Age of Audiences and Median Age of the Population.

Contrarily, we could also easily chart an Orchestra audience that is growing *younger* if we select the studies with the oldest Median Ages from the past to show a trend in the opposite direction that Crisis folks are showing. For example, take the survey with the Audience Median Age of 49, which was obviously published sometime before 1978, then follow that up with the 1985 Audience Survey by the Minnesota Orchestra which shows a Median Age of 48 (which is still significantly higher than the 1982 SPPA Median Age of 40) and follow that with the 1992 SPPA Median Age which is 45 and we’ve just shown a trend for an audience that is getting younger.

Rather, what we should do is take the totality of all studies and see how they fit with what we know to be the case with studies of most any sort–that the largest studies (e.g. Baumol, McKinsey, NEA SPPA) will be near the average and that small studies (such as the individual Orchestra Audience Surveys) will be spread on both sides of the average. This is actually one of techniques used to show whether or not there is publication bias within a field. A scatterplot, (i.e. Funnel Plot – symmetric inverted funnel shape) arises when we have ‘well-behaved’ data. We have the large studies which show a consistency in the rising median age of audiences to the rising median age of the population as a whole; and we have significantly more smaller studies which show median ages of audiences significantly below or above that expected average.

The wildly divergent results for Audience Median Ages goes back as far as the early Grant and Hettinger which show lower than expected Median Ages (the LA Philharmonic and Grand Rapids Symphony audience Median Ages were 33 and 27, respectively) and the study mentioned above in DiMaggio and Useem’s set (MA = 24) all the way up to the DiMaggio/Useem survey (MA = 49) and the aforementioned 1985 Minnesota Orchestra study. Note also that in 1989 the Minnesota Orchestra produced another survey which resulted in an audience Median Age of 55–also well above the average we get from the larger studies.

That the Meidan Age of Orchestras is rising isn’t in doubt–we should be surprised to find it not rising with the Median Age of the population as a whole. That it is rising at a faster rate than the population isn’t problematic either as it would have to be the case if it were rising proportionally to the population. That it is rising at a faster than expected proportional rate isn’t at all clear and only through selection bias of the studies and surveys, as well as a naive understanding of statistics and how data sets behave, can we be led to such a conclusion.

For more posts on the Aging of Orchestra Audiences, please visit this link.

______________________________________

**REFERENCES**

Baumol, William and William Bowen (1966). *Performing Arts, The Economic Dilemma: a study of problems common to theater, opera, music, and dance*. New York: Twentieth Century Fund.

DiMaggio, Paul and Michael Useem (1978). “Cultural Democracy in a Period of Cultural Expansion: The Social Composition of Arts Audiences in the United States.” *Social Problems*, Vol. 26, No. 2, pp. 179-197

DiMaggio, Paul, Michael Useem and Paula Brown (1978). “Audience studies of the performing arts and museums: a critical review.” National Endowment for the Art, Research Division Report #9. <<arts.gov/publications/audience-studies-performing-arts-and-museums-critical-review>>

National Endowment for the Arts SPPAs and summaries may be accessed at the NEA publications page here: http://arts.gov/publications

## One thought on “The Hazards of Drawing Conclusions about Median Ages from Orchestra Audience Surveys”