Special tabulations of the 2014-2015 Annual Arts Basic Survey (AABS) and the 2012 Survey of Public Participation in the Arts (SPPA)
Civilian, non-institutional population, 18 years and older
Estimates are reported for 50 states and the District of Columbia
Conducted by the U.S. Census Bureau for the National Endowment for the Arts
State-level arts participation rates for the following activities:
- Attending live music, theater, or dance performances;
- Attending art exhibits;
- Going to the movies;
- Visiting buildings, neighborhoods, or other sites primarily for their historical or design value;
- Reading literature;
- Personally performing or creating artworks;
- Using TV, radio, and/or the Internet to consume art or arts programming
The SPPA, and the related, short-form AABS, have routinely provided estimates of arts participation rates for 32 states (and 11 metropolitan areas). [i] These specific 32 states are those for which the SPPA and AABS sample sizes, at the state level, are large enough to provide reliable estimates of arts participation, as determined by the U.S. Census Bureau.
Although it has been valuable to report estimates for these particular states, users of the NEA’s arts-participation surveys, as well as NEA senior staff, expressed interest in obtaining arts participation rates of all U.S. states, and the District of Columbia.
In response, the NEA’s Office of Research & Analysis entered into an agreement with the U.S. Census Bureau to investigate methods that may be used to produce “small area estimation” of arts participation. Using the AABS, and the 2012 SPPA module covering media use, the Bureau investigated two “synthetic” methods of estimating state-level arts participation, as well as “direct estimation” from the two surveys.
Experimentation with synthetic methods involved statistically predicting, at the state level, arts participation rates via: (1) regression analysis; and (2) a method that attempted to proportionally distribute national arts participation estimates to each state.
The Census Bureau, however, rejected the synthetic estimates generated from each of these methods. The mean squared errors (MSE) of the estimates were negative, suggesting the estimates generated from the two synthetic models were statistically unstable.
Instead, direct estimation from the AABS and SPPA proved a more successful method of calculating state arts participation rates.
To understand this technique, consider that selection of the 32 states for which SPPA and AABS estimates are routinely reported is based only on state sample sizes and the corresponding variance in state populations. Direct estimation, alternatively, takes into account responses to the survey questions. For small-area estimation of arts participation rates, the Census Bureau accepts estimates with associated coefficients of variation (CVs) of 30 percent or lower. [ii]
For example, the AABS shows that 17.6 percent of Nebraska’s adults attended an art exhibit in 2015. The CV for this estimate is 17.5 percent, a value well below the 30 percent maximum CV recommended by the Census Bureau. In that same year, the AABS indicates that 3.8 percent of the state’s adults went to a live book-reading, storytelling, or poetry event. The CV for this estimate, however, is 34.5 percent. Consequently, for this form of arts participation, direct estimation did not yield an acceptable estimate for Kansas.
This ADP uses direct estimation of the 2013-2015 AABS, and the 2012 SPPA, to report state-level estimates of arts participation rates for which the associated CVs for all states are 30 percent or lower.
The results indicate that western states such as Colorado, Utah, and Washington, along with certain eastern states, including Maryland and Vermont, tend to participate in the arts at above-average rates. A number of southern states—Florida, West Virginia, and Mississippi, for example—generally report below-average arts participation.
On the whole, however, most U.S. states exhibit arts participation rates close to the U.S. average.
[i] For a list of the 32 states captured by the SPPA and AABS, please see NEA Arts Data Profile #5, States of Engagement: Arts Participation by U.S. Geography.
[ii] The coefficient of variation (CV) is the ratio of the standard deviation to the mean or average.