Previous: Completion Rates and Attrition
Because the survey was Internet-based and open to the public, the respondents are not a random sample of the population of musicians. However, one can observe three factors to check the representativeness of our survey sample.
First, one can look at response rates by partner organizations. Table 2 reports the approximate membership of several music organizations, many of which partnered with us to promote the survey; the number of respondents that reported being a member of each organization; and the calculated response rate for each music organization. Some of the organizations, especially the larger ones, include both individuals and organizations (such as publishers and arts presenters) within their reported membership rolls. Thus, the response rates I have calculated are only a rough estimate.
The estimated response rates are nonetheless informative about the sample. For instance, the AFM—the largest musicians’ union—participated at a much higher rate than other organizations, 2.9%. This makes sense based on the AFM’s relatively eager cooperation with the research team. The response rate from the National Academy of Recording Arts and Sciences (“NARAS”)—the organization behind the Grammy awards—was also high, at 2.3%. Other organizations with participation rates above 2% include Chamber Music America, Early Music America, Folk Alliance, American Music Center, Jazz Education Network, American Composers Forum, and the Association of Performing Arts Presenters.
Based on the organizations whose membership participated at the highest rates, the sample is likely to have overrepresentation from the classical and jazz genres. This is reinforced by the relatively high concentration of classical and jazz musicians within AFM. [Note 16] On the other hand, our sample does have substantial representation from other genres. Across the entire sample, 48% of respondents listed genres other than classical and jazz as primary. But it is important to keep the classical and jazz focus in mind when interpreting the aggregate statistics reported in this Article.
As a second type of check for representativeness, one can compare some of our aggregate statistics to those from government surveys of the labor market.
The Occupational Employment Statistics, produced by the Bureau of Labor Statistics, report the hourly wage distribution for the category “Musicians and Singers.” [Note 17] The government’s figures only pertain to musicians who are employees; self-employed workers are not part of the analysis. [Note 18] The government estimate of the mean wage for musicians is $31.74 per hour, with a median of $22.99. [Note 19]
The survey asked respondents for the number of hours spent on music per week, total income, and percentage of income derived from music. [Note 20] From those three questions, I have calculated an estimate of hourly wages. Among the subset of respondents in the sample who collect some part of their income as salaried musicians (usually as orchestra players), the estimated mean wage is $28.91 per hour, with a median of $20.07. The proximity of the survey estimate to that of the Bureau of Labor Statistics provides some confidence in the representativeness of the sample.
Third, one can compare our results within particular genres or roles to the results of previous studies conducted within those genres or roles. The scholar who has done perhaps the most similar in spirit to our own study is Joan Jeffri of the Research Center for Arts and Culture. Her 2009 study of composers collected some of the same variables we have collected.[Note 21] The 1,347 individuals in Jeffri’s sample appear to play instruments and engage in live performances in addition to composing. [Note 22] Similarly, the 2,660 respondents to our survey who report doing at least some composing play many other roles as well, such as recording, performing live, doing session work, teaching, or orchestra playing. An exact apples-to-apples comparison is not possible, but some questions in each study sought the same information.
The composers in our sample look similar to those in Jeffri’s sample for variables including income distribution, percentage of income from recordings, percentage of income from songwriting royalties, age, gender, race, and ethnicity, and hours spent on music per week. [Note 23] The participants in Jeffri’s survey also reported a mix of attitudes about unauthorized downloading, which accords with our results discussed below. [Note 24] The main differences between the statistics collected in the two studies are that the composers in Jeffri’s sample are more focused on the classical and new music genres and accordingly receive more revenue from commissions and grants. [Note 25] Overall, a comparison between the two studies suggests that the studies’ findings are largely similar where the questions asked overlap. Previous studies do not provide points of comparison for the survey questions that focused on specialized revenue streams, copyright law, or other detailed institutional features of the music industry. But where points of comparison do exist, the results generally support the validity of our survey estimates.
______[Note 16]: In our sample, 53.6% of AFM members reported classical as their primary genre, along with 17.7% reporting jazz. [Note 17]: Occupational Employment Statistics: Musicians and Singers, BUREAU LABOR STATS. (2011), (last visited Jan. 28, 2012). [Note 18]: Id. [Note 19]: Id. [Note 20]: The specific questions are Question 5, Question 16, and Question 17, respectively. We instructed respondents to answer the income questions as individuals (for example, “What’s your personal annual income?”), even if they chose to answer other questions from the perspective of their band or ensemble. [Note 21]: JOAN JEFFRI ET AL., TAKING NOTE: A STUDY OF COMPOSERS AND NEW MUSIC ACTIVITY IN THE UNITED STATES (2009). [Note 22]: Id. at 2–3 (showing that live performances account for 15% of professional composers’ income and 24% of nonprofessional composers’ income). [Note 23]: See id. at 2 (demographics); id. at 23–27 (all other variables). [Note 24]: See id. at 14–15. [Note 25]: Id. at 17, 23.