Are rich voters more likely to favor candidates from the Republican Party? The richest states by per capita income are Maryland, Alaska, New Jersey, Connecticut, and Massachusetts. Of these five, four are pretty reliably “blue” – i.e. leaning Democrat – with Alaska being the outlier. The five poorest states by per capita income are Alabama, Kentucky, Arkansas, West Virginia, and Mississippi, all of which are “red” – typically Republican. So the rich actually tend to vote Democrat, right? Wrong. That inference is an example of what statisticians call the ecological fallacy: it is wrong to make inferences about individuals (in this example, voters), based on the characteristics of groups (in this example, states). In fact, richer individuals, other things equal, tend to vote Republican. Even though a higher proportion of voters in Maryland vote Democrat than we find in Mississippi, within each state, the higher the income of the individual, the more likely the voter chooses the Republican candidate.
Today, ArtsJournal links to a study from Southern Methodist University that aims to refute the following claim that was made during the debate over appropriations to the National Endowment from the Arts:
Federal subsidies for the National Endowment for the Arts, the National Endowment for the Humanities, and the Corporation for Public Broadcasting can no longer be justified. The activities and content funded by these agencies go beyond the core mission of the federal government, and they are generally enjoyed by people of higher-income levels, making them a wealth transfer from poorer to wealthier citizens.
Now, whether you support increased funding for the NEA or not, I had always thought it commonly known that people of higher income levels tend to have higher participation in the arts. Indeed, econometric evidence for that claim is given by the NEA itself (see page 55 of this study, and look at participation by income quartile).
So how does the SMU team try to refute the claim? They look to the characteristics of the cities that host the arts organizations receiving NEA grants, relative to the cities of arts organizations that do not:
Based on a comparison of median household income, the community wealth characteristics of NEA grant recipient organizations and those of non-recipient organizations are remarkably similar. Specifically, there is no significant difference in median household income in communities with an NEA-funded organization and those without. This finding indicates that there is no bias in NEA grant-making either towards or against organizations on the basis of the median household income of the surrounding community.
But that does not refute the claim about arts funding being a transfer from poorer to richer citizens. If the opera company in a city of average income receives a grant that, on a per capita level, is much the same as the grant received by the opera company in a city with a higher average income, it might still be the case that in each city the opera attendees are primarily from the top income quartile.
In a second part of the study, the authors claim the following:
When looking across all arts organizations and their resident communities, there is no statistically significant relationship between total physical attendance and median household income, nor is there a significant relationship between free attendance and median household income. Since no relationship was found for median income and attendance, the extremes of the income spectrum (poverty and wealth) were analyzed for their relationships to attendance. As shown in Chart 2, there is a positive correlation between attendance at local arts organizations and the percentage of households below the poverty line; as the percentage of households below the poverty line increases, an increase in attendance at local arts organization is observed. We observe a similar positive relationship between attendance and the percentage of households with incomes greater than $200,000. Thus, the evidence indicates that arts organizations serve diverse audiences, with the poor and the wealthy benefitting from the arts more or less equally.
Again, the claim does not hold up. One cannot take figures for aggregate arts attendance, and community income distribution, and make any inference about who are the individuals that attend the arts.
I do not understand the motivation here. There is an international scholarly literature on the predictors of individual participation in the arts, and research on the topic sponsored by the NEA itself. The SMU study is on the wrong track.
Given our plutocratic system of funding the arts, they will always be skewed toward the wealthy.
Even in the wealthier states you mention, the vast majority of the people are either middle or working class. Such generalized demographics say little about the political views of the very wealthy who fund the arts and little about who attends the arts.
Voting Democrat or Republican bears little relationship to public arts funding, since neither party has given it significant support.
Attendance at opera and orchestra performances is skewed toward higher income brackets because the prices for decent seats are so expensive.
Subsidized arts allow for lower ticket prices, and over time create a much wider demographic for participation that crosses class boundaries. This is clearly proven in all European countries.
Arts organizations cater to the wealthy because they fund the arts with their donations. Board members are often selected exactly because they are wealthy donors. Inevitably, an ethos of opulence begins to surround many major arts organizations.
When the arts are funded by the wealthy, arts organizations are concentrated in financial centers where the wealthy live. Other large metro areas where the vast majority of Americans live remain culturally impoverished (cities like Albuquerque, Tulsa, Spokane, Tacoma, Columbus, Tampa, etc..) Public funding systems distribute arts funding more democratically.
Performing arts organizations are located in our cities which often contain the poorest demographics in the country. The arts help revitalize our urban environments.
These problems can only be solved with long term approaches. Only 1 in 4 males in Detroit, for example, has a high school education, and half of those are functionally illiterate. This city’s major arts institutions are funded by a small number of individuals in the city’s wealthy suburbs (which contain some of the richest counties in the USA.) As long as the funding system for the arts reinforces a stark class system (and that is also racially informed,) we will face severe problems with the demographics of arts participation.
Professor Rushton offers quick criticism of our research. We welcome criticism of our research because we all benefit from open, healthy exchanges. However, when criticism misreads or misrepresents our analyses and results, we feel the need to clarify.
Professor Rushton claims that the study commits an ecological fallacy to draw inferences about individual behavior (in our research, individual arts attenders) based on characteristics of groups (in our research, poor and wealthy households). We report correlations between household wealth characteristics and arts attendance.
To illustrate the shortcomings of our research, Professor Rushton offers an analogy with respect to wealth and tendency to vote Republican or Democrat. He offers a link to an online article titled “Rich Man, Poor Man; Rich State, Poor State.” After reading the article, it appears that Professor Rushton does not understand our methodology and/or the findings reported in the article he references.
Professor Rushton compares our analyses to analyses compiled at the state level, which might lead you to conclude that “the rich actually tend to vote Democrat.” He then observes that, in fact, “richer individuals, other things equal, tend to vote Republican.” He fails to note that the article finds that “rich counties tend to vote Republican with greater likelihood than poorer counties.” The takeaway from the results reported in the article is that researchers can expect greater distortion in results the more they aggregate data away from the level of analysis.
Our level of analysis was the relationship between attendance at arts organizations and the household (not individual) wealth characteristics of the surrounding community. To create an appropriate voting analogy for our analyses and results, which are spatially adjusted to account for proximity between the arts organizations and community households, you would have to examine the relationship between voting results at individual polling places and the household wealth characteristics of the surrounding community. One might criticize this approach because there is noise in the household-level measurement; for example, if one person in the household votes Republican and another person votes Democrat. But perhaps even Professor Rushton would accept that this analysis would produce a reasonably accurate estimate of the propensity of wealthier households to vote Republican versus Democrat.
A similar criticism regarding noise could be leveled at our research; specifically, perhaps household size is related to wealth. If, for example, poor households contain more people than wealthy households, wealthy people could be more likely to attend an arts event (as Professor Rushton claims), but attendance at the household level could be equal.
Furthermore, Professor Rushton alludes to the idea of “other things being equal.” Our objective in conducting the current report was to make the results as accessible as possible. Accordingly, we used simple correlation analyses without “controlling” for anything. We point this out as a limitation but note that we have conducted additional analyses where we do control for nearly everything – competitive characteristics, community characteristics, and cultural policy characteristics. The findings, which are available on the NCAR website, suggest that an increasing percentage of wealthy households actually leads to higher contributed revenue but lower levels of total engagement (including attendance) at local arts organizations. The inference is that wealthy households may be more likely to participate with and support local arts organizations, but they also have many conflicting priorities and substitutes, including extensive travel opportunities, that limit their attendance at local arts organizations.
Finally, Professor Rushton illustrates his point by referencing opera. Our data take into account a much more diverse spectrum of arts and cultural organizations, from opera to arts education organizations, community-based organizations, and museums of all types. What is lost from any conversation about the arts that focuses solely on traditional high-art sectors is that it leaves out the great diversity and number of arts and cultural organizations that are not located in wealthier parts of town, do not target wealthy individuals, and attract high levels of attendance, paid and free.
I appreciate Professor Voss’s response to my criticism of the study by her and her colleagues. Unfortunately, her response does not deal with the central problem with the study.
The first paragraph of the report’s “summary of findings” is as follows:
But the study does *not* address the argument. Looking at household income averages (and extremes) in the communities that host arts organizations receiving grants does not allow the researcher to make inferences about whether NEA-funded activities are “generally enjoyed by people of higher-income levels.” The ecological fallacy is not a trivial issue in this case, and it must be addressed seriously. This study does not do that; indeed it doesn’t even hint that there might be an issue.
The authors of the study actually have an interesting data set and could do interesting things with it. By all means let us study the household income characteristics of cities with organizations that receive NEA grants. But this data will not address the argument that arose in the NEA funding debate that the authors claim to refute. It doesn’t hold up.
I requested a correction from the Times of the line suggesting that the study finds that NEA funded events are “just as likely” to be attended by poor as wealthy attendees. This was the response. Curious what you, Michael make of it, and if Zannie finds it accurate.
Dear Mr. Marcus,
Thank you for your email. In response to your question, I would direct you to the boldfaced “Finding 2” in the study at hand. Your question is addressed more fully and with greater nuance in the study text itself, of course, but here is a distillation.
Finding 2: There is no relationship between arts attendance and the median household income of the local community. However, attendance increases with increases in the percentage of households below the poverty line and with increases in the percentage of households with incomes above $200,000.
Regarding our claim that the researchers “found that arts grants led poorer people to attend an event just as much as those in higher tax brackets,” it seems evident from this finding that poorer people are just as likely to attend an event as wealthier people, as based on the statement that “there is no relationship” between arts attendance and median income. The increases in attendance with increased income (above $200,000) are mirrored by increases in attendance with decreased income (below the poverty line), so the effects are neutralized.
We hope this explains, and thanks for taking the time to write. We certainly appreciate your careful reading.
Best,
Louis Lucero II
Assistant to the Senior Editor for Standards
The New York Times
As might be expected, reactions to the study break down along party lines. Right-wingers hate to see their old “elitism” argument thrown into question and liberals like the idea that public funding moves toward equality. The liberal “New Republic” thus has a very positive review of the study here:
http://www.newrepublic.com/article/116482/nea-study-proves-rich-arent-only-ones-enjoying-arts
It quotes a key point from the study: “Free attendance shoots up in communities where it gives a grant, from an average of 16,000 patrons to 40,931, and per capita attendance nearly doubles, from 2.3 percent to 4.4 percent. The study does not argue that the arts are need-blind—anyone who has paid for a ticket at Lincoln Center knows that isn’t true. But it does show that, far from being ‘a wealth transfer from poorer to wealthier citizens,’ federal funding for the arts guarantees that when you erect a theater in a mixed-income community, the seats don’t all go to the one percent.”
In short, public funding lowers ticket prices thus making seats more affordable to a wider demographic.
Another here’s another analogy regarding the general arguemnt: Far fewer children from poor families attend college than those from middle class and upper class groups. So should we eliminate the funding for all state universities because it distributes educational funds upward? That would, by the way, include Indiana University where Mr. Rushton teaches….
The New Rubublic is citing a correlation not a causation. And it is also assuming that poor people are more likely to attend these free events. Lincoln Center does a lot of free events and a lot of rich people attend them. What’s needed here is hard audience data. And it exists. It’s actually generated by the NEA and those numbers look bad. Do all the seats in a newly erected theater go to the 1%, no. Do 40% of the seats for the performance of a musical go to people in households in the top 5%, yes. And that’s not effective use of that money.
This did not become a political football until the 2014 budget proposal, where the GOP realized they had a good argument. The Huffpo piece I linked to, which also contains damning hard data, was from 2010, and huffpo is hardly a right wing mouthpiece.
When per capita attendance increases because tickets are free or less expensive, that is causation.
I have to agree with David: the SMU study is not addressing the same question as that raised by the House Republicans. It isn’t about comparing COMMUNITIES servedor not served by the NEA, but rather WHO THE PATRONS ARE. As Roadside Theatre’s Dudley Cocke wrote in “Arts in Democracy,” “With most (80 percent) of its audience drawn from the top 15 percent of the income scale, the assembled spectators for the typical not-for-profit professional theater production don’t look like any community in the U.S., except, perhaps, a gated one.” Holly Sidford’s research into the distribution of foundation money found the same thing: 55% of foundation money going to the arts went to the richest 2% of arts organization — organizations that serve wealthy, white Americans. Arts leaders have ignored this, preferring to justify the status quo. Now it is time to pay the piper.
I wouldn’t be surprised if counties that received NEA funding were not especially wealthy, since performing arts organizations and large museums are often located in inner cities where the demographic is poor. Unfortunately, the people who attend these institutions tend to come from richer suburban counties where incomes are higher.
All the same, the NEA funding, as paltry as it is, trickles out to the communities where the organizations are located. Arts participants, for example, go to local restaurants, pay parking garages, and support institutions that have local staff, among other things. The presence of the arts institutions usually helps revive otherwise stressed communities.
When people attack the NEA for distributing wealth upward, their arguments could not be more specious. An analogy illustrates the point. The best health care is for the wealthy who can pay for it. Many can’t even afford health insurance and thus have little health care at all. So maybe we should stop funding medical research since it disproportionately benefits those with higher incomes. Of course, everyone needs medical care. And everyone needs to live in a society with a healthy cultural life.
This is great breakdown Mr. Rushkin, and a worthwhile exchange with Ms. Voss. Below is my piece on the issue in today’s issue of The Federalist. As I point out in the article the ultimate issue is what is going to be best for American art. But in the mean time, whatever the actually distribution of arts attendees is, we need to stop funding shows that working people can not afford. http://thefederalist.com/2014/02/07/yes-working-families-are-paying-for-the-1-percents-caviar-art-dreams/
Some of this debate reminds me of the Sidford study from a couple of years ago, which suggested that arts grants in general were directed toward organizations serving the wealthy, therefore the grants were targeted to the wealthy. But on closer examination, it was noticed that the study did not account for the widespread phenomenon of upper-class-serving organizations receiving grants specifically for the purpose of serving disadvantaged populations. Which suggests that to identify the relative wealth of arts clients is a chore that is much more challenging than a simple demographic macro-analysis. You’d have to look at every grant the NEA gives and exactly who benefits from it.
In the end, you’d probably find that many recipient populations were not well off while others were. The “high price of the arts” theory, as the previous commenter alleges, makes little sense when one considers that many museums are free but have primarily upper-class constituents. But for every 100 “rich people” partaking of the visual art prior to their evening at the wine bar, there are a few (2? 5? 10?) poor or nearly-poor people who are attending the museum, drawing inspiration from the art, perhaps moving up the socioeconomic ladder ultimately by being pulled by the association with art and artists. This is a phenomenon that many discuss but that is probably more identifiable anecdotally than has been quantified.
In the end, leaving snide comments like “1% caviar dreams” aside, one has to ask if the purpose of government is to avoid benefiting people with middle and upper incomes. If so, then a whole host of tax breaks need to be eliminated immediately. Crop and flood insurance should no longer be subsidized by the government for wealthy landowners or for those with beach houses on Cape Cod. Businesses that have become successful should no longer be eligible for any government incentives. Do we really want to go there? I don’t.
The NEA’s funding is 0.0036% of the 2013 US budget – 36 thousands of one percent. Or as a simple fraction, it is less than 1/27,000th of the budget. Not much is going up to the rich, and even less to our communities. It’s almost grotesque to give such an infinitesimal sum any sort of serious analytical discussion.
NEA funding is only one part of federal tax money to the arts, tax expenditures through exemptions for individual, corporate and foundation giving brings the total federal dollar amount up significantly. And many of those donors have an even worse record than the NEA in this regard as shown below. If Artistic Directors of major NFPs are going to pay themselves hundreds of thousands of dollars and charge exorbitant prices for their shows, working families should not be paying for it. http://www.huffingtonpost.com/2011/10/10/arts-funding-report_n_1003065.html
If we’re concerned about Federal funds going to the rich we might consider that the 900 billion spent on the bank bailout would fund the NEA at is current 140 million budget for the next 6428 years. (Divide 900 billion by 140 million.)
Below are some examples of other high salaries for arts executives in 2009, as reported in the New York Times. The sums are probably a good bit higher now. See:
http://www.nytimes.com/2010/04/26/arts/26comp.html?_r=1&pagewanted=all
* Peter Gelb at the Met makes 1.3 million.
* Reynold Levy’s annual compensation to run Lincoln Center tops $1 million.
* Carnegie Hall pays Clive Gillinson more than $800,000.
* Glenn D. Lowry, director of the Museum of Modern Art, earned $2.7 million in the year that ended in June 2008, including several one-time bonuses and the cost of his apartment in the tower beside the museum.
* Occasionally institutions will also pay bonuses tied to performance or longevity, like the $3.25 million given in 2006 to Philippe de Montebello to recognize his 30-year service to the Metropolitan Museum of Art.
* On top of his $940,000 salary, Michael Kaiser of the Kennedy Center earned a $150,000 bonus, as well as other benefits, for 2009.
* Zarin Mehta’s most recent compensation, for fiscal year 2010, is $807,500. In the fiscal year ending in August 2008 he earned 2.67 million. This reflected his salary in addition to eight years of accumulated deferred compensation.
* Timothy Rub, the director of the Philadelphia Museum of Art earns $450,00.
* George Steel, the general manager and artistic director of New York City Opera receives $360,000 – and from an opera house that just shut down its next season due to a lack of funds.
*Deborah Borda, Executive Director of the LA Phil makes $799,970 per year plus bonuses that raise her income to 1.5 million dollars.
These high US salaries are a reflection of America’s system of funding the arts by the wealthy for the wealthy. In Europe, where the arts are funded almost exclusively by governments, these sorts of salaries are very rare. The governments would find them unaccountable.
The vast majority of executives of nonprofit arts organizations in the U.S. make salaries well below $100,000. The handful of stratospheric paychecks cited here is completely nonrepresentative of the conditions under which arts managers work. It is emphatically not a rich person’s field. Calculating the tax deductions taken by donors to the arts is fair game in terms of public support, but that money is by definition not a transfer of wealth from poor to rich, since poor people don’t itemize on tax returns and therefore can’t take a deduction.
“The vast majority of executives of nonprofit arts organizations in the U.S. make salaries well below $100,000. The handful of stratospheric paychecks cited here is completely nonrepresentative of the conditions under which arts managers work.”
Exactly. A funding system by the wealthy concentrates funding for delux institutions in a few financial centers where the wealthy live. For example, salaries for musicians in a few top orchestras are between 130-150k per yer, while the average salary for musicians in regional orchestras is only 13k per year.
The USA only has 3 cities in the top 100 for opera performances per year. The budget of the Met is over 300 million per year for a seven month season, while for its European counterparts of equal quality it is 150 million per year for 11 month seasons. The Met treats itself to the most famous stars and lavish sets while opera is virtually non-existent in the rest of the country. In short, the wealthy treat themselves luxuriously while the rest of the country is neglected.
There are a multitude of explanations that could possibly account for the findings in the SMU study. The major flaw of the study, then, is that it presents the link between the findings and their implications unambiguously, when really there could be a number of different factors driving the conclusions.
Let’s take the first finding: “NEA grants do not favor arts organizations in wealthier communities; instead,
funding is more often awarded to economically diverse communities with a higher percentage
of households that are wealthy and a higher percentage of households that are below the
poverty line.” The term “economically diverse” might give readers a warm and fuzzy feeling that NEA grants are going to communities where the distribution of wealth is more equal. Alternatively, readers may interpret the finding that NEA grants are going to organizations in communities with larger income gaps between the rich and poor. The point is that the finding can mean a lot of things, and it’s inaccurate to suggest that the finding means only one thing without even bringing up all of the other possible explanations.
The study also fails to define its terms, making the findings even more ambiguous. What is a “community”? How was it defined? There are certainly differences between a neighborhood, a city, a county, etc. and plenty of researchers have spent countless hours (me included) trying to understand the differences. Perhaps if the study used an alternative definition its finding would change?
Furthermore, there are a series of limitations with the study’s data that could seriously affect the results. CDP collects self-reported attendance data. In other words, organizational representatives fill out forms indicating how many people attended their organization. Obviously, there’s an incentive to over-report who’s coming. And anyone who’s worked in an arts organization, or is familiar with the sector knows that tracking attendance is not the sector’s strong-suit. Also, does the data include repeat attendees? I’m sure it varies from one organization to the next, but again, overall, most attendance data doesn’t take into account whether the attendees are double (or triple, or quadruple) counted. That being said, is it fair to report attendance data and come to conclusions about “community” attendance when we don’t even know who’s included? I don’t think so. Speaking as someone who used to work at the NEA and was privy to reading grantee reports, which indicated attendance figure, I can confirm that attendance data overall is not to be trusted. There’s a clear incentive for organizations to over-report how many people they’re serving.
I could go on to point out other flaws in the study, but I won’t because the point is clear — while the study contributes to our understanding about the distribution of NEA grants in geographic areas, it does not refute the claim that NEA funding disproportionately caters to the wealthy. The study presents a series of summary statistics that can be used to formulate hypotheses that can then be tested using more robust methods. Unfortunately, the media tend to turn to what’s easy to report and thus impede the public’s understanding of how arts polices affect our communities.
The irony is that a private funding system tends to create institutions whose ticket prices are expensive. When these organization receive small amounts of public funding, it does not allow them to lower ticket prices, so the public funds aid the wealthy who can afford expensive tickets. Those who work to limit public funding thus create self-fulfilling prophecies.
If public funding in the USA ever increases to the levels found in all other developed countries it will lower ticket prices and gradually create a wider demographic for arts participation.
BTW, there’s no reason to assume the study might “give readers a warm and fuzzy feeling that NEA grants are going to communities where the distribution of wealth is more equal.” It clearly says just the opposite, that grants often go to communities with higher than average dichotomies — communities with higher proportions of both low and high incomes. That is characteristic of the inner-city areas where many major arts institutions are based.
It may be that only the rich go to arts events in those communities, but over time larger grants would allow for lower ticket prices and better outreach programs that would widen the demographic. The study documents where the grants are landing geographically. And it sites examples of how the grants increased attendance — by whoever. For those interested in improving the catastrophic condition of our urban environments, the study is useful. For those opposed to arts funding, it will merely be a thorn in the side.
I would have second Joanna’s call for greater detail in the unit of analysis. Anyone who is familiar with the complexities of spatial analysis will tell you that how you define “community” will have a dramatic affect on the data resources you are able to pull-in and, consequently, your effect sizes. The report makes a quick reference to geographies defined by the NCAR “spatial model”
http://mcs.smu.edu/artsresearch/questiondocsadditional/modeling-arts-culture-ecosystem-0
A quick trip to that link tells me that the spatial boundaries are based on zip codes. Unfortunately, there is no the reference years for the zip code files (which matters because zip code boundaries change often) and information given as to the method for normalizing data within zip code polygons. Apart from these technical issues, I would strongly urge SMU to explain to the average reader that there can be radical variability in land areas covered by zip code polygons. As an example, Tonopah, Nev., has the largest ZIP code in the lower 48 states. It is basically the size of my home state of Maryland. At the other extreme, there are zip codes in NYC that are the size of two city blocks!
I want to make it clear that I appreciate the work that the authors put not this paper. As a matter of fact, it looks to me that the lead author is a student (bravo SMU). That said, much more could be done to provide readers with the basic information necessary to assess the methods that were employed and, by extentionk, to substantiate very strong claims that are made.
Thank you very much, Carlos, for raising the question of a need for more information on our spatial modeling. As you suggest, larger geographic areas lead to less precision in spatial estimates, but relatively little arts activity occurs in these locations. Our analyses focus primarily on high-arts-activity, high-spatial-density zip codes and census tracts. And the spatial model uses centroid latitudes and longitudes to calculate distances.
When data are available at the census tract level (which is true for most socio-demographic data such as population, income levels, racial characteristics, etc.), we use the census tract centroid. When data are available at the zip code level (which is true for most business data such as number of businesses, business revenues, etc.), we use the zip code tabulation area centroid as established by the census bureau (https://www.census.gov/geo/reference/zctas.html). We always seek to incorporate data at the lowest level of aggregation because it provides the greatest locational precision.
For example, business data are available at the county and zip code level (http://www.census.gov/econ/cbp/). We incorporate the zip code data because the approximately 40,000 zip codes provide greater locational precision than the 3,000+ counties. Likewise, we incorporate socio-demographic data for the 70,000+ census tracts rather than the 3,000+ counties to achieve greater locational precision.
As the number of zip codes and census tracts suggest, census tracts are nearly twice as precise as zip codes. The average (i.e., arithmetic mean) land area for a census tract is a little over 100 square kilometers or 10 kilometers by 10 kilometers. Assuming normal census tract boundaries, the centroids for two adjacent census tracts are roughly 10 kilometers away. As described on our website, the formula for the spatial model would discount the characteristics of a census tract 10 kilometers away by 50%. That means that half of the population and half of the competition of the adjacent census tract would be incorporated into the trade area for the focal census tract, and vice-versa.
However, this analysis is skewed by the very large geographic size of remote census tracts that feature few people and little arts activity. For example, the median-sized census tract is less than 5 square kilometers or roughly 2.25 kilometers by 2.25 kilometers. When the centroids for two adjacent median census tracts are 2.25 kilometers away, the spatial model discounts the characteristics of the adjacent census tract by only 5%, so that 95% of the population and competition of the adjacent census tract would be incorporated.
Zip codes work in the same fashion but are less precise, given the smaller number. Irregular boundaries may also reduce precision but only slightly because centroids tend to smooth out the boundary irregularities. We encourage everyone to read the section on our website related to Building a Spatial Model.
http://mcs.smu.edu/artsresearch/questiondocsadditional/modeling-arts-culture-ecosystem-0
We also encourage you to view the spatial model video: http://mcs.smu.edu/artsresearch/about/building-spatial-model .
Thanks, Zannie this is helpful, although it raises other questions. From what you wrote above, the analysis was not national in scope but rather skewed toward areas with higher land area density? Is that correct? That is what I gather when you say that your: “analyses focus primarily on high-arts-activity, high-spatial-density zip codes and census tracts”.
Can you tell us if this this segmentation is an intentional part of the analysis, with a particular cut-off or are you saying that this is a de facto consequence of the spatial distribution of the organizations in your study? I just hadn’t seen anything in either report qualifying the spatial analysis in this way before.
Also, can you clarify what you mean by “high-arts-activity” ? Is this a reference to frequency of arts opportunities and performances or is this a reference to the status honor associated with the art form? Either way, it would be good to know how this distinction was operationalized and what impact that may have had on the findings.
-Carlos
Carlos, please let me know if this provides sufficient clarification. We study arts organizations where they exist. We attempt to include every arts organization in the US. But there are parts of the US where virtually no one lives and formal arts activity is not recorded. Consider, for example, that in the 20 largest county tracts in the US, the average number of people per square mile is much less than 1. And no formal arts activity is recorded. So when we say that our analyses focus primarily on high-arts-activity, high-spatial-density locations, it is not because we intentionally focus on these areas, it is because that is where formal arts activity is recorded. Please let me know if you have other questions or if I haven’t been clear. Thanks again, Zannie
Definitions are important. The focus on “formal arts organizations” masks a continuous and ongoing bias against rural areas “where virtually nobody lives.”