Read first:
The Detroit Art Ecosystem Part I
The Detroit Art Ecosystem Part II: Funders, institutions, and decision makers
In Part II of this report, we used network analysis techniques to identify the individuals, families, and organizations that collectively determine how art and culture resources are allocated within the Detroit art ecosystem. We then identified systematic forms of network bias and used these to determine opportunities for increasing the impact of art and culture funding.
The previous installment identified that the decision-makers directly connecting funding sources to large cultural institutions were almost entirely composed of the city’s political and economic elite. This is an overarching structural bias of the network.
In this installment, we analyze these decision-makers to identify additional bias by race and/or gender. As in Part II, we consider two network configurations: with and without the densely interrelated development and placemaking ecosystem. For each of these cases, we statistically test for bias in representation and influence by race and gender. This comes to a total of eight (two (case1/case2) x two (gender/race) x two (representation/influence)) tests.
Test for bias in Influence
To test for bias in influence, for example, between Males and Females or White and BIPOC decision-makers, we use the T-test to compare the average weighted degree centrality for the two samples. A one-sided T-test is used, explicitly testing whether (i) White decision-makers have more influence than BIPOC decision-makers and (ii) Male decision-makers have more influence than Female decision-makers.
Test for bias in representation
To test for bias in percentage representation, for example, by gender or race, we calculate the p-value to test for a statistical difference between the sample percentage and the estimated population percentage for Metropolitan Detroit (which we consider the tri-county area).
The estimated population percentages for Male/Female, White/BIPOC are determined from the latest census data for Macomb, Oakland, and Wayne Counties.
Based on this, the estimated population percentages for the Tri-County area are as below:
BIPOC = 33.8% | White = 66.2%
Female = 51.2% | Male = 48.8%
Results
The average weighted degree centrality is consistently higher for white decision-makers over black decision-makers and for male decision-makers over female decision-makers, indicating that overall, white male decision-makers have more influence.

The sample population of white decision-makers (as a percentage) is consistently higher than the percentage in the overall population, and the sample population of male decision-makers is consistently higher than the percentage in the overall population, indicating that white males have greater representation in the decision-making process.

Statistical Significance
Testing at a probability level of p<0.05, seven of the eight individual tests are statistically insignificant.

The one statistically significant difference is gender representation in Scenario 1 (where the development and placemaking ecosystem is included.) This might indicate a gender bias where female members of the economic and political elite are more likely to be involved in art and culture, whereas participation in development and placemaking is still biased towards men.
Conclusions
The results show a consistent bias towards white male decision-makers in representation and influence; however, these differences are not statistically significant in most cases. When combined with the conclusions of Part 2 of this study, this may indicate that the network has some capacity to self-regulate in terms of racial and gender bias while remaining consistent in its overall bias that decision-makers almost entirely come from the political and economic elite.
Copyright Essay’d 2025