adwel81
commented
2 years ago I agree with revisiting thresholds for higher ed enrollment, and whether educational enrollment should be such a significant piece of the scoring process (i.e., a screening criteria for whether other indicators pass a threshold.
It looks like current enrollment is correlated with the age distribution, and may be inadvertently filtering out disadvantaged communities near state and community colleges and universities Based on New York State data: Areas with older populations tend to have lower current enrollment. And, it looks like there may be an inverse relationship between current enrollment and attainment. On average, areas with lower current enrollment have higher high school attainment (though not college).
This metric may also filter out high-need areas near colleges/universities with lots of students who are not permanent residents. Many of these neighborhoods may be high-BIPOC or high-Limited English Proficiency communities, which I understand is not a criteria, but many stakeholders will want to see those areas considered.
Below is analysis of all census tracts in New York state. The first table is by Age quintile (using % over Age 65). The second is by Higher Ed Enrollment quintile (from the CEJST data) For NY it looks like low educational enrollment on its own is not a clear (or linear) measure of disadvantage - It seems to reflect the age distribution.
The first map shows tract 50007000400 in the Old North End of Burlington Vermont. This area has low income, many long-term renters, poor housing quality, high linguistic isolation (thriving immigrant community), and relatively poor health outcomes. However it fails to pass any of the thresholds because of high educational enrollment, due to high numbers of student-renters attending local colleges. Rents are high due to students, landlords don't have to invest much in building or energy improvements, therefore housing quality and energy insecurity are low.
The second and third maps are tract 36067003500 in Syracuse NY. Not a DAC per CEJST but a DAC per NY definition. High poverty and unemployment, high rent and energy burden, near highway and industrial facilities, high rate of disabilities and premature deaths...But younger population & doesn't pass higher ed enrollment filter.
I would encourage reviewers to (a) look at other lower-income student-heavy neighborhoods (esp in older cities with community and state colleges), and (b) look at these data relationships in other states, to see how pervasive this is.
I'm guessing (speculating!) that perhaps low enrollment in higher ed was identified as a potential correlate or proxy for race/ethnicity since the tool is not allowed to consider race/ethnicity, and perhaps this generally works, with exceptions like places around colleges and universities. If the goal is to measure historical discrimination or systemic factors that led to lower savings/wealth, educational, employment or health outcomes, or social/political capital, I'm curious if there are any other proxies from the Census (e.g. single-parent households?) or federal program data (e.g., social assistance like SNAP, TANF, SSI?), or anything like voter participation or census non-response rates (...as a potential proxies for social capital?) that could serve a similar purpose? As an interim step, perhaps decreasing the higher ed enrollment threshold and/or considering proximity to colleges/universities could help?
emma-nechamkin
Our goal is to modify income in tracts that have under some high threshold of students (80% is my starting number!) so they can continue to be included.
May require refinement going forward.