Tierrabyte: 2020 Census May Find San Antonio ‘Hard to Count’

Print Share on LinkedIn Comments More
(From left) Juana Arellano and Casandra de Leon hold one another while listening to speakers protest the Trump administration's decision to end DACA.

Bonnie Arbittier / Rivard Report

Protestors in San Antonio hold one another while listening to speeches opposed Trump administration's decision to end DACA, September 2017.

San Antonio may be at risk for an undercount in the 2020 decennial census, an analysis by the Rivard Report found. Following last month’s decision by U.S. Secretary of Commerce Wilbur Ross to add a new citizenship question, state and local demographers worry the changes will deter people in areas with large Hispanic populations from responding.

It’s not yet known, to the public anyway, what the final question will be. Some groups, especially immigrant advocates, are concerned that responding to the question could lead to deportation, though it’s unclear if that would be an intended or unintended outcome.

U.S. residents are required to fill out the survey by law, though the bureau has not prosecuted non-responders since 1970. The results of the census are published anonymously to the public, but the census itself is not anonymous: respondents are asked for their name, social security number, and address on the questionnaire.

Every 10 years, new census data is used to determine how congressional districts will be drawn. The data also determines how more than $600 billion in federal funding is allocated to states for over 100 programs like medicaid, food vouchers, head start programs, Pell Grants, and transportation projects, among others.

The 2020 census will also include an option to take the survey online, raising further concerns that citizens who don’t have access to the internet will not participate.

If the changes result in fewer responses, they could threaten the state’s projected gain of three congressional seats in Congress. Texas won an additional four seats due to population gains counted in the 2010 Census.

San Antonio ranks 24th out of 155 metropolitan areas for the largest population of unauthorized immigrants according to a recent Pew Research Center study. With a 62 percent Hispanic population, and nearly 20 percent of residents lacking broadband internet at home, according to American Community Survey estimates, much of San Antonio is at risk to be undercounted.

About 25 percent of Bexar County’s population live in “hard-to-count” neighborhoods according to a recent City University of New York study.

In the visualization above, each circle represents one census tract in Bexar County. We graphed all census tracts by the likelihood of residents to respond to the census, and the percentage of hispanics living in those tracts. You can also filter the graph by each tract’s level of broadband access, to discern additional patterns.

How likely residents in a census tract are to respond to the decennial census is measured by the “low response score” determined by the bureau following the 2010 census. Based on the likelihood that residents will not respond to the survey due to a variety of factors, the bureau found that most of the state’s residents live in areas exceeding the national average for such scores. Earlier this year, the U.S. Census Bureau released a map of its findings.

“It’s feasible that [the citizenship question] will have some impact on hard-to-count populations,” said State Demographer Lloyd Potter, noting that a citizenship question has been on the American Community Survey, administered by the U.S. Census Bureau, since 2005.

Potter said that the introduction of the question at the last minute and the politicization of the 2020 census by the media could skew results.

“We’re basically throwing in a question at the end that hasn’t gone through a rigorous testing and assessment process,” he said. “It just messes with standards that have been pretty well established … I think that’s the thing that I’m most concerned about.”

Josué Romero, 20, a Southwest School of Art student, was given temporary status under the Deferred Action for Childhood Arrivals (DACA) program. After being released by Immigration and Customs Enforcement (ICE) under charges of marijuana possession, Romero has since returned to coursework and is now establishing a career as an artist.

Romero, who will be filling out the census for the first time in 2020, described facing a choice to risk suspicion from the government by answering the citizenship question or to participate, and represent a community needing both resources and recognition.

“I feel comfortable participating,” he said, “because it does more for our representation in the state and the greater community. I think, while this is a bureaucratic process, it is a way to establish my place in an affirmative way, and to challenge those political agendas that are being pushed.

“How do I give myself firm ground to stand on when it’s not given to me?” he asked. “I have to make that for myself.”

The decennial census will be held on April 1, 2020.

4 thoughts on “Tierrabyte: 2020 Census May Find San Antonio ‘Hard to Count’

  1. The article appears to be more a collection of anecdotes masquerading as data-based analysis without context.

    What, according to the CUNY study is a “hard-to-count” neighborhood? The article cites the category without definition of the term. What, according to the Census Bureau, is the “national average”? There’s a line in the graph, but the data in that graph is suspect based on discrepancies from the text.

    CUNY says 73% or less mail-in rate qualifies as “Hard to Count” and does say that of approximately 25% of Bexar County’s population live in neighborhoods that fall below this line. The CUNY study does, however, opt not to map any Census tract with a population below 100. This article omits that context, but includes low-population tracts that add to the statistical noise.

    The Rivard Report text claims the “visualization” includes Census Tracts between 1,200 and 8,000. However the lowest “Low Response Score” (7.4%) and the highest (43.3%) both have populations significantly less than 1,200 (25 and 301, respectively). Including the low-sample “noise” in the visualizations unnecessarily expands the Y-axis and presents “extremes” where they are not necessarily relevant. The inclusion is also inconsistent, since there are other low population tracts not displayed – what is the actual low-end cutoff?

    The graphing also induces irrelevant and anecdotal “correlations” with race and broadband access. Using this data, a tract with high broadband penetration (>80%) has the lowest census response (43.3%) while a tract with low penetration (<20%) has the highest response displayed (7.4%). They don't meet the text's criteria for display, but they are shown in the visualization anyway. The lowest non-zero response rate (6.9%) was 0% Hispanic; followed closely by the second lowest non-zero rate (7.3%) @ 100% Hispanic. These two tracts are not displayed in the visualization, however they are in the map.

    I don't know what any of that really means; which probably means it doesn't mean very much. It's numbers without context or filters, plotted on a map and a graph.

    • Hello, thank you for your comment. I hope I am able to address your thoughts here:

      1. From the CUNY study, here’s their definition of a “hard-to-count” census tract: “For the purpose of this map, a census tract is considered hard-to-count (HTC) if its self-response rate in the 2010 decennial census was 73% or less. If 73% or fewer of the tract’s households that received a census questionnaire mailed it back to the Census Bureau, it is shaded in light orange-to-dark red as a hard-to-count tract on the map.”

      2. We did not use any data or definitions from the CUNY study for our visualization, but included a link to their study for additional context. We mapped all census tracts in Bexar county regardless of their population size. We used Low Response Scores (LRS) for each census tract in Bexar county in the U.S. Census Bureau database (2016) available here: https://www.census.gov/research/data/planning_database/. We obtained the national average for LRS response rates from the same database.

      We used American Community Estimates (2016) for hispanic populations available here: https://factfinder.census.gov/faces/nav/jsf/pages/download_center.xhtml.

      We used Form 477 FCC (2016) data to obtain broadband internet access rates available here: https://www.fcc.gov/general/broadband-deployment-data-fcc-form-477.

      3. Thank you for pointing out the error regarding our visualization. As you pointed out, we erroneously wrote that the census tracts contained in the visual were bounded by a population size of 1,200 to 8,000. You correctly noted that is not the case. The error has since been corrected.

      4. We did not use a cut-off or exclude any census tracts from the analysis. Data points you pointed out are actually contained in both the map and the visualization. They are quite difficult to see however, as the lowest non-zero response rate (6.9) is a tiny point that falls along the zero axis. Each circle’s size is based on the total population size for the census tract that it represents. As the total population size of 4 being very small for this census tract relative to the others it becomes quite difficult to see. However we chose not to arbitrarily inflate this circle’s population size to make it more visible. The other point you mentioned, (7.3% @100% Hispanic) was obstructed by our bexar county map in the top right corner. I’ve since reduced the size of that map so that this point is visible.

      Thank you for pointing out these concerns, thoughtful contributions from our readers help make our journalism better.



    • Can you explain, perhaps, similarities or differences between three large population tracts in your visualization?

      Tract 48029161400 has the largest population shown @ 9,945 (largest circle), broadband between 60-80% penetration, is approx 20% Hispanic. It’s response rate was poor, @ 31%. It’s the “least Hispanic large tract” and it’s on the SW side of town.

      Tract 48029160100 also has a large population @ 5,794 (similar size circle), broadband @ 40-60%, is 99% Hispanic, also a poor rate @ 31%. It’s the “most Hispanic large tract” and it’s near downtown.

      Tract 48029162007 has a large population @ 9,945, broadband @ 80-100%, is 56% Hispanic, but has the best “large tract rate” @ 18.5% and is on the far west side.

      What can or should a reader conclude from looking at the characteristics you’ve selected in your visualizations?

      Also, since you’ve injected race into the data while excluding any racial perspective other than Hispanic – unlike the CUNY study from which you extract the 25% HTC value – what can a reader conclude about a majority Black tract (per CUNY study terminology), like 48029130800 on the East side @ 53% Black, which also has a “Hard-to-Count” rate of 31% per CUNY data?

Leave a Reply

Your email address will not be published. Required fields are marked *