Abstract
Digital redlining is the modern day continuation and perpetuation of historic, institutionalized racism. In this experiment, ten locations within the City of Cleveland were selected, with two locations for each of five income groups. Two internet speed tests for download speed and upload speed were conducted in the ten locations over three trials for both Verizon and AT&T. A linear regression analysis was conducted to measure the association between median income and internet speed. Verizon download, AT&T download, and Verizon upload speeds all trended downward as median income increased. A statistically significant correlation of (p<.05) was found for these three out of four measures. These findings were surprising given the hypothesis was the opposite of these based on the expectation that more impoverished areas would have slower service. Further investigation should collect data of the speed of wifi provided internet resources rather than cellular data speed and at a greater number of areas over a variety of times of day.
Introduction
Redlining was a racist practice which prevented black families from purchasing property in white neighborhoods and lending to invest in black neighborhoods (Doyle 2021). Digital redlining is a modern practice that perpetuates racial inequities by providing high quality internet and cellular service exclusively to wealthier neighborhoods. Its effects have become particularly evident since the COVID-19 pandemic, when students were required to have a reliable internet connection to attend school (McCall et al. 2022). Digital redlining has had a profound effect throughout the evolution and development of the internet. Past studies conducted in rural environments show that limited internet access is not exclusively connected to income, but also race, which was seen when internet access increased with an influx of white families in the area (Friedline, Naraharisetti, and Weaver 2019). However, information regarding this crucial topic is still limited, and the topic isn’t widely discussed. Thus, this study was conducted in an attempt to rectify this area of insufficient knowledge. The experiment sought to explore how cellular data quality correlates to the income and racial makeup of various neighborhoods around the Cleveland area. Data was collected from ten different locations within Cleveland, and the cellular data speed was recorded at each location. These data were then compared to the median income as a proxy for the racial makeup of the given neighborhood.
Methods
The experiment involved traveling to ten different neighborhoods within the City of Cleveland. Using data from Statistical Atlas (The Demographic Statistical Atlas of the United States 2018.), two neighborhoods were randomly selected in each of the following income categories: high income ($45K-$54K), high mid income ($37K-$45K), mid income ($29K-$37K), low mid income ($22K-$29K), and low income area ($13K-$22K). The two lower income zones align with historically redlined areas (McEwen 2018). The internet speed of the cellular data was taken six times at each of the locations using Google’s internet speed test. The trials were performed for both Verizon and AT&T. The designated neighborhoods were as follows, in decreasing order of median income: Riverside, Kam’s Corner, Downtown, Jefferson, Lee-Miles, West Boulevard, Buckeye-Shaker, University District, Kinsman, Central. The cellular data was measured with two internet providers to represent a broader range of user experiences. The internet speed test returned two numbers measured in megabits per second (Mbps): download speed (Mbps) and upload speed (Mbps). Download and upload speed measure the rate at which information is downloaded and uploaded from the internet, respectively.
Linear regression tests were completed for all four parts of the data: AT&T upload speed, download speed, as well as Verizon upload and download speed. Regression was performed to discern the connection between median income and internet speed.
Results
As median income increased, cellular data speeds trended down (Figures 1-4). AT&T download values ranged from 53.6Mbps to 523.6Mbps (Fig. 1). AT&T upload values ranged from .8Mbps to 41.3Mbps (Fig. 2). Verizon download values ranged from 2.7Mbps to 168.9Mbps (Fig. 3). Verizon upload values ranged from .2Mbps to 28.4Mbps (Fig. 4). Verizon download, AT&T download, and Verizon upload speeds all trended down as median income increased. Linear regression testing between median income and internet speeds for both Verizon and AT&T returned significant results for these three out of the four data sets. AT&T download speeds vs median income had a p-value of 0.012., Verizon download speeds vs median income had a p-value of 0.003, Verizon upload speeds vs median income had a p-value of 0.005, but AT&T upload speeds vs median income returned a p-value of 0 .368. The r2 value for AT&T download speeds vs median income was 0.183, for Verizon download speeds vs median income was 0.487, for AT&T upload speeds vs median income was 0.005, and for Verizon upload speeds vs median income was 0.45.
Figure 1: Median Income vs. Verizon Download Speed
Figure 2: Median Income vs. Verizon Upload Speed
Figure 3: Median Income vs. AT&T Upload Speed
Figure 4: Median Income vs. AT&T Download Speed
Discussion
The study found that internet speed decreases as the median income of neighborhoods in Cleveland increases. These findings are surprising, as the hypothesis was the inverse based on expectations that poorer communities would have worse service. Three out of four data sets indicated this, including the download and upload speeds from a Verizon phone and the download speeds from an AT&T phone. Though for AT&T this effect is noticeably reduced, and includes one of the four data sets which doesn’t indicate a convincing connection between income and internet speed. These data indicate that the cellular data speed reduces as the income increases. The hypothesis was incorrect, and the opposite effect was observed. It's difficult to know why this is the case. The reviewed research indicates that internet speed would increase with median income, but the opposite was observed. Some potential reasons for this include chance, such as the areas with lower income were closer to cellular towers. Another possibility is that population density, which is generally higher in lower income areas, is connected to cellular data speed. Another limitation that may have caused this fault is that we only took data for each income value from two geographic areas, and at one time of day. If more samples were taken in different geographic regions at different times, the chance of coincidence would be greatly reduced. Future studies should investigate wifi internet speed rather than cellular internet speed. Cellular data likely fluctuates due to a wide range of unpredictable variables, while wifi provided internet speed is likely more predictable and more indicative of true internet access disparities.
Bibliography
Madrigal, Alexis C. “The Racist Housing Policy That Made Your Neighborhood.” The Atlantic. The Emerson Collective, May 2014. https://www.theatlantic.com/business/archive/2014/05/the-racist-housing-policy-that-made-your-neighborhood/371439/
McCall, Terika, Kammarauche Asuzu, Carol R. Oladele, Tiffany I. Leung, and Karen H. Wang. “A Socio-Ecological Approach to Addressing Digital Redlining in the United States: A Call to Action for Health Equity.” Frontiers in Digital Health 4 (2022). https://www.frontiersin.org/journals/digital-health/articles/10.3389/fdgth.2022.897250.
Friedline, Terri, Sruthi Naraharisetti, and Addie Weaver. "Digital Redlining: Poor Rural Communities’ Access to Fintech and Implications for Financial Inclusion." Journal of Poverty 24, no. 5-6 (2019): 517-541. https://doi.org/10.1080/10875549.2019.1695162 .
“Household Income in Cleveland, Ohio.” Statistical Atlas. The Demographic Statistical Atlas of the United States. https://statisticalatlas.com/place/Ohio/Cleveland/Household-Income
McEwen, Colin. “REDLINING’S LEGACY: Crime and poverty higher in Cleveland neighborhoods that faced lending restrictions decades ago.” THINK Magazine. Case Western Reserve University. https://case.edu/think/fall2018/redlining.html