Background and Question
One of the first things I noticed about living in Flatbush was how infrequently the B41 bus came and how frequently unmarked white vans came. I’d often wait over 15 minutes for the bus even though the B41 is the primary route for accessing Downtown Brooklyn, as well as critical connections to other parts of Brooklyn and Midtown Manhattan. When the bus did come, it inched slowly down Flatbush Avenue, especially during rush hour. It eventually felt impossible to depend on. Such has been noted by MTA Chair and CEO Janno Lieber, who, along with Mayor Eric Adams, have plans to improve transit service for various New York communities. Flatbush has been labeled a Bus Priority Project because of its criticality as a cross-borough transportation corridor, its high number of average daily bus riders, slow bus speeds, and the number of people killed or seriously injured on Flatbush Ave between 2016-2021 (Flatbush Avenue Bus Priority Plan).
I came to find out that much more frequent, unmarked white vans were “dollar vans” that shuffled folks up and down Flatbush Ave, letting you off wherever you needed to go until the Barclays Center. When I worked at the Navy Yard, I ended up taking these vans right past Grand Army Plaza, grabbing a coffee, and walking the remaining 30 minutes to work. To get home, I’d take the Yard’s shuttle to Barclay’s Center and another dollar van back home. As journalist Aaron Reiss noted in his article mapping and detailing this system, “dollar vans and other unofficial shuttles make up a thriving shadow transportation system that operates where subways and buses don’t—mostly in peripheral, low-income neighborhoods that contain large immigrant communities and lack robust public transit” (Reiss, “New York’s Shadow Transit”).
It’s this last point that I’d like to explore in this report. Are commute times to work in Flatbush longer than those of other Brooklyn neighborhoods? Where are the longest commute times in Brooklyn? What is the median household income of these communities? Is there a relationship between median household income and commute time to work?
Materials
For this project, I used Microsoft Excel for data cleaning, Social Explorer for data collection and mapmaking, and Datawrapper to make a chart. I used data from the 2019 American Community Survey 5-Year Estimates. I used “Table A09003: Average Commute to Work (In Min)”; “Table A09002B: Travel Time to Work for Workers 16 Years and Over 45 Minutes (In 15 Min Intervals) – Cumulative (More);” and “Table A14008: Average Household Income (In 2019 Inflation-Adjusted Dollars).”
Methodology
Data exploration and cleanup were the bulk of this project. The 2019 ACS data described above provided information for all of New York City. I filtered the data to only show values of Brooklyn and chose to visualize census-tract data because it was granular enough to allow neighborhood-specific exploration.
The only other option was visualization by county, and since all of Brooklyn falls under Kings County, this view would not have been useful for my visualizations. During this phase, I calculated that the average commute time in Brooklyn was 45 minutes and chose to use this figure as a point of comparison.
Because there are so many census tracts in Brooklyn, I was worried that my choropleth maps would look too crowded, since each tract would represent a data value. To help make my maps more legible, I chose only 6 classes for the cutpoints and grouped the data by quantiles, which resulted in maps that showed sufficient variation amongst the various data points.
My choice of color scale depended on the emphasized variable. For “Average Commute to Work (In Min),” for example, I chose a sequential scale from light yellow to dark orange, with the darker hues signifying longer commute times. “Average Household Income” uses a green sequential scale where darker hues represent a higher median income. Orange’s cultural association with negativity and green’s cultural association with money proved useful here. For my map showing the percentage of workers who commuted over 45 minutes to work, I used a diverging color scale from dark blue to dark red. The diverging scale allows visualization using 50% as a midpoint – the redder a tract appears, the larger the percentage of workers commuting over 45 minutes. The bluer the tract, on the other hand, the smaller the percentage.
I then decided to place these maps side by side for easy comparison of commute times and income across Brooklyn as a whole and between different Brooklyn census tracts. The green line outlines zip code 11226, an area of Flatbush.
For my chart, I chose split bars because I was showing two different types of data in different units – average household income in USD and average commute time in minutes by census tract. I visualized the data of the five highest and lowest earning Brooklyn tracts to not overwhelm the reader. There are essentially two bar charts – average household income by census tract and average commute time by census tract. These are then grouped by where they rank relative to the average commute time in Brooklyn. Grouping the bars allows the reader to see variations within the data and how different tracts relate to each other. I used green bars for household income and yellow bars for commute time as fairly neutral colors. Even though it may not be clear to everyone where these census tracts are, I chose this labeling because zip codes and neighborhood names often contain multiple census tracts.
Results, Interpretation, and Visualizations
I chose to create three side-by-side maps to compare all of the variables I was interested to one another. Those variables are: average commute time in minutes, average household income, and percent of workers commuting over 45 minutes to work. The maps appear to show a relationship between income and commute time, but statistical analysis would be needed to explore this topic further. As mentioned previously, I aimed to reveal both overall patterns in Brooklyn and more granular community-level patterns particularly through the maps visualizing the percentage of workers who commute for longer than 45 minutes.
The Datawrapper chart visualizes data from ten census tracts – the five highest- and lowest- earning. The data is grouped by whether the census tract’s commute time is above or below the borough average.
Reflection and Future Work
Comparing volume when variables have different units of measurement and thus different scales presented a problem while building my chart in Datawrapper. I also wanted to visualize 3 variables – community, commute time, and income. I look forward to learning more about how to better approach this task. Because Datawrapper only allows a certain amount of editing and customization, I feel the bars showing household income do not communicate much meaningful information on their own since their size is arbitrary, though as a whole I do think the chart appears to show patterns between commute time and income.
Future work could include incorporating transportation methods and availability and quality of public transportation.