Board of Elections or Bored of Elections?

Addressing low voter turnout by proposing specific locations for expanded polling site capacity in New York City

Varisa Tanti, Kavyaa Rizal, Daniel Wexler, and Carlos Miranda · Columbia University

INTRODUCTION

Could New York City's Board of Elections do more to improve voter turnout, or are New Yorkers simply bored of elections? In the 2021 election season, only 23% of registered voters  cast ballots  in New York City’s general election, the lowest turnout rates in the city’s history. While numerous variables may have contributed to the historic lows–such as the newly implemented ranked choice voting or skepticism over effective public governance–New York City’s election efforts must remain focused on improving the electoral process through pro-voter reforms.

The New York City Campaign Finance Board has addressed some causes of low voter participation by emphasizing paid time off to vote, extending early and absentee voting opportunities, and allowing online voter registration. These demonstrate the city's interest in encouraging voter participation and are evidence of a marked effort to improve voting conditions for individuals; however, continued low voter turnout make evident the need for continued improvements to the electoral process, with regard to voter engagement and voting infrastructure.

This project seeks to understand, and address, low turnout rates through a spatial dimension. Specifically, this research aims to deploy a spatial methodology which identifies and proposes new polling site locations, by considering New York City's existing election infrastructure. An analysis that considers proximity to and accessibility of polling sites, as well as different poll site worker capacities, can reveal where the Department of Finance should open new polling sites to best serve New Yorkers who have an interest in exercising their right to vote.

METHODOLOGY

Conducting a Multi-Criteria Decision Analysis (MCDA) was the most appropriate approach, given this research's interest in identifying potential new polling site locations. MCDA is a method which can account for the plethora of different variables which factor into the site-selection process. It does this by weighting different selection criteria within the analysis. The result is a collection of sites which are assigned numeric scores based on how well the location meets the set of specifications.

A set of three conceptual categories, all related to voter turnout, served as the impetus for this project's MCDA: poll site accessibility, voter behavior, and poll site capacity. These were selected under the pretense that these three factors have the potential to affect voter turnout and were feasibly measurable, given the resources and timeline within the scope of this project.

  • Poll site accessibility was operationalized using service area network analysis, which incorporated a walkable distance of 0.25 miles from the poll site.
  • Voter behavior was operationalized using emerging hotspot analysis, which allows for a spatial-temporal measurement of participation patterns over time.
  • Poll site capacity was operationalized by calculating normalized voter-per-worker ratio for each polling site.

The logic, assumptions, and procedures associated with these three metrics will be expanded upon in subsequent sections. In addition to these three metrics, land use data was included within the MCDA to account poll site designations, which are based on permanent residential addresses. Lastly, only sites that could realistically provide physical space for a polling site, such as vacant storefronts, public schools, and places of worship, were considered in the MCDA.

01 NETWORK COVERAGE

Creating a network

A quarter-mile walking service area was generated per polling site. The New York City Campaign Finance Board assigns individuals to polling sites based on their election district, which is determined by their permanent residential address. Thus, it is reasonable to assume that polling sites should be within walking distance for all voters. In reality, however, individuals within a given election district may be unevenly served by their polling site, which this method aims to address, visualize, and quantify.

Calculating areas beyond the network

The polling site service area was overlaid atop the election district boundary, making it possible to calculate the percentage of the election district that is beyond the quarter-mile walking service area for the polling site. Conducting this analysis for every polling site and election district allows for the accessibility of polling sites' to be characterized, quantified, standardized.

Of note, there are boundless ways to conceptualize and measure accessibility. This operationalization simply represents one physical dimension of accessibility.

Assigning coverage for each election district

The percentage of the election district outside of the poll site's service area was incorporated into the weighting for the MCDA, with priority given to election districts with a higher percentage of area beyond the quarter-mile service area. In other words, if a large percentage of an election district was not within a quarter-mile walking distance to the polling site, it was given preference and was considered more favorably for a new polling site than other election districts.

Figure 1

Figure 1 provides a graphical representation of key findings from the network analysis. The results indicate, for all five boroughs, population density has an inverse relationship with polling site service coverage. This means, generally, areas with higher population densities are also seeing higher numbers of polling sites.

This map illustrates the population density and polling site service area coverage per election district. The varying degrees of coverage provided by the polling site service areas in relation to the corresponding population density within those districts underscores that there still remains areas that have high population densities that are not being serviced with enough sites. It is evident that a majority of the districts (depicted in light green) exhibit a high level of network coverage that aligns with high levels of population density. However, there are certain districts (depicted in dark green) characterized by high population density but inadequate network coverage. These election districts harbor a larger voting population, but the areas beyond reasonable distance to polling sites are relatively unaddressed.

02 EMERGING HOTSPOT ANALYSIS

Hotspot analysis per election

The first step in conducting the hotspot analysis was aggregating voter data in NYC between 2008-2018 by census tract and election year. This data was further aggregated into a hexagon mesh layer in order to standardize geographic boundaries given changes the in census tract delineations within the studied decade.

A Getis Ord Gi* (hot spot) analysis was then conducted for each year, providing clusters of high voter participation (hot spots) and low voter participation (cold spots) for each election type, as seen on the right.

Space-Time Cube & Emerging Hotspot Analysis

To more effectively capture voter turnout across all 11 election years, turnout rates and behavior patterns were used in the Space-Time Cube. Cube bins were the geographic hexagons and each time slice was an election year. An Emerging Hot Spot Analysis was conducted to visualize the Space-Time Cube as a 2D representation (below).

The Emerging Hotspot Analysis highlights trends of high value and low value clusters over time. Here, an Intensifying Cold Spot indicates a cluster that has been a statistically significant cold spot for 9 of the last 10 years, including 2018, the final year considered, and that the intensity of clustering of low counts in each year is increasing overall. A Persistent Cold Spot indicates a cluster that has been a cold spot for 9 years. A Consecutive Cold Spot indicates a cluster with a single uninterrupted run of at least two years. A Sporadic Cold Spot indicates a cold spot for the final year, with a history of also being an on-again and off-again cold spot and a Diminishing Cold Spot a cluster that has been a cold spot for 9 years including the year, while the intensity of clustering of low counts in each year is decreasing overall. The cold spots explained above are most prioritized in the upcoming MCDA.

03 ADDITIONAL PARAMETERS

Land use: Because polling sites are assigned to individuals based on proximity from their permanent addresses, only residential areas were considered for the new polling sites. Commercial and industrial areas are not appropriate for new polling sites since voters are location-bound to areas near their homes.

Worker capacity: In order to approximate poll site capacity assigned by the city, the number of workers allocated to each polling site was sourced from the New York City Board of Elections. Then, the number of eligible voters per poll site worker was used to generate a standardized voters-per-worker ratio for each election district. Wait times at polling locations pose a potential hindrance for voters; thus, polling sites which are understaffed, relative to the voter demand, can be flagged as opportunity sites for improved voter access. Consequently, within the MCDA, areas with a higher voter-per-worker ratio were weighted with higher priority than those with lower values.

MULTI-CRITERIA DECISION ANALYSIS

BEST POTENTIAL SITES

I.S. 093 Ridgewood is a large public middle school in the residential Ridgewood neighborhood of Queens that serves just over a thousand students. The election district that I.S. 093 would serve has a population of approximately 1,200 people, of which 51% are eligible voters. Approximately 45% of the population in this election district identifies as Latinx, 6% identifies as Asian, and 1% identifies as Black. This election district has a high poverty rate; an estimated 98% of the population live below the New York poverty line.

Harry S. Truman High School is a large public high school serving over 2,000 students and is located in Co-op City, a cooperative housing complex, in the Eastern Bronx . The total population of the election district that this polling station would serve is approximately 2,000 people, of which 80% are eligible voters. The approximate demographic breakdown of this district is as follows: 60% of the population identifies as Black, 29% identifies as Latinx, and 3% identifies as Asian. This election district also has a high estimated poverty rate, with approximately 90% of the population living below the poverty line.

P.S. 200, also known as The Benson School, is a large elementary school in the Bath Beach neighborhood of Brooklyn and serves approximately 1,200 students Pre-K to 5th grade. The election district in which the Benson school is located has a population of approximately 1,000 people of which 61% are eligible voters. Approximately 49% of the population identifies as Asian, 9% identifies as Latinx, and less than 1% identifies as Black. Our estimates indicate that approximately 100% of the population in The Benson School’s election district lives below the poverty line.

CONCLUSION

This Multi-Criteria Decision Analysis has provided us with valuable insights regarding specific sites that are most suitable for use as polling sites in New York City. Careful consideration has been given to factors such as polling site capacity and access. The various methodologies, including network analysis and emerging hotspot analysis, have facilitated our understanding of different areas in terms of their proximity to polling sites, polling site capacities, and changes in voter participation behavior over time. It is important to emphasize that this research has employed highly specific methodologies to measure access, capacity, and voter behavior changes over time. It should be noted that using different methodologies, calculation measures, or indicators in a Multi-Criteria Decision Analysis may yield different outcomes. Moreover, the strength of an MCDA relies heavily on the quality of its data and methodologies. Limitations within the datasets utilized, such as temporal gaps in data, modifications in spatial delineations overtime, and the necessity to infer or make assumptions, influence the final results of the MCDA. Nevertheless, the methods presented in this analysis have the potential to enhance the city's comprehension of service provision and trends within its existing election infrastructure. Ultimately, we aspire for this study, along with similar research endeavors, to contribute to the city's endeavors in incorporating more robust spatial methods into its election planning and site selection processes.

Sources

  1. "All Places Of Worship." June 30, 2020. Private Member, GeoPlatform ArcGIS Online. [Shapefile]. Available:  https://hifld-geoplatform.opendata.arcgis.com/datasets/geoplatform::all-places-of-worship/explore?location=40.710610%2C-73.833947%2C12.00 
  2. "Borough Boundaries." January 13, 2013. Department of City Planning (DCP), NYC Open Data. [Shapefile].  https://data.cityofnewyork.us/City-Government/Borough-Boundaries/tqmj-j8zm 
  3. "Building Footprints." May 3, 2016. Office of Technology and Innovation (OTI), NYC Open Data. [Shapefile].  https://data.cityofnewyork.us/Housing-Development/Building-Footprints/nqwf-w8eh 
  4. "LION." January 1, 2013. Department of City Planning (DCP), NYC Open Data. [Shapefile].   https://data.cityofnewyork.us/City-Government/LION/2v4z-66xt  
  5. New York City Board of Elections. Annual Report 2018, New York City, NY, 2018. 
  6. "NYC Election Districts." Jan 29, 2013. Department of City Planning (DCP), NYC Open Data. [Shapefile]. Available:   https://data.cityofnewyork.us/widgets/h2n3-98hq?mobile_redirect=true  
  7. "Primary Land Use Tax Lot Output (PLUTO)." February 2, 2020. Department of City Planning (DCP), NYC Open Data, [Shapefile].  https://data.cityofnewyork.us/City-Government/Primary-Land-Use-Tax-Lot-Output-PLUTO-/64uk-42ks 
  8. "School Point Locations." September 22, 2011. Department of Education (DoE), NYC Open Data. [Shapefile].  https://data.cityofnewyork.us/Education/School-Point-Locations/jfju-ynrr 
  9. "Voter Analysis 2008-2018." May 9, 2022. Campaign Finance Board (NYC Open Data). [Shapefile]. Available:   https://data.cityofnewyork.us/City-Government/Voter-Analysis-2008-2018/psx2-aqx3  
  10. "Voting Poll Sites." October 31, 2018. Board of Elections (BOENY), NYC Open Data. [Shapefile]. Available:   https://data.cityofnewyork.us/City-Government/Voting-Poll-Sites/mifw-tguq  

Figure 1