How NYC 311 Pays Attention to the Poor

On a recent trip to New York City a friend of mine made the comment that city officials seem to prioritize the needs of the city’s ultra-wealthy over those most reliant upon city services. In my time researching NYC’s service delivery mechanisms I’ve found this sentiment to be completely untrue.

I’ve condensed a more readable version of a deep-dive I conducted into the subject back in 2015, the full report can be found on my website.

tl;dr

Two years ago I sourced 925,391 NYC 311 service requests (Jan. 2015 — June 2015, via NYC Open Data Portal), extracted each request’s geolocation metadata, and intersected them with city parcel data.

3 Interesting Results

  • If a service request is made in a wealthier neighborhood, then those incidents hold a marginally higher chance of going unresolved
  • Incidents assigned to the departments of Parks & Recreation or Health & Mental Hygiene have the worst resolution rates amongst all city agencies
  • In the 6-month period, there was a service request made for every 9 NYC residents with a total resolution rate of 97.44%

By all accounts the face of NYC service delivery did an outstanding job, no matter who made the request, or what it was.

NYC 311 Development

Widely considered standard operating procedure in today’s major cities, the concept of a city call-in service request system is still relatively young. First seen in 1996 (Baltimore) and 1999 (Chicago), the service was quickly adopted by US cities, big and small.

Under former-Mayor Michael Bloomberg NYC launched its own program in March of 2003 — now the largest in the world, NYC 311 accommodates 180+ languages, integrating with 300 city agencies, and fields over 50,000 call per day.

NYC 311 has answered over 158 million calls to date, 22.2 million more than the next 26 largest systems combined.

Research Methodology

In the proceeding analysis, NYC 311 volume and resolution has been examined against the backdrop of the incident’s originating density (household density), income (median income), and assigned agency.

In addition, incidents within a 500-foot radius of New York Housing Authority Developments (NYCHA) have been segmented to evaluate service response time to those most reliant upon them.

Due to the incredible volume of the NYC 311 system, this analysis will focus on a 6-month snapshot in time — January 1, 2015 through June 30, 2015. An additional 3-month grace period has been applied through September 30, 2015 to allow for service resolution of late-term requests.

Data Sources

Information for this analysis came from a variety of sources, which together helps paint a vivid depiction of the performance of the NYC 311 agency.

  • NYC Open Data Portal: 311 Service Requests (call data), Borough Boundaries, NYCHA Developments, Street Centerlines, 2010 Census Tracts
  • American Community Survey (2013): All Borough Populations, All Country Household Income

Analysis

My analysis specifically focused on the resolution (closed incident) rate of the NYC 311 service amongst census tracts (household density, median income) and NYCHA developments. At no point did I position NYC 311 at an analytical disadvantage or hold a position of existing discrimination.

Before breaking down the findings its prudent to describe the impact a magnitude of change represents in this dataset. Due to the quantity of requests into the system, even a 1.00% difference is equivalent to over 9,000 incident reports — small percentages = big numbers.

Service Adoption & Volume

Amongst all income classes NYC 311 service usage is consistent, ranging from 10.00% to 11.98%. Accounting for requests made within the 500-foot radius of a NYCHA development, adoption spike to 14.75% — almost 32% higher than the city-wide average.

In almost every case the NYC 311 system received substantially more of its volume from lower income populations; however, when adjusted for population density a more consistent usage rate is prevalent.

Staten Island ($145,084–250,000+) volume was 138.60% — it was excluded from the chart above to decrease extreme outliers impacting the legibility of the above chart.

Noticeably high usage is found in the following areas:

  1. Staten Island, $145,084–250,000: 138.60%
  2. Brooklyn, $145,084–250,000: 19.08%
  3. Bronx, NYCHA Developments: 18.79%

Service Resolution Rate

When comparing NYC 311 volume to population density we find the two metrics to be almost perfectly correlated — while correlation likely doesn’t equal causation in this case, the relationship is useful in hypothesizing two potential characteristics of the service:

  1. Higher Income Areas = Lower Resolution Rates
  2. Lower Income Areas = Higher Resolution Rates

After accounting for the underlying population changes, plotting a General QQ Plot chart, and mapping the relationships to area income levels we find a hypothesis confirming result.

There is a definitive positive trend between increasing income levels and decreasing resolution rates.

Succinctly, incidents located in the highest income levels hold a 48.60% higher chance of remaining unresolved than those originating in the lowest income levels.

NYCHA Developments hold a comparative resolution rate to that of the city-wide average; therefore, the housing developments actually receive better overall service than high income areas.

Household Density

When controlling for household density, NYC 311 adoption rates hold at similar levels to that area’s median income adoption rate. However, examining density classes separately shows a downward trend in request volume as density levels increase — dropping from 11.75% (high) to 9.33% (low) — suggesting a potentially negative relationship between increasing density and increasing NYC 311 request volume.

We cannot take this result at face value. In fact, by further scrutinizing the underlying population of these density zones show the lowest 3 levels of household density hold 86.00% of the city’s population.

By repeating the General QQ Plot for both total request volume and non-resolved requests no discernible relationship can be drawn between request volume and household density.

Responsible Agency Performance

Another interesting variable to explore is the resolution rate (performance) of agencies assigned by NYC 311 with request completion. Below is a table depicting agency service burden and their overall completion performance.

Abnormally low resolution rates (orange) and agencies bearing the most burden (green) have been highlighted.

The Department of Parks & Recreation (DPR) stood out with a 60% resolution rate, ~30% below the median rate. In fact, DPR is responsible for 41.64% of all outstanding unresolved service requests. Many things might explain this fact, some examples: longer service completion timelines, environmental impact assessments, etc.

The Department of Health & Mental Hygiene (DOHMH) was another standout with a 72.74% resolution rate. Similar to DPR, DOHMH was responsible for a large portion of outstanding requests, 31.21%.

DPR and DOHMH bear the joint responsibility of 72.85% of all NYC 311 requests listed as unresolved in the period examined. The two agencies only account for 5.59% of all incident reports, further signaling them out for potential underperformance.

Congratulations is owed to the agencies most burdened with NYC 311 requests, they are among those with the highest resolution rates in the city.

Conclusion

NYC 311 is widely regarded for its customer service, responsiveness, and comprehensiveness; thus, its no surprise no “smoking gun” was discovered when the agency’s performance was scrutinized. Some final bullets:

  • NYCHA residents use the NYC 311 service at a higher rate with comparable performance to city-wide averages
  • When controlling for underlying population, NYC 311 s usage at similar rates regardless of population income levels
  • Household density has little impact on NYC 311 resolution rates
  • Residents in the highest income levels are 48.60% more likely to have incidents go unresolved compared to those in the lowest income level
  • Agencies responsible for the most service requests hold the highest performance — noticeable under-performers are DPR and DOHMH

Further Study

At the time of this report the request data for the NYC 311 iOS and Android application were not available via the NYC Open Data Portal. A Freedom of Information Law (FOIL) Request was made to obtain the preliminary (unstructured, incomplete, and non-geolocated) dataset for the native apps.

After reformatting this preliminary data (2015), results for the same 6-month period as this analysis were as follows:

  • 69,975 total app requests vs. 925,391 call-in requests
  • 71.17% app resolution rate vs. 97.44% call-in resolution rate
  • iOS devices accounted for ~70% of all app requests
  • Brooklyn (34.07%) and Manhattan (27.17%) were by far the highest users of the native apps

Since tracking inception in 2013, all native NYC 311 mobile applications received 217,941 (ending June 2015). In the 6-months of this analysis, the NYC 311 call-in centers fielded 4.25x that volume.

Articles Referenced

New York City Government. “The NYC 311 Story…your city. your needs. your number.” Government Technology (2009).

Marc LaVorgna, Kamran Mumtaz, and Nick Sbordone. “Mayor Bloomberg Commemorates Ten Years of NYC311, The Nation’s Largest and Most Comprehensive 311 Service.” New York City Government (2013).