Research question

My research began with the question: Which city agencies respond to the most 311 service requests annually, and how does this request volume relate to their annual budgets?

The answers to these questions could factor into city-wide decisions around agency resourcing and budgeting. They could help the city government — as well as advocacy / special-interest groups — understand whether agencies are over- or under-funded when it comes to addressing city residents’ complaints.

The analysis

Below, I’ll break down each of my visualizations, describing the directions my research took me along the way.

As I dug into the data, I noticed that, perhaps unsurprisingly, the NYPD is by far the most active agency in handling 311 requests.

I also suspected that the NYPD is one of the more well-funded city agencies, which may better equip it to handle that volume of requests.

I wanted to pursue this line of thinking to see if it holds up across agencies. As it turns out, 311 service request volume is not directly proportional to an agency’s budget: some agencies are very well-funded relative to their request volume, and others not so much.

The Department of Housing Preservation & Development (HPD) stands out as being the No. 2 respondent to 311 service requests (with around half the volume of the NYPD) — and yet its annual budget ($1.8B) is less than a third of the NYPD’s.

This may have an effect on HPD’s ability to resolve 311 service requests in a timely manner. For instance, among the 5 agencies that receive the most requests, it takes HPD by far the longest (~15 days on average) to close out a ticket.

As I homed in on the Department of Housing Preservation & Development, I began to wonder what kinds of 311 requests it deals with the most — and when.

When HPD requests are plotted throughout the year, it becomes clear they have an element of seasonality: they reach a low point in the summer and a high point in winter.

The primary type of complaint driving this winter surge is heat/hot water requests. Starting in October each year, the city requires building owners to provide adequate heat to residents — and when they don’t, the city hears about it. (The city requires owners to supply hot water year-round.)

By zooming in on just 1 month — November, the peak of HPD request volume in 2023 — and on heat/hot water requests in particular, we can see that spikes in requests track closely with drops in temperature. In essence, there’s an inverse relationship between temperature and request volume: tenants tend to complain more about heat and hot water when the temperature falls.

This might seem unsurprising, but it’s important information for the city to know. By being aware of the tight relationship between temperature and 311 request volume, HPD could potentially seek additional funds and/or dispatch more staff ahead of winter storms and expected drops in temperature as a way of proactively supporting the city’s residents.

Data & design decisions

In general, I tried to tell one story per chart. Once I’d figured out what that story (or takeaway) was, I toyed around with various chart types and designs until I settled on ones that I felt told the story in the most direct manner.

I also decided to use different formats throughout — to add variety and continually engage the viewer — while sticking with relatively basic chart types that aren’t likely to confuse many readers.

My chart titles also progress from more neutral (the first two charts) to a more explicit point of view (the final three charts). I felt it was more important early on to be very clear about the shape and scope of the data. Further along, after establishing a foundation and shared understanding, it felt appropriate to be more descriptive of the narrative behind the data.

Next steps

There are many future directions I could take with this data and research question. For one, I could stick with the current research focus but expand the data being included — for instance, by seeing if the trend in HPD request volume (down in summer, up in winter) holds up consistently over the past decade, not just in 2023.

More interesting, I think, would be to see how shifts in budget over time relate to an agency’s ability to respond to 311 requests. If NYPD has a larger or smaller budget from one year to the next, does that impact how long it takes for the police to resolve a request?

I also noticed, in the initial steps of data analysis, that the Taxi & Limousine Commission (TLC) takes an average of 53 days (!) to resolve 311 requests. What gives?! Answering that would require digging deeper into the data.