Methodology of crime rate around migrant shelters

For the reporting story please see here.

Summary

  • The crime complaint data is counted at the block group level. The size of the block group varies, and can be seen from the map on the right. It is approximately 1-2 blocks. 

  • Complaint data is used other than arrest data, as it better represents the safety concerns of nearby residents. All categories of crime are counted.

  • The location of each crime is recorded as the nearest intersection by the NYPD.

  • For a few incidents, the location of the coordinates can’t match with the borough that the original data provides. These are excluded from the data analysis, and the maximum error is below 1%.

  • FOIL request to the Mayor’s Office and the DHS for open dates of all the asylums in the city is attached below.

Data Diary

The goal of this data analysis is to show the change of crime data around some shelters related to immigrants in NYC, and see if it backs up the safety concerns of nearby residents who protest against asylums.

0. Original dataset

The crime data all comes from NYC Open Data, provided by the NYPD. Complaint data is chosen instead of arrest data because it better represents the safety concerns of nearby residents, and its historical data map is easier to access. Data is collected at the level of intersections. Complaint data in 2023 is only available before Sept. 30. Complaint data in 2022 is filtered and downloaded from this map

The address of immigrant navigation resource centers comes from the Mayor's Office of Immigrant Affairs. Official press release shows that these sites were announced to open on Nov. 30, 2022. 

The open dates and addresses of other protested asylums come from news reports. 

1. From coordinates get GeoIDs

Original data provides longitude and latitude (EPSG 4326) for each complained crime incident. Dhrumil’s code is used to get GeoIDs from coordinates, specifically the census-geocode.ipynb.

The format of GeoID is 15 digits, for example: 360470074002001

State:36, County: 047, Tract: 007400, Block: 2001

Cut slice from the GeoID to fit the required format for Datawrapper:

NYC Census Block group: 604700740020

NYC Census Tract: 047007400

2. Data cleaning - methodological choice

Remove those data where coordinates are None, or zero, or generate errors in the code (very rare, 1 or 2 in 400,000). 

The original data has a column for the borough, which is described as “The name of the borough in which the incident occurred” from the data dictionary. From the GeoID, county code is also generated. A small part of them can’t match. 

(Those that don’t match are simply excluded from this data analysis. The maximum error is below 1%.)

The county code for the 5 counties in NYC:

Bronx: 005

Brooklyn: 047

Manhattan: 061

Queens: 081

Staten Island: 085

3. Select the time scale

The 9 immigrant resource navigation centers are announced to open on Nov. 30. The 2023 data is available for 9 months, from Jan to Sept. For the 2022 data to compare, 9 months before Nov. 30 is selected, that is March to Nov. 

For other shelters, take the Brooklyn sunset park recreation center as an example. The exact open date is hard to find in the news. But the newly built shelter was being protested on Aug 6, 2023. Two months before and after Aug 1 are selected to do an approximate comparison.

4. Data analysis and Visualization

Block group is the most accurate map level in datawrapper. The number of complaint data within the time scale is counted for each block group. Use the latter time scale to subtract the previous time scale, and get a difference, which is mapped through the whole NYC. 

The value in the map shows the change of complaint data number. Positive value (in red) means crime increases. Negative value (in blue) means crime decreases. Choose the maximum number for a better coloring, and set 0 to be in the center.

Different open dates have different time scales, and the map varies. 

Zoom in at the location of known migrant shelters:

FOIL Data

The FOIL request received a response from DHS, stating that they cannot disclose the addresses of shelters under NY Public Officers Law §87(2)(a), as addresses and information that could identify applicants or recipients of public assistance are confidential.

Below are their responses, including data indicating the number of sites designated as DHS Emergency Sanctuary-specific locations as of December 2023, along with an updated version from February 22, 2024.

December 2023

February 22, 2024

Previous
Previous

A Driver’s Life: Waiting, Waiting, Waiting for UberEats

Next
Next

Does crime increase around migrant shelters? Just illusion