How long does it take the police to respond to people calling 911 in New Orleans, compared to in Seattle? What about in the wealthier areas of those cities, compared to their poorer areas? How often do people call the police about suspicious activity in Detroit? And how often are those suspicions founded? What about in areas of Dallas?

Questions like these are often at the heart of people’s concerns, and their calls for criminal justice reform in cities across America. But they have often been answered purely anecdotally, thanks in part to a lack of transparency and sophistication in how policing data is released to the public. There has been no national aggregated dataset of 911 calls, or even a national standard for collecting and reporting such data.

Until now.

Pretty soon anyone with a computer and a healthy sense of curiosity will be able to find the answers to such questions, and ask plenty more, thanks to a new open 911-call data initiative being spearheaded by the Vera Institute of Justice and the analysts at Two Sigma Data Clinic.

Thanks to newly cleaned and restructured accessible and user-friendly data sets, community and grassroots groups in five American cities will be able to compare 911 data, based on open datasets released by Charleston, Dallas, Detroit, New Orleans, and Seattle, with more cities’ data expected to come online, soon.

Vera and Data Clinic will be hosting the data on an open source platform so that anyone can look into far more accessible and understandable versions of existing open-datasets. These cleaned and standardized versions will make it easier to assess patterns and trends in policing based on 911 calls and officer-initiated interactions. Cleaning and analysis techniques coupled with the corresponding code for doing so will be hosted on the Data Clinic github site.

Our organizations have spent months cleaning up and standardizing the data between the different department so that it’s possible to compare and contrast across jurisdictions. The hope is that people in different cities will be able to equip themselves with evidence about 911 based policing outcome compared to tactics and outcomes being employed elsewhere.

With the data in a consolidated format, we can start to ask interesting questions through targeted queries. For example, does the type of 911 calls change based on specific neighborhood characteristics?

Using the updated New Orleans dataset, we can quickly produce visualizations that, for example, show the breakdown of call outcomes, the ratio of officer-initiated calls, and predominate call types for census tracts with different income distributions. In New Orleans, for example, the data shows a higher percentage of 911 calls about drugs and violent crime in areas with the lowest median income, compared to the highest:

The data also show that officers tend to initiate more interactions with the public on suspicion of some crimes in the poorer neighborhoods, than in the richer ones. And vice versa. In particular, officer-initiated drug interactions are higher in poorer neighborhoods, according to the data:

And the data also shows that police interaction with the community tends to be unfounded slightly more often in poorer communities than richer ones:

This sort of analysis is just the tip of the iceberg, and we’ll be blogging about how best to use the data here. We’re expecting people to reach out so that they can ask more questions.

We would love to start a conversation about how Vera and Two-Sigma can support your 911 data exploration in your city, and how you can use the data aggregation and cleaning system to make the process easier. You can find out more about the project by emailing Frankie Wunschel at fwunschel@vera.org, and you can read more about the underlying methodology, here.