Inside Hoodline's data-driven approach to crime reporting

Hoodline’s coverage of crime trends in U.S. cities is part of our local news automation project. For more general information on our data- and tech-driven news experiment, and the stories we’re producing, see the overview blog post here. Read on for more information about our crime coverage.

Where does the data come from?

For our crime stories, we use data provided by crime monitoring sites, mainly derived from incident reports uploaded by police departments and other law enforcement agencies. Our crime trend templates currently use data from SpotCrime and CityCop, two crime data aggregation sites. Both provide data with basic details about crimes reported in each city, including the type of offense, date and time, and location (down to the nearest block or intersection, for privacy).

We're always looking for new data sets to improve the geographic coverage and scope of our crime stories. If you’re a company or organization with locally relevant data that could help to create local crime stories, we want to hear from you—please email us at partner@hoodline.com.

How do we analyze the data, and what do we choose to report?

We analyze the data to identify straightforward trends, such as which types of offenses are increasing or decreasing most each period, and where in the city most reported crimes occurred.

Our team includes social scientists with research experience in public policy issues, as well as experienced local reporters who have covered major crime beats. While the local publications with which we partner still cover individual crime stories about the latest shooting or drug bust, our technical resources and expertise make us better able to report on larger-scale trends, including when and where crime might be going down.

Total crimes in Chicago per day for a three-month period, compared to the previous year.

We compare the latest incident counts to previous periods and—forthcoming in our monthly trend reports—to the same time last year, in order to identify whether current ups and downs appear to represent meaningful change rather than normal fluctuations or seasonal patterns. At the same time, our goal is to regularly report the latest developments to our readers. Even short-term swings in crime levels matter to local residents, as well as the latest count of major offenses, even if those counts do not fall outside the norm. We aim to provide enough statistical context while keeping our articles brief, readable, and timely.

Does the data capture everything that's happening in my city?

Police departments throughout the U.S. are under pressure to increase transparency and to systematically monitor and track what they’re doing with data, in order to reduce bias and improve the effectiveness of the services they provide. This pressure has made more data on local crime and police responses publicly available, and creates incentives for law enforcement agencies to ensure that the data is accurate, timely, and complete.

At the same time, there remain pervasive challenges with policing and public safety in the U.S. The crimes that are committed, incidents that get called in, calls that police respond to, reports that get written up, and ultimate outcomes are all influenced by many institutional and societal factors that go far beyond the data we have available to analyze.

We are therefore able to report on levels of and changes in official crime incident reports, broken down by offense, time, and location. But we remain aware that these numbers are not necessarily an exact match or definitive proof of what crimes were actually committed, or the areas and time periods that are seeing the highest and lowest rates of actual crime.

Monthly crime incidents in Chicago vs. one year ago, by most common offense types.

There is a wide and growing field of research and data analysis on police data, because it does capture important information about crime and police responses. We believe that that information is useful and valuable to local communities, but we continue to study how it should best be used, and what more we can do to improve what we are able to report.

What more can we do to address these concerns?

We are continuing to seek new and better ways to supplement and add context to official police reports, and are exploring additional sources of crime reports from social media and other distributed signals from local populations. At the same time, reports of crimes from a local population contain their own challenges, and may be less consistent and authoritative than reports determined credible by police officers, while continuing to suffer from residents’ or social media users’ own agendas and biases.

Deconflicting mentions of the same crime or offense, and validating that such signals are real, remain important challenges that each data provider handles differently. We continue to work with a number of current and potential data providers to find the best resources to use to identify, analyze, and report on meaningful developments and trends in local crime and public safety, as well as across the full range of other areas of local news that we cover.

You can help us improve our local crime stories.

Do you have thoughts on the local crime story that brought you here, or feedback about the information in this blog post? As this experiment in data-driven local news evolves, we welcome any and all feedback to help us improve. What are we missing in our reporting? What would make these stories more interesting or useful? Which of our stories are most engaging or valuable to you? What are your overall thoughts on what we’re doing in general?

Let us know what you think, right here.