The Trump Administration has relied on the false argument that immigrants bring crime into America to drive its immigration policies: restrictions to entry, travel and visas; heightened border enforcement; plans for a wall along the border with Mexico. We showed that these claims are not accurate.
In a large-scale collaboration across four universities, a team of researchers led by Robert Adelman of the State University of New York at Buffalo studied 200 local metropolitan statistical areas across the country over four decades, concluding that there is no correlation between higher immigrant populations and higher crime rates. Building upon the data from this report, The Marshall Project collaborated with the authors to extend the data to 2016, and then used that data as a basis for a further in-depth time-series analysis and visualization of trends in immigrant populations versus violent crime rates.
The Marshall Project analysis found that a large majority of the metro areas studied have more immigrants today than in 1980, and fewer violent crimes. As of 2016, the latest year data was available, crime fell more often than it rose even as immigrant populations grew almost across the board. The top ten places with the largest increases in immigrants all had lower levels of crime in 2016 than in 1980.
Our reporting also took a close look at violent crime, breaking it down into its components to find that growth in immigrant population again failed to produce increased rates of assaults, robberies or murders. Most areas experienced drops in all types of violent crime, accompanying widespread increases in immigrant population.
The Marshall Project then built a visual story to communicate these findings and expose the underlying data city-by-city. The project goal was to not simply ask readers to believe the statistical results of the original academic paper, or the results of our further time-series analysis – but to give them the tools to see the data for themselves, from their own cities, and make up their own minds.
What makes this project innovative?
Our challenge in reporting this story was to address a highly controversial topic and cut through false claims and cherry-picking to show that, despite rhetoric to the contrary, immigration does not drive crime. Our approach to this challenge was to design the analysis to communicate transparently with readers. In addition to citing the academic result that statistical models showed no correlation between immigration and crime rates, we showed this lack of correlation more viscerally by framing it as change over time. This allows readers to see the trends literally diverging: more immigrants are associated with less crime. Our goal was to try to reveal the data in as transparent, accessible and usable a form as possible, to encourage readers to look for themselves and make up their own minds. We try hard to build reader participation and trust, but it is also very hard to measure – we continue to struggle with this story and with others in determining when we have succeeded and when we have failed.
What was the impact of your project? How did you measure it?
Our story added needed data-driven foundation to the public discussion around immigrants and crime, and the debate on how immigration policy should be shaped. Viewed widely, the article was shared thousands of times within the first 12 hours alone, including by Chelsea Clinton, Katie Couric, Chris Hayes, Khoi Vinh, Peter Baker, Jim Acosta, Maggie Haberman and Michael Barbaro. The combined social reach of those in the media who tweeted the story reached 9.5 million in that first day. Politicians and members of law enforcement have also shared the story in policy discussions since, including Beto O'Rourke, Sam Liccardo, Art Acevedo, Veronica Escobar, the Philadelphia DA’s office, and the Office of the Ohio Public Defender. The article has been viewed over 50,000 times on our site, and as of our last check with our partner in July of 2018, had about 300,000 views on the New York Times’s The Upshot page. Multiple other news organizations picked up or cited the work, including the Washington Post, NPR, The Atlantic, The LA Times, The Guardian, GQ, Washington Monthly, The Chicago Tribune, Salon, The Hollywood Reporter, ABC, Daily Kos, FiveThirtyEight, and dozens of others. Flagg was interviewed by Hari Sreenivasan on the PBS Newshour, which also featured visuals from the project adapted for television. Months after its publication, the story continues to be a leading fact check for the anti-immigrant claims by the Trump administration and other politicians across a range of news organizations. In November, the new Netflix weekly talk show “Patriot Act with Hasan Minhaj” featured our data visualizations and findings in a segment.
Source and methodology
Anti-immigrant rhetoric was widespread before the 2016 presidential election, and continues to this day. The Marshall Project wanted to investigate some of the more common claims related to immigrants, particularly those citing a connection to crime. While looking into data that might show a correlation – or lack of correlation – between immigration rates and crime rates, we found Adelman and his team’s research, and decided to collaborate with the authors to extend the data and build upon their analysis. We collected 2016 immigrant population numbers from the American Community Survey one-year estimates, and crime figures from the F.B.I. Uniform Crime Reporting Program metropolitan area data sets. Our main difficulty was in handling the metropolitan areas that had changed over time. In the years since 2010, many areas grew to include additional regions or split into multiple smaller ones. The Marshall Project referenced metro area technical documentation and consulted with the study authors to determine when a larger area was still an appropriate match to the original described in the paper. When an area split into components, raw data from each was added to calculate rates approximating the original region. When no reasonable approximation to the original area could be found, it was marked as missing for 2016. Missing data was also a problem for a small number of areas for some years. When an area was missing information for a certain year, that year's data was linearly interpolated using figures from the closest year(s) available. This process resulted in a dataset of immigrant population and violent crime rates, including assaults, robberies and murders, for each of the years 1980, 1990, 2000, 2010, and 2016. We also compiled the corresponding national data for those years. In addition to citing statistical results from Adelman and his team showing no correlation between crime and immigration rates, we did an in-depth time analysis to characterize the diverging trends of immigration and violent crime, focusing particularly on how each has changed since 1980. We performed this analysis for overall violent crime, as well as for its component parts of assaults, robberies and murders.
We used R for analysis. Design and development was done using D3.js and Adobe Illustrator.
This project was done by Anna Flagg. We are very thankful for the help of Robert Adelman and his team of researchers, especially Charles Jaret and Gail Markle.