On Sept. 20, 2017 Hurricane Maria devastated Puerto Rico — the strongest hurricane to hit the island in the last century. In the weeks after the storm, the government insisted that only a few dozen people had died, yet reporting on the ground by the Center for Investigative Journalism (CPI) suggested hundreds had perished.
With no reliable official death toll, CPI, Quartz and the AP teamed up to build a database of victims. “Hurricane Maria’s dead,” the resulting database and website co-published by the three news organizations in both English and Spanish, is the most extensive record, in both size and detail, of individuals whose deaths are associated with the hurricane, based on CDC guidelines. As far as we know it is also the only large-scale effort by a media organization to use an online survey to record natural disaster deaths. It lists victims by name and includes demographic details, such as age, place of residence, and in most cases, cause of death.
The project started as an attempt to overcome the Puerto Rican government’s refusal to provide information about the hurricane’s death toll. It repeatedly denied requests for mortality statistics and other data and documents, such as death certificates and burial permits, that could help the public understand the extent and causes of the death toll.
It was clear early on that it would be impossible to get at the scale of the death toll through traditional reporting methods alone. So we decided to partner up. Each organization brought unique strengths to the process without which the scale and depth of the project would have been impossible to achieve. By working together, we enhanced our individual expertise in key areas of our project, such as on-the-ground reporting and data analysis, and were able to cover more ground. The collaboration also gave our project both a local and international perspective.
The first step was to reach out to the public for help with an online survey. We verified the survey deaths by matching the victims’ names with government death records that CPI eventually obtained through a lawsuit, and through more than 300 phone interviews with victims’ relatives, all of which were copy edited and translated into English and Spanish.
The project took nearly a year. It required designing and building a data collection and management system from scratch. The crowd-sourcing aspect necessitated a large-scale data validation and verification process that required automating certain tasks. We partnered with Document Cloud for hosting and information extraction.
It also involved assembling a cross-regional team of reporters from the three partners and volunteers to follow up with hundreds of grief-stricken families who’d lost loved ones. We recruited students from the Carlos Albizu University in San Juan, Interamerican University and University of Puerto Rico, who volunteered to do data extraction and conducted many of the verification calls.
We collected more than 500 cases through the online survey and 300 personal interviews, all of which were copy edited and translated into English and Spanish.