UPDATE 12/30/2022: The AJAM projects server is no longer live. Files available upon request.
This is one of the best data projects I've ever been able to work on.
Investigative journalist Greg Palast and a team of others led the charge on this reporting. Over several months and many requests, they successfully FOIA'd for lists of alleged "double voters" from the Interstate Crosscheck program. The Crosscheck program has amassed a list of nearly 7 million by matching names from roughly 110 million voter records of participating states. Palast got copies of the list from three of the 28 participating states: Washington, Virginia and Georgia.
Examination of the lists revealed a host of problems. The names were a simple first and last match -- a different middle name is ignored, as is a junior or senior -- and the lists unfairly targets minorities with similar names:
The three states’ lists are heavily weighted with names such as Jackson, Garcia, Patel and Kim — ones common among minorities, who vote overwhelmingly Democratic. Indeed, fully 1 in 7 African-Americans in those 27 states, plus the state of Washington (which enrolled in Crosscheck but has decided not to utilize the results), are listed as under suspicion of having voted twice. This also applies to 1 in 8 Asian-Americans and 1 in 8 Hispanic voters. White voters too — 1 in 11 — are at risk of having their names scrubbed from the voter rolls, though not as vulnerable as minorities.
I helped build the graphics and the lookup database where users can explore the list of names that show up on the Virginia and Georgia lists.
The database uses flat files so that users typing into the box above get a Google-predict feeling; it also helps prompt the user to know they're typing a name with results. We also added name suggestions at the bottom of the box. These randomize between the top 25 most common last names.
The results look like this:
We wanted people to be able to see the names -- we hid addresses and other details -- but we wanted them to easily see the mismatches that occur because Crosscheck ignores middle initials. We also used data from the U.S. Census 1,000 most common surnames to help show the likely race or ethnicity of the people with this name, since minorities were disproportionately represented on the lists.
Palast won the Sidney Prize from the Sidney Hillman Foundation for this article. The project was also a finalist for the Online News Association's 2015 awards (planned news/events, large), a 2015 "General Excellence" Data Journalism winner, and a finalist in the 2015 EPPY awards for "best investigative/enterprise feature."