Jonathan Auerbach, who is currently the statistician for the New York City Council, recounted lessons learned from a Data Science for Social Good project focused on 311 service calls in Chicago. Auerbach’s team uncovered the unique, local level of program knowledge needed to understand and evaluate each department’s data before using it in any planning. In one example, data showed that service requests about overgrown grass were particularly high in some neighborhoods in south Chicago.
In an interview with city employees, Auerbach learned that the high number in one area was due in great part to the persistence of one resident (“Oh, that’s Nate”) spotting and reporting overgrown lots. Data at the city neighborhood level is not so rich as to overcome an individual’s influence. Auerbach also learned that weedy lots in some neighborhoods are popular places for criminals to dump drug-dealing evidence, giving those service calls a different context from similar calls from a low-crime area.
Auerbach believes city transactional data can be used in planning, but analysts should consider how the data is collected and used in a department’s daily operations before extracting any higher level of meaning from it. He also joked that Big Data at the city level could be defined as “any data that can’t be pulled easily into a spreadsheet.”
Auerbach’s talk provided a great lead-in for the presentation by the Census Bureau’s Elizabeth Accetta: “The Census of Governments: Cautionary Tales in Analyzing Government Data.” Accetta detailed the incredible amount of research and local knowledge that goes into creating a set of government finance and employee data that is comparable across wildly different state and local government entities. Drawing conclusions from the data requires familiarity with differences in governmental structures, fiscal years, and operations. The Census Bureau has started to release preliminary data from the 2012 Census of Governments; more will be released throughout 2014 and 2015.
American Community Survey: The Elephant in the Room
The Census Bureau’s American Community Survey (ACS) program has come under heavy attack from some members of Congress who have responded to some constituent privacy complaints. Congressmen have threatened to eliminate the ACS, a significant survey program designed to replace the former decennial census long form.
Defending ACS was a huge focus of the 2012 APDU meeting. This year, James Treat of the ACS Office at Census detailed the extent to which the Census Bureau is working with Congress to address constituent concerns and demonstrate the value of ACS information to each congressional district. The new American Community Survey Users Group—a nongovernmental group—was announced at the meeting as a group seeking “to improve understanding of the value and utility of ACS data and to promote information sharing among data users about key ACS data issues and applications.”
In addition to its professional educational role, APDU provides informed advocacy for the support of federal statistical programs. The last two sessions of the annual meeting were The Road Ahead: The Challenges Faced by Federal Statistical Agencies and Defining the Data User Advocacy Agenda. Interested ACS users can consult a slide deck with session notes and follow APDU for updates.