b'ManagementTable 2: Summary of key characteristics of case studies World Banks Social Observatory APHIS Public Outreach LA Express ParkInterdisciplinary team of WorldAPHIS (agency) and researchersU.S. Department of Bank researchers, governmentat Fors Marsh Group (contractor) Transportation (main funder), Actors and stakeholders actors at various levels in IndiaLos Angeles City, Xerox-PARC (from leadership of Tamil Nadu(contractor)down to villages), womens self-help groupsTarget users Rural residents, mostly womenDomestic and internationalMotoriststravelers Big data source forPilot census survey of 32,000AQIM (Agricultural QuarantineReal-time parking occupancy quantitative analysis residents in Pudhu Vaazhvu,Inspection Monitoring) datasetdata from 6,000 smart metersTamil NaduResearchers are embedded inInput from agency experts at allRapid ethnographic method Thick data source forcommunities to observe andstages of the research process;(compressed periods of qualitative analysislisten to deliberations; directresearchers visited airports andobservation, interviews, direct participation by women self-helpinterviewed agency officers andparticipation, and videotaping)groups in designing survey travelers P-tracking: survey designed withTargeted public outreach andDemand-based parking pricing, direct inputs from local women;social marketing campaign, withpaired with mobile apps for Product collected data was visualizedmessages tailored by location andviewing and booking parking and shared with communities tosegment of travelers spacesinform deliberation and decision makingSource: Yuen Yuen AngLessons LearnedThick data can inform the analysis of big data. Thick data Big data is a means to an end, rather than an end. Thealso informs the analysis of big data. In the APHIS case, ongoing focus on big data may compel public managersdata scientists worked closely with agency experts to to feel that they need to do something with the data,understand the construction and quality of the data sets, whether or not this is necessary or useful. In the U.S,seeking inputs for the design of statistical models and recent federal plans that link big data to the nationsinterpretation of results. The data scientists who worked strategic assets may unintentionally reinforce suchwith APHIS found: Because these results were to inform pressures. But while public agencies should seriouslythe development of a campaign that would not exist in consider big data as part of their toolkit, they should nota vacuum but instead would be implemented in a mix use big data just for its own sake. All three case studiesof policy, political, and budgetary influences, the final illustrate big data as a means to an end, rather than ansolution could not be determined by the machine-based end in and of itself. analysis of the results alone.Thick data can identify unexpected problems orMixed analytics can offer both scale and depth. Given previously unexpressed needs. Clearly, thick data andthe different advantages and functions of big data and ethnography can complement big data analysisthethick data, the best research teams and technology harder work arises in specifying how. Governments candesigns typically use mixed analytics (big data and thick miss obvious problems if they rely only on big data anddata) and mixed research methods (quantitative and analytics, as best illustrated by the case of LA Express Park.qualitative). In addition, they feature an interdisciplinary Thick data also proves particularly useful for informing theteam of specialists, not just data scientists. The World collection of big data, as seen in the World Banks SocialBank brought in economists, sociologists, behavioral Observatory project. If public agencies invest time andscientists, and information system specialists. effort to collect big data without first inquiring what usersDesigning LA Express Park involved both engineers and care most about, they may measure the wrong things. ethnographers.94 www.businessofgovernment.org The Business of Government'