Jon Schwabish Discusses Big Data for Public Policy at the Urban Institute

[MUSIC PLAYING][Jon Schwabish Discusses Big Datafor Public Policy at the Urban Institute]

JON SCHWABISH: My name is Jon Schwabish.[Jon Schwabish, PhD Senior Fellow Urban Institute]I'm a senior fellow at the Urban Institute,a nonprofit research institution in Washington, DC.I actually have two jobs at the Urban Institute.Half my time is spent in what's called the Income and BenefitsPolicy Center.That group tends to focus on low income programs, sofood stamps, TANF, WIC, social security, disability issues,

JON SCHWABISH [continued]: education as well.My research focuses on food stamps, the SupplementalNutrition Assistance Program, SNAP,for those who are not familiar, also disability issues.And so that's a fairly standard research department.So we are writing papers.We are doing reports.I also help run the Social Genome Project,

JON SCHWABISH [continued]: which is a microsimulation model that takes people from birthall the way through middle age.That's half of my job.The other half of my job, I'm in the communications department,where I help researchers and colleagueswith their presentation skills, and with their datavisualization work, and generallyjust how to communicate their work,be it in their standard reports, be it on our urban blog,

JON SCHWABISH [continued]: or be it in their seminars, or workshops,or talking to funders, or other practitioners, or policymakers.Overall, Urban is a nonprofit research institution.We're about 500 people, spread across 10 different divisions.We focus on a wide array of-- primarily US domestic policies

JON SCHWABISH [continued]: and programs.We have a justice program, a health program,a housing program, obviously the low income program.So all these different groups work together.We have a lot of different cross--what we call cross-center initiativesto try to bring these groups together, where sometimesthey might be siloed off.But we're trying to bring them together.

JON SCHWABISH [continued]: And overall, what we're trying to dois elevate the debate around important public and socialpolicy issues.So we communicate to lots of different groups.And as we do so, what I think we are uniquely good atis thinking very carefully about our different audiences.So we have the big academic PDF-style report.We have issue briefs.We have a blog.We have really great data tools.

JON SCHWABISH [continued]: We have narrative features that we put on our website.So we have all these different communication methodsand mechanisms.And we're trying to think very carefullyabout how our content best reaches our audiencesand it helps them do their jobs.So it's a great organization.We have lots of different tools, techniques, and technologiesthat we're now trying out.We're moving a lot of our simulation models

JON SCHWABISH [continued]: into the cloud, which I think is a new techniquefor a lot of nonprofit research institutions.We're using social media data.We're using credit card data.Of course, in addition to all the regular economic datawe get from the Census Bureau and other government agencies.I actually think that we've been workingwith big data for a long time.I guess it depends obviously on how you define big data.

JON SCHWABISH [continued]: So I've used data from the IRS and the Social SecurityAdministration, where I'll have a sample of hundredsof thousands of people, with dataall the way back to the 1950s.So when you multiply that out, it'smillions of data points, millions of records.But I think what we need to do is go beyond thatand think what is big, what is huge data, where--maybe it's real time or it's intersecting

JON SCHWABISH [continued]: with different people.So social media data or data coming outof businesses and corporations.And to harness that information and beable to process it and analyze it,we need to be thinking carefully about different tools.We need to be thinking about how wemight use new technologies, like moving our models from where wehave them currently sitting on our desktop machines,on our local hard drive, to maybe using the cloud

JON SCHWABISH [continued]: or using different servers that are fasterand allow us to do more with the models.So we might be using multiple processors, for example.And I don't think researchers needto feel like that's something that theyneed to do specifically.That there are experts in those fieldsand we can go use those people and askthem to help us to figure out the best way to do that.The biggest challenge social science researchers

JON SCHWABISH [continued]: face with big data is they're probably notused to working with that amount of dataand trying to take that amount-- that size of data and allthe different dimensions and winnow it down to a story,or to an argument, or to a point.We often feel like we have to show all of our dataall the time when we're presenting it.And with big data especially, that's really hard to do.

JON SCHWABISH [continued]: So I think there's a big communication effortchallenge that researchers to have to overcome.And obviously, there's a technical challenge.And I think a lot of researchers--at least in the field of economics and public policy,we're used to opening up Stata, or SaaS, or SPSS.And those tools may not necessarilywork when we're talking about real-time Twitter feeds comingin, with hundreds of thousands of observations.

JON SCHWABISH [continued]: So there's lots of challenges.And there's probably-- like the technical challengeis a big one.But also then the communication challenge,because there's lots of different stories, and topics,and arguments that we want to make,that maybe we're not used to doing right now when we're--have smaller data or even qualitative data.