Today’s world revolves around electronics: iPhones, 4K TVs, chips in credit cards, GPS mapping to get you from here to there and an ever exponentially increasing access to data storage and computational power. On a daily basis, we are living in and exposed to a world of big data; documenting so much of what we do with electronic markers. Home security systems scan for movements in our homes, phone apps can identify our location at any given time, and vendors track merchandise and services sent to our living spaces. It is the age of digital information and electronic footprints.

Social scientists find relevance and importance in using data markers to connect the dots of human behavior in logical ways that improve our understanding of society and provide essential information-driven observations to decision-makers.

At this year’s Association for Public Policy Analysis and Management Fall Research Conference, some of the Census Bureau’s leading social scientists will highlight the ways they harness big data. Topics include linking food stamp administrative records to survey data to help understand poverty; using tax records to examine racial differences in earnings for men or how parenthood affects the gender wage gap; and using time-use data and administrative records on earnings to study the payoffs of job search intensity on employment.

Highlights:

Can we use state data to improve poverty measurement?

Liana Fox and colleagues link state administrative records on Supplemental Nutrition Assistance Program (SNAP, formerly known as Food Stamps) with Census Bureau household survey data to re-estimate supplemental poverty. They find large underreporting of SNAP by survey respondents and its impact on poverty measurement matters. Pooling data on four states, underreporting of SNAP in the Current Population Survey leads to a 0.6 percent higher estimate of the Supplemental Poverty Measure. They show that precision in national measures of supplemental poverty can be improved by harnessing big data from the states.

Are our assumptions about race and returns to earnings all wrong?

Using the Current Population Survey linked to tax records, Michael Gideon and co-authors find that the effects of age and education on the earnings gap for black and white men depends on source data and that the impact of changing source data differ by race. Their study advances the literature by identifying the impact of data source on inequality measurement, in particular the black-white earnings gap.

What role does parenthood play on the gender wage gap?

Marta Murray-Close and her coauthor Eunjung Jee use linked data to estimate the wage gap between childless women and men, the motherhood wage penalty and the fatherhood wage bonus in the 1990s and 2000s. They use the Survey of Income and Program Participation linked to tax records and a new measure of labor-market experience developed with the linked data to determine key drivers of the overall gender wage gap in both the 1990s and 2000s.

Should assistance programs focus on job search intensity to help people get jobs?

Using the American Time Use Survey and administrative records on wages, Mark Klee and Lewis Warren find no evidence of returns to the intensity of job search for the average unemployed individual. In an attempt to explain this surprising result, they explore whether any type of unemployed individual exhibits returns. They show positive returns to job search for less educated individuals but not for their more educated counterparts. Their results suggest that job search intensity is not always a good indicator for re-employment outcomes.

As you can see, social scientists at the Census Bureau delve into big data to help inform interesting, relevant public policy questions. In a time of information abundance, we search for data, methods and strategies to harness already existing information to improve the quality of our household survey data and examine ways in which big data and administrative records can help answer big questions in social science.

Come see us at the conference! We love to chat about our data and look forward to talking with you.

###