In the age of smartphones with cameras and sensors, data are generated by our every move, whether physical or virtual. The sum of this information, commonly called “big data,” may create new opportunities for social researchers to learn about human behavior. To better prepare students for this ever-evolving world, the Program in Survey Methodology at the University of Michigan’s Institute for Social Research is introducing data science to its curriculum and a concentration in data science for students looking for more depth. The first cohort of survey methodology graduate students whose studies incorporate data science began the program in Fall 2017.
Our choices create information that, for better or worse, can be accessed and analyzed — by humans, and computers. For social researchers, these new data sources are more complex and varied than the data created by traditional methods such as surveys. Analyzing and harnessing these data sources require new skills and tools unheard of just years ago, such as text mining, data mining, and machine learning.
“We hope to prepare the next generation for the workforce they’ll find themselves in, which is clearly changing,” said Frederick Conrad, director of the Michigan Program in Survey Methodology Program. “Our goal is to prepare survey methodologists for this new world but also to give them the basic tools to adapt. The data science industry is rapidly developing right now, and we want our students to be able to work with new sources of data and to critically evaluate them for their quality.”
The survey methodology curriculum at U-M is changing in two ways: Data science principles will be taught in existing core courses for all survey methodology students; and students will now have the option to pursue a concentration in data science, in addition to the two existing concentrations in social science and statistical science.
U-M research associate professor Brady West said the decision to supplement the survey methodology program with data science curriculum began after talking with employers of former survey methodology graduates. Virtually 100% of U-M survey methodology graduates are employed in their field after graduation, but employers said that, going forward, ideal candidates will be trained in advanced analytical techniques as well as in traditional survey methods. West led a curriculum committee comprising survey methodology faculty members to determine the best way to supplement core courses with data science skills and which electives were best for the data science concentration.
“The spirit of creating the new data science curriculum is to direct students to courses we think will enhance their training, and it’s motivated by external demand for these skills,” said West.
That external demand is what pushed Preeti Gill to pursue the data science concentration at U-M this fall. A senior programmer analyst at Mathematica Policy Research, Gill remembers being asked to perform a research forest analysis (a type of machine learning), a technique to which she had no previous exposure. She holds a BS in Industrial Relations from Cornell University and a MA in Sociology from Stanford University, and says most of her professional background is in administrative data. After using tuition reimbursement from Mathematica to take a class in advanced statistics at U-M, Gill says her eyes were opened to a “whole new side of social science research that was exciting and interesting to me.”
“Through my work I began catching wind of how things are changing in the social science industry,” said Gill. “I want to be better able to work with large data sets and have the ability to teach myself to adapt throughout my career.”
To learn more about the data science concentration and the Michigan Program in Survey Methodology, visit http://psm.isr.umich.edu/.