Volume 25/1996-1997 contents | Duke Policy News Online | Sanford Institute


"I take the tools they learned in economics and apply them to policy questions such as food stamps, minimum wage and international trade."

RESEARCH

Anti-Poverty Policymaking

Politicians love to talk about eliminating welfare, expanding employment and improving education. But economist Becky Roselius wants to know which social programs actually work to alleviate poverty, and why.

Roselius, one of the Institute's newest assistant professors, received her Ph.D. from the University of Chicago, where she was a doctoral fellow of the Center for the Study of Urban Inequality. Her research focuses on effective evaluation of social programs, such as job training. She looks at the costs and benefits of education and training programs, and their policy implications.

For example, Roselius re-evaluates non-experimental methods for evaluating JTPA, the federal job training program for low-income, low-skilled workers. The original, experimental evaluation found low success rates for youth, and only moderate success for adults. Although experiments are considered the most accurate method to evaluate social programs, experiments can have inherent biases and can be politically problematic and disruptive to ongoing programs. Roselius and her colleagues at the University of Chicago compare several alternative, non-experimental methods to the experimental results and uncover many of the underlying flaws in non-experimental methodology. The flaws arise in part from problems with bias. This type of bias occurs because participants in the program are compared to similar persons who did not participate but differ in unobservable ways that effect program outcomes. Roselius' studies reveal that better methods can overcome selection bias and other flaws with non-experimental methods.

Roselius' current and future research will continue to apply and improve on these methods in settings where selection bias is likely to arise. Her newest research will estimate the impact computers have on learning in school and productivity at work. The methods honed in her previous research will attempt to control for the selection bias that arises when comparing grades and productivity of persons who use computers to similar persons in similar situations who do not use computers.

These research issues dovetail nicely with the questions Roselius raises for her undergraduate and graduate public policy students. In her undergraduate class on Economic Analysis and Public Policy, Roselius teaches students to use economic tools to evaluate the impact of public policies.

"I take the tools they learned in their introductory economics class," she says, "and apply them to policy questions such as food stamps, minimum wages, international trade, and labor force participation decisions." Students are encouraged to communicate with Roselius and each other between classes, via a class newsgroup on the Institute's web page. In this way, students can post a question, or an article to the whole class, for email discussion and further discussion in the next class. The students also learn to use the web to find data to analyze policy questions and to find articles related to their interests.

In her graduate class on Poverty and Policy, Roselius teaches her students to critically evaluate arguments and evidence in support of (or against) various policy recom-mendations to alleviate poverty. Students also discuss the role values play in shaping policies related to welfare, job training, housing, enterprise zones, and education.

"The key word for this class is 'evidence,'" Roselius says. "We ask, 'What evidence is presented? Where are the weaknesses? What rival explanations could explain it?' Only then can we evaluate the effectiveness of policies our value systems lead us to promote."



Volume 25/1996-1997 contents | Duke Policy News Online | Sanford Institute