Durham, N.C. – UNC Center for Community Capital research associate Kim Manturuk advised health researchers at a national conference on key issues to consider when selecting research samples to minimize bias in their findings.
Key among them, Manturuk said:
- Researchers should be conscious of how sample selection problems can bias the results of causal models.
- Techniques that match treatment and control observations include propensity score analysis, Mahalinobis metric matching and coarsened exact matching.
- Regardless of the methods used, no statistical technique can overcome poor data and, therefore, selection must also be considered when developing data collection strategies.
Manturuk’s remarks came during a panel discussion at the “Social Determinants of Health Disparities,” co-sponsored by the Research Network on Racial and Ethnic Inequality at Duke’s Social Science Research Institute and the Duke Global Health Institute. The national conference explored factors that influence health disparities between racial and ethnic groups on a national and global scale.
Manturuk’s panel discussed methodological issues in health research. She presented material about statistical approaches to handling selection bias with a focus on sample matching techniques.
The UNC Center for Community Capital is the leading center for research and policy analysis on the transformative power of capital on households and communities in the United States. The center is part of the College of Arts and Sciences at the University of North Carolina at Chapel Hill. Its in-depth analyses help policymakers, advocates and the private sector find sustainable ways to expand economic opportunity to more people, more effectively. For more information, visit www.ccc.unc.edu or call (919) 843-2140.
Topics(s): Other, Research Design