Amstat News invited John Phillips—associate commissioner of the Office of Research, Evaluation, and Statistics—to respond to the following questions so readers could learn more about him and the agency he leads.
John W. R. Phillips joined the Social Security Administration as associate commissioner of the Office of Research, Evaluation, and Statistics in February. Prior to joining SSA, he served as chief of the Population and Social Processes Branch of the National Institute on Aging, leading an extramural research program and serving as a federal project scientist for the U.S. Health and Retirement Study. Phillips earned a PhD in economics from Syracuse University and completed a postdoctoral fellowship at the University of Pennsylvania.
What about this position appealed to you?
The position captures the elements that motivated my career choices to date. The Office of Research, Evaluation, and Statistics (ORES) is a federal statistical unit within the Social Security Administration responsible for the production and dissemination of research and data on Social Security programs. I began my federal career 17 years ago as an economist in a division within ORES, so the opportunity to return in a leadership role was an exciting opportunity for me. Over my career at SSA and the National Institutes of Health, I worked to develop research and data that contribute to better understanding the mechanisms that influence the welfare of older Americans. It seems like a great fit.
Describe the top 2–3 priorities you have for ORES.
Two critical priorities for our agency are to educate the public about Social Security programs and to accelerate the use of data-driven decision making. ORES plays important roles in supporting these objectives. Our statisticians and researchers collaborate with our information resources team to produce publications reporting information about important aspects of the program. Trends in applications and benefits, changes in international social insurance policy, and the relationship between earnings and mortality are a few examples. Our data team produces extracts of administrative data to support analysis conducted by federal and academic researchers using protocols intended to manage disclosure risk. My intention is to continue to encourage and publish meritorious research through our intramural and extramural programs, expand our capability to conduct research using administrative records while protecting confidentiality, and build the evidence base to support both policy making and education of the public about our programs.
What do you see as your biggest challenge(s) for ORES?
Changes in technology are providing new opportunities to conduct research and share data while also producing new challenges to protect it. Effective data sharing expands the pool of researchers conducting both novel analyses and replication studies. We need to evaluate new opportunities and partnerships to enhance the quality of and access to research data while enhancing effective protections from inappropriate disclosures. Further, new data management and visualization techniques can fundamentally change the way we organize, share, and use program records for research. Managing our desire to do more to improve data for research in the current budget environment is a big challenge.
What kind of support from the broader statistical community do you look for?
Being a federal statistical unit has benefits, including belonging to a network of other federal statistical units with excellent leadership. For example, I have had the opportunity to meet with the agency representatives from the Interagency Council on Statistical Policy (ICSP). The participants bring a wealth of statistical experience and share many of the same objectives as ORES. I hope to continue to engage groups of statistical experts such as ICSP and the National Academies of Sciences Committee on National Statistics to learn the best strategies to achieve the statistical objectives of ORES.
Prior to your tenure, what do you see as the biggest recent accomplishment of ORES?
ORES has accomplished a great deal in recent times. Our research unit has produced important findings about the social determinants of health, and our grant program has published award-winning research on saving. Our statistical, international, and publication teams combine to produce a significant number of valuable online publications a year. Our data team produces important extracts supporting social research on our programs, such as the administrative data linkage to the Health and Retirement Study. All these accomplishments contribute to our objectives to educate the public about Social Security programs and to accelerate the use of data-driven decision making. That said, a significant accomplishment for ORES would be our relatively recent designation as an official statistical agency by OMB in 2009. The designation both affords the enhanced confidentiality protections to data the agency acquires for exclusively statistical purposes and connects SSA to a valuable network of federal statistical units.
One way to make data available and protect individual identity is via micro-aggregation. Micro-aggregation can be designed into a statistical analysis strategy. Decide on a response variable. Decide on a predictor variable. Select a small number of covariates that you would like to control for and cluster on these variables making many relatively small clusters, micro-aggregation. So long as each cluster is large enough, individual identity is well protected. Now within each cluster test the relationship between response and predictor, say simple linear regression. Now look at if and how the relationship changes from cluster to cluster.
A recent publication makes the claim that the longer you work before retiring the longer you will live. True? Causal? How to test aspects of the claim?
Sharp question: If we moved the retirement age up,would we extend life?