MPC News - Fall 2016
Member Profile: Theresa Osypuk
When your community determines your health
Associate Professor of Epidemiology Theresa Osypuk is using new tools and methods to rethink the way inequality relates to health in the U.S. Using data from the U.S. Department of Housing and Urban Development’s Moving to Opportunity for Fair Housing demonstration project, Osypuk is revealing that the health gains in families who have relocated away from poverty vary.
Theresa Osypuk’s interest in inequality began early in her career, when she began working for the Ad Council. As she and her colleagues were creating the public service announcements (PSAs) to communicate and persuade Americans to promote better health, she grew frustrated. “By the time that we were implementing these campaigns, the research was old.” Adding to the frustration was the competition to reach the audience for the advertising. The United States, unlike in other countries in Europe, relies on the benevolence of the media companies in order to get those messages out. “The networks donate the time to air public service messages, so when you compare PSA airtime with that of paid advertising, it’s small, and therefore difficult to reach enough people for effective campaigns,” Osypuk explained this summer. PSAs are also most effective to promote attitudinal or behavioral change at the individual level; but such changes may not be the most effective way to improve health at the population level.
These realizations, along with the desire to do more to shift the levels of inequality in America, led Osypuk to change the angle of her work, from health communications to research and policy change. She is critical of the national view of health in the U.S., that simply visiting your health practitioner will keep you in good health. “Even though health care is critical for treating disease once it occurs, lack of healthcare is not the most important factor causing health in the first place,” she explains. “The most important causes of health overall are the everyday conditions to which each of us is exposed over our lives, and these social determinants of health are largely outside of the health care sector.”
Osypuk argues that changing the social determinants of health -- the conditions in which we live, including housing, neighborhood, education, employment, family context – is necessary to shift population health, since these are fundamental causes of health and health disparities.
Osypuk’s work uses the findings of previous research to enhance human health and well being. One her projects examines the Moving to Opportunity (MTO) housing mobility program. In 1992 the federal government conducted an experiment in an effort to determine whether or not families moving from high-poverty neighborhoods to lower-poverty neighborhoods would improve their economic and employment prospects. The study enrolled 4,600 volunteer low-income predominantly racial minority families with children headed by single mothers living in high-poverty public housing developments.
“The goal was to reverse poverty and promote economic autonomy,” Osypuk explains, “and what they found was that the MTO program produced few effects on these primary study measures of reducing poverty or promoting economic autonomy in those families. Although health was not an original outcome domain that program planners anticipated would be affected by the MTO program, it was added to the quantitative evaluation after MTO participants discussed health benefits of the program in qualitative research. The quantitative evaluation then included health outcomes for all participants and found large effects on health for those families, particularly in mental health of mothers and teenage girls.”
Interestingly, the MTO program of moving to lower-poverty neighborhoods into private rental housing had the opposite effect for boys. “The mental health of older boys was surprisingly harmed by the move,” she elaborates. This finding has spurned interest in work on gender specific inequality studies. “We don't know why this program had opposite effects for boys and girls,” Osypuk says. “It could be that girls are at an increased risk of sexual harassment in high-poverty neighborhoods, so moving out of high poverty neighborhoods away from that sexual risk improved the mental health of girls. But these mechanisms seem gendered, and the reduction in sexual risk might be less relevant for boys. We know that children are socialized very young to cope with and live in high -poverty neighborhoods, and these behaviors that boys have learned, may be maladaptive in lower poverty areas they move to. It may also be that in those single-parent families, when boys moved away, they lost their male mentors.” She and her colleagues are testing some of these potential mechanisms in the NIH grants that she is leading on this project.
Osypuk’s research takes data from that original study and attempts to understand what is happening in the health of those individuals, so that policy can better address positive health interventions. In a recent paper in the American Journal of Public Health, Osypuk and her colleagues examined the MTO data to understand whether certain adolescent subgroups differentially benefited from the MTO programs, in terms of their mental health: psychological distress and behavior problems. Understanding the varying effects of the program may help to improve the program in the future, in terms of supplying additional services, or to inform the eligibility for the program.
Osypuk and her colleagues found that the city in which the program was implemented, along with gender and age, affected the results as the strongest modifiers of the programs.
Using this secondary data from the MTO, Osypuk and her colleagues get to integrate all of the research innovations that epidemiologists have made over the past 10-20 years in causal methods to strengthen the inferences that her team makes.
“The experimental design of MTO gives us that comparison group for the treatment group, we have an understanding of what our group would have experienced (if not for the treatment). Because of the existence of this alternate universe, we can say more strongly that this treatment actually caused these outcomes.”
To get these results Osypuk and her colleagues are applying machine learning techniques that have not previously been implemented much in fields like demography, public health and epidemiology. She explains, “It’s a different way of examining the patterns in the data, and the first time the method has been applied to preserve the intention to treat principle to preserve the experimental design and its strong causal inferences.”
The implications for policymakers from the research is clear on one note, a one-time or one-size adjustment will not fix all problems for deeply-struggling families. Osypuk elaborates, “Our research is finding that some subgroups need more assistance to benefit from housing relocation, particularly those families with health problems or those that experience violent crime. We also need to provide more help to adolescent boys, to help them succeed from housing relocation, by providing a housing intervention, but also services from the education and health sectors. Investment to improve health must occur across sectors, because social factors such as housing, education, family context, and criminal activity are such strong predictors of health. Therefore policies addressing these factors may be powerful levers to improve population health and reduce health disparities.”
Over the past 16 years, MPC has developed the world's largest integrated population databases, spanning 100 countries, 300 years, and billions of records. These data are an indispensable and irreplaceable element of the world's statistical heritage, and caring for them is a serious responsibility. It is much easier to raise money to create new databases than to maintain existing ones. Accordingly, for the past several years leaders of the MPC data projects have been focusing on sustainability planning.
To manage long-run sustainability, we have reorganized our administrative structure. We established the IPUMS Center for Data Integration (IPUMS Center) to focus on the sustainability issue. Vice President Herman established a new Institute for Social Research and Data Innovation (ISRDI), which will oversee both the IPUMS Center and MPC, along with the Minnesota Research Data Center (MnRDC). The Vice President appointed me to serve as the first Director of ISRDI, and I am also serving as Interim Director of the IPUMS Center.
MPC will continue to be a University-wide interdisciplinary center for population research. We will continue to bring together members to foster intellectual interaction and research collaborations, provide research facilities, support proposal development, and provide training in population research methods. In short, we will continue to offer the same services to our members as we have for the past 16 years.
Rob Warren has been appointed as the new Director of MPC. Rob has been an MPC Faculty Member since coming to the University of Minnesota in 2002, and he is currently PI or co-PI on several major grants, including IPUMS CPS and an NIH R01 to link the 1940 U.S. Census Data to five modern surveys of health and aging. Rob has served as MPC Training Director for the past five years, and I am confident that MPC will continue to grow and thrive under his direction.
Fall 2016 Seminar Series Schedule
MPC seminars are held from 12:15 to 1:15 PM on Mondays in the MPC Seminar Room (50 Willey Hall)
Fall 2016 MPC Workshop Schedule
MPC workshops are held from 12:15 to 1:15 PM on Fridays in the MPC Seminar Room (50 Willey Hall)
Free Data Training Workshop Schedule
Workshops are free, open to the public, and held in computer classrooms on the campus of the University of Minnesota on Fridays from 12:30-2:30 p.m. Registration is required.
Get to know the employees at the MPC. In this issue, we learn more the MPC IT team. Due to the large size of this internal group, we will be selecting a few employees at a time to profile over the next year. Below is our first crop of colleagues.
Technology at the MPC
“Good IT is invisible,” says MPC IT Core director Fran Fabrizio. “You want the users to have the idea that it’s a magic black box.” Though the intent is for the technology behind IPUMS and the other MPC data tools to seem effortless, Fabrizio understands the extent of the human work goes that goes into producing good technology. Getting 2.6 terabytes of data out to users each week requires no small amount of technology behind it.
“Simply put, we do two things, we get data ready for delivery and we build sites and tools that get the data to the user.” That simple task requires the work of 25 people to deliver data to 70,000 global users each week. The IT unit at the MPC is divided into three groups: data delivery, data production, and operations, with Fabrizio at the lead.
It’s a huge IT staff for a population center, but Fabrizio considers it a skeleton crew when compared to his private industry peers. “We try to be as efficient as possible so that we can get the most out of our grant money.”
The IT core recently made the investment in user experience. “It’s a new focus for us,” Fabrizio notes. The user experience work has been primarily focused on the Terra Populus website. “We hired our first dedicated UX designer, Alex McWhinnie, last year to help us have a user-centered design process when we build our sites. We want to think through issues with a usability perspective, and get feedback about our work from user testing so that we can be data-driven about design.”
Fabrizio and his group are looking into the future of data access, as well as the design of the sites themselves. “Right now we are giving a lot of consideration into how we get data to our users,” he says. The MPC data websites were first launched in the 1990s as a way to get data to users quickly. While this was a significant improvement from the previous method, filling out a form and waiting for data to arrive in the mail, as technology has changed we now have the potential for alternate delivery methods. The group is now researching new methods to get data to users, including giving IPUMS users open access to data and metadata via API. Fabrizio says, “Our current method, via the websites, will continue to be the primary way we deliver data, but it can be constraining to some of our users, particularly those who want to interact with our data via their own software. We need to think about how we can get that data out to them more easily, more efficiently. There are a lot of technically difficult issues, and it’s going to take us a while to figure this out.”
The sheer amount of data that the MPC is on track to release by 2020 -- 2 billion records -- is spurring the IT group’s progress. “If you consider how much the size of the data has grown over the years, it’s easy to understand how scalability plays a role in the future of the MPC,” Fabrizio explains, “Our current tools were created with assumptions of data sizes that we have long grown past.”
“It’s an interesting situation to be in,” he continues. “We are happy that there is a continuous need for our products, and we are focused on finding faster, easier and more friendly ways to continue our mission -- getting data to users.”
Fran Fabrizio has been the MPC’s IT Director since 2012 and is a 20-year veteran of the industry. After a couple of software development roles in the private sector, Fran moved to higher ed IT in 2003. He has the great fortune of working with an incredible team at the MPC, where he splits his time between management and technical duties, continuing to write software regularly. After work, he spends his time with his wife and son, riding his bike, and following his favorite sports teams. Fifty weeks of the year are also spent anticipating and planning for the two weeks of the Minnesota State Fair, which he will attend at least five times annually.
Lap Huynh works in IT supporting the work of the Historical Census Data team. Lap joined the MPC in 2006 after working at Boston Scientific for 10 years as a process development technician. Lap’s colleagues at the MPC keep him busy during the day, and when he heads out of the office he works at his second job as a personal driver for his kids, where he is responsible for getting them to all of their extracurricular activities.
IT Operations Manager
Willy Lee is part of the IT operations team. For the past year and a half, Willy has been taking care of the servers and the services that the IT group depends upon at the MPC. His 20-year tenure in web technology predates the existence of Google. When he isn’t stoking the servers at the MPC, Willy enjoys cooking, trying new restaurants, and cycling.
Team Lead, Microdata Projects
Marcus Peterson is an MPC IT veteran. Marcus started with the MPC in 2002 writing data reformatting programs for the first IPUMS-CPS. Over the years he has worked on the original IPUMS-USA website and helped develop the first IPUMS-CPS, IPUMS-International, and NAPP sites. After noticing many commonalities in the metadata structure and website functionality across these microdata projects, Marcus developed a general codebase in Java (which has since been ported to Ruby) for building IPUMS microdata web applications. These days he manages a team of developers, with whom he provides support and builds new features for the MPC's ever-growing list of IPUMS projects. As a new dad, Marcus doesn’t have much time for movies and concerts, but he makes time to run (and sometimes race).
User Experience Designer
Alex McWhinnie joined the MPC in 2015 as the Center’s first User Experience Designer (UX). As an advocate for the users of TerraPop and other MPC data projects, Alex designs user-centric workflows that help shape how those users access MPC data. His day is typically spent doing sketching, prototyping, or user testing. Prior to joining the MPC, Alex worked at MIT in Boston, taught UX at Suffolk University, and raised chickens and bees on the weekends. Now, his spare time is focused on cycling and smashing Mario Kart records.
Jacob Wellington has been at the MPC for just over two years. He joined the MPC in 2014 as a Ruby on Rails developer for the NHGIS project. He previously worked on marketing teams building websites, and is pleased to be working for the University. If Jacob had learned not to play with fire, he may never have taken up juggling flaming torches with his wife. There may be juggling clubs at his desk right now for a (flame-free) demonstration.