Laboratory

MaxHel Center

At a Glance Projects Publications Team

Research Area

Methods

Research in the MaxHel Center is supported by advanced methodological tools from several disciplines, including demography, statistics, and genetics research. Center knowledge on these tools is partially based on our in-house expertise and partially on expertise that we can leverage from our institutional and individual collaboration partners. The Max Planck Institute for Demographic Research and the Population Research Unit have an extensive knowledge base on demographic and statistical methods. In projects that require advanced methods and knowledge of genetics research and polygenic score construction, we collaborate extensively with the Institute for Molecular Medicine Finland (FIMM) and its directors Mark Daly and Samuli Ripatti, as well as with experts from the University of Bristol, such as George Davey Smith, who has been pioneering genetic epidemiology.

The Center also focuses on advancing research methods in particular at the intersection of demographic techniques, causal inference, and genetic epidemiology. This intersection is at the core of where progress could facilitate important advances in research on social inequalities in health. A key tool of demographic analysis, the multistate life table, is used mainly as a descriptive method in current research. Attempts to incorporate causal thinking or the so-called counterfactual causal revolution to multistate life-table analysis have been limited. Further, life-table techniques and genetic data have not been integrated. The same is also true for the integration of genetic methods and other methods for longitudinal or causal analysis. Moreover, the combination of machine-learning techniques with these methods to deal with population heterogeneity has not been explored much. We advance research on all these topics, with the ultimate goal of producing both better descriptions of social health inequalities and the causal processes that generate these inequalities.

Our methods research is embedded in an extraordinary data infrastructure that includes detailed generational and kin linkages, genetic information, quasi-natural experiments, longitudinal data spanning over half a century, and virtually no attrition or loss to follow-up. This context allows us to combine the best of both worlds in terms of description (we observe the full population over time, not a sample or cross sections) and causal inference (we can simultaneously explore various fixed effects, instrumental variables, and genetic designs).