Daniel J. Benjamin, Ph.D. - US grants

DESCRIPTION (provided by applicant): Older Americans face the dilemma of complex financial and health decisions combined with a high incidence of cognitive decline with age. One of the important policy proposals for dealing with the effects of cognitive limitations on economic decision- making is to set up institutions, such as automatic 401(k)'s, which default people in to good choices. The difficulty with this approach is that knowing what a good choice is for an individual depends crucially on knowing her preferences. In particular, for retirement saving and investment choices, it is crucial to know the values of three preference parameters: risk aversion, time preference and the elasticity of intertemporal substitution (EIS). Even if policy ignores individual differences and creates one version of the default, a good choice for everyone in the group still depends on at least knowing the distribution of preferences in that group. This project will extend our understanding and further our identification of these preferences and how the expression of these preferences is affected by cognitive abilities. The specific aims are to: (1) Understand at a high level of analytic detail inconsistencies in measures of risk aversion and begin to do so for time preferences and the elasticity of intertemporal substitution; (2) Gauge the extent to which cognitive limitations (as opposed to non-standard preferences) are responsible for inconsistencies in different measures of risk preferences, and also time preferences and the elasticity of intertemporal substitution; and (3) Develop methods to overcome the difficulties cognitive limitations create in measuring these preferences-in particular by eliciting reasoned preferences rather than using untutored preferences, and implement these methods on a large sample of adults in the prime years for retirement saving. The more general aim is to develop techniques for identifying reasoned preferences for the even larger set of preferences that matter for difficult decisions that older Americans face, even beyond the retirement saving decision. PUBLIC HEALTH RELEVANCE: The appropriate financial or health decision for an individual often depends in part on the individual's preferences, such as willingness to take risks. This project will study how cognitive abilities interact with measures of preferences that are relevant for financial and health decision-making, and will develop improved measures of these preferences that can be used to help people make the decisions that are best for them.

? DESCRIPTION (provided by applicant): The Health and Retirement Study (HRS) has the potential to become a critical resource in social-science genetics in general, and the behavioral genomics of aging in particular, due to its relatively large sample size, its rich longitudinal behavioral measures, and the availability of dense genomic data for approximately 13,000 older Americans, with data on 7,000 more on the way. The proposed research will use the HRS's rich phenotypic, genetic, and environmental data to pursue two complementary strategies. One is to incorporate HRS data into large consortium meta-analyses of behavioral phenotypes conducted under the auspices of the Social Science Genetic Association Consortium (SSGAC), which the applicants organize. The second strategy is to use the HRS data to test specific hypotheses arising from the consortium's findings and to shed light on the genetic architecture-i.e., the joint distribution of genetic effect sizes and allele frequencies-of the rich set of behavioral phenotypes measured in the HRS. Our general aim is to use the phenotypic, genetic, and environmental data from the HRS to significantly advance understanding of behavioral genomics in general, and of the economic behavior, health, and well- being of older Americans in particular. Among the goals of our proposal are: (a) discoveries of specific genetic polymorphisms that are associated with important behavioral outcomes, psychological characteristics, and economic preferences; (b) analysis of biological pathways that underlie these associations; (c) development of polygenic scores (indexes of many polymorphisms) that, when constructed with weights estimated in large samples, will have substantial predictive power for behavioral phenotypes; (d) identification of behavioral mechanisms (i.e., endophenotypes) that mediate associations with specific polymorphisms and polygenic scores; (e) analysis of the genetic architecture of a range of phenotypes measured in the HRS; and (f) examination of gene-environment interactions.

? DESCRIPTION (provided by applicant): When science is used to inform practice, careful measurement based on clear conceptualization is the beginning of wisdom. In particular, gauging how well people are doing is basic to figuring out whether efforts to improve people's lives are succeeding. Cost-benefit analyses at the micro level and GDP at the aggregate level face serious challenges in dealing with non-market goods. (Both cost-benefit analyses and GDP account for some non-market goods but miss many others.) This matters particularly in monitoring the well-being of older individuals because non-market goods, such as health and the pleasure from social relationships and leisure-time activities, become especially important at later ages. The overall objective of this proposal is to lay a stronger foundation for extensive us of subjective well-being (SWB) data to monitor changes in overall health and well-being. While the proposed research focuses broadly on understanding SWB data and how it can best be used, the cornerstones of this proposal are (1) developing new sources of data on multidimensional SWB, and (2) developing a set of methods and the necessary complementary data to aggregate SWB data across different aspects of well-being into an overall measure of well-being. This proposal builds on a substantial body of prior work, as well as on many of the conclusions and recommendations of the report of the NIA's Panel on Measuring Subjective Well-Being in a Policy-Relevant Framework (Stone and Mackie, 2013). Phase 1 (Years 1-2) of the proposed research is focused on understanding and improving SWB questions (especially in health dimensions) and methods for aggregating the responses into overall measures. Phase 2 (Years 2-5) is focused on creating and developing web panels, with two waves of data on the levels of a large number of dimensions of well-being. The purpose of Phase 1 is to make the quality of the data collection in Phase 2-and the subsequent analysis-as high as possible. The many SWB measures that will be collected on the Understanding America Study (UAS) will include versions of all the commonly used SWB questions currently carried on major surveys, making direct comparisons possible. The new data resources will be from the UAS, a relatively new web panel founded by Arie Kapteyn at USC, and other data sources, including a New Zealand well-being survey being planned by the government of New Zealand. The data would be publicly available. An important virtue of the UAS and similar surveys is that, without needing a prior connection to the UAS, any serious researcher can easily arrange to have additional complementary data collected, at a reasonable cost in terms of additional funding.

The Social Science Genetic Association Consortium (SSGAC) is a research network that provides a platform for large-scale, interdisciplinary collaborations on genome-wide association studies of behavioral phenotypes. Because credible results often require large discovery samples, a primary function of the SSGAC is to conduct genome-wide association studies conducted in different datasets and then rigorously combine the results. To this end, the SSGAC also promotes the collection of harmonized and well-measured phenotypes across datasets and the development of new methodological tools. The SSGAC is committed to disseminating its research findings, as well as accompanying software and manuals, to the research community. The overarching goal of this proposal is to maintain and develop the core infrastructure of the SSGAC. The resources produced by the SSGAC will continue to serve the field of behavioral genomics in general, and the behavioral genomics of aging in particular. The Specific Aims are: ? Conduct genome-wide association studies of health and aging-relevant behavioral phenotypes in much larger samples that have now become available. Currently planned studies include analyses of educational attainment, general risk tolerance, and life satisfaction, all in unprecedentedly large samples. In addition, we will undertake large-scale studies of additional phenotypes as data becomes available. ? Facilitate researchers' use of SSGAC results by (a) maintaining published results on the SSGAC's website, and (b) providing participating datasets with certain unpublished results, in accordance with the SSGAC's Data Availability Policy. ? Develop a ?Repository of Polygenic Scores,? which are variables constructed from genetic data that are useful to researchers. We will set up the Repository to provide polygenic scores to participating datasets. ? Update and expand the Repository as new genetic-association results become available. We will also welcome any new datasets who wish to participate. Each release will be accompanied by documentation that clearly describes methods used and the underlying data. ? Create and disseminate software that will allow datasets to create their own polygenic scores, using the same harmonized methodology as the Repository but without needing to share their genetic data with the SSGAC. The software will be publicly available on a GitHub repository featuring a Q&A forum where potential users can ask questions. We will also prepare user-friendly software manual.

Project Summary/Abstract This proposal, ?Genome-Wide Analyses of Health and Well-Being Phenotypes,? is a competing renewal for a currently funded, three-year R01. The current R01 is focused on advancing the research of the Social Science Genetic Association Consortium (SSGAC), an interdisciplinary collaboration for conducting large-scale genetic studies of behavioral phenotypes, which is directed by three of the applicants. All of the aims of the R01 were achieved. Its results are being widely used in health and aging-related research in social-science genetics, medical genetics, and epidemiology. This competing renewal proposes to continue the work of the SSGAC. In brief, we propose to: ? Conduct genetic-association studies of health and aging-relevant behavioral phenotypes in much larger samples that have now become available. We will complete our study of dietary intake begun under the current R01. In addition, we will undertake large-scale studies of additional phenotypes, including physical activity and self-reported general health. Each of these projects will identify genetic variants associated with the phenotype, analyze biological pathways that underlie these associations, and construct polygenic scores (indexes of many genetic variants) that can have substantial predictive power for the phenotype. ? Develop a more powerful method for joint analysis of multiple phenotypes, which will be able to (a) estimate the fraction of genetic variants associated with some set of phenotypes but not others, and (b) identify genetic variants likely to be associated with some set of phenotypes but not others. The results of applying the method will help disentangle different mechanisms by which the genetic variants matter for the phenotypes and will enable the construction of more predictive polygenic scores for each phenotype. ? Apply this method to shed light on the shared and unique genetic pathways that influence educational attainment and (late-onset) Alzheimer's disease. This analysis will: (a) identify many new genetic variants associated with Alzheimer's disease; (b) generate more predictive polygenic scores for Alzheimer's disease, facilitating earlier diagnosis and treatment; (c) shed light on hypotheses related to the underlying mechanisms driving the genetic relationship between Alzheimer's disease and educational attainment; and (d) enable biological annotation of genetic variants identified to affect Alzheimer's risk but not educational attainment and which thus may operate through more direct biological pathways on disease risk.