Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 1
Biosocial Surveys Introduction James W. Vaupel, Kenneth W. Wachter, and Maxine Weinstein New kids on the block! This volume spreads the word. Population-based sample surveys that combine demographic, social, and behavioral data with biological indicators have arrived in town. The subtitle of the precursor to this volume, Cells and Surveys, was the question “Should biological measures be included in social science research?” In practice, that question seems to be already answered: yes. Social surveys that include the collection of biological data have proliferated since that volume’s publication in 2001. Are these new studies going to be our friends? Speculation gives way to assessment, as publications emerge from the dozen or more large-scale social surveys that have moved into collecting biological measurements and materials ranging from grip strength to DNA. Unlike biomedical surveys and clinical studies, always rich in biological indicators, these social surveys probe demographic characteristics, economic and health-related behaviors, resources, constraints, and life-course transitions in depth. Consequently, they hold out promise for shining a spotlight on interactions between biological factors and “environment” in its personalized complexity. They hold out promise for distinguishing causal pathways that turn out only statistically significant in experiments and trials from causal pathways that have real “oomph” in the squishy, messy, diverse, adapting world of whole populations. But along with a chorus of greeting are glances of doubt. These new arrivals on the social science scene bring out feelings of caution among many, especially around analysis of genetic indicators in the study of
OCR for page 2
Biosocial Surveys complex social traits. As practical challenges to the inclusion of biological indicators are mastered and as technology rushes ahead, questions of meaning call for ongoing debate. Like the earlier one, this volume was sponsored by the Behavioral and Social Research Division of the National Institute on Aging and put together under the auspices of the Committee on Population of the National Research Council. Cells and Surveys was influential, but not because the book caused social scientists to add biological measures to their surveys. Doing that was an idea whose time had come. Rather, Cells and Surveys was influential because it provided an authoritative overview of the kinds of biological information that could be collected and how and what the problems and pitfalls were. In this volume the focus shifts to what has been and can be learned. The authors of most of the chapters have firsthand experience with incorporating biomarkers into social science research and have followed the rapid development of the endeavor. They have thought long and hard about the critical issues. Although there are still many open questions, the time seems ripe for taking stock, not for advocacy but for aspiration to mature judiciousness and nuanced wisdom. The book emerges from a series of discussions and interchanges that started in 2005 and ended in mid-2007. A workshop, held in June 2006, enabled many of the authors to exchange information and perspectives. Participants in the workshop considered (and the contributions to the book reflect) a few fundamental questions: What has been learned from what has already been done? What is the place of genetic information in social research? What new concepts and methods are being developed or need to be developed? Biodemography (National Research Council, 1997; Carey and Vaupel, 2005) and its related disciplines have expanded rapidly; one book cannot begin to provide a comprehensive summary of its findings and challenges. We had to pick and choose. Our focus in this volume is on the inclusion of biomarkers in social science surveys: what does (or should) one include and how does (or can) one use the data. The volume grew as we proceeded; in particular, we found ourselves asking more questions about the collection and analysis of genetic information. We also made a conscious choice not to solicit manuscripts on some subjects: notably absent are chapters that specifically focus on ethics. This omission is not because we felt the concerns were unimportant, but because the enormity of the subject made it unfeasible to include. However, both explicitly and implicitly, almost every chapter discusses concerns relating to privacy, informed consent, and the appropriate treatment of study participants and their information. Hunting for genes that determine health or behavior is not the pre-
OCR for page 3
Biosocial Surveys occupation of the authors of these chapters. This may surprise readers who follow the news about science but who have not kept up with social science research that includes biological indicators. Most of the research being carried out in conjunction with such surveys does not even involve DNA: the biological indicators pertain to grip strength, pulmonary functioning, clinical measurements of various substances in blood, saliva, or urine, blood pressure, heart rate variability, weight and height, perceived age, and various other physiological and anthropometric measures of risk factors, exposures, and health outcomes. It is sometimes argued that scientists should wait for some more advanced technology before trying to discover genes that influence some trait. The darling of the moment is genome-wide association studies, an approach that is powerful and rapidly becoming less expensive. This argument, however, is largely irrelevant for social science research that incorporates biological markers. Some social science surveys have collected DNA, but the intent has not been to discover genes that determine health or behavior. The genetic information that is collected generally pertains to genes that have already been discovered to have (or reportedly have) important effects. Social scientists can verify that the effect is indeed important: this is what Harald Göring cogently argues is a more appropriate task for social surveys than gene discovery (Chapter 11), and this is what Kaare Christensen and colleagues report as one of the many uses of data from surveys of elderly Danish twins, nonagenarians, and centenarians (Chapter 1). Social scientists can also use the genetic information to better estimate environmental and behavioral effects on health: George Davey Smith and Shah Ebrahim describe an ingenious method to exploit genetic variants to make causal inferences about environmental variables (Chapter 16). Other chapters describe how knowledge about a person’s genotype can be used to study how genes and nongenetic variables interact to produce behavioral and health outcomes. Accounting for genetic variation among individuals permits more accurate estimation of how environmental and behavioral factors influence health and longevity. As summarized by David Abrams (quoted by Davey Smith and Ebrahim), “The more we learn about genes, the more we see how important environment and lifestyle really are.” These critically important areas of research should not be delayed because better methods of gene discovery are being developed. A recurring question throughout the volume pertains to the utility of DNA markers for survey-based social and behavioral analysis. Skeptics predominate over enthusiasts. It seems appropriate, however, to include here one version of an enthusiast’s perspective. The key phrase in this discussion is “complex traits.” Many contributors explain why it is a mistake to look for genes or for simple genetic determinants of complex traits.
OCR for page 4
Biosocial Surveys Cautiously put—and most contributors put it cautiously—this argument is a counterbalance to the extravagance of the press and of scientists seeking celebrity billing. The gene for financial success, for tennis stardom, for obesity, for inanity, or for extroversion is not a target for reputable research. There is a continuing need for social science to test and bury false claims. Vehemently put, however, this kind of argument about complex traits can amount to a radical critique. Suppose the decisions and outcomes now coming onto the explanatory horizon of social science are driven by large assemblages of interacting factors, each with a hunk of its own variation. Then even the substantial sample sizes of existing surveys will be too small to pin down the proliferating combinations. Multiway tables with too many multiways have too many cells to estimate. Nonlinear functional relationships with too many arguments are not identifiable with modest batteries of questions. This argument sees the enterprise of survey research coming up against its limits. Two modes of complexity need to be kept quite separate. One is complexity in the basic units that act as determinants. Single-nucleotide polymorphisms or single genes or epistatic pairs or triples or regulators or promoters may none of them be the right units to be studying. But there may still be some sort of complex units, gene networks, regulatory feedback systems—something else, perhaps as yet undiscovered, for which relevant causal variation can be specified by a modest number of bits of information. Low-dimensional structure may be hiding in high-dimensional systems. This kind of complexity makes the problems of social scientists hard but soluble. Progress in genetics should translate into progress in behavioral analysis. The other mode of complexity is the one implicated in the radical critique: many interlinked factors, each with variation of its own, each only sometimes a rate-limiting factor, each only sometimes a catalyst or inhibitor in feedback loops. This kind of complexity is akin to the complexity of ordinary daily life. Relevant variation irreducibly requires large numbers of bits of information to specify. If such complexity is the rule in the behaviorally relevant structure of the genome, prospects for using genetic indicators from social surveys would seem daunting. There are, however, countervailing reasons for encouragement. A complex structure with enough interacting but partly independently varying elements can often be modeled by a random system. In such a system, laws of large numbers often operate, and many detailed interactions cancel out. Physicists successfully model complex particle systems with random matrices whose spectra obey simple, discoverable laws. Genetic influences that figure importantly in trends or in population dif-
OCR for page 5
Biosocial Surveys ferentials may prove to be accessible to research, without the necessity of fully untangling the underlying complex causal structure. We go to the beach and ask, “Why does this wave surge up far enough to wash away our sandcastle?” Hopeless. But we could ask, and answer, the other question, “Why do waves surge up far enough to wash away our sandcastle?” Pursuing knowledge of causal processes and pathways is important, but usable knowledge depends ultimately on finding something simple in the picture. The goal is not a secular theology of predestination. We are not aspiring to understand the full determinants of complex traits or tell out the person-specific reasons for each person’s path to troubled or successful aging. Survey research has to be a gamble on simplicity, no less so as we start to try to take advantage of our first samples of DNA. The kind of complexity that might defeat social science research is also a kind of complexity that might have defeated natural selection. The genomic information that mattered for behavior and reproductive success over evolutionary time should not have been so contingent that selective pressure could not impinge on it. Environments matter, and it seems plausible that some good part of existing functional genetic variation persists because variation has been adaptive in variable environments. Despite the complexity of complex traits, what natural selection manages to notice might be out there for us to find. Why hesitate to look? So it is that some social scientists hope that genes might be discovered that influence behavior—and that they can play a role in their discovery. Parts of Chapter 15 by Daniel Benjamin and colleagues point in this direction, albeit with cautious recognition of problems and difficulties. In particular, the authors hypothesize that genes with known effects on neurotransmission pathways or on memory (and some other cognitive functions and abilities) might influence labor force participation, wealth, and other economic behaviors and outcomes. The four chapters in Part II provide detailed explanations of how problematic and questionable such a quest is. The reader can make a judgment about whether Benjamin et al. are brave or reckless. The more general point is clear: most social science research using biological information does not pertain to genes, and the research that does include genes largely focuses on using this information to better assess nongenetic influences. WHAT HAVE WE LEARNED SO FAR? Part I, roughly “What we have learned from what we have already done,” includes chapters on six different research programs, a chapter that reviews other relevant surveys with an emphasis on biological indicators, and a chapter that more generally reviews the theory and practice
OCR for page 6
Biosocial Surveys of biomarkers in social science research on health and aging. The first three chapters cover large, well-established studies of health and aging that have long included biological information. Kaare Christensen and colleagues describe the main research directions and findings of the Danish studies of elderly twins, nonagenarians, and centenarians. Michael Marmot and Andrew Steptoe provide a review of two important English surveys, Whitehall II and ELSA (the English Longitudinal Study of Ageing). Jack Chang and colleagues describe and discuss the Taiwan biomarker project. These three chapters provide a wealth of details and insights based on deep experience. The directors of the Health and Retirement Study, one of the largest and most widely used surveys of aging, were considering the addition of biomarkers at the time of Cells and Surveys (Weinstein and Willis, 2001). The study is cautiously adding the collection of a few markers. The reasons for caution and lessons from recent experience are discussed by David Weir. Robert Wallace calls social scientists’ attention to the Women’s Health Initiative, a very large, intricate set of clinical trials. He describes the difficulties in managing such an endeavor: the complexity of the nested studies and the extent of the collaborative networks constitute a Petri dish for nurturing administrative complications. Wallace emphasizes the potential value of the Women’s Health Initiative for social science research, a potential that has not yet been realized. Duncan Thomas and Elizabeth Frankenberg demonstrate that it is indeed possible for social scientists to use clinical trials to learn about behaviors and social outcomes. The experiment involved adults in Indonesia who were randomly assigned to two groups, one receiving iron supplements and the other a placebo. The health of those receiving the iron supplements improved. Thomas and Frankenberg studied how this affected workers’ productivity and time allocations. Not every significant study is or could be represented in this volume—a testament to how widespread the collection of biomarkers has become. Among others, important studies that do not have their own chapters include MIDUS (National Survey of Midlife Development in the United States), the Women’s Health and Aging Study, SHARE (Study of Health and Retirement in Europe), and SAHR (Stress, Aging, and Health in Russia). The chapters by Jennifer Harris, Tara Gruenewald, and Teresa Seeman and by Douglas Ewbank provide overviews of some of these other studies and their connections with the studies covered in the earlier chapters. Harris and colleagues critically review biomarker research from community and population-based studies, with an emphasis on physiological parameters and genes. Ewbank reviews theory and practice, with
OCR for page 7
Biosocial Surveys an emphasis on studies of social behaviors and environments. He concentrates on studies of the effects of chronic social and psychological stress. Taken together, the eight chapters on what we have learned so far include discussion of findings that have been made possible by the combination of biological and social data. They also include candid commentaries on the logistical problems that the researchers encountered. These hurdles include consent processes; challenges in the field; difficulties in storing, archiving, and reusing specimens; and management of collaborations across institutions. Ewbank reminds us that it is probably too early to draw generalizations from the studies. A corollary to this caution is that however great the temptation, it is also too early to restrict ourselves to “cookie cutter” surveys that simply replicate the biomarker collections of previous studies. Still, the broad conclusion is clear: the studies to date provide new ways of understanding the interactions and joint effects of behavior, the social environment, and physiology on health. Of particular interest for future understanding of these effects is the growing number of studies that are being done outside the United States. For example, as noted by Marmot and Steptoe, the collection of biological and social data in both England and the United States has contributed to understanding of the higher rates of morbidity in the United States. The chapters by Harris et al. and Chang et al. note differences between results from the MacArthur study in the United States and the Taiwan biomarker project; these variations have the potential to open a window onto macro-level influences on the links between social experience and health. As additional studies are added to the “arsenal”—nascent and ongoing studies, for example, in Japan, Korea, mainland China, Mexico, and Russia—there will be opportunity to explore these factors. WHAT ABOUT GENETICS IN SOCIAL SCIENCE RESEARCH? Traditionally, social scientists have studied people’s behavioral, psychological and demographic characteristics, how these characteristics interact, and how they are affected by the social environment. More recently, social scientists have begun to collect and analyze data on individuals’ health and on physiological and morphological factors that affect health. In the past few years, it has become possible not only to study phenotypes—observable traits of individuals—but also to study genotypes, the genetic makeup of individuals. This ability presents a new set of scientific opportunities, practical difficulties, and ethical challenges. Part II offers an introduction to these issues. We expect that this primer will be of particular interest and value to social scientists. The chapter by Mary Jane West-Eberhard provides a “reader’s guide
OCR for page 8
Biosocial Surveys for how to relate genes to phenotypic traits … to better interpret research results and public discussions…” She explains why it is deeply misleading to claim that there is a gene for a complex trait, such as a gene for obesity or laughter. She stresses the role of the environment in shaping the action of genes and the enormous intricacy of the genetic architecture of complex traits, such as longevity or intelligence. George Vogler and Gerald McClearn provide a complementary perspective on the limited opportunities, subtle complexities, and numerous pitfalls in analyzing genetic markers in social science research. They dismiss the outmoded nature versus nurture distinction and outline various types of correlations and interactions among genetic and environmental factors. Harald Göring covers many of the same issues of polygenes and complicated gene-environment interactions and correlations. He does so, however, from a different perspective: he focuses on explaining why large-scale social surveys are not well suited to the discovery of genes that influence complex traits related to behavior or health. Social surveys can, however, sometimes play an important role—not in gene discovery but in the validation of purported discoveries. Many of the published discoveries are false positives. Many do not accurately estimate effect sizes. And many are based on study of special populations that are not representative of the general population. Large social surveys can provide additional information to address these deficiencies. The fourth chapter on genetics in social science research, by Kenneth Weiss, also emphasizes that many genes may influence a trait and that those genes interact with each other and with the environment in extremely complex ways. Furthermore, “genes typically harbor tens to hundreds or more alleles,” genetic variants that differ from individual to individual. Somatic mutations (that is, changes in genes over the course of life) and “stochasticities of countless sorts” further complicate analysis. Weiss also raises some uncomfortable and far-reaching ethical concerns. His questions are particularly important for demography, economics, and other domains of social science that have strong influences on policy formulation. The hazards of entering into an arena with a history of political and social abuse should not be ignored. Understanding the profound complexities of analyzing genes in social science research is enhanced by the complementary perspectives of the four chapters in this section. The chapters are a tutorial for social scientists, valuable for naïve researchers intrigued by publicity about obesity or longevity “genes,” and even more valuable for experienced social science researchers who have recognized the usefulness of incorporating (nongenetic) biological indicators in surveys and are now considering studying genes.
OCR for page 9
Biosocial Surveys WHAT’S NEXT? Part III explores concepts, methods, and modeling strategies that are emerging as biomarkers are increasingly collected. Although incorporation of biological indicators in social surveys has burgeoned over the past several years, the field is still young. New kinds of biomarkers and new ways of thinking about rich combinations of social and biological data are needed. Stacy Tessler Lindau and Thomas McDade provide an overview of minimally invasive and innovative methods for collecting biological information in population-based research. Some of the strategies they discuss are being implemented. For instance, information from MRIs is being collected by a subproject headed by Richie Davidson as part of the MIDUS II study led by Carol Ryff and Burt Singer. Extensive electrocardiogram data from Holter monitoring of heart patterns over a 24-hour period are being gathered from a sample of hundreds of older individuals in Moscow as part of the study of Stress, Aging, and Health in Russia (SAHR) under the direction of Maria Shkolnikova. The use of Holter monitoring represents an innovative application of a well-established method to collect information on the dynamics of physiological variation. Most of the biological indicators used to date are based on a single measurement at a single time. Multiple measurements two or more years apart are sometimes taken in longitudinal surveys, and these data can be valuable in reducing measurement error as well as in tracking trends. Temporal patterns of variation over hours or days in an individual’s physiology are often much more informative than point values. Holter monitoring and other methods for continuously assessing a person’s biological activity and exposures to environmental influences will undoubtedly become a major feature of future social surveys, especially if comfortable, minimally invasive instruments can be developed to gather such data. The salience of nutrition and its relationships with social and economic characteristics are increasingly being recognized by social scientists; the research described by Thomas and Frankenberg (Chapter 7) is an example. The potential importance of genetic influences on an individual’s response to nutrition and biomarkers for nutritional intake are described by John Milner and colleagues in their chapter on nutrigenomics. Given the well-documented and extensive problems in collecting complete and reliable data on diet (see, for example, comments in the chapter by Davey Smith and Ebrahim), the development of sensitive and specific biomarkers of nutritional intake would be a major advance. The chapter by Daniel Benjamin and colleagues describes the nascent field of genoeconomics. The field includes three kinds of contributions. “First, economics can contribute a theoretical and empirical framework
OCR for page 10
Biosocial Surveys for understanding how market forces and behavioral responses mediate the influence of genetic factors. Second, incorporating genetics into economic analysis can help economists identify and measure important causal pathways (which may or may not be genetic). Finally, economics can aid in analyzing the policy issues raised by genetic information.” As an example, the authors present their ongoing work exploring candidate genes that may influence decision making and their associations with economic characteristics. As the authors acknowledge, it is a daunting enterprise, one that underscores the relevance of the warnings, both scientific and ethical, raised in Part II. The social scientists whose work is reviewed in this volume are primarily interested in biological indicators in order to better understand how other kinds of factors—behavioral, social, environmental—affect health and aging. Social scientists would like to understand causal relationships. It is, however, difficult to distinguish causation from correlation, association, or reverse causation in social science research. This difficulty is documented in the chapter by George Davey Smith and Shah Ebrahim. One strategy for identifying causal links is to conduct randomized, controlled experiments or clinical trials. Social scientists have only occasionally been able to do so, but Thomas and Frankenberg describe how a clinical trial with iron supplements could be used to study how health influenced workers’ productivity and time allocations in Indonesia. Davey Smith and Ebrahim describe an alternative strategy, in which quasi-randomization is produced by use of instrumental variables related to individuals’ genetic variants. How can social scientists move beyond simplistic associations to more sophisticated understanding of mechanisms and causal linkages? This overarching question is also the topic of the final two papers of the volume. John Cacioppo and colleagues consider the difficulties of interdisciplinary research that crosses biological and social levels of organization. They discuss how a scholar “can productively think about concepts, hypotheses, theories, theoretical conflicts, and theoretical tests” in multilevel investigations. They describe concepts to aid thinking about the “mapping of biological measures to social and behavioral constructs in surveys.” John Hobcraft also considers general issues in the way one models and thinks about research questions. Social science research is increasingly drawing on multiple levels of biological, behavioral, social, and environmental observation; disparate sources of data; and diverse perspectives and areas of expertise. Hobcraft argues that “progress in understanding human behavior (or health) requires an integrative approach.” He recommends “greater attention to pathways within the individual and their interplays with the processes and progresses whereby the individual interplays with multiple contexts over the life course,” concluding that “a concentration
OCR for page 11
Biosocial Surveys on chains or sequences of events, greater awareness of contingent relationships … and elaboration of partial mid-level frameworks or mechanisms” is required. Such an approach is difficult, and most research studies will continue to focus on fragments. Including biological indicators in social surveys will, however, encourage broader integrative thinking and produce deeper understanding of mechanisms and causal linkages. This volume documents how difficult and how promising the endeavor is—and provides practical advice and wise insights. The report concludes with an appendix that provides biographical sketches of contributors to this volume REFERENCES Carey, J.R., and Vaupel, J.W. (2005). Biodemography. In D.L. Poston and M. Micklin (Eds.), Handbook of population (pp. 625-658). New York: Kluwer Academic/Plenum. National Research Council. (1997). Between Zeus and the salmon: The biodemography of longevity. Committee on Population, K.W. Wachter and C.E. Finch, Eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press: . National Research Council. (2001). Cells and surveys: Should biological measures be included in social science research? Committee on Population, C.E. Finch, J.W. Vaupel, and K. Kinsella, Eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press. Weinstein, M., and Willis, R.J. (2001). Stretching social surveys to include bioindicators: Possibilities for the Health and Retirement Study, experience from the Taiwan Study of the Elderly. In National Research Council, Cells and surveys: Should biological measures be included in social science research? (pp. 250-275). Committee on Population, C.E. Finch, J.W. Vaupel, and K. Kinsella, Eds. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.
OCR for page 12
Biosocial Surveys This page intentionally left blank.