Cover Page

METHODS IN SOCIAL EPIDEMIOLOGY

SECOND EDITION

 

 

Edited by

 

J. Michael Oakes

Jay S. Kaufman

 

 

 

 

titlepage_fmt

For
Maddy and Henry
and
Amelia, Julian, Louis, and Sol

Tables and Figures

Tables

  1. 4.1 Impact of Alternative Resource Measures on Poverty Rates
  2. 4.2 Distribution of the Poor by the Amount of Their Unmet Needs (1994)
  3. 5.1 Impact of Shifts in the Distribution of Health on Selected Measures of Inequality
  4. 5.2 Calculation of Group-Based Ranking in the Cumulative Distribution of Education
  5. 5.3 Total, Within- and Between-Group Inequality in Body Mass Index by Education and Gender
  6. 5.4 Means and Regression Coefficients for Weight (kg) for Covariates
  7. 5.5 Components of the Mean Difference in Weight (kg) for Blacks and Whites
  8. 5.6 Determinants of Socioeconomic Inequality in Physical Activity Among Men in the Americas Region
  9. 5A.1 Common Inequality Measures Based on Disproportionality Functions
  10. 6.1 Properties of Segregation Measures
  11. 8.1 Principles of Community-Based Participatory Research
  12. 9.1 Four Classic SNA Studies
  13. 9.2 Glossary of SNA Concepts
  14. 9.3 Adjacency Matrix Corresponding to Figure 9.1
  15. 11.1 Anova Partition of the Sums of Squares in a Single Cross-Section Group Randomized Trial Having Unit as a Random Effect
  16. 11.2 Anova Table for the Repeated Measures Analysis in a Nested Cohort Design or the Pre/Post-Test Analysis in a Nested Cross-Section Design
  17. 11.3 Impact on the Factor (t1−a/2 + tpower) in Power Calculations as a Function of df Available
  18. 12.1 Covariate Imbalance Across Exposure Groups Prior to Matching
  19. 12.2 Reduction in Covariate Imbalance after Matching on the Propensity Score
  20. 13.1 Prevalence of Heavy Episodic Drinking by Wave in the Analyzed Subsample of the National Longitudinal Study of Adolescent Health
  21. 13.2 Baseline Predictors and Time-Varying Marijuana Use across Adolescence to Adulthood in Predicting the Intercept and Slope of Log Odds of Heavy Episodic Drinking
  22. 13.3 Respondent Childhood and Adult Characteristics by Adult Neighborhood Disadvantage, African American Respondents, in the Panel Study on Income Dynamics (PSID), 1999 (n = 251)
  23. 14.1 Difference-in-Differences in Potential Outcomes
  24. 14.2 Difference-in-Differences in Regression Coefficients
  25. 15.1 Fixed Effects, Random Effects, and Hybrid Fixed Effects: Linear Model Equations and STATA/SAS Code for Linear and Logistic Models
  26. 15.2 Multilevel Example Results: Black–White PTB Disparity Within and Between Neighborhoods
  27. 17.1 Illustration of Parametric Substitution Estimator Implementation for the CRD and CPAR of the Relation Between Physical Abuse and Psychopathology

Figures

  1. 1.1 Conceptual Framework for Multilevel Thinking
  2. 2.1 Fundamental Graph of Public Health
  3. 4.1 Census Poverty Rate by Age—1966 to 2012
  4. 5.1 Average Body Mass Index and Kernal Density Estimates by Years of Completed Education for Women Aged 25 to 64
  5. 5.2 Proportion of Individuals under Age 65 with Health Insurance, 1998–2009, by Race-Ethnicity
  6. 5.3 Percentage of Stunted Children for Poorest and Richest Wealth Quantiles, Selected Countries
  7. 5.4 Diverging Scenarios for Absolute and Relative Inequality Trends
  8. 5.5 Hypothetical Life Expectancy for Three Social Groups with Varying Population Sizes in Two Different Societies
  9. 5.6 Graphical Example of a Lorenz Curve for Health
  10. 5.7 Relative and Absolute Health Concentration Curves for Daily Smoking in Brazil and Dominican Republic, 2002
  11. 5.8 Income-Based Slope and Relative Index of Inequality in Current Smoking
  12. 5.9 Graphical Depiction of Blinder–Oaxaca Decomposition
  13. 6.1 The Checkerboard Problem
  14. 6.2 The Modifiable Areal Unit Problem
  15. 7.1 Cluster Map from Concept Mapping of Urban Neighborhood Factors and Intimate Partner Violence
  16. 7.2 Neighborhood Stabilization Factors and IPV Cessation. Diagram of the Relationship Between Items Drawn by Participants
  17. 7.3 Neighborhood Monitoring Cluster and IPV
  18. 9.1 Example 7-Node Network
  19. 9.2 Selection and Influence Processes
  20. 11.1 Variance of a Unit Mean as a Fraction of Within-Unit Variance, σ2, Plotted Against the Number of Members per Unit, at Different Levels of the Cluster Effect, VCR
  21. 11.2 Relationship Between the Detectable Difference (Δ) and Power
  22. 12.1 Conceptual Diagram of Target Values and Causal Contrast
  23. 12.2 Fictitious Graph of Overlap in Propensity Scores
  24. 12.3 Overlap in Propensity Scores by the Neighborhood Exposure Group
  25. 12.4 Effect Estimates as a Function of Caliper Width
  26. 13.1 Growth Model for Heavy Episodic Drinking from Adolescence to Adulthood in a Subsample of the National Longitudinal Study of Adolescent Health
  27. 13.2 Hypothetical Disease Prevalence Across Age in a Cross-Sectional Study
  28. 13.3 Hypothetical Disease Prevalence by Age and Time Period
  29. 13.4 Hypothetical Disease Prevalence by Time Period and Birth Cohort
  30. 13.5 Period and Cohort Effects on Asthma Prevalence in the United States 1997–2011 Using a Cross-Classified Random Effects Model
  31. 13.6 Time-Dependent Confounding
  32. 13.7 Time-Dependent Confounding with Common Cause of L and Y
  33. 13.8 Conditioning on Baseline Outcome Status
  34. 13.9 Differential Loss to Follow-up
  35. 13.10 Histograms Denoting the Distribution of Stabilized Inverse Probability of Treatment Weights by Level of Neighborhood Disadvantage
  36. 14.1 Graphical Example of DD Estimate
  37. 15.1 Between- and Within-Cluster Variation and Potential for Cluster-Level Confounding
  38. 15.2 Decision Tree for Random Effects, Fixed Effects, or Hybrid Model Selection
  39. 16.1 Mediation Model in Baron and Kenny (1986)
  40. 16.2 Causal Diagrams Showing Conditions Needed to Identify Total Effects, Controlled Direct Effects, and Natural Direct and Indirect Effects
  41. 16.3 Causal Diagram of the Effects of a Hypothetical Conditional Cash Transfer Program on Children's Height-for-Age
  42. 16.4 Causal Diagram Showing the Mediation Model with Two Mediators of Interest
  43. 16.5 Causal Diagram Showing Unmeasured Confounding of the Meadiator-Outcome Relation by the Confounder U
  44. 16.6 Causal Diagram Illustrating a Non-Differential Error in the Measurement of the Mediator
  45. 17.1 Directed Acyclic Graph or DAG of the Causal Relations Between the Exposure A, Outcome Y, and Confounding Variable W
  46. 18.1 Definitions of Terminology Applied to an Example Causal DAG, with Corresponding Causal Assumptions and Implied Independencies
  47. 18.2 A DAG to Illustrate Identification of Paths Connecting Variables and Covariates That Block Paths
  48. 18.3 A DAG under Which Conventional Confounding Rules Fail
  49. 18.4 A DAG for Selection Bias
  50. 18.5 An Example Illustrating Inclusion of a Measurement Error in Exposure and Outcome, with the Outcome Measurement Error Influenced by the Value of the Exposure
  51. 19.1 Causal Diagrams Depicting a Valid Instrument
  52. 19.2 Causal Diagrams Depicting Variables That Are Not Valid Instruments
  53. 19.3 Characterization of Individuals Based on How the Instrumental Variable or Random Assignment Affects the Exposure or Treatment Variable
  54. 19.4 Example Contrasting ITT, IV, and Population Average Causal Effect in Two Populations
  55. 19.5 Sample Size Required to Achieve 80% Power at α = 0.05 with Improvements in the First-Stage Association

About the Editors

Jay S. Kaufman, Ph.D., is a Professor in the Department of Epidemiology, Biostatistics, and Occupational Health, McGill University. Dr. Kaufman's research focuses on social determinants of health and health disparities, and estimating the causal effects of population interventions.

J. Michael Oakes, Ph.D., is a Professor in the Division of Epidemiology and Community Health, University of Minnesota, and Director of the Robert Wood Johnson Foundation's Interdisciplinary Leaders Program. His research and teaching interests include social epidemiology, quantitative methodology, and research ethics, and he has received the school's highest awards for teaching as well as advising and mentoring.

About the Authors

Jennifer Ahern, Ph.D., M.P.H., is the Associate Dean for Research and Associate Professor of Epidemiology at University of California, Berkeley, School of Public Health. She examines the effects of the social and physical environment, and programs and policies that alter the social and physical environment, on many aspects of health (e.g., violence, substance use, mental health, and gestational health). Dr. Ahern has a methodological focus to her work, including application of causal inference methods and semi-parametric estimation approaches, aimed at improving the rigor of observational research and optimizing public health intervention planning. Her research is supported by a New Innovator Award from the National Institutes of Health (NIH), Office of the Director.

Kate E. Andrade, M.P.H., is a doctoral candidate in the Division of Epidemiology and Community Health, University of Minnesota. Her interests include applied research methods for social epidemiology, causal inference, and consequential epidemiology. Her dissertation work is exploring different analytic techniques in neighborhood effect studies.

David M. Betson, Ph.D., Associate Professor of Economics and Public Policy, College of Arts and Letters, University of Notre Dame. His research examines the impact of government on the distribution of income and wealth in the United States with a particular focus on the measurement of poverty. He was a member of the NRC Panel on Poverty Measurement that in 1995 issued a series of recommendations that has led to the new Supplemental Poverty Measure.

Melody L. Boyd, Ph.D., is an Assistant Professor of Sociology at The College at Brockport, State University of New York. Her research focuses on urban poverty, housing, neighborhoods, race, and social policy.

Magdalena Cerdá, Ph.D., is an Associate Professor of emergency medicine at the University of California at Davis School of Medicine. In her research, Magdalena integrates approaches from social and psychiatric epidemiology to examine how social contexts shape violent behavior, substance use, and common forms of mental illness. Her research focuses primarily on two areas: (1) the causes, consequences, and prevention of violence and (2) the social and policy determinants of substance use from childhood to adulthood.

Stefanie DeLuca, Ph.D., is an Associate Professor of Sociology at the Johns Hopkins University. Her research uses sociological perspectives to inform education and housing policy. She has carried out mixed-methods studies that incorporate qualitative research into experimental or quasi-experimental designs. Her new book address the children of the Moving to Opportunity Study as they transition to adulthood in Baltimore: Coming of Age in the Other America.

M. Maria Glymour, Ph.D., is an Associate Professor at the University of California, San Francisco, Department of Epidemiology and Biostatistics. Dr. Glymour's work focuses on evaluating social determinants of healthy aging, emphasizing methods to overcome causal inference challenges in observational data.

Peter J. Hannan, M.Stat., was a Senior Research Fellow in the Division of Epidemiology and Community Health in the School of Public Health at the University of Minnesota. Mr. Hannan's research interests included methodological issues with clustering in community trials, multiple imputations, Bayesian statistical analysis, and correspondence analysis. He was involved with the Minnesota Heart Health Program, was a statistical consultant to David Murray's classic text “Design and Analysis of Group Randomized Trials,” and has done statistical analysis and power calculation sections for many group randomized trials implemented in the Division, and collaborated on a number of methodological papers in his research interest areas. He is widely recognized as a leader in the design and analysis of community trials. Mr. Hannan died from natural causes on September 28, 2015.

Sam Harper, Ph.D., is trained in epidemiology at the University of South Carolina, the US National Center for Health Statistics, and the University of Michigan. His research focuses on measurement and analysis of social and economic determinants of health using routinely collected data and the use of quasi-experimental and experimental study designs to inform policy. He is currently an Associate Professor in the Department of Epidemiology, Biostatistics & Occupational Health at McGill University.

Ashley Hirai (Schempf), Ph.D., is a Senior Scientist at the Maternal and Child Health Bureau. In this role, she applies technical expertise in perinatal epidemiology, GIS, and advanced research and evaluation methods to inform and improve various programs and initiatives. Her research focuses on perinatal disparities and policy-relevant strategies to reduce inequality.

Alan E. Hubbard, Ph.D., is the Head of Biostatistics at University of California, Berkeley, School of Public Health. Dr. Hubbard is the Principal Investigator of a study of statistical methods related to patient-centered outcomes research among acute trauma patients (PCORI), head of the computational biology Core D of the SuperFund Center at UC Berkeley (NIH/EPA), as well a consulting statistician on several federally and foundation projects, including a study to measure the impacts of sanitation, water quality, hand washing, and nutrition on child growth and development. He has published over 200 articles and worked on projects ranging from molecular biology of aging, wildlife biology, epidemiology, and infectious disease modeling, but most of his work has focused on semi-parametric estimation in high-dimensional data. His current methods-research focuses on statistical inference for data-adaptive parameters.

Barbara A. Israel, Dr.P.H., M.P.H., is Professor of Health Behavior and Health Education in the School of Public Health at the University of Michigan. Dr. Israel has extensive experience conducting, evaluating, disseminating, and translating findings from community-based participatory research (CBPR) projects in collaboration with partners in diverse communities. Her research interests and publications are in the areas of: the conduct of CBPR; the evaluation of CBPR partnerships; the social and physical environmental determinants of health and health inequities; the relationship among stress, social support, control, and physical and mental health; and evaluation research methodologies.

Pamela Jo Johnson, M.P.H., Ph.D., is Associate Professor, Center for Spirituality and Healing, with graduate faculty appointments in the Divisions of Epidemiology and Community Health and Health Policy and Management, School of Public Health, University of Minnesota. She is a health services epidemiologist who focuses on social disparities in health and healthcare; access to healthcare; and complementary and alternative medicine (CAM). Her current work is focused on CAM use in diverse populations, well-being promotion in midlife, and integrative health services research. She is particularly interested in the measurement and methodological issues inherent in each of these areas.

Saffron Karlsen, Ph.D., is Senior Lecturer in Social Research at the Centre for the Study of Ethnicity and Citizenship at the University of Bristol. Her work examines the processes by which ethnicity becomes meaningful in people's lives: as aspects of personal identity and in relation to particular social outcomes, such as health and socioeconomic position. This work has examined, in particular, the influence of power imbalances on ethnic inequalities, evidenced in different forms of racist victimization and social inclusion/exclusion.

Katherine M. Keyes, Ph.D., is an associate professor of epidemiology at the Columbia University Mailman School of Public Health. Katherine's research focuses on life course epidemiology with particular attention to psychiatric disorders and injury, including early origins of child and adult health and cross-generational cohort effects on substance use, mental health, and chronic disease.

Paula M. Lantz, Ph.D., M.S., is Professor of Public Policy and Associate Dean for Research and Policy Engagement in the Gerald R. Ford School of Public Policy at the University of Michigan, where she is also Professor of Health Management and Policy in the School of Public Health. Professor Lantz is an elected member of the National Academy of Medicine. As a social demographer/epidemiologist, her research focuses on public policies and other interventions aimed at improving population health and that address social inequalities in health over the life course. She is currently conducting research regarding the potential of social impact bonds/pay for success strategies in addressing the social determinants of health in low-income communities.

John Lynch, Ph.D., is a Professor of Epidemiology and Public Health, University of Adelaide, Australia. John's research focuses on improving health and development outcomes for disadvantaged children through conducting pragmatic randomized control trials, analyses of large cohort studies, and whole-of-population linked government and non-government administrative and service data.

Lynne C. Messer, Ph.D., is a social, environmental, and reproductive/perinatal epidemiologist whose substantive work focuses on the social-structural determinants of maternal and child health disparities within the Developmental Origins of Health and Disease framework. Methodologically, her work entails better-defining neighborhood environments, developing environmental exposure measures for a variety of health outcomes, and social network analysis. She is also interested in the psychosocial mechanisms through which socio-environmental exposures result in health disparities for women and children. She is an associate professor in the OHSU-PSU School of Public Health. She earned her Ph.D. from the Epidemiology Department (2005) and her M.P.H. from the Department of Health Behavior and Health Education (1995) at the University of North Carolina.

Arijit Nandi, Ph.D., is an Associate Professor jointly appointed at the Institute for Health and Social Policy and the Department of Epidemiology, Biostatistics, and Occupational Health at McGill University. He holds a Canada Research Chair in the Political Economy of Global Health. In his research he is primarily interested in understanding the effects of social interventions on health and health inequalities in a global context. Part of this work applies causal mediation methods to examine the mechanisms through which social inequalities in health are engendered. A former Robert Wood Johnson Health and Society Scholar at Harvard University, Dr. Nandi received a Ph.D. from the Department of Epidemiology at the Johns Hopkins Bloomberg School of Public Health.

James Yzet Nazroo, Ph.D., is Professor of Sociology and Director of the ESRC Centre on Dynamics of Ethnicity at the University of Manchester. He has been investigating ethnic inequalities in health for more than 20 years, with a focus on the role of socioeconomic inequalities, racism and discrimination, area deprivation, and ethnic concentration. Central to this has been the study of the changing ways in which certain identities are radicalized and how this varies over time, over the life course and across contexts.

Margaret O'Brien Caughy, Sc.D., is the Georgia Athletic Association Professor of Family Health Disparities in the Department of Human Development and Family Science at the University of Georgia. Dr. Caughy's research combines the unique perspectives of developmental science, epidemiology, and public health in studying the contexts of risk and resilience affecting young children. She is particularly interested in race/ethnic disparities in health and development and how these disparities can be understood within the unique ecological niches of ethnic minority families. Dr. Caughy has been the principal investigator of several studies focused on how inequities in neighborhood structural characteristics and social processes affect the cognitive development, socioemotional functioning, and early academic achievement of young children in diverse race/ethnic groups. Another theme of her research has been methodological, specifically methods related to measuring neighborhood context and the utilization of these measures in models explaining child developmental competence using multilevel and structural equations modeling methods.

Patricia O'Campo, Ph.D., is Professor of Epidemiology at the Dalla Lana School of Public Health Sciences at the University of Toronto and holds the Chair for Intersectoral Solutions to Urban Health Problems. She is co-lead on the University of Toronto's Healthier Cities Hub, a research and education unit dedicated to work in partnership with community organizations to improve the health of those residing in urban settings. As a social epidemiologist, she has been conducting research on the social and political determinants of health and health inequalities for over 25 years. Dr. O'Campo's work often focuses on upstream determinants of health, quantifying the impacts of structural issues and social programs, and working to propose concrete solutions. She co-edited the book Rethinking Social Epidemiology: Toward a Science of Change (2011, Springer), which calls for stronger evidence for and evaluations of interventions to address health inequities.

Sean F. Reardon, Ph.D., is the endowed Professor of Poverty and Inequality in Education and is Professor (by courtesy) of Sociology at Stanford University. His research focuses on the causes, patterns, trends, and consequences of social and educational inequality, the effects of educational policy on educational and social inequality, and in applied statistical methods for educational research. In addition, he develops methods of measuring social and educational inequality (including the measurement of segregation and achievement gaps) and methods of causal inference in educational and social science research.

Angela G. Reyes, M.P.H., is the founder and Executive Director of the community-based Detroit Hispanic Development Corporation, which was established in May 1997 and has since grown to provide several state-of-the-art programs in the Southwest Detroit community. Ms. Reyes is herself a resident of Southwest Detroit, where she has been active in the community for more than 30 years. Ms. Reyes is a national speaker on issues affecting her community, including youth gangs and violence, substance abuse, community activism, and cultural competency.

Amy J. Schulz, Ph.D., is Professor of Health Behavior and Health Education and the Associate Director of the Center for Research on Ethnicity, Culture and Health in the School of Public Health at the University of Michigan. Dr. Schulz has extensive experience conducting community-based participatory research with a particular focus on etiologic and intervention research to address social determinants of health inequities. She contributes considerable expertise in engaging diverse partners in the development, implementation, and evaluation of multilevel interventions to promote health and address environmental factors linked to health, and in the evaluation of partnership characteristics and their associations with partnership effectiveness.

David A. Shoham, Ph.D., is an Associate Professor and Director of the M.P.H. program at Loyola University Chicago. He received his Ph.D. in Epidemiology from UNC Chapel Hill in 2007, where he focused on social epidemiology. His current research focuses on applying social network analysis to understand healthcare teams, health behavior, and prevention of chronic disease.

Erin C. Strumpf, Ph.D., is an Associate Professor in the Department of Economics and the Department of Epidemiology, Biostatistics and Occupational Health at McGill University. Her research in health economics focuses on measuring the impacts of policies designed to improve the delivery of healthcare services and improve health outcomes. She examines the effects on healthcare spending and health outcomes overall, and on inequalities across groups.

Eric J. Tchetgen Tchetgen, Ph.D., is a Professor of Biostatistics and Epidemiologic Methods at Harvard T.H. Chan School of Public Health, Departments of Biostatistics and Epidemiology. Professor Tchetgen Tchetgen conducts methodological research in causal inference and missing data problems.

Tyler J. VanderWeele, Ph.D., is Professor of Epidemiology in the Departments of Epidemiology and Biostatistics, at the Harvard School of Public Health. He holds degrees in mathematics, philosophy, theology, finance, and biostatistics from the University of Oxford, the University of Pennsylvania, and Harvard University. His research in causal inference concerns how we distinguish between association and causation in the social and biomedical sciences and the study of the mechanisms by which causal effects arise. His current empirical research is in the areas of perinatal, psychiatric, and social epidemiology; various fields within the social sciences; and the study of religion and health. Dr. VanderWeele serves on the editorial boards of Epidemiology, The American Journal of Epidemiology, Journal of the Royal Statistical Society Series B, Journal of Causal Inference, and Sociological Methods and Research. He is also Editor-in-Chief and co-founder of the new journal Epidemiologic Methods. He has published over 200 papers in peer reviewed journals, is author of the book Explanation in Causal Inference: Methods for Mediation and Interaction published by Oxford University Press, and will also be an author on the fourth edition of the epidemiologic methods text Modern Epidemiology.

Stefan Walter, Ph.D., is a Research Specialist at University of California San Francisco (UCSF), Department of Epidemiology and Biostatistics. Stefan Walter is an expert in Mendelian randomization analysis and genetic epidemiology. His research focuses on the relationship between cardiovascular risk factors such as obesity and diabetes and cognition and dementia.

Jennifer L. Warlick, Ph.D., Associate Professor of Economics and Public Policy and Director of the Poverty Studies Program, College of Arts and Letters, University of Notre Dame. Her research and educational interests are to examine the causes and consequences of poverty in the United States and developing nations from a multidisciplinary perspective.

Preface

This text addresses many important methodological issues faced in contemporary social epidemiologic research. The motivation for assembling this material is to increase the potential for social epidemiology to contribute meaningfully to public health knowledge and policy through stronger and clearer methodological foundations. It has been 10 years since the publication of the first edition of this book, and yet social epidemiology remains a nascent enterprise, and the methodologic approaches that characterize work in this subdiscipline are still rapidly evolving. New techniques are continually being developed or borrowed from other disciplines. Nonetheless, the bulk of published research in this area is still made up of studies for which the inferential content is modest at best. Some of this ambiguity in interpretation arises from a weak conceptual orientation about the logic underlying many common methods. This is especially true of regression, which is seldom taught with a focus on causal inference.

Without improvements in standard analytic practice, social epidemiology risks being dismissed as naïve or simplistic by policymakers as well as by the wider scientific readership. Popular imagination and scientific credence are extended readily to the rapid developments in molecular biology and genetics, even though their relevance for public health concerns remains largely speculative. In contrast, the questions posed in social epidemiology have immediate relevance for the most important public health concerns, and yet the results of such studies rarely have the necessary clarity and robustness to alternate explanations, such as confounding and measurement error, that would allow them to enter meaningfully into the public and policy debates. This dilemma will not be solved overnight with the introduction of some exciting new statistical model, but rather slowly, over time, with the training of more careful thinkers and more assiduous analysts.

This volume is intended as a methods text, and so is unlike the handful of recent books on social epidemiology and the social determinants of health, which focus on substantive findings.

For this reason, little attention is paid to existing knowledge about social epidemiologic relations, except by way of motivation or worked examples. It is our intention, however, that this text will compliment these substantive efforts by providing a more thorough investigation of the techniques we use to gather subject matter knowledge in this field, and ways in which this research process can be improved.

Is there really a need for a separate text devoted entirely to social epidemiological methods? Why should the interested reader not just rely on the many outstanding methods texts available for epidemiology as a whole? We believe that social epidemiology as a distinct subdiscipline comprises several phenomena that are not very well addressed by traditional epidemiological texts. Foremost among these are human volition, social interaction, and collective action. Since epidemiology is a population science, it is indeed ironic that mainstream epidemiology texts say so little about human interaction, social forces, or social scientific research and understanding more generally. In noting this, we certainly do not intend to minimize the importance of medical or biological knowledge or research; there can be no doubt that these disciplines are also vital to epidemiology. Our point is only that something is missing. A more complete epidemiology includes the social, the biologic, and the quantitative, and yet the first of these, which most distinguishes our field from clinical medical investigation, is almost entirely neglected in texts written in the modern period (for example, since the appearance of Kupper, Kleinbaum, and Morgenstern's Epidemiologic Research in 1982 and Miettinen's Theoretical Epidemiology in 1985). Furthermore, we emphasize that this is obviously not a complete methods text, if such a thing were even conceivable. It is not meant to replace the traditional epidemiology texts, statistical analysis texts, or other foundational works or training. Rather, it augments these works by providing a collection of insights and some original research into the particular challenges facing the study of social relations and institutions on health.

We hope this second edition continues to serve as a learning guide, a reference tool, and a stepping stone for conceptual advancement. Our target audience remains second-year epidemiology doctoral students—those who have some basic training in epidemiologic methods and the capacity and interest to extend these to settings in which the exposures are social phenomena or related to the same. Accordingly, we encouraged contributing authors to write penetrating and cutting-edge chapters that are nonetheless accessible to non-methodologist readers. Since chapter lengths were necessarily limited, we also asked our authors to include abundant citations through which interested readers might continue their study in greater detail.

The text is loosely organized into an introduction and two sections: (Part One) measures and measurement and (Part Two) design and analysis. Kaufman and Oakes's introductory chapter addresses the state of social epidemiologic methodology and important focus areas. The first section, on measures and measurement, comprises six chapters. There must be no doubt that better conceptualization of study quantities and measurement of these quantities is fundamental to any scientific advance. First, Oakes and Andrade consider the construct of socioeconomic position and its central role in social epidemiology. Next is an important chapter on the measurement and analysis of race and racial discrimination by Karlsen and Nazroo; much more work is needed in this area and this chapter moves us forward with greater precision and clarity. Betson and Warlick's chapter on measuring poverty comes next. The most enduring finding in all of health research is that poverty is not healthy, and this chapter serves as a much-needed reminder that such a seemingly simple idea as poverty is anything but simple to operationalize. Following this, Harper and Lynch contribute an essential chapter in measuring health inequalities. Once again, the deep issues here are difficult and these authors help us to recognize and better appreciate the subjective aspects of these measures. Because residential segregation remains overlooked in much of epidemiology, we wanted to include a cutting-edge discussion of the construct and current thinking in this volume. Reardon's chapter not only fills the gap but offers practical insights into how such measurement can and should be done. Finally comes a chapter on measuring neighborhood constructs by O'Campo and Caughy, who carefully consider methods and issues that should move us beyond naïve reliance on census data for community measurement. Taken together, the chapters in this section greatly strengthen social epidemiology's foundation by clarifying and extending the measurement tools available to social epidemiologists aiming to understand how social processes interface with health.

The second and larger block of chapters includes 12 contributions on research designs, data analysis, and related issues. The first chapter, by Lantz and colleagues, is special in that it concentrates on community-based participatory research. Such an approach appears to blend well with our view of social epidemiology and merits more attention. Following this is Shoham and Messer's thoughtful and informative chapter on understanding, measuring, and analyzing social networks. This chapter should help fill a major gap in the current literature and help strengthen formal approaches to networks. Next comes a chapter that is new to the second edition by Boyd and DeLuca on qualitative methods. As the authors show, qualitative methods are a critical part of understanding etiologic processes in context, and developing explanations that are sufficiently deep and rich to address the complexities of social life.

Given the centrality of randomization to quantitative studies, the chapter by Hannan on design and analysis of community trials is a key resource. Observational studies of community effects are meant to mimic exactly these kinds of designs, and so an appreciation for the conduct of such studies is a necessary foundation for all multilevel work. We remain convinced that cluster-randomization remains woefully misunderstood and neglected by social epidemiologists.

Next comes a chapter by Oakes and Johnson on propensity score matching, a technique that relies on measured covariates, but permits several advantages over standard regression modeling, including balance checks, non-parametric contrasts, and restriction to regions of the data in which causal inference is most secure. Cerdá and Keyes follow with another chapter that is new to the second edition, focusing on life course models and analyses. It is one of several chapters in the new edition that contain examples of coding in standard software packages, which we hope will make the material more readily accessible for readers who want to put these ideas into practice.

The next two chapters deal with clustered data, as encountered in life course designs like those described by Cerdá and Keyes as well as in community or neighborhood multilevel designs like those described by the chapters by Hannan and by Reardon. The chapter by Strumpf and colleagues develops the fixed effects model and shows its relation to the econometric technique called “differences in differences,” which is especially appropriate for studying the causal effect of social interventions or policy changes. The second of these chapters, by Hirai and Kaufman, covers random effects and fixed effects models, as well as a “hybrid model” that seeks to take advantage of the best aspects of each of these.

Two more new chapters follow, which were not in the first edition because they represent methods that were not yet a part of the standard toolkit just 10 years ago, but are sufficiently developed to be now applied widely. The first of these chapters, by Nandi and VanderWeele, focuses on effect decomposition and mediation, topics that have had a long history in the social sciences, but only recently received a solid methodologic treatment in epidemiology. The next new contribution is a chapter by Ahern and Hubbard on standardization methods, which are in fact old tools that have been resurrected by the causal inference community in the first decade of the twenty-first century, and which offer notable advantages for population scientists in making flexible inferences that are not constrained by arbitrary scale choices and which free epidemiologists to choose more readily interpreted population contrast measures.

The design and analysis section is completed with two chapters by Glymour that were present in the first edition, but which receive considerable revision and updating in this edition. The first of these chapters on instrumental variables analysis reflects an explosion of interest in this and related techniques for identifying causal effects when some confounders remain unmeasured. The second Glymour chapter is on causal diagrams, which have also become a mainstay of epidemiologic practice over the last decade, especially so in social epidemiology.

No preface is complete without acknowledgments. As in the assembly of all such works, we find ourselves in the debt of many—in fact, too many to mention—but a few merit extra special thanks from both of us. First, we gratefully acknowledge the remarkable group of contributing authors; their hard work and positive attitudes nearly made this project fun all over again. Next, we owe a special debt to our publisher Andy Pasternack and his colleagues at Jossey-Bass. Andy encouraged us to undertake the first edition and he remained remarkably patient as we missed several self-imposed deadlines. Later, Andy began to work with us on the second edition, but he did not survive to see this work completed. We still get excellent support from Jossey-Bass, but we miss Andy and remember him fondly for his important role in making this book possible from the very beginning.