Volume 14, Issue 2

The Health and Retirement Study (HRS) is widely use for research on the well-being of the elderly. This paper assesses the quality of economic and financial variables in the HRS. I find the coverage is comprehensive and the quality of the data is uniformly high. Thus the HRS has earned its position as the most widely used data source for research on retirement, saving adequacy, pension policy and a host of other aging-related topics. I identify two general areas that continue to merit special attention. The first is measurement error, particularly errors arising from item non-response and from inaccurate respondent reports of the ownership and level of assets. The second is the quality of the pension data. Where appropriate, I make suggestions for improving economic measures in the HRS.

As a nationally representative cohort of middle-aged and elderly adults with longitudinal data spanning nearly two decades, the Health and Retirement Study (HRS) is an important resource for researchers studying the dynamics of health insurance coverage in the United States and the relation of insurance coverage to the use of health services and to health outcomes. We assessed the strengths and limitations of currently available HRS data for such research, focusing on survey items in these domains and Medicare claims data that have been linked to HRS survey data. The process for researchers to obtain Medicare claims has greatly improved in recent years. The additions of biomarkers (e.g. blood pressure and serum cholesterol) and objective measures of physical functioning for HRS participants have also been notable improvements. We propose changes to the HRS to enhance its value for health services research, particularly regarding the effects of health care reform as the Patient Protection and Affordable Care Act of 2010 is implemented over the next decade.

This paper reviews out-of-pocket (OOP) medical expenditure measures collected in the Health and Retirement Study (HRS). Medical expenditures are an important cost of poor health. Medical expenditure measures are important for understanding retirement decisions, financial preparation for retirement, and predicting the consequences of health care reform, particularly Medicare reform. Despite the comprehensiveness of the HRS, there are always limitations to what can be learned from population interviews. To assess the quality of current HRS measures of OOP spending, we compare various measures of OOP spending across survey waves to the Medical Expenditure Panel Survey (MEPS) and Medicare Current Beneficiary Survey (MCBS), two surveys that expend considerable resources on measuring both OOP spending and total medical expenditures. Such comparisons make it possible to identify potential bias in the HRS data and to improve HRS measures of OOP. We find that the HRS produces good quality and useful data on OOP spending.

The Health and Retirement Study (HRS) is an important national resource for policy makers and investigators across a wide range of disciplines, and it is critical that the study collects the best information possible on the health status of its participants within the constraints of the survey design, and without compromising the integrity of the sample. Potential directions for the collection and analysis of biomarker data in future waves of HRS are discussed, with a primary focus on blood-based biomarkers. Advantages and disadvantages of various methods for collecting blood in the home are considered, with particular attention given to the strengths and weaknesses of dried blood spot (DBS) sampling. DBS sampling has been widely applied in recent biosocial surveys due to the low cost and burden associated with sample collection, but these benefits need to be weighed against challenges associated with quantification in the laboratory. Attention is also given to additional biomarkers that may be of relevance to HRS, and that would expand the survey’s current focus on obesity and metabolic syndrome. Measures of inflammation, pathogen exposure, reproductive function, stress, and epigenetic modifications are suggested as potentially productive future directions for the study. In addition, the analysis concludes with the following recommendations for HRS: Continue to collect DBS samples, but consider alternatives; implement enhanced procedures for quality control; calibrate DBS results against plasma values, and invest in methods development.

We review a diversity of genome-wide association studies (GWAS) with particular emphasis on precision in specifying phenotypes. This implies that examination of any specific phenotype involves considering the likely genetic contributions to it from the entire genome. We consider a variety of phenotypes specifiable with data from the Health and Retirement Survey (HRS). However, evidence from other large population studies is also incorporated as part of the process of developing and refining pathway representations from the genome thru a hierarchy of intermediate endpoints to behavioral, cognitive, and economic phenotypes. Any causal modeling focused on genome-phenotype connections must, of necessity, include consideration of intermediate endpoints (endophenotypes) as mediators of such associations. We also discuss metabolic and gene expression consequences of gene-environment interactions as a next research step beyond GWAS, not only for HRS but also for an integrated set of human population surveys that can provide much more statistical power than any one of them used alone. A variety of concrete examples based on physiological, psychological, sociological, and economic outcomes are carried along throughout our discussion.

Subjective well-being is comprised of both evaluative (life satisfaction) and hedonic (affect) components, and there has been a call to include both aspects of well-being in large-scale surveys. This paper presents a rationale for the feasibility of including a brief measure hedonic well-being based on the measurement of yesterday’s affect and experience. It discusses issues of the distinctiveness of hedonic well-being from life satisfaction, the sensitivity of a single day’s affect, the sample sizes required for detecting group differences, and the experiential content that could also be collected to extend the value of affective reports. I conclude that a brief assessment is feasible and, in conjunction with measures of evaluative well-being, could add to our understanding of well-being in populations.

This paper reviews several methods for measuring how people spend their time, and how they feel during these different activities, and argues that some of these methods could be well suited for large scale longitudinal surveys. Because time use methods allow for the quantitative assessment of the dynamics of human experience, they provide opportunities to explore numerous research questions that cannot be readily answered with more traditional summary measures of well-being. In the last decade or two, techniques have become available that can capture the dynamics of time use and well-being in ways that also reduce methodological problems such as biased recall. The paper describes several such methods, with a discussion of how they are implemented, and a comparison of their relative strengths and weaknesses. In the final section, the paper describes how the addition of time use measures can enhance national surveys such as the Health and Retirement Study by a) opening new avenues of research, and b) clarifying previous findings from more traditional measures of well being.

Personality tests are being added to large panel studies with increasing regularity, such as the Health and Retirement Study (HRS). To facilitate the inclusion and interpretation of these tests, we provide some general background on personality psychology, personality assessment, and the validity of personality tests. In this review, we provide background on definitions of personality, the strengths and weaknesses of the self-report approaches to personality testing typically used in large panel studies, and the validity of personality tests for three outcomes: genetics, income, and health. We conclude with recommendations on how to improve personality assessment in future panel studies.

I discuss the data in the Health and Retirement Study (HRS) that can be used to study family change and intergenerational family relationships and offer suggestions about what might be done to enhance the uses of the HRS family data going forward. A number of family demographic behaviors are altering the family context of more recent cohorts of the HRS. These family changes need to be well-captured in the data collection and and should also inform future design decisions. Changes include the higher rates of childlessness, delayed marriage and childbearing after age 30 among the Baby Boom cohorts just now being enrolled in the HRS. Cohorts coming into the study also have higher rates of (lifetime) labor force participation on the part of women and much higher rates of nonmarital childbearing, marital disruption, and informal cohabitation than the original HRS cohorts. There is also great heterogeneity in family patterns by race and class among Baby Boom cohorts. Many of these changes increase the value of collecting family data in the HRS but also complicate the collection of useful data on transfers between parents and children or among siblings. I offer four suggestions for enhancing data collection in the HRS in light of these family changes. These include the following: 1) Reevaluate decisions about the family members on whom to gather information, particularly the decisions about when to collect data on siblings. 2) Consider collecting more information directly from each spouse (and perhaps expanding the definition of “spouse” to include cohabiting partners as cohabitation is on the rise among cohorts now entering the HRS). 3) Begin to experiment with interviewing adult children of HRS respondents and consider broadening the content to include more on parent-child relationship quality and/or on everyday activities and exchanges. 4) Make the family data in the HRS easier to access and use.

This part of the mid-term review of the Health and Retirement Study (HRS) provides an overall assessment of the utility of HRS data for research targeting the nature and influence of social connectedness. As one of the major dimensions of the social aspects of psychosocial influences, social connectedness is among the most complicated in terms of definition, conceptualization, and measurement. However, the century-long body of theory and findings couple with a recent resurgence of research on the critical impact of these ties for health, illness, and health care to call for an examination of the richness in and limitations of current HRS data.This assessment is comprised of three broad steps: 1) an overview of the nature of social connectedness, and of the dimensions and methodological approaches that can and have been used in studying health, health care, and aging; 2) the range, strengths and limitations of the HRS data on each approach; and 3) suggestions for potential directions to increase the utility of data collected and further research contributions from the HRS. While no tabular listing of items relevant to social connectedness is presented, the sets of items that tap this notion are referenced throughout.Overall, the HRS represents one of, if not the most impressive data sets regarding the ability to examine the influence of social connectedness on health, illness and health care. Given different theoretical and methodological traditions of social connectedness (e.g., the local or ego-centered perspective; social support perspective; social capital perspective; Pescosolido 2006a), the HRS either currently offers a way to tap into various views of social connectedness or holds the potential to do so. Specifically, the HRS includes four kinds of social connectedness data: socio-demographic proxies that represent a tie (e.g., marital status) with detailed data on the nature of the bond; social support batteries which offer respondent perceptions of the overall positive and negative aspects of sets of relationships; eco-centric tie data, which provide a list of names or roles that can provide support (i.e. latent ties); and networks of event response in which respondents list individuals who were called upon (e.g., activated ties) under certain conditions. Given the individually-based and national scope of the HRS, the collection of full or complete network data is not feasible at present.Four strategies could improve the collection and use of social connectedness data in the HRS. First, data collection sections that are explicit or implicit ego-centric name generators or activated ties lists could be expanded and refined to provide more complete data. Under the “looping” structure of the HRS, both the ego-centric and event response batteries can serve as a foundation for expanded network batteries. Second, given the increasing role of social media in contemporary American lives, the HRS section on the use of technology should be reviewed and expanded to tap into virtual ties. Third, locator data designed to improve follow-up of the HRS samples can form the basis of a network roster and for analyses of the dynamics of ties and its influence on health and health care. Fourth, while it is not possible to “go back” and recapture data about social connectedness, a sub study which targets the named “social convoy” over a person’s life (defined only as time in the HRS) would provide invaluable data that could not be collected from any other existing study. That is, while subject to a variety of criticism (e.g., telescoping effects), the ability to collect data on extent of turnover and the reasons for shifts in social connectedness would allow an analysis of the impact of social network dynamics in later life, potentially reveal key turning points in social network support, and offer targeted points of interventions for fostering the social connectedness that has, to date, been shown (in the HRS and other studies) to be so essential to health and well-being.

Since 2002, the average number of questions asked per respondent in the Health and Retirement Study (HRS) has risen by 39 percent, from 413 to 581. Yet there is little or no understanding of which questions, or how many in total, should be included–and more importantly, maintained–in longitudinal surveys. In this paper, we propose a simple approach to assessing the value of survey questions: journal citation counts. A sample of journal articles and book chapters published in 2006-09 (N = 206) is used to document which questions, and categories of questions, were used most and least frequently. A disproportionate number of published articles used a relatively small number of questions regarding health, wealth, income, and employment. By contrast, several categories of questions were rarely used, and many specific questions were never used. This evidence-based approach to measuring the value of survey questions can have applications for other surveys beyond the HRS.