Creating an ideal data set would be very costly, but it may represent the only strategy likely to produce sufficient data to elucidate completely the relationship between workplace risk factors and workers’ health and employment patterns in later life. A more limited and less expensive alternative is to modify existing longitudinal and nonlongitudinal surveys so they contain crucial information about workplace health risks. Another alternative is to collect information on a convenience sample for which longitudinal record gathering is less costly in contrast to a nationally representative, random sample with periodic in-person or telephone survey updates. One possibility is to conduct thorough baseline interviews in a cohort of workers and recent retirees from a large national employer, such as the U.S. government. Personnel and other administrative records and less frequent in-person interviews would be used to construct lifetime work histories and measure subsequent employment and retirement patterns in the cohort. The size of the government workforce would also permit targeted sampling for better assessment of demographic subgroups. It would likely, however, exclude the possibility of assessing a full range of occupations.
Recommendation 1: New longitudinal data sets should be developed that contain detailed information on workers’ employment histories and the specific demands of their jobs, as well as objective information on the health and safety risks to workers in the job. If cost makes it impossible to create a nationally representative, longitudinal survey focused on workplace health and safety, a less expensive alternative is to create a new longitudinal data set using a convenience sample in which information gathering is less costly, for example, a representative sample of workers at a large national employer, such as the U.S. government.
The risk of workplace injury or illness or disorder varies both across and within occupation and industry, and workers’ exposure to such risks varies across the course of their lives. Therefore, analyses that attempt to explain life course health outcomes or that use health characteristics as variables to help explain major life course transitions such as retirement should have good information on these health and safety risks.
However, otherwise richly detailed socioeconomic surveys such as the Health and Retirement Study or the Panel Study of Income Dynamics, which contain detailed information on the health characteristics of their respondents, lack information on the health and safety risks that workers face in their current or past jobs. A National Research Council (2001) report has strongly encouraged longitudinal research to disentangle and illuminate the complex interrelationship among work, health, economic status, and family structure. Without capturing the independent effects of