Longitudinal Study

A longitudinal study involves the repeated collection of data over time. When the data are collected over time from the same cases, this can also be described as a panel study. Longitudinal studies can also be retrospective if they involve the tracing of archived records over time (e.g., medical records for a group of patients). In contrast, a cross-sectional study involves the collection of data at only one point in time. Longitudinal designs offer unique advantages; by being able to track changes to cases over time, researchers can develop a more nuanced view of dynamic effects and can begin to develop more powerful tests of causality. Longitudinal designs can also incorporate other methodological dimensions, such as a hierarchical (or multilevel) structure. This enables the differentiation of effects of compositional factors (i.e., characteristics of cases or individuals) and contextual factors (i.e., characteristics of groups or areas) over time.

For example, the British Household Panel Survey has collected data on a representative sample of the United Kingdom since 1991. The study's sample consists of more than 5,000 households, and includes over 10,000 individual interviews per year. The same individuals are interviewed year after year, and if a member of a household forms a new household, the new household is incorporated into the survey as well. It includes measures of physical and psychological well-being, as well as measurements of income, employment status, individual demographics, and household composition.

The Longitudinal Survey of Immigrants to Canada offers an example of a longitudinal study of shorter duration. It consists of a sample of 20,000; participants are interviewed at three different times: at six months (wave 1), two years (wave 2), and four years after landing in Canada (wave 3). Data from this survey can be used to examine how new immigrants adjust to life in Canada over time; more specifically, this survey offers data on how the health of new immigrants changes [Page 1029]after arriving in Canada, enabling analysis of underlying social mechanisms affecting changes in health. Other examples of longitudinal studies include the U.S. Panel Study of Income Dynamics and the Canadian Longitudinal Study on Aging.

However, while offering researchers powerful advantages (such as the ability to examine dynamic effects), longitudinal designs also have inherent limitations. These include the problems of mobility, population change, and repeated interviewing effects. The notion of mobility is problematic if respondents move during the life of the study; it can be very expensive to track mobile respondents, and unless care is taken to track and document such moves, the results of the panel may be biased. Population change is problematic if it results in the panel sample itself losing its representativeness as time goes on; this may be a considerable problem over lengthy fol-low-up periods. Repeated interviewing effects, or panel conditioning effects, suggest that respondents may be influenced by their participation in the survey and, over time, begin to give biased responses.

Longitudinal studies are administratively complex, costly to implement, and require a degree of institutional stability if the study is to collect data for an extended period of time. However, these disadvantages are offset by the unique perspective that longitudinal studies offer on dynamic effects. As such, longitudinal studies have been particularly important in the study of the health effects of poverty and unemployment, wherein the length of time respondents live in poverty or are unemployed may be a critical factor in influencing their health.

...