Longitudinal Research

Longitudinal studies are those in which measurement occurs at more than one point in time. These can use qualitative, quantitative, or a combination of methods. The purpose of longitudinal research is to determine the mechanisms and predictors of change or stability in phenomena over time. Most theories on aging focus on the movement of individuals or groups through time, explaining processes such as adaptation, health decline, and accumulating disparity. However, much of the existing research on health and aging has used cross-sectional study designs, where all of the data are collected at one time point. Longitudinal studies are beneficial for establishing causality; measuring change; and separating age, period, and cohort effects.

There are three main types of longitudinal study designs: repeated cross section, panel, and rotating panel. First, repeated cross-section studies are those with similar measurement at two or more time points on different samples drawn from the same population. This study design is most appropriate for measuring change in a population over time. Examples of this design include the National Health Interview Survey that draws a new random sample representing the United States' population each year. Because the samples represent the population at that time point, it is reasonable to expect that the demographic composition (e.g., age, race, gender) of the sample would vary over time as well. Other examples are the National Home and Hospice Surveys and the Survey of Consumer Finances.

Second, panel studies are those that collect data from the same respondents at two or more points in time, generally on the same or related topics. This study design is most appropriate for measuring change within individuals. Examples of panel studies include the Health and Retirement Study, the Established Populations for Epidemiologic Studies of the Elderly, and the Longitudinal Study on Aging. Panel studies typically use a prospective design where respondents are followed forward through time and asked questions about their current state of being. However, some data may be collected retrospectively by asking respondents to recall events or information in their past or between observation points. Integrating retrospective questions in a prospective panel study increases the window of observation, but these responses can be subject to recall bias. Respondents tend to be most accurate about significant events in their lives such as marriage, birth of children, and hospitalization.

The third type of longitudinal study design is arotating panel, which is a combination of the previous two study designs. Like a panel study, this type of design collects a sample of respondents and follows them for a specified length of time that is generally fairly short. Then, consistent with the repeated cross-section design, a new representative sample is recruited and followed for the same length of time as the first sample. These separate samples often overlap in waves, allowing one to compare the two cohorts at the same time point. The advantage of this approach is that one is able to measure change in a population and change within individuals simultaneously. It also minimizes the impact of study attrition over time because each selected respondent is followed for a short period of time, even if the study continues for decades. An example of this design is the Consumer Expenditure Survey, which draws a new sample every quarter and follows those respondents for 15 months. After five consecutive quarters, the sample is dropped. The study continues to rotate new cohorts of respondents into the survey, and they are then followed over time.

...