Longitudinal Design

Traditional Longitudinal Designs

Longitudinal designs can be categorized in several ways but are defined primarily on differences in initial sample (e.g., age homogeneous or age heterogeneous), number of occasions (e.g., semiannual or intensive), spacing between assessments (e.g., widely spaced panel designs or intensive measurement designs), and whether new samples are obtained at subsequent measurement occasions (e.g., sequential designs). These design features can be brought together in novel ways to create study designs that are more appropriate to the measurement and modeling of different outcomes, life periods, and in capturing intraindividual variation, change, and events producing such changes.

In contrast to a cross-sectional design, which allows comparisons across individuals differing in age (i.e., birth cohort), a longitudinal design aims to collect information that allows comparisons across time in the same individual or group of individuals. One dimension on which traditional longitudinal designs can be distinguished is their sampling method. In following a group of individuals over time, one might choose to study a particular birth cohort, so that all the research subjects share a single age and historical context. As an extension of this, a variety of sequential designs exists, in which multiple cohorts are systematically sampled and followed over time. K. Warner Schaie's Seattle Longitudinal Study used such a design, in which new samples of the same cohorts are added at each subsequent observational “wave” of the study. More typical, however, is an age-heterogeneous sample design, which essentially amounts to following an initial cross-sectional sample of individuals varying in age (and therefore birth cohort) over time. Cohort-sequential, multiple cohort, and accelerated longitudinal studies are all examples of mixed longitudinal designs. These designs contain information on both initial between-person age differences and subsequent within-person age changes. An analysis of such designs requires additional care to estimate separately the between-person (i.e., cross-sectional) and within-person (i.e., longitudinal) age-related information. Numerous discussions are available regarding the choice of design and the associated strengths and threats to the validity associated with each.

Another dimension along which traditional longitudinal designs can differ is the interval between waves. Typical longitudinal studies reassess participants at regular intervals, with relatively equal one or several-year gaps, but these might vary from smaller (e.g., half-year) to longer (7-year or decade). Variations in this pattern have been used because of funding cycles; for example, intervals within a single study might range from 2 to 5 or more years, and to creative use of opportunities, such as recontacting in later life a sample of [Page 741]former participants in a study of child development or of army inductees, who were assessed in childhood or young adulthood. Intensive measurement designs, such as daily diary studies and burst measurement designs, are based on multiple assessments within and across days, permitting the analysis of short-term variation and change.

...