Cross-Sectional Design

The methods used to study development are as varied as the theoretical viewpoints on the process itself. In fact, often (but surely not always) the researcher's theoretical viewpoint determines the method used, and the method used usually reflects the question of interest. Age correlates with all developmental changes but poorly explains them. Nonetheless, it is often a primary variable of concern in developmental studies. Hence, the two traditional research designs, longitudinal methods, which examine one group of people (such as people born in a given year), following and reexamining them at several points in time (such as in 2000, 2005, and 2010), and cross-sectional designs, which examine more than one group of people (of different ages) at one point in time. For example, a study of depression might examine adults of varying ages (say 40, 50, and 60 years old) in 2009.

Cross-sectional studies are relatively inexpensive and quick to conduct (researchers can test many people of different ages at the same time), and they are the best way to study age differences (not age changes). On the other hand, a crosssectional study cannot provide a very rich picture of development; by definition, such a study examines one small group of individuals at only one point in time. Finally, it is difficult to compare groups with one another, because unlike a longitudinal design, participants do not act as their own controls. Cross-sectional studies are quick and relatively simple, but they do not provide much information about the ways individuals change over time.

As with longitudinal designs, cross-sectional designs result in another problem: the confounding of age with another variable—the cohort (usually thought of as year of birth). Confounding is the term used to describe a lack of clarity about whether one or another variable is responsible for observed results. In this case, we cannot tell whether the obtained results are due to age (reflecting changes in development) or some other variable.

Confounding refers to a situation in which the effects of two or more variables on some outcome cannot be separated. Cross-sectional studies confound the time of measurement (year of testing) and age. For example, suppose you are studying the effects of an early intervention program on later social skills. If you use a new testing tool that is very sensitive to the effects of early experience, you might find considerable differences among differently aged groups, but you will not know whether the differences are attributable to the year of birth (when some cultural influence might have been active) or to age. These two variables are confounded.

What can be done about the problem of confounding age with other variables? K. Warner Schaie first identified cohort and time of testing as factors that can help explain developmental outcomes, and he also devised methodological tools to account for and help separate the effects of age, time of testing, and cohort. According to Schaie, age differences among groups represent maturational factors, differences caused by when a group was tested (time of testing) represent environmental effects, and cohort differences represent environmental or hereditary effects or an interaction between the two. For example, Paul B. Baltes and John R. Nesselroade found that differences in the performance of adolescents of the same age on a set of personality tests were related to the year in [Page 314]which the adolescents were born (cohort) as well as when these characteristics were measured (time of testing).

...