Longitudinal Studies

In general terms, longitudinal studies are research designs dealing with the problem of time and change. To give a proper definition of longitudinal studies it is necessary to follow some steps. First, it has to be acknowledged that longitudinal research is not a single technique to collect or analyze social sciences data (such as, for instance, Likert scales to survey attitudes or regression analysis). Longitudinal research is a set of techniques matched according to different objectives. Second, it is useful to contrast longitudinal with cross-sectional studies, that piece of social research not only most specialists—but also most laymen—are acquainted with. In cross-sectional studies, data are collected for each case (individuals, organizations, countries, etc.) only once; they refer to a single time interval (that is the same for all cases). An example could be a telephone survey about ethnic consumption where data are collected once and refer to a single period. By contrast, in longitudinal studies, collection takes place more than once and—above all—refers to different periods (days, years, etc.). Third, an important distinction is between collection occasions and time points. Collection occasions are [Page 872]the number of times the researcher enters the field (for instance, organizing a new survey or experiment); time points are the number of periods the data refer to (i.e., in household panels, it is usual to ask the occupational condition for every month of the year preceding the interview). Collection occasions and time points usually coincide (in the year Y0, information is collected about the interval Y0; in the year Y1, about the interval Y1; and so on). But it is not always the case: in a single interview, a subject can be asked to think back on his or her life with regard to family, work, education, et cetera (a standard practice in life-course research). Finally, cases can change from one collection occasion to the other. Not all authors agree on this statement: some restrict longitudinal studies to designs where cases are the same (Rose 2000). Only these research designs provide the chance to study change at the individual level (see below). Here the alternative position on the argument is adopted: it is useful to consider the different features of the full range of techniques used in collecting and analyzing data for different time periods. These last two points are useful to better understand the longitudinal research designs that are introduced later in this entry.

At the end of these steps, it is possible to adopt the following definition:

According to Greg J. Duncan and Graham Kalton (2002, 50–78), the different longitudinal designs encountered on a research field can be classified as

...