Panel

The term panel refers to a research design in which the data are gathered on at least two occasions on the same units of analysis. Most commonly, that unit of analysis is an individual in a survey. In a two-wave panel survey, respondents at time t are reinterviewed at time t + 1; in a three-wave panel, they would be reinterviewed yet again at time t + 2. Sometimes, the units of analysis of the panel are aggregates, such as nations. For example, a sample of European countries might be measured at time t and again at t + 1. A special value of a panel design is that it incorporates the temporality required for strong causal inference because the potential causal variables, the Xs, can actually be measured before Yoccurs. In addition, panel studies allow social change to be gauged. For example, with repeated interviewing over a long time, panel surveys (unlike cohort designs) have the potential to distinguish age effects, period effects, and cohort effects.

A chief disadvantage of the panel approach is the data attrition from time point to time point, especially with panel surveys in which individuals drop out of the sample. A related problem is the issue of addition to the panel, in that demographic change may, over time, suggest that the original panel is no longer representative of the population initially sampled. For example, as a result of an influx of immigrants, it may be that the panel should have respondents added to it to better reflect the changed population.