panel study

panel study A study that provides longitudinal data on a group of people, households, employers, or other social unit, termed ‘the panel’, about whom information is collected over a period of months, years, or decades. Two of the most common types of panel are age-cohorts, people within a common age-band, and groups with some other date-specific common experience, such as people graduating from university, having a first child, or migrating to another country in a given year or band of years. Another type is the nationally representative cross-sectional sample of households or employers that is interviewed at regular intervals over a period of years. Because data relate to the same social units, change is measured more reliably than in regular cross-sectional studies, and sample sizes can be correspondingly smaller (often under 500), while remaining nationally representative, as long as non-response and sample attrition are kept within bounds. These are the key problems for panel studies, as initial samples are eroded by deaths, migration, fatigue with the study, and other causes. Another problem is that people become experienced interviewees, leading to response bias. For example, they may report ‘no change’ since the previous interview, so as to avoid detailed questioning on changes that have in fact occurred.

Data are usually collected through interview surveys with respondents in the panel, with other informants (such as parents, doctors), with their spouses and other members of their household. With the respondent's permission, data from administrative records may be added, such as information from educational or medical records, which are usually more precise than the respondent's recollection. A panel element is sometimes added to regular cross-sectional surveys, and rotating sample designs are a hybrid between panel study and regular survey.

Because they yield longitudinal data, panel studies offer possibilities for examining the relationships between individual life-histories, cohort effects, and period effects due to social change. A number of specialized techniques have therefore been developed for analysing these data (see, for example, J. S. Coleman , Longitudinal Data Analysis, 1982
). Nick Buck et al. ( eds.) , Changing Households (1994)
, is a good example of the range of substantive issues that can most usefully be addressed using panel data—in this case those from the British Household Panel Study. Topics include, for example, changes in household composition, patterns of residential mobility, and stability in voting intentions over time. This volume also contains a useful overview of the methodological problems involved in conducting panel surveys.