Design and conduct of surveys (cont.)
Panel (cohort) studies
- These designs involve two or more rounds of data collection from the same panel or cohort of units.
- They are more logistically complex than single-round surveys because it is necessary to track the units over time and contact them repeatedly and then to link together the information collected about them upon each occasion.
- Multi-round surveys vary greatly in the frequency with which data are collected (daily, weekly, monthly, annually, etc.) and in the overall duration of the study (1 week to a lifetime).
- The approach is suited to the investigation of dynamic processes. For example, one could measure income inequalities at a point in time in a single-round survey, but only a panel study would be capable of monitoring economic mobility at the individual level.
- Studies that collect longitudinal data also make it possible to link earlier exposures to later outcomes and can be analysed by methods that address some of the difficulties involved in identifying causal processes using observational data.
Can you think of some drawbacks of using this type of survey?
Multi-round studies typically experience substantial attrition of the sample due to losing track of units and refusals. Such loss to follow up of units not only reduces the size of the sample but may cause it to become increasingly unrepresentative of the population from which it was drawn.
Panel studies may also become unrepresentative because either the characteristics of the units alter over time as a result of participating in the study or they provide less and less accurate information. This phenomenom is often termed conditioning or panel conditioning.