7.3 Cross-sectional and Longitudinal Studies
Cross-sectional study designs essentially offer a snapshot in time, whereby outcomes are observed (or measured) in all participants at the same time without manipulating any of the study conditions. This type of design is useful for gathering large amounts of information about participants and comparing variables within that time point. For example, to assess attitudes or current behaviours at a single time-point, and comparing across demographic groups. The downside to this type of design is that it is difficult to control for other possible influencing factors (although they can be measured and statistically accounted for it would increase the sample size needed for robust analysis), and it is not possible to determine cause and effect. This means that it is not possible to determine the exact impact that the intervention has had on the outcome measures. For example, we may capture behavioural frequency after intervention but not know what the frequency was among the sample before our intervention, thus we cannot say that the intervention directly caused the behavioural frequency.
Longitudinal designs, similarly to cross-sectional studies, are observational. However, in longitudinal designs outcome measures are taken at multiple time points and analysis can track changes over time. This means that longitudinal designs can help to establish cause and effect. Owing to the need for more resource to run an effective longitudinal study (e.g. to get a large enough sample accounting for attrition, collecting data at multiple time points, and allocating resource to a project over a much longer time frame) it can be useful to run a cross-sectional study to establish links and associations between variables of interest before deciding whether a longitudinal study is worth conducting.
Example: Public safety campaigns and events
Since public safety campaigns and community events are targeting a wide audience and could be accessed by anyone within the local community, there are a variety of factors beyond the control of the research team that may influence customer behaviour. Additionally, it may not be feasible to contact specific individuals before and after a campaign or event for a reliable longitudinal study. Therefore, for this type of intervention activity, the recommended approach is to pre-test messaging based on behaviour change theory, and then to assess at a single time point what customer attitudes and behaviours are. This could then be rolled out into a longitudinal design to assess whether there are changes over time (however it would still not be realistic to suggest that this could be directly attributed to the intervention activity specifically). Given that campaigns and events have a wide reach but likely lower impact, proportionately cross-sectional studies are likely the most cost-effective evaluation method.