2.3.3 Cross-sectional vs. event-based datasets

Unlike cross-sectional datasets that are built through the command import, this is not possible through the command import-event. Data extraction at the event level will always result in different numbers of records per individual, and it will make little sense to link such extracts into a common data set. A new dataset must therefore be created for each event-based data extraction (see section 2.3.2).

The purpose of event-organized data extraction is, as mentioned, to make calculations based on events over time through the command collapse. This will transform the event-based dataset into a unit-level data set (one record per individual) with the aggregated statistical measure being the new variable value (measured over the specified time span), allowing the variable to be linked with cross-sectional datasets for further analysis.

Section 2.8 describes the method for linking datasets together.