In the Joining layer data recipe, we discussed that joins only append additional columns to existing features (1:1 or n:1 relationships). Using joins, it is, therefore, not possible to model 1:n relationships, such as "one zip code area containing n schools". These kinds of relationships can instead be modeled using relations. This recipe introduces the concept of relations and shows how you can put them to use.
To follow this exercise, load zip code areas and schools from zipcodes_wake.shp and schools_wake.shp.
Relations are configured in Project Properties. The dialog is very similar to the join dialog:
Define the two layers (Referencing/Child and Referenced/Parent), as well as the fields containing the common values/IDs. As you want to model "one zip code area contains n schools," the zip code dataset is the parent layer and the school dataset is the child layer. The connection between both datasets is established based on the zip code fields...