Doing many-to-many merges

Many-to-many merges have duplicate merge-by values in both the left and right DataFrames. We should only rarely need to do a many-to-many merge. Even when data comes to us in that form, it is often because we are missing the central file in multiple one-to-many relationships. For example, there are donor, donor contributions, and donor contact information data tables, and the last two files contain multiple rows per donor. However, in this case, we do not have access to the donor file, which has a one-to-many relationship with both the contributions and contact information files. This happens more frequently than you may think. People sometimes give us data with little awareness of the underlying structure. When I do a many-to-many merge, it is typically because I am missing some key information rather than because that was how the database was designed.

Many-to-many merges return the Cartesian product of the merge-by column values. So, if a donor ID appears...