If the data source was created specifically for this mail merge project, it might contain exactly the records you want and no others. If that's your situation, you can skip this whole section on filtering. (Stick around, though, if you want the records sorted in a certain way.)
Most of the time, however, people do mail merges using generic databases that are designed for more than just a single mail merge. For example, you might draw recipients from your Outlook contact list or a SQL database containing the personnel listings for your entire company. To avoid wasting paper and generating a lot of unneeded copies of the merge, you can apply a filter to the data set so that it contains only the records you want.