I think this is an illustration of the lean programming / “what’s the simplest thing that could possibly work?” outlook that Mark B talks about in The Tinderbox Way.
If you had a large number (many hundreds, or thousands) of author-and-book combos, it would be worth the programming time and de-bugging to work out an automated system for seeing, in the Attribute Browser, only those authors with multiple books.
But in terms of efficiency and input/output ratios, I bet that an easier way would be either (1) as Mark A says, just ignore the ones with only one book, or (2) create some new Boolean attribute called, for instance, $MultipleBooks. Then when you have an author you know has multiple books – or that you see, in a quick scan in Attribute Browser, has many books – you could use a QuickStamp (or a rule or an agent) to flag that author as “true” for $MultipleBooks. Then it’s very easy to filter the attribute browser by that value, or an agent, or some other device.
Main point: if you have a giant data base, automating this process may be worthwhile. For a smaller set of data, the time it would take just to flag them manually is probably a lot less than the time it would take to work out the automation. FWIW.