Analysts often find themselves working with less than perfect development and/or validation samples, and data issues typically affect the interpretation of default prediction validation tests.