What data was *not* collected? Who was not sampled? What variables were not measured? - What made the missing data inconvenient, unavailable, or methodologically difficult? - Could the missing data change the conclusions?