Measuring crime. Twice each year, the National Crime Victimization Survey asks a random sample of about 49,000 households whether they have been victims of crime and, if so, the details. In all, nearly 100,000 people answer these questions. If other people in a household are in the room while one person is answering questions, the measurement of, for example, rape and other sexual assaults could be seriously biased. Why? Would the presence of other people lead to overreporting or underreporting of sexual assaults?

AppendixLO1