In the U.S., there are two sources of national statistics on crime rates: (i) the FBI's Uniform Crime Reporting Program, which publishes summaries on all crimes reported to police agencies in jurisdictions covering virtually 100% or the population; (ii) the National Crime Survey, based on interviews with a nationwide probability sample of households.

In 2001, 3% of the households in the sample told the interviewers they had experienced at least one burglary within the past 12 months. The same year, the FBI reported a burglary rate of 20 per 1,000 households, or 2%. Can this difference be explained as chance error? If not, how would you explain it? You may assume that the Survey is based on a simple random sample of 50,000 households out of 100 million households.