ABSTRACT
This chapter explores statistical properties of crime estimates from the Federal Bureau of Investigation’s Uniform Crime Reporting (UCR) program, which has collected counts of criminal offenses from states and law enforcement agencies since 1930. UCR statistics include only crimes known to the police. Law enforcement agencies can only report statistics about crimes that come to their attention. Measurement error for a law enforcement agency’s crime count is the difference between the agency’s count and the count that would be obtained if every crime complaint known to the agency were correctly classified and recorded. Data entry errors, coding errors, and data collection systems affect the estimates. In large police departments, reports often go through multiple levels of review and an error can be introduced at any of those levels. Systematic measurement errors tend to recur in one direction. They cause a statistic to be consistently larger than the true value, making the statistic biased.
