Comparing death certificate data with FBI crime reporting statistics on U.S. homicides

Both the National Center for Health Statistics (NCHS) Mortality System and the Federal Bureau of Investigation (FBI) Uniform Crime Reporting System measure the numbers and characteristics of homicide in the United States. There are important differences, however, in both the substance and the quality of the information that the two systems collect. The NCHS mortality system reported an average of 9 percent more homicides nationally than did the FBI crime reporting system during the 1976-82 period. Variations did occur in the average ratios of the frequencies of homicides reported by the two systems across age, race, and sex subgroups and geographic areas. The major source of the ascertainment difference between the NCHS and the FBI systems is thought to be incomplete voluntary reporting to the FBI by participating law enforcement agencies and lack of reporting by nonparticipating agencies. The proportions of homicides among corresponding demographic categories in the two systems is remarkably similar despite the difference in ascertainment. This congruence of the distributions of reported homicides supports the idea that inferences drawn from analysis of variables in one of these systems will be valid for the population reported on by the other system.