A senior researcher at RTP’s Research Triangle Institute has published a recent federal report noting slight inaccuracies in the government’s homicide data. Titled “The Nation’s Two Measures of Homicide,” the report outlines the process in which homicide statistics are collected by two agencies: the Federal Bureau of Investigation, which receives its data from state-level law enforcement agencies; and the Centers for Disease Control and Prevention, which receives its data from death certificates.

Taken together, the measures provide a fairly comprehensive understanding of homicide. But according to criminologist Duren Banks, who co-wrote the report for the U.S. Department of Justice, “Overall, when you’re looking at one or both of these sources, they really have limitations and you can’t just take them at face value.”

The FBI analyzes homicides through its Supplementary Homicide Reports, which are based on data submitted voluntarily by municipal law enforcement agencies. The reports offer detailed information, including victim, offender and weapon characteristics, as well as jurisdiction, date and relationship between offender and victim. But since the reports are voluntary, the total number of homicides included in each year’s report is slightly inaccurate. (The reports also don’t account for federal homicides, which include murders in federal prisons, on military bases and on Indian reservations.)

The CDC analyzes homicides through National Vital Statistics System’s Fatal Injury Reports, which include data derived from death certificates. In homicide cases, medical examiners and coroners typically determine the manner of death, but there are variations on the way they work, and the way in which homicide subcategories are determined. (The FBI’s definition of homicide is the “Willful killing of one human being by another, including murders and nonnegligent manslaughters,” whereas the CDC’s definition of homicide is “Injuries inflicted by another person with intent to injure or kill by any means.”)

Moreover, according to the RTI report, studies suggest that fatal injury reports may undercount American Indian and Alaska Native deaths by 20 to 30 percent, and Asian deaths by 11 to 13 percent.

“Nationally the FBI and CDC measures show really similar trends, but if you look at the state-to-state level, we see a pretty big variation,” says Banks. “There are a lot of discrepancies and things we don’t know about, because we rely on the voluntary reporting from the law enforcement agencies and the medical examiners’ interpretations of cause of death.”