| FBI estimates suggest a 4 percent increase in homicides in 2021 over 2020. But this is within the margin of error, and thus not statistically significant. "There were an estimated 22,900 murders in the US in 2021, with a lower bound of 21,300 and an upper bound of 24,600 murders," notes Vox. "The FBI estimates that there were 22,000 murders in the US in 2020, with lower and upper bounds between 21,000 and 23,000. Given the bounds of those estimates, murder — like Erwin Schrödinger's famous cat — could have been up 17 percent or down 7 percent, and there is no way to know for sure which is right."
"Estimation has long been used to produce the official annual reported crime statistics generated by the FBI," the agency points out. But a new data system and the dearth of agencies reporting presented new challenges and required "a new set of statistical procedures."
The result is an FBI crime report with even less reliable data and less conclusive than usual.
In most years, around 95 percent of the population is covered by agencies submitting data. This year, only around 64–65 percent was covered (the FBI gives both figures, in different spots). The discrepancy lies in changes to the way the FBI collects and reports crime statistics.
For decades, the agency collected and reported on data via a widely used Summary Reporting System (SRS) and a lesser-used National Incident-Based Reporting System (NIBRS). Some agencies have used the NIBRS since the '90s, and the proportion using it has steadily increased, but it has always represented a much smaller portion of data than the old SRS system.
This year, for the first time, the FBI's Uniform Crime Report includes only NIBRS data. "As a result, it may seem like crime is rising and it will be tricky, at best, to compare the new data to the previous year for cities that were previously submitting data in the now obsolete format," notes Kelly McBride at Poynter. "It will also be easy for those who want to confuse the public or sow fear by creating disinformation to do so."
Under the old system, departments only counted the most serious crime in a series of related crimes. If someone robbed a store and assaulted the clerk using an illegally owned gun, only the assault would be counted. But under the newer system, police can log up to 10 crimes per incident.
So, comparing NIBRS crime data to SRS data may suggest a "rise" in certain crimes created entirely by new methods of crime reporting.
"The transition from UCR to NIBRS is a setback for crime reporters all around the country," Dylan Purcell, a data reporter at The Philadelphia Inquirer, told Poynter. The new NIBRS data "will be fertile ground for those who want to distort or exaggerate crime trends for political or commercial reasons," suggests McBride. "And, it lays bare a dirty secret about counting crime in the United States: As a nation, we keep horrible, incomplete data that makes it impossible to get an accurate sense of the scope or impact of crime."
For instance, in terms of homicides actually recorded in 2021, we're looking at 13,537 incidents, reported by 11,794 law enforcement agencies using the NIBRS system last year. This is up from 9,630 homicides reported through NIBRS in 2020. But in 2020, only 9,993 agencies submitted NIBRS data. And while agencies reporting homicides in 2021 covered 64-65 percent of the population, agencies reporting in 2020 covered just 53 percent.
In 2019, NIBRS data covered only 47 percent of the population; in 2018, just 40 percent; and in 2017, just 33 percent. In 2012 it was just 29 percent, in 2002 just 20 percent, and in 1992 just 6 percent.
One could accurately say that there has been an enormous spike in homicides recorded through NIBRS data. But this tells us nothing about the number of homicides actually happening.
McBride also notes that "there are no checks and balances on the gathering of this data. No third party routinely audits law enforcement agencies to ensure they are counting crimes accurately. As a result, there are inconsistencies from agency to agency about what counts as reported crimes. And there are numerous examples of police using bureaucratic smokescreens to deliberately manipulate crime numbers, sometimes to make crime seem scarier, sometimes to make crime seem to be less of a problem, sometimes to make it seem like they are solving more crimes, and sometimes to undercount specific types of crimes."
For what it's worth, the FBI says that its estimates show violent crime in 2021 remained "at or near the 2020 level" (a level that was up 5.6 percent over 2019) and "violent and property crime remained consistent between 2020 and 2021."
"The aggregate estimated violent crime volume decreased 1.0 percent for the nation from 1,326,600 in 2020 to 1,313,200 in 2021," it says in a summary. "The robbery rate decreased 8.9 percent from 2020 to 2021, which heavily contributed to the decrease in overall violent crime despite increases in murder and rape rates at the national level." But "it is important to note that these estimated trends are not considered statistically significant by NIBRS estimation methods," the agency states. "The nonsignificant nature of the observed trends is why, despite these described changes, the overall message is that crime remained consistent." The uncertainty in the data means that politicians—who have made crime a big issue in the upcoming election—can basically use the new statistics to support their position—whatever their position is. And while there's little in the new data to support our nouveau crime panic, there's also little to stop the renewed race to appear toughest-on-crime and the rejection of burgeoning criminal justice reform policies. Relatedly: | 
No comments:
Post a Comment