By Taylor Smith and Citlalli Florez
WASHINGTON D.C. – The National Academy of Sciences held a panel workshop this past week to explore and measure crime rates during the COVID pandemic—key panelists shared their vital research findings.
John Macdonald, a professor at the University of Pennsylvania and member of the National Academy of Sciences Committee for Law and Justice, said, “We hope this workshop provides a measured discussion of how we measure crime rate changes during the pandemic and the tradeoffs between various metrics,” he introduced.
He explained there has been a dramatic increase in homicide which started at the beginning of the pandemic, and there was a 15-30 percent increase in the number of people on the street at risk of being robbed or assaulted.
NAS has also reported a 27.4 percent increase in crimes of this manner just between 2019 and 2020. Because of these issues, Macdonald explained “given that most homicides are committed with firearms, it is important that we have a firmer understanding of the role of guns.
Before turning over the floor, Macdonald said shootings and homicides tend to be geographically clustered.
Jeff Asher, co-founder and criminal analyst of AH Analytics, said, “We can somewhat track it and, I think, start to understand what it was that potentially caused it by understanding when exactly it happened.”
His findings were consistent with the beginning of the previous presenters. He and his team had measured in 2020 that murder was elevated in the beginning of the year relative to the average, which was the very beginning of the pandemic in America.
Then it was found the rate of murder is dramatically increased in the middle of the year and remains at an elevated rate through the end of the calendar year. Based on these findings, they assumed the most likely case would be that this elevated rate maintained through most of 2021 and then slightly decreased at the beginning of 2022.
After giving the background of statistics, Asher then moved on to discuss the issues in measuring the problem at hand, noting the number of murders committed with a firearm had nationally risen quite dramatically; the rate was low in the 20th century, then hit 77 percent by 2020 and was likely more than 80 percent in 2021.
This is problematic information, though, because according to Asher, there is a lot of uncertainty in this data due to measurement inaccuracies. He elaborated on this fact in a summary of policing data sources.
A summary reporting system is precise relative to other methods, but it is very slow and vague so data that comes out is not available in a timely enough manner to be relevantly impactful.
Asher said the National Incident Based Reporting System—or NIBRS—is also unreliable because it is imprecise and slow, but it is specific. Quarterly Data from the FBI is problematic because it is incomplete and vague, but it is fast.
Because of these downfalls, Asher explained, researchers are extremely limited in their data collection for such an extreme matter. Given that most data is self-reported to these agencies, there is much opportunity for inaccuracies.
In 2021, these very systems produced data with an enormous margin of error; in some states, the range was between 17 percent up and seven percent down.
“It is important that we understand that we have only some insight as to what is happening nationally, but it is so difficult to be accurate on such a large scale,” insisted Asher in response to these findings.
He concluded by reminding listeners that this information sheds light on the fact that until there is a system that properly and accurately measures firearm violence, there cannot be a system that functions well enough to effectively eliminate such violence.
The panel next heard from Jillian Carr, an associate professor of Economics at Purdue University, who presented information on domestic violence crime data. She based her studies in Chicago in order to centralize her findings with more accuracy than she would on a national scale.
She explained domestic violence crimes are extremely under-reported relative to other types of crimes; she offered an explanation for this being that it is unlikely for people to report violence done to them by someone with whom they are in any kind of relationship or living or financial situation.
These circumstances were exponentiated during the pandemic because the stay-at-home order forced people back into their homes during hours when they may typically have been elsewhere.
Carr explained Chicago prioritized domestic violence 911 calls as emergencies that receive an officer immediately. This made it easy for them to measure the number of domestic violence 911 calls increased by about 7.4 percent at the time the stay-at-home order was enacted.
These findings made sense, but then Carr shared more shocking statistics, that even though there were more calls, the number of cases reported by the officers responding to the scene actually decreased by 6.8 percent, and the number of arrests made by those officers at the incident decreased by a dramatic 26.4 percent.
An explanation Carr gave for this was that reports increased due to not only the incapacitation of the families that were already experiencing domestic violence in their own homes, but also the incapacitation of their neighbors; this created more opportunity for witnesses to report the incidents—especially considering most domestic violence incidents happen on weekdays—which in turn greatly increased the report rate.
This fact in combination with the fact that the pandemic was reaching its peak, the City of Chicago apparently asked officers to “reconsider the way they were enforcing certain laws,” so as to protect the safety of themselves and their families when entering other people’s homes; Carr highlighted that this problem, however, is not just a domestic violence issue nor is it just a pandemic issue.
Richard Rosenfeld, Curators’ Distinguished Professor Emeritus, University of Missouri, St. Louis, spoke next on criminology and the pandemic, citing a quotation from Stickle and Felson that the pandemic has been “the largest criminology experiment in history.”
In his study, they used Google cell phone data to understand the dispersion of activities away from the home in correlation with crime. Rosenfeld reported they received mixed results from this study in which they hypothesized that more time at home provides more opportunities for crime which will in turn generate a higher crime rate.
Rosenfeld then touched on the pandemic and robberies. He said that in his team’s research, they did not find significant results at the beginning of the pandemic; in fact, the robbery and larceny rates were both lower relative to pre-pandemic times.
They found that as the hours spent at home increased from Feb. 2020 through the end of the year, unemployment rates increased, closely corresponding to the time at home. Also increasing along with these two factors was the average homicide rate.
“The structural break of significant rise did not begin in the beginning of the pandemic or even in the first week or two. It began in the weekly data in the final week of May and into June,” Rosenfeld reported. “Recall that George Floyd was murdered by Minneapolis police on May 25, 2020.”
Rosenfeld also noted their findings supported the suggestion the pandemic did not have an impact on domestic violence rate, in direct contrast to the findings heard previously from Carr.
He addressed Carr’s contradictory findings by saying, “I would also ask you to consider the consistency or stability, the absence of change in light of Jillian’s comments about what has probably been happening with domestic violence calls and police reporting of those calls during the pandemic.”
The expectation when looking at crime trends and the extent of residential duration during the height of the pandemic, was for crime rates to decrease with more time spent at home. The exception would be for domestic violence.
The panel model results, he said, contained control for seasonal and period effects. There was support for the hypotheses of the decrease in aggravated assault, robbery, and larceny. There was limited support for the decrease in homicide, gun assault, burglary and drug offenses.
Little to no support was given for the hypothesis of the reduction in domestic violence and motor vehicles. These results are based on police data that displayed a rising trend of motor vehicle thefts. The trend did not begin at the beginning of the pandemic but rather closer to the widespread social unrest after the murder of George Floyd in 2020.
There have also been some trend reversals present in the data. Homicide rates from 2022 have been down five percent compared to the first six months of this year and last year. For several cities, homicide rates have been down four to five percent.
He also said, in contrast to robberies and property crimes, during the first six months of this year, robberies were up by 20 percent. In 2020 robberies were going down, therefore there has been a sizable increase in robbery. Larceny was down in 2020 with a significant rise now present in the first half of this year. There has also been an increase in burglaries during the first half this year as well.
Because of the events of recent years, Rosenfeld made another hypothesis that has yet to be studied. “When we began to see these crimes increase at the end of last year, those increases corresponded with some lag with the abrupt and large increases in inflation.”
Incident level crime data is accessed from the open data portals of local police departments. Jeff Asher also used police based data.
Rosenfeld said there are advantages and disadvantages to using the data provided by the police.
He noted an advantage of using such data would be its timeliness factor. Incident-based data can be recorded almost immediately; the data is also comprehensive and covers a large range of crimes in a reasonably large number of cities.
A disadvantage, he added, would be that the data could change over time. Because of the need for timeliness, the data has to be recorded relatively soon after the collection period. However, the data could change over time.
Aggravated assault can be reclassified to homicide, he said. Criminal homicide can be reclassified as justifiable. Data collection from the databases are not regularly updated which means such changes may not be reflected in the data presented for the study.
As a summary of the results there were several things found in the study. Homicides and serious assaults rose immediately after the murder of George Floyd. Robberies and property crimes, except for motor vehicle theft, dropped at the beginning of the pandemic. At the height of the pandemic, robberies and property crime rates rose and homicide rates dropped.
It was found that increases in these offenses were associated with increases in residential duration. Everything else was not equal during the summer of 2020.
COVID-19 reduced police presence and response. There was also a decline in police legitimacy after the murder of George Floyd.
To be able to obtain reliable data, there has to be a reliance on private entities, according to the professor, adding this shouldn’t have to be the case in order to replace federal statistics for crime.
Federal systems are supposedly run like a charity and are based on voluntary contributions, he explained, and said until this has changed, there will be issues around the collection of data for years to come.
He said one of the general themes for each presentation is the discretionary aspect of how crime data is collected. An officer may be pulling back from sharing data, or the agency may lack the capability to consistently report data. Such an occurrence may lead to wrong conclusions.
According to the panelists, it’s difficult to tell how much police recording played a role in declines in robberies, property crime, or larcenies, noting it’s possible that robbery went up, but police recordings may have declined. Citizen reports to the police may have also declined.
Police departments have lost officers which may lead to longer response times, as one panelist explained how the response time in New Orleans went from 50 minutes in 2019 to 145 minutes this year.
As the response time gets longer the crime is allegedly less likely to get reported, said the panel, noting a correlation to the reports by police officers of citizens being gone on arrival. This may create artificial deflating of certain types of crimes.
Finally, the panel suggested declines and willingness to report crime further into the pandemic period and less trust of police may also inflate the effects of the pandemic as well.
