The union claims police supervisors in the department manipulate crime data to make it appear violent crime has fallen considerably compared to last year.
“When our members respond to the scene of a felony offense where there is a victim reporting that a felony occurred, inevitably there will be a lieutenant or a captain that will show up on that scene and direct those members to take a report for a lesser offense,” Fraternal Order of Police Chairman Gregg Pemberton said. “So, instead of taking a report for a shooting or a stabbing or a carjacking, they will order that officer to take a report for a theft or an injured person to the hospital or a felony assault, which is not the same type of classification.”
The union has been gathering evidence for some time now by looking at reports and talking with officers all over the city, Pemberton said.
Found my way here through the interview with Paul Krugman, and happy to subscribe. In another post, Krugman, noting that the relationship of economic data is far from exact, says, “this is economics not physics.” So too crime data.
Jeff notes that DC tracks serious assaults differently that the FBI defines aggravated assault for UCR. Across river in Virginia, there are degrees of assault, including what generally would be deemed a hate crime that may not meet the UCR definition. With the best of faith, there’s always going to be some sanding and planing in moving information from the 18,000 U.S. law enforcement agencies, mostly local, through the State to the FBI. Arriving in a final fashion to citizens about two years later.
The National Incident Based Reporting System (NIBRS) is intended to improve on UCR, particularly in timeliness, but it’s worth emphasizing that 75% of those agencies have less than 50 officers and the IT infrastructure to fully implement NIBRS has been a fiscal struggle (in GA where I live for example).
Agencies use more real time data, many in variations of NYPD’s CompStat, and many put that up for citizen to view which creates opportunities for what I believe Jeff is about. A citizens’ CompStat as a balance to the common misperceptions about crime in the United States.
Thank you for this incredibly insightful post and for exploring how categorization can result in differing interpretations of the same data. It is thus possible for experts to disagree despite sharing data sets.
I was interested in the contrast between incidents/victims counts. I can see how both are useful units of measurement, though they could, perhaps, produce different counts. One could imagine several victims in a single incident (for example) or am I misreading this difference?
The other interesting point, for me, is the effect that categorizing can have on the conclusions one derives from data sets.
New Orleans’ “Metropolitan Crime Commission” website has the data presented in a logical way to make it easy to analyze. So murders are way down but still at 2019 levels…but they make it clear that it includes the Bourbon Street terrorist attack. So all of the other crime numbers are significantly below 2019 levels which means murders are way below 2019 levels if the terrorist from Houston that was in the Army isn’t included in the 2025 data.
Almost all of the crime figures you present have tripled over the last 10 years, but you suggest that because they’ve fallen in the last few years that they’re meaningfully in decline… compared to what?? And how do current DC crime figures compare to other major metros? Spoiler alert- some of the highest in the entire country by orders of magnitude.
Is there a way to do a GIS map tracking the spread of violent crimes in DC over past decades? Also an overlay on avg housing costs, neighborhood racial demographics? Not enough people cared in the early 90s when ward 7 & 8 were a war zone, now DC crime is international news.
Looking at the newest UCR release it seems like DC and several other states (and individual departments) have started reporting/reporting more consistently Simple Assault (available on the drop down menu of crimes). DC's change is striking, no numbers recorded till late 2016, and then maxing out under 150, until august '21 where the numbers jump to consistently 1k and more.
Some random comparisons, Denver police department has been reporting these numbers to FBI since 2005, and that the NYPD didn't start submitting these numbers to FBI till 2023. I assume each individual department's reporting has to do with the saga of NIBRS transitions.
Does the jump in D.C. lend any credence to the 'book cooking' stories being posted in this comments section and elsewhere? Can Simple Assault be any kind of meaningful indicator despite not being an index crime? Is it too inconsistent of a stat to figure into analysis and/or something that might be more meaningful in the longer term post NIBRS transition?
The union claims police supervisors in the department manipulate crime data to make it appear violent crime has fallen considerably compared to last year.
“When our members respond to the scene of a felony offense where there is a victim reporting that a felony occurred, inevitably there will be a lieutenant or a captain that will show up on that scene and direct those members to take a report for a lesser offense,” Fraternal Order of Police Chairman Gregg Pemberton said. “So, instead of taking a report for a shooting or a stabbing or a carjacking, they will order that officer to take a report for a theft or an injured person to the hospital or a felony assault, which is not the same type of classification.”
The union has been gathering evidence for some time now by looking at reports and talking with officers all over the city, Pemberton said.
There are always going to be discrepancies in the numbers. That's why you use multiple sources. Regardless, there is no data to justify a military crackdown on the city.
The most honest thing to do is to acknowledge the discrepancies, but point out that all of the trends are down or at the very least, plateaued. There's no indication in the data to suggest that DC is going through some unusual crisis in crime necessitating a military occupation. Further, shouldn't the people who actually live in DC have the final say? If DC were out of control with rampant criminal gangs, wouldn't the people be thankful for the intervention? Instead, they are protesting.
Good analysis. If I recall correctly there can be a reporting delay in MPD data releases, but it shouldn’t be more than a few weeks. that doesn’t seem like it would explain the discrepancy
another thing to look out for is the many police forces in DC. usually feds will call up MPD to process. if that’s changed and feds are processing arrests themselves, that may affect reporting. don’t think we’ve seen reporting to indicate that this is the case, but may be something in the
The violent crime total metric is possibly misleading. Assuming that’s the total for the year and not a side by side comparison for the first 8 months -we are only in August so that doesn’t account for the remaining 4 months of crime. A more accurate way to compare would be to divide the total by the months. For 2024 it’s 2116/12 =176.3 per month.
2025 it’s 1572/8 =196.5 per month. This would indicate violent crime is in fact on the rise.
The data divergence began right after Muriel Bowser appointed the recently-hired DEI chief, Pamela A. Smith, to the MPD chief of police position, so she very likely played a part in it.
I’m open to the possibility that the new chief changed reporting policy, and it’s showing up in the stats. Maybe.
I’m not open to the slur that she’s a “DEI chief”. What do you know about this chief’s qualifications or why she was hired? If you jumped to that conclusion because of her identity, that’s way off base - and the suggestion that chiefs of various identities haven’t manipulated statistics is wildly at odds with reality.
Can we stick to individuals, qualifications, and analysis, and try to leave red-meat politics out of it?
I didn't intend for "DEI chief" to be a slur. She was initially hired by MPD in 2022 to fill a newly-created chief DEI officer role, which she worked briefly before getting promoted to assistant chief and then police chief in 2023.
Jeff, you correctly point out that "As explained on DC’s website, “These numbers are based on DC Code Offense definitions and do NOT reflect Part I crime totals…” And then you go off and base your conclusions on the FBI crime totals. But what if MPD is more careful about classifying crimes based on DC Code Offenses -- because that's their statutory metric -- and is less careful about reporting within the FBI definitions?
Except for the year 2009, when for some reason a lot of crimes were not entered into the database, the MPD database CrimeCards is extremely accurate for 2008 to the present. Also, the MPD annual reports, which provide all the DC Code Offense numbers back into the 1960s, are accurate. You have no reason to doubt them.
Now, I know you wanted to talk about DC crime stats because of Trump, but maybe you have no business casting doubt on MPD numbers in the middle of an authoritarian takeover?
He clearly demonstrated that the numbers were roughly in alignment (though universally understated by MPD) for several years, which indicates that there is no classification issue.
The FBI has no reason to falsify or misconstrue DC crime data, and they simply publish the numbers they receive from MPD, so it can't possibly be an inadvertent reporting error by the FBI. The discrepancy is caused by MPD. The question is whether it's deliberate.
Maybe you should get a grip on your TDS so you can start thinking properly.
Or maybe he has every right to cast doubt over all of it because thats the point of this article. The questions is “Is one side or the other being more truthful?”, if so “Why is one reporting different numbers?”.
Putting your head in the sand because you don’t like what one of them could mean is a little obtuse. I’m not saying I’m for what’s happening, I’m just saying that criticizing someone for bringing to attention the discrepancy because you don’t like what Trump is doing is not an intelligent way to go about sharing or reading data and sharing it.
From NBC News in Washington July 18, 2025:
The union claims police supervisors in the department manipulate crime data to make it appear violent crime has fallen considerably compared to last year.
“When our members respond to the scene of a felony offense where there is a victim reporting that a felony occurred, inevitably there will be a lieutenant or a captain that will show up on that scene and direct those members to take a report for a lesser offense,” Fraternal Order of Police Chairman Gregg Pemberton said. “So, instead of taking a report for a shooting or a stabbing or a carjacking, they will order that officer to take a report for a theft or an injured person to the hospital or a felony assault, which is not the same type of classification.”
The union has been gathering evidence for some time now by looking at reports and talking with officers all over the city, Pemberton said.
Found my way here through the interview with Paul Krugman, and happy to subscribe. In another post, Krugman, noting that the relationship of economic data is far from exact, says, “this is economics not physics.” So too crime data.
Jeff notes that DC tracks serious assaults differently that the FBI defines aggravated assault for UCR. Across river in Virginia, there are degrees of assault, including what generally would be deemed a hate crime that may not meet the UCR definition. With the best of faith, there’s always going to be some sanding and planing in moving information from the 18,000 U.S. law enforcement agencies, mostly local, through the State to the FBI. Arriving in a final fashion to citizens about two years later.
The National Incident Based Reporting System (NIBRS) is intended to improve on UCR, particularly in timeliness, but it’s worth emphasizing that 75% of those agencies have less than 50 officers and the IT infrastructure to fully implement NIBRS has been a fiscal struggle (in GA where I live for example).
Agencies use more real time data, many in variations of NYPD’s CompStat, and many put that up for citizen to view which creates opportunities for what I believe Jeff is about. A citizens’ CompStat as a balance to the common misperceptions about crime in the United States.
Thank you!
Thank you for this incredibly insightful post and for exploring how categorization can result in differing interpretations of the same data. It is thus possible for experts to disagree despite sharing data sets.
I was interested in the contrast between incidents/victims counts. I can see how both are useful units of measurement, though they could, perhaps, produce different counts. One could imagine several victims in a single incident (for example) or am I misreading this difference?
The other interesting point, for me, is the effect that categorizing can have on the conclusions one derives from data sets.
New Orleans’ “Metropolitan Crime Commission” website has the data presented in a logical way to make it easy to analyze. So murders are way down but still at 2019 levels…but they make it clear that it includes the Bourbon Street terrorist attack. So all of the other crime numbers are significantly below 2019 levels which means murders are way below 2019 levels if the terrorist from Houston that was in the Army isn’t included in the 2025 data.
Almost all of the crime figures you present have tripled over the last 10 years, but you suggest that because they’ve fallen in the last few years that they’re meaningfully in decline… compared to what?? And how do current DC crime figures compare to other major metros? Spoiler alert- some of the highest in the entire country by orders of magnitude.
Is there a way to do a GIS map tracking the spread of violent crimes in DC over past decades? Also an overlay on avg housing costs, neighborhood racial demographics? Not enough people cared in the early 90s when ward 7 & 8 were a war zone, now DC crime is international news.
Could you add a comparison to other big cities? Only FBI data maybe?
Hey Jeff, long time fan, first time commenter.
Looking at the newest UCR release it seems like DC and several other states (and individual departments) have started reporting/reporting more consistently Simple Assault (available on the drop down menu of crimes). DC's change is striking, no numbers recorded till late 2016, and then maxing out under 150, until august '21 where the numbers jump to consistently 1k and more.
Some random comparisons, Denver police department has been reporting these numbers to FBI since 2005, and that the NYPD didn't start submitting these numbers to FBI till 2023. I assume each individual department's reporting has to do with the saga of NIBRS transitions.
Does the jump in D.C. lend any credence to the 'book cooking' stories being posted in this comments section and elsewhere? Can Simple Assault be any kind of meaningful indicator despite not being an index crime? Is it too inconsistent of a stat to figure into analysis and/or something that might be more meaningful in the longer term post NIBRS transition?
That’s almost certainly just an artifact of the NIBRS transition in each agency.
From NBC News in Washington July 18, 2025:
The union claims police supervisors in the department manipulate crime data to make it appear violent crime has fallen considerably compared to last year.
“When our members respond to the scene of a felony offense where there is a victim reporting that a felony occurred, inevitably there will be a lieutenant or a captain that will show up on that scene and direct those members to take a report for a lesser offense,” Fraternal Order of Police Chairman Gregg Pemberton said. “So, instead of taking a report for a shooting or a stabbing or a carjacking, they will order that officer to take a report for a theft or an injured person to the hospital or a felony assault, which is not the same type of classification.”
The union has been gathering evidence for some time now by looking at reports and talking with officers all over the city, Pemberton said.
But are the numbers real? I see contrarians citing this headline from Channel 4 NBC Washington.
DC police commander suspended, accused of changing crime statistics
There are always going to be discrepancies in the numbers. That's why you use multiple sources. Regardless, there is no data to justify a military crackdown on the city.
I agree, but I need some ammunition to counter the claim about the DC police commander.
The most honest thing to do is to acknowledge the discrepancies, but point out that all of the trends are down or at the very least, plateaued. There's no indication in the data to suggest that DC is going through some unusual crisis in crime necessitating a military occupation. Further, shouldn't the people who actually live in DC have the final say? If DC were out of control with rampant criminal gangs, wouldn't the people be thankful for the intervention? Instead, they are protesting.
Good analysis. If I recall correctly there can be a reporting delay in MPD data releases, but it shouldn’t be more than a few weeks. that doesn’t seem like it would explain the discrepancy
another thing to look out for is the many police forces in DC. usually feds will call up MPD to process. if that’s changed and feds are processing arrests themselves, that may affect reporting. don’t think we’ve seen reporting to indicate that this is the case, but may be something in the
second half of 2025
The violent crime total metric is possibly misleading. Assuming that’s the total for the year and not a side by side comparison for the first 8 months -we are only in August so that doesn’t account for the remaining 4 months of crime. A more accurate way to compare would be to divide the total by the months. For 2024 it’s 2116/12 =176.3 per month.
2025 it’s 1572/8 =196.5 per month. This would indicate violent crime is in fact on the rise.
It is a comparison with the same period in previous years not the full year.
The data divergence began right after Muriel Bowser appointed the recently-hired DEI chief, Pamela A. Smith, to the MPD chief of police position, so she very likely played a part in it.
I’m open to the possibility that the new chief changed reporting policy, and it’s showing up in the stats. Maybe.
I’m not open to the slur that she’s a “DEI chief”. What do you know about this chief’s qualifications or why she was hired? If you jumped to that conclusion because of her identity, that’s way off base - and the suggestion that chiefs of various identities haven’t manipulated statistics is wildly at odds with reality.
Can we stick to individuals, qualifications, and analysis, and try to leave red-meat politics out of it?
I didn't intend for "DEI chief" to be a slur. She was initially hired by MPD in 2022 to fill a newly-created chief DEI officer role, which she worked briefly before getting promoted to assistant chief and then police chief in 2023.
https://www.nbcwashington.com/news/local/dc-police-commander-suspended-crime-statistics/3959566/?utm_medium=email&utm_placement=newsletter&user_id=66c4bb7e5d78644b3a93e62d
Jeff, you correctly point out that "As explained on DC’s website, “These numbers are based on DC Code Offense definitions and do NOT reflect Part I crime totals…” And then you go off and base your conclusions on the FBI crime totals. But what if MPD is more careful about classifying crimes based on DC Code Offenses -- because that's their statutory metric -- and is less careful about reporting within the FBI definitions?
Except for the year 2009, when for some reason a lot of crimes were not entered into the database, the MPD database CrimeCards is extremely accurate for 2008 to the present. Also, the MPD annual reports, which provide all the DC Code Offense numbers back into the 1960s, are accurate. You have no reason to doubt them.
Now, I know you wanted to talk about DC crime stats because of Trump, but maybe you have no business casting doubt on MPD numbers in the middle of an authoritarian takeover?
He clearly demonstrated that the numbers were roughly in alignment (though universally understated by MPD) for several years, which indicates that there is no classification issue.
The FBI has no reason to falsify or misconstrue DC crime data, and they simply publish the numbers they receive from MPD, so it can't possibly be an inadvertent reporting error by the FBI. The discrepancy is caused by MPD. The question is whether it's deliberate.
Maybe you should get a grip on your TDS so you can start thinking properly.
Or maybe he has every right to cast doubt over all of it because thats the point of this article. The questions is “Is one side or the other being more truthful?”, if so “Why is one reporting different numbers?”.
Putting your head in the sand because you don’t like what one of them could mean is a little obtuse. I’m not saying I’m for what’s happening, I’m just saying that criticizing someone for bringing to attention the discrepancy because you don’t like what Trump is doing is not an intelligent way to go about sharing or reading data and sharing it.