Select Page

The Math and Logic of Bloomberg’s Collective Punishment of minorities.

The Math and Logic of Bloomberg’s Collective Punishment of minorities.

On June 28, 2013, the New York Post reported:

Mayor Bloomberg claimed that people of color should be stopped and frisked more — not less — while whites are stopped too frequently.

“I think we disproportionately stop whites too much and minorities too little. It’s exactly the reverse of what they say,” Bloomberg said on his weekly radio show, in response to the City Council passing two bills aimed at reining in the controversial policing tactic.

“I don’t know where they went to school but they certainly didn’t take a math course. Or a logic course.”

“Disproportionate” is a popular word on a website called Stormfront. Stormfront, for those who don’t know, is a white supremacist watering hole. “Disproportionate statistics” on such fringe sites tend to involve racial/religious/ethnic minorities and are usually anecdotal. Which is to say, the argument is built around something a person saw or heard or thinks they saw or heard. Modern civilization is built on studies and surveys conducted in an organized, scientific manner and published by governments, academics, NGOs, think-tanks, corporations et al. One “disproportional” argument, however, is not entirely anecdotal and is based on two pieces of data from two federal agencies. and appears to carry the weight of the US government behind it. “Appears” is crucially important.

US Census Bureau data says roughly 13% of the US population is Black / African American. Another government entity, the FBI, reports that the nationwide murder rate was 5.3 murders per 100,000 people in 2017; and that in approximately half of all homicides, the perpetrator is black. These individual numbers are combined to arrive at the “disproportionate” statistic, which reads thus: blacks are only 13% of the population, but account for 50% of all murders.

Both parts of the 13-50 “anomaly” (as many genteelly refer to it) are reliable numbers, not made up, or anecdotal. Somebody long ago observed a number in the Census Bureau’s data, saw a factoid in the FBI spreadsheets, and did a mash up to create a powerful, but fundamentally wrong, argument. This hybrid argument is in fact an ‘anecdotal’ observation from two different datasets.  There are ways to combine different studies into a ‘meta study’ but picking one single fact from each is not one of them.

When presented with the 13-50 “anomaly”, many people infer that perhaps blacks are more likely to commit a violent crime than whites.  The Census Bureau also publishes data that shows minorities face higher rates of unemployment, are paid less than their white counterparts, have no disposable income to save, own less property, and… the list goes on, but they seem to play little or no part when crime is discussed at the policy table. Policies crafted carefully to minimize challenges in court, and politically crucial, to avoid charges of racism. Mayor Bloomberg caused jaws to drop when, in 2013, he dispensed with the polite code understood and expected at the policy table, and came out sounding like a person ranting on Stormfront.

But before we get to Mayor Bloomberg’s rant, let’s look at ratio/proportion examples that do not involve race, and see how using a ratio observed in one group and blindly applying it to another group yields absurd results. 

Studies show that same-sex sexual activity in prisons has a higher incidence than in the general population. Leaving aside how much is consensual, coerced, or violent rape, incidence estimates vary from 30% to 70%. Nobody suggests homosexuality occurs at the same rates in the nearly all-male US senate as it does in a prison. The behavior is put in context, understood to arise from situational factors at play in prisons, and that specific sets of social and environmental pressures give rise to certain behaviors and that the ratio can be applied elsewhere where similar social and environmental factors exist, not otherwise.

Let’s take another example. If criminal behavior is traced to skin color/genes, proponents of 13-50 can be forced into some strange corners. Russia has nearly no blacks and its murder rate was 9.2 per 100,000 in 2016. Many times US whites’ murder rate of 2.65 per 100,000 (half of the overall rate of 5.3). Does that mean American whites are genetically better than Russian whites?  Cameroon, in central Africa, with a nearly all-black population, has a murder rate of 1.4 per 100,000. Now what? American whites, 85% more murderous than Cameroonian blacks, are, as a group, genetically more violent than Cameroonian blacks? This nonsense can go on.

When 13-50 creeps into the policy room, there seems to be no attempt to factor in socioeconomic factors that engender crime, such as income and educational levels, family structures, substance abuse, access to jobs, and so on. People who appear to believe that blacks are more genetically prone to crime run for and get elected to office, set policy, hire police forces, and use the power of the state to implement policies that reflect their thinking. This was demonstrated in the streets of New York by former mayor Bloomberg, in the biggest racial profiling machine built in the western world. A juggernaut that processed 5 million people in a city of 8.3 million. Nearly half a million New Yorkers every year for 11 years.

Here is more on Mayor Bloomberg’s thinking in a New York Times report dated June 28, 2013:

“They just keep saying, ‘Oh, it’s a disproportionate percentage of a particular ethnic group,’” he said dismissively of the practice’s critics. “That may be, but it’s not a disproportionate percentage of those who witnesses and victims describe as committing the murder.  …

Aides to Mr. Bloomberg furnished police statistics indicating that 87 percent of police stops were of blacks and Latinos, but that more than 90 percent of murder suspects were identified as being either black or Latino. Nine percent of police stops were of whites, and 7 percent of murder suspects were identified as white.”

Mayor Bloomberg denigrated the math and logic skills of city council members trying to overhaul stop and frisk, contending that his police were looking for people who matched a description provided by the victim or eyewitnesses. What was NYPD reporting, and how does it line up with Mayor Bloomberg’s claims?

NYPD officers filed an extensive report covering details of a stop-and-frisk, including reasons for the stop, demographic information, whether contraband or a weapon was found, if force was used, and so on. Sidebar: people from outside New York assume that a “stop” means police pulled over a vehicle. The average New Yorker does not own a car. The vast majority of stops were by police sitting in patrol cars in minority neighborhoods, observing people doing people things in their neighborhoods, and picking candidates for stop and frisk.

The ten reasons listed that could form the basis for a stop:

Carrying Suspicious Object Actions Indicative Of A Drug Transaction
Fits A Relevant Description Furtive Movements
Casing A Victim Or Location Actions Of Engaging In A Violent Crime
Suspect Acting As A Lookout Suspicious Bulge
Wearing Clothes Commonly Used In A Crime Other

NYPD data shows there were 4.98 million stops from 2003 to 2013. There is no data for 2002, Mayor Bloomberg’s first year in office. Fits A Relevant Description is reported as Y, for Yes, in 867,804 instances. That is 17% of the total 4.98 million stops.

In addition to the above ten reasons for a stop, NYPD could also check off ten Additional Circumstances of a stop:

Report By Victim/Witness/Officer Change Direction At Sight Of Officer
Ongoing Investigation Area Has High Crime Incidence
Evasive Response To Questioning Time Of Day Fits Crime Incidence
Associating With Known Criminals Sights Or Sounds Of Criminal Activity
Proximity To Scene Of Offense Other

NYPD data shows 630,447 instances where Report By Victim/Witness/Officer was checked off. That is 13% of the total 4.98 million stops. 

Fits A Relevant Description and Report By Victim/Witness/Officer were both checked off in 399,725 stops, and adjusting for double-counting, leaves a total 1,098,526 stops where NYPD could possibly have been acting on a description about the race of the suspect. That’s 22% of 4.98 million.

Mayor Bloomberg claimed 90% of murder suspects were reported by victims or witnesses as black or Latino, and those stopped matched the same percentage black or Latino. NYPD data says that police acted on a description (from any source) in at most 22% of cases. What the mayor claimed years later in retrospect does not match the reports being filed by police in (near) real time.  

The only rational conclusion, using basic math and logic, can be that the mayor set a quota of 90% of stops to be black or Latino, and NYPD’s Ray Kelly almost met that quota. Police met the quota they were set and while fulfilling their targets also noted that 22% of those they stopped met some kind of description. “We disproportionately stop whites too much and minorities too little” begins to make sense when you consider that 7% of murder suspects were claimed to be white, and that must have been the target set for NYPD; but whites actually made up 9% of stops, 2% more than the target. Minorities were stopped only 87% of the time, instead of an expectation of 90%, causing hizzoner to get querulous. There is no record of the “data guy” directing NYPD bosses to stop dragnetting so many innocent New Yorkers. Police seemed to be following orders and filling their quotas while also recording that 22% of those they stopped matched any sort of description.

Mayor Bloomberg repeatedly invoked the race of murder suspects, specifically, for his 90% metric. Let’s flesh out what he wanted us to look at with numbers.

Homicides in NYC, 2003 to 2013 (from Wikipedia):

2003 597
2004 570
2005 539
2006 596
2007 494
2008 522
2009 471
2010 534
2011 515
2012 414
2013 332

That’s a total of 5,584 homicides over 11 years. Mayor Bloomberg and his police top brass stopped and frisked 4,984,393 people, in a specific race ratio, in order to catch suspects in 5,584 murders. That’s about 893 innocent people stopped and frisked for each murder, with no basis other than their skin color as a match.

The FBI reports 16,503 murders nationwide in 2003 and 12,253 in 2013. Taking the lower number as representative of each of 11 years from 2003 to 2013, it works out to an estimated 134,783 murders nationwide. If Mayor Bloomberg’s math and logic, and ratio/proportion of 893 stops per murder, were used by the FBI, 120 million Americans would have had to stopped and frisked as potential suspects. What does that kind of disproportionality, compounded by racial targeting, say about Mayor Bloomberg?

A race-based dragnet that catches innocent minorities in the millions raises valid questions, not just of racism, but whether issues of collective responsibility or collective punishment are in play. “Throw them against the wall and frisk them” – the violence in the words betrays an animus that is hard to miss. Arbitrary collective punishment is not alien to the US. Most glaringly, the collective punishment of millions of Iraqis with crippling sanctions, for no known crime that they committed, against anybody.

The New York Times article again:

“They are fabricating outrage over an absolutely accurate comment,” said Marc La Vorgna, the mayor’s press secretary, adding that critics should be labeled “professional purveyors of outrage.” 

That was in 2013. Well after a federal judge characterized Mayor Bloomberg’s program as riven by racial profiling, needing a federal monitor to oversee fixes.  What does it take for a mayor to look at reports produced by his subordinates that show innocent people being targeted in the tens of thousands every month, month after month? What does it take for that mayor to call a halt, pending a review? A moral compass. Decency. Compassion. And, yes, a capacity for outrage.


All numbers pertaining to the stop-and-frisk program under Mayor Bloomberg cited above were extracted from data provided by NYPD here. The relevant R code is here on my Github page.

About The Author

Leave a reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.