On Vendettas and the George Floyd Effect

Homicides in Chicago 2018-2024 and in Cook County 1999-2020

May 25, 2025 marked the 5^th anniversary of the infamous encounter between George Floyd and the police in Minneapolis, Minnesota. The George Floyd fetishists dutifully reminded us of this on social media.

Some discussion on social media concentrated on the question of whether the George Floyd spectacle inspired an increase in violent crime, including, most importantly, an increase in black-on-black homicide. There is also some persisting debate about whether there is a “Soros district attorney effect”: Did such homicide increase principally in jurisdictions featuring district attorneys or “state’s attorneys” that had been elected to office on a wave of funding from George Soros?

In this essay, I update analysis about homicides in Chicago and extend analysis to Cook County, Illinois. Cook County encompasses the city of Chicago, but Chicago encompasses most of the population of Cook County. The basic results?:

1. The George Floyd effect was obvious in the earlier analysis. It is still obvious, but …

2. The volume of homicides in Chicago has finally, after five years, exponentially decayed and reverted to pre-George Floyd levels.

3. One can find hints of a George Soros effect in the Cook County data, but they are just hints. I would argue that the statistical results are not strong enough to assert a George Soros effect. Indeed, one surprising result is that …

4. The volume of homicides in Cook County started to increase appreciably in 2015 and especially in 2016. That was just before the “Soros DA” Kimberly Foxx assumed the office of state’s attorney for Cook County in December 2016. In fact, …

5. The volume of homicides had increased discernibly during the tenure of the previous state’s attorney (2009-2016) after having declined discernibly under the tenure of the state’s attorney before that (1999-2008). The volume of homicides had been diminishing after the “Soros DA” assumed office before increasing sharply in the immediate post-George Floyd environment.

6. There is some suggestion in the data that vendettas can explain some of the volume of homicides. Killings in one month inspire killings in the following month.

A more thorough exploration of the George Soros phenomenon would involve examining data from a representative set of jurisdictions. Would a “Soros DA effect” tend to show up in places that had featured Soros DA’s like Philadelphia and Los Angeles but not in other jurisdictions? That makes for work for another day.

In any case, let me very briefly recount the George Floyd context:

Everyone knows about the developments that succeeded the George Floyd spectacle. Among other things, police officer Derek Chauvin was convicted under Minnesota state law for “unintentional murder,” and the Black Lives Matter crowd inspired rioting in Minneapolis and other, major American cities. This was the stuff of the “fiery but mostly peaceful” rioting that resulted in $billions in damage and some non-trivial number of fatalities.

The autopsy report of May 25, 2020 indicated “No life-threatening injuries.” More pointedly, a memorandum of May 26, 2020 from the Hennepin County Attorney’s Office reported that “The autopsy revealed no physical evidence suggesting that Mr. Floyd died of asphyxiation.” These inconvenient conclusions did not discourage the beatification of George Floyd and a summer of rioting and killing in his name. Neither did these conclusions frustrate, according to our friends at the Claremount Institute, “contributions and pledges to the BLM movement” in excess of $80 billion. (!) MacKenzie Scott (formerly MacKenzie Bezos) alone contributed about $1.7 billion to BLM and BLM-adjacent “Organizations Driving Change” as a way of redressing the “heartbreak and horror” of the events of “the first half of 2020.”

Enough of that. On to the data and results.

In earlier essays (1, 2, 3), I recounted the seasonal pattern of homicides in American cities. Historical evidence suggests that this seasonality goes back to the post-bellum South (late 19^th century) and became a thing in the industrialized cities of the North after the first wave of the “Great Migration”—the migration of laborers, both black and white—from the South to the North. The migration really got going with the industrial boom that attended war production during the First World War. It continued through the 1920’s before disappearing during the Great Depression (the 1930’s). Migration picked up again as production to support the war effort during the Second World War also picked up.

The pattern of seasonality is strong. In the warmer months, blacks in certain urban communities tend to socialize in the evenings. This socializing creates opportunities for fraught encounters. Someone makes a stinging joke. Someone else chooses to take offense. That other someone pulls a gun and looses off. A bystander takes a bullet. The incident inspires a “beef” (a vendetta). In the colder months, folks are more likely to stay home. The volume of homicides drops by as much as 50%.

The following graph maps the volume of homicides in Chicago by month for the years 2018 through 2024. I use data compiled by the Chicago Sun Times of “Every homicide victim in Chicago”. The data only go back to 2018, and the data are ostensibly partitioned by race, but I have discovered that the race data can be incomplete, so in this graph I only analyze the total number of homicides.

The solid black line, the “Fitted Total,” indicates a regression line. That regression analysis involved a “non-linear regression” that featured a damped sine wave, a linear trend line, and a discrete George Floyd dummy variable. The dashed line indicates the linear trend, and the discrete jump in the trend indicates the George Floyd effect.

Chicago Results:

1. I cannot look for a “George Soros DA” effect in these data from Chicago, because Kimberly Foxx, the “George Soros DA,” remained the “state’s attorney” for Cook County from December 1, 2016 through to December 1, 2024.

2. That said, one can see an obvious George Floyd effect. In the immediate aftermath of the George Floyd affair, the baseline number of homicides in Chicago in a given month increased from about 40 to 72, an 80% increase.

3. In 2025, the rate of homicides has reverted to pre-George Floyd rates. The monthly rate of homicides may have jumped up discretely post-George Floyd, but the baseline monthly rate did decay by about 6.5 homicides a year. After four years, that would amount to a decrease of about 26 homicides in the the monthly rate. About 12 x 26 = 312 fewer people were victims of homicide in 2024 than would have been the case had the immediate post-George Floyd rates had prevailed.

4. These Chicago data do not exhibit an obvious “vendetta” effect. I did toss a one-month lag of the dependent variable (“total homicides”) into the regression equation, but I get no action from it. All the other coefficient estimates hardly moved with inclusion of the one-month lag.

Now, let’s examine the data from Cook County. Data from 1999 through 2020 are available from the CDC’s “WONDER online databases”. (Many thanks to a reader for suggesting these data to me.)

Here’s a look at the monthly volume of homicides in Cook County. The red line is a 12-month moving average.

These data make contact with the consecutive service of three state’s attorneys, one of whom was a “Soros DA”. Richard Devine served as the state’s attorney for Cook County from 1996 through November 2008. Anita Alvarez served as the state’s attorney from December 1, 2008 through November 2016. “Soros DA” Kim Foxx served as the state’s attorney from December 1, 2016 through to December 1, 2024.

The moving average reveals that homicides had averaged more than 60 per month until 2004. It then declined to about 50 per month. In 2015, however, the volume of homicides started to increase such that, by the end of 2016, it was averaging more than 70 per month. Kim Foxx then took over, but, by late 2017, the baseline rate of homicides had declined. It then declined below 60-per-month until increasing sharply with the George Floyd phenomenon.

In the following graph, I feature a regression exercise that itself features (1) a constant linear trend, (2) dummy variables distinguishing each of the three different state’s attorneys, and (3) a George Floyd dummy variable. The exercise also captures the seasonality of the volume homicides by mapping homicides against the average monthly low temperature. (I did not map a sine wave against these data, because I could not get the regression algorithm to converge.)

The regression line and the underlying, piece-wise linear trend line reveal important structure, but they also miss some important structure.

The regression exercise picks up on the seasonality of homicides and shows a steady decline in the volume of homicides pre-George Floyd. It identifies a discrete jump in homicides post-George Floyd, and it suggests that there was a discrete jump in the volume of homicides with the succession of Kim Foxx to the office of state’s attorney for Cook County. But, note what the regression line misses: It misses the big jump in homicides in 2016. Indeed, it is not obvious that Kim Foxx’s tenure had anything to do with the increase in homicides pre-George Floyd.

To get a better fit to the data, I imposed a more flexible regression equation. This more flexible equation estimated linear trends specific to each of the three state’s attorneys, and it included a one-month lag of the dependent variable (“total homicides”) among the explanatory variables. Note what we get:

The up-and-down regression line does a much better job of capturing the big increase in homicides in 2016. Further, it does not identify a sharp increase in homicides after Kim Foxx assumed office. If anything, homicides started to moderate in 2018 and 2019 before increasing sharply with the George Floyd phenomenon.

Cook County Results:

1. The seasonality of homicides is not very obvious when only considering homicides of non-black (“Other”) victims. It becomes obvious once one includes “Black” homicides in the analysis. Black homicides drive the seasonality.

2. A Vendetta effect: Every four homicides in a given month would inspire about one extra homicide in the following month.

3. On average, homicides were declining at a rate of about 3 per year during the tenure of Richard Devine (1999-2008 in the data).

4. On average, homicides increased at a rate of about 3 per year during the tenure of Anita Alvarez (2009-2016).

5. There is no discernible “Soros DA” effect in that the baseline volume of homicides did not exhibit a statistically-discernible discrete jump during the tenure of Kim Foxx.

6. Further, there is no discernible “Soros DA” effect in that the monthly homicide rate did not exhibit a steady increase after Kim Foxx assumed office. If anything, the results suggest that the monthly rate started to decay over time.

7. The George Floyd effect is obvious with homicides increasing by about 24 per month in the last half of 2020.

What to make of all of this?

I admit some satisfaction to finding some surprises in the data. Surprises help keep one honest with oneself and with others.

I did expect to find a “Soros DA” effect, and, while one could run regression analyses that would appear to weakly support such a thing, there are better regression analyses that would suggest that any effect—if any effect at all—is really too small to tease out. It would be easier to argue that “Soros DA” Kim Foxx inherited a situation that was already aggravated. Then things seemed to be cooling down. Why? Because murderers were doing a good job of killing each other off and there were fewer candidate murderers left standing? Something else? But, then there was the George Floyd business, and the volume of murders increased sharply and immediately. Perhaps the state’s attorney did a bad job of containing the violence, or maybe there was really nothing to do about it.

What else? The volume of homicides in Chicago (and likely Cook County as a whole) seems to have reverted to pre-George Floyd levels by 2024, and one can wonder whether the monthly volume will yet go lower.

I may yet to do some work on CDC data from Philadelphia County and Los Angeles County.