Why Homicide Counts Can Be Misleading

Most public combats between criminal justice reformers and law-and-order conservatives are as stylized as Kabuki theater.

The champion of law-and-order enters, stage-right, and demands more aggressive enforcement, brandishing a hair-raising news story that stars the latest incarnation of Willie Horton.

A reformer steps toward the footlights, deprecates “mere anecdotes,” and unleashes “data-driven” or “evidence-based” findings to argue for de-incarceration or eliminating money bail.

Recently, Thomas Hogan, a veteran prosecutor and an adjunct fellow of the Manhattan Institute, circulated an article, “De-Prosecution and Death: A synthetic control analysis of the impact of de-prosecution on homicides, ” that has been accepted for publication in a peer-reviewed journal, Criminology & Public Policy.

Hogan’s article seems (at least at first) to subvert the data-versus-anecdote tradition.

Hogan, formerly the elected District Attorney of Chester County in Philadelphia’s suburbs, has Philadelphia progressive DA Larry Krasner in his sights. (Hogan had listed Krasner among America’s five worst prosecutors in an earlier article, marshalling familiar tough-on-crime talking points.)

His new article has attracted attention by appropriating the reformers’ own “evidence-based” procedures and applying them to Krasner’s efforts to cut prosecutions for categories of crimes in his city.

Hogan concludes that his numbers show a “causal association” between Krasner’s de-prosecution strategies and a rise in homicides—during one period, an additional 74.9 homicides per year.

According to Hogan, “The public in Philadelphia will have to make a normative choice between a reduction in the number of prosecutions and an increase in homicides.”

From my legal tribesman’s perspective, the mechanism Hogan employs looks like a spitting competition staged on the shores of Cape Cod, its winner determined by measuring mean high tides on the coast of Japan—a Rube Goldberg device.

But constructing a “synthetic Philadelphia” by crunching the numbers on a collection of similar cities which have not elected “progressive prosecutors” is an accepted social science technique.

Although Hogan’s attempt has been criticized, it has survived a peer review.

So, Hogan gets credit for showing his work. Whether he would have chosen to publish it if the results had come out otherwise is, ultimately, beside the point.

The Limits of ‘Policy-Based Evidence’

The article joins publications from all sides of the crime debate that offer what John Roman, a senior fellow at NORC at the University of Chicago, has aptly called “policy-based evidence-making”: the product of “Knowing what you should do and finding evidence to support it—solipsistic, opaque and subjective evidence.”

Hogan’s product suggests that the reformers, accustomed to asserting that “the road to reform is paved by data” had better take account of the possibility that data’s road can run in more than one direction.

Many will see the fact that a dedicated polemicist such as Hogan has elevated his efforts to an “evidence-based” level as a sign of progress.

For me, the answers Hogan offers are less interesting than the question he asks.

The criminal justice system is where the state applies its monopoly on coercive violence. The question posed by Hogan’s study is “How much of that violence do we employ?”

His article offers a stream-lined application of the core ideology of liberal imperialism.

We, Here, work to civilize Them, There. (Hogan had warned against the danger that Philadelphia’s “blight,” encouraged by Krasner’s leniency, could spread to suburban counties like his own.)

We impose Control to allow for  self-government at some later moment (to be determined by Us). The key will be finding a prosecution “sweet spot.” Applying just enough state violence will curtail homicidal private violence in territories populated by the Other.

This sweet spot is located on an axis that stretches between two poles: “no prosecution” and “prosecute everything,” and it constitutes the outcome of a zero-sum calculus. Less prosecution always means more homicides; more prosecution, fewer.

Hogan vs Krasner

There is a macabre contest here between the policies of two white prosecutors—Hogan (Chester County is 84 per cent white) and Krasner (Philadelphia is only 34 per cent non-Latino white)—in which the score will be kept by counting the bodies of urban homicide victims, who are (overwhelmingly) poor, minority residents.

People might argue that the increased prosecutions Hogan advocates are launched for the city’s residents; although others might argue that they are launched at them. You can’t claim the battle is being fought with them.

Each of the “data points” arming Hogan (and his critics) represents a complex and tragic human story, now radically essentialized—reduced to “x kills y.”

From aggregations of these simplified moments a vision emerges: an urban Philadelphia, framed as a territory of fulltime predators and fulltime prey, earning attention only when they are killers or killed.

Maybe this could be explained in part by Thomas Hogan’s personal preoccupations, but it also reflects an endemic vulnerability of any data-centric enterprise that allows itself to systematically de-value narrative.

As University of Richmond law professor Erin Collins cautions in a series of articles, the “evidence-based” paradigm “Reinforces a hierarchy of knowledge and promotes epistemic injustice: only the findings of the researchers matter, while those whose experiences are often the subject of such studies will not be heard.”

The battle between Hogan and his adversaries—especially after it is digested and infiltrated into policy and media debates—cloaks a tacit etiology that is essentially Newtonian. Effects follow inevitably from causes; sequences are linear; time is reversible.

The world’s complexity is reduced to inputs (e.g., prosecutions) and outputs (e.g., homicides).

If Safety is our goal, we have to recognize that we are dealing with swirling conditions and influences that don’t flip switches, but that interact, and then bend probabilities.

Inadvertent Harms

Among the things lost in Hogan’s article is any sense of iatrogenesis. What harms are caused when we multiply prosecutions?

If it is true that “the children do the time with the parents,” how many killers and victims in this generation were impacted by War on Drugs prosecutions in the last one? How many current homicides were influenced by the stigma and unemployability fed by mass incarceration initiatives and promiscuous prosecutions in the past?

What roles did housing discrimination, income inequality, and lack of early childhood education play?

What other interventions do we overlook while we fight over moving the needle up and down the Prosecute/De-prosecute axis?

Why do we assume—as Prof. Erin Collins warns us that the  “evidence-based” paradigm does—a scarcity of other ameliorative tools?

The fact is, there is a broad range of community safety initiatives that could be explored.

Violence isn’t he only weapon we have to cut crime. It turns out that even “greening” vacant lots is influential in crime prevention. Violence is simply the only one we seem willing to discuss in our debates.

 Restoring the Narratives

 Stop for a moment. See an urban homicide as a fully human story.

Read, for example, Jennifer Gonnerman’s account, in “A Daughter’s Death,” of the shooting of a budding New York basketball star, its precursors, and the radiating concentric circles of harms in its aftermath.

You will see how much we miss when a death is reduced to a data point. No quantity of data or statistical technique will fill this gap.

This is not only a moral question of bearing respectful witness. And it is not an argument for living without data.

We have to begin to build the capacity to treat the individual event as a unit of interest. We have to see that simply reducing the issue to the number of prosecutions forces us toward a dead end.

Faced with the question of how “de-prosecution” might lead to increases in homicides, Thomas Hogan is forced to hypothesize: maybe it’s police demoralization, or more contacts between victims and killers, or fewer experienced prosecutors. But examination of individual events would reveal the answers.

There are initiatives, such as the pioneering efforts of the Milwaukee Homicide Review Commission, that show us a way out: collaborative, all-stakeholders’ analyses of particular sentinel events that open up productive areas of inquiry and can address these questions.

These efforts will have to be data-informed. They will supply platforms for interrogating the data. They will include the people most affected. They will identify new areas where data must be harvested.

Hogan’s “De-prosecution and Death” shows that unless data is complemented by a searching interest in the lives of the people experiencing the harms, and of the frontline workers attempting to provide safety, the data alone won’t be much help.

A myopic focus on data will produce harms of its own.

Reading articles that allow people to believe that more public violence is an answer—in fact, is the answer—tells us something about where we are.

“More public violence” is no answer to anything; it only lands us on a treadmill.

We have to recognize that data alone cannot help us move forward. As Tricia Wang has put it, big data needs thick data.

James Doyle

It may be that Thomas Hogan’s achievement in “De-prosecution and Death” is not the one he seemed to hope for—that is, demolishing Larry Krasner with an evidence-based “gotcha”—but illustrating Wang’s point.

If we want to understand the relationship between prosecution—and “de-prosecution”—and homicides, we will have to bring everyone’s scrutiny to bear on specific events, and to understand them in all of their singularity, ambiguity, and complexity.

James M. Doyle is a Boston defense lawyer and author, and a regular columnist for The Crime Report. He welcomes comments from readers.