Marty A. Davidson II

Book Projects

Working as Intended (with Frank R. Baumgartner and Kaneesha R. Johnson)

***Currently in production at University of Chicago Press***

In this book, we make use of a comprehensive database covering every arrest over seven years in a single state, providing a comprehensive look into a state’s criminal legal system from start to finish. How many people are arrested for which types of crimes? Who are those people: rich, poor, young, old, male, female, black, white, Native American, or some other race, residing in urban, rural, small towns, or a big city? What are they charged with? What are the consequences of these interactions with the criminal justice system? Are individuals punished in similar manners for similar crimes? Do the poor and the wealthy come out of the process with similar outcomes? Who gets a lawyer, who defends themselves, and who is assigned legal assistance by the court system? How often do people go to trial before a jury, and how often are they found innocent at trial? Do those who plead guilty before trial come out better than those who maintain their innocence? What punishments are eventually meted out? Are there systematic or idiosyncratic disparities in who receives harsher or more lenient treatment, and who escapes law enforcement altogether, though they may violate a law?

By describing from head to toe the internal workings of the North Carolina criminal legal system, we seek to understand how the state constructs social orderings. Which groups does the legal code target and which ones does it protect? Which behaviors are outlawed, and which crimes have enhanced punishments depending on the characteristics of the victim? Did the state legislature intend for this targeting (and protection) to occur, or do any observed disparities represent an unintended consequence, a simple artifact of differential behavior? That is, we seek to understand whether observed disparities are mere coincidences, or whether they represent the result that the state legislature had in mind when it outlawed this or that behavior.

Disparate Impact: In a first section consisting of multiple chapters, we focus on documenting the disparate impacts of various parts of the criminal code. We break down the population into more than 50,000 identity-based groups based on race, age, sex, and geographic place of residence and show vast differences in annual rates of arrest. We show how different groups tend to be arrested for different types of crimes. We show how policing patterns lead to tremendous levels of surveillance in some areas but how the police rarely venture into some other neighborhoods. We document that these patterns are consistent with the historic "redlining" patterns of bank loans in the 1930s: Those areas previously described as "undesirable" developed various levels of under-investment, lack of jobs, and concentrated poverty and marginalization, and these are still visible today in high arrest rates, particularly for those who are marginalized by race. In sum, we document various forms of disparate impact in this set of chapters and review the empirical evidence about the court system and how it works. This section of the book relies on analysis of millions of court records and highly sophisticated geo-spatial and statistical analyses.

Discriminatory Intent: Having documented various specific forms of disparate impact, we move in the next group of chapters to evaluate the intent of the legislature when various laws were enacted. We first explain existing scholarly and judicial definitions of discriminatory intent; demonstration of such intent renders a law unconstitutional. We then expand on these definitions to consider the likelihood that vague or overly broad laws would be expected to be implemented by law enforcement in ways that are detrimental to marginalized individuals, particularly Black people. Finally, we assess the implications of inaction by the legislature when evidence of disparate impact is widely known or presented directly to the legislature. Inaction in such a situation when a feasible revision to the law is possible constitutes a signal that the disparity is welcome, not repugnant, to those who control the legislature. In this section we then review historical and archival records to discuss the origins of the traffic code (1937), laws about protesting (1969), capital punishment (1976, 2009, 2011), gangs (1990s), drugs (1980s, 1990s), structured sentencing with its emphasis on prior points and the creation of a class of "habitual felons" (1994), and other examples. In each case, we show that race was a fundamental part of the conversation during the time when the study commissions were evaluating the problem to be solved, that racial differences that would likely lead to disparate outcomes were known to the legislative drafters at the time of crafting, and / or that legislative drafters were in close contact with law enforcement actors, but not with representatives of the communities most likely to be affected by the laws.

The result of this combination of statistically sophisticated and qualitatively rich analysis is a troubling portrait of a system donig exactly what it was designed to do. We present the racial disparities in the criminal legal system not as a flaw, but as the intent of the system. In the end, we propose reforms that could keep us all safter, provide greater confidence in the justice system, reduce the scope of the law, and promote racial equity. Our assessment of criminal legal system makes clear that that system is embedded in other systems, such as housing segregation, economic and educational disparities, transportation and health-care systems, that all contribute to unequal outcomes. While our focus here is on the criminal legal system, any solution to these issues that would be likely to work should also address the broader structural issues of which the criminal legal system is but one part.

Book Website: (Click Here)

Deadly Justice: A Statistical Portrait of the Death Penalty (with Frank R. Baumgartner, Kaneesha R. Johnson, Arvind Krishnamurthy, and Colin P. Wilson)

Deadly Justice is a comprehensive examination of the record established through 40 years of experience with the “new and improved” death penalty. In 1972, the U.S. Supreme Court invalidated all existing death penalty laws in its landmark Furman v. Georgia decision. The Court was particularly concerned about the arbitrary and capricious administration of the penalty. Four years later in Gregg v. Georgia (1976) the Court approved a system with special guidelines to reduce or eliminate the problems earlier identified. The book poses a simple question: Has the modern system worked as intended? Have the states successfully targeted only a narrow class of particularly heinous crimes and the most deserving criminals for the ultimate punishment, or do various elements of caprice, bias, and arbitrariness continue to make the application of the death penalty akin to “being struck by lightning” as the Court noted in Furman?

With chapters focusing on homicides, race and gender dynamics, aggravators and mitigators, geography, reversals, delays on death row, exonerations, methods of execution and botches, last-minute stays of execution, mental illness, public opinion, cost, deterrence, and evolving standards, the book offers a comprehensive overview of our nation’s modern experiment with capital punishment. At a time when other countries have abandoned judicial execution, the U.S. attempted to fix its own deeply flawed system.

The book’s empirical focus provides hard statistical evidence that not only has the modern system retained the vast majority of the issues that concerned the Justices in Furman, but several new problems have arisen as well: cost, botched lethal injections, decades of delay, geographic concentration in just a few jurisdictions, enormous rates of reversal, and last minute stays of execution. Thus, if anything, the modern death penalty not only fails the Furman test, but it scores even worse than the historical death penalty which was declared unconstitutional in 1972. Efforts to repair the system have failed.

Book Website: (Click Here)

Published Peer-Reviewed Articles

Integrating Data Across Misaligned Spatial Units (with Yuri Zhukov, Jason S. Byers, and Ken Kollman)

Abstract: Theoretical units of interest often do not align with the spatial units at which data are available. This problem is pervasive in political science, particularly in subnational empirical research that requires integrating data across incompatible geographic units (e.g., administrative areas, electoral constituencies, and grid cells). Overcoming this challenge requires researchers not only to align the scale of empirical and theoretical units, but also to understand the consequences of this change of support for measurement error and statistical inference. We show how the accuracy of transformed values and the estimation of regression coefficients depend on the degree of nesting (i.e., whether units fall completely and neatly inside each other) and on the relative scale of source and destination units (i.e., aggregation, disaggregation, and hybrid). We introduce simple, nonparametric measures of relative nesting and scale, as ex ante indicators of spatial transformation complexity and error susceptibility. Using election data and Monte Carlo simulations, we show that these measures are strongly predictive of transformation quality across multiple change-of-support methods. We propose several validation procedures and provide open-source software to make transformation options more accessible, customizable, and intuitive.

Published in 2023 in Political Analysis. Article Link: (Click Here)

Working Papers

The Problem of Dynamic Spatial Sampling and Geofence Surveillance (with Jason S. Byers)

Abstract: Geofencing surveillance poses a dynamic spatial sampling problem. Law enforcement must establish geofence perimeters to identify a relevant suspect. This requires identifying a sampling region around a surveillance site and counting the number of intersecting individuals as proxied by geolocation tags. Law enforcement commonly constructs sampling regions with fixed distance intervals or fixed polygon boundaries. This generates privacy concerns as considerations for constructing these perimeters do not factor in the local density of human activity, such as pedestrian flows or traffic patterns. This increases the risk of selective expansion where agencies attempt to extend their data collection beyond what a warrant previously approved. This paper attempts to balance law enforcement's needs for surveillance with individual level privacy by proposing a set of optimal radius estimators. These plug-in estimators use the empirical distribution of human activity patterns to estimate an optimal radius. Given a surveillance site and set of point densities, the optimal radius generates surveillance perimeters that adapt to local conditions. We discuss the implications of applying this estimator to policing surveillance efforts and how law enforcement can use algorithms to better protect the privacy of its citizens.

arXiv Link: (Click Here)