Racialized Victim Gender Differences in Capital Decision Making in Pennsylvania

Abstract

The death penalty has historically been used to communicate moral outrage and fear about the murders of some kinds of victims more than others. A large body of research has inquired into the effects of defendant race, and to a lesser extent victim race in capital punishment. However, there is much less research on how victim gender influences capital punishment decisions, and even less research on how victim gender and race might intersect to influence such decisions. This study examines the role of victim gender and race in: prosecutors’ decisions to seek the death penalty, prosecutors’ decisions to retract death filings, and jury or judge decisions to sentence defendants to the death penalty. We utilize detailed data on first degree murder convictions in 18 Pennsylvania counties, 2000–2010. The intersecting gender and race of victims conditioned each of the death penalty outcomes. In addition, whether the victim had children conditioned the effects of victim gender/race.

Keywords:

• Death penalty
• gender disparity
• racial disparity
• sentencing
• prosecutorial discretion

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 There appears to be little research on the effects of victim or defendant characteristics on decision to retract death filings. We could find only two studies; Baldus et al. (1997) found that prosecutors were less likely to retract death filings when the victim was black, while Ulmer et al. 2020a) found the opposite.

2 In the course of collecting data for this study, we had the opportunity for discussions with prosecutors in each county we visited. While this did not constitute a formal qualitative exploration, several interesting themes emerged.

3 Due to the large number of Philadelphia cases, the Philadelphia District Attorney’s Office required that we limited our data collection to cases charged from 2005 to 2010, as a condition of cooperation. This yielded 331 first degree murder convictions, and resulted in the omission of 250 cases prior to 2005.

4 In the case of multiple murder victims, the victim race/ethnicity variable indicates whether any of the victims were white or black. For victim gender, the variable is coded “1” if any victim was female.

5 The following statutory aggravators did not occur with enough frequency to be included: (1) Victim was a firefighter, police officer, prison officer, judge; (2) The defendant was contracted to kill the victim; (3) Victim was held hostage or for ransom; (4) Victim was killed during an aircraft hijacking; (5) Victim was a nongovernmental informant; (6) Defendant had been convicted of voluntary manslaughter; (7) Victim was in third trimester of pregnancy; (8) Defendant was under a restraining order protecting the victim.

6 While it would be valuable to examine the role of defendant gender, we do not have adequate numbers of first-degree murder cases involving female defendants, and do not have adequate variation in death penalty outcomes among those females, to pursue the role of defendant gender further in our analyses. There were also 62 cases with Hispanic defendants, and 87 cases with Hispanic victims in the data. Due to insufficient variation in the many control variables and death penalty decision outcomes among the Hispanic defendant cases, and insufficient gender variation among the Hispanic victim cases, we omit cases with Hispanic defendants and victims from the data. Finally, there were an extremely small number of cases with defendants and/or victims of other races, and these were omitted.

7 Our death sentencing models yield almost identical results for the gender/race of victim coefficients of interest whether we control for the aggravators as independently coded or as filed by the prosecutor. When we use the aggravators as filed, their coefficient become unstable due to lower numbers of cases per aggravator variable.

8 Rosenbaum and Rubin (1985) recommend the standardized difference statistic to assess balance. Standardized difference values between −.20 and .20 are said to indicate acceptable balance (Rosenbaum & Rubin, 1985). In all our propensity score weighted models presented below, standardized difference statistics for the large majority of our covariates were between −.20 and .20, and covariate balance was improved over the raw, unweighted cases. Full balance statistics for all the covariates in our propensity models for each comparison are available on request.

Additional information

Notes on contributors

Jeffery T. Ulmer

Jeffery T. Ulmer is Professor of Sociology and Criminology at Penn State University. His work spans topics such as courts and sentencing, criminological theory and symbolic interactionism, religion and crime, and violent crime rates.

Lily S. Hanrath

Lily S. Hanrath is a Research Fellow at the Robina Institute at the University of Minnesota. Her research interests include racial and gender disparity in criminal justice outcomes, actor decision-making, and the interplay between different divisions of the justice system.

Gary Zajac

Gary Zajac is the founding Managing Director of the Criminal Justice Research Center and Associate Research Professor at The Pennsylvania State University. His research focuses on courts, corrections, sentencing and policing.