Lethal leverage: false confessions, false pleas, and wrongful homicide convictions in death-eligible cases

ABSTRACT

Substantial attention to wrongful convictions has been developing since the 1990s. This research examines known cases of exonerations and factors that contribute to known wrongful convictions. The issues explored include false confessions, false guilty pleas, and death jurisdiction that impact known wrongful homicide convictions. Literature shows that false confessions can affect relationships with defense attorneys, jury perceptions, other evidentiary analyses, and sentencing outcomes. This research measures the effect of death-eligibility on false confessions and guilty pleas and how those admissions of guilt affect the likelihood of receiving a death sentence. Using data from the National Registry of Exonerations, we estimate probabilities of false confessions, guilty pleas, and death sentences based on multiple predictor variables. Results reveal three major findings. First, African Americans accused of homicide in known wrongful conviction cases, where the homicide victims were mostly female, where false and misleading forensic evidence and official misconduct were present, are statistically most likely to have falsely confessed. Second, guilty pleas were mostly likely produced in death-eligible jurisdictions, net of control factors. And third, death sentences were most likely to have been imposed on male exonerees where the homicide victims were mostly female, where official misconduct was present, and who had falsely confessed.

KEYWORDS:

• Wrongful convictions
• false confessions
• death penalty
• guilty plea

Acknowledgments

The authors wish to thank the anonymous reviewers for their feedback on an earlier draft of this paper, and the journal editors for their assistance throughout this process.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. http://www.law.umich.edu/special/exoneration/Pages/about.aspx.

2. See http://www.law.umich.edu/special/exoneration/Pages/ExonerationsContribFactorsByCrime.aspx for current data.

3. https://www.bjs.gov/content/pub/pdf/p17.pdf.

4. http://www.law.umich.edu/special/exoneration/Pages/about.aspx.

5. https://www.innocenceproject.org/cases/.

6. We also estimated a series of models exploring if gender of victim moderated the association between gender of exoneree and the various outcomes, and race of exonerees and the various outcomes. Research indicates that the gender of the victim (females) is an important predictor of the overall sentence. With offenders who have been accused of hurting female victims often getting harsher sentences (Phillips et al., 2012; Williams et al., 2007). Gender of victim did not significantly moderate the association between gender of exonerees or race of exonerees and the various outcomes. Due to space we do not present result in paper, but they are available upon request.

7. We estimated a series of chi-square analyses to explore if variables were independent of one another. Appendix Table A1 provides the cross-tabs and chi-square for independence for each of the independent and control variables and the outcome false confessions. Appendix Table A2 provides the cross-tabs and chi-square for independence results for each of the control and independent variables and the outcome plea. Finally, Appendix Table A3 provides the cross-tabs and chi-square for independence for each of the independent variables and predictors for the outcome death sentence.

8. To calculate the percent change in the probability of an outcome in logistic regression, we utilized the formula $100(\mathrm{e}\mathrm{x}\mathrm{p}({\beta }_k$)-1), where ${\beta }_k$ is the logistic regression coefficient (not presented but available upon request).

9. Appendix Tables A4 and A5 provide the results for the penalized maximum likelihood logistic regression results. Given that false confession, pleas, and death sentences are all rare events, it is necessary to control for the rarity. The overall conclusions are the same, and the odds ratio are relatively close. The penalized maximum likelihood logistic regression is slightly more conservative compared to the logistic regression both in the odds ratios and p-values.

Additional information

Notes on contributors

Karcin Vick

Karcin Vick (M.A. 2018) is a police officer in the Locust (NC) Police Department. This research is based on her thesis.

Kimberly J. Cook

Kimberly J. Cook is a professor of Sociology & Criminology at the University of North Carolina Wilmington.

Meghan Rogers

Meghan Rogers is an Assistant Professor of Criminology at the University of North Carolina Wilmington.