Transcript Document

Comment on
Ashley Rubin
The Declining Death Penalty in Eighteenth-Century London?
The Role of Ecological Fallacy in Lenience-Based Accounts
Dan Klerman
Conference on Empirical Legal Studies
October 25, 2013
Before 1718
Prosecutions
% of
prosecutions
Death
sentences
Probability of
death penalty
Theft
8,090
74%
1,243
15%
Other crimes
2,885
26%
212
7%
1,455
13%
Total
After 1718
Theft
Other crimes
Total
10,975
Prosecutions
% of
prosecutions
Death
sentences
Probability of
death penalty
29,839
81%
2461
8%
6,811
19%
688
10%
3149
9%
36,650
Theft is vast majority of prosecutions
So decline in probability of death sentence for theft
Drives overall statistics
Masks increase in probability of sentence for other crimes
Relatively small change in case composition
Is it helpful to describe this as ecological fallacy?
Contrast Simpson’s paradox
Simpson’s Paradox
(hypothetical data)
Before 1718
Prosecutions
% of
prosecutions
Death
sentences
Probability of
death penalty
Theft
8,000
80%
1600
20%
Other crimes
2,000
20%
200
10%
1800
18%
Total
After 1718
10,000
Prosecutions
% of
prosecutions
Death
sentences
Probability of
death penalty
Theft
2,000
20%
500
25%
Other crimes
8,000
80%
1200
15%
1700
17%
Total
10,000
Before 1718
Prosecutions
% of
prosecutions
Death
sentences
Probability of
death penalty
Theft
8,090
74%
1,243
15%
Other crimes
2,885
26%
212
7%
1,455
13%
Total
After 1718
Theft
Other crimes
Total
10,975
Prosecutions
% of
prosecutions
Death
sentences
Probability of
death penalty
29,839
81%
2461
8%
6,811
19%
688
10%
3149
9%
36,650
Theft is vast majority of prosecutions, so decline in probability of death sentence for
theft
Drives overall statistics
Masks increase in probability of sentence for other crimes
Relatively small change in case composition
Is it helpful to describe this as ecological fallacy?
Contrast Simpson’s paradox
Is it problematic to say that legal system became more lenient overall?
Control for Case Strength
• Maybe non-theft cases got stronger after 1718
– So, given cases of equal strength, system was more lenient
even in non-theft cases
• Maybe theft cases got weaker after 1718
– So, given cases of equal strength, system got harsher in theft
cases
• Potential controls
–
–
–
–
–
–
–
Value & type of stolen goods
Sex, age & status of offender
Relationship between offender & victim
# of witnesses
Prior convictions
Whether prosecutor or defendant had a lawyer
Literacy
Identify Actors and Mechanisms
• Jury
– Pious perjury increased (only theft?)
– Other ways for juries to affect sentence
• Judge
– Sentencing discretion of judge
• More transportation for theft, but not for other crimes
– More or fewer requests for pardons
– Greater or lesser stringency in administration of benefit of clergy
• Legislators
– Greater percentage of non-theft crimes eligible for death penalty
– Changes in eligibility for transportation and/or benefit of clergy
• Neighbors
– Less petitioning for pardons?
• Prosecutors
– Less charging of capital offenses
Other Questions
• Why focus on pre/post 1718?
– Could test for other breakpoints
– Could test for overall/continuous trend
• Death sentence versus execution?
• Decision to prosecute
– Coroner’s data for homicide
Conclusion
• Important example of benefit of quantitative
analysis of legal history
• Convincing
• Additional analysis could make truly outstanding
– Control for case quality
– Explain which actors & mechanisms were responsible for
increase/decrease in leniency