{ Brandon Buskey, Deputy Director, Criminal Law Reform Project, ACLU - - PDF document
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides The Need and the Path For Reform { Brandon Buskey, Deputy Director, Criminal Law Reform Project, ACLU 1 Overview Why do we need bail reform? How can
Brandon Buskey, Deputy Director, Criminal Law Reform Project, ACLU
Why do we need bail reform? How can we achieve pretrial reform? What role should risk assessments play, if
any?
1 2
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
What can we learn from Washington’s experience with the death penalty?
Death Penalty
Tries to identify
“worst of the worst”
Most serious
consequence after conviction
Pretrial Detention
Predicts “worst of the
worst” risks of danger
Most serious decision
prior to conviction
3 4
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
1972: Supreme Court strikes down death
penalty as
Racist & Arbitrary 1974: Supreme Court reinstates, Today, fewer people on death row, but
still
Racist & Abitrary See Gregory
Trying to replace a system that is Arbitrary: majority of felony
defendants detained because of bail
Discriminatory: African Americans Receive higher bail More likely to be detained on bail
5 6
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
7 8
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
48 hours in jail can lead to lose of: Employment, Housing, Child custody
Causal effect on case outcomes 4X’s more likely to be sentenced to jail 3X’s more likely to receive prison 3X’s longer jail sentence 2X’s longer prison sentence 25% more likely to plead guilty
9 10
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Increases recidivism
More likely to be arrested
following release
Increases failure to appear
Less likely to return to court
Nationally: 1 of 5 people incarcerated are pretrial 2 of 3 people in a local jail are pretrial Washington 7‐8 of 10 people in jail are pretrial Places like King, Pierce, and
Spokane counties
11 12
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Only 1.9% of Felony Defendants re‐
arrested for new violent offense
Baughman and McIntyre, Predicting Violence, Tex.
13 14
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
“In our society, liberty is the norm, and detention prior to trial or without trial is the carefully limited exception.”
“It is better that ten guilty persons escape than that one innocent suffer.” Blackstone 15 16
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Flight Stack v. Boyle Dangerousness US v. Salerno
17 18
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
False Negatives
Mislabeling
someone as “not risky”
False Positives
Mislabeling
someone as “risky”
(Presumption of Innocence) +
(Right to Pretrial Release) = Greater Concern for False Positives?
19 20
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Standard for Guilt
Issue: What you did Proof Beyond a
Reasonable Doubt
Err on the side of
acquittal Standard for Release
Issue: What you
might do
Detention only if
“necessary”
Err on the side of
release
Judges Worry More About False Negatives 21 22
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
“Deadly shooting of Dallas officer reignites bail reform debate”
23 24
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Pretrial feedback is biased toward detention; Over time, detention will naturally increase
“People of color in the United States, particularly young black men, are often assumed to be guilty and dangerous…. As a consequence of this country’s failure to address effectively its legacy
the history that created it have significantly shaped every institution in American society, especially our criminal justice system.”
‐ Bryan Stevenson
25 26
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
27 28
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Limit Detention to
Those charged with serious offenses Whose risk cannot be managed
Everyone Else Goes Home,
Pretrial supports; or Release conditions
Personal
recognizance
Court reminders Voluntary treatment
referrals
Unsecured bond Capable defense
counsel
29 30
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
“Overconditioning”
Increases failure
Conditions that are not
evidence‐based
Drug testing GPS monitoring Secured money bail
Unsecured and other alternatives are
more effective
Denies or delays release Ability to pay is inadequate: People overestimate ability to pay “Second look” is rare and too late
31 32
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
3 Critical Issues
33 34
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Be clear about your goals
Increase release? Reduce bias?
Be sure you can do it “right”
Necessary data and capacity Public input
Be sure you’re meeting goals
No “Zombie Predictions”: Evaluate
pre‐ and post‐implementation data
Feedback from stakeholders
35 36
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Flight and Violence are rare; rare events
are inherently hard to measure
RAI’s don’t measure flight or violence,
but:
Risk of single failure to appear Risk of a single new arrest
What is the failure rate for PSA
violence flag?
Answer: 8.9%
37 38
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
High risk doesn’t mean strict conditions
E.g., High FTA risk better addressed by: Court reminders? Transportation? Treatment?
What can they not do? 39 40
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Not individual risk! Think: Profile
Ex: Person A looks like Group 2
May help with sorting decisions
E.g., release w/o hearing
41 42
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Evidence‐based: robust studies
showing something works
Being based on data is not “evidence based”
We don’t yet know if risk
assessments “work”
More accurate than judges? Less biased than judges? Improve judicial decision‐making?
Megan Stevenson
Another struggle with the Legacy 43 44
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
“There is no such thing
Prof. Sandra Mayson
will carry into the future
make of the past?
45 46
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
What would a risk assessment tell you?
47 48
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
The Pro Publica Debate 49 50
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Compass ProPublica Predictive Parity (True
Positive & Negative)
Same Score = Same % Re‐
arrest
E.g., High risk = 20% re‐arrest
(violence)
Industry standard
False Positive & Positive
Parity
AA’s not re‐arrested 2X more
likely mis‐labeled high risk
Whites rearrested 2X more
likely mislabled “low” risk
The ProPublica disparities are unavoidable
Confirmed by Center for Court Innovation
Why?
AA’s more likely to be arrested Tools more likely to predict AA’s arrests Higher proportion of mistaken predictions
No tool can satisfy ProPublica and Compass
51 52
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
All stem from The Legacy
Disparate police behavior (e.g., marijuana
arrests, “stop‐and‐frisk”)
Structural racism and socioeconomic inequality
53 54
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020
Limit use to those:
Charged with serious offenses
Identified in the highest risk category???
Center for Court Innovation
Disaggregate risk Regular validation with new data
No “zombie predictions”
Test predictive accuracy and false positive rates Drop “High Risk” classifications
Bias in, Bias out Assessing Risk Assessment Predicting Violence Center for Court Innovation
55 56
Defender Initiative 10th Annual Conference Session 3 - MJC Bail REform Pannel BB slides February 28, 2020