sponsored links
TED@BCG San Francisco

Anne Milgram: Why smart statistics are the key to fighting crime

October 30, 2013

When she became the attorney general of New Jersey in 2007, Anne Milgram quickly discovered a few startling facts: not only did her team not really know who they were putting in jail, but they had no way of understanding if their decisions were actually making the public safer. And so began her ongoing, inspirational quest to bring data analytics and statistical analysis to the US criminal justice system.

Anne Milgram - Criminal justice reformer
Anne Milgram is committed to using data and analytics to fight crime. Full bio

sponsored links
Double-click the English subtitles below to play the video.
In 2007, I became the attorney general
00:12
of the state of New Jersey.
00:15
Before that, I'd been a criminal prosecutor,
00:16
first in the Manhattan district attorney's office,
00:19
and then at the United States Department of Justice.
00:21
But when I became the attorney general,
00:24
two things happened that changed
the way I see criminal justice.
00:26
The first is that I asked what I thought
00:30
were really basic questions.
00:32
I wanted to understand who we were arresting,
00:34
who we were charging,
00:37
and who we were putting in our nation's jails
00:39
and prisons.
00:41
I also wanted to understand
00:42
if we were making decisions
00:44
in a way that made us safer.
00:45
And I couldn't get this information out.
00:48
It turned out that most big criminal justice agencies
00:51
like my own
00:55
didn't track the things that matter.
00:56
So after about a month of being incredibly frustrated,
00:58
I walked down into a conference room
01:02
that was filled with detectives
01:04
and stacks and stacks of case files,
01:05
and the detectives were sitting there
01:08
with yellow legal pads taking notes.
01:09
They were trying to get the information
01:12
I was looking for
01:13
by going through case by case
01:14
for the past five years.
01:16
And as you can imagine,
01:18
when we finally got the results, they weren't good.
01:20
It turned out that we were doing
01:23
a lot of low-level drug cases
01:24
on the streets just around the corner
01:26
from our office in Trenton.
01:28
The second thing that happened
01:30
is that I spent the day in the Camden,
New Jersey police department.
01:32
Now, at that time, Camden, New Jersey,
01:35
was the most dangerous city in America.
01:37
I ran the Camden Police
Department because of that.
01:40
I spent the day in the police department,
01:44
and I was taken into a room
with senior police officials,
01:46
all of whom were working hard
01:48
and trying very hard to reduce crime in Camden.
01:50
And what I saw in that room,
01:53
as we talked about how to reduce crime,
01:55
were a series of officers with a
lot of little yellow sticky notes.
01:57
And they would take a yellow sticky
and they would write something on it
02:01
and they would put it up on a board.
02:04
And one of them said,
"We had a robbery two weeks ago.
02:06
We have no suspects."
02:08
And another said, "We had a shooting in this neighborhood last week. We have no suspects."
02:10
We weren't using data-driven policing.
02:15
We were essentially trying to fight crime
02:17
with yellow Post-it notes.
02:19
Now, both of these things made me realize
02:22
fundamentally that we were failing.
02:24
We didn't even know who was
in our criminal justice system,
02:27
we didn't have any data about
the things that mattered,
02:31
and we didn't share data or use analytics
02:34
or tools to help us make better decisions
02:36
and to reduce crime.
02:38
And for the first time, I started to think
02:40
about how we made decisions.
02:43
When I was an assistant D.A.,
02:45
and when I was a federal prosecutor,
02:46
I looked at the cases in front of me,
02:48
and I generally made decisions based on my instinct
02:50
and my experience.
02:52
When I became attorney general,
02:54
I could look at the system as a whole,
02:56
and what surprised me is that I found
02:57
that that was exactly how we were doing it
02:59
across the entire system --
03:01
in police departments, in prosecutors's offices,
03:03
in courts and in jails.
03:06
And what I learned very quickly
03:08
is that we weren't doing a good job.
03:11
So I wanted to do things differently.
03:14
I wanted to introduce data and analytics
03:16
and rigorous statistical analysis
03:19
into our work.
03:21
In short, I wanted to moneyball criminal justice.
03:22
Now, moneyball, as many of you know,
03:25
is what the Oakland A's did,
03:27
where they used smart data and statistics
03:29
to figure out how to pick players
03:30
that would help them win games,
03:32
and they went from a system that
was based on baseball scouts
03:34
who used to go out and watch players
03:37
and use their instinct and experience,
03:38
the scouts' instincts and experience,
03:40
to pick players, from one to use
03:42
smart data and rigorous statistical analysis
03:44
to figure out how to pick players
that would help them win games.
03:46
It worked for the Oakland A's,
03:50
and it worked in the state of New Jersey.
03:52
We took Camden off the top of the list
03:54
as the most dangerous city in America.
03:56
We reduced murders there by 41 percent,
03:58
which actually means 37 lives were saved.
04:01
And we reduced all crime in the city by 26 percent.
04:04
We also changed the way
we did criminal prosecutions.
04:08
So we went from doing low-level drug crimes
04:11
that were outside our building
04:13
to doing cases of statewide importance,
04:15
on things like reducing violence
with the most violent offenders,
04:17
prosecuting street gangs,
04:20
gun and drug trafficking, and political corruption.
04:22
And all of this matters greatly,
04:26
because public safety to me
04:28
is the most important function of government.
04:30
If we're not safe, we can't be educated,
04:33
we can't be healthy,
04:35
we can't do any of the other things
we want to do in our lives.
04:36
And we live in a country today
04:39
where we face serious criminal justice problems.
04:41
We have 12 million arrests every single year.
04:44
The vast majority of those arrests
04:48
are for low-level crimes, like misdemeanors,
04:50
70 to 80 percent.
04:53
Less than five percent of all arrests
04:54
are for violent crime.
04:56
Yet we spend 75 billion,
04:58
that's b for billion,
05:00
dollars a year on state and local corrections costs.
05:02
Right now, today, we have 2.3 million people
05:06
in our jails and prisons.
05:09
And we face unbelievable public safety challenges
05:11
because we have a situation
05:13
in which two thirds of the people in our jails
05:15
are there waiting for trial.
05:18
They haven't yet been convicted of a crime.
05:20
They're just waiting for their day in court.
05:22
And 67 percent of people come back.
05:24
Our recidivism rate is amongst
the highest in the world.
05:28
Almost seven in 10 people who are released
05:31
from prison will be rearrested
05:33
in a constant cycle of crime and incarceration.
05:35
So when I started my job at the Arnold Foundation,
05:39
I came back to looking at a lot of these questions,
05:41
and I came back to thinking about how
05:44
we had used data and analytics to transform
05:46
the way we did criminal justice in New Jersey.
05:48
And when I look at the criminal justice system
05:51
in the United States today,
05:53
I feel the exact same way that I did
05:54
about the state of New Jersey when I started there,
05:56
which is that we absolutely have to do better,
05:58
and I know that we can do better.
06:02
So I decided to focus
06:04
on using data and analytics
06:05
to help make the most critical decision
06:07
in public safety,
06:10
and that decision is the determination
06:11
of whether, when someone has been arrested,
06:13
whether they pose a risk to public safety
06:16
and should be detained,
06:18
or whether they don't pose a risk to public safety
06:19
and should be released.
06:22
Everything that happens in criminal cases
06:23
comes out of this one decision.
06:25
It impacts everything.
06:27
It impacts sentencing.
06:29
It impacts whether someone gets drug treatment.
06:30
It impacts crime and violence.
06:32
And when I talk to judges around the United States,
06:34
which I do all the time now,
06:36
they all say the same thing,
06:38
which is that we put dangerous people in jail,
06:40
and we let non-dangerous, nonviolent people out.
06:43
They mean it and they believe it.
06:47
But when you start to look at the data,
06:49
which, by the way, the judges don't have,
06:50
when we start to look at the data,
06:53
what we find time and time again,
06:55
is that this isn't the case.
06:57
We find low-risk offenders,
06:59
which makes up 50 percent of our
entire criminal justice population,
07:01
we find that they're in jail.
07:04
Take Leslie Chew, who was a Texas man
07:07
who stole four blankets on a cold winter night.
07:09
He was arrested, and he was kept in jail
07:12
on 3,500 dollars bail,
07:15
an amount that he could not afford to pay.
07:17
And he stayed in jail for eight months
07:20
until his case came up for trial,
07:22
at a cost to taxpayers of more than 9,000 dollars.
07:24
And at the other end of the spectrum,
07:28
we're doing an equally terrible job.
07:30
The people who we find
07:32
are the highest-risk offenders,
07:34
the people who we think have the highest likelihood
07:36
of committing a new crime if they're released,
07:38
we see nationally that 50 percent of those people
07:40
are being released.
07:43
The reason for this is the way we make decisions.
07:45
Judges have the best intentions
07:49
when they make these decisions about risk,
07:50
but they're making them subjectively.
07:52
They're like the baseball scouts 20 years ago
07:55
who were using their instinct and their experience
07:57
to try to decide what risk someone poses.
07:59
They're being subjective,
08:02
and we know what happens
with subjective decision making,
08:03
which is that we are often wrong.
08:06
What we need in this space
08:09
are strong data and analytics.
08:10
What I decided to look for
08:13
was a strong data and analytic risk assessment tool,
08:15
something that would let judges actually understand
08:17
with a scientific and objective way
08:20
what the risk was that was posed
08:23
by someone in front of them.
08:24
I looked all over the country,
08:26
and I found that between five and 10 percent
08:27
of all U.S. jurisdictions
08:29
actually use any type of risk assessment tool,
08:31
and when I looked at these tools,
08:34
I quickly realized why.
08:35
They were unbelievably expensive to administer,
08:37
they were time-consuming,
08:40
they were limited to the local jurisdiction
08:41
in which they'd been created.
08:43
So basically, they couldn't be scaled
08:45
or transferred to other places.
08:47
So I went out and built a phenomenal team
08:49
of data scientists and researchers
08:51
and statisticians
08:53
to build a universal risk assessment tool,
08:55
so that every single judge in
the United States of America
08:58
can have an objective, scientific measure of risk.
09:00
In the tool that we've built,
09:04
what we did was we collected 1.5 million cases
09:06
from all around the United States,
09:09
from cities, from counties,
09:11
from every single state in the country,
09:12
the federal districts.
09:14
And with those 1.5 million cases,
09:16
which is the largest data set on pretrial
09:18
in the United States today,
09:20
we were able to basically find that there were
09:21
900-plus risk factors that we could look at
09:23
to try to figure out what mattered most.
09:27
And we found that there were nine specific things
09:29
that mattered all across the country
09:32
and that were the most highly predictive of risk.
09:34
And so we built a universal risk assessment tool.
09:37
And it looks like this.
09:40
As you'll see, we put some information in,
09:42
but most of it is incredibly simple,
09:45
it's easy to use,
09:47
it focuses on things like the
defendant's prior convictions,
09:48
whether they've been sentenced to incarceration,
09:51
whether they've engaged in violence before,
09:53
whether they've even failed to come back to court.
09:55
And with this tool, we can predict three things.
09:58
First, whether or not someone will commit
10:00
a new crime if they're released.
10:02
Second, for the first time,
10:04
and I think this is incredibly important,
10:05
we can predict whether someone will commit
10:07
an act of violence if they're released.
10:09
And that's the single most important thing
10:11
that judges say when you talk to them.
10:13
And third, we can predict whether someone
10:14
will come back to court.
10:16
And every single judge in the
United States of America can use it,
10:18
because it's been created on a universal data set.
10:21
What judges see if they run the risk assessment tool
10:25
is this -- it's a dashboard.
10:28
At the top, you see the New Criminal Activity Score,
10:30
six of course being the highest,
10:33
and then in the middle you
see, "Elevated risk of violence."
10:35
What that says is that this person
10:37
is someone who has an elevated risk of violence
10:39
that the judge should look twice at.
10:41
And then, towards the bottom,
10:43
you see the Failure to Appear Score,
10:44
which again is the likelihood
10:46
that someone will come back to court.
10:47
Now I want to say something really important.
10:50
It's not that I think we should be eliminating
10:53
the judge's instinct and experience
10:55
from this process.
10:58
I don't.
10:59
I actually believe the problem that we see
11:00
and the reason that we have
these incredible system errors,
11:02
where we're incarcerating
low-level, nonviolent people
11:05
and we're releasing high-risk, dangerous people,
11:08
is that we don't have an objective measure of risk.
11:11
But what I believe should happen
11:14
is that we should take that
data-driven risk assessment
11:15
and combine that with the
judge's instinct and experience
11:18
to lead us to better decision making.
11:21
The tool went statewide in Kentucky on July 1,
11:24
and we're about to go up in a
number of other U.S. jurisdictions.
11:28
Our goal, quite simply, is that every single judge
11:31
in the United States will use a data-driven risk tool
11:33
within the next five years.
11:36
We're now working on risk tools
11:38
for prosecutors and for police officers as well,
11:39
to try to take a system that runs today
11:42
in America the same way it did 50 years ago,
11:45
based on instinct and experience,
11:48
and make it into one that runs
11:50
on data and analytics.
11:52
Now, the great news about all this,
11:54
and we have a ton of work left to do,
11:56
and we have a lot of culture to change,
11:58
but the great news about all of it
12:00
is that we know it works.
12:01
It's why Google is Google,
12:03
and it's why all these baseball teams use moneyball
12:05
to win games.
12:08
The great news for us as well
12:10
is that it's the way that we can transform
12:11
the American criminal justice system.
12:13
It's how we can make our streets safer,
12:16
we can reduce our prison costs,
12:18
and we can make our system much fairer
12:20
and more just.
12:22
Some people call it data science.
12:24
I call it moneyballing criminal justice.
12:26
Thank you.
12:29
(Applause)
12:30

sponsored links

Anne Milgram - Criminal justice reformer
Anne Milgram is committed to using data and analytics to fight crime.

Why you should listen

Anne Milgram is focused on reforming systems through smart data, analytics and technology. She is currently a Professor of Practice and Distinguished Scholar in Residence at New York University School of Law, where she is building a Criminal Justice Innovation Lab, dedicated to using data and technology to transform the American criminal justice system. She also teaches seminars on criminal justice policy and human trafficking. Milgram began her career as a criminal prosecutor, serving in state, local and federal prosecution offices.  She then became the Attorney General of the State of New Jersey, where she served as the Chief Law Enforcement Officer for the State and oversaw the Camden Police Department.

Though her work, Milgram seeks to bring the best of the modern world -- data, technology and analytics -- to bear in an effort to transform outdated systems and practices. Milgram is centered on creating a paradigm shift in how we think about innovation and reform in the criminal justice system and beyond.

Milgram graduated summa cum laude from Rutgers University and holds a Master of Philosophy in social and political theory from the University of Cambridge. She received her law degree from New York University School of Law.

sponsored links

If you need translations, you can install "Google Translate" extension into your Chrome Browser.
Furthermore, you can change playback rate by installing "Video Speed Controller" extension.

Data provided by TED.

This website is owned and operated by Tokyo English Network.
The developer's blog is here.