Jeff Deutch: Profiling (In)justice

preroll music

Herald: Welcome Jeff with a warm applause
on stage. He works for Tactical Tech

applause

and will talk about a bias in
data and racial profiling

in Germany compared with
the UK. It’s your stage!

Jeff: Right. Thank you! Yeah, okay!

My presentation is called
“Profiling (In)justice –

– Disaggregating Data by Race
and Ethnicity to Monitor

and Evaluate Discriminatory Policing”.
In terms of my background:

I’ve done research, doing
mostly quantitative research

around the issues of racial
discrimination for a long time.

In New York, at the Center for
Constitutional Rights I was working on

looking at trends and levels of

use-of-force by police against civilians,
and also on stop-and-search

against civilians. And then more
recently for the last 18 months or so

I’ve been working as a research
consultant at Tactical Tech,

looking at issues of data politics and
privacy. So this is kind of like a merger

of these 2 areas. In terms of what this
presentation is gonna be about:

there’s gonna be 3 takeaways. First, that

we’re dealing with the issues of privacy
and also [freedom from] discrimination.

And both are fundamental human rights.
But there’s tension between the two.

And important questions to think about
are: “When do privacy concerns exceed

or take precedence over those of
discrimination, or vice versa?”

Two: That data is political, both in the
collection and aggregation of data;

but also in terms of having the
categories of being created.

And then, three: That data ethics are
a complex thing, that things aren’t

so black-and-white all of the time.
So what is racial profiling?

The term originates from the US.

And it refers to when a police officer
suspects, stops, questions, arrests or…

you know, or… at other stages (?)
of the communal justice system

because of their perceived
race or ethnicity. After 9/11

it also refers to the profiling of Muslims
or people perceived to be Middle Eastern.

And in German there is no direct translation,
so the term ‘Racial Profiling’ (quotes)

is used a lot in parliamentary hearings
and also in court documents.

So the problem that we’re gonna talk
about is that because of the lack of data

in Germany there’s no empirical
evidence to monitor and evaluate

trends in discrimination.
This is creating problems

for both civil society in terms of looking
at these levels and trends over time,

but also from an individual perspective
it becomes difficult for people

to file complaints. In Germany the only
way to file a complaint officially

is to go to the police department,
which introduces power dynamics,

you know, challenges and additional
barriers. But also if you’re an individual

you have to show that there’s a trend,
right? That you are part of another,

a long standing story. And without this
data it becomes difficult to prove

that that’s happening.
So what we’re needing,

or what some people are calling
for, is having this data

at a state and a sort of national level.
And this ratio that I’m putting here,

referring to policing, is looking at the
rate at which people are stopped

over the census figure of the
demographic share of the population.

And you really need both; the first
being on the police side and

the second being on the census. So
that, you know, if you only have one,

if you only have the rate at which police
were stopping people then you actually

can’t see if this is discriminatory or
not. And if you only have the census

then you can’t see that, either.
So you really need both.

The European Commission, the International
Labour Organisation and academics are all

calling for these… the creation of
standardized and comparable data sets.

And I’m not gonna read these out,
but I can go back to them later

if you’re interested. But what I’m gonna
talk about is comparing the UK

to that of Germany. So in Germany,

in 1983 there was a census; or there
was an attempt to making a census.

But due to wide-spread resentment
and disenfranchisement,

fears of surveillance and lack of
trust in state data collection

there was a big boycott. Or people
deliberately filled in forms wrong.

In some cases there were even
bombings of statistical offices.

Or people spilled coffee over census
forms to try to deliberately ruin them.

As a couple of other presentations at the
conference have already said

this was found to be an
unconstitutional census.

Because of the way that
they were framing it.

Comparing the census to
household registrations.

And so the census was delayed until 1987,

which was the most recent census until
the most recent European one in 2011.

This Supreme Court decision
was really important

because it established this right
for informational self-determination.

Very important for privacy
in terms of Germany.

You know, until today. So what kinds
of information is being collected?

In Germany we have pretty standard kind
of demographic information things

like gender, age, income, religion. But
what I want to talk about in particular

is country origin and country citizenship.

Which are used to determine a person
of migration background. And

this term ‘person of migration background’
generally refers to whether you,

your parents or your grandparents
– the first, second or third generation –

come from a migrant background. Right, and

this term is used oftentimes as a proxy
for ethnic or for racial diversity in Germany.

And this is problematic because
you’re using citizenship as a proxy

for looking at racial and ethnic identity.
And it also ignores the experiences

and identities, the self identities
of people who don’t fall into

this ‘first, second or third generation’,
right? People who may identify

as Black German, let’s say. But
of fourth, fifth or sixth generation.

They’re just ignored in this
data set. So they fall out.

Also, it’s difficult to measure these at
a national level because each state

has different definitions of what
constitutes a migrant background.

So we don’t have this at a national level
but also within states there’s no way

to compare them. Of course, not
having that data doesn’t mean

that there’s no racism, right?
And so in 2005 e.g. we see

that neo-Nazi incidents have increased 25%

– the NSU case coming out but still
going on in court proceedings.

The xenophobic attacks but also the way
in which these crimes were investigated

– at a state and at a federal level –
and the way that it was botched,

in addition to showing that
racism now in general

is at a higher rate than it has been for
the last 30 years. And much more recently

seeing the rise in arson attacks on
refugee centers. There’s been

over 200 attacks this year so far.
You know, all of these showed

that not collecting this data doesn’t
mean that we don’t have a problem.

So, the UK by comparison: In 1981,
there was the Brixton riots,

in an area of London.
And these arose largely

because of resentment towards
the way that police were

carrying out what they called ‘Sus Laws’.
Or people being able to be stopped

on suspicion of committing
a crime, carrying drugs,

having a weapon etc. and so forth.
And so in the aftermath of the riot

they came up with this report called the
‘Scarman report’. And this found

that there is much disproportionality in
the way that Police were carrying out

their stop-and-search procedures.
So for the first time this required…

or one of the reforms that was
instituted was that UK Police started

to have to collect data on race
or ethnicity during the stops.

When they stop a person they have to start
collecting this data. And then you have

a baseline that’s being established.
Around the same time in the UK

we have the 1981 census.

And in society they were having
a lot of debates around

whether or not they wanted to have this…

collecting this baseline national level
(?) figure, because we need these 2 things

for this ratio in order to monitor and
evaluate levels of discrimination.

But, you know, there was
a lot of opposition to this.

And many found it to be (quote)
“morally and politically objectionable”.

But not for the reason you’d think.
People found objections to it

not because of asking these question,
but because of the way that the question

was phrased, with the categories that
were being used. And they did surveys

between ’75 and about ’95, and found that

among marginalized communities
and in minority ethnicity groups

there was actually a lot of support
for collecting this kind of data.

They just wanted to have it phrased to
be different. And so ’91 they started

to collect the data. They put this
‘race question’ in. And here I have,

in 2011 – the most recent census –
some of the kinds of categories

that they wanted to also include.
And they’ve changed over time.

So e.g. like ‘White Irish people’ felt
that they also were discriminated against.

And they experienced things differently
than white British people, e.g.

So having things broken down
further would be helpful for them

in terms of highlighting discrimination
that each specific demographic faces.

So around that time ’91, ’93 we
have the murder of Stephen Lawrence

in an unprovoked racist attack. Nobody
was ever convicted of that. But

what’s important is that we have this
‘Macpherson report’ that came out.

And it developed a lot of recommendations,
70, and most of them were adopted.

One: to be collecting this at a national
level, and to be comparing these.

In 2011 they stopped mandating
that you had to collect this data,

at a national level. So none of the
data from then going forward

can actually be trusted. Some
forces continued to do it,

but not all of them. So you can’t actually
compare them between forces.

In the same year we have these London
riots. The Guardian and LSE put out

a report called “Reading the Riots”. Where
they did a lot of interviews with people

who participated. And they found that
most of the people who participated

had feelings of… that they
were mistreated by Police.

Or that there is racial discrimination
in terms of the policing practices.

That they weren’t being
treated with respect.

So to put some data to that:
Before this was removed

there… it was 2 different types of
stops in the UK. Those PACE stops,

where you stops with reasonable suspicion.

And among that you have e.g. black people
stopped at 7 times the rate of white people.

Asian people – Asian referring to (?)(?)(?)(?)
Southeast Asian in the UK –

at twice the rate. And ‘Section 60 stops’:
where you don’t have to actually have

reasonable suspicion. And when you don’t
need to have that you have much, much

higher rates. 26.6 times the rate of white
people black people are being stopped at.

But the State Department even coming
out and they’re saying: “There’s

no relationship between criminality
and race… criminality and ethnicity”.

In fact it’s like: If people are being
stopped at these rates it’s…

it’s in the wrong direction. You have
white males in particular who are

fending at higher rates. Who are using
drugs at a higher rate. Who are

possessing weapons at a higher rate.
But that’s not who’s being stopped.

There is a connection though between
race and ethnicity and poverty.

So you can see here, they call it like
BAME groups, or ‘Black, Asian and

Minority Ethnicity’. And you can see
that among like wealth and assets:

it’s much, much lower for non-white
households. Unemployment rates

are much higher as well.
Income is much lower.

So I like making maps. And I think
maps are really cool. ’Cause you can

tell stories when you overlay a lot
of data with it. So on the left

I put by different borough in London
where people are actually being stopped.

Per 1,000 people in 2012.
And on the right I put

where the crime is actually occurring.
And this is coming from UK Police.

And so you can see that where people
are being stopped isn’t exactly

where the crime is actually happening.
And if you’re seeing this stop-and-search

as a crime preventing tactic then we
have to question why this isn’t lining up.

Going back to this ratio:

earlier I mentioned like – having the rate
at which one group is being stopped

over that share of the total population.

And we can take it a step further
and we can compare that to…

between different demographic groups.

And when using census figures
combined with police figures,

we can do things like looking on the left.
I mean this disproportionality ratio,

so the rate at which black groups
as a share are stopped

versus the total population, compared
to white groups are stopped.

And you can see the darker areas
is where you have a higher rate.

So when we’re talking about those
‘7 times, or 26 times more likely’

these are those areas that we’re
talking about. And so the darker areas:

you can see that when compared to poverty,

people are stopped… there’s
greater disproportionality ratios

in wealthier areas than there are in
poorer areas. And this is kind of

a way, you could say, almost
of perceiving people of colour

as others who shouldn’t belong in
these areas. It’s also… you can…

when combined with other census
information you can see that you have

more discrimination, you have more
disparities in areas that are more white

and also less racially diverse.

So this is kind of all on the
same kind of a message.

But if it works fine? – It doesn’t.
UK Police is saying that

at most they have a 6%
arrest rate of all stops.

And arrests are not conviction rates.

Looking for weapons we have like less
than 1% of a positive search rate.

And the European Human Rights
Commission e.g. has called for reform

of these practices. The UN has called
for reform of these practices.

And they instituted like
a reform that called for

having a 20% arrest quota. And so that
could either go positively or negatively.

Making a higher quota means that you
could be just arresting more people

that you’re stopping. More likely, or
hopefully it means that you have

a higher justification or grounds
for stopping a person.

So these are the kinds of things you can
do in the UK, with these kinds of data.

In Germany, you can’t. But I wanna
highlight there’s this one case

in Koblenz in 2010.
There was a student of…

unnamed, black student who
is stopped travelling on train,

and who was asked to show his ID.
And he refused. And he said: “No,

I’m not gonna do that. This is
reminiscent of Nazi era tactics”.

And so he was charged with slander.
And he was brought into court.

And the police officer, when it
was in court, said, (quote):

“I approach people that look like
foreigners, this is based on skin colour.”

And so this is for the first time
the police have admitted that

their grounds for doing immigration
related stops are based on

perceived race or ethnicity.
The judge sided with the police.

That this was good justification,
like it was good grounds.

But a higher court ruled
that that wasn’t the case.

They said: “Yeah,
this is unconstitutional,

you can’t actually do it,
it violates the constitution.”

No person shall be favoured or disfavoured
because of sex, parentage, race,

language, homeland, origin,
faith, religious… etc.

Just as a side note there’s been a large
movement to remove this term ‘race’

from that part of the constitution
since it’s been put in.

And also the court dismissed the slander
charge. They said: “No, this student…”

like he’s actually able to critique
the police, you know, in this way.

But after we have the response
by the police union,

the head of the police union
at the time, who said (quote):

“The courts deal with the law in
an aesthetical pleasing way, but

they don’t make sure their judgments
match practical requirements”.

And so what this means is: we see
that according to the police union

– this isn’t official response, but this
is from the Police Union itself –

they say that we need to
profile. We need to do this.

Or else we aren’t able to do
immigration related stops.

That’s crazy. They also…
I mean, at the same time

when they were doing these parliamentary
hearings they institute these mandatory

inter cultural trainings for police
officers. And these are kind of

like a one-day training where
you go and learn all about

how to deal with people from different
cultures. But in some of the interviews

that I was doing they said: “Okay, well,
this isn’t an inter cultural issue.

This is a racism issue”.

People aren’t just coming from other
places. These are Germans,

these are people who grew up here. These
are people who live here. Who know

how to speak the language.
Who were born and raised…

And we need to be dealing
with this in a different way.

However, in this time, we see that
racial profiling has become part of

the national conversation. And so this
is the sticker that somebody put up

in Berlin, in a U-Bahn.
It says: “Attention…,

we practice RACIAL PROFILING while
checking the validity of your ticket”.

It’s not real, but it looks…
I think it’s kind of cool.

When they were doing this in
these Bundestag hearings…

they released data for Federal Police
for 2013. This is the first time

that we have any data that’s released.
No data has ever been released

based on State Police stops.
They say that they’re not actually

collecting the information, so they
don’t have anything to show. Of course

the figures that are released from the
Federal Police are not disaggregated

by race and ethnicity.
But what does this actually show?

So, most of the stops,
over 85% are border stops.

Border being within ca. 30 km
of the German border.

So this is actually taking into account
most of the German population.

But if we’re doing these immigration
related stops then… if we break it down

by offense – in the last two, these are
the immigration related offenses

that people are committing – and
we have less than, at most,

maybe 1% of people who
are found to be a positive,

meaning that they’re found to be violating
some kind of offense. It’s – again,

it’s not a conviction, right?
And people can challenge this.

E.g. like you don’t have to have your
ID on you in all times. You can

present it later, and the
charge can go away.

But if we have such low
rates of positive searches

then like why is this happening? Why
do we feel that with such good data,

and knowing, as good researchers,
why are we continuing this as a practice?

On one of the other interviews that I was
doing they found that okay well:

You know, we know this is ineffective.
But this has the effect of criminalizing

our communities. And
whether or not this is true

is an argument for why we should maybe
have this kind of data to show that

this is or is not actually occurring.
Of course, European Commission

against racism and intolerance and the UN
have said: “Well, even among this at most

1% positive rates these are
not being distributed evenly, and

you have people of certain groups that are
being stopped at rates higher than others,

particularly black and other
minority ethnicity groups.”

Okay, so, going back, why…
into the initial question…

If we have both freedom from
discrimination and the right to privacy

as these human rights how
do we address this tension?

And how do we make sure that we’re
making the right decision in terms of

which takes precedence? And so I came…
or I’ve thought of 3 different reasons

why this isn’t happening. The first
being a series of legal challenges.

Things that are preventing
us from implementing this

from a legal basis. And the first…
you know there’s 3 exceptions

that would allow for this
data to be collected.

(1) The first being if there’s a provision
in EU directive that calls for collecting

this kind of a data. And within that
(2) if you have the consent of the person

the data is subject, let’s say.
Consent is kind of a difficult thing

and we could have a whole conversation
just about that on its own.

If you’re being stopped by police officer
to what extent can you actually consent

to the data that’s being collected?
But this is put in place

as one of the mandatory
legal requirements.

Or (3) if there’s an exception in
the hopefully soon to be finalized

EU Data Protection law that
allows for collecting data

if it’s in the public interest. So you
could argue that we need to be collecting

this data because monitoring
and evaluating discrimination

is a problem that we need
to solve as a society, right?

Two: As a lot of people here at
the conference are talking about:

there’s a lot of distrust in terms
of collecting data by the state.

Particularly sensitive data. But I mean
as many of us are already aware

this data is already being collected. And
this doesn’t mean that we should maybe

collect more just for the
sake of collecting data.

But in terms of sensitive data –

we’re collecting things also like medical
data. And medical data sometimes

is interesting for looking at trends
in terms of the illnesses,

and where illnesses spread. And you can
look at this as also possibly a way of

using sensitive data for highlighting
and monitoring public problems.

And, (3), we have these
challenges in determining

which kind of categories
we should put in place.

But, like the UK, if something
were implemented in Germany

I feel as though this would change over
time as other groups also want their data

to be collected… or not!

So that’s kind of where
I’m at. I think that

there are no easy answers in terms of
whether we should or should not do this.

But I think that at the very least
we should be starting to have

these conversations. And I think that
it’s important to start having these

conversations with communities
themselves who are being targeted,

or feel they’re being profiled.
So, thank you!

applause

Herald: It was an awesome talk. I think
there might be 5 minutes for questions.

There are mics over there and over
there. And whoever has a question,

like in the front rows,
I can come walk to you.

Question: Thank you very much.
I’m just wondering in terms of…

are you sort of creating this…

Jeff: I’m sorry, I can’t hear you…

Question: Sorry, of course… I’m sort
of curious in terms of how you’re

creating the disproportionate demographics
where there will be birth, including

other kinds of information, such as sex,
age, time of day they’re stopped.

Because there’s possibly
unemployment bias as well…

Jeff: I’m sorry, I still can’t
actually hear you.

Question: Sorry… whether it’d be
worth including, say, other details

about people, such as their sex, their
age, maybe the time of day that

these stops are happening. As there may
be a bias towards the unemployed.

If that’s possible, do you think,
with the UK census data?

Jeff: So you’re asking: Do I feel as
though we should also be including

other kinds of demographic data?
Yeah. I mean I do, but I think that

I shouldn’t be the one who’s deciding how
to implement these programs. And I think

that we should be speaking with
the communities themselves

and having them give their opinion. So if
this is something that those communities

who feel that they’re being targeted
or being discriminated against

want to include then I think that they
should be taken into account. But

I don’t know that I should be
the one who’s deciding that.

Herald: Okay, next question
over there, please.

Question: To this ratio you’ve been
talking about: So you compare

census data to – as you
said in the definition

in the first slide –
perceived ethnicity or race.

So it is an attribution of the
persons themselves in a census

compared to attribution per
police officers. And those

won’t necessarily match, I’m not
sure. So I was just wondering

whether you could comment on
that a bit. And this is related

to the second question when it comes
about: We don’t get this data

maybe from the police, because it’s
difficult for the state to collect it.

But maybe we could get the data from
those which suffer from discrimination

in the first place. So do you see any
possibility for public platforms…

So I was reminded of this
idea from Egypt, HarassMap (?)

which is about sexual harassment
of women. That just made visible,

with maps, similar to what you do,
actually where this happened,

when this happened, and how this happened.
But it’s been the people themselves

speaking out and making this
heard. And I was wondering

whether that may be another source of the
data you would be needing for your work.

Jeff: So the first question was talking
about whether we should be using

‘self-identified’ vs. ‘perceived’,
right?

Yeah, I mean they may not line up, right?

People can be perceived in a way
different than they identify.

Some groups in Germany
are calling for both.

They’re calling for kind of like
a two-ticket mechanism

where you have people who
put how they self-identify

and also how the Police are identifying
them. If we’re looking for patterns

of discrimination then it may actually
be more interesting if we’re looking at

how people are perceived.
Then, how people self-identify.

But I think it’s important to take both
into account. And for the second question,

I’m sorry, I kind of forgot what that was.

Question: Like asking the
people themselves for data

when they suffer from discrimination
or [are] being stopped more.

Jeff: Yeah, no, I mean I think that’s a
great idea. And there was a survey

that was actually just done,
that was doing just that!

The findings haven’t been released,
but it just finishes up. And it’s looking

at different types of experiences of
discrimination that people are having.

There’s also organisations like
social worker organisations

that have been collecting
this data for a long time.

Having hundreds and hundreds
of cases. Yeah, thanks!

postroll music

Subtitles created by c3subtitles.de
in the year 2016. Join, and help us!