►
From YouTube: Census Update Meeting | Dec, 3, 2020
Description
Speakers:
James Whitehorne, chief, Census Bureau Redistricting & Voting Rights Data Office
Karin Mac Donald, director, Statewide Database, University of California-Berkeley
A
A
Welcome
everyone.
I
appreciate
you
all
taking
the
time
to
be
with
us
today
to
learn
a
little
bit
more
about
the
census.
In
my
own
life,
when
the
election
was
over
my
personal
friends
and
even
my
boss,
I'm
not
talking
about
tim's
story.
It's
my
other
boss
at
ncsl
said
to
me
boy.
You
must
be
relieved
because
the
election
is
over
and
now
things
are
going
to
settle
down.
That
must
feel
great.
A
Well
you
all
on
this
call
are
my
group
of
folks
who
understand
why
the
election
didn't
mean
a
chance
to
settle
down
and
relax,
there's
more
stuff
coming,
and
it's
the
census
followed
right
by
redistricting.
A
So
I
feel
like
I'm
amongst
friends,
who
only
got
to
take
a
quick
weekend,
breathe
breath
and
then
onward
to
new
things.
So
we
are
fortunate
today
to
have
james
whitehorn
from
the
census
bureau.
He
comes
to
us
frequently
and
I
want
to
say
he
needs
no
introduction,
but
the
introduction
I'll
offer
is
that
I
consider
him
a
friend
and
almost
a
party
he's
he's
adjacent
he's
to
our
program.
A
I
feel,
as
it
was
almost
one
of
us,
and
then
we
have
carl
mcdonald
from
california
who
runs
the
statewide
database
and
she
also
has
been
with
ncsl
through
many
years,
possibly
many
decades.
I'm
not
quite
sure
she
and
I
go
back
for
10
years,
so
we've
got
real
expertise
on
the
call,
and
I
know
many
of
you
have
expertise
too,
and
probably
some
of
you
have
questions.
If
that's
the
case,
then
go
ahead
and
put
them
in
the
chat
box
down
at
the
bottom.
A
We
will
answer
those,
maybe
one
or
two
after
james
talks,
but
mostly
we're
going
to
save
them.
For
the
end
two
more
details,
we
do
have
a
hard
stop
at
one
hour
from
now.
That's
one
o'clock
mountain
time
and
I
do
believe
that
there
are
reporters
on
the
call,
so
I
just
know
that
that's
happening
and
that
we
are
recording
this
and
we
will
make
it
available
to
ncsl
folks
after
the
fact
as
well.
B
All
right,
so
I
I
you
you
made
me
laugh
a
little
bit
there
wendy,
because
I
sometimes
feel
the
same
sort
of
the
feeling
as
the
the
census
is
conducted
and
when
the
collection
operations
come
to
a
close
there's,
a
collective
sigh
of
relief
from
the
people
who
have
to
deal
with
the
field
operations,
and
then
those
of
us
who
deal
with
post-processing
and
data
dissemination
and
and
and
those
types
of
things
are
sort
of
waving
a
little
bit
saying
you
know,
wait
we're
not
done
we're
not
done,
but
that's
sort
of
just
an
internal
thing
from
the
last
couple
years.
B
But
but
I
want
to
thank
both
you
and
christy
for
again
giving
us
this
venue
that
we
can
keep
your
members
informed
of
the
current
activities
of
the
census
bureau,
but
especially
for
those
activities
that
are
involved
around
redistricting
and
voting
rights
data
products,
I'm
going
to
go
through
an
assortment
of
topics
today.
Some
are
new
and
some
are
ones
that
I've
touched
on
before.
B
But
I
wanted
to
kind
of
bring
everyone
up
to
date
on
what
we're
going
to
talk
about,
and,
of
course,
I
I
do
try
to
start
with
whatever
the
sort
of
elephant
in
the
room
at
the
the
time
we're
doing
these
presentations
is-
and
I
know
that
one
of
the
highest
profile
census
to
use
right
now
is
regarding
congressional
apportionment.
B
Unfortunately,
there
is
not
a
lot
that
I
can
say
beyond
what
you've
likely
already
heard.
The
census
bureau
is
committed
to
delivering
the
apportionment
counts
as
close
to
our
statutory
deadline
as
possible
and
we're
working
to
ensure
the
quality
of
the
data
behind
the
apportionment
counts
and
taking
the
time
necessary
to
fix
any
problems
that
have
come
up.
B
We
are
able
to
do
this
work
in
an
expedited
manner,
by
making
adjustments
to
how
this
work
is
performed
and
just
a
couple
of
the
things
that
we've
been
able
to
do
one
is
we
were
able
to
eliminate
a
secondary
check
of
our
master
address
file.
We
have
consistently
tested
the
master
address
file
leading
up
to
and
through
the
census.
B
We
have
modified
how
we
update
the
master
address
file
to
do
an
in
office
address
canvassing,
which
involves
some
testing.
So
there
was,
there
was
enough
testing
that
had
been
done
to
the
master
I
just
filed.
This
secondary
check
didn't
seem
that
it
was
necessary.
B
B
So,
typically,
when
we're
preparing
the
data
for
calculating
apportionment,
there
are
other
activities
that
we
do
that
are
required
of
the
processing
that
are
are
needed
for
the
purposes
of
generating
their
later
redistricting
data,
but
they
don't
they.
They
aren't
required
for
producing.
Just
the
state
level
counts
that
we
need
for
creating
apportionment
numbers,
and
so
what
we've
done
is
we've
sort
of
decoupled.
B
Those
and
I'll
mention
this
later
in
the
pro
in
the
presentation
as
well:
we've
decoupled
those
activities
and
deferred
the
ones
that
are
specifically
just
for
redistricting
to
later
to
lessen
the
amount
of
activities
that
needed
to
be
conducted
to
get
us
to
the
apportionment
counts.
B
They
present
data
on
numbers
of
establishments,
employment,
annual
payroll
sales,
broken
out
by
employment
and
sales
size
of
the
establishment
or
firm,
and
that
data
was
actually
published.
Today
we
have
the
service
annual
survey
or
sas
table
package.
This
provides
statistics
on
revenue,
payroll
sources
of
revenue,
expenses,
exports,
inventory
revenue
from
electronic
sources
and
then
other
selected
industry,
specific
data
for
the
service
industries
and
this
data
set
will
be
published
next
friday.
B
This
new
product
uses
exclusively
existing
administrative
records
and
census
data
to
provide
demographic
characteristics
for
non-employer
businesses
by
geography,
industry,
receipt,
size,
class
and
legal
form
of
organization,
and
then
finally,
we
have
one
that
we're
quite
proud
of
which
is
the
small
business
pulse
survey.
This
is
one
of
the
two
pulse
surveys
that
we're
doing,
which
collects
near
real-time
data
on
the
effects
of
the
coronavirus
pandemic
on
small
businesses.
B
On
the
demographic
side
of
the
census
bureau,
we
have
several
data
products
coming
out
this
month
as
well.
One
that
came
out
yesterday
was
the
americas,
families
and
living
arrangements
table
package,
which
provides
new
estimates
on
the
changing
trends
and
the
number
of
parents
with
children
under
age
18
living
with
them.
The
release
also
highlights
the
similarities
and
differences
in
living
arrangements
among
fathers
and
mothers.
B
B
Another
very
critical
data
set
that's
coming
out
next
week
and
one
that's
important
in
the
world
of
redistricting
and
voting
rights
is
the
latest
five-year
estimates
from
the
american
community
survey.
The
survey
covers
more
than
40
social,
economic
housing
and
demographic
topics
such
as
homeownership
rates
and
costs,
health,
insurance
and
educational
attainment.
B
This
is
the
data
set,
that's
often
used
by
experts
to
predict
apportionment
accounts
and
then
finally-
and
one
of
the
reasons
I
wanted
to
run
through
these
product
releases-
is
the
upcoming
release
of
the
2020
demographic
analysis.
Estimates
on
december
15th,
and
I'm
going
to
talk
about
those
a
little
more
later
in
the
presentation.
B
B
B
B
We
had
about
the
same
percent
of
proxy
respondents
the
last
decade.
If
a
census
taker
couldn't
get
a
response
directly
from
a
household
member.
After
three
visits,
they
tried
to
get
that
information
about
the
address
from
a
knowledgeable
proxy
respondent,
such
as
a
neighbor,
a
landlord
or
a
building
manager,
and
then,
of
course,
for
vacant
addresses.
We
have
to
use
a
proxy
response
because
no
one
lives
at
those
addresses
proxy
responses
are
subject
to
the
same
quality
checks.
B
B
B
A
James,
I
we
don't
hear
you.
Are
you
on
mute.
A
B
Okay
and
then
did
you
hear
about
post
enumeration
survey.
B
B
B
Then.
The
second
measure
that
I
want
to
highlight
a
little
more
is
the
demographic
analysis
estimates,
as
I
mentioned
earlier.
These
are
expected
to
be
published
on
the
15th
of
december.
B
B
These
estimates
were
are
developed
from
historical,
vital
statistics,
estimates
of
international
migration
and
medicare
records,
they're
independent
from
the
2020
census,
and
they
can
be
used
to
calculate
net
coverage
error,
which
is
one
of
the
two
main
ways:
the
census
bureau
traditionally
measures
coverage
of
the
census,
the
deputy
director
of
the
census,
bureau
ron
jarmon
and
the
associate
director
for
demographic
programs
tori
valkov,
along
with
several
other
census
bureau
experts
will
be
doing
a
news
conference
on
this
on
the
day
of
their
release,
and
I
have
that
information
on
this
slide
deck
and
the
slides
I've
shared
with
wendy
and
christy
so
that
they
could
share
them
with
the
attendees
after
the
session
is
over.
B
B
Jurisdictions,
examples
of
the
types
of
tables
that
are
produced
for
the
demographic
analysis,
from
the
2010
census
or
from
from
the
last
decennial
census,
where
the
total
u.s
resident
population
by
age,
the
black
us
resident
population
by
age,
non-black,
u.s,
resident
population
by
age,
total
population
by
asian
sex,
then
the
hispanic
u.s
resident
population
by
age
and
non-hispanic,
and
then
there
were
others.
I
also
included
the
link
here.
B
So
if
people
want
to
go
and
look
and
poke
around
with
those
tables
when
they
get
the
slide
deck,
they
can
jump
right
there
and
take
a
look
as
an
example
of
what
you
can
find
in
those
reports.
This
is
the
table
for
the
total
us
population
by
age.
It
illustrates
the
low
middle
mid,
the
blow,
the
low
middle,
the
middle,
the
high
middle
and
the
high
estimate
for
single
years
of
age,
and
although
the
table
I
have
in
my
slide
is
truncated
just
for
purposes
of
fitting
on
the
screen.
B
In
addition
to
the
post
enumeration
survey
and
the
demographic
analysis,
we
also
assess
the
quality
of
the
census
by
evaluating
how
well
census
operations
were
conducted
by
looking
at
how
the
census
was
conducted,
we
can
get
a
sense
of
the
quality
of
the
results
in
the
coming
months.
The
bureau
will
conduct
extensive
evaluations
of
2020
census
operations
and
will
make
these
available
to
the
public
as
we've
done
each
decade.
B
B
Quality,
so
I'm
going
to
switch
gears
and
talk
a
little
bit
about
our
disclosure
avoidance
plans.
All
the
2020
census
data
products,
with
the
exception
of
apportionment,
will
have
disclosure
of
women's
techniques
applied.
We've
done
this
for
at
least
the
last
several
decades.
The
difference
this
decade
is
that
we're
moving
away
from
our
old
techniques
to
something
called
formal
privacy
or
differential
privacy,
which
injects
noise
into
the
data
to
protect
respondents,
while
still
maintaining
the
data
as
fit
for
use
by
our
data
users.
B
So
yes,
as
I
said
in
my
last
presentation
on
this,
this
does
mean
that
the
redistricting
data
will
have
differential
privacy
applied,
we're
continually
updating
information
about
our
efforts
on
differential
privacy
and
have
a
specific
web
page
dedicated
to
providing
that
information
which
is
provided
in
the
last
link
of
this
bullet
as
well.
So
you
will
get
that
in
the
slide
deck.
B
If
you
want
to
check
that
out
now,
I'm
not
going
to
go
through
each
of
these
entries
today,
because
I've
I've
walked
through
them
in
previous
updates
that
we've
gone
we've
done
together.
But
what
I'm
going
to
point
out
today
is
that
the
latest
materials
we're
providing
to
illustrate
our
progress
as
well
as
point
out
a
report
that
I
had
previously
not
mentioned
so
for
those
who
who
may
not
have
heard
this
last
time.
For
a
brief
background,
we've
been
producing
demonstration
data
products
using
the
2010
census
data
as
its
source.
B
One
important
piece
of
research
that
I
have
not
previously
mentioned
in
past
presentations
is
a
report
that
was
published
on
september
4th.
This
report
takes
several
section.
5
voting
rights
cases
and
20
to
25
runs
of
our
disclosure
avoidance
system
and
uses
that
to
simulate
these
redistricting
plans
for
the
purpose
of
evaluating
the
variability
of
the
underlying
data
coming
out
of
the
differential
privacy
application.
B
There
were
some
additional
enhancements
that
occurred
at
that
time,
but
they
were
mostly
things
like
hardening
of
the
code
to
a
production
code
code
base
so
more
like
back-end
activity,
back-end
cleanup,
the
ppa,
this
ppmf
was
also
converted
to
tables,
and
it's
also
available
on
the
same
link
from
the
previous
slide
for
the
ipums
nhgis
group.
That
has
been
doing
that
work
for
us.
B
Our
most
current
activity
as
far
as
differential
privacy
has
been
the
official
setting
of
the
invariant
parameters
for
the
2020
group.
One
products
group
one
products-
is
sort
of
a
new
term
for
this
decade
because
of
how
we're
organizing
things
as
we
go
through
the
differential
privacy
mechanism
and
so
for
those
of
you
who
aren't
aware
of
what
what's
included
in
that
it's,
the
pl
94171
redistricting
data.
B
It's
the
demographic
profiles
that
come
out
right
after
the
registering
data,
the
demographic
housing
and
characteristic
file,
which
is
not
an
exact
match
to
what
used
to
be
called
summary
file
one.
B
However,
with
the
identified
risks
that
make
differential
privacy
essential
to
protecting
respondent
data,
a
decision
was
made
that
those
could
no
longer
be
reported
unprotected.
Removing
that
constraint
does
not
mean
a
value
won't
be
the
same.
It
just
means
that
it's
not
guaranteed
to
be
the
same,
so
the
invariants
that
have
been
set
for
the
2020
census
group
1
products
are
the
total
population,
but
at
the
state
level
the
total
housing
unit
count
at
the
individual
census
block
level
and
the
number
of
group
quarters
facilities
by
type
at
the
census
block
level.
B
So
this
is
not
the
one
for
the
group
quarters.
I
want
to
just
be
clear.
This
is
not
talking
about
the
population
in
the
group
quarters.
This
is
talking
about.
The
number
of
actual
group
quarters
by
their
type
will
be
held
in
variant
at
the
block
level,
and
I
have
a
little
more
clarity
on
that
on
a
future
slide.
If
people
are
curious
as
to
what
group
quarters
facilities
are
and
what
their
types
are,.
B
So
now
what
comes
next,
in
december,
the
disclosure
avoidance
team
will
conduct
experimental
runs
of
the
system
to
help
evaluate
and
tune
it.
The
experiments
will
examine
how
various
allocations
of
the
privacy
loss
budget
by
geographic
level
and
by
query
set,
would
impact
accuracy
and
fitness
used
for
the
data
from
these
experiments.
We'll
produce
a
new
set
of
metrics
to
allow
our
users
to
assess
the
improvements
made
during
the
test.
B
But
next
in
early
2021,
the
census
body,
that's
responsible
for
making
decisions
about
differential
privacy
parameters.
Our
data
stewardship
executive
policy
committee
will
determine
the
final
privacy
loss
budget
for
the
2020
census
and
its
allocation
toward
the
different
data
products,
including
the
redistricting
data.
B
B
You
know
it's
been
mentioned
publicly
by
the
bureau
that,
in
order
to
deliver
abortion
accounts
as
soon
as
possible,
we
had
to
decouple
some
activities
that
would
have
happened
together
but
were
needed
for
redistricting
and
not
apportionment.
That's
what
I
was
mentioning
earlier
and
when
we
were
talking
about
some
of
the
efficiencies
to
to
be
able
to
compress
the
abortion
and
schedule,
we
will
be
working
to
provide
the
registration
tabulations
as
close
to
their
statutory
deadline
as
possible.
B
For
folks
who
are
still
preparing,
we
do
still
have
the
prototype
data
products
available,
and
the
link
to
them
is
also
on
this
slide
up
under
the
prototype
data
part
of
the
calendar
and
as
soon
as
we
have
dates
and
timelines
for
the
final
tabulation
release.
We'll
make
sure
that
we
communicate
that
wide
and
far
through
ncsl
to
our
official
stakeholders
and
to
the
public.
B
B
And
then
this
group
quarters
population
by
group
quarters
type.
All
of
these
tables
will
be
available
at
the
individual
census
block
level
and,
as
I
mentioned
before,
about
group
quarters
types
what's
being
held
in
variant.
Is
these
types
and
the
number
of
these
group
quarters
in
a
block
will
be
held
in
variant.
So
things
like
correctional
institutes
for
adults,
nursing
homes,
college
universities,
certain
housing,
military
quarters,
those
types
of
things
will
will
not
change
when
we
process
the
data.
B
And
finally,
I
just
wanted
to
revisit
the
citizen
voting
age
by
race
and
ethnicity
or
cvap
product.
The
annual
immune
american
community
survey-based
evap
tabulation,
has
often
been
used
when
analyzing
redistricting
plans
and
we've
been
producing
this
data
set
from
the
acs
every
year
since
2011,
and
I
can
confirm
that
we
will
publish
this
annual
special
tab
from
the
acs
in
early
february.
B
The
smallest
unit
of
geography
in
this
will
be
block.
Group
it'll,
be
the
2010
block
groups,
because
it's
coming
from
that
acs
five-year
estimate
that's
being
released
next
week
and
that
data
set
uses
the
2010
statistical
geography
and
then
the
other
cvap
that
you've
heard
about
this
decade
or
for
this
decennial,
the
one
based
on
the
2020
census
and
using
administrative
records
for
the
citizenship
characteristics.
B
It's
the
schedule
for
that
is
still
under
development,
because
much
like
redistricting
is
hooked
to
the
end
of
apportionment.
The
cvap
from
administrative
records
needs
the
redistricting
data
schedule
to
be
able
to
start
working
on
their
production
timeline,
so
they're
sort
of
in
the
same
boat.
We
are
with
redistricting
waiting
for
a
few
of
those
inputs.
B
B
A
Well,
the
first
thing
I
want
to
do
is
say
that
it
always
amazes
me
how
many
products
the
census
produces.
Our
group
tends
to
think
about
two,
your
reapportionment
data
and
your
pl94
171.
Those
of
us
who
are
more
advanced
remember
that
there's
some
geographic
information
that
comes
before
that,
but
you've
shared
with
us
many.
I
bet
you've
mentioned
two
dozen
different
products
are
coming
out
from
the
bureau.
A
So
that's
a
lot
of
productivity,
we're
seeing
there
there's
one
question
from
or
adeline
mentioned
that
maybe
the
reason
the
housing
units
are
held
in
variant
is
because
they
were
released
some
time
ago.
Could
you
just
address
that
real,
quick
and
then
I'm
going
to
turn
it
over
to
karen.
B
I
I
don't
know
if
that's
the
sole
reason
I
mean
the
the
number
of
housing
units
is
an
indicator
that
is
used
in
our
account
question
resolution
project
when
people
appeal,
if
they,
if
they
feel
that
there's
an
issue
with
coverage,
errors
or
boundary
errors
in
their
data.
They
also
need
to
know
the
number
of
housing
units
to
be
able
to
make
that
appeal.
B
I
think,
because
we
in
some
of
our
interactive
update
programs,
that
we
did,
we
released
the
number
of
housing
units,
so
that
may
be
part
of
what
they're
talking
about
from
mike
ratcliff
and
then
the
new
mexico
data
users
conference.
I
don't
have
a
solid
answer
for
that,
but
I
know
that
those
are
two
of
the
considerations.
A
All
right-
and
I
am
going
to
ask
hanse's
question
while
you're
still
with
us,
because
I
think
it
might
be
super
easy
and
that
is
do
you
know
if
you're
going
to
be
doing
a
rolling
release
on
the
state
data,
or
will
it
be
all
at
once,
which
was
what
we'd
heard
the
last
time
it
was
likely.
B
Well,
I
still
am
the
plan
that
I'm
I'm
going
to
put
forward
to
census
executives
for
approval.
As
soon
as
we
lock
down
the
last
two
unknowns
that
we
have
to
to
do
to
finish,
that
schedule
we'll
have
the
single
national
release
of
data
and
then
it'll
be
up
to
them
to
either
approve
that
that
plan
or
we
will
have
to
readjust
that
plan.
A
C
A
And
say
that
I
know
that
your
assignment
was
to
respond
to
james,
but
I
think
you
can
do
more
than
just
respond.
So
would
you
take
it
away?
Please.
C
Yes,
good
morning,
everybody
thank
you
very
much
for
inviting
me.
Actually,
I
had
two
assignments.
My
first
assignment
was
to
put
a
powerpoint
together,
and
then
I
had
another
one
that
talked
about
responding
to
james.
So
by
that
point
I
already
had
my
powerpoint,
and
so
I
might
as
well
share
it
with
you.
I
would
like
to
caveat
this
presentation
by
saying
that
this
is
going
to
be
less
technical.
C
This
is
a
user's
perspective
and
I
thought
I
might
break
down
some
of
the
terminology
that
james
has
been
talking
about
and
I'll
go
through
some
of
the
things
that
james
was
saying.
So
just
one
second,
please.
While
I
share
my.
C
Screen,
okay,
so
anticipating
the
2020
census
data.
I
was
struggling
a
little
bit
with
with
the
title
for
this
talk.
I
initially
wanted
to
call
it
things
that
keep
me
up
at
night,
and
here
they
are
I'd
like
to
talk
about
four
things
just
in
general,
and
they
are
data
quality
issues
with
census
disclosure
avoidance
system.
C
When
can
we
finally
get
started,
because
that
is
something
that
everybody
is
wondering
and
then
talk
about
other
data,
so
this
is
going
to
be
pretty
quick
from
a
user
perspective.
Why
are
we
concerned
about
data
quality?
Let's
remember
what
happened
in
this
census.
I
mean
we
all
know
that
census
is
already
an
incredibly
difficult
undertaking.
C
One
of
the
concerns
that
many
states
had
were
related
to
the
fact
that
six
percent
of
the
overall
population,
according
to
the
fcc,
lacks
broadband
access
and
in
rural
areas
that
is
more
pronounced,
so
that
is
25
percent.
We
also
know
that
there
are
differences
in
internet
access
for
minority
populations
and
so
forth,
so
there
was
a
concern
that
was
translated
by
many
or
responded
to
by
many
states,
in
particular
california.
C
There
were
also
concerns
about
things
like
unsecured
computers,
potential
hacking,
phishing
and
so
forth,
the
kind
of
stuff
that
just
happens
or
could
happen
potentially
when
lots
of
people
are,
you
know,
accessing
a
particular
site
and
there
are
lots
of
people
sitting
out
there.
That
would
like
to
take
advantage
of
that.
Then
we
had
covet,
of
course,
that
threw
a
wrench
into
all
of
the
operations
and
then
moving
on
there
were
various
issues
reported
that
had
to
do
with
data
collection
and
data
assignments
that
were
pushed
to
enumerators
via
apps.
C
We
saw
a
lot
of
this
in
the
media
if
you're
looking
for
you
know,
reports
from
actual
enumerators
and
so
forth.
This
has
also
translated
in
some
of
the
lawsuits
already
and
then
there
are
questions
about,
for
example,
student
populations
that
are
on
the
move
and
duplications
and
how
the
census
bureau
would
deal
with.
Duplications.
C
Just
remember
that
census
day
is
april,
1
of
the
year
ending
with
zero
and
if
you're,
looking
at
the
timeline
of
covid
and
the
fact
that
a
lot
of
students
were
in
student
housing
at
that
point
and
then
had
to
move
out,
many
moved
back
to
their
parents.
There
was
a
lot
of
confusion
about
where
the
census
forms
should
be
filled
out.
C
You
know,
people
that
deal
with
student
populations
can
talk
about
this
for
days
about
how
universities
are
not
giving
complete
data
sets
to
senses
to
be
able
to
aid
in
the
deduplication
efforts
and
so
forth.
So
there
are
definitely
some
concerns
about
just
that
whole
issue
of
data
collection
and
so
forth,
and
then
we
also
know
that
there
were
fewer
non-response
follow-up
visits
and
some
of
that
had
to
do
with
covet.
There
were
a
lot
of
locked
apartment
buildings.
C
C
Then
we
also
had
changing
and
abbreviated
timelines,
and
I
know
I'm
taking
a
trip
down
memory
lane
with
all
of
you,
but
you
know
we
first
had
an
extension
in
the
data
collection
and
processing
timelines,
and
then
that
was
undone
so
overall,
there
is
really
significantly
less
time
for
data
processing
and,
let's
all
remember
that
there
are
significantly
more
people
in
the
u.s
that
had
to
be
counted
for
this
census
than
there
were
in
2010.
C
So
add
to
that
significantly
less
time
for
data
processing
and
looking
at
the
plant
data
processing
schedule
pre-pandemic,
which
was
five
months,
then
the
census
bureau
requested
to
cove
it
making
that
six
months
and
then
the
actual
time
allotted
being
less
than
three
months
that
most
certainly
can
somebody
can
can
give
you
some
nightmares.
C
According
to
the
commerce
department
inspector
general,
the
streamlined
data
processing
under
the
accelerated
census
plan
poses
a
myriad
of
risks
to
accuracy
and
completeness-
and
this
is
of
course,
what
we're
all
worried
about.
C
So,
let's
talk
about
disclosure
avoidance
and
differential
privacy,
it
might
be
good
to
just
kind
of
go
through
this
terminology
a
little
bit,
because
I
think
at
this
point
there's
an
assumption
that
everybody
really
has
internalized.
What
that
all
means.
I'm
not
sure
that
that's
true.
I
have
to
remind
myself
always
what
that
means.
So
I
thought
it
might
be
helpful
to
have
this
slide
and
there
are
some
experts
on
this
call.
So
I'm
hoping
I'm
getting
this
right.
C
So
what
is
a
disclosure
avoidance
system
they're,
basically,
methodologies
and
techniques
that
are
applied
to
collected
data,
to
protect
the
respondents,
privacy-
and
it's
also
referred
to
as
privacy
protection
systems,
what
is
differential
privacy,
so
that
is
the
new
methodology
that
is
implemented
that
has
been
implemented
now
by
the
census
bureau.
It's
part
of
the
new
disclosure
avoidance
system
and
in
other
words,
when
we
talk
about
noise
that
is
injected
into
the
collected
data.
That
is
essentially
what
differential
privacy
is.
C
C
That
is
a
statistical
process
that
corrects
some
of
the
counts
by
geography,
so
that
happens
after
you
apply
the
differential
privacy
noise
and
what
it
does
is
it
eliminates
negative
values,
so
some
of
the
blocks
after
you,
after
the
census,
I
should
say
after
the
census
applies
differential
privacy
may
have
negative
values,
so
there
is
negative
five
people
living
in
it
and
then
post-processing
would
look
at
those
blocks
and
they
basically
correct
it.
C
So
what
is
the
problem
or
how
big
is
the
problem
and
I've
seen
some
of
my
colleagues
in
particular
kim
and
some
others
and
jeff
already
post
some
of
these
issues
that
I'm
talking
about
here
in
the
chat
and
it's
basically,
how
do
we
know
how
big
the
problem
is?
The
problem
in
general
is
that
the
new
disclosure
avoidance
system
is
still
a
work
in
progress,
and
I
appreciate
all
of
the
stuff
that
census
is
doing
right
now.
C
I
really
appreciate-
and
I
think
we
all
do-
the
incredible
collaboration
and
james's
work,
and
you
know
the
willingness
to
put
data
sets
out
there
for
general
consumption
and
for
people
to
evaluate,
but
I
think
the
proof's
also
in
the
pudding
that
this
is
a
work
in
progress
and
perhaps
that
census
could
benefit
from
a
little
bit
more
time
to
to
iron
out
some
of
the
issues
that
are
happening
to
remind
you
census
released
four
data
sets
for
evaluation
and
made
changes
after
feedback
was
received.
C
After
from
the
research
community
and
those
data
sets,
they
took
2010
census
data
and
applied
the
new
privacy
measures.
So
that
is
what
we
could
compare.
The
data
sets
are
referred
to
as
dps
one
through
four
so
demonstration
product,
one
through
four.
If
you
see
that
that
is
what
those
are,
they
were
released,
starting
in
october
2019,
there
were
some
really
great
maps
made
by
caliber
corporation
that
are
still
on
their
website.
C
So
if
you
look
at,
I
think
you
can
google
caliber
corporation
differential
privacy
maps
and
they
pop
up,
and
they
illustrate
what
happened
in
this
first
release.
So
I'll
give
you
some
of
the
quick
talking
points.
The
findings
were
that
there
was
a
bias
in
there
that
moved
total
population
from
the
urban
to
the
rural
areas.
C
The
effect
was
less
in
the
urban
areas,
but
pretty
significant
in
the
rural
areas
it
distributed
also
minority
populations
from
high
concentration,
urban
areas
to
low
concentration,
rural
areas,
and
with
this
assessment
I
am,
I
am
benefiting
from
analyses
that
have
been
done
by
many
of
my
colleagues,
and
I
just
want
to
do
a
particular
shout
out
to
steven
ochoa
from
maldef
who's
done
an
incredible
amount
of
work
on
these
on
these
data
sets
and
has
shared
findings
with
many
of
us,
so
differential
privacy
demonstration
product
two
all
right,
different
demonstration
product
two
was
released
in
july
2020
and
that
one
was
better
in
total
population
distribution,
but
it
was
still
very
similar
to
the
release.
C
C
It
took
the
zero
population
blocks
and
and
held
them
constant,
so
that
through
differential
privacy,
other
data
could
not
be
sprinkled
to
to
those
blocks,
and
that
was
an
interesting
data
set
actually
to
evaluate
it,
also
made
high
concentration,
minority
areas
more
minority
and
lower
concentration,
minority
areas,
less
minority
and
then
a
differential
privacy
for
demonstration
product.
I'm
sorry
demonstration
product
for
release
november
2020.
C
There
wasn't
a
lot
of
time
to
evaluate
this,
yet
it
appears
to
do
better
on
total
population,
but
it
again
increases
minorities
in
high
concentration,
minority
areas
and
it
decreases
minorities
in
lower
minority
population
areas,
and
that
is
similar
to
release
3.
so
in
all
releases.
In
summary,
the
smaller
the
geography,
the
larger
the
changes
in
total
population-
and
this
may
really
have
significant
effects
on
redistricting
of
small
jurisdictions.
C
I
know
we're
generally
talking
about
you,
know,
legislative
seats
and
all
that,
but
city
councils
and
so
forth
are
definitely
going
to
see
an
effect
if
this
continues.
So
the
new
disclosure
avoidance
system
definitely
has
certain
biases
that
are
being
worked
out.
The
question
is,
can
they
be
fixed
in
time
and
how
will
we
know,
as
james
said
in
the
interest
of
time,
they're
not
going
to
put
out
another
demonstration
data
set?
So
that's
a
big
question
and
the
census,
you
know
again
will
not
release
another
test.
C
C
For
example,
if
your
state
needs
to
reallocate
prisoners
to
their
last,
their
last
known
residential
address,
you
could
start
collecting
those
data
sets
and
clean
those
data
sets,
and
I
will
tell
you
from
experience
cleaning
of
that
data
set.
California
has
to
do.
It
is
a
lot
of
work,
so
get
started
on
this.
If
you
haven't
already,
if
you
don't
have
election
and
voter
registration
data
already,
but
you
will
need
it
start
to
collect
it,
you
can
set
up
hearing
schedules,
websites,
draft
informational
materials.
Do
translations
and
accessibility
tools,
get
everything
ready.
C
This
can
all
be
done
on
the
front
end
oftentimes
these
things
kind
of
get
done
when
the
data
are
already
released,
but
many
of
these
things
can
be
done
on
the
front
end
and
one
very
important
thing
is
figure
out
how
to
conduct
hearings
during
covet.
I
think
that's
going
to
keep
all
of
us
busy
start
hearings
to
collect
public
input.
That's
also
possible
before
the
data
are
released
and
then,
if
communities
of
interest
are
a
criterion
in
your
state,
you
can
start
to
collect
those.
C
Also
also
remember
you
will
have
the
geography
already
available
and
that
is
definitely
going
to
be
a
big
plus
once
the
geography
is
out
in
february
latest.
If
you
have
a
commission,
you
can
start
trainings,
you
can
start
the
trainings
for
the
commission,
there's
usually
a
lot
of
bureaucracy
there.
That
needs
to
happen
before
people
are
brought
up
to
speak.
You
can
hire
staff
and
contractors
and
so
forth
and
then,
as
james
said,
utilize,
the
citizen
voting
age
population
special
tabulation.
C
We
were
able
to
get
the
get
the
census
with
james's
help
to
release
that
one
more
time
they
were
not
going
to
do
that.
Thank
you
very
much
for
everybody's
help,
kim
in
particular
and
james.
This
will
be
released
in
february
and
you
could
potentially
start
with
some
preliminary
voting
rights
assessments
so
takeaways
really
quickly.
The
census
is
not
over.
Everybody
needs
to
stay
informed
and
involved,
and
ask
questions.
C
Ask
questions
about
data
accuracy,
support
census
staff
incenses
operation
by
requesting
that
they
be
given
sufficient
time
to
produce
the
best
possible
data
and
the
delete
the
delays
in
the
census
data
release
do
not
mean
that
you
cannot
get
started
with
your
redistricting
activities.
Prepare
other
data
sets
collect
supplemental
data
and
utilize
the
cvap
dataset.
If
you
want
to
do
some
preliminary
vra
assessments
and
stay
safe
and
happy
redistricting.
Thank
you
very
much.
C
A
Thank
you
very
much
for
those
of
you
who
are
new
to
redistricting
karen
and
james.
Just
did
a
very
fine
job
of
trying
to
make
a
dense
topic
available
to
all
of
us.
So
I
really
appreciate
sort
of
the
the
thought
that
went
into
making
this
information
available,
and
I
know
that
james
and
karen
would
both
be
happy
to
talk
to
people
offline
as
well,
so
you
can
get
their
emails
from
us.
I'm
I'm
seeing
the
nodding
heads
that
yes,
they're
they're
willing
to
do
that.
We
have
a
few
questions.
A
One
is
sort
of
a
legal
question
and
it's
addressed
to
you
james,
but
I
think
james
or
karen
could
answer
it
and
then
I'm
going
to
come
back
to
the
more
technical
questions
and
that
came
from
benjamin
he's
asking
what
are
your
thoughts
on
new
jersey's
recently
passed
constitutional
amendment
to
delay
redistricting
the
data
isn't
received
by
february
15th
do
either
of
you
have
an
opinion
or
something
to
offer.
My
own
thought
on
that
was
that
they
had
good
forethought
in
giving
themselves
some
flexibility.
How
about
you
too?.
B
B
So,
if
that's
when
they
felt
that
they
needed
to
have
the
data
in
their
hands
to
be
able
to
to
do
the
work
that
they
need
to
do
with
the
plans
this
decade,
even
before
our
our
coveted
delays,
we
wouldn't
have
been
able
to
get
them
the
data
by
that
time
frame.
So
if
that's
what
they
felt
they
needed
and
that's,
it
should
work
well
for
them.
A
Right
and
now,
let's
move
on
to
some
of
the
more
technical
questions
and
this
one
came
in
while
you
were
talking
karen
and
that
was,
can
you
explain
why
differential
privacy
has
less
impact
in
urban
areas,
but
also
distributes
minority
population
from
high
concentration
areas,
which
tend
to
be
urban
to
lower?
Can
can
you
do
that
again
and
and
if
it
makes
more
sense
for
james
to
feel
bad,
I'm
sure
that's
fine
too.
C
I
don't
know
that
I
can
explain
it
quickly.
I
think
that
there
is
another
seminar
coming
up
in
january,
where
wendy
and
christy
promised
that
we
would
have
more
time
to
explain
things,
and
I
think
we
should
focus
on
on
explaining
it
there
and
perhaps
kind
of
illustrating
it
a
little
bit
so
james.
If
you
think
you
have
a
quick
way
of
explaining
that,
please
go
it.
B
I
would
say
with
it
with
differential
privacy,
it's
a
steep
learning
curve
and
there
is
no
real
quick
way
for
for
any
of
the
explanations.
But
but
I
think
your
suggestion
might
be
a
good
one.
A
Well-
and
that
was
such
a
nice
segue
car-
and
I
thank
you
for
mentioning-
we
do
have
a
redistricting
seminar
january
6
to
the
8th
that
is
online
and
we'll
put
the
registration
in
the
chat
box,
our
intended
audience
our
legislators
and
legislative
staff.
But
everyone
is
welcome
to
participate
with
that
james
if
you
can
do
invariants
for
housing
units
and
group
quarters
at
the
block
level.
Why
not
also
for
total
population
that
comes
from
kim.
A
B
Yes,
so
you
know
we
we
did
discuss
it
and
we
looked
at
it
and
holding
the
total
population
invariant
for
the
population
would
affect
other
things
about
population,
so
housing
units
don't
necessarily
affect
the
population
characteristics,
but
if
we
were
to
hold
the
total
population
in
the
total
voting
age
population
invariant
at
the
block
level,
as
we've
done
in
the
past,
then
the
noise
that
goes
into
the
data
to
help
protect
the
respondent's
privacy
would
have
to
be
distributed
elsewhere
and
the
only
elsewhere
there
when
it
comes
to
the
person
records,
is
the
race
and
ethnicity
characteristics.
B
So
the
the
decision
was
made
that
it
would
be
better
overall
for
the
purposes
of
redistricting,
to
try
to
to
not
not
have
that
effect
and
and
relax
that
that
that
requirement.
A
Right,
thank
you
and
then
mickey
asks-
and
this
is
probably
for
you
james,
but
either
any
recommendations
on
best
ways
for
localities
to
involve
community
groups
to
understand
communities
of
interest
and
karen
you've
done
a
lot
of
that
kind
of
stuff.
Maybe
maybe
we
will
start
with
you?
How
do
you
involve
the
local
people
in
identifying?
What
are
communities
of
interest.
C
I'm
going
to
make
a
suggestion
that
we
start
that
we
start
a
user
group
that
starts
communicating
on
that
question
and
also
raisa.
If
I
am
pronouncing
your
name
properly
asked
about
recommendations
about
conducting
redistricting
public
hearings
during
covet,
I
think
everybody
is
going
to
be
crop
grappling
with
this.
This
is
going
to
be
one
of
the
top
issues.
C
I
think
that
redistricting
bodies
have
to
deal
with
and
I
think
it
would
be
good
to
get
a
group
together
that
perhaps
meets,
and
maybe
ncsl
can
set
that
up
and
talks
about
best
practices.
There
have
definitely
been
some
developed
already
and
I'm
happy
to
share
what
california
has
done.
I
would
say:
look
at
your
census,
outreach
and
start
piggybacking
onto
that.
That
would
be
my
quick,
very
quick.
Very
you
know,
broad
answer,
but
there's
a
lot
of
detail
after
that
there's
also
apps
out
there
that
can
be
used
and
so
forth.
So
wendy.
C
A
I
I
hear
you
loud
and
clear,
because
we've
been
sort
of
saying
public
input,
public
input,
we
don't
quite
have
that
landed
yet
and
maybe
the
first
way
to
land
it
is
to
find
out
what's
going
on
in
the
states,
and
that
would
be
by
bringing
people
together.
If
any
of
you
have
an
idea
or
just
just
give
us
what
you've
got
right
there
in
the
chat
box
today,
please
on
on
things
that
were
working
in
your
state
and
chris
asks
something
about.
A
B
Yeah,
so
you
know
I
I
I
am
empathize
with
chris's
question
and
michael
chernick's
follow-up
question.
I
am
also
frustrated
with
not
being
able
to
provide
more
concrete
information
about
the
timeline
for
when
you'll
receive
that
data.
B
We
are
the
the
processing
and
the
creation
of
different
data
products
from
the
census.
Bureau
follows
a
chain
of
activities,
and
there
is.
There
are
proceeding
activities
going
on
right
now
in
order
to
create
the
apportionment
counts
that
have
to
the
the
the
final
date
for
when
certain
outputs
from
the
of
course
or
from
the
the
apportionment
product
production
line
are
available,
are
needed
so
that
we
can
hook
the
redistricting
activities
to
that
final
date,
because
that
has
the
inputs
that
we
need.
B
We've
made
a
lot
of
progress
on
identifying
the
subsequent
activities
that
we'll
have
to
do
to
sort
of
put
humpty
dumpty
back
together
again
since
we've
decoupled
things
that
we
would
have
already
done
by
that
point.
So
we
know
what
we
have
to
do
there.
We
know
the
sequencing
of
activities
there's
one
or
two
little
pieces
of
uncertainty
that
we
have
to
get
ironed
out,
and
then
we
actually
have
to
propose
that
plan.
Before
we
can.
B
B
A
I've
I've
been
trying
to
work
for
most
of
this
year
on
new
words
to
replace
uncertainty
and
unprecedented
and
such,
and
we
are
just
in
that
world
of
uncertainty
and
unprecedented
everything,
so
I
do
need
to
close
us
out
promptly.
So
I
want
to
thank
james
and
karen
very
much
for
sharing
with
us
and
for
making
yourselves
available
to
talk
with
anyone
who's
on
this
call
separately
and
the
message
we
got
about
public
input.