Making Oregon Count 2020 – The Census Undercount and Hard-to-Count Individuals

so I am Nick Chun I am a demographer at the Population Research Center I manage the Oregon Population Forecast Program With me here is my colleague Randy Morris at the Portland Institute metropolitan
studies today we’re going to be going over is the hard to count map I now
realize in hindsight the actual description of this breakout session
probably wasn’t the most detailed it probably could have been more
detailed will be going over the hearts account map that PRC created that was
funded through cefco and I’ll be touching on some of the other the couple
other maps that are available that were mentioned in some previous presentations
or panels particularly the the CUNY HTC 2020 map and the Rome map from the
census so we’ll be going over like those three maps each one of them’s has their
own strengths and weaknesses and so it’ll be up to the user to to judge
when it’s when it’s appropriate to use one over the other so first thing I
wanted to start out is really differentiating the difference between
undercount and hard to count and so this
terminology is often this terminologies often use interchangeably we probably
seen some statistics that are out there where it’s some that are talking about like
oh like the asian population in 2010 was under counted by 46 percent it’s
like that’s that’s not true it means is that there are hard to count
now even though those two terminologies are like very close together they mean
two completely separate up they mean two separate things and so hard to count
means that individuals are less likely to respond by mail in 2010 or in this
case by mail or internet it means that if they’re less likely to respond by
mail or internet that makes them more likely to be under counted more likely
to be missed and so why is it important to look at the hard to count populations
well for those of you who are here in literally any of the other sessions I’m
sure it’s been hammered in your head that it’s important to look at because
disproportionately populations that are hard to count or
or are missed are usually a lot of racial minorities and it costs more resources
resources to follow up and so ideally we would want people to just to answer the
form the first time because federally the the funding I think Dick like in a
previous interview I asked our previous panel asked is there enough resources to
guarantee that there will be enough a follow-up money for follow-up and I
guess the the best answer is just to get people to answer the form. Undercount
again that means individual not counted in the census in the 2010
post-enumeration survey it showed that there was practically a zero percent net
over count or under count so no statistical it was a statistically
established that there was a net over count or under count but obviously that
varies by racial and ethnic groups and so blacks were at the nation level
african-americans were under counted at two percent renters had one percent and
children at five percent or as their counterparts such as white alone
non-hispanic homeowners and an older and older populations were over counted so
in order to when determining who is hard to count really there was like this
broad framework that really hits on four things it’s a people that are hard to
locate and so hard to locate could be for housing units that are not on the
grid or who work that were missed in the LUCA or that were missed when the
census was trying to get a master address list hard to contact so hard to
contact that could that really goes over into maybe highly mobile people who are
in RV parts or in transient locations those are hard to locate hard to
persuade and so hard to persuade are people who may have low civic engagement
or are distrustful of government and then finally have hard to interview
so this is people who are households that speak a different language that
isn’t on the census so obviously these four are not mutually exclusive hence
the Venn diagram that I shamelessly stole but sourced down over there that
shows that you know a lot of hard to count populations and communities fall
within one or really multiple of these themes and so for the hearts account
groups I’m just going to go there’s children rural residents individuals of
color immigrants and highly mobile persons this is not a full list this is
probably about the fifth or sixth time that you’ve heard someone say this so I
won’t go into too much more detail on that
but for what we are when we go into the second part of this presentation when
Randy will take over we’re going to show the 2020 hard to count map that PRC produced
funded through cefco and this is different from the two other maps that
people are familiar with so there is the CUNY hard to count map which I’m just
gonna pull up over here so here’s the CUNY Hard to Count map so
what this is showing is 2010 mail response rates for the nation the core
plus areas are the areas that are shaded in color those areas represent the
bottom quartile or the lower or the bottom 25% of tracks with the lowest of
response scores at the nation at the national level and so I think that was
at 73 percent and so obviously this is on a national scale right when we look
at the ROAM map those sorry when we look at the ROAM map this is again a national
scale that uses a multiple regression model to predict low response scores so
these are not actual response this is a predicted value based on a
model that’s very important to remember the 2010 is the actual rates that ROAM
is predicted rates and that’s using a national model and so what we wanted to
do and what we did was we created a statewide map that was using kind of
elements from both of these maps or we use the quartile method to identify hard
to count locations but looking at the Oregon average so instead of 73% it was 76%
but using more problems and visualizations to make it a little
easier to read and then for the ROAM which was using American Community
Survey data at the bloc group or census tract level which if any of our data
users are here you’ll know that that’s extremely problematic because of the
data reliability and so the first iteration that we have on the website
right now is more like the hard to count 2020 map where it’s just showing the
historic rates ideally in the near future we want to add a capability that projects out the expected change or what the the new me low return
rate will be similar to ROAM but using in my opinion a more robust methodology
for doing so with that in mind I will let Randy take over to show you the
features of our map if I can get this going hi there like Nick said my name’s Randy
Morris and with Portland IMS The institute of Portland metropolitan studies I’m the
community GIS project leader there and they asked me to help visualize some of
this interesting data that’s coming out from the hard to count and so I’ll show
you what we came up with and explain some of the rationales for the
visualizations if I can find it in this mess so you go to the census 2020 now
website it says right here I have this own link Oregon HTC map this is the
website that’s going to have all the presentations up on board so all the
notes so I’ll do my best to stay out of the way the map only describe some of
these features. we want we did our best to make this available at various
screen sizes and these are the kinds of details I would go into in a more
cartography centric presentation so I’ll stick to the functionality wherever I
can we’ve got general information on the
left tells you sort of why the map is doing what it’s doing and how its
dealing with general information most of the actual information comes up or is
related through these side panels here so it gives us go ahead and click on
census tracts map is fully zoomable pan-able as any good online map should be and if you come in here and click on one
of the tracts what you get is first the reading of the expected or the the 2010
return rate which is how these were analyzed
whether or not we’ve identified it as an urban or rural and then the part that I
had the most fun with to be honest is the visualization here under hard to
count indicators so this is one of the things I really like to try to bring to
the table is drilling down into the data so so it’s hard to count but why is it
hard to count and this is something that Nick and I have been experimenting with
in various places in this case we have red dots that pop up next to the
particular indicator that is in the quartile quartile right now is in the
quartile that is indicative of to to count so in this case if there’s a
high population of children under five then that if it’s in the top quartile
then that’s going to come up with the dot and so on and so forth so a couple
of them I believe are low quartile like occupied housing units okay so most of
them are all high quartile with one look quartile but those change based on what
you click on so you can get an instant sort of readout of what’s going on in
that particular census tract and we’re always trying to expand that capability
we did a social vulnerability map with that sort of concept or order as well
then was there anything else there? yeah so if you click on that again okay so he clicked on the select feature
and that selects the particular census tract then you get a directed way out
here that shows you the district by race in that particular census track
now this is being sort of compressed because of the screen size so if you’re
on a larger screen all of these labels will show up but in this case you have
to hover over it to find out what it is and then of course you have a legend
here and an explanation of the esoteric you know descriptions that you find in a
lot of these census race and ethnicty broadcasts so that’s one thing I showed
this is my wife and she’s like what does this mean what does this mean what is it
you know AIAN which of course some of us early in my career of dealing with this
stuff I instantly saw Asian that’s what I saw when I looked at that and I think
that’s an important thing for us to help clarify with people so always working on
that as well this also has the capability of doing multi select so you
can select using that more complex select tool in the top corner you can
select multiple census tracts the other thing you can do is you know and it
gives you an aggregate I’m still testing this to make sure it’s it’s on the
numbers but so far so good the other thing you can do if you clear that is
we’ve got a couple simple selections and query tools at the top so you can select
by county say your your census work is going on in Kings County or Baker County
or something like that automatically selects that County zooms to it and then
filters all the communities within that County so you can do a simple visual
query and selection drilling down into particular
obviously this one doesn’t come up as one of the the core symbolized
hard to counter areas but you can still get that information so get the specific
information from the census tract relevant to the HTC 2020 so you can do
as many of those as you want trying to think if there’s any other functions
yeah that’s pretty much it like Nick said one of the things we’re hoping to
do is model the arc account for 2020 and offer that as a second layer or a second
map that we that you can compare with the original one and and then have a
better idea of maybe where you’re going to have a problem getting those counts
yeah that’s my part in this so I guess we’ll just opens up questions
or discussions it’s just about the hard to count population or any of the three
maps that was brought up I may not be an expert in the other two but I can at
least to do the best answer questions about the one that we made at PRC I have a question have you guys thought much about the
fact that some of this data could be used to target certain groups or
individuals and what do you guys think about that yeah that’s a really good
question that’s actually one of the reasons why we I mean that’s with with
all types of data particularly when it’s oriented towards social justice you’re
run that risk when you identify populations that are vulnerable then
those are depending on who is using the data that’s that’s that information can
be used for nefarious purposes I mean that’s just the reality of of working
with social science data I would say that the way that we are trying to
mitigate that is by making sure that this information is aggregated no lower
than the census tract level for confidentiality reasons you’ll on the
ROAM data set I believe they have data at the bloc group level or they modeled
some of the estimates at the block group level which is again problematic because
of the reliability of those estimates but also problematic in terms of the the
kind of like the geographic precision of the unit of analysis and so it’s not the
most satisfying answer but the way that we’re trying to I guess like the way
that we’re putting a stance on is making sure that the estimates that we produce
are at a geographic level that is broad enough where ideally they specific
communities like cross sections or intersections won’t be targetted my question is I am going to piggyback on
your concern because we are seeing the rise of richness you know the racist
right racist groups here in the state in our city downtown you know they’re
taking possession of our of our streets and so I as a member of a quote-unquote
minority group that is still you know funny that it’s now 20% of the US
population in Oregon I’m worried about the specificity of your data because we
could keep that data you know and really reflect who’s out there but I cannot you
know some somehow be safe from being targeted you know in my own home my
mother is disabled how would I know my friends who are gays lesbians trans
women how would I know that they’re not going to be targeted because you know in
the pursuit of science we’re trying to get you know a true count but are we
giving weapons to people who don’t have my own view of my destiny my life you
know my security so so I wanted to make sure what can you tell us in terms of
keeping that privacy and keeping the social net you know in place yeah I think that’s a I think that’s a great question and again this kind of comes
back to the double-edged sword of using data-driven decisions right on the one
hand if we don’t count these populations if we don’t highlight the injustice or
the inequities is then those inequities don’t exist on our system they don’t see
it that’s not measure on the other hand by pointing them out like you said it
can open up and create vulnerabilities and you use them an Avenue
nefarious purposes I will say for the data that we’re showing it’s all publicly
available so this is not our responsibility
we don’t have the purview to restrict this information this information is
also put out by the Census Bureau we’re actually providing less information than
other than other groups because we don’t find that the estimates are
reliable or that they’re necessarily conceptually valid so that’s the rigor
that we implement but in terms of the actual information like someone’s like
race ethnicity or nationality that information is already out there and a
one thing for the census is that it doesn’t ask the question about the
orientation it will ask about your household relationship this will be
covered more by my director Dr.. Jacob Jurgevich in the following panel but
it actually goes over household relationships I know that’s kind of a
tangent but I just wanted to point that out that is a great question and that’s
a conversation that we literally have almost daily in our office we’re all
very concerned with that and we’ve run into it in in a variety of different
research projects I can give you one example I’ll give you a bunch of
examples too that are very relevant to this or the point in time count I was
involved in that a couple years ago and as a as a researcher it was it was
infuriating on one level to have to tear all that information up about the
particular points geographically especially as a mapper right that’s what
I do I help them map in order to find the people for the point in time count
and then we destroyed the information afterwards and I was internally very
happy about that but also like I said just infuriated at destroying data and
not you know what I mean so that struggle is there and then the
other thing I was going to mention is in a website that we also have there in
IMS called neighborhood pulse I put together a story in partnership with one
of our researchers Uma, Uma Krishnan who now works for city of Portland and
the story is about the Somali population in Portland and we had many
back-and-forth conversations with their representatives the people who
ostensibly represented the the groups that wanted to do this story with us
with that specific discussion about do you want to be mapped and how how should
we protect this information like how much do you want out there President
Trump had just been elected and that really really set off these
conversations and they decided they would rather be counted they would
rather be out this is who we are this is how many we are so that they were
participating in that you know in that great social experiment of getting money
essentially they were trying to get funds like everybody else to build
community centers and things like that but as long as the city or the county or
whoever only thought there were 2,000 of them rather than 10,000 they weren’t
getting the kind of allocation that they needed to do the things they needed for
their kids so it we have this conversation we struggle with that
question all the time and all the GRAs and research students that come in and
work with us struggle with that as well just so you know it’s not a not a new
question for us either so thank you so I’m coming in a little bit new learning
about the census in terms of the city side I work for the city of Vancouver
I’m just wondering if the question of race was completely removed how that
impact how we get funds and allocate funds in terms of the removal of race and so
yeah so I believe the race or the population count by race is actually in
the pl file so that’s one of the files that is legal where the census is
legally required to produce for working men but your to your question yeah what
if that was taken out yeah I mean it would definitely affect some it affects
a could affect on multiple things one is in terms of representation so in terms
of like a APANO know if you’re if we’d have someone from here from APANO and
okay right so for APANO right like when you’re applying for grant or you’re
applying for a federal grant one of the one of the things that they need to know
is not just the total population of your area but like the area or the population
of interest and so the Asian American Asian American and
Pacific Islander population and you know even though we may be only looking at
2010 or 2020 data the thing to remember is that all inter-censal estimates so like 2011 2018 where we’re at now that
was based off of their relationships and and the denominator that was in 2010 and
so even though like you know this happens only every every ten years like
quite literally all social science estimates all of our estimates are based
off of the decennial census and so if we don’t get that population count by
race that means we can’t get all these other estimates by race like meaning how
far are these other these other estimates by race like median household
income poverty all of those become much more subject to to scrutiny and and they
become less reliable and so yes it matters proportionate but it also
matters for research in general this is just a curious question of why
you decided to use race like black alone rather than and in combination with yeah
that’s a that’s an excellent question oh one of the reasons is because of the
way race is operationalized in the census and so you’ll notice therefore a
lot of these things we’ll see white alone non-hispanic black alone not Hispanic
everything and the reason we do that is because that’s the only way to get to a
hundred percent – like when you add it up it adds up the total population
without over counting anybody are missing anybody so if you were to do
black alone or in combination and you sum that up for all the races not only
would you be excluding Hispanics but the total pocket but the sum of that would
exclude what would exceed the total population because they’re double
counting people right now obviously there are issues when you’re looking at
black alone non-hispanic when thirty-seven percent of black Oregonians
checked another box other than black so yeah you’re potentially excluding a
lot of populations but at this point the depending on, no method is perfect when
you’re when you’re looking at this and for the purposes of this map we wanted
to make sure that the estimates of the subgroups added up to the total pop is right if I just speak here? Nope we do need you oh yeah can you speak to just creating the math
and the colors used for the hard to count I’ve seen a few hard to count Maps
and I’m disappointed to see that the hard to count populations are indicated
by the red color in particular when we sort of unpacked that history and
understand that it has to do with being highly mobile or not being like being
excluded from civic activities and so it doesn’t seem if anything it should be
green that’s where the money is yeah it is problematic absolutely this
is essentially a first pass based on the Oregon version of this map so we we went
strictly with their symbology but I struggle with that as well because part
of me says says the you know the stoplight is actually useful for calling
attention but at the same time wait I understand that’s something that we’re
going to continue sort of field testing and that kind of feedback actually helps
a lot so if you have any any more specific reasoning for that I’d be glad
to hear it maybe afterwards and I will take that into consideration absolutely
but that is foundational and I appreciate you calling that out even
though we’d already sort of thought about redoing red becomes a very
dangerous color to use these days you can blame it on me I default into
existing realm he broke the cardinal rule for cartography you never use blue
or red unless you’re doing a political map don’t do red and green because we’re
colorblind there’s like there’s like like really if
we want to go deep into it like there’s only like maybe five colors that color
ramps that you could use for this map but we of the time constraints that we had
like we did not invest the time the resources into that so this isn’t directly related to Oregon
but when you work showing the National maps at the county level northern
Wisconsin was something that stood out maybe not necessarily on that map but the other one do you know what was going on
up there or why that particular area I just talked to somebody about that I am
not entirely sure when we click on this I don’t know if this data was like
marked out because of confidentiality or something yeah. I think that’s an area that uses a different enumeration Yeah that sounds right And so one of the different so one issue this happens in rural places that
was brought up in previous panels is that if you have a PO box right in in
2010 the census wouldn’t deliver mail to a PO box which is problematic if
you’re in a rural area where like literally ninety percent of households
have a PA boxes so that requires more interaction conversation census to get
those forms out that would be like a special case that deviates from the
normal process of enumeration I’m not sure if that’s one of the reasons why
this is grayed out because it’s not consistent methodologies of how the
other estimates were derived We’ve got a few of those in Oregon as well well if nobody has any other questions
about hard to count or just questions in general like just go we have we have
time where people want to go see another one they can like you won’t hurt my
feelings like it’s fine please feel free to stop me I have a question about so we
know that some areas so I’m I’m actually working in Fairview in wood village and
when we pulled up on up on the map you can see that they had a pretty low
response rate in Oregon and the third lowest response rate it
I think sixty three point one or 61.3% very low so as I look at as I work in
those communities something I wonder about is we know that they have a low
male response rate we know that probably some people weren’t counted can we get estimates like about how many
people weren’t counted or I don’t know what I’m exactly trying to say but we
know like we know there’s a question mark and do you guys have like tools to
kind of get at that? Are you referring to an under count or a hard to count? so there was not so there was not an official under count which I understand
to be like a process right yes we would as if I’m with the city so we would I
would say hey I think there was an official or I think there was an
undercount in 2020 and I go through an official process where a determination
is made right whether its was under counted or not yeah that didn’t happen
in 2010 but because of the I just wonder like what am I trying to say yeah but
how come we have to count people I know that’s in the Constitution but we know
that some people don’t get counted so why can’t we just count them anyways
like without having to go through the census process that’s a we have the data
sorry yeah I mean I think one of the things I think it’s a good question
especially with a data with a lot of administrative records and sources out
there that like I’m sure as time passes on one of the questions that will come
up with the census is like all this like multi-billion dollar process is it
really worth it when we have administrative records that we could
just join I would say that there are two issues to that one is that
administrative records don’t have all the information that the that the census
is going to ask for at least not like all of them in one administrative
records so it may have your age it may have your your sex but they may not have
your race or it may not have your address of location the again like it’s
literally written into law that there needs to be an actual enumeration I
think one of the challenges in with an undercount like you said like if there’s
an undercount in 2010 you can actually we’ve PRC’s actually done this in the past
where a city believed that their census counts too low and we’ve actually
conducted another census for them and based on those results if that turns out
to be higher than the actual census population I believe there’s some sort
of process for emending it or or or changing the official count thank you
one of the things that I wanted to find out from being here today is how will
you deal with populations that one are moving and to whose nationality and
presence is not always there we have maybe hundreds of Mexican nationals not
only Mexicans are people from Thailand people from Western Europe who work here
with visas but that are here present in the country although there may not ever
have a nationality so in talking about trout though I’ve done work for cities
there as a community engagement liaison and I find it so difficult to find
at least a gathering place you know the communities the social the places for
social gathering are not really there maybe only the the library and you can
tell for sure there’s spaces to walk that are huge and not a lot of
engagement into places where you can be safe so you know I don’t know if this
the forum to ask for that you know as best ways to try to count those people
but I know that these are important important populations because many times
you have children you have mothers of sometimes children that are unborn there
should be counting because this there’s people you know this children you know
again in as we go south you know to where Salem and Woodburn we have a ton
of people living in in camps and so in egg farms and things like that and so
how do we count not just the people that we have today but the ones that will be
here you know ten years in I know okay okay I think that final your your last
statement so the census regardless of one’s
nationality will count you if you are reside in Oregon on April 1st so if you
are even a temporary worker I know there’s a bunch of temporary workers I
think in northern Lake County that you know they may only be here for four or
five months but if they’re her on April first they get counted for their population
the other thing that I would say is I’m not for in terms of making sure that
vulnerable populations get counted one of them is particularly around
transitory locations one of the things that our Center did is creating a
statewide inventory of transitory locations across Oregon and so they’re
transitory locations that includes camps includes hotels motels that people
actually that’s their primary residence obviously that inventory I can say
objectively speaking is not complete it’s not a complete list but it’s
because we’ve never had one in Oregon before and so we we created one and
ideally we hope we hope to build on that the future to improve the outreach the
button but again like in terms of like for council council tutorial vacations
that that’s like less as researchers we can’t do that that’s really on the
ground work that’s why it’s really important to get boots on the ground and
get trusted community partners to engage and and identify those areas without you
know without stepping over and and violating some ethical or or
confidential information in terms of understanding the number of populations
and the future like future populations that’s that’s actually my job so I
produced the population forecasts for the for the counties and the cities so
that for planning reasons it’s more difficult to do it by race so I know
there’s a lot of talk about creating projections by race for across the race
groups problematic for a few reasons one is how I told you about how race is a
very complicated construct to measure ends it may not necessarily align with
what we assume to see on what we see in the ground so again like for blacks if
you just do block alone not Hispanic you’re isolating the thirty seven
percent of African Americans who checked another race and so for a recent
particular it’s difficult to forecast because of the changing definition and
the changing way that people record their their race that’s one of the
reasons I would say your question about safe spaces really the best one that I
think we can advocate for right now is the libraries because libraries are
long-established they’re pretty much considered safe spaces across the board
by everyone and they don’t correct me if I’m wrong they don’t seem to be places
that are targeted at this point which you know means that that it’s probably
continuing to be a safe space to the 2020 census some of my work is
around establishing libraries as you know further centers of education and
things like that and everything I’ve seen so far says take your folks to the
library get them you know get them counted don’t expect the government to
find some other safe space because at this point it’s it’s not in the works
and it’s not there’s not enough time. Vivian Bide reporting for the Pacific Sentinel most
of the research that’s been presented today regarding hard to count
populations particularly unhoused populations focuses on engagement with
service providers what is your approach to getting an accurate or near accurate
count of those who are not actively engaging with services? so is this like
a place specific like if it’s a cit y or county that doesn’t have said services
services more this is more about This is more about counting the population that for
whatever reason is not engaging with the service partners that you would normally
reach out to. yeah yeah that’s a question that I that is not in my I don’t have
the expertise to answer that question unfortunately I would say that the way
that that’s done would you if you if you don’t group community members or
community organizations or or trusted neighborhood partners in that
that would be the first place that the census or the city would point to into
identifying the location of of camps okay or the location of these tran of
these transient locations to identify where they are there’s really not a
really a good answer if you’re not on if you’re not engaged in those services
I think the census uses a different enumeration technique that’s actually
going to be talked about in hopefully in the next session I don’t think it was
this session that they’ll go over it I I think if you if if you were to attend
that session the person who is up here answering would be able to answer
question more effectively. A bit of a follow-up question this may also be outside your
purview to my knowledge the 2020 census test forms and even the 2010 census
forms did not include an option for the unhoused how are these figures actually
reported considering there is no there’s no fitting answer for this population. that’s a great question actually I think Mark could probably answer that question
because that you went over he went over that in the previous panel so Mark do
you wanna take take a swing all right this particular question is given that
there is no option to just answer unhoused
on the census forms how are we actually taking account of unhoused individuals
and how are they being classified considering there is mandatory reporting
on all questions so there I’m gonna actually go to deep dive but so they’re enumerated during our
service based enumeration operation so it’s sort of it’s very similar to the
point in time count you know that the state does and then the data from that
count is included in a sort of subcategory of the group quarters counts
that you’ll see that you can see I don’t know if you went over this but on the
American Community Survey for example or the census data as well I guess anymore
decennial census but you can go on and see there’s sort
of included under a group for folks that were enumerated in a emergency or a
transitional shelter for example and then they’re otherwise included in a
group of like folks who are living outdoors and enumerated during targeted
non shelter or outdoor locations that operation so the decennial census itself
doesn’t really provide a good figure for people experiencing homelessness because
as you were saying there’s no information on the form that would
really capture that hoping this is helping to answer your question a little
bit but essentially that information is included under the group quarters
breakdown in the in the census because of course the the actual homeless
enumeration and the actual counts of people experiencing homelessness comes
from housing and Housing and Urban Development that’s the official count for people experiencing homelessness in the U.S. anybody else? okay thank you everyone

Leave a Reply

Your email address will not be published. Required fields are marked *