From the archive, originally posted by: [ spectre ]

http://www.gnocdc.org/repopulation/

ACCIDENTAL CENSUS
http://adage.com/article?article_id=130414
Valassis Database a Boon to Repopulation Efforts
BY Jack Neff  /  August 19, 2008

BATAVIA, Ohio — To people who have no use for junk mail, what’s
happening in New Orleans may come as a surprise. Direct mail could
help rebuild a city still struggling to recover as the three-year
anniversary of Hurricane Katrina approaches.

New Orleans has found a novel use for the massive database used by
Valassis Communications’ RedPlum direct-mail operation. Normally used
to send promotional circulars to virtually every household in the
U.S., it’s now being used to track the speed of recovery in the
Crescent City.

Overall, the Valassis data indicates New Orleans had 146,174
households receiving mail in June 2008, still down 28% from the
203,457 receiving mail in June 2005, two months before the Aug. 29,
2005, hurricane and resulting flood.

Groups focus their resources
The nonprofit Greater New Orleans Community Data Center is using the
Valassis data to track the city’s repopulation progress block by
block. Availability of the data already has resulted in volunteer
groups in some hard-hit neighborhoods, such as Holy Cross in the Lower
Ninth Ward, diverting funds and volunteer hours they had planned to
spend on street-by-street repopulation surveys to actual rebuilding
efforts instead.

Another group, Kingsley House, is using the block-level data to
identify repopulated parts of the Central City neighborhood to enroll
children in health insurance.

And neighborhood groups in other parts of town, such as Gentilly, are
using the data to show repopulation efforts have been successful as
they try to persuade businesses such as supermarkets to come back,
said Denice Warren Ross, deputy co-director of the nonprofit Greater
New Orleans Community Data Center, which is working with Valassis on
the project.

Online tools
On Tuesday, GNOCDC launched a section of its website using the
Valassis data along with Google Maps and Street View to provide a
graphic depiction of the city’s repopulation progress — or in some
sections, the lack thereof.

“One thing our neighborhoods have struggled with since the storm is
trying to determine who’s back and what do they need to do to get back
home,” Ms. Ross said. “That’s a lot of work for a neighborhood leader
to take on when they’re trying to rebuild their own home. Data and
tools like this allow the neighborhood to do more with their very
limited resources.”

The data also serves as a graphic reminder of how much of New Orleans
remains devastated — such as a section of the Lower Ninth Ward
directly adjacent to where a barge broke through a flood wall. Only
one of 25 homes there at the time remains today.

“The rest of the block just looks like a prairie,” Ms. Ross said.

In another case, though, the Valassis data shows a block where there
were no households before Katrina but 80 today. That’s likely the
result of a Federal Emergency Management Administration trailer park
still operating three years later, Ms. Ross said.

Another block showing particularly high population density in the
Valassis data is revealed by Google Street View to be an apartment
complex that was fortuitously built atop a parking garage.

GNOCDC has linked the Valassis data with Google Maps and Street View
to provide visual evidence and get a better idea of what’s behind the
numbers. Google’s Street View images come from a series of August 2007
photographs, though, so the Valassis data is more current.

Information trove
The group had been working with current Valassis data since early last
year, but what it lacked was a comparison to pre-Katrina New Orleans.
Valassis executives were preparing to develop a model to come up with
pre-Katrina household numbers when they found the June 2005 data
sitting in the company archives. It was pure luck, as Valassis, which
generally only has an interest in current household data, generally
doesn’t keep historical data.

“Since the storm, nothing was easy, and we couldn’t believe they just
had a snapshot in time of June 2005,” Ms. Ross said. “It was really
such a gift to us, because it allows us to put the current numbers of
households receiving mail in context.”

While New Orleans did have access to 2000 census data, it was outdated
even before Katrina and didn’t provide a close comparison to the
currently available Valassis data. GNOCDC has been paying what
Valassis Director-Data Solutions Mark Gundersen terms a “nominal fee”
to cover part of the operational and processing costs for the current
data, essentially what the organization had budgeted for the work.
Valassis has provided the historical data and consultation on how to
use that data for free.

“Where Valassis really put themselves out there for the greater good
is that they let us publish the technical details of how their mailing
list works and doesn’t work for this application,” Ms. Ross said.
“It’s documented in enough detail that another city could follow
it … and ramp up rather quickly.”

Valassis gets its data through weekly updates from the U.S. Postal
Service via a non-exclusive relationship, Mr. Gundersen said. But
because it sends out RedPlum mailers nationally to all of its
addresses on at least a monthly basis, Valassis tends to have the most
complete and up-to-date address database in the U.S. The post office
has similar data, but is restricted by law in how it can share the
data, he said.

Hard data helps
While anecdotal information about how much or how little New Orleans
has come back is plentiful, Valassis is helping provide hard data to
back up the claims, Mr. Gundersen said.

“When they came to us, we felt that it was right for us to help them
out,” he said. “People think the rebuilding is done. The volunteers
have dried up. But the need there is still very real, and they need
data to fight for federal funding for rebuilding efforts.”

‘ACTIVELY RECEIVING MAIL’
http://www.gnocdc.org/repopulation/aboutUSPSdata2.html
“Households actively receiving mail”

This data is important because:
1. There are no magic techniques for finding out exactly how many
people are in New Orleans. There’s no way to scan the planet for life
forms like on Star Trek. And, the U.S. Census Bureau’s methods for
estimating population can’t keep up with the extraordinary situation
we have here post-Katrina.
2. Direct surveying is resource-intensive, and methodologies across
neighborhoods are often not comparable.
3. Data on households actively receiving mail is frequently updated
by the USPS, and readily available through several channels. It is
already being used by government officials and others who are making
decisions about New Orleans. So we think you should know about it,
too.

Data experts never expect any particular data set to be perfect.
Instead they try to identify the limitations and find a way to work
with them. The following information will help you to do just that.

What does “actively receiving mail” mean?
Every month, the USPS reports the number of residential addresses that
are actively receiving mail. Remember, their data is a count of
housing and other living quarters—not mail, and not people. Most of
the addresses are for houses and apartments; some are “group quarters”
such as dorm rooms. “Actively receiving mail” means that someone has
picked up the mail at that address within the last 90 days—that is,
the mail isn’t piling up. Typically, a residence identified as
“actively receiving mail” has someone actually living in it.
Sometimes, someone nearby, like a landlord, may be picking up the mail
even though no one is living there. There are also many houses in post-
Katrina New Orleans where people are living and mail is not being
delivered.

Though the total counts of households receiving mail are available,
federal law prohibits the USPS from sharing their actual address list
with the public. In order for us to map “residences actively receiving
mail” block-by-block, we purchased a comprehensive mailing list
database from Valassis.

What is Valassis data?
Valassis’ mailing list is as close to the actual USPS address list as
anyone can get. They start with a complete list of mailing addresses
for the entire country. Valassis is qualified to receive weekly
address updates from the Postal Service, which allows them to double-
check and enhance their database, and work cooperatively with the USPS
to improve address quality.

How accurate are these counts for each block?
Because we are using the Valassis data for a different purpose than
mailing, we started by taking a close look at some of the counts
ourselves. We found that in some blocks, the number of addresses
actively receiving mail was higher than the actual number of occupied
housing units, and in some blocks it was lower. Still, we found that
it was a very helpful indicator of how many housing units might be
occupied in a particular block.

Some reasons why numbers in a block might be an OVERcount:
* Seasonal homes such as boathouses on the lake, or condos on St.
Charles might receive mail though nobody lives there full-time.
* A family may be on the verge of moving in, but not yet actually
living in the house.
* A house may not be occupied, but a neighbor, relative or owner
picks up the mail regularly.
* Addresses from a nearby block might have been assigned in error
to this block. (This type of geocoding error occurs in ALL maps.)

Some reasons why the numbers in a given block might be an UNDERcount:
* Residents may have moved back home, but have chosen to get their
mail at a PO Box only.
* Itinerant workers (including international immigrants) often do
not receive mail.
* Squatters living in vacant or damaged homes would not receive
mail at those addresses.
* Homeless living in cars or tents do not have a physical address
that qualifies for mail delivery.
* Addresses belonging in reality to one block, might be assigned
in error to a nearby block (Again, this type of geocoding error occurs
in ALL maps.)

TIP: All data has flaws, and the smaller the area you’re looking at,
the more visible those flaws will be. If you combine data from
multiple neighboring blocks, the counts for the larger area will be
more accurate, particularly because this will resolve many of the
geocoding inaccuracies.

Why can’t I get the address level data?
You can get the address level data by purchasing it directly from
Valassis. We cannot share the address level data according to our
purchase agreement with Valassis. You can contact them through their
web site at http://www.valassislists.com or e-mail at:
Neworleansdata [at] valassis [dot] com

Technical Documentation
http://www.gnocdc.org/repopulation/TechDoc.html
In this rapidly changing post‐Katrina environment, standard estimates
of population are insufficient. Valassis data offers a new indicator
of population that is timely and available for small geographic areas.
But this data is quite experimental. Users should be well aware of the
available research about this data. This report provides a description
of the Valassis Lists data and the methodology we used to create our
counts. Over the next year, we will continue our analysis of the
Valassis Lists data and document our findings in updated or additional
versions of this technical document.
http://www.gnocdc.org/repopulation/ValassisTechDoc.pdf

Related research…
Using U.S. Postal Service Delivery Statistics to Track Population
Shifts Following a Major U.S. Disaster. Plyer and Hodges. 2008.
This paper examines the impact of Hurricanes Katrina and Rita on U.S.
Postal Service monthly residential address counts for parishes in
Louisiana. Pre-hurricane comparisons between USPS counts and Census
estimates establish a baseline. Then, starting with August 2005, month
by month address counts are examined to assess their effectiveness in
reflecting population displacement and recovery at the parish level.
http://paa2008.princeton.edu/abstractViewer.aspx?submissionId=80655

Using U.S. Postal Service Delivery Statistics To Track the
Repopulation of New Orleans & the Metropolitan Area. Plyer with
Bonaguro. 2007.
This Research Note explains why U.S. Postal Service Delivery
Statistics are needed and useful for tracking repopulation in a post-
disaster context, and documents some of the limitations of the data as
a measure of repopulation. It provides initial comparisons of
indicators based on this USPS data and census estimates for the New
Orleans metro area, along with sub-parish analyses for Orleans and St.
Tammany parishes.
http://www.gnocdc.org/reports/GNOCDC_research_note_May07.pdf

UPCOMING! The accuracy of USPS data on residences actively receiving
mail as a measure of neighborhood repopulation following a
catastrophic U.S. disaster

Allison Plyer of the Data Center will be conducting research in the
summer of 2008. The objectives will be to:

* assess the correspondence between USPS counts of active
residences by carrier route and on-the-ground counts of occupied
housing units in post-disaster neighborhoods;
* explore the reason for any observed differences between these
two data sets, including differences in operational definitions;
* assess the extent to which group quarters are included in USPS
counts of active residences.