January 29, 2010

Counting Every Address in the Census: Joseph Salvo at Pew Research Center

Joseph Salvo, New York City’s in-house demographic consultant, spoke at a Jan. 21 forum on the 2010 Census at the Pew Research Center about how building a strong address list is a key task to ensure that no one is missed in the census count. Salvo, director of the Population Division at the New York City Department of City Planning, discussed how officials in his city reviewed and expanded the list of addresses to be contacted by 2010 Census officials. He also suggested ways to improve the address-review process.

In the following edited excerpts, ellipses have been omitted to facilitate reading.

The Master Address File is the foundation for the census. Everyone needs to be tied to a location. It’s the basis for representation. Irrespective of what vehicles you use to collect data, you need to be tied to an address of some sort.

The creation, review or correction of this master list of addresses is a critical process, and one that is very near and dear. I want to tell you a little bit about what we’ve done in New York City and raise some issues, get you thinking as we move forward.

The current system as it exists essentially provides for systematic updates of this master list of addresses. It’s done through the U.S. Postal Service, where the Postal Service delivers information to the Census Bureau, and the Census Bureau updates new housing units that come on line. And there are field efforts in rural areas that are used to update the address list.

More haphazard updates (a euphemism), are done because of what I call discovery in current strategies. A form goes out in the Current Population Survey or the American Community Survey; a result comes back, undeliverable–the Census Bureau needs to look into it. And sometimes they discover that there’s another address or that an address needs to be modified.

And then, of course, there’s the full canvass prior to the decennial census, which occurred in 2009. The Local Update of Census Addresses program, the LUCA program, which was instituted as a result of the 1994 Address List Improvement Act, provides local governments with the capacity to review their address list.

Now, in all fairness to the Census Bureau, they haven’t been doing this too long.  If you go back to 1970 and 1980 when we saw a lot of mail-out/mail-back [of census forms] for the first time those decades, they created a list and then kind of dropped it after the decennial census.

Starting in 1990 in the push for the American Community Survey, the Census Bureau recognized that they needed to have a continuous list – “continuous” underlined three times.  A list that essentially is put in place and is used to draw a sample for the American Community Survey, and as a basis for the decennial census. So the Census Bureau is learning and needs input from the localities, needs input from researchers on what a continuously updated master address file essentially means.

The address list is a tremendous database.  When people think about an address list, they may not get too excited. I do, because what you have there  is kind of a longitudinal history of a whole bunch of addresses in the United States. There’s this massive list that needs to be honed for the purposes of a survey, for example, the ACS.

Local governments participate in this LUCA program, and that’s a major input.  But resources vary dramatically.

In New York City, our LUCA operation took place over three years, involved about 60 people, leveraged the resources of the largest planning department in the country. I had to get a $250,000 grant for three separate field operations to work on the address list, to get it to a point where we felt like we had something that was real.

Standards at a local level vary.  Many municipalities in this country have one person who does building permits, who does zoning, who does all the things that are done in a planning department or done in a local town.

LUCA technically takes place over a 45-day period.  Forty-five days. People who’ve been working in the LUCA program–we started our LUCA work in 1998 for the 2000 Census–know that you cannot do anything in 45 days. Even small municipalities have difficulty with that.

Now, we submitted to the Census Bureau, as a result of our LUCA program, 196,973 addresses.  These were called “adds.”  These were addresses that we felt were missing from the address list.

We also submitted about 200,000 corrections to the address list–5.7% of the Master Address File. I just want you to note here the biggest chunk was in Manhattan and it’s about 6.8%, almost 7% of the Manhattan addresses. Keep that in mind.

Now, [take a look at this] map. [see slide] Look at all of those dark blues.  Five hundred or more housing units in a census tract.  This is a census tract map in New York City, 2017 census tracts.  Look at all of the activity in Manhattan–a lot of it new stuff, new buildings.

Now, what actually happened?  The Census Bureau took 121,000 of our addresses.  They went out and did their canvass and they concluded that 121,000 of these addresses were actually there.  We submitted 36,000 recently in our appeal, another process that you go through where you have a second shot at getting those addresses in.

If our appealed addresses are accepted, 157,000 addresses would have been added as a result of this process–or 4.5% of our total. [Note: This number later was revised down to about 130,000 addresses.]  Now, this is great, but if I tell you about the difficulties, what is the phrase that was used earlier:  ugly?  LUCA is ugly.  There are so many things that can happen. There are so many problems. There are so many resource-intensive processes you need to go through in order to do it.

How do local governments help, though?  We can help by telling the Census Bureau about the subdivision of existing buildings, for the purposes of all kinds of creative layers of occupancy. Overcrowding in many of our buildings is a function of people doubling up and all kinds of cutups of existing buildings. We can provide timely address information on new construction and new patterns of occupancy.

What I would like to suggest is that we all begin to think about what continuous address list updating means.  It means you empower the Census Bureau’s regional office to conduct a regular MAF [Master Address File] review with representatives of local governments.

Right now I have 60 people in my agency who have signed a form saying that they are going to maintain the confidentiality of the address list.  We may have to alter Title 13 [the law governing privacy of census information] to do this continuous updating.

The question is, how do you do that and maintain the confidentiality of the census process, the image that we portray accurately that the census is a confidential operation?  You get local government geographers involved.  My staff and I probably know more about our address list than many people in the Census Bureau because we focus on New York exclusively.

You do this year after year after year and what you end up with is an extremely valuable data set that you learn more and more about so you can more accurately mine it.  Again, I have this image of a Google-like algorithm being created that somehow sorts through and figures out which addresses to use, which representation for an address or for a property should be used for a survey.

What you do is you create a more inclusive and equitable process.  This one-shot stuff does not work.  What we have right now is a situation which is very uneven.  Even though participation in the LUCA program is up a bit, the fact is that many governments cannot participate because they don’t have the resources.

Many governments would like to participate, but they can’t.  If this program is instituted through the state data center network or through some network of people in the regional census offices, over time you will build a data set so that by the time we get to 2020 we will have a basis for the decennial census. We’ll know how to do an extract of the Master Address File and, bottom line, we may not require a total canvass of all addresses in the United States as a way to rectify this.

This resource over time will feed the American Community Survey, make it better.  It needs to be something that we all have to think about.  Again, there is a legislative issue here:  Title 13 protects those addresses for good reason.

Earlier postings from this event include the transcript and audio of a presentation by Census Bureau Director Robert Groves, and an edited transcript from speaker Constance Citro.