FEATURE1 November 2010

Census boss on answering the $400 billion question


Population counts from this year’s US census will be used to allocate $400bn of federal funding to local areas. As the Census Bureau prepares to deliver the results, we speak to its director, Bob Groves, about the challenges of surveying more than 300 million people and the lessons that have been learned along the way.

“I think it’s probably fair to say that every census faces at least one problem that had not been faced by the prior”

Research: The census count took place seven months ago, on 1 April. What stage are you at in the process now?

Bob Groves: The last form was processed through optical scanning machines on September 8, so at this moment we are going through a set of very, very large files. We have about seven billion answers to process, and we’re examining those files to make sure that we understand the data, cleaning the data when there are problems, all pointing ahead to delivering to the President by the end of December population counts at state level.

You must feel a lot of pressure as the person in charge of this huge undertaking that only happens every 10 years.

I think it’s probably fair to say that every census faces at least one problem that had not been faced by the prior. In this census, prior to my arrival at the Census Bureau, there were technological developments that did not go as planned, so there were changes in the design in the last two years of the decade that made things more rushed than was desired.

The US census has been controversial since the 1970s or so, which has to do with the use of population counts for the setting of the congressional districts, so the stakes are high on these counts for political influence. The US government returns to local areas about $400 billion based on population counts, so people care and controversies arise.

What preparatory work took place before the count to make sure that it was as comprehensive and fair as possible?

There are a variety of changes in the design. One is a pretty massive change: the long form was removed from the decennial census and transformed into a continuous sample survey going on throughout the decade which we call the American Community Survey. That left a decennial census which was short form only. That’s good news with regard to likely cooperation or participation of the American public, and we did achieve [response] rates that were much greater than we anticipated.

We also introduced for the first time in mail to a large number of houses a bilingual form, Spanish on one side and English on the other, which was a recognition of a non-trivial component of the population that’s more comfortable reading and answering in Spanish.

The other thing to note with regard to innovation is that this really was a multilingual census. We had six different languages fully translated, we advertised in 28 different languages, we had language assistance guides in 59 different languages, we had a staff that spoke over 130 languages. The US is clearly a multilingual society and we recognised that in this census in a way that we are hopeful has improved the participation of those language groups.

“A successful census has to be viewed as a local event even though it’s a national event. The local community has to embrace it as something they care about”

It’s important to know that in the US we count all residents whether they are citizens or not, so we are reaching out to people who normally would avoid contact with all government officials because they would fear their deportation. And that’s a challenge to us. The technique that seems to be ubiquitously successful is if we find trusted voices in the diverse communities throughout the country who are willing to ally themselves with us and get the word out about how the census, although a government activity, is really quite separate from any enforcement activities of any government and actually acts to benefit all residents whether they’re citizens or not, because of the return of these federal funds for services like schools and hospitals and so on.

I’m willing to speculate right now that the over 250,000 organisations throughout the country that were willing to ally themselves with us are part of the ingredients of success. It was just amazing. These were sometimes groups that were as small as a residents’ association in an apartment building. They were multinational firms, they were local service providers to folk who are homeless. All of these folks united to try to get the message out, and those messages, once delivered by them, are much more powerful than any message we could give.

A successful census has to be viewed as a local event even though it’s a national event. The local community has to embrace it as something they care about.

A lot of the success must come down to having good people on the ground.

We were the beneficiaries of a very unusual labour market this decade. The recession in the United States produced unemployment among a set of people with unusually good skills. We hired them, and they are the success story of many of the operations of the decennial census. We had people with skills and work experience that were unprecedented. It was the smoothest operation in terms of the execution by personnel that I’ve ever witnessed – almost every line of our operations was under budget and ahead of schedule.

The Bureau is making an effort to make the process easy to understand and to make the data accessible. You’re running a blog, you have videos on YouTube, and you’re launching a new system for looking at census data online. Is that a big part of the job?

It is a quite deliberate effort on our part to use modern media to get our statistics in the hands of more and more diverse populations. We are aware at the Census Bureau that no one medium works anymore to reach our diverse society, and we have to worry about multiple languages, multiple media, different platforms to get the word out.

It must be a challenge to make sure that your communications don’t appear to be politicised, because a lot of the topics that you’re talking about are inevitably quite contentious.

The success of our work at the Census Bureau is totally dependent on whether the populous views the statistics as credible. So we must maintain an absolutely neutral stance on this, and much of my day is involved with making sure that in fact and in perception we attain that image.

When you were picked out for the role earlier in the year there was some controversy over things you had said in the past about using statistical adjustments for parts of the population that tend to be undercounted. Did you have concerns that your role might become politicised before you even began?

It was easy for me, I guess, within my own mind not to have those fears on that particular issue. The Supreme Court in the United States had a ruling at the end of the 90s that took off the table the option of statistical adjustment for reapportionment purposes, and I was able to say that. I think it’s important to separate in my role statistical and technical issues from political issues and to keep good boundaries between them. I must keep my statistical mind open, but follow those rules as interpreted by the court.

“There will be an internet option in 2020 for sure. Having said that, no one has been able to tell me what the internet of 2020 is going to look like”

What do you think will be done differently at the next census in 2020?

There are a couple of things. One is that we in this country spend a lot of money on a census; our rate of inflation of cost exceeds that of most countries. Cost efficiencies must be the key watchword, I think, of planning for the next census, and that has to be done without diminution in the quality of the census. That’s the key challenge, I think.

Secondly, we’ve already committed to a vision that would be a heavily mixed mode census design – there will be an internet option in 2020 for sure. Having said that, no one has been able to tell me what the internet of 2020 is going to look like. So we have to remain pretty nimble in terms of evaluating and choosing applications and platforms for that.

We also believe that one source of cost efficiency gains would be consideration of administrative records in ways that haven’t been used before. So we are examining three issues: the data quality of administrative records, the legislative support for such use, and the reactions of the American public to that proposal. Would they view this as a wonderful gain in efficiency on the part of federal government or would they worry about threats to privacy in the use of those records?

It is the season of censuses around the world these next two years – I just visited my friends at the Office for National Statistics in the UK who are preparing for the 2011 census, and my friends in Canada. The problems of the developed societies are quite similar dealing with the diverse new immigrant groups, trust in government statistics, all of this while trying to be efficient. So we, in some sense, share the experience.