Washington, D.C.
October 18-20, 1995
Martha Farnsworth Riche, Director
U.S. Bureau of the Census
When Duane Webster sent me an invitation to speak to ARL, he wrote, “While there is a range of institutional characteristics represented in ARL, they all face the common pressures of leading a large organization through an era of extraordinary change in information technology, user needs, and economic models.” We can say the same thing about the Federal government. Today I will talk specifically about the Census Bureau, the government’s major data collector for statistics about, first, our people, and, secondly, our economy.
At the Census Bureau we collect 75 percent of the data that goes into the national economic accounts, such as the Department of Commerce’s reports on retail sales or housing. A third of our business is from other federal agencies, collecting data regarding employment rates and health care issues, for example.
Over the last two decades both of the areas we measure have grown. Our population has grown by 22 percent and our economy has grown by nearly 70 percent in real dollar terms, but government spending on statistics has remained level during that whole time. Furthermore, not only have our population and economy grown, but they have become more diverse. They are changing, and we are very much in the situation of a tailor trying to measure someone for a suit while that person is running full tilt.
What happened with our predecessors, and I’ll speak for my colleagues at the Bureau of Labor Statistics and the Bureau of Economic Analysis as well, is that they kept waiting for new resources. New initiatives were put in the budget every year, but they would get cut out and postponed. People in our field noticed that our economy was changing and that something needed to be done about it, but Congress just didn’t give us the money to make the changes we needed.
It was clear to us working with the Clinton administration that, if our predecessors couldn’t get the resources, we sure weren’t going to get them at a time when the government was talking about cutting the deficit. So we knew we needed to rethink our processes.
Let me give you an example of how much delayed maintenance there is on statistics. One of the things we do is maintain a coding system called the Standard Industrial Classification (SIC). This system enables us to take data from all the economic activity in the United States and put it together in a coherent statistical whole, like the Gross Domestic Product.
That coding system was last conceptually redesigned in 1973. As a result, the fastest growing industries in America today do not fit easily into its calculations, and must be placed into the “all other” category, making it even more difficult for our colleagues in the private sector to adjust to the nation’s changes. So we created a new industrial classification system, directed by Office of Management and Budget (OMB), that will not only reflect our current economy, but will be a North American industrial classification system, working together with Canada and Mexico.
We have other areas needing delayed maintenance, as well. For instance, the OMB is particularly concerned about poverty threshold, which it defines, but we produce. If we are going to do the work that needs to be done on pressing public issues like that, we have to find the money. So we have set priorities, tightened our mission, and examined our activities to see what projects we can stop doing, in order to do what needs to be done.
That brings me to our subject today, data dissemination and transmission. What we found when we examined our own data transmission and dissemination, especially in light of this new information and technological environment, is that we were spending too much money delivering too little data to too few people and taking far too long to do it. That is what led us to electronic transmission.
Let me give you an example of the dimensions. Last year we sold 40,000 printed reports of our data. We are currently getting 60,000 hits a day on our Internet site.
This is extremely exciting to us, because it reveals a whole new audience. For example, I recently went around the country to tell our data users about the current state of statistics in Washington, D.C. In one town a dean of a small college told me, “I never would have gone to the library to get census data. There are so many volumes that I wouldn’t have known how to find my way through them. But now I can go online and to get data, and I’ve even opened computer folders for my colleagues.” Both the Equal Opportunity Office and the admissions office had a folder. He said, “Now I can get the data I want quickly, because I don’t have to go through all the data I don’t want.”
We are a labor intensive agency, and electronic transmission allows the analysts’ time that had gone into preparing each carefully checked statement and number in the printed reports to be turned to looking at how to measure poverty, how to measure race and ethnic origin, or how to get an SIC code for the OMD.
However, I do recognize that we are in the dawn of a new age, and with that comes problems. Archiving is one of those problems. In our case, where we have millions of numbers out of a given data file, what happens to that when the technology moves on? Will people be able to read it? It is a very important issue, and I look forward to partnering with whatever agency or organization takes leadership in that effort.
Another issue that is very important is what will happen when data, and the tables and appendices that educate you how to think about data, is disseminated electronically into footnotes, asterisks, and confidence intervals. Our chief statistician, who is at the Office of Management and Budget, and I are also very concerned about statistical literacy in a time when we are giving people data in a way they have never had it before.
That is where we stand right now at the Census Bureau.