purpose of politicizing an election or getting more people registered to vote, 30-some-thousand people, you have no trust in the people, because you know they will do anything for political reasons.

When you have people who will take personal FBI files and use them for political reasons, only, you have no trust at all. I can assure you. And what we have here is a question that really is not about the census, it is totally about money and power. And you, yourself, I think said there was $180 billion as far as distribution here. How can we trust anybody to guess at numbers when they continually will lie, will do things like give away citizenship to felons, not even check 200,000 people on their backgrounds for political reasons only. How can you expect us to trust that?

Mr. PRICE. Well, I think you can trust it because the Census Bureau has an impeccable reputation for its independence and its statistical

Mr. LATHAM. So did the FBI, so did the INS, before this and it became a political tool for them to use for political purposes only. Mr. PRICE. In this instance, whatever the statistical design that is going to be used in 2000, it will be open for public review for a year in the next year. There will be no changes after that is done.

USE OF SAMPLING

Mr. LATHAM. There are laws on the books having to do with privacy and also with citizenship and they were totally ignored and I would not expect, because of the character of people involved, for any change to happen here. And I am just telling you it is not directed at you but I am just telling you that it will be used-we have no reason to believe it will not be, because of past experience and a track record. There is no question about it.

I have, to get more to the sampling question, my understanding is that in the proposal that when you get to 90 percent, you will sample for the last ten percent, is that about right or is that the plan?

Mr. PRICE. The plan is for the 60,000 tracts in the country that each of those we will get a mailed response, let us say at 70 percent. Then for the 30 percent that did not send their forms back, we will sample two-thirds of that to get up to, so that we will have information on 90 percent. The combination of the mail response and the sample will assure you of the information on 90 percent of the housing units in each tract.

Mr. ROGERS. Would the gentleman yield on that?

Mr. LATHAM. Yes.

Mr. ROGERS. How will you know when you have gotten to 90 percent since you do not know what 100 percent is?

Mr. PRICE. It is based on, as you say, the crucial Master Address File. To the extent that you have got so many housing units, when you draw your sample you eliminate, in effect, 10 percent of the tract's housing units from the housing units that have not responded by mail.

And then, for all of those remaining housing units, you go to send the enumerators out, to go collect information as if they were doing 100 percent in the old methods.

Mr. LATHAM. That is the question I had, how do you know when you are at 70 percent if you do not know what 100 percent is, unless you actually count them.

Mr. PRICE. It is 100 percent of the mailing of

Mr. LATHAM. Right, which you cannot show us. You have not got it ready yet and it is going to be right up to the 12th hour when that is finally available and if it is wrong then we are done.

Mr. PRICE. We need the best possible mailing addresses whether we do sampling or not, but we really need it if we are not going to do sampling.

INTEGRATED COVERAGE MEASUREMENT (ICM)

Mr. LATHAM. I would suggest there is some level of confidence that needs to be a part of the equation. And I have not heard anything to give me confidence.

How do you know in a household if they say there are four people in there and there are actually 20, how do you know there are 20, if you are guessing?

Mr. PRICE. We are not guessing. When we do the ICM

Mr. LATHAM. What do you do, surveillance or something?

Mr. PRICE. No, no. We use only information we actually collect in the sample. The ICM is an independent sampling operation, it is called, Integrated Coverage Measurement, and we go to 750,000 housing units. We pick up some missing housing units that are not in the standard MAF, so we can get some correction there. But we also send people to every housing unit in the ICM and if we do not get mail back on those we send somebody there. They ask more questions, and spend more time there and they find out more often that there is a college student that may have been registered twice, or there is a person who was there but was not put on the first form.

That information is compared to the first set of information and reconciled. To the extent that people then admit that there was somebody there that they did not admit to the first time, that is information that we then use.

Mr. LATHAM. Well, the example you used was they may say there are four people but there may be 20 people there.

Mr. PRICE. If they do not say

Mr. LATHAM. I mean how do you know that there are 20 people there for your guessing?

Mr. PRICE. We do not know. All we know is that the first time we went and they told us there were two people, and the second time we go and they say there are four people. And then we go back and try to reconcile, well, you said two and then you said four, which is it? And we get sometimes, often enough, that it was four but that is evidence that we use that there are more people there than the two that we would get in the old-fashioned method.

Mr. LATHAM. So, with your guessing, you could say because they said there were two and there were four, that actually there were 20 in there?

Mr. PRICE. No, we say there are four.

Mr. LATHAM. Where did you get the example before about 20 people in the house?

Mr. PRICE. We would only calculate 20 people when we got some solid information that there were 20 people. We do not—in the sample survey we only use the data that is collected.

Mr. LATHAM. I had a personal experience just in January. I was in my apartment with my wife here in the District and a census person came by and we were moving out at the end of the month and she went through the question, put it in her computer, and she was going to check back in three weeks and we were going to be out the next week. In her computer system she could not do that. She had to know how she could get a hold of me at that address with my phone number here when we had told her five times we were moving the next week.

There is something in the system that is not taking into account the realities of life here.

Mr. HOLMES. Well, I guess I will try and address that one, sir. What you are referring to is one of our sample surveys.

Mr. LATHAM. That is why I have no real confidence in the sampling.

Mr. HOLMES. Well, but again, the sample, doing a sample survey is radically different from doing a census using statistical methods. The reason it is different is that in a census we make an attempt or provide an opportunity for everybody to participate. Only after you have gone through all kinds of efforts to get everyone to participate, do you at that point apply a sample.

Now, what you have described in terms of the survey is just the opposite. Rather than making an attempt to get in touch with everyone we select a small group and each of those particular addresses are the ones that are part of the sample and it is the responsibility of that person to make sure that they get the information from the person that is there.

I do not question the fact that there may have been a problem with the instrument and she could not go forward or she could not go backwards to get some information but that situation that you described is a little bit different than the census.

Mr. LATHAM. You are saying that that was part of sampling.

Mr. HOLMES. No, sir.

Mr. LATHAM. Well, that is what you just said.

Mr. PRICE. There are two different kinds of surveys.

Mr. LATHAM. You are sampling the population.

Mr. PRICE. Some surveys the sample is based on the address, where you just come back and survey whoever is living there. Mr. LATHAM. Right.

Mr. PRICE. Other surveys, you are supposed to follow the person. This apparently was a survey where they were directed to try to follow the person.

HIRING CENSUS TAKERS

Mr. LATHAM. Let me just make my point and that is that the results would not have any validity based on my personal experience. We have virtually zero, in reality, unemployment. Where are you going to get the people, say in Sioux City, Iowa, to do the sampling? Are they going to be competent?

Mr. PRICE. They will have to hire many more people if we do not do sampling. If we have to get 100 percent of housing units instead

of 90 percent of housing units, the difference between 70 percent and 90 percent is two-thirds as much as the difference between 70 and 100.

So, the number of people who have to be hired is going to be much tougher to meet in those Sioux City areas that have low unemployment.

Mr. LATHAM. I think your results are going to be skewed, if you do not have competent people to begin with. I do not know where you are going to find them. It is critical to accuracy if you are using the basis of sampling whether the results at 70 percent are wrong or at 90 percent are wrong. I do not know where you are going to find the people to ensure accuracy in this process.

Mr. PRICE. That is, as the GAO has said in its report, that is one of the biggest challenges we face and when you have to find 100,000 more people in the country not only is it going to be hard to find the people but their ability to do the work may be not as good as the first 250,000 people. So, to the extent that there is a problem in the labor market it tilts our decision in favor of sampling.

2000 COMPUTER PROBLEM

Mr. LATHAM. One question I have asked virtually every panel that we have had through the process this year is regarding the year 2000 computer problem, obviously, we are looking at something that is very important in the year 2000. What if the computers implode here?

Mr. PRICE. One of the things the IG talks about is all the new software design we are doing. But when you are doing new software design then you can avoid that problem. In large part, the Census Bureau and the BEA, which is another part of the ESA, are well ahead of most other Government agencies and they are aware of these problems. They have inventoried them and they are confident that by next year, early next year, we will have replaced those programs that are a problem and we will be able to test them and be ready well in advance of January 2000.

But we are creating whole new software that will not have that problem.

Mr. LATHAM. You can assure us of that.

Thank you, Mr. Chairman.

Mr ROGERS. Mr. Dixon?

POLITICAL ENVIRONMENT OF CENSUS

Mr. DIXON. Thank you, Mr. Chairman.

I would like to welcome both of you gentlemen here today and as the Chairman has indicated, it certainly is a hot seat. It is probably a hot seat, in my opinion, because as I understand, from both of your backgrounds and the people that are here, that you have spent considerable time in developing the science of how to count people. And on this side, it is the politics of the issue. And, so, when the science and politics hit together there is always some kind of clash.

But I really do not think, Mr. Price, that you get it. I think you are beating your head against the wall and I only speak for myself on this. You have been challenged this afternoon as to the accuracy

of sampling, notwithstanding what other scientists and professionals in the area say about sampling. You have been accused of not being prepared today and the strong inference from reports is that you will never be prepared to do the job of enumeration.

You have been told directly that the Committee or at least some Members of the Committee are concerned about political influence. In fact, one Member, Mr. Kolbe said, this is a political issue. So, I do not think you get it because, unless there is some drastic change you are not going to be allowed to sample. If the courts say that it is constitutional that moves you no further forward.

Because, as you said, the under count involves, in the main, Latinos and African-Americans. The perception is that they would be inclined to vote Democratic. There are other concerns about States that know they have had declining population, losing representation. So, you do not get it.

You are not going to be able to do sampling, one. And two, when it comes out and is disclosed that a lot of people were not counted, you are already being set up to say that you were inefficient and not prepared. That message is very clear today. Now, my interest, coming from a State that has gained population, is that we have the most accurate census available. But, in the real world, we will not be able to get that. So, rather than enumeration being a contingent issue, for my interest I want it to be your main interest.

And I want you to realize that when you come before this Committee, in the main, you come to a very hostile environment on what is a political question. And from that perspective, I do not think you are prepared today.

Because I think, not as a professional, but in the real world you have your head in the sand. You are still under the belief that if a 1,000 scientists demonstrated the accurate way to count that you can turn this Congress around on the issue.

Now, unless there is a drastic change in the next two years, you are not going to be able to do that. So, setting that issue aside, I still want the most accurate count, Mr. Holmes, that you can make. I suggest to you that you get busy because it is going to be a short count and you are going to get blamed for it and there are going to be a lot of reports waived around and a lot of testimony about forecasting. I tell you what happened in 1990 will happen in the year 2000 and you have heard that here today.

I do not mean to lecture to you at all, but you must see in the real world what is happening. And you are trying to fight back without a sword. You are trying to fight back with the logic of the science, you know, that is who are you going to believe me or your lying eyes?

Now, that is what is occurring in this room. And I am not suggesting that there is anything_evil about it; it is a political issue. You are not prepared today. These reports say you are not prepared. I remember 1990 and the disaster there suggests to you it is coming for 2000, you are going to take the heat for it.

And, although, we are not challenging your integrity, we are concerned with somebody out there in politics at the last minute changing the numbers. The handwriting is on the wall. So, I do not have any questions for you except get busy with the enumeration process. Do not look at it as a contingency. Look at it as a reality