home
in espanol
mission statment
news
board of directors
A.
Dr.
Joe
reports to congress
contact us


Friday, February 23, 2001

Excerpts from Board Member Mark Neuman`s Appearance on NPR`s "Talk of the Nation"


[To Listen to the entire show, Click here

JUAN WILLIAMS (HOST): Mark Neuman, let me begin with you and ask about the upset that you could hear in the congresswoman`s voice. She feels as if the president betrayed her. She feels as if this was done in the dark of night, so to speak, while the bombing of Iraq was going on and the president was visiting Mexico. Is this fair or do you think that, in fact, the president is playing politics?

Mr. MARK NEUMAN, (US Census Monitoring Board): Well, the secretary of Commerce has had the authority by law for the last 70 years to determine the form and content of the Census and make decisions related to it. The last secretary of Commerce had decided that he wanted to delegate it to a political appointee, who was the director of the Census Bureau at the time, and that appointee has since departed the Census Bureau. So I don`t think it`s such big news, and the bottom line is that Secretary Evans has not made a decision about whether or not to adjust the Census. He`s going to weigh all the facts, including the recommendations of career professionals at the Census Bureau.

But I`m sure most members of Congress wouldn`t want to leave up to career professionals at the Pentagon how many missiles we should buy. These are matters of national policy. The secretary of Commerce believes that someone who`s confirmed by the Senate, appointed by the president, ought to be making a decision of this importance for the country.

WILLIAMS: Mark Neuman, I think that you characterized the former head of the Census Bureau as a political appointee. Clearly, the Democrats have been suggesting that that person was more of a scientist, an impartial observer of events, and less likely to be influenced by politics. No one could make that claim about Donald Evans. I mean, clearly, he`s a political player and the former campaign manager for President Bush. So is it unfair then to say that this is going to be more of a politicized decision-making process now with Mr. Evans involved, or do you think it was also politicized when we had the former Census director making the call?

MR. NEUMAN:  I agree with you, and it was very interesting to contrast. Secretary Evans wants to study the numbers. We still don`t have a recommendation from the career professionals about whether we could improve the accuracy of the Census count by using statistical adjustment, so he`s not made a decision. Yet, it appears that Dr. Prewitt before he`s had a chance to evaluate all the data, has already said we need to adjust the Census. So I think Secretary Evans wants to listen to both sides of the argument, wants to review all the data, and the great thing about the secretary of Commerce, I think, in this process is that we`re going to know the reasons why he chose to make a decision one way or another. There will be a sense of transparency, an openness, and I think that`s what the American people want.

WILLIAMS: So you don`t think the deal is done? You think this is still open and it`s possible that Donald Evans may decide to have an adjusted count?

MR. NEUMAN: That`s a possibility.

WILLIAMS: All right.

MR. NEUMAN: He`s keeping an open mind.


WILLIAMS: And, in fact, when you look at the numbers in the House of Representatives--I believe there are about 10 or 11 seat difference between Republicans and Democrats at this time--it is very important to see exactly which areas of the country will be given more seats in the aftermath of the 2000 Census and which areas of the country will lose seats. And also, we have redistricting coming where new district lines will be drawn for congressional areas, and again, that will be impacted by the Census count. So there are lots of areas where there`s a real consequence to what happens with the Census, right, Mark?

MR. NEUMAN: Well, I think we need to look at some of the facts and compare what we know about the 2000 Census and what we knew about the 1990 Census. We went from, in 1990, a black undercount rate of 4.57 percent, and we`ve reduced that 67 percent, down to 1.60. And this is a personal issue in my family. I`m a son of a Central American immigrant, so I have two younger brothers who don`t speak any English. They`re affected by the undercount. I`m looking at the Latino undercount, in 1990 was almost 5 percent. We cut that to 2.2 percent, a 60 percent reduction. And American Indians living on reservations, 12.2 percent down to 2.77 percent, a 75 percent reduction.

Remember, Juan, it`s not just counting everyone in the Census that`s important. It`s counting them at the right address. And most people think that statistical adjustment, if you have a neighborhood that is undercounted at a rate of 40 percent, that what adjustment does is correct that undercount, puts the 40 percent back where they were missed. It doesn`t do that. That`s what we know for certain about the method...

WILLIAMS: Well, what does it do?

MR. NEUMAN: It spreads that undercount throughout the country in different regions based on these adjustment factors that are assigned to the post-strata, which are individual adjustment factors based on this process of statistical estimation. A great example would be if you asked Sandy Allen, who`s a very hard-working councilwoman here in Ward 8 in Washington, DC, does she care most about how many people get counted in Washington, DC, or how many people get counted in Ward 8, you can bet she`s much more concerned with how many people get counted in her neighborhood.

And that`s really the issue for adjustment. I`m very familiar with the 3rd Ward and the 5th Ward in Houston, Texas. And what we found in 1990, looking at the data, it was a 45 percent undercount in the 3rd Ward, which is a heavily African-American low-income group of neighborhoods. Then we looked at the Rice University-South Hampton area, and there was a 5 percent overcount. Yet the adjustment process would have added back the same number of people to both neighborhoods, so we only went from a 45 percent undercount to a 41 percent undercount. And we made an overcounted neighborhood even more overcounted. That`s the...

WILLIAMS: You mean overcounted by sort of making it an actual count, but only a 5 percent correction in the black and Latino area, the poor people in Houston.

MR. NEUMAN: No, actually, we added the same number of actual people to both wards.

WILLIAMS: Right. But I`m saying, since you had a 45 percent undercount in one area and a 5 percent overcount in another--oh, you added them to both areas.

MR. NEUMAN: Exactly.

WILLIAMS: So you made it a 10 per--I get it, you even made it worse in some ways.

MR. NEUMAN: So there`s a bigger overcount in the overcounted neighborhoods which are all white, in this case, and there is very, very little reduction in the undercount in the most undercounted neighborhoods.

WILLIAMS: Very quickly, Mark, how do they know the degree of the undercount? Since they can`t find the people to count them in the first place, how do they know that they were missed?

MR. NEUMAN: Well, that`s based on this survey, this poll, of 314,000 households. And we`re going to base--we`re going to match the results of that poll to the actual Census and we`re going to...

WILLIAMS: Oh, you do the polls separately from the Census.

MR. NEUMAN: Right, so we`ve already had a full head count, and now we`re going to compare that with this poll. And then we`re going to always assume that the people counted in the survey is the correct result. And then we`re going to assign adjustment factors to 448 separate groups of people based on different characteristics.


WILLIAMS: All right. Thanks for your call, Jerry. I`ll take that crab dinner. Now, Mark, when you hear Jerry say, look, you know what the deal is, you know exactly how Mr. Evans is going to rule, do you say, `Oh, you know, I guess you`re right, Jerry; I mean, there`s no getting around it. In two weeks, Mr. Evans will announce no estimating, no adjusting, no sampling, he`s just going to go with the raw head count`?

MR. NEUMAN: We`ve already had sampling. What we did was we went out and took this survey of 314,000 households. The work that the bureau is doing to evaluate the 2000 numbers, that`s not being interfered with in any way. And, in fact, Don Evans went out to the Census Bureau two days after he was sworn in, this was--no one could believe it out at Suitland, Maryland. He went out there to really praise the bureau, praise the work that they are doing, and you talk to anybody in the career staff at the Census Bureau, and they`ll tell you that this secretary of Commerce is not interfering with their work to evaluate the data at all.

WILLIAMS: You said there`s already been sampling?

MR. NEUMAN: There has already been this--the largest survey in history which was supported in order to evaluate the Census. The question is not...

WILLIAMS: Wait, wait, wait. The Census--you mean the largest Census in history, or do you mean the survey independent...

MR. NEUMAN: There was a post-enumeration survey called the Accuracy and Coverage Evaluation.  

WILLIAMS: But how does that involve any sampling or adjusting for people who may have been missed by the Census takers?

MR. NEUMAN: The debate is whether the bureau is going to use the results of this survey of 314,000 households to reformulate the population by creating adjustment factors based on the bureau`s assumptions about the undercount. So the ACE adjustment is based on results taken from this 314,000 person survey.

WILLIAMS: Let`s take a look at the e-mail. Here comes one and it`s from Richard McFarlane in Orange, California. And Richard writes, `I thought the Supreme Court said statistical sampling was unconstitutional and that only a direct head count could be used as a Census. Did I misunderstand?`

MR. NEUMAN: Well, the Supreme Court of the United States said that you have to actually take a full head count. What the Clinton administration originally proposed is to count 80 percent of the population and then take a survey for the remaining 20 percent. And then combine that data to give you an estimate. So now what the Supreme Court forced the administration to do was, in fact, to take the full enumeration and now we have this post-enumeration survey, this ACE. And of course, remember that survey that was taken after the Census is also conducted by Census Bureau employees.

WILLIAMS: But wait, wait. I`m still confused. Did the court say you have to have a head count and no enumeration allowed?

MR. NEUMAN: Right.

WILLIAMS: No sampling.  Let me ask you, Mark, do you agree [that all scientists favor sampling]?

MR. NEUMAN: Oh, a very prominent scientist on the National Academy of Sciences panel studied this issue. Charles Schultz, a former adviser to Jimmy Carter, told the US Commission on Civil Rights, where I testified, `You shouldn`t use any of those numbers,` meaning adjusted numbers, `at the block level. There will be a lot of errors and probably some added errors at the block level because of the sampling problem.` So what he`s saying is, yes, maybe the adjusted data will be good for federal funding purposes where using it at the city, county or state level, or maybe it should have been done at the apportionment level, but what we know is that there`s a false promise. If we think that we`re going to correct the undercount for blacks and Latinos and American Indians and Asians at the neighborhood level, I can promise you that no methodology that`s available to the Census Bureau is capable of doing that.


[END Excerpts]




to Web ...
People, not statisti...