1260 1 Q. If it is closer to zero, then that suggests that 2 the regression analysis, in fact, does not show a high 3 degree of correlation between two variables? 4 A. No. 5 MR. SUBAR: Your Honor, with your permission, 6 could I approach the easel? 7 THE COURT: Sure. 8 MR. SUBAR: I submit that although I am going to 9 draw another couple of graphs, they will be very simple. 10 Q. Professor Cain, you see, don't you, on the easel 11 -- and if you don't, I would invite you to approach the 12 easel -- two graphs that I have drawn? 13 A. Yes, I see them. 14 Q. On each graph there is a series of dots? 15 A. That's correct. 16 Q. On the upper of the two graphs the dots are 17 arrayed in more or less a straight line; would you agree 18 with that? 19 A. That's correct. 20 Q. In the lower of the two graphs the dots are 21 arrayed all over the place, if you will? 22 A. Yes. 23 Q. With regard to the upper of the two graphs, if 24 you knew nothing else about the matter studied by that graph 25 other than the fact that that graph represents the results 1261 1 of a regression analysis, would you say that the R squared 2 statistic would be closer to 1 or closer to zero? 3 A. That would be closer to 1. 4 Q. With regard to the lower one, would it be closer 5 to 1 or closer to zero? 6 A. I assume you were trying to represent randomness 7 there. Indeed, if there is randomness in that 8 representation, that's right, it would be closer to zero. 9 Q. Take a look at plaintiff's exhibit 586, if you 10 would. Would you happen to know the R squared statistic 11 with regard to that graph? 12 A. It's about .15. 13 Q. The reason that that R squared statistic is 14 closer to zero than to 1 is because each of the hundred dots 15 on that graph represents a couple of thousand blocks, is 16 that not correct? 17 A. No. It represents 298 blocks. 18 Q. This graph represents all of the blocks in the 19 State of California that are populated, doesn't it? 20 A. No. It is ones that have -- yes, populated and 21 have Democratic registration -- has voters in them. They 22 have to have voters in them. It doesn't represent 23 unpopulated, nonvoting areas. 24 Q. But the point is that each of those dots 25 represents a large number of blocks? 1262 1 A. Right. 2 Q. Within any one of those dots the blocks could 3 vary greatly with regard to undercount or presumed 4 undercount, is that not correct? 5 A. That's correct. But -- okay. 6 Q. Could you take a look, please, at Plaintiff's 7 Exhibit 583. On the second page of Plaintiff's Exhibit 583 8 there are five columns, the last two of which are labeled 9 Dem Ratio and Dem Rat 2, is that correct? 10 A. That's correct. 11 Q. What is the definition of the term Dem Ratio? 12 A. That's the Democratic/Republican -- Democrats 13 over total voters. 14 Q. What is the definition of Dem Rat 2? 15 A. That was an attempt to -- it's the square of the 16 term. It is an attempt to see if some form other than a 17 straight line is appropriate. 18 Q. What is the square of .6? It is not 1.2, is it? 19 A. No, it is not. 20 Q. The square of .5, which is on the second row of 21 that column, isn't 1.017, is it? 22 A. No. 23 Q. There is an error here, isn't there? 24 A. Yes. It is not one that I relied on. I didn't 25 use that function or form, so I didn't spend a lot of time 1263 1 looking at it. I am not sure what it is. 2 Q. Dr. Cain, in your view, it is only Republicans 3 who oppose the concept of adjustment that is the subject 4 matter of this trial, isn't that right? 5 A. Only Republicans? 6 Q. Isn't that what you said at your deposition? 7 A. I said that Republicans particularly with an 8 interest in redistricting saw the political consequences of 9 this and were opposed to the adjustment, that's right. I 10 suspect, again, John Q. Public probably didn't have a view. 11 Q. But, in fact, there are plenty of Democrats who 12 oppose adjustment, do you know that? 13 A. You mean Democrat citizens? 14 Q. Democrats in official positions. 15 A. I don't know what survey -- I have no idea what 16 the distribution is. 17 Q. Do you know that the State of Wisconsin is an 18 intervenor defendant in this case appearing through the 19 Attorney General of that state? 20 A. Yes. 21 Q. Do you know that the State of Oklahoma is an 22 intervenor defendant in this case, opposing the adjustment 23 sought by plaintiffs? 24 A. Yes. 25 Q. Did you know that the Attorney General of the 1264 1 State of Wisconsin is a Democrat? 2 A. No, I don't know the Attorney General. 3 Q. Could you please turn to -- 4 THE COURT: Can we stipulate that he is? 5 MR. SUBAR: I would be happy to do that, your 6 Honor. 7 THE COURT: Thank you. He is. 8 MR. SUBAR: If the plaintiffs wish any 9 confirmation of that fact, they can simply look at a piece 10 of paper that I am about to show them. 11 MR. SOLOMON: He is if it is in one of these 12 pieces of hearsay that is subject to the objection. Yes, we 13 will stipulate to that. 14 MR. SUBAR: I take it the plaintiffs will also 15 stipulate to the fact that the Governor, Lieutenant 16 Governor, Secretary of State, and Attorney General of the 17 State of Oklahoma are Democrats, and that both houses of the 18 legislature of the State of Oklahoma are much more heavily 19 Democratic than Republican. Again, if plaintiffs wish -- 20 MR. SOLOMON: I can't stipulate to anything close 21 to that. 22 THE COURT: Sure you can. 23 MR. SOLOMON: Without seeing the document that I 24 have now been provided. What is it that you want us to do 25 now? Who is it? 1265 1 MR. SUBAR: Oklahoma, the Governor is a Democrat, 2 the Lieutenant Governor is a Democrat, the Secretary of 3 State is a Democrat, the Attorney General is a Democrat, and 4 in the Senate of the State of Oklahoma Democrats outnumber 5 Republicans by 37 to 11, and in the House Democrats 6 outnumber Republicans by 69 to 32. 7 MR. SOLOMON: You have read 705 correctly. 8 MR. SUBAR: Thank you. 9 A. I might point out that these are states without 10 minorities, substantial concentrations of minorities. 11 THE COURT: Republicans are minorities. 12 THE WITNESS: Yes. 13 Q. Not to belabor a very elementary point, Dr. Cain, 14 but presidents in this country are elected on the basis of 15 the number of the electoral votes that they get with so many 16 being allocated to each state, is that correct? 17 A. You are referring to the Electoral College, yes. 18 Q. Yes. And the allocation is based on the census 19 numbers, is that not correct? 20 A. Indirectly, yes, that is right. 21 Q. In your view, you can categorize states as to 22 whether they are more likely to send a Republican delegation 23 to the Electoral College or a Democratic delegation to the 24 Electoral College, is that not correct? 25 A. You could do that. 1266 1 Q. If you were to do that, in your view, you would 2 say that California and Pennsylvania are swing states, while 3 Wisconsin tends to vote Democratic and Arizona tends to vote 4 Republican, is that not correct? 5 A. We talked about that in the deposition, that's 6 correct, Mr. Subar. 7 Q. Did you know, Dr. Cain, that had there been an 8 adjustment last July, California and Arizona would each have 9 gained a seat in Congress and therefore in the Electoral 10 College, and Pennsylvania and Wisconsin would each have lost 11 a seat in Congress and therefore in the Electoral College? 12 A. I have read that, yes. 13 MR. SUBAR: I have nothing further. 14 THE COURT: Any redirect? 15 MR. SOLOMON: A trivial amount. 16 REDIRECT EXAMINATION 17 BY MR. SOLOMON: 18 Q. Mr. Subar read to you from your deposition about 19 giving an opinion on adjustment. He stopped reading where I 20 am going to pick up, on page 67 to 68. As a lay person, he 21 asked you, do you have a view. "It would appear to me 22 reading the papers that there is good reason to believe that 23 the adjusted data should be used." 24 Did you give Mr. Subar the reasons why you 25 believed that at your deposition? 1267 1 A. Why I believed the adjusted data should be used? 2 Q. Yes. 3 MR. SUBAR: Objection, your Honor. This calls 4 for his lay view, not his expert opinion, by the terms of 5 the question. 6 MR. SOLOMON: No. 7 Q. Did you give him the reasons that you had at the 8 deposition? 9 A. Yes, I did. 10 MR. SUBAR: Objection. 11 THE COURT: Did he give the reasons, without 12 saying what they are. 13 Q. You drew a distinction repeatedly, Mr. Subar I 14 don't think was listening, between this not being a 15 gerrymandering issue but this being a bias or 16 malapportionment issue. 17 A. That's correct. 18 Q. Could you sharpen that for us, please? 19 A. Again, Mr. Subar focuses on two things. He 20 focuses on the apportionment, which affects the states, and 21 he focuses on the gerrymandering issue, which is how the 22 lines actually get drawn. Once again, my point is that the 23 far more important effect, the effect that would be known, 24 is based on simple intuition, which is, the more of your 25 guys you have and the fewer of their guys they have, the 1268 1 better off you are in redistricting, because you have more 2 strength that you can spread around and create additional 3 seats or influence seats or whatever. 4 So what we are talking about is a 5 malapportionment that systematically along racial and ethnic 6 lines that was produced because it has partisan effects. It 7 debases, fundamentally debases the votes of those 8 individuals at every single point up and down the ladder, 9 not only in Congress, not only the assembly level, not only 10 in the state senate level, but all the way up and down in 11 water districts. 12 We do this election after election, we do this 13 decade after decade. We have been fundamentally debasing 14 our right to vote that was recognized by the court through 15 this malapportionment. So it is an inequity which is 16 perpetuated as a result of undercount. 17 Q. Last question. Mr. Subar tried in one sentence 18 to recapture all of your testimony and asked whether it was 19 just your testimony that you believe that the partisan 20 effects were knowable to Secretary Mosbacher. Based on what 21 you know, based on inferences that you are prepared to draw 22 as an expert, based on the tome that you have read and your 23 experience as a political scientist, were the partisan 24 effects of the decision against adjustment known to 25 Secretary Mosbacher? 1269 1 A. Yes. 2 MR. SOLOMON: Thank you. 3 MR. SUBAR: One question, your Honor. I think it 4 will be just one question. 5 RECROSS-EXAMINATION 6 BY MR. SUBAR: 7 Q. Your views with regard to the so-called 8 malapportionment, as you used the term, assume that the 9 results of the PES are correct, is that right? 10 A. Again, it assumes that a consensus of 11 statisticians and advisory panels believe that it is 12 correct, that's right. 13 MR. SUBAR: Nothing further, your Honor. 14 THE COURT: Thank you very much, Dr. Cain. You 15 may step down. 16 (Witness excused) 17 THE COURT: We will recess for lunch, resume at 18 2:15. 19 (Luncheon recess) 20 21 22 23 24 25 1270 1 AFTERNOON SESSION _________________ 2 2:15 p.m. 3 MR. ZIMROTH: Good afternoon, your Honor. Your 4 Honor, the plaintiff's next witness is Barbara Bailar, B A I 5 L A R. With your Honor's, my former colleague at the 6 Corporation Counsel's office David Goldin, will examine. 7 THE COURT: Mr. Goldin, welcome. 8 MR. GOLDIN: Thank you, your Honor. 9 THE COURT: Mr. Goldin would you call Dr. Bailar. 10 MR. GOLDIN: Dr. Bailar, if you would take the 11 stand, please. 12 BARBARA BAILAR, 13 called as a witness by the defendant, having 14 been duly sworn, testified as follows: 15 THE COURT: Would you state your full name and 16 spell your last name for the record. 17 THE WITNESS: Barbara Bailar, B A I L A R 18 THE COURT: Mr. Goldin. 19 MR. GOLDIN: Thank you, your Honor. 20 DIRECT EXAMINATION 21 BY MR. GOLDIN: 22 Q. Dr. Bailar, where are you currently employed? 23 A. At the American Statistical Association. 24 Q. What position do you hold there? 25 A. I'm the executive director. 1271 1 Q. How long have you held that position? 2 A. Four years. 3 Q. Since 1988? 4 A. Yes. 5 Q. What are your responsibilities in that position? 6 A. As the executive director, I run the office. I 7 also do a fair amount of work getting grants together for 8 the kinds of things that the profession wants to pursue. 9 For example, ethics for statisticians, teaching of 10 statistics in high schools. I also put on professional 11 meetings, continuing education, things for the professional 12 growth of statisticians. 13 Q. Before becoming executive director of the 14 American Statistical Association, where were you employed? 15 A. At the census bureau. 16 Q. When did you join the bureau? 17 A. 1958. 18 Q. What was the first position that you held there? 19 A. A grade 7 mathematical statistician. 20 Q. What was the next position that you held at the 21 bureau other than mathematical statistician of whatever 22 grade? 23 A. An assistant division chief at the research 24 center for measurement methods. 25 Q. Did you at some point become chief of that 1272 1 division? 2 A. Yes, I did. 3 Q. When was that, roughly? 4 A. Early in the seventies. 5 Q. What was the last position that you held with the 6 bureau? 7 A. Associate director for statistical standards and 8 methodology. 9 Q. When did you become associate director for 10 statistical standards and methodology? 11 A. 1979. 12 Q. What were your duties in that position? 13 A. It was to oversee the research done at the census 14 bureau and to examine the methodology done in all of the 15 census's surveys. So there were two division chiefs who 16 reported to me in the statistical research division and the 17 center for survey methods research, the chief mathematics 18 statisticians. And then there was a group throughout the 19 census bureau in different divisions who had a dual 20 reporting responsibility, and they reported to me as well as 21 to their own division chief. 22 Q. To whom did you report? 23 A. I reported to the deputy director and the 24 director. 25 Q. How much associate directors were there at the 1273 1 bureau? 2 A. At that time five. 3 Q. Have you received any honors or awards in your 4 career? 5 A. Yes. I received the silver medal from the 6 Department of Commerce, outstanding performance ratings. I 7 was elected president of the American Statistical 8 Association by its members. I was elected president of the 9 International Association of Survey Samplers. 10 Q. Are you a fellow of the American Statistical 11 Association as well? 12 A. Yes, I am. 13 Q. Have you published any articles or papers 14 concerning statistical adjustment of the census? 15 A. Yes. 16 Q. Let me ask you to turn not binder before you for 17 what has been marked for identification as Plaintiff's 18 Exhibit 709. Do you have that? 19 A. Yes, I do. 20 Q. Is that a copy of your current curriculum vitae? 21 A. Yes. 22 Q. Do you have any particular areas of expertise 23 within statistics? 24 A. Survey methodology and nonsampling errors on 25 surveys. 1274 1 MR. GOLDIN: Your Honor, I would proffer Dr. 2 Bailar as an expert. 3 MR. GLASS: Your Honor, I am assuming this is as 4 an expert in statistics. 5 MR. GOLDIN: That is correct. 6 MR. GLASS: With two caveats, I have no 7 objection. At Dr. Bailar's recent deposition, which was a 8 scant two weeks ago, she disclaimed any expertise in 9 smoothing or loss function analysis. With the exception of 10 those two, if we exclude those, I have no objection to her 11 being accepted as an expert. 12 MR. GOLDIN: With those caveats, the 13 understanding that we are offering Dr. Bailar as an expert 14 in statistics generally and without representation that she 15 is equally familiar with every nook and cranny of the 16 discipline, those are the terms on which we offer her. 17 THE COURT: I will qualify Dr. Bailar as an 18 expert in statistics generally. 19 Q. Dr. Bailar, have you received any compensation 20 from plaintiffs for testifying today or for any work you 21 have done in connection with this litigation? 22 A. No. 23 Q. Do you have any expectation that you will receive 24 any compensation in that connection? 25 A. None. 1275 1 Q. Dr. Bailar, are you familiar with the concept of 2 differential undercount? 3 A. Yes. 4 Q. Was differential undercounting a problem in the 5 1980 census? 6 A. Yes, it was. 7 Q. How did the census bureau know that? 8 A. The census bureau has known about differential 9 undercount since the 1950s, different ways of measuring it. 10 In 1980 it used some of those same methods and saw that 11 there was still a differential undercount. 12 MR. GOLDIN: Let me backtrack one moment and 13 offer into evidence Plaintiff's Exhibit 709 which Dr. Bailar 14 identified a few moments ago as being her CV. 15 MR. GLASS: No objection. 16 THE COURT: 709 is received. 17 MR. GOLDIN: Thank you, your Honor. 18 (Defendant's Exhibit 709 for identification was 19 received in evidence) 20 Q. In preparing for the 1980 decennial census, had 21 the bureau taken any measures to address the problems of 22 undercount, differential undercount, in conducting that 23 census? 24 A. In 1908. 25 Q. In 1980, before the 1980 census was actually 1276 1 conducted? 2 A. Yes. The bureau had looked at different methods 3 of publicizing the census and trying to make people more 4 aware that there was a census going on, had community 5 workers trying to get people aware of the census in the 6 local areas. 7 Q. Did those programs succeed in eliminating either 8 the undercount or the differential undercount? 9 A. No, they did not. 10 Q. At the the 1980 post-enumeration program was 11 conducted, what role was envisioned for it in connection 12 with the 1980 census? 13 A. It was supposed to be as an evaluation of the 14 quality of the census. 15 Q. Was the post-enumeration program successful as an 16 evaluation of the quality of the census? 17 A. As an evaluation program, I think it was 18 successful. It showed the kinds of people that were missed 19 in the census. It showed that there was a differential 20 undercount. It gave us some numbers that showed that there 21 was an Hispanic undercount, something that we thought had 22 occurred but really didn't have any evidence from before. 23 So in many ways it was successful. 24 Q. In your opinion, could the PEP, if we can 25 abbreviate it that way, have been validly used for a 1277 1 statistical correction of the 1980 census? 2 A. No, not in my opinion. 3 Q. Why is that? 4 A. There was a lot of error in the PEP from 5 different kinds of survey procedures. There was a problem 6 with missing data, a problem with matching, and it was a lot 7 of error in a program that you are going to try to use to 8 correct. 9 (Continued on next page) 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 1278 1 Q. Can you describe specifically what the problem 2 with missing data was or what the problems with missing data 3 were? 4 A. With missing data, it was a problem first because 5 the PEP was built on another survey that was going on, the 6 current population survey, and that survey has a certain 7 amount of missing data, nonresponse, and so you have to take 8 that amount into consideration, and then you do this survey 9 and you get an additional amount on top of it so there was a 10 range of missing data in the 1980 PEP of approximately nine 11 to ten percent. 12 Q. And what was the problem, what were the problems 13 with matching count in the 1980 PEP? 14 A. With matching error, part of the problem was a 15 direct result of having to do everything manually. 16 The way the files were set up, the census files 17 in Jeffersonville, there were long stacks where when you 18 were going to look for a form to try to match it, you just 19 had to walk through these aisles, look through the right 20 box, try to sort through the box. Though you might start 21 out in the morning with the best will in the world to try to 22 match things, you know, as the day wore on, your enthusiasm 23 became less and less to find that form, so it was just 24 inherent in the procedure that there was a certain amount of 25 matching error, and that permeated the studies. 1279 1 Q. Did the 1980 PEP suffer from other limitations? 2 A. It had problems, of course, of correlation bias, 3 it had things about erroneous enumerations, the quality of 4 the addresses. 5 Q. Was adjustment of the 1980 decennial census 6 considered by the Bureau? 7 A. Yes, it was. 8 Q. Who at the Census Bureau participated in making 9 that decision? 10 A. There were a group of senior staff people at the 11 Census Bureau who were very much involved in that decision. 12 Q. Did you play any role? 13 A. Yes. I was one of those people. 14 Q. What role did you play? 15 A. Well, I was really, as the statistical research 16 person, I was advising the director on the policy and I 17 participated in all the decisions, all of the conferences, 18 all of the meetings that we had in this -- on this topic. 19 Q. Who was the director of the Bureau at that time? 20 A. Vincent Barabba. 21 Q. Would you please take a look at a document which 22 has been marked for identification and should be in your 23 binder as Plaintiff's Exhibit 606. 24 Do you recognize that document? 25 A. Yes. 1280 1 Q. Would you tell us what it is? 2 A. It's the Federal Register statement on the census 3 undercount adjustment basis for decision, the decision not 4 to adjust the 1980 census. 5 Q. And this is the announcement that there is going 6 to be a decision made on that subject? 7 A. This is that there -- the way the decision will 8 be reached, the critical assumptions and the general 9 guidelines for making the decision. 10 MR. GOLDIN: Your Honor, I would offer 11 Plaintiff's Exhibit 606. 12 MR. GLASS: No objection, your Honor. 13 THE COURT: 606 is admitted. 14 (Plaintiff's Exhibit 606 marked for 15 identification was received in evidence.) 16 BY MR. GOLDIN: 17 Q. Dr. Bailar, I would ask you to take a look at the 18 following document in your binder, which has been marked as 19 Plaintiff's Exhibit 607. 20 A. Yes. 21 Q. Do you have that? 22 A. Yes. 23 Q. Can you tell us what that is? 24 A. This is the Federal Register notice of the 25 Bureau's position on adjustment of the 1980 census counts 1281 1 for underenumeration. 2 Q. Can you tell us briefly what the position is 3 which is described in this announcement? 4 A. That the Bureau will not make an adjustment for 5 the 1980 census, and it gives the underlying reasons for 6 that decision. 7 MR. GOLDIN: Your Honor, I would offer 8 Plaintiff's Exhibit 607. 9 MR. GLASS: No objection, your Honor. 10 THE COURT: 607 is received. 11 (Plaintiff's Exhibit 607 marked for 12 identification was received in evidence.) 13 BY MR. GOLDIN: 14 Q. In connection with the decision process reflected 15 in Exhibit 607, what method did the Bureau considered for 16 adjusting the system? 17 A. The Bureau considered the post-enumeration 18 program and demographic analysis. 19 Q. Did the Bureau decide the demographic analysis 20 could appropriately be used to adjust the 1980 census? 21 A. No. 22 Q. Why not? 23 A. Demographic analysis gives estimates only at the 24 national level, it gives estimates only for the black and 25 nonblack population, so it gives no detail on Hispanic, 1282 1 Asian-American, any other groups, so it didn't seem useful 2 for adjustment. 3 Q. Did the Bureau determine that the PEP would 4 appropriately be used to adjust the census? 5 A. The Bureau decided that it could not be used for 6 adjustment. 7 Q. You described before your reasons for believing 8 that the PEP could be used for that purpose. 9 Could you briefly describe the Bureau's views for 10 arriving at that conclusion? 11 A. Well, the Bureau's reasons and my reasons were 12 the same reasons; they were the amount of error that was in 13 the PEP and the worry that you could be adding error rather 14 than deleting error by using it as a correction tool. 15 Q. Was there any other method available for the 16 Bureau to consider in 1980 apart from the PEP and 17 demographic analysis? 18 A. No. 19 Q. What role did the Commerce Department play in the 20 decision on whether to adjust the 1980 census? 21 A. None. 22 Q. Do you know how that came to be the case? 23 A. Well, the Secretary has delegated responsibility 24 for taking the census, taking surveys to the director, and 25 in this case the he emphasized that delegation and there was 1283 1 a separate document signed by Secretary of Commerce 2 Klutznick delegating that authority to Vincent Barabba. 3 Q. Would you please take a look at the document in 4 your binder which has been marked for identification as 5 Plaintiff's Exhibit 604. 6 A. Yes 7 MR. GOLDIN: To clarify the record, I will note 8 that it is actually two documents, one of which covers the 9 first three pages of this double-sided photocopy and the 10 other one of which covers the last three pages. 11 I am assuming the document appears in the same 12 format in everybody's binder. 13 Q. Have you seen those documents previously, Dr. 14 Bailar? 15 A. Yes. 16 Q. Can you tell us what they are? 17 A. This is the departmental organization series 18 which delegates authority to the director of the Census 19 Bureau and then also describes the functions of the Bureau. 20 Q. Was it your understanding when you served as 21 associate director of the Bureau that authority to make all 22 decisions concerning the conduct of the decennial census had 23 been delegated to the director of the Bureau as described in 24 these departmental organizational orders? 25 A. Yes. 1284 1 MR. GOLDIN: Your Honor, I would offer 2 Plaintiff's Exhibit 604. 3 MR. GLASS: No objection. 4 THE COURT: 604 is admitted. 5 (Plaintiff's Exhibit 604 marked for 6 identification was received in evidence.) 7 BY MR. GOLDIN: 8 Q. Dr. Bailar, would you look at the following 9 document in your binder which has been marked for 10 identification as Plaintiff's Exhibit 605. 11 A. Yes. 12 Q. Is that a document which you have seen 13 previously? 14 A. Yes. 15 Q. How did you come to see this document? 16 A. Well, as one of the senior staff at the Census 17 Bureau, I was aware this document existed and had seen it. 18 Q. What is this document? 19 A. It's a memo from Philip Klutznick to Vincent 20 Barabba through the chief economist, Courtenay Slader, 21 talking about the 1980 census and delegating the authority 22 for making the decision on a statistical adjustment for 23 undercount to the director of the Census Bureau. 24 Q. Philip Klutznick at that time was Secretary of 25 Commerce? 1285 1 A. Yes. 2 Q. Was Secretary Klutznick's instruction that 3 Director Barabba was to make the decision whether to adjust 4 the 1980 census consistent with the overall division of 5 responsibilities between the Department of Commerce and the 6 Census Bureau in matters concerning the decennial census? 7 A. Yes. It was quite consistent with the 8 departmental orders. It just underlined that the Secretary 9 was delegating the authority for anything to do with the 10 adjustment of the undercount 11 MR. GOLDIN: Your Honor, I would offer 12 Plaintiff's Exhibit 605. 13 MR. GLASS: Your Honor, it is part of the 14 administrative record. We have no objection. 15 THE COURT: 605 is admitted. 16 (Plaintiff's Exhibit 605 marked for 17 identification was received in evidence.) 18 BY MR. GOLDIN: 19 Q. What process did the Bureau follow to make its 20 decision whether to adjust the 1980 census? 21 A. The Bureau went through a long involved procedure 22 on this topic. 23 The director wanted it to be as open a process as 24 possible. He had a workshop on the undercount in September 25 of 1989, where he invited in numerous people, both pro and 1286 1 anti adjustment, people from different government places and 2 so forth -- 3 MR. GLASS: Your Honor, I would object to the 4 extent the witness is testifying about what the director 5 wanted. I don't think she knows what the director wanted. 6 I believe there is no foundation for this testimony. 7 THE COURT: Help me. 8 THE WITNESS: Well, I think I did know what the 9 director wanted. 10 THE COURT: How did you know? 11 THE WITNESS: Because the director and I were in 12 consultation about this. 13 THE COURT: I will permit it. 14 A. So we had a workshop in the fall of 89 on this 15 adjustment issue where we explored the adjustment. 16 Then there was a conference held in February of 17 1990 where a lot of the people were invited to talk, give 18 papers on different aspects of adjustment, the pros and cons 19 and how it would affect different things that use the 20 census. 21 And then there was another workshop in the fall 22 of 1980 with the same group of people who had been at the 23 conference, the workshop the year before, and exploring this 24 in even more detail. 25 And then, of course, there was all the discussion 1287 1 with advisory committees and other outside groups as the 2 director went around the country speaking to different 3 groups. 4 Q. Just to clarify the record slightly, I think I 5 heard you say at one point in that answer that something 6 happened in September 1989. 7 Did you mean 1979? 8 A. I'm sorry, 1979. 9 Q. This was the process that culminated in a 10 situation against adjustment? 11 A. That's right. 12 Q. Was that decision challenged in litigation? 13 A. Yes. 14 Q. Can you tell us roughly or precisely, if you 15 know, how many lawsuits there were challenging that 16 decision? 17 A. I believe it was 37. 18 Q. And what happened in the end in those lawsuits? 19 A. The Census Bureau didn't adjust. 20 Q. The Census Bureau prevailed in all of those 21 lawsuits? 22 A. That's right. 23 Q. What role did you play in defending the decision 24 not to adjust in that litigation? 25 A. I helped construct the defense and I was a 1288 1 witness for the defense. 2 THE COURT: In both trials? Actually, there were 3 two trials. 4 THE WITNESS: In both Detroit -- 5 THE COURT: I mean here. Wasn't there one before 6 Werker -- 7 MR. ZIMROTH: One in 1980 and the retrial in 8 1984, but I think the witness testified that there were 9 other -- 10 THE COURT: I know. She said she testified in 11 the one here. I wanted to find out if it was both of the 12 ones her. 13 THE WITNESS: Yes, it was. 14 Q. And you testified elsewhere as well? 15 A. Yes. 16 Q. Where was that? 17 A. Detroit. 18 Q. Now, apart from litigation concerning the issue 19 of the statistical adjustment based on demographic analysis 20 or the PEP, did you submit an affidavit in connection with a 21 lawsuit brought by the State of Indiana concerning the 1980 22 census? 23 A. Yes, I did. 24 Q. And that lawsuit was Orr against Boldrige? 25 A. Yes. 1289 1 Q. What did that lawsuit involve? 2 A. That lawsuit was focused on the use of imputation 3 in the census where imputation was used when the census 4 taker couldn't decide whether a housing unit was occupied or 5 vacant. 6 If it had been left alone, all of those housing 7 units would have been really inputed as having no people in 8 them, and so the Census Bureau, based on a statistical 9 model, went forward with an imputation procedure, first of 10 all, deciding whether the unit was occupied or vacant, and 11 if the it was occupied then actually imputing people into 12 that household. 13 Q. Can you tell us, how does the Bureau perform this 14 technique of imputation? 15 A. It relies on the household that is closest in the 16 neighborhood and gives the household whose status isn't 17 known the same tenure status, either occupied or vacant, and 18 then if it is occupied it reimputes -- it imputes those 19 same people into that household as well. 20 Q. What was the State of Indiana seeking in that 21 lawsuit? 22 A. Indiana had lost a Congressional seat because of 23 the imputation and had lost the seat in Florida, and so they 24 were saying that the use of statistical methods in the 25 census shouldn't happen and that only the count without 1290 1 imputation should be used. 2 Q. What was the outcome of that lawsuit? 3 A. The Census Bureau prevailed. 4 Q. Had you agreed with the Census Bureau's decision 5 to use imputation in the 1980 census? 6 A. Yes, because if you didn't do that, it would be, 7 again, it would be imputting and acting if if those people 8 didn't exist and yet there was evidence that those people 9 did exist. 10 Q. And apart from that adjustment of the 1980 11 census, did you agree with the Bureau's decision not to 12 adjust the 1980 census on the basis of the PEP or 13 demographic analysis? 14 A. Yes, I did. 15 Q. When did you first hear concern expressed at the 16 Bureau about a differential undercount in the 1990 census? 17 A. During the middle of the 1980 census. 18 Q. And what form did you hear that concern 19 expressed? 20 A. Well, it was obvious during the taking of the 21 1980 census that there was going to be an undercount in 1980 22 no matter how the Bureau tried to put together a big 23 publicity program, it couldn't seem to take care of that 24 undercount. It was obvious that something new had to be 25 developed, the same old procedures were not going to take 1291 1 care of it, so worrying about the undercount always occurs 2 throughout the whole census period. 3 Q. What steps did the Bureau take to address that 4 concern about the 1990 census? 5 A. Even while the 1980 census was going on, some 6 research programs were put in place, the forward trace 7 studies, for example, to get a way of trying to measure 8 undercount that might be more effective. 9 So beginning right with 1980, the thought was 10 there that as 1990 was approached, the Bureau needed to be 11 ready to do something about the undercount. 12 Q. When did active work at the Bureau begin on the 13 question of the feasibility of adjusting the 1990 census? 14 A. Well, an undercount research staff was put in 15 place in 1984, but, as I said a minute ago, actually work 16 was started during the 1980 census itself. 17 Q. Why did work begin so early in the decade in 18 connection with a census that was not to occur until 1990? 19 A. Everything takes a long time with the census and 20 when you are going to try to work out a research program and 21 then go from the research to implementation and testing and 22 getting budgets together, it takes a long period. 23 Q. Did the Bureau establish an undercount steering 24 committee? 25 A. Yes. 1292 1 Q. What was that committee? 2 A. That committee was composed of two associate 3 directors and some division chiefs, all of whom had some 4 responsibility for parts of the census, to guide and 5 coordinate the work on undercount research. 6 Q. When was the undercount steering committee 7 established? 8 A. I think it was 1985. 9 Q. What was the function of the undercount steering 10 committee? 11 A. The function was to review the work from 1980 to 12 decide what work had to be finished in connection with the 13 1980 census and then to coordinate the work leading towards 14 1990. 15 Q. Who at the Bureau was responsible for conducting 16 the actual work on undercount research and techniques for 17 adjustment in 1990? 18 A. Well, there were a couple of groups. 19 One of them was the undercount research group in 20 the statistical research division headed by Howard Hogan, 21 and then there was work also on undercount done by another 22 part, another group of people in the statistical research 23 division, Cary Isaki and his people, and then there was some 24 work done with demographic analysis in the population 25 division. 1293 1 Q. To whom did Dr. Hogan report? 2 A. Kirk Wolter. 3 Q. And to whom did Dr. Wolter report? 4 A. To me. 5 Q. What was the function of the undercount research 6 staff? 7 A. They were supposed to review different methods of 8 estimating an undercount, come up with a recommendation for 9 one to be used for 1990 and they were supposed to then, once 10 there was a recommendation, to test it in the test censuses, 11 evaluate it and then make a recommendation about how we 12 should approach this in the 1990 census. 13 Q. Did the Bureau arrange for anyone outside the 14 Bureau to conduct or review research on undercount and 15 adjustment? 16 A. Yes. The Bureau was very much interested in 17 having a very open process so it signed joint statistical 18 agreements with several universities getting people to work 19 on different aspects of the undercount. It also contracted 20 with the National Academy of Sciences for a special panel to 21 review the undercount problem. 22 Q. What is a joint statistical agreement? 23 A. It's an agreement between primarily universities 24 and the Census Bureau to do joint research on the topic of 25 interest to each party. 1294 1 Q. Why did the Bureau enter into those agreements 2 with outside academicians? 3 A. Well, there are lots of reasons. 4 One is it's a good idea to get a different 5 perspective on these problems by having groups outside of 6 the Census Bureau work on them, and a lot of these people 7 are very noted researchers and if they are interested, it 8 seems a good idea to take advantage of using their skills. 9 And it also makes sense to try to explain to the technical 10 community what it is the kinds of work that you are doing 11 and to have them be responsible for a part of it as well. 12 Q. Apart from the joint statistical agreements with 13 university statistics departments, did anyone else outside 14 the Bureau conduct research or review the Bureau's research 15 on the issues of undercount and adjustment? 16 A. Well, the National Academy panel did and advisory 17 committees to the Census Bureau. 18 Q. How was the National Academy panel established? 19 A. A contract was signed between the Census Bureau 20 and the National Academy to create a special panel to review 21 undercount and look towards the 1990 census. 22 Q. What was the mechanism for creating that panel? 23 A. There was a contract. 24 Q. Who was it that was responsible for setting up 25 the panel? Under whose auspices was that done? 1295 1 A. That was the committee of the national statistics 2 at the National Academy that set that up. 3 Q. And who was the head of the committee on national 4 statistics at that time? 5 A. Steven Feinberg. 6 Q. When was the National Academy panel established? 7 A. Sometime around the mid-80s. 8 Q. How did the panel fulfill that contract? 9 A. They were doing a very thorough review of the 10 census. They put out one big volume called the bicentennial 11 census, but they did other reviews of the work that the 12 Census Bureau was doing on the adjustment question. Quite 13 often, at least twice a year, sometimes three times a year 14 they met to consider those problems. 15 Q. I believe in your responses a moment ago you 16 mentioned review buy outside advisory committees. 17 A. Yes. 18 Q. What were they? 19 A. The American Statistical Association has an 20 advisory committee to the Census Bureau, and they reviewed 21 this, and the population advisory committee also reviewed 22 it. 23 Q. How were those committees kept apprised of the 24 Bureau's work? 25 A. Those committees meet twice a year and at each 1296 1 meeting they were given a presentation and background 2 materials on the work that had been accomplished so far and 3 then asked for their comments, and sometimes they made 4 recommendations as well. 5 Q. Did the Bureau keep Congress apprised of its 6 research and plans for 1990? 7 A. Yes. The Bureau had frequent testimony before 8 Congress. 9 Q. Did the undercount research staff come to focus 10 on a particular technique or approach as the one that showed 11 promise for use in adjusting the 1990 census? 12 A. Yes. They selected the post-enumeration survey. 13 Q. Did the undercount research staff do work on 14 redesigning the PES for 1990 as compared to what the 1980 15 PEP had been? 16 A. Yes, they did a lot of work that way and took it 17 from the realm of being just piggybacked on another survey 18 to being a separate survey that was to be used to estimate 19 omissions and erroneous enumerations. 20 Q. Apart from having it be a separate survey, what 21 other changes did the undercount research staff make in the 22 designing of the PES? 23 A. Well, in the design of the survey was one. 24 In the use of the matching, because it is such a 25 tedious process, one of the first thing that it decided to 1297 1 do was to develop an automated matcher so that the cases 2 where there was no question about the two samples, the 3 census and the P-sample, being a match could have been 4 gotten out of the way without people spending a lot of time. 5 That program worked extremely well, it was tested 6 in one of the earlier test censuses. 7 Q. When was the PES tested? 8 A. It was tested in Tampa in 1985, in East Los 9 Angeles in 1986, in Meridien, Mississippi in 1986. 10 Q. Was one of those an overall test of adjustment 11 related operations? 12 A. Yes. That was TARO, the adjustment related test 13 of adjusted related tests in East Los Angeles. 14 Q. What was the objective of those tests? 15 A. The objectives were to try to see if the PES, the 16 flaws that had been identified in 1980 had been taken care 17 of, if it was technically feasible. 18 In Tampa it was really concentrated on seeing 19 whether the automated matching program worked; in East Los 20 Angeles it was to see if the whole procedure worked and 21 adjusted data could be produced and what it looked like and 22 do an evaluation of it; and in Meridien, Mississippi it was 23 to see if the matching could work in a rural area as well. 24 Q. What were the result of those tests? 25 A. The results were that the PES was working very 1298 1 well. 2 Q. Did the Bureau establish any deadlines or 3 decision points in connection with deciding whether to 4 correct the 1990 census? 5 A. Yes. 6 Q. What were those decision points? 7 A. The Bureau decided that it would make -- there 8 would be two decision points: 9 One would be in the spring of 1987 and it would 10 be a decision on technical and operational feasibility and 11 whether the Bureau could do a PES, have it evaluated and be 12 satisfied that it was a -- under good technical control. 13 Then the second decision point would be in the 14 fall of 1990. In the meantime, a series of technical 15 standards would be developed and shared with the public, and 16 on the basis of those standards the PES would be reviewed in 17 the fall of 1990 and a decision would be made that if 18 nothing major had gone wrong, then the adjustment could go 19 forward. 20 Q. Would you please take a look at a document in 21 your binder that has been marked for identification as 22 Plaintiff's Exhibit 596. 23 A. Yes. 24 Q. Do you have that? 25 A. Yes. 1299 1 Q. Is that a copy, albeit a not very good one, of 2 the testimony that you delivered on July 24, 1986 before the 3 United States House of Representatives Subcommittee on 4 Census and Population? 5 A. Yes. 6 MR. GOLDIN: Your Honor, I would offer 7 Plaintiff's Exhibit 596. 8 MR. GLASS: Your Honor, we have no objection, but 9 I would note that the copy that we have seems to be missing 10 some material here. 11 MR. GOLDIN: The portion to which Mr. Glass is 12 referring involves written responses to supplemental 13 questions postdating the testimony that was actually 14 delivered and it's not an area of the exhibit about which I 15 intend to question Dr. Bailar in this examination. 16 That said, we will attempt to locate and supply 17 the missing pages, which I should say are missing from my 18 copy and I presume from the court's and the witness' as 19 well. 20 THE COURT: With that understanding, I will admit 21 596. 22 MR. GOLDIN: Thank you, your Honor. 23 (Plaintiff's Exhibit 596 marked for 24 identification was received in evidence.) 25 BY MR. GOLDIN: 1300 1 Q. Dr. Bailar, in this testimony, did you describe 2 to Congress the two-stage decision process that you just 3 mentioned? 4 A. Yes. 5 Q. And was that process approved by the Department 6 of Commerce? 7 A. Yes. 8 Q. How do you know that? 9 A. All testimony given by Bureau employees had to be 10 reviewed by the department and by the Office of Management 11 and Budget. It was reviewed and no changes were made in it. 12 Q. This testimony in particular was reviewed? 13 A. Yes, it was. 14 Q. And it was permitted to be delivered in unchanged 15 form? 16 A. That's right. 17 Q. Let me call your attention now to page 17 of that 18 testimony under the heading, "1987 decision." 19 Do you see there the paragraph that begins, "In 20 early 1987, we will decide on the statistical and 21 operational feasibility of adjustment"? 22 A. Yes. 23 Q. Would you please read into the record that 24 paragraph of four sentences? 25 A. "In early 1987, we will decide on the statistical 1301 1 and operational feasibility of adjustment. This is not a 2 decision about whether the adjusted numbers will be the 3 official 1990 census counts. What appears feasible in 1987 4 may or may not be feasible in 1990. It will depend on the 5 research and testing yet to be done and actual experience in 6 the census." 7 Q. So in the spring of 1987, there was to be a 8 decision on the statistical and operational feasibility of 9 adjustment. 10 What did you mean by the statistical feasibility 11 of adjustment? 12 A. Whether you could carry out a PES with low levels 13 of error, whether the matching error looked relatively 14 small, whether there were small amounts of missing data, 15 whether it was under good statistical control. 16 Q. What did you mean by the operational feasibility 17 of adjustment? 18 A. Whether the PES could be fitted into the on-going 19 Census Bureau operations and be finished by December 31, 20 1990. 21 Q. Would you go now to the top of page 18 in your 22 testimony where there is a paragraph beginning, "If the 23 decision is that we are statistically and operationally 24 capable of adjustment." 25 A. Yes. 1302 1 Q. Would you read that paragraph, please? 2 A. "If the decision is that we are statistically and 3 operationally capable of adjustment, then we will plan work 4 to permit adjustment, unless the final results do not meet 5 technical standards that we are currently developing. 6 Although we cannot say definitively at this time what our 7 determination will be, our research thus far on undercount 8 measurement has been promising." 9 Q. Was the Bureau going to make a commitment of 10 resources if the spring 1987 decision was positive? 11 A. Yes, the Bureau would have to commit resources 12 and go forward asking for budget. 13 Q. What kind of resources aside from financial would 14 the Bureau be committing? 15 A. Space in census offices, computer resources, 16 people. 17 Q. Now let me call your attention to the following 18 page, page 19 under the heading, "Standards for adjustment," 19 and there is a paragraph there which I would ask you to read 20 starting at the bottom of page, "The standards upon which 21 the judgment decision is to be based." 22 MR. GLASS: Your Honor, I would object. The 23 document really does speak for itself. We have admitted it 24 into evidence. I don't see the purpose to be served by 25 having the witness read selected excerpts of this. 1303 1 THE COURT: The purpose is to save me the trouble 2 of having to read it. On that basis alone, I will overrule 3 the objection. 4 A. That last paragraph? 5 Q. Dr. Bailar, if you will forgive the strain on 6 your eyes, could you make out the last paragraph on the 7 bottom of page 19? 8 A. "The standard upon which the adjustment decision 9 is to be based must depend upon demonstrable observable 10 results, excuse me, of the census." 11 I don't have my glasses on, either, so this is a 12 real test. 13 "-- and of the coverage measurement studies. 14 These would be such things as the measured differential 15 census -- " I think I lost the page here. 16 Q. Do you have it at the top of page 20? 17 A. I am going from the bottom of page -- I lost 18 where I was. 19 THE COURT: You were at the bottom of page 19 and 20 about to go to the top of page 20. 21 A. "These would be such things as the measured 22 differential census uncounted by race, ethnicity or 23 geographic level, indicators of the quality of the data 24 collected in the coverage measurement survey and indicators 25 of the quality of the matching operation. The determination 1304 1 of how to weigh these observable results would be based upon 2 conceptual measures of census data accuracy." 3 Q. Just to be clear, in that paragraph, what you are 4 calling the coverage measurement study or coverage 5 measurement survey is the PES, is that correct? 6 A. Yes, it is. 7 Q. Were the standards to be developed in advance of 8 the actual December 1990 decision? 9 A. Yes. 10 Q. Why was it planned that they be developed in 11 advance? 12 A. Because the Census Bureau wanted to get them out 13 there to the public and have public comment on those 14 standards, wanted to make sure that they were clearly 15 understood. 16 Q. Now I want to turn to one last excerpt. From 17 this testimony on page 24 there is a paragraph in the middle 18 of the page. I can simply read this myself. 19 THE COURT: Do it. 20 MR. GOLDIN: I will. 21 Q. "Second, in December 1990 or as soon as the 22 coverage measurements have been obtained, we will evaluate 23 the measurements in light of the specified technical 24 standards. This evaluation will be looking for any 25 substantial unforeseen errors that might have occurred or 1305 1 other unanticipated serious flaws in the coverage 2 measurement process that would cause us to view the adjusted 3 data as further from the truth than the unadjusted data. 4 Assuming the standards are met and such flaws have not 5 occurred, we would release the adjusted data, but if our 6 review indicates that serious errors occurred, then we would 7 release the unadjusted data." 8 Dr. Bailar, was all of the language that you have 9 read aloud and that I have just read aloud from this 10 testimony submitted for approval by the Department of 11 Commerce? 12 A. To the department, yes. 13 Q. Was this the only occasion on which the Bureau 14 presented testimony about the two decision point framework 15 before Congress? 16 A. No. Director Keane also presented testimony. 17 Q. Did the Bureau advise the National Academy panel 18 of the two decision point framework? 19 A. Yes. 20 Q. Would you please turn in your binder to the 21 document that has been marked for identification as 22 Plaintiff's Exhibit 611. 23 Is this a document that you have seen before? 24 A. Yes, it is. 25 Q. What is this document? 1306 1 A. This is a letter from the Director Jack Keane to 2 John Pratt, who was the chair on that panel on decennial 3 census methodology. 4 Q. And in this letter, fourth paragraph, Dr. Keane 5 describes the May decision as being the turning point. 6 Is he referring to the decision on whether to 7 proceed with the plans for adjustment? 8 A. Yes. 9 MR. GOLDIN: Your Honor, I would offer 10 Plaintiff's Exhibit 611. 11 MR. GLASS: No objection, your Honor. 12 THE COURT: 611 is admitted. 13 (Plaintiff's Exhibit 611 marked for 14 identification was received in evidence.) 15 BY MR. GOLDIN: 16 Q. Dr. Bailar, in preparing for its decision in the 17 spring of 1987, did the Bureau review results of the 1986 18 test of adjustment related operations? 19 A. Yes. 20 Q. Did the Bureau also review results of the other 21 1985 and 1986 tests of components of the PES? 22 A. Yes. 23 Q. What conclusions did the Bureau draw from its 24 review of those tests? 25 A. The Bureau decided that the PES had been improved 1307 1 considerably over the 1980 experience, that the level of 2 missing data had been reduced, that the matching error was 3 considerably reduced and that the level of fabrication was, 4 again, very small and decided that the Bureau knew how to 5 control this kind of a survey. 6 Q. Were those results reviewed outside the Bureau? 7 A. Yes, they were. 8 Q. By whom? 9 A. They were reviewed with the National Academy 10 panel, they were also reviewed by the two advisory 11 committees that I mentioned earlier. 12 Q. What conclusions did the National Academy come 13 to? 14 A. The National Academy was very positive about the 15 research and encouraged the Bureau to go forward. 16 Q. Would you please take a look at what has been 17 marked for identification as Plaintiff's Exhibit 598. 18 Can you tell me what that document is? 19 A. These -- this is a letter to John Keane from 20 Benjamin King, who was the chairman of the panel at that 21 time, on two recommendations directed to the May 87 decision 22 on the feasibility of adjustment. 23 Q. And are those the recommendations that you were 24 just mentioning in your testimony? 25 A. Yes. 1308 1 Q. You testified a moment ago that the American 2 Statistical Association advisory committee also made a 3 recommendation? 4 A. Yes. 5 Q. Would you look at what has been marked for 6 identification as Plaintiff's Exhibit 599? 7 A. Yes. 8 Q. Is that the recommendation to which you were 9 referring? 10 A. Yes, on the feasibility of adjustment. In fact, 11 there is more than one, the post-enumeration survey as well. 12 Q. And would you look at the following document 13 which has been marked for identification as Plaintiff's 14 Exhibit 600. 15 A. Yes. 16 Q. Is that -- 17 A. Those are the recommendations from the committee, 18 the population advisory committee. 19 MR. GOLDIN: Your Honor, I would offer 20 Plaintiff's Exhibits 598, 599 and 600. 21 MR. GLASS: No objection, your Honor. 22 THE COURT: All three are admitted. 23 (Plaintiff's Exhibits 598, 599 and 600, 24 respectively, marked for identification were received in 25 evidence.) 1309 1 BY MR. GOLDIN: 2 Q. Dr. Bailar, in the spring of 1987, did the Bureau 3 make the decision when it arrived at the first decision 4 point? 5 A. Yes. 6 Q. Who participated in making that decision? 7 A. The undercount steering committee, the deputy 8 director and the director of the Census Bureau. 9 Q. When did their deliberations occur? 10 A. In May of 1987. 11 Q. Was that on the schedule contemplated in your 12 1986 testimony? 13 A. Yes. We had talked about spring of 87. 14 Q. What was the decision that was made? 15 A. That the Bureau should go forward with putting 16 the machinery in place to have -- to be able to do an 17 adjustment. 18 Q. Would you please turn to the document that has 19 been marked for identification as Plaintiff's Exhibit 597. 20 A. Yes. 21 Q. Is that a document which you have seen 22 previously? 23 A. Yes, it is. 24 Q. What is it? 25 A. It's a draft document that Sherry Courtland at 1310 1 the Census Bureau and I put together describing what the 2 decision on adjustment was, and we were getting this ready 3 for the director to use for meetings downtown and then to be 4 able to use for press release later on. 5 Q. When you referred to meetings downtown, just to 6 locate the context in terms of district geography, what are 7 you referring to? 8 A. I am referring to meetings at the Department of 9 Commerce. 10 Q. Where is the Bureau itself located? 11 A. Out in Maryland. 12 Q. Proceeding on in this document from the first 13 portion to which you have referred, eight pages in there is 14 a chunk of text which begins with the stamped date May 22, 15 1987, and then the phrase, "On May 20, senior staff." 16 Do you see that? 17 A. Yes. 18 Q. Would you tell us what that portion of the 19 document is? 20 A. This was a document that was put together right 21 after the steering committee met when it was going to go 22 through its -- try to decide on its recommendation about 23 technical feasibility and these were -- this was a summary 24 of the questions we asked, what the major recommendations 25 were and the general conclusions. 1311 1 Q. And then another three pages into the document 2 there is another section also stamped May 22, 1987 which 3 begins, "Discussion outline adjustment decision." 4 Do you see that? 5 A. Yes. 6 Q. What is that portion of this document? 7 A. That was, that was really -- this was put 8 together really to help the director focus on all of the 9 options with the decision, and so it gave the different 10 options and listed the pros and cons of each option. 11 Q. Was this a document that you prepared or assisted 12 in the preparation of? 13 A. Yes. 14 Q. And all of this material was put together for the 15 director of the Census Bureau? 16 A. That's right. 17 Q. Who was the director of the Census Bureau in the 18 spring of 1987? 19 A. John Keane. 20 MR. GOLDIN: Your Honor, I would offer 21 Plaintiff's Exhibit 597. 22 MR. GLASS: No objection, your Honor. 23 THE COURT: 597 is admitted. 24 (Plaintiff's Exhibit 597 marked for 25 identification was received in evidence.) 1312 1 BY MR. GOLDIN: 2 Q. What was the basis for the Bureau's decision in 3 the spring of 1987? 4 A. The basis was really the success of the census 5 tests, the test, first of all, in Tampa, which showed that 6 the matching software program was working very well, the 7 test in Meridien, Mississippi which showed you can do a PES 8 in a rural area, and especially the test of the adjustment 9 related operations in Los Angeles which showed that the 10 Bureau knew how to keep control of the post-enumeration 11 survey and keep the levels of error quite low. 12 Q. Was there consensus at the Bureau that the 13 reduction or elimination of differential undercount in the 14 1990 census was a primary goal in planning for that census? 15 A. Yes. 16 Q. Did you agree with the Bureau's 1987 decision on 17 adjustment? 18 A. Yes. 19 Q. You testified earlier that you had agreed with 20 the Bureau's decision against adjusting the 1980 census. 21 Was your position in 1987, in your view, 22 inconsistent with the position that you had taken earlier? 23 A. No. In my view it was very much consistent. 24 My problem in 1980 had been with the quality of 25 the data that came from the PEP program and that because of 1313 1 the poor quality I believe that that data could not be used 2 for an adjustment. 3 But after we set a research program in place, we 4 tested it, we knew what we were doing, I was convinced then 5 that the errors had been taken care of and so to me it was 6 always a technical question, and when those technical 7 problems were taken care of, I had no problem with going 8 forward with an adjustment. 9 Q. Did you believe in 1987 that adjustment would 10 necessarily improve the accuracy of the census for every 11 block? 12 A. No. 13 Q. Was that a factor in your decision in 1987? 14 A. No. 15 Q. Was that consistent with the view you had taken 16 in 1980? 17 A. No. 18 Q. What changed? 19 A. In 1980, when we were looking at the PEP, I was 20 really focusing on the amount of error that was in the PEP 21 and listening to the plaintiffs talk about how they wanted 22 to adjust the census with a synthetic adjustment and I was 23 worried about what that meant, but I was so -- I was really 24 caught up with the amount of error that was in the PEP and I 25 wasn't really concentrating on what would happen with the 1314 1 blocks and so forth. 2 After that, several people told me that I should 3 re-examine that point of view about the block level data 4 having to be corrected, that every block should be better 5 after an adjustment -- 6 Q. Was that a position you had taken publicly, that 7 every block had to be improved? 8 A. It was very close to that. It was in an 9 affidavit that was used in the 1984 case. 10 I started looking at the uses of census data, I 11 started talking to people about who used the block census 12 and found out that it really isn't used except when it is 13 aggregated to higher levels, and we knew that the PES could 14 make an improvement at higher levels of aggregation. 15 So it seemed to me then that my earlier position 16 had been incorrect and I revised it. 17 Q. Did you take into consideration in that process 18 of revision what you do about the level of error in the 19 unadjusted census at the block level? 20 A. Yes. I had done some work there to look at what 21 happens to census statistics at the block level, and it's 22 very interesting that within the Census Bureau among people 23 who have some familiarity with the actual census, it's quite 24 well known that the block level data are just full of error, 25 that if you ask another person to go out in the next day and 1315 1 recount a block that you get very different results. 2 So the census blocks are not well counted to 3 begin with, there are differences of interpretation, and to 4 have to think about making a PES that is going to improve 5 greatly on what the census itself can't do very well at the 6 block level didn't make a lot of sense considering the uses 7 of that block level data. 8 Q. Did you believe in 1987, at the time of the 9 Bureau's decision, that adjustment would improve the 10 distributive accuracy of the census counts? 11 A. Yes. 12 Q. Was that a factor in your decision? 13 A. Yes, it was, but in a different kind of a way. 14 I'd worked at the Census Bureau a long time and I 15 worked in a lot of different censuses, and when you talk to 16 people who worked in those censuses, they have to focus on 17 getting the most accurate count at each level of geography 18 and really push to get that job done and done well, they 19 don't think about if I don't do as well here, maybe I'll 20 improve the distributive accuracy. 21 That would be the wrong thing for the Census 22 Bureau to count on. The Census Bureau really has to do the 23 best job it can everywhere it's working. So the philosophy 24 is that where you are getting the best possible counts, the 25 numeric accuracy, you are also increasing the distributive 1316 1 accuracy. 2 Q. So numeric accuracy is what you shoot for and 3 that's the way that you get to distributive accuracy? 4 A. Yes. 5 Q. You mentioned a consensus among the senior staff 6 on this first decision in 1987. 7 Was there any dissent from that consensus? 8 A. Yes. 9 Q. Who were the dissenters? 10 A. Charles Jones. 11 Q. What was the nature of his dissent? 12 A. He was very much concerned with operational 13 feasibility and the timing question, but he also had some 14 questions about the technical feasibility as well. 15 Q. Did anyone else dissent on the subject of 16 technical feasibility? 17 A. No. 18 Q. Did anyone else have concerns about timing? 19 A. Yes. 20 Q. What were those concerns? 21 A. That there was a tight time schedule, that the 22 Census Bureau starts off with the census in April and it's 23 always running right up against the deadline at the end of 24 December, that no matter how much time you have, that's the 25 amount of time it is going to take to do the census, so that 1317 1 in doing a PES you have to build that into the census as 2 well, and so there were concerns about whether they could do 3 both operations and get them both done by the end of the 4 year. 5 Q. So that was a concern about the production of 6 adjusted data by December 31, 1990? 7 A. Yes. 8 Q. What was the director's decision in the spring of 9 1987? 10 A. The director decided that the Census Bureau 11 should go forward with the adjustment. 12 Q. Was that decision announced publicly? 13 A. No. 14 Q. Why not? 15 A. The Census Bureau and the director -- well, the 16 director decided that he should inform the Under Secretary 17 and Secretary of Commerce what the decision was before he 18 announced it publicly, and when he did so, it was never 19 announced. 20 THE COURT: Who was the director at the time? 21 THE WITNESS: John Keane. 22 Q. Who was the Under Secretary with whom he 23 consulted? 24 A. Robert Ortner. 25 Q. Let me call your attention to a document which 1318 1 has been marked for identification and should appear in your 2 book as Plaintiff's Exhibit 232. 3 Do you see that document? 4 A. Yes. 5 Q. What is that document? 6 A. This was a briefing document for the director, 7 Director Keane, for a meeting with Robert Ortner to discuss 8 this decision on the undercount. 9 Q. Did you participate in the drafting of this 10 document? 11 A. Yes, I did. 12 MR. GOLDIN: Your Honor, I would offer 13 Plaintiff's Exhibit 232. 14 MR. GLASS: Your Honor, I suppose at this point I 15 should reassert the continuing objection that my colleague 16 Mr. Millet asserted I guess during the Wolter testimony, 17 which is that this all goes to the 1987 Commerce decision 18 and, of course, that is irrelevant in this action, because 19 we are dealing with the 1991 decision. The 1987 decision 20 was vitiated by the stipulation and order in this case. 21 THE COURT: I find it helpful background if 22 nothing else. I will overrule the objection and admit it. 23 232 is admitted. 24 (Plaintiff's Exhibit 232 marked for 25 identification was received in evidence.) 1319 1 MR. GLASS: Your Honor, the only other thing I 2 would note about this document is that there seem to be some 3 handwritten comments in it on the third page. 4 MR. GOLDIN: Let me ask Dr. Bailar if she knows 5 whose handwritten comments those are. 6 THE WITNESS: Yes, I do. Those are Director 7 Keane's 8 BY MR. GOLDIN: 9 Q. Dr. Bailar, do you know when Director Keane made 10 his decision on adjustment? 11 A. When the director made his decision? 12 Q. Yes. 13 A. It was in May of 1987. 14 Q. Was that after the meeting of the undercount 15 steering committee on May 20? 16 A. Yes, it was. 17 Q. And the director thereafter met with Under 18 Secretary Ortner on June 2? 19 A. Yes. 20 Q. And at that meeting, did the director discuss his 21 decision with Under Secretary Ortner? 22 A. Yes, he did. 23 Q. Were you at that meeting? 24 A. Yes. 25 Q. You were at the June 2 meeting? 1320 1 A. I believe I was. 2 Q. Does this briefing or report, Plaintiff's Exhibit 3 232 set forth the basis for the director's decision? 4 A. Yes. 5 Q. And what was the basis for his decision? 6 A. That given the evidence from the past record of 7 the undercounting in the census, he was convinced that there 8 was going to be a differential undercount in 1990. He 9 wanted to go forward and try to correct that by improving 10 census procedures if he could, but since that hadn't worked 11 in the past, he wanted to have the adjustment ready to take 12 care of the problem, and based on the results of the test 13 censuses, he thought that the Bureau could do that. 14 Q. And did he tell that to Under Secretary Ortner at 15 the June 2 meeting? 16 A. Yes, yes, he did. 17 Q. What was Dr. Ortner's reaction at that meeting? 18 A. Dr. Ortner asked a few questions about this, but 19 voiced no strong objection. 20 Q. Was there a subsequent meeting on this subject 21 involving Director Keane and Dr. Ortner? 22 A. Yes, there was another meeting on June 8. 23 Q. And who else was at that meeting? 24 A. The Deputy secretary and several of the 25 departmental officials. 1321 1 Q. Who was the deputy Secretary at that time? 2 A. Clarence Brown. 3 Q. What happened at that meeting? 4 A. At that meeting, the group that was assembled did 5 not share Dr. Keane's enthusiasm for moving forward with the 6 undercount adjustment and compared this to trying to adjust 7 for the people who didn't vote in an election, and on that 8 basis decided that there would be no adjustment in the 1990 9 census. 10 Q. Did Dr. Keane agree with that decision? 11 A. No. 12 Q. Were the other participants in that meeting from 13 the Department of Commerce? 14 A. I'm sorry? 15 Q. The participants in that meeting other than Dr. 16 Keane were from the Department of Commerce? 17 A. Yes. 18 Q. Was that decision made public at the time of that 19 meeting on I believe you said June 8? 20 A. No. 21 Q. Was it made public at some later time? 22 A. Yes. 23 Q. When was that? 24 A. October 1987. 25 Q. Dr. Bailar, before the announcement of the 1322 1 Department of Commerce decision, was there further work done 2 at the Bureau on the development of the standards for the 3 adjustment decision that you had described in your 1986 4 testimony? 5 A. Yes. 6 Q. First of all, let's look at where those standards 7 stood when the director made his decision. 8 Would you please take a look at what has been 9 marked for identification as Plaintiff's Exhibit 683. 10 A. Yes. 11 Q. Is that a document you have seen before? 12 A. Yes. 13 Q. What is that? 14 A. These are the prototype standards for decennial 15 census adjustment by Howard Hogan and Mary Mulry who were on 16 the undercount research staff. 17 MR. GOLDIN: Your Honor, I would move the 18 admission of Plaintiff's Exhibit 683. 19 MR. GLASS: No objection, your Honor. 20 THE COURT: 683 is admitted. 21 (Plaintiff's Exhibit 683 marked for 22 identification was received in evidence.) 23 BY MR. GOLDIN: 24 Q. Were those standards presented to the outside 25 advisory committees? 1323 1 A. Yes. 2 Q. Now would you take a look at the preceding 3 document in your binder, which has been marked for 4 identification as Plaintiff's Exhibit 682. 5 Is that a document that you have seen previously? 6 A. Yes, it is. 7 Q. What is that document? 8 A. Preliminary census adjustment standards, a draft, 9 by Howard Hogan and Mary Mulry, dated October 1987. 10 Q. And how did you come to see that document? 11 A. These people worked in the statistical research 12 division and I saw this document as part of my job with 13 reviewing the work. 14 MR. GOLDIN: Your Honor, I would offer 15 Plaintiff's Exhibit 682. 16 MR. GLASS: No objection, your Honor. 17 THE COURT: 682 is admitted. 18 (Plaintiff's Exhibit 682 marked for 19 identification was received in evidence.) 20 BY MR. GOLDIN: 21 Q. Did the October 1987 standards continue to refine 22 the work presented in the April version? 23 A. Yes. 24 Q. Did the October standards specify particular 25 approaches to measuring identified sources of error in the 1324 1 PES? 2 A. Yes. The October document went so far as to 3 looking at the sources of error that can occur in a PES and 4 then setting out particular kinds of studies to examine 5 whether that kind of error actually did occur, and these are 6 the -- these are the P studies that were carried out as 7 part of the evaluation of the 1990 PES. 8 Q. When you say that these are the P studies, could 9 you explain more specifically what you mean? 10 A. Well, later on, when the Bureau decided that 11 they would have to have some kind of technical standards, 12 P studies were used to evaluate the quality of the PES. 13 They were actually mentioned, those particular P studies 14 were mentioned as far back as October 1987. 15 Q. And when you say those particular P studies, you 16 are referring to the identification of P1, for example, as 17 the analysis of reasonable alternatives? 18 A. For missing data, that's right. 19 Q. And that carries through with the other P studies 20 that are referred to in Plaintiff's Exhibit 682? 21 A. That's right. 22 Q. Those were, in fact, the P studies that were 23 ultimately conducted by the Bureau? 24 A. That's right. 25 Q. You testified earlier that the Commerce 1325 1 Department's decision not to correct the 1990 census had not 2 been made public in the spring of 1987. 3 When was it finally revealed? 4 A. October 1987. 5 Q. And in what form was it finally revealed? 6 A. It was a press release by the Under Secretary 7 Ortner. 8 Q. Would you take a look at what has been marked for 9 identification as Plaintiff's Exhibit 475. 10 A. All right. 11 Q. Is that the press release? 12 A. Yes, it is. 13 MR. GOLDIN: Your Honor, I would move the 14 admission of Plaintiff's Exhibit 475. 15 MR. GLASS: No objection, your Honor. 16 THE COURT: 475 is admitted. 17 (Plaintiff's Exhibit 475 marked for 18 identification was received in evidence.) 19 BY MR. GOLDIN: 20 Q. Dr. Bailar, have you previously had occasion to 21 read Dr. Ortner's statement, Plaintiff's Exhibit 475? 22 A. Several times. 23 Q. Do you agree with the arguments contained in that 24 statement? 25 A. No. 1326 1 Q. What is the basis for your disagreement? 2 A. Well, the very beginning of the statement ignores 3 completely the differential undercount that has existed in 4 the census. 5 Dr. Ortner talks about how well the census is 6 going to be conducted in 1990 and that there will be no need 7 for an adjustment. He goes on to say that adjusting the 8 numbers is unlikely to improve on the 99 percent or better 9 count of the population which we expect to achieve without 10 saying that even if it had been a 99 percent success, that 11 could still have had a large differential undercount. 12 He goes on to say that adjustment would be 13 controversial even amongst statisticians and that there is 14 no unique system generally accepted by the professional 15 statistical community. 16 That's true about almost every decision that is 17 made in the census. The idea that you would get perfect 18 consensus on whether to mail out mailback census, whether 19 you would have agreement even that the unit of enumeration 20 should be the housing unit when we have problems with the 21 homeless, there is nothing in the census that isn't 22 controversial, and yet peopel go forward, they have to make 23 decisions and do their work. 24 He goes on to say that adjustment would be based 25 on a survey and that the survey would sample a relatively 1327 1 small number of households. 2 At that time the survey was planned to be 300,000 3 households. That's the largest sample survey that had ever 4 been proposed at that point. It's a large survey. 5 And he goes on to say it would be conducted 6 during the summer months when many people are on vacation or 7 changing residents. 8 There are many people who are missed in the 9 census who don't go on vacation. 10 And then he goes on to say, "No matter how 11 careful and unbiased is the adjustment process, it will 12 raise suspicion in the public's mind about the reliability 13 and integrity of the census." 14 It also raises suspicion in the public's mind 15 about deciding three years ahead of the date that there will 16 be no adjustment in the census. 17 And then he worries that this is going to reduce 18 the willingness of the public to respond to the census. 19 There is no evidence for that, no work was done. 20 There was no research on the part of the Department of 21 Commerce to come to this kind of dedicion. 22 And his final statement is that adjustment is a 23 threat to the customary process of reapportionment which has 24 been one of the foundations of our political system and that 25 adjusting the count may creat the appearance of changing the 1328 1 numbers to achieve a desired political outcome. 2 Not deciding or deciding against adjustment may 3 also creat that appearance. 4 There was absolutely no evidence for any ot these 5 points. There was no research done and I don't agree with 6 any of it. 7 THE COURT: Take a break? 8 MR. GOLDIN: Yes. 9 (Recess) 10 11 12 (Continued on the next page.) 13 14 15 16 17 18 19 20 21 22 23 24 25 1329 1 Q. Dr. Bailar, during the summer and first part of 2 the fall of 1987, before the press release that we have just 3 been looking at came out, were there occasions when you were 4 asked in public settings whether there had been a decision 5 made on adjusting the 1990 census? 6 A. Yes. 7 Q. What were those occasions? 8 A. There were professional meetings where people 9 were talking about the census, and so people would rise up 10 in the audience and ask about the decision in front of the 11 decennial census panel of the National Academy, and of 12 course the advisory committees. 13 Q. What did you say on those occasions? 14 A. I said that the decision was not yet made. 15 Q. Was that true? 16 A. No. 17 THE COURT: It turned out to be. 18 THE WITNESS: That's right. 19 THE COURT: Interesting question. 20 Q. Was that what you had in mind? 21 A. No. 22 Q. Did it give you concern to have to answer those 23 questions in that way? 24 A. Yes. It had never happened before at the census 25 bureau, and I had worked there a long time. It was a 1330 1 wonderful place. It was like an open university where 2 people could talk about everything they wanted to. They 3 could have public disagreements. You could just voice your 4 opposition to various points of view. 5 Now we were under this not being able to speak 6 about something. I was a person who was often contacted by 7 others because of my role there in the methodology area, so 8 I got a lot of questions about this, and I found it 9 extremely uncomfortable. 10 Q. When the Department of Commerce decision was 11 finally made public, what was the impact on bureau personnel 12 of the announcement of the department's decision? 13 A. There was a general feeling of frustration that 14 the decision had been taken out of the hands of the census 15 bureau, that all of the work that had been done and under 16 lots of time pressure had been just ignored, that no 17 additional research had been done to show this other point 18 of view. And the fact that you couldn't talk openly about 19 things anymore, it changed the nature of the place. 20 Q. When did you leave the bureau? 21 A. January 2, 1988. 22 Q. Why did you leave? 23 A. I had been uncomfortable with staying there with 24 the decision that it was clear to me that I was going to be 25 called upon to defend, since I had to defend the earlier 1331 1 decision, and it was not one that I felt comfortable about 2 defending. 3 Q. Was that a position that you took publicly, that 4 explanation for your departure? 5 A. Yes. 6 Q. When did you state that position publicly? 7 A. In testimony before Congress in March of 1988. 8 Q. When you left the bureau, was work ongoing 9 concerning the possibility of an adjustment in 1990? 10 A. No. 11 Q. Why not? 12 A. Because the decision had been made not to adjust 13 the census. 14 Q. Was there a point at which work on adjustment 15 resumed at the bureau? 16 A. Yes, after the stipulation and order. 17 Q. That was in 1989? 18 A. Yes. 19 Q. Was there further work done on the standards for 20 the adjustment decision? 21 MR. GLASS: Objection, it is not clear as to how 22 the witness would know the answer as to what was going on at 23 the census bureau after she left. 24 Q. Dr. Bailar, were you aware of work going on at 25 the bureau after your departure? 1332 1 A. Yes, I was. 2 Q. How did you become aware of work being done at 3 the bureau after your departure? 4 A. I attended research conferences. I attended 5 meetings where the work was discussed, meetings of the 6 American Statistical Association. There were many occasions 7 where this work was discussed. 8 Q. Were you aware of any further work being done on 9 the standards for the adjustment decision? 10 A. No. 11 Q. Were you aware of any further work being done on 12 those standards after the stipulation and order in this case 13 in 1989? 14 A. Yes. 15 Q. What work was done to further the development of 16 those standards after 1989? 17 A. There was work, I don't know how much work was 18 done, but there were some technical guidelines that were 19 released that looked very similar to the ones that had been 20 circulated in October of 1987 discussing the P studies. 21 Q. Were those guidelines incorporated into the 22 Secretary's guidelines for the 1991 decision? 23 A. Not into the eight guidelines that the Secretary 24 issued, no. 25 Q. Do you know what happened to those guidelines, 1333 1 the technical guidelines? 2 A. Those guidelines were actually used by the census 3 bureau. They carried out the P studies that were described 4 in those technical standards, and they were used by the 5 census bureau to look at the quality of the PES. 6 Q. Are you familiar with the results of the P 7 projects for the 1990 PES? 8 A. Yes. 9 Q. What do those results show? 10 A. They show that the level of the error in the PES 11 was held to very low amounts and that the general quality of 12 the PES was excellent. 13 Q. Dr. Bailar, let me just go back in your testimony 14 for a moment. You testified earlier about a meeting on June 15 2, 1987, that Dr. Keane attended with Under Secretary 16 Ortner? 17 A. Yes. 18 Q. Do you recall that testimony. Yes, I do. 19 Q. Have you had a moment to reflect further on it? 20 Do you recollect whether you were at that meeting? 21 A. Yes. I was not at that meeting. 22 Q. How did you come to know what happened at that 23 meeting? 24 A. From Dr. Keane. 25 MR. GLASS: I guess under those circumstances, 1334 1 your Honor, we would have to strike that testimony as 2 hearsay. 3 MR. GOLDIN: Dr. Keane is a party opponent. I 4 think that his account of the meeting is an admission. 5 MR. GLASS: It would be hearsay on hearsay, it 6 seems to me. 7 THE COURT: Keane is a party? 8 MR. GOLDIN: Yes. Dr. Keane was sued as a party 9 in the complaint as originally framed as director of the 10 census bureau and continues to be a party in the litigation, 11 unless Mr. Glass represents -- 12 MR. GLASS: He was not personally served. 13 MR. GOLDIN: Yes, he was. 14 MR. GLASS: Not in his individual capacity. 15 THE COURT: Can I intrude on this discussion? Is 16 Dr. Keane a named defendant in the action? 17 MR. GOLDIN: Yes. 18 THE COURT: He is also something in the census 19 bureau at the time? 20 MR. GOLDIN: He was the director of the census 21 bureau. 22 THE COURT: As against him it is admissible 23 because it is an admission by a party. As against the 24 census bureau, it is also an admission because it is made by 25 an officer, director, or employee of the party within the 1335 1 scope of his authority. It is both, therefore, an admission 2 and admissible, and the objection is overruled. 3 MR. GOLDIN: Thank you, your Honor. 4 Q. Dr. Bailar, in 1987, the bureau had planned, had 5 it not, that it would adjust the census in 1990 if technical 6 standard were met, meaning that no substantial unforeseen 7 errors or unanticipated serious flaws in the PES had 8 occurred? 9 A. Yes, that's right. 10 Q. Have you read the Undercount Steering Committee's 11 recommendation to Secretary Mosbacher? 12 A. Yes. 13 Q. Does that recommendation which we have previously 14 introduced into evidence as Plaintiff's Exhibit 54 apply 15 standards to guiding the decision-making process as the 16 bureau had anticipated in 1987 would be done at the second 17 decision point? 18 A. Yes, very much so. They followed the P studies 19 and took evidence from the P studies about the general 20 quality of the PES and used that in making the decision. 21 Q. Have you read the recommendation of the census 22 bureau director Barbara Bryant to Secretary Mosbacher? 23 A. Yes. 24 Q. Does that recommendation, also introduced into 25 evidence as Plaintiff's Exhibit 55, apply standards for 1336 1 guiding the decision-making process as the bureau had 2 anticipated in 1987 would be done at the second decision 3 point? 4 A. Yes. She very much follows that same line of 5 reasoning using the quality of the PES to bolster her 6 recommendation. 7 Q. Have you read the Secretary's decision? 8 A. Yes. 9 Q. Does the Secretary's decision apply standards for 10 guiding the decision-making process as the bureau had 11 anticipated in 1987 would be done at the second decision 12 point? 13 A. No. It varies from that in several ways. 14 Q. In what ways does it vary from that? 15 A. The Secretary seems to almost be able to divorce 16 himself at all from knowledge of what the census bureau 17 does. 18 MR. GLASS: Your Honor -- 19 THE COURT: Sustained. Don't go that far. 20 THE WITNESS: I'm sorry. 21 A. The census secretary doesn't seem to recognize 22 the work of the census bureau, either in general or 23 specifically as regards the guidelines and technical 24 standards, and then seems to harken back to the 25 recommendation by Under Secretary Ortner. What I mean by 1337 1 that is the census bureau is the leading agency in the world 2 for doing sample surveys. It knows how to select samples. 3 It knows how to do surveys. The Secretary seems to overlook 4 that and questions a sample survey. 5 He overlooks the fact -- he talks about cutting- 6 edge research and mentions things like post-stratification. 7 Census bureau uses post-stratification in almost every 8 survey of population that it carries out. It is not a new 9 technique. It's an old technique and one that the census 10 bureau pioneered over 20, 30 years ago, and has sort of 11 exported to the rest of the world, who looks at these 12 surveys and follows them. 13 The work on imputation and the use of imputation, 14 the Secretary seems to worry about the level of missing 15 data, which is quite minimal in the PES, and then says 16 something to the effect that the levels of missing data are 17 on the same order of magnitude as the undercount itself. 18 This isn't unusual. And there is nothing that says that 19 this is a problem. 20 If you look at the unemployment survey that the 21 census bureau carries out every month and has for years and 22 years, it uses imputation, and it has about a five-percent 23 nonresponse rate. For years and years the overall level of 24 unemployment in this country was about 5 percent. 25 If you look at income data, income data has a 1338 1 very high nonresponse rate, on the order of 20 percent. 2 When you look at many of the percentages that are calculated 3 on the basis of those data, they are much smaller than the 4 amount of missing data. So this isn't a good rationale to 5 be used by the Secretary in saying that this is cutting edge 6 research. It's done all the time, it's done by the census 7 bureau, and it is a well accepted statistical survey 8 procedure. 9 Then the Secretary seems to ignore the kinds of 10 standards and guidelines that the census bureau set to 11 follow and what its path had been set out to be. The census 12 bureau had these technical guidelines, said it was going to 13 evaluate the quality of the PES. It did so. 14 Census bureau should feel very proud to have done 15 a survey, a large survey, with the kind of quality that was 16 exhibited in that survey. It is almost unheard of to be 17 able to do a survey, to get that nonresponse rate down, and 18 then to be able to carry out the matching with such small 19 levels of error. It was marvelous. 20 The Secretary said they did a good job, passed it 21 off, and went on to something else. He raises other kinds 22 of questions that he is bringing in, but just sort of 23 ignores what the census bureau has set as the procedure and 24 that had been agreed on more or less by others who had 25 reviewed this. 1339 1 Then in parts of his testimony, he harks back to 2 Bob Ortner's decision, and you see these striking 3 resemblances to 1987, where he worries about the fact that 4 this is controversial and that it could get people not to 5 participate in the year 2000, where there is absolutely no 6 evidence -- you talk about a survey that was carried out. 7 Anybody who knows survey research can tell you that in a 8 survey where you ask people what they are going to do in ten 9 years, you should have saved your money, it's worthless. 10 MR. GLASS: Your Honor, I would object to that. 11 The witness is not qualified as an expert on survey 12 research. 13 THE COURT: All right. I'll strike it. It's 14 harmless. 15 MR. GOLDIN: What's that, your Honor? 16 THE COURT: I said I'll strike it. She doesn't 17 care about that. 18 MR. GOLDIN: Your Honor, just so the record is 19 clear, I believe that the witness did testify before she was 20 proffered as an expert that one of her particular areas of 21 expertise within statistics was survey research. 22 THE WITNESS: Yes. 23 THE COURT: Fine. If you want it in, we will 24 leave it in. 25 THE WITNESS: Thank you. 1340 1 THE COURT: It's in. 2 MR. GOLDIN: Skilled advocacy. 3 A. I have lost my train of thought. Excuse me. 4 MR. GOLDIN: Skilled advocacy. 5 A. In general, it harkens back to that earlier 6 decision, and it has overtones and echoes to it. So it is 7 hard to believe that a really good faith effort was done 8 again, and it makes me think that the Secretary's decision, 9 by just ignoring completely all the work that the census 10 bureau has done, all the things that it had set in place, 11 the good work that it actually carried out, and then harking 12 back to these overtones from 1987, that it was an 13 unreasonable decision. 14 MR. GOLDIN: Thank you, Dr. Bailar. 15 Your Honor, I have no further questions at this 16 time for this witness. 17 THE COURT: Thank you. 18 You are Mr. Davis? 19 MR. GLASS: I am David Glass, your Honor. 20 THE COURT: Mr. Glass. 21 CROSS-EXAMINATION 22 BY MR. GLASS: 23 Q. Good afternoon, Dr. Bailar. Good to see you 24 again. 25 A. Hello. 1341 1 Q. Let's start with a few preliminary matters. 2 THE COURT: You are going to have to hold your 3 voice up. I can't hear you, and I doubt the reporter can. 4 Of course, he doesn't want to. 5 MR. GLASS: That better? 6 THE COURT: Not really. 7 MR. GLASS: I will shout. 8 Q. Dr. Bailar, you testified, as a preliminary 9 matter, taht you were not receiving any compensation for 10 your testimony here and expected to receive none? 11 A. That's right. 12 Q. Do you know whether that is true for all of 13 plaintiff's witnesses in this case? 14 A. I don't know. 15 Q. You testified about the imputation that was 16 conducted in 1980 for the 1980 decennial census. Do you 17 know how many people were added to the census count through 18 imputation in 1980? 19 A. 762,000. 20 Q. Do you know how many people would be added to the 21 census for 1990 were the statistical adjustment method that 22 the Secretary rejected to be approved? 23 A. Somewhere between 4 and 5 million people. 24 Q. Thank you. You testified a second ago that the 25 Secretary's decision of July of 1991 harkened back to what 1342 1 Under Secretary Ortner said in 1987 in his decision. Do you 2 have any evidence that the Secretary gave any consideration 3 to Dr. Ortner's decision from 1987? 4 A. Only that the wording seems very familiar. 5 Q. But you have no specific evidence that you can 6 refer to that the Secretary actually referred to and relied 7 on Secretary or the near's decision? 8 A. No. 9 Q. Just the overtones and echoes? 10 A. That's right. 11 Q. In fact, was Under Secretary Ortner still in the 12 commerce department in 1991 when the Secretary made his 13 decision? 14 A. No. 15 Q. Dr. Bailar, you appear here today as a proponent 16 of adjustment, is that correct? 17 A. That gives me a little trouble. 18 THE COURT: As you sit here today, you think an 19 adjustment should be done? 20 THE WITNESS: Yes. 21 THE COURT: Better word. 22 MR. GLASS: Fine. 23 Q. You testified against the proposed adjustment of 24 the 1980 decennial census in the litigation that ensued from 25 that, isn't that correct? 1343 1 A. Yes. 2 Q. So that the Court can determine the amount of 3 weight to give to your present testimony, I would like to 4 contrast your present position on the adjustment issue with 5 certain views on that issue that you expressed before you 6 embarked on the road to Damascus, so to speak, and changed 7 your position. To do that, I would like to refer you to 8 Defendant's Exhibit 102, which is in the binder on the table 9 right next to you. 10 MR. GLASS: Your Honor, may I approach? 11 THE COURT: Yes. 12 A. I'm sorry. What number was it? 13 Q. Exhibit 102, please. 14 A. I don't have anything like that. 15 Q. It is toward the end. 16 A. The last one I have is 3. 17 THE COURT: Right in front of 3. 102 precedes 3. 18 Misenumeration. But we can adjust it. 19 MR. GLASS: We can both adjust it and smooth it, 20 your Honor. 21 A. I have 2. 22 Q. 2 is good enough. 23 THE COURT: Flip the tab that has 2 on it, and 24 you will see an exhibit number on the first page after that. 25 THE WITNESS: Thank you. 1344 1 Q. This exhibit is, is it not, an article published 2 in the journal of the American Statistical Association in 3 March of 1985? Is that correct? 4 A. Yes. 5 Q. It is an article by Professors Ericksen and 6 Kadane, and it is entitled "Estimating the Population in 7 Census Year: 1980 and Beyond"? 8 A. Yes. 9 Q. Following that, that article goes from page 98 to 10 109, and then following that are a series of comments on the 11 Ericksen and Kadane article, isn't that correct? 12 A. Yes. 13 Q. The first of those comments was written by you, 14 isn't that correct? 15 A. Yes. 16 Q. In their article, Professors Ericksen and Kadane 17 present a proposed methodology for the statistical 18 adjustment of the census, is that correct? 19 A. Yes. 20 Q. And the methodology that they propose bears some 21 strong similarities to the adjustment method that is 22 currently before the Court, isn't that correct? 23 A. No. 24 Q. It is not correct? 25 May I refer you to page 106 of this exhibit 302 1345 1 and refer you to the righthand column, down at the very 2 bottom of the page. There, two lines up from the bottom, 3 the following sentence appears. "We see the census 4 occurring in four basic steps: (a) the construction of 5 local address registers and alternative lists, (b) the mail 6 out-mail back procedure as occurred in 1980, (c) follow-up 7 activities, and (d) list matching with estimation." 8 Isn't it a fact, Dr. Bailar, that the central 9 aspect of the adjustment method before the Court, just as it 10 was the central aspect of the Ericksen/Kadane model, is list 11 matching with estimation? 12 A. This isn't the same kind of list matching at all. 13 Q. It is a different list, is it not? 14 A. That has all the difference in the world. 15 Q. But after you acquire the list, then you have 16 list matching with estimation, and you come up with dual 17 system estimators, isn't that a fact? 18 A. There are several differences here. This is not 19 the current methodology. There are changes in timing. 20 There are changes in the quality of the lists and the number 21 of lists that are being brought together. It has nothing to 22 do with the current method under consideration. 23 Q. But it is a fact, is it not, that both methods 24 involve the method of dual system estimation? 25 A. I haven't looked at this lately. I'll take your 1346 1 word for it that it involves dual system estimation. It 2 could involve more than dual system estimation, though. I 3 have heard multiple lists are used. 4 Q. Dr. Bailar, isn't it a fact that list matching 5 with estimation is the same as dual system estimation? 6 A. No. 7 Q. It is not the same? 8 A. It doesn't have to be the same. 9 Q. How is it different? 10 A. There could be list matching with dual estimation 11 the way it is currently being discussed. There could be 12 list matching of multiple lists, which they are talking 13 about here. And there could be an estimation scheme that 14 goes beyond dual systems, it could be many systems. 15 Q. In either event, regardless of the number of sets 16 of additional lists that you matched with results of the 17 census, in both schemes you are matching census result to 18 some other results, isn't that correct? 19 A. But it makes a big difference, how many lists and 20 what kinds of lists and where they came from and how you are 21 going to do the estimation. 22 Q. But the list matching is the same principle in 23 both regardless of the number of lists that you doing, isn't 24 that a fact? 25 A. Well, I can't agree to that because it is not the 1347 1 same. You don't have the same information in the lists. 2 You don't know that you have got the same kinds of 3 characteristics to help you do the matching. You have all 4 kinds of different problems with different kinds of lists. 5 It's a different system. 6 Q. But the process of matching is the same, is it 7 not? 8 A. The process of matching may not be the same 9 because you may have to match on different kinds of 10 characteristics. 11 Q. Let me refer you to the comment that you had on 12 the Ericksen/Kadane proposal here. Let me refer you first 13 of all to page 110 of your comment, to the lefthand column 14 on that page, and to the third full paragraph. That reads 15 as follows: 16 "Some deep questions about policies and goals 17 must be raised about a shift to a statistical census." You 18 then go on to list four of these deep questions of policies 19 and goals. Then you go on to say, "These are questions that 20 no one can now answer, but the possible public perceptions 21 associated with a statistical census concern me." 22 The first of the deep questions of policies and 23 goals that you ask in that paragraph is the following. 24 "What happens to the census when counting each individual is 25 no longer viewed as important by the residents because the 1348 1 census bureau will estimate the population?" 2 Dr. Bailar, as you sit here today, what answer do 3 you give to this deep question as it applies to the proposed 4 adjustment of the census before the Court? 5 A. You are asking me about this sentence that I 6 wrote in connection with an entirely different methodology, 7 and asking me what relation it has to the method before the 8 Court? 9 Q. I am asking you what answer it has to this 10 question as it pertains to the adjustment methodology before 11 the Court. 12 A. It has no importance for this current adjustment 13 because there has been nothing in this current adjustment to 14 make each individual no longer viewed as important by the 15 residents at all, because the census bureau did make them 16 feel important, the count did go forward. There has so far 17 not been an agreement there will be an adjustment, but it is 18 certainly not going to have an effect after the fact. 19 Q. So it is your opinion that there will be no 20 impact on the census if an adjustment goes forward pursuant 21 to the method before the Court, there will be no impact 22 because counting each individual will no longer be viewed as 23 important by the residents? 24 A. What says -- 25 Q. You are saying that this question that you pose 1349 1 here, this deep question, is no longer applicable to the 2 proposed method of adjustment before the Court? 3 A. Are you talking about what census in 1980 census? 4 Q. The 1990 census. 5 A. The 1990 census? 6 Q. That's right. 7 A. Of course. What effect can you have? Going back 8 two years, what effect can it possibly have on people's 9 behavior two years ago? 10 Q. And it will have no effect on the 2000 census? 11 A. I believe that is true too. 12 Q. It is your testimony that this is a question that 13 one can now answer? 14 A. It is a question that was raised about an 15 entirely different method of doing an undercount adjustment, 16 a different kind of procedure for taking a census, and it 17 has no relevance, as far as I can see, to deal with the 18 current method under discussion. 19 Q. In 1985, when you wrote this article, you said 20 that this was a question that one could not now answer, and 21 your testimony is that one can now answer this question? 22 A. One can certainly answer about the kind of 23 adjustment process that we are talking, since the census has 24 already gone forward and the adjustment has not been decided 25 upon. 1350 1 Q. Is it a question that one can now answer as to 2 upcoming censuses? 3 A. I believe so. 4 Q. Let's take a look at the second question, deep 5 question of policies and goals that you raised here in this 6 paragraph. That is, if a statistical census goes forward, 7 "does the census lose its legitimacy?" 8 What answer do you have to this question that you 9 pose as to the statistical method presently before the Court 10 in terms of upcoming censuses? 11 A. There is no idea at all to try to do away with 12 the counting of people in a census. I think everyone agrees 13 that it is important to do a good census, and that the only 14 place that this adjustment comes in is at the final -- to 15 take care of the final few percentage points at the end of 16 the census process. 17 So in my mind, because this is a very different 18 process here that would have changed the fundamental nature 19 of the census, this question about the census losing its 20 legitimacy has no relevance to what we are talking about. 21 What we are talking about is a census that counted about 98 22 percent of the population, and we are trying to take care of 23 that final 2 percent. Doing that, it seems to me, does not 24 cause the census to lose its legitimacy. 25 Q. This is a question that one can now answer as to 1351 1 the possible loss of legitimacy for the census? 2 A. Since I wrote this about something completely 3 different and you are asking me to apply it to the current 4 situation, yes, I think I can answer it. 5 Q. The third deep question that you pose here is, 6 "Does the census become suspect because we use" -- 7 THE COURT: What page are you on now? 8 MR. GLASS: I'm on page 110 of the witness's 9 article. 10 THE COURT: All right. I have the article. 11 MR. GLASS: I am in the lefthand column, and I am 12 in the third full paragraph. 13 THE COURT: Thank you. 14 Q. The third of the, quote unquote, deep questions 15 of policies and goals that you say must be raised about a 16 shift to a statistical census is the following: "Does the 17 census become suspect because we use confidential 18 information from many sources to add people who did not come 19 forward to be counted?" 20 Dr. Bailar, as you sit here today, what is your 21 answer to this question as it applies to the adjustment 22 method before the Court and as it may impact on future 23 censuses? 24 A. This question has absolutely no relevance in 25 connection with the current adjustment method being proposed 1352 1 because confidential information from many sources was not 2 used. It has absolutely no relevance. 3 Q. So you are saying that because information from 4 only one source, i.e., the PES, would be used to adjust, 5 there would be no danger of the census becoming suspect 6 because of the addition to the census count of people who 7 did not come forward to be counted? 8 A. This was talking about using lists that come from 9 many different sources, some of which may be confidential, 10 such as welfare lists, and whether people will get nervous 11 about the appearance of a Big Brother out there matching 12 lists. 13 What we are talking about with this current 14 adjustment method is using a PES conducted by the census 15 bureau, the same agency that conducted the census, and did 16 it under their own confidentiality provisions, Title 13. I 17 don't think this has any relevance about people worrying 18 about this question at all. This was written about 19 something different. 20 Q. It is a fact, is it not, that if the adjustment 21 method before the Court is approved, people will be added to 22 the census who did in the come forward to be counted? Isn't 23 that so? 24 A. Some of them will be added who didn't come 25 forward to be counted and some will be added because the 1353 1 census procedures didn't allow them to be counted. 2 Q. And it is your opinion as you sit here today that 3 the fact that a certain number of people will be added to 4 the census who did not come forward to be counted will not 5 cause the census to become suspect? 6 A. That's right. 7 Q. Is that, in fact, a question that one can now 8 know? 9 A. I'm sorry. I have lost track of what your 10 question was. 11 Q. At the very end, at the passage that we have been 12 reviewing on page 110 here, you talk about these deep 13 questions of policies and goals and then say, these are 14 questions that one cannot now answer. 15 A. Yes. 16 Q. I am asking whether one can now answer whether 17 future censuses will become suspect if people are added to 18 the 1990 census count who did not come forward to be 19 counted. 20 A. And the reason that they are added is because of 21 the use of confidential information? 22 Q. Simply the fact that people are added who did not 23 come forward to be counted. 24 A. That is only a part of the thought expressed 25 here. But no, in my opinion, the census does not become 1354 1 suspect. 2 Q. Let's take a look at the fourth of these four 3 deep questions that you ask here, and that is as follows. 4 "Is the result of a statistical census more suspect than a 5 participative census because there are fewer people directly 6 enumerated and those fewer people have greater influence 7 over the resulting population estimates?" 8 As you sit here today, what is your answer to 9 this deep question that you pose as it applies to the method 10 of adjustment before the Court and as it would apply to 11 censuses taken in the year 2000 and beyond? 12 A. I hate to keep saying the same thing, but, you 13 know, this was a different kind of census that we are 14 talking about here. Under that kind of census structure, 15 there could have been many fewer people directly enumerated. 16 That isn't the case here. We had 98 percent of the people 17 directly enumerated. 18 If I go back to the only evidence that I know of 19 that exists, there are two sources of evidence on this. One 20 is that there was a major adjustment of the census in 1970, 21 and over a million people were added to the census. As far 22 as I know, and I do think I do know the answer to this, no 23 one was worried about the census becoming suspect because 24 these people were added. 25 The second source is that people are added to the 1355 1 census through the imputation process in every census that I 2 know of, and again the census doesn't become suspect because 3 a statistical procedure was used. In fact, I think it 4 legitimizes the census because it recognizes that those 5 people who are there were not counted and should have been 6 counted and now the census bureau is doing something to try 7 to remedy the situation. 8 Q. How many people did you say were added to the 9 census through imputation in 1970? 10 A. I believe I said 762,000. 11 Q. How many did you say would be added to the census 12 for 1990 if the method before the Court were approved? 13 A. I said between four and five million. 14 Q. Thank you. Dr. Bailar, would you characterize 15 the deep questions of policies and goals that you list here 16 as technical or nontechnical? 17 A. Nontechnical. 18 Q. If the Secretary found certain nontechnical 19 questions to be relevant to his decision on the statistical 20 adjustment issue, that would not render his decision 21 improper, would it? 22 A. I think the Secretary probably was within the 23 realm of what he should have considered looking at some 24 nontechnical issues, but not at the expense of the technical 25 issues. 1356 1 Q. Let me direct your attention in this article, 2 this comment of yours that we are looking at, to the 3 following paragraph. This would be page 110, lefthand 4 column, fourth full paragraph. There is a sentence in there 5 toward the end of it where you say the following. "Error in 6 the proportionate distribution of population among states 7 and other jurisdictions is more important than absolute 8 error." 9 Dr. Bailar, does this continue to be your 10 opinion? 11 A. Yes. 12 Q. Moving up to the top of the righthand column on 13 the same page, you make the following statement there, 14 beginning three lines from the top. "In a statistical 15 census, the outcome from the census and from the adjustment 16 would both be known, compared, and argued over. The lawsuit 17 filed by the State of Indiana is a harbinger of the future." 18 By the lawsuit filed by Indiana, I believe you 19 were talking about the 1980 lawsuit that you testified to 20 earlier, isn't that correct? 21 A. Yes. 22 Q. Do you continue to agree with the statement that 23 you make here, that if the statistical adjustment method 24 were approved, the outcome from the census and from the 25 adjustment would both be known, compared, and argued over? 1357 1 A. Are you talking about right now, the 1990 2 results? 3 Q. The 1990 results, and I am asking what the impact 4 would be for future censuses if that methodology is then 5 applied to future censuses. 6 A. Do I think lawsuits will be filed? 7 Q. Do you continue to think that this is a problem, 8 that the outcome from the census and from the adjustment 9 will be known, compared, and argued over, and that the 10 lawsuit filed by the State of Indiana will be a harbinger of 11 the future? 12 A. The lawsuit filed by the State of Indiana, which 13 the census bureau won, no other lawsuit that I know of was 14 filed in that regard and on that issue. So I don't know 15 whether it is or whether it isn't. 16 Q. Is it your opinion that if the adjustment 17 methodology is improved, what we can look forward to is an 18 unending series of lawsuits about the census? 19 A. Oh, I don't know whether there will be or not. I 20 am sure that there will be lawsuits. After all, we are here 21 on one right now. Until the issue gets settled and 22 adjustment becomes a part of the census the way the 23 imputation procedure is, until it is fully integrated into 24 the census, there probably will be lawsuits. But once it 25 becomes a full part of the census, there is no reason to 1358 1 believe that it will be. 2 Q. Let me refer you on the same column on page 110, 3 to the third full paragraph in the righthand column, above 4 the heading "The 1980 Census." There you make the following 5 statement. "But when error is intolerable in a 6 participative sense based on counting, would error based on 7 a statistical census be tolerated"? 8 What is your answer to this question as we sit 9 here today? 10 A. I think in this paragraph I was referring to 11 thresholds. Cities have certain thresholds, and if they get 12 over a certain count, they get certain benefits. In those 13 cases I say no error is tolerable, and no error is tolerable 14 no matter how you do the census. 15 Nobody is willing to believe that these numbers 16 are absolute the way they are recorded generally by the 17 census. There is a range of error in the census. Everyone 18 knows that. When a threshold is reached, it seems unfair to 19 people because they know there is error in the census. 20 Q. Isn't it a fact, Dr. Bailar, that the answer to 21 the question that you pose here is set forth in the 22 following sentence, and that is, "An appearance of greater 23 accuracy would be required of a statistical census, since 24 adjustments of unknown accuracy would be superimposed on 25 counts of reduced completeness"? 1359 1 A. But then, as I have said before, I am not talking 2 about -- you aren't talking about a statistical census. You 3 are talking about the 1990 procedure. So we are not talking 4 about an adjustment of the size that we are talking about in 5 the area of a statistical census. 6 Q. So it is your testimony that an appearance of 7 greater accuracy would not be required for the statistical 8 method before the Court for the reasons set forth in that 9 sentence? 10 A. I don't believe so. 11 Q. Let me refer you to the next page, page 111, the 12 lefthand column, the second full paragraph. The statement 13 appears about six or seven lines down. "It has been the 14 bureau's policy, consistent with the law, to take as 15 complete a census as possible, given time, money, and 16 personnel constraints." 17 Do you disagree that the foregoing describes the 18 approach that the bureau took in conducting the 1990 19 decennial census? 20 A. Except for not incorporating the adjustment. 21 Q. But aside from that? That the bureau, consistent 22 with the law, took as complete a census as possible, given 23 time, money, and personnel constraints? 24 A. Certainly they had the time, money, and they did 25 have personnel constraints. Yes, I believe they did take as 1360 1 complete a census as possible. 2 Q. You do not contend, do you, that the bureau made 3 a conscious effort to undercount blacks, Hispanics, or other 4 advantaged groups in conducting the census? 5 A. No, there wasn't a conscious effort to do so. 6 And this isn't conscious, but the census bureau recognizes 7 that they have not for 50 years been able to count minority 8 populations. So when they continue to take a census in the 9 same manner as earlier censuses, it does lead one to 10 question whether there is being a good-faith effort made to 11 take a count of those populations that don't conform to 12 these middle class expectations of a census. 13 Q. But it is not your testimony, is it, or your 14 position that the bureau set out to undercount those 15 populations, that it intended to do so? 16 A. No. 17 Q. It is not your contention, is it, that the 18 decision of the Secretary not to approve a statistical 19 adjustment was based on a desire on his part to discriminate 20 against blacks, Hispanics, or other disadvantaged groups? 21 A. No. 22 Q. Let me refer you back in your article here to the 23 same paragraph that we were looking at, that is, the 24 lefthand column on page 111, second full paragraph. I would 25 refer you to the sentence at the end of that paragraph which 1361 1 reads as follows. "The bureau could not have adopted a 2 policy of less counting and more estimation in 1980, since 3 that would have been a sharp departure from our traditional 4 legal mandate as well as from societal expectations." 5 Isn't it a fact, Dr. Bailar, that that sentence 6 refers just as clearly and as accurately to the 1990 7 decennial census and its proposed adjustment as it does to 8 the 1980? 9 A. Not at all. 10 Q. Isn't it a fact that a statistical adjustment of 11 the 1990 census would constitute a sharp departure from the 12 bureau's legal mandate? 13 A. No. When I wrote that, the bureau was still in 14 the -- it had a policy in 1980 that it did not believe it 15 could adjust the census for apportionment and redistricting. 16 It would only talk about adjusting the census for funding 17 allocations. 18 So when I wrote this, I was still writing it with 19 that idea in mind, that the traditional legal mandate was 20 that you couldn't use an adjustment for apportionment 21 redistricting. 22 Q. Dr. Bailar, I know that you are not a lawyer 23 here, but you were referring to a legal mandate. Can you 24 refer me to the specific document which embodies the change 25 in the legal mandate between the time that you wrote this in 1362 1 1985 and today? 2 A. I don't know where the legal mandate was. I know 3 Vincent Barabba referred to it in some of those Federal 4 Register documents that we were looking at earlier, that 5 this is the way that he believed this policy was 6 interpreted. Then we listened to all the Court trials 7 during 1980 and were advised by our lawyers -- 8 I am talking about our, I am talking about the 9 census bureau. I still have a little trouble sometimes 10 divorcing myself there. 11 -- told by the lawyers that given what came out 12 of the court trials in 1980, the bureau's interpretation had 13 been incorrect. That is my basis for going on it. 14 Q. Isn't it a fact that those court trials took 15 place before this article was written? 16 A. Yes. But I hadn't talked to the lawyers fully 17 about what had come out of all this. We had talked about 18 specific cases, but we hadn't talked about what the whole 19 group of trials meant and what it meant for this policy. 20 Q. Let me ask you about your statement here that 21 statistical adjustment would constitute a sharp departure 22 from societal expectations. Isn't it a fact that that 23 statement applies just as clearly to the proposed adjustment 24 before the Court as it did to the proposed adjustment of the 25 1980 decennial census? 1363 1 A. I don't think so, because the stage was different 2 before the 1990 census. There was a lot more discussion, a 3 lot more people realized what the problems were. There was 4 much more knowledge of the differential undercount of 5 minority populations. So societal expectations might very 6 well have been that the bureau should do whatever it needed 7 to do to correct that problem. 8 Q. Have you done any research on societal 9 expectations vis-a-vis the statistical adjustment of the 10 census? 11 A. I think societal expectations are a very 12 difficult area of research, so no, I haven't tried that. 13 Q. You have no basis for your conclusion that an 14 adjustment of the census at this point in time would not be 15 a sharp departure from societal expectations? 16 A. No more than I had for saying this when I said it 17 earlier. I hadn't done the research then either. 18 Q. That may be the point. 19 Looking down in the next paragraph after this, 20 you make the following statement. 21 THE COURT: What page are you on now? 22 Q. Page 111, lefthand column, third full paragraph, 23 right at the middle of the page. 24 THE COURT: Okay. 25 Q. Dr. Bailar, you make the following statement. 1364 1 "'Curbstoning,' the fabrication by people by enumerators, 2 has always occurred in census and in surveys as well." 3 Isn't it a fact, Dr. Bailar, that this statement 4 applies to the PES, that there was curbstoning there? 5 A. Yes. And there was curbstoning in the census as 6 well. 7 Q. That's fine. I am simply asking you about the 8 PES. 9 Let me ask you about a statement that you make in 10 the next column of this article on page 111, third full 11 paragraph from the top. It is rather a long quote here. I 12 will lead it as slowly as I can. 13 "There are at least three basic assumptions to 14 the use of dual-system estimation based on matching studies 15 to give unbiased or consistent estimates: Both systems must 16 include no erroneous inclusions, for such erroneous 17 inclusions must be corrected or deleted prior to estimation; 18 the matching procedure must be perfect or close to it, with 19 little or no ambiguity; and the two systems must be 20 statistically independent or have a valid basis for 21 estimating the effect of correlations." 22 By the way, before I go on to the substance of my 23 question here, does this refresh your recollection as to 24 whether what Ericksen and Kadane were talking about was 25 dual-system estimation? 1365 1 A. This says it is dual-system estimation. 2 Q. That's right. Isn't dual-system estimation what 3 is before the Court currently? 4 A. Yes. 5 Q. So there are substantial similarities between the 6 statistical adjustment method proposed by Ericksen/Kadane 7 and the statistical adjustment method before the Court? 8 A. I'm not convinced that they don't have something 9 else in mind as well, because they are talking about more 10 than one list. But I will grant you that at some point 11 perhaps they are talking about dual-system estimation. 12 Q. Fine. Based on this passage that I read to you, 13 does it continue to be your view that the use of the 14 dual-system estimation based on matching studies to give 15 unbiased or consistent estimates requires the matching 16 procedure to be perfect or close to it, with little or no 17 ambiguity? 18 A. The matching procedures have to be very good. 19 They do not have to be perfect. And of course no one these 20 days really requires an unbiased estimate. The census 21 bureau surveys, I don't think there is any survey in which 22 they put out unbiased estimates. 23 Q. We are talking here about ambiguity in the 24 matching process. 25 A. Yes, but you are asking me if this is a basic 1366 1 assumption, and I am telling you that if you are putting it 2 all on the basis of you have to have unbiased estimates, 3 that is not a consideration that is really important in 4 estimation. 5 Q. What I am asking you here is whether the use of 6 dual-system estimation to give unbiased or consistent 7 estimates requires the matching procedure to be perfect or 8 close to it with little or no ambiguity. 9 A. It requires it to be a very good process, as the 10 1990 PES showed the matching process was. 11 Q. How much ambiguity in the matching process is 12 permissible, if your view? 13 A. I don't know, because it really depends on a lot 14 of factors. I would have to know more than a simple number 15 to know what kind of ambiguity there are there was, what 16 kind of groups were involved in it, and so forth. 17 Q. So you have no opinion as to the amount of 18 ambiguity as we sit here that is permissible? 19 A. No. I would have to have a lot more detail. 20 Q. Do you know as you sit here today how much 21 ambiguity in the matching process existed in the matching 22 between the PES and the census results? 23 A. No. I only read the summary of the P studies 24 that seemed to say that the matching was very good. 25 Q. So you neither can tell us what the permissible 1367 1 amount of ambiguity is in the abstract, nor the amount of 2 ambiguity that actually occurred in the matching between the 3 PES and the census? 4 A. The amount of ambiguity has to be looked at in 5 the context of several other facts, as I have just mentioned 6 to you. 7 Q. But you do not know the answer to that? 8 A. No. 9 Q. Let me refer you ahead two full pages here, to 10 page 113 of this article, to the lefthand column. Down at 11 the bottom there is a paragraph that starts with the Arabic 12 numeral 1. I am looking at the paragraph that is right 13 above that. That gets into some of the technical 14 assumptions required for smoothing. Let me read you some of 15 this. 16 "The proposal to combine regression and sample 17 estimates and use the resulting estimates not only for 18 estimates at the state and city level, but as the basis for 19 adjustment for all other smaller areas, is intriguing and 20 one that the bureau is looking at carefully for 1990. 21 However, some critical questions must be resolved at least 22 in part before the methodology could be adopted. 23 "1. Are the underlying assumptions of such a 24 model met? There are several layers of assumptions, as 25 pointed out by Freedman (1984) as a witness for the 1368 1 government in the New York lawsuit. First is an assumption 2 that the estimates of net census error are unbiased or 3 substantially so. The second is that the true net census 4 errors are linearly related, perhaps by use of a 5 transformation, to a set of explanatory variables. Other 6 assumptions are of the expected values of the error terms in 7 regression equations are zero, that the error terms have 8 equal variances, and that they are uncorrelated over areas." 9 Dr. Bailar, the first assumption that you 10 identify here as one that has to be met is an assumption 11 that the estimates of net census error are unbiased or 12 substantially so. Can you tell us as you sit here today 13 whether this assumption is met for the adjustment model 14 before the Court? 15 MR. GOLDIN: Your Honor, this question is plainly 16 outside the side of the direct. At the time that I 17 proffered Dr. Bailar as an expert, Mr. Glass specifically 18 made a point that she should not testify during the course 19 of her direct examination with respect to smoothing. I 20 agreed to that. I adhered to that. And there was, in fact, 21 no testimony concerning smoothing on the direct. I 22 therefore believe that this question is plainly outside the 23 scope. 24 MR. GLASS: Your Honor, I am not asking about 25 smoothing in the abstract. I am asking about what the 1369 1 witness said in 1985 as to what assumptions had to be met 2 for smoothing to be valid. I simply want to ask her whether 3 those assumptions are met. 4 THE COURT: Since you fenced off smoothing from 5 the direct examination, it strikes me as unfair to open the 6 gates to that fence on cross. I am going to sustain the 7 objection. 8 MR. GLASS: Okay, fine. 9 THE COURT: It is now five o'clock, Mr. Glass. 10 If you can finish, we will go ahead. If you are not going 11 to finish -- 12 MR. GLASS: I have maybe another half hour's 13 worth of questions. I am perfectly willing to resume 14 tomorrow morning and finish tomorrow morning. 15 THE COURT: Most important, what about the 16 witness? 17 THE WITNESS: I would like to finish tonight if I 18 could, so could go back home. 19 THE COURT: Do you want to stay? 20 THE WITNESS: Yes. 21 MR. GOLDIN: That is what I was going to say, 22 your Honor. 23 THE COURT: Let's stay. 24 MR. GLASS: In this case, I will try to move this 25 forward as rapidly as I can. 1370 1 Q. Dr. Bailar, let me refer you ahead to page 114 in 2 the article, the lefthand column, the second full paragraph. 3 There you make the following statement. "Anyone who has 4 done list matching speaks of it as a time-consuming and 5 error-prone operation." 6 Dr. Bailar, isn't it a fact that this statement 7 applies to the matching that took place between the P sample 8 and the E sample that the bureau conducted? 9 A. In 1908? 10 Q. In 1990. 11 A. A lot has changed since I wrote this. The bureau 12 developed a computer matching software program that takes 13 care of at least 80 percent of the cases and only leaves the 14 20 percent behind that are the more difficult cases but that 15 people can concentrate on. And also because of more 16 automation, they don't have to go into those terrible stacks 17 and hunt around in boxes, and so forth. 18 So a lot of what I said here, certainly the 19 time-consuming part of it has been cut down, and I believe 20 the error proneness has also been cut down. 21 Q. Looking at the 20 percent of the matching work -- 22 that is your figure, the 20 percent -- where the people 23 don't have to go back in stacks and look through all the 24 boxes, that, in fact, is still an error-prone operation, is 25 it not? 1371 1 A. But the 20 percent that is left is not as 2 difficult as it was in 1980. It really has been, through 3 the automation, improved. That is not to say that some of 4 the matching cases aren't more difficult. But you don't 5 lose your energy during the day when you have to keep 6 running back into the stacks and pulling boxes down. You 7 can do it through a computer. You can look at listings on 8 computer, and so forth. It is a much easier operation. 9 Q. So you are saying that there are no errors now in 10 the 20 percent that is not computer matched? 11 A. No, that is not what I said. I said it is not as 12 time-consuming and error-prone an operation as it was in 13 1980. Of course, some errors can be there. 14 Q. Let me direct your attention to the top of the 15 righthand column on page 114, in the carryover paragraph 16 there. There you say the following. "Adjustment must be 17 used to correct residual problems, not to replace an 18 accurate enumeration, and it must be done only if there is 19 widespread professional agreement that it will increase 20 accuracy." 21 Dr. Bailar, you haven't been here and you will 22 not be here for the government's case, but the witnesses who 23 will testify for the government are imminent statisticians 24 such as David Freedman, Paul Meier, Leo Breiman, Ken 25 Wachter. 1372 1 Isn't it a fact that if all of the government's 2 witnesses say that they are opposed to adjustment, that it 3 would not increase accuracy, that widespread professional 4 agreement does not exist on the adjustment question? 5 A. I think widespread professional agreement does 6 exist, especially among those people who have been involved 7 with this kind of procedure, with the census, with looking 8 at the results. You know, professional agreement doesn't 9 mean that everybody has to agree. It does mean that there 10 has to be some kind of consensus. 11 I believe there is a consensus, especially among 12 people who have done work for the census bureau, know those 13 operations, know what the PES is, and have some history of 14 being affiliated with a long relationship with the census 15 bureau. For example, the National Academy panel. 16 Q. Assuming that the government's experts do have 17 substantial familiarity with the process and with the 18 issues, it is still your testimony that widespread agreement 19 exists as to whether an adjustment will increase accuracy? 20 A. I believe there is a consensus on this, even 21 though some people still don't agree. 22 Q. Some people still haven't seen the light. 23 THE COURT: Is there any one thread that unifies 24 the people he has mentioned? Are they of a particular 25 philosophical school? Are they bound together in some way 1373 1 that gives them a unity against the rest of the profession? 2 I know I am not reaching the point too well. Do you see 3 what I am getting at? 4 THE WITNESS: I know what you are getting at, but 5 I don't see that unity. 6 Q. Let me simply ask you one final question about 7 your article here. I refer you to the last paragraph in the 8 article. That paragraph begins as follows: "There must 9 also be some agreement on the standard of accuracy to be 10 used." Then skipping down to the end of the paragraph, you 11 ask the following question: "Would there need to be 12 assurance that 90 percent of the areas are better off after 13 adjustment or 99 percent? These are questions that need to 14 be considered." 15 Dr. Bailar, what is your answer to the number of 16 areas that have to be better off after adjustment for 17 adjustment to be valid? Is it 90 percent or is it 99 18 percent? 19 A. Whether it is 90 percent or 99 percent, I think 20 if you look at the kind of adjustment that was done in the 21 1990 census, when at least two-thirds of the people are 22 living in areas where the adjustment is improving things, to 23 me that is a clear indication that we could go forward. 24 90 percent, does that mean 90 percent by just a 25 strict count or does it mean that 90 percent of the 1374 1 population totals? I think when I wrote this I was raising 2 those questions. But I think when you actually apply those 3 questions, you have to think deeper about what you are 4 talking about, whether it is numeric, absolute numbers, 5 proportions, what. 6 Q. Is it still your belief that there must be some 7 agreement on the standard of accuracy to be used before a 8 statistical judgment is appropriate? 9 A. I think it would have been a good idea before the 10 census started if there had been some agreement on the 11 standard of accuracy. That was precluded by having to stop 12 work on the standards, by not being able to go out and 13 discuss these things with the professional community. 14 Because of that, we didn't get that assurance ahead of time. 15 But I think the census bureau's Undercount 16 Steering Committee has made a good case for the fact that it 17 is improving accuracy, distributive accuracy as well as 18 numeric accuracy. 19 Q. Is it your testimony, though, that it is still 20 important that there must be some agreement on the standard 21 of accuracy? 22 A. Well, it would be nice. But it didn't happen in 23 this case because of factors that were beyond the control of 24 the census bureau. 25 Q. Does such agreement now exist? 1375 1 A. It exists certainly among the Undercount Steering 2 Committee, who came to a recommendation about what they 3 believed, that it was increasing distributive as well as 4 numeric accuracy. I believe it exists. In people who have 5 followed this that used to be on the decennial census panel, 6 I believe that they agree with a standard of accuracy. 7 I can only say it is too bad that work on this 8 was stopped during 1987 so that that standard couldn't have 9 been clearly articulated. But it was stopped, so you have 10 to make do with what exists. 11 Q. In exact, what you are saying is that agreement 12 exists among those who agree with you, but among those who 13 don't agree with you, there is no agreement on an standard 14 of accuracy? 15 A. I haven't heard them say anything about what 16 their standard of accuracy is. 17 Q. Fine. Let me simply ask you a few brief 18 questions about some of the exhibits that were admitted into 19 evidence or discussed during your testimony, and then we 20 will have done with here. 21 Let me refer you to Plaintiff's Exhibit 596, 22 which was your testimony before Congress. I would refer you 23 to page 26 of your written statement. There in the second 24 paragraph on the page you say the following: "Adjustment is 25 a highly complex and controversial subject. We are working 1376 1 on many technical issues that we must resolve before we can 2 undertake adjustment. But we must also develop an approach 3 to adjustment that will achieve widespread public support 4 and understanding." 5 Dr. Bailar, does it continue to be your belief 6 that adjustment must achieve widespread public support and 7 understanding before it is permitted to go forward? 8 A. When I wrote this, we were looking toward -- this 9 was in 1986 and we were looking towards a process of 10 educating the public about what an adjustment would mean, 11 and so forth. As I mentioned a few minutes ago, that was 12 precluded by the action of the Department of Commerce just 13 ruling out a possible adjustment, so work fell behind in 14 that. In fact, I don't believe that any work went forward 15 by the census bureau after that to try to get a public 16 understanding of what adjustment was. 17 So the fact that I said that it is important, 18 yes, it would have been important, it would have been nice 19 to have that. But we are not in that situation because of 20 events that were beyond the control of the census bureau. 21 Q. Is it still important that such widespread public 22 support and understanding exist? 23 A. Actually, I guess I am not so sure of that 24 anymore, since I have noticed that from the reports of the 25 NLRC survey, that people don't even understand very much 1377 1 about what a census is, much less what an adjustment is. So 2 it doesn't seem to affect what they think about it. 3 Q. Do I misremember, or during your direct testimony 4 didn't you say that nobody would put any stock in that 5 particular survey? 6 A. That's true. On the other hand, that is the only 7 evidence that exists on this. 8 Q. Let me refer you to Plaintiff's Exhibit 604 here. 9 Those are the delegation orders from the Secretary of 10 Commerce to the director of the census bureau. I would 11 refer you to the later of these two orders, the July 22, 12 1987 order. I would refer you to section 2.02. The last 13 sentence of that reads as follows. "The director," that is, 14 the director of the census bureau, "shall report and be 15 responsible to the Under Secretary for Economic Affairs." 16 Dr. Bailar, doesn't that sentence indicate that 17 the director of the bureau is not a free agent and that his 18 or her decisions can be countermanded at any point by the 19 department, by the Secretary of Commerce? 20 A. I didn't think this order meant that the director 21 was a free agent, but it does delegate responsibility to the 22 director for the conduct of censuses and surveys, especially 23 in technical areas. 24 Q. So your testimony is that the decisions of the 25 director can be countermanded by the commerce department? 1378 1 A. Obviously. They have been. 2 Q. Finally, let me just ask you this one last 3 question. Isn't a fair characterization of this lawsuit as 4 follows: That what plaintiffs are asking the Court to do 5 here is, "to take sides in a dispute among statisticians, 6 demgraphers, and census officials concerning the 7 desirability of making a statistical adjustment to the 8 census head count"? 9 A. I'm sorry. Could you repeat that? 10 Q. Sure. The plaintiffs are asking the court "to 11 take sides in a dispute among statisticians, demgraphers, 12 and census officials concerning the desirability of making a 13 statistical adjustment to the census head count." Isn't 14 that all that is really going on here? 15 A. No. 16 Q. No? Okay. 17 A. Maybe I don't understand your question, but to me 18 it's no. 19 MR. GLASS: Thank you. I have nothing further, 20 your Honor. 21 THE COURT: Any redirect? 22 MR. GOLDIN: Very briefly, your Honor. 23 REDIRECT EXAMINATION 24 BY MR. GOLDIN: 25 Q. Dr. Bailar, during your cross-examination I heard 1379 1 you to say, I believe I was following this quite carefully, 2 that there were 762,000 whole person imputations in the 1970 3 census. 4 A. I'm sorry. 5 Q. Let me ask you whether that was what you meant to 6 say. 7 A. That was 1980. 8 Q. Was there an adjustment of the 1970 census? 9 A. Yes. 10 Q. What was that adjustment? 11 A. There was imputation in that census of the same 12 kind that we were talking about in 1980. Then there was an 13 adjustment for two different parts of the 1970 census. One 14 of them was the problem that when a small sample was 15 revisited of households that had been enumerated by census 16 takers as vacant, when they were revisited it was found that 17 they were occupied. So based on that sample, an adjustment 18 of the census was done. 19 Then another part of an adjustment was in the 20 rural areas where the there was a study called PEPOC, and I 21 can't for the life of me remember what it stands for. 22 Q. Post-enumeration post office check or something 23 like that? 24 A. Something like that. They went back to look at 25 addresses there and again had to add people in. I believe 1380 1 it was well over a million people who were added in as an 2 adjustment to the 1970 census. 3 Q. Let me call your attention again to Defendant's 4 Exhibit 102, which Mr. Glass questioned you about. Do you 5 still have that in front of you? 6 A. Yes. 7 Q. I want to compare the Ericksen/Kadane article 8 which is the major feature of that exhibit with the actual 9 PES-based adjustment with respect to 1990. Would you please 10 turn to page 106 of the Ericksen/Kadane article. 11 A. All right. 12 Q. Do you see the righthand column there, a series 13 of bulleted paragraphs? 14 A. Yes. 15 Q. About two-thirds of the way down the page? 16 A. Yes. 17 Q. In those paragraphs Ericksen and Kadane propose 18 in their model that, first, "People would only be included 19 in the census count if certain minimal information, a name, 20 and from two to four demographic characteristics were 21 obtained." 22 Do you know whether that principle was followed 23 with respect to the 1990 census and the PES? 24 A. In the census it was not followed. 25 Q. Going down to the next one. "Attempts would be 1381 1 made to count everyone, but these would be limited to the 2 mail out-mail back phase of the census and to some amount of 3 follow-up. Intensive but selective searches for the hard to 4 count people would be made thereafter." 5 Does that apply to the 1990 census? 6 A. No. 7 Q. Next paragraph. "Alternative lists of the 8 population would be constructed for each district office. 9 These would be sampled and matched against the census count 10 to estimate an omission rate." 11 Was that procedure followed for the 1990 census 12 and PES? 13 A. Not at all. 14 Q. The final bulleted paragraph reads, "A sample of 15 the census count would be checked for erroneous inclusions 16 and for the consistent errors in measurement of important 17 characteristics such as race, Hispanic origin, and income." 18 Does that paragraph apply to the 1990 census and 19 PES? 20 A. Yes. The sample was checked for erroneous 21 inclusions. I don't know about the consistent errors in 22 measurement of the characteristics. 23 Q. Dr. Bailar, when you wrote your comment on this 24 article, had you seen the results of the 1985 test census? 25 A. No. 1382 1 Q. Had you seen the results of the 1986 tarrow? 2 A. No. 3 Q. Had you seen the results of any other 1986 test 4 census? 5 A. No. 6 Q. Dr. Bailar, in your cross-examination Mr. Glass 7 asked you about defendants' experts. During the period of 8 time that you served as associate director of the census 9 bureau, were you familiar with the range of opinion among 10 the American statistical community concerning the adjustment 11 issue? 12 A. Yes. 13 Q. During that period of time from 1980, let's bring 14 it up through the present, did Dr. Freedman serve as an 15 expert witness for the government in census-related 16 litigation? 17 A. Yes, in connection with the 1980 litigation. 18 Q. It is your understanding that he continues to 19 serve as an expert witness for the government in this 20 litigation? 21 A. Yes. 22 Q. Did Dr. Wachter serve as an expert witness for 23 the government in census-related litigation concerning the 24 1980 census? 25 A. Yes. 1383 1 Q. Is it your understanding that he continues to 2 serve as an expert witness for the federal government in 3 this litigation? 4 A. Yes. 5 Q. During the 1980s, to your knowledge, did Dr. 6 Breiman obtain any information or guidance from the bureau 7 concerning the area of adjustment research and planning for 8 an adjustment of the 1990 census? 9 MR. GLASS: Objection. It calls for speculation. 10 MR. GOLDIN: To the extent that Dr. Bailar knows. 11 THE COURT: To the extent you know. 12 A. During the time I was at the census bureau, any 13 question that was related to adjustment, anything that 14 related to research, would have been referred to me, and we 15 had no communication with Dr. Breiman. 16 Q. Did you have communication with other 17 statisticians? 18 A. Yes. 19 Q. Approximately how many statisticians would you 20 say you had communications with on that subject during your 21 tenure in the 1980s? 22 A. I'd say somewhere between 50 and 75. 23 Q. Do you know with what faculty Dr. Freedman is 24 associated? 25 A. University of California-Berkeley. 1384 1 Q. Do you know with what faculty Dr. Wachter is 2 associated? 3 A. University of California-Berkeley. 4 Q. Do you know with what faculty Dr. Breiman is 5 associated? 6 A. University of California-Berkeley. 7 Q. Do you know whether there Freedman is being 8 compensated for his work in this litigation? 9 A. Do I know what he is being compensated? 10 Q. Do you know whether he is being compensated? 11 A. Yes, I do. 12 Q. Is he? 13 A. Yes. 14 Q. Do you know whether Dr. Wachter is being 15 compensated? 16 A. Yes, he is it. 17 Q. Do you know whether Dr. Breiman is being 18 compensated? 19 A. Yes. 20 Q. Do you know whether Dr. Meier, before the current 21 phase of this litigation, has been identified by the 22 defendants as an expert witness for the government in 23 connection with census litigation? 24 A. No, he has not been. 25 Q. Do you know whether Dr. Meier is being 1385 1 compensated in connection with his work in this litigation? 2 A. Yes. He is. 3 Q. During the period of time that you were associate 4 director of the bureau, did you receive any inquiries for 5 information or guidance from Dr. Meier concerning the 6 proposed adjustment with the PES? 7 A. I think I had one letter from Paul Meier. He 8 wasn't asking for information. He was telling me that he 9 objected to an adjustment of the census. 10 Q. Did you have any letter from him in which he 11 asked for any information or guidance on the subject? 12 A. No. 13 Q. Other than Dr. Freedman, Dr. Wachter, Dr. 14 Breiman, and Dr. Meier, are you aware of any statistician in 15 the United States who is familiar with the bureau's research 16 in this area and is opposed to an adjustment of the 1990 17 census? 18 A. Yes. 19 Q. Who is that? 20 A. William Kruskal. 21 Q. Anyone else? 22 A. No 23 Q. Do you know whether Dr. Kruskal was one of the 24 Secretary's nominees to the special advisory panel? 25 A. Yes. He was. 1386 1 MR. GOLDIN: I have nothing further. 2 MR. GLASS: Your Honor, a very brief recross. 3 MR. GLASS: Your Honor, I must say that as an 4 alumnus both of the University of California at Berkley and 5 its law school, I am delighted with the Berkeley 6 representation here. 7 RECROSS-EXAMINATION 8 BY MR. GLASS: 9 Q. Let me ask you this, Dr. Bailar. Can you tell me 10 where Bruce Cain, one of the plaintiff's witnesses, hails 11 from? 12 A. No. I don't know him. 13 Q. Let me ask you this question. In the comment 14 that I was questioning you about before, I was going to ask 15 you about some of these assumptions that go to smoothing. 16 There you make the following statement. You ask before the 17 regression analysis can go forward, "Are the underlying 18 assumptions of such a model met? There are several layers 19 of assumptions as pointed out by Freedman (1984) as a 20 witness for the government in the New York lawsuit." 21 That is the same David Freedman we are talking 22 about here that Mr. Goldin just asked you about, isn't that 23 correct? 24 A. Yes. I mentioned that he was in the litigation 25 in 1980. 1387 1 Q. Your reliance on what he wrote here indicates, 2 does it not, that you have no questions about his 3 qualifications as a statistician or his reputability or his 4 esteem in the statistical community? 5 A. I have no questions about his reputation in the 6 statistical community. I was answering a question that he 7 never asked for any information from the census bureau about 8 these matters in between lawsuits. 9 Q. Let me just ask you one or two brief questions 10 about the statistical adjustment that you say took place of 11 the 1970 decennial census. It is a fact, is it not, that 12 that was not an adjustment that was intended to deal with 13 the differential undercount? 14 A. It was an adjustment of the census for the 15 undercount. Whether the differential was part of it or not, 16 no one knew at that time. 17 Q. But the purpose of it, the express purpose, was 18 not to eliminate the differential undercount as we have been 19 addressing it in this lawsuit? 20 A. It was to eliminate the undercount. 21 Q. It is a fact, is it not, that simply in terms of 22 raw numbers of people added through the imputation in 1970, 23 that we are dealing with a much smaller number than we are 24 at the present time, isn't that correct? 25 A. It's a smaller number. 1388 1 Q. It is on the order of a million as opposed to 2 your testimony it was between four and five million, isn't 3 that correct? 4 A. Yes. But I don't understand why you are so 5 concerned about the size of the number, because it is the 6 methodology that counts and not the number. 7 MR. GLASS: I have nothing further of this 8 witness. 9 MR. GOLDIN: Nothing further. 10 THE COURT: Thank you Dr. Bailar. You have been 11 very enlightening. We excuse you. 12 (Witness excused) 13 THE COURT: We will recess for the day. We will 14 resume tomorrow at 9:30. 15 MR. MILLET: Are we in any danger of the 16 plaintiffs resting tomorrow, so we can make our travel 17 plans? 18 THE COURT: How are you doing, Mr. Zimroth and 19 Mr. Rifkin? 20 MR. ZIMROTH: I think it is unlikely. If it is 21 at all possible, it will be towards the very end of the day. 22 But more likely it will be Thursday. 23 THE COURT: Have a nice evening. We will see you 24 tomorrow at 9:30. 25 (Adjourned to 9:30 a.m., May 20, 1992)