Carey v. Klutznick, 508 F. Supp. 420 (S.D.N.Y. 1980)

*421 Cravath, Swaine & Moore, New York City, for all plaintiffs except the State of New York and Hugh L. Carey; Frederick A. O. Schwarz, Jr., Robert A. Rifkind, David A. Barrett, Roger D. Einerson, John A. Redmon, Michael J. Malone, Roger H. Cummings, Mark P. Schnapp, Ronald P. Mysliwiec, New York City, of counsel.

Robert Abrams, Atty. Gen. of the State of New York, New York City, Peter Bienstock, Shiela Abdus-Salaam, Daniel Berger, Asst. Attys. Gen., New York City, of counsel, for the State of New York and the Hon. Hugh L. Carey.

Allen G. Schwartz, Corp. Counsel of the City of New York, New York City, Mary McCorry, New York City, of counsel, for the City of New York and the Hon. Edward I. Koch.

John S. Martin, Jr., U. S. Atty., S. D. New York, New York City, Michael H. Dolinger, Jane E. Booth, Steven E. Obus, Thomas D. Warren, Gaines Gwathmey, III, Susan M. Campbell, Asst. U. S. Attys., New York City, of counsel, for defendants.

Stay Granted December 30, 1980. See 101 S. Ct. 799.

WERKER, District Judge.

Plaintiffs^[1] commenced this suit against several defendants including the Bureau of the Census^[2] to obtain declaratory and permanent injunctive relief requiring the defendants to adjust census figures for New York City and New York State in a manner *422 that would reflect more accurately their true populations. Plaintiffs seek this relief on the ground that Census Bureau mismanagement of the census in New York further exacerbated what normally is a substantial undercount of the New York population. This action having been tried on the merits before the Court on November 12-14 and 17-21, 1980, and all the evidence having been duly considered, judgment is rendered for plaintiffs. Pursuant to Fed.R.Civ.P. 52(a), my findings of fact and conclusions of law follow.

The budget for the 1980 decennial census was approximately $1,008,000,000 for the full ten-year period preceding the census and approximately $670,000,000 for 1980. Tr. at 1116. The methodology of the 1980 census is set forth in numerous Census Bureau manuals. See Defendant's Ex. 22. Briefly, the census was to be carried out by a mail-out, mail-back procedure. Respondents were to provide the Census Bureau with information concerning the number of persons within their household. Several follow-up procedures designed to detect persons and households who were missed in the original count either because a form was never sent to them or because they failed to return a form also were planned. Finally, there were to be certain last resort techniques and other imputations made by the Census Bureau to complete the count. In order to accomplish the proposed procedures, approximately 270,000 persons were temporarily employed by the Bureau across the nation to supplement the staff of several thousand permanent Bureau employees. Tr. at 1115.

In large cities, mailing lists were compiled from address lists purchased from commercial mailing firms. In outlying areas the Bureau utilized a pre-list procedure whereby enumerators went out to the field and listed every housing unit in a given area. Tr. at 1128-29. It was anticipated that the commercial mailing lists would be incomplete so the Bureau designed procedures to improve the quality of the lists. There were to be three Post Office checks and a precanvass check by enumerators. Id. at 1129. "After the [Advance] Post Office check, all of the postal corrections were [to be added] to the list, and the list was then [to be] geocoded ...."^[3]Id. at 1131.

Based on the address lists so compiled, questionnaires were to be mailed out to every listed household in late March 1980. Id. at 1137. Every questionnaire that was returned to the local district office was to be matched to the Master Address Register ("MAR"). Deficiencies identified in the MAR's with respect to number of units reported in the structure were to be assigned for recanvass as were questionnaires that appeared incorrect because of the manner in which certain questions were answered. The recanvass operation undertaken at this stage was known as Follow-up I. Essentially, Follow-up I was to consist of series of procedures for field follow-up by enumerators to provide for enumeration of those households that did not receive or return a census questionnaire in the mail and for recanvass of those units which had "failed edit."^[4]Id. at 1140.

Follow-up I was to commence in mid-April 1980. Questionnaires were to be filled out for any units successfully enumerated in the course of these procedures and the information obtained was to be included in the MAR's. If after four attempts, an *423 enumerator was unable to obtain information concerning a unit, the enumerator was to attempt to obtain that information from persons with knowledge of the status of the unit such as neighbors or janitors. This was known as last resort information. If this procedure still resulted in a dearth of information concerning the unit, the crew leader of the district office was to attempt to enumerate the unit. If this effort similarly proved unproductive, the unit was to be classified as a verified refusal and no further attempt was to be made to enumerate the unit. Id. at 1142-43. At the end of Follow-up I there was to be a merge of the MAR's and the Follow-up Address Registers ("FAR's") which were kept in the district offices after which the information in each book was to be identical. There were target dates for the completion of Follow-up I but they were not deadlines and the procedure was to continue in each district office until "every housing unit had been enumerated or had been classified as vacant, found to be no longer existent ... or where ... visits had been made and no information had been obtained, a so-called unclassified unit." Id. at 1146.

At the end of Follow-up I, preliminary figures as to number of units, vacant units and population were to be disclosed to local governmental units for a voluntary procedure known as "local review." The purpose of the program was to permit local officials to compare census figures with their own population and housing figures in order to identify possible problem areas and present challenges to Census Bureau numbers. So long as the local census office was still open and was in a position to investigate the challenges, it was to review the challenges and recanvass the challenged areas if necessary. Id. at 1147-51.

The next step in the census operation was known as Follow-up II. This phase of the enumeration involved a recheck of unclassified units, units classified as vacant or non-existent, and units that had "failed edit." Id. at 1152-54. A non-household source check was to be conducted in large cities to improve coverage of minority populations. Id. at 1252. In this procedure names and addresses from Motor Vehicle driver's license lists and of legal aliens from the Immigration and Naturalization Service were to be compared against census data received during Follow-up I. In New York City, names and addresses from public assistance files were also to be used in the non-household source check. Id. at 1154-56. Follow-up II "was supposed to end when the procedures were all completed. However, the census activities were running behind schedule around the country so that a firm time schedule was developed for each district office, depending on the status of the census in that particular office ...." The closing dates were not adhered to in many cases. "[D]istrict offices were told that in order to reach this closing date, that they could truncate some activities, but they were specifically told they could not truncate any activity which had to do with coverage and, therefore, if some activity was still going on that had not been or could not be completed by that time, the office was kept open until such time as it could be completed, unless it was of a relatively small nature, and if so, the regional census office was asked to complete the activity." Id. at 1157.

If at any time after April 1, 1980, unlisted housing units were discovered during the course of any field follow-up operations, the procedures called for those units to be added to the MAR's and enumerated. Tr. 217-18; D's exhibit 22.

At the end of field operations, all of the questionnaires and address registers were to be sent for processing to Jeffersonville, Indiana. At this point, the final population totals were to be tabulated from the questionnaires, and population and other characteristics of any units that remained unclassified were to be imputed based upon the characteristics of the immediately preceding unit. Tr. 1163-64.

Despite the rather detailed procedures outlined above, the district offices often experienced problems in enumerating the *424 population. The problems were of several different types. Some recurred frequently, thus enabling the court to draw the inference that these or similar problems were encountered throughout New York City and other large cities within the state as well as in some less populous areas of the state.

The first identifiable problem concerns the manner in which the MAR's were compiled for parts of New York City. As discussed above, the MAR's for urban areas were compiled from commercial mailing lists. Use of commercial mailing lists presented several sources of error. First, the commercial lists were incomplete and contained inaccuracies. Plaintiff's Ex. 334 at 134. This rendered "[t]he overall coverage provided by the purchased lists for the Nation's four largest urban areas" including New York at only about 59 percent. Id. at 136. The commercial lists frequently lacked apartment unit designations. This absence posed a serious problem in urban areas because "[t]he 1970 census data showed that approximately 50 percent of the people in central cities of major urban areas live in multiunit structures." Id.

The Post Office review procedure was insufficient to correct many of the defects in the address lists. "Bureau experience has shown that the quality of Postal Service reviews is related to the quality of the address list provided.... [T]he Postal Service can be expected to improve a list by only about 25 percent." Id. The other procedure designed to improve the MAR's compiled from commercial lists, the precanvass by enumerators was the only quality control check for the Postal Service review. Tr. at 1197. This was also of limited effectiveness for several reasons. First, "[t]ime to accomplish the precanvass [was] very limited because" its commencement was delayed. Instead of commencing in the third week of January 1980, precanvass began on February 21, 1980. Plaintiff's Ex. 334 at 139. Other problems experienced in this procedure included delay in receipt of address registers, grossly deficient address registers, illegible, incomplete and inaccurate maps, and lack of trained enumerators due to the reversal of the order of several procedures. Tr. at 915, 265, 268-270, 273. This delay reduced the time available for correcting the MAR's and "some quality controls that could help the pre-canvass operation, such as the Postal Service's review of precanvass corrections and monitoring, did not take place." Furthermore, "time [was] very limited before census day to react to problems with the precanvass operation." Plaintiff's Ex. 334 at 139-40. Finally, the pre-census local review of address lists was cancelled by the Census Bureau in at least 25,000 local governmental units including the City of Rochester, New York because of delays in completing other preparatory procedures. As a consequence, local help in finding omissions in the address lists prior to the mailing of questionnaires was bypassed. Plaintiff's Ex. 14 at 8; Tr. at 587.

Problems were also experienced with the mail-out, mail-back procedure. Despite several Postal Service efforts to identify additional units that were not on the address listings, the Census Bureau nevertheless failed to update their lists. This required duplicative effort on the part of the Postal Service and when the questionnaires were mailed out at the end of March 1980, the Census Bureau still did not provide forms for 78,000 addresses in the Bronx, Brooklyn, Manhattan and Staten Island. Plaintiff's Ex. 15 at 84-86. The Postal Service continued to deliver forms it received from the Census Bureau through mid-April, 1980. Id. Instead of placing the forms in the appropriate mailbox, many forms were simply left in apartment building lobbies creating innumerable problems from a failure to return forms at all to returning forms for units other than those for which they were intended. Tr. at 876. Id. at 90-92. Problems were also encountered with respect to Spanish language questionnaires. All forms originally delivered by the Census Bureau were in the English language. If a respondent desired a Spanish questionnaire, he was to call a telephone number listed on the front of the English form and a Spanish language form would be mailed to him. In one district office in the Bronx, the local *425 post office returned these Spanish forms because the carriers were under the impression that they were duplicates. Id. at 87-89. Finally, no quality control was exercised by the Regional Census Center ("RCC") or national headquarters over Postal Service delivery of the questionnaires. Tr. at 1028.

Another problem encountered with the procedures developed by the Census Bureau was in the area of advertising. The Bureau chose not to utilize paid advertising in the 1980 census and relied instead on public service announcements and free advertising for heightening public awareness to census operations. Tr. at 1055. Despite complaints from the minority community and others that the minority media was less able to provide pro bono advertising, the Acting Assistant Regional Director of the Census for the New York Region, Richard Bitzer, nevertheless failed to consider whether the claims were accurate. Id. at 1056. Furthermore, in addition to limited advertising with respect to the April 1, 1980 mail-out, there was little advertising for subsequent follow-up procedures such as the Harlem recount which took place in the autumn of 1980 and the "Were You Counted Campaign." The lack of publicity for the Harlem recount was attributed to concern that "the Census Bureau is involved in a lawsuit with the City, and if we advertise that we have to get all this work done, that it is not finished, it is going to make us look like we are wrong." Tr. at 82-83, 875.

The overall mail return rate for New York City was 72%. In parts of the city the return rate was considerably lower, decreasing to only 48% in Bedford-Stuyvesant.

Another problem experienced in carrying out the 1980 Census occurred in staffing at both management and enumerator levels. First, it was not a prerequisite that district office managers in the New York Region have prior supervisory experience or prior experience in census-taking in the field. Tr. at 1029. Thus, supervisory staff including some district office managers in decentralized offices^[5] and crew leaders lacked experience in supervising large groups. They were not given adequate training by the Census Bureau with respect to the special problems of enumerating the population in New York. Tr. at 639-640. This factor and factors such as severe time and budget pressures imposed by the National and Regional headquarters caused supervisors to be unresponsive to problems made known to them by their staffs. See Tr. at 66-67, 106-13, 184, 541, 649-50. Crew leaders and enumerators frequently were unconcerned about the necessity of carrying out procedures in the proper manner, Tr. at 289-290, 482-84, and enumerators were often untrained in certain difficult procedures such as Mission Night.^[6] Tr. at 642, 681-82. The final problem and a pervasive one was that of understaffing and high turnover rates in temporary employees. Staffing levels contemplated by the Census Bureau for field operations rarely were achieved. Tr. 270, 555-57, 687, 1011-12; Plaintiff's Ex. 15 at 146-47. High personnel turnover occurred at both the enumerator and supervisory level. Plaintiffs Ex. 15 at 116, 129; Plaintiff's Ex. 71.

Another problem previously alluded to is that of Census Bureau deadlines. Whether due to time or budget constraints, there was a definite pattern of pressure on enumerators, crew leaders, and district office managers to finish the job without regard to the quality of the count that would be rendered. See, e. g., Tr. at 304-07, 663, 686-87, 699-700. Closing dates were scheduled by the National headquarters for all *426 district offices apparently without regard for the status of operations in the individual offices. Tr. at 1239-41. District office managers were told that they would be jointly responsible for costs incurred by remaining open beyond the identified closing dates. Plaintiff's Ex. 211. When some district office managers resisted pressure to close because there was still a substantial amount of work to be done, they were overridden by RCC supervisors. See, e. g., Tr. at 336-37. These pressures caused enumerators and crew leaders to falsify work, classify units as vacant, and make up answers on questionnaires. See, e. g., Tr. at 64-66, 80-82, 472-74, 522-23, 530-31.

As a result of time pressure, Spanish questionnaires were not mailed out to those who had requested them, responses in the Spanish language forms were not transcribed to English forms in order to permit them to be processed by Census Bureau computers, damaged forms were not transcribed, and the MAR's and FAR's were not reconciled. Tr. at 202-05, 249-51; Plaintiff' Ex. 260. In addition, there were insufficient bilingual enumerators during follow-up operations. Tr. at 95-96.

The Census Bureau's decision to pay enumerators at a piece rate rather than an hourly rate was one of the causes of the high turnover rate among enumerators. Plaintiff's Ex. 294; Plaintiff's Ex. 315 at 13-14.

It also appears that quality control procedures failed to identify many of the foregoing problems and that certain procedures were never utilized or were curtailed. See, e. g., Tr. at 651-55. There were several instances in which the Census Bureau was notified that people had not been counted but who were not followed up by the Bureau and were left uncounted. Tr. at 141-56, 174-76.

The effectiveness of the non-household source check in New York City was severely inhibited by the Census Bureau's decision to eliminate 428,210 addresses which represented the addresses of persons under the age of 16 years, a segment of the population that had a high undercount rate in 1970. Tr. at 1208-12.

Finally, the local review procedure was not fully implemented. The Census Bureau did not respond to many local review challenges, Tr. 592-95, and provided insufficient information to allow local governments to identify all missed units. Furthermore, in some districts, Census Bureau review of local challenges was carried out simultaneously with other important operations resulting in insufficient staff for the multiple operations. Tr. at 320.

The decennial census attempts to count the entire population of the United States. The census, however, is not a headcount in which each and every person residing in the United States on a given date is counted by the Census Bureau. Rather, it is a survey of the population that through the responses of one member of each household attempts to enumerate the entire population. This method concededly fails to accomplish this, and the resultant underenumeration in the past has been compensated for by imputing population and their characteristics to the results achieved by survey enumeration. See Plaintiff's Ex. 29 at 16-18.

As alluded to above, undercounts have been experienced in prior decennial censuses. "[T]he rate of undercounting was 2.7% for the 1960 Census (8.0% Blacks and 2.0% for whites) and 2.5% for the 1970 Census (7.7% for Blacks and 1.9% for whites). Thus, "undercounts were disproportionately large in the 1950, 1960 and 1970 Censuses for certain minority groups compared to the population at large. The Census Bureau has found that disproportionate undercounts existed for Blacks in the 1950, 1960 and 1970 Censuses and for Hispanics in the 1970 Census." Pre-Trial Order, Stipulated Facts Nos. 1-2.

"[M]ost socially and economically disadvantaged minority groups, including Blacks, Hispanics, other minorities and aliens living lawfully in the United States [are] undercounted to a higher degree than the national average ... because members of such *427 groups have many of the same sociological characteristics that lead to undercounting of Blacks." Plaintiff's Ex. 235 at ¶ 5. These characteristics include:

1. poverty, lack of education and poor language abilities;

2. irregular living arrangements such as several families sharing one living unit; shifting of child-care among several distantly related females; single men with no clearly fixed place of residence; living with a group related by kin or common place of origin; taking in household boarders;

3. a fear that revealing information about family or group members may jeopardize eligibility for government income-transfer programs, may provoke a landlord, may violate health or zoning codes; and

4. general distrust to (sic) all government operations.

Id. at ¶ 6. Illegal aliens are also affected by the foregoing characteristics and in particular fear discovery of their unlawful status by the government. Id. at ¶ 10; see Tr. at 836-49.

A disproportionate number of the classes of persons that have been more severely undercounted in the past resided in New York City in 1970 as compared to the nation as a whole. Pre-Trial Order, Stipulated Fact No. 3. "[A]ccording to the 1970 census, 11.1 percent of the total population of the United States was Black and 4.6 was of Spanish Origin. In New York City, 21 percent of the population was Black and 15 percent of Spanish Origin. The New York State population was about 12 percent Black and 7 percent Spanish Origin." Plaintiff's Ex. 235 at ¶ 8. The Census Bureau was aware before the 1980 Census that New York City contained "hard to enumerate" areas. Pre-Trial Order, Stipulated Fact No. 9. Furthermore, the Bureau knew that "[d]uring the decade of the seventies, certain demographic changes ... occurred which would make taking the 1980 census inherently more difficult than taking the 1970 census. In particular, a larger proportion of the population in 1980 will be Black and a larger proportion will fall in the young adult ages, groups which were more difficult to count in 1970." Plaintiff's exhibit 327 at 1. Since New York City and New York State contain disproportionate numbers of the groups mentioned above as well as other groups which are inherently difficult to enumerate, it is reasonable to conclude and I do so conclude that taking the census in New York in 1980 was more difficult than taking the 1970 census.

Based on the above, the Census Bureau undertook to utilize several procedures including the non-household sources check and the local review program for the first time for the purpose of improving Census Bureau coverage of the entire population. Plaintiff's Ex. 327 at 1; Tr. at 1252-54.

Preliminary estimates from the Census Bureau indicate a national "measured undercount" of only 0.4 percent of the population. Affid. of Jacob Siegel, sworn to Nov. 17, 1980 at ¶ 2. The "measured undercount" a term of recent vintage, see Plaintiffs' Memo on Recordings of October 16-17, 1980 Conference at Census Bureau at 17, is determined by subtracting the gross overcount from the gross undercount. The Census Bureau nevertheless concedes that there must have been some net undercount in 1980 and that it probably is about one or two percent of the total population, more or less. Tr. at 1396-97. Moreover, while the national gross undercount and gross overcount may cancel each other out there is no evidence that this will occur at the sub-national level. Rather, the evidence indicates that the incidence of an overcount and an undercount in the same geographic area is unlikely because the factors which produce an overcount are different from those that cause an undercount. Plaintiff's Ex. 193 at ¶¶ 46-47. Furthermore, the Bureau's estimates with respect to the "measured undercount" are speculative because they are based on a comparison of up-dated 1970 population figures for individual places and the preliminary 1980 counts for those places. Finally, the "measured undercount" figure does not include the illegal alien population *428 which may number as many as 5 million persons. Id. at 1362-63.

While the "measured undercount" for the nation has been reduced to almost zero by the latest estimates, the Census Bureau has no evidence that there has been a decrease in the differential undercoverage of the groups that traditionally are "hard to enumerate." Plaintiff's Memo on Recordings of October 16-17 at 3; Tr. at 1374. There is, however, evidence from independent surveys that the populations of New York State and New York City which contain relatively large numbers of the "hard to enumerate" were undercounted in 1980 in differentially high proportions to that of the nation as a whole.

Conservative estimates of the undercount in 1980 for New York City ranged from 517,000 to 650,000 and for New York State ranged from 772,000 to 905,000. Plaintiff's Ex. 235 at ¶ 19. The results of a random phone survey of New York City and New York State residents conducted between August 20, 1980 and September 11, 1980 by the independent public affairs research organization, Penn & Schoen, indicate that 6% of the population of New York State and 8% of the population of New York City have not been counted by the Census Bureau. Plaintiff's Ex. 269 at ¶ 4. While the defendant's have challenged the reliability of this survey I find that it is entitled to some weight in demonstrating the magnitude of the undercount in New York. But see Defendant's Ex. 52. Other evidence of the undercount in New York are the results of "Were You Counted Campaigns" in New York City and Rochester. The results of the New York City "Were You Counted" campaigns show that 18,109 or 11.2% of the 161,718 applicants or recipients interviewed by 41 Income Maintenance Centers, 8 recertification offices and 16 non-public assistance Food Stamp Offices of the New York City Human Resources Administration had not been counted. A mail survey conducted of all 302,000 households in the New York City Income Maintenance Program indicated that of the 281,130 households that responded, 217,325 stated that they had returned a census form or were visited by a census taker, 41,175 did not answer the question and 22,720 or 8% of the respondents stated that they did not return a census form or were not visited by a census enumerator.

Defendants have submitted a preliminary report on the processing of "Were You Counted" forms for New York City. See Affid. of William Starr, sworn to Dec. 1, 1980. The results indicate that 9,796 or approximately two-thirds of the 15,158 individuals who claimed they were not counted lived in households from which the Census Bureau had already received a mail questionnaire or a field return questionnaire showing an equal or greater number of persons. See Ex. B, lines 4 and 5. Of the remaining one-third or 5,362 persons, 83.04 percent of the individuals contacted by the Census Bureau were found to have been missed by the census. Of 2,123 persons contacted by the Bureau 1,763 were added to existing census figures. If the 3,239 individuals who were not contacted by the Census Bureau are deemed to have been missed at the same rate as persons already reached by the Bureau, this would result in the addition of 2,688 (83% of 3,239) individuals to existing census counts. The total figures would then reveal that of 15,158 individuals who claimed they had not been counted, 4,451 or 29.4% were in fact not counted. Of course, many more individuals have submitted "Were You Counted" forms in New York City and approximately 30 percent of that population were most likely missed by the Bureau.

In the City of Rochester "Were You Counted Campaign," 7 to 10 percent of those contacted stated that they had not been enumerated. Tr. 600, 602. Of those who claimed they had been missed by the Census Bureau, approximately 60% were verified by the Census Bureau as not having been enumerated. Tr. at 600-607; Plaintiff's Ex. 320.

In response to the order of the District Court for the Eastern District of Michigan,^[7] the Bureau has prepared a report on the feasibility of undertaking a "statistically defensible"^[8] method of adjusting population figures from the 1980 census. The Bureau's conclusion as stated in its report which has been received in evidence in this case is that at this time it is unable to develop a "statistically defensible" method of measuring the national undercount, if any, in 1980. Defendant's Ex. 23 at 3. According to the Bureau, "[s]ince the present measurement of the undercount is unreliable even at the national level, no method of adjusting for the undercount is statistically defensible for 1980." Id. at 35. The Bureau asserts, however, that "if good undercount data were available, some ... methods [of adjustment] would be defensible at the national level, some at the State and standard metropolitan statistical area ("SMSA") level, but none would be defensible at lower levels of geography." Id. The Bureau nevertheless submitted a proposal for adjusting census data in the Detroit case should the court order the Bureau to adjust.

The principal issues before the Court are (1) whether there was a disproportionate undercount of the population of New York City and New York State in the 1980 census; (2) whether the Census Bureau has mismanaged the 1980 census either in the manner in which it was planned or in execution or both; (3) whether mismanagement, if any, contributed to the undercount of the population of New York; (4) whether there are methods of adjusting the 1980 census that will produce more accurate population counts for New York than the unadjusted figures; (5) whether defendants are required to utilize a reasonable and scientifically valid statistical, survey or sampling method to adjust for the undercount in the population tabulations reported to the President pursuant to 13 U.S.C. § 141(b); (6) whether the Bureau must report census figures to the President by the December 31, 1980 deadline even though those figures are not adjusted for the disproportionate undercount.^[9]

Plaintiff's claims as to mismanagement are directed to the plans and methodology developed by the Census Bureau for carrying out the 1980 census as *430 well as to the manner in which the census was conducted. In reviewing the methodology devised by the Census Bureau and the manner in which the decennial census was carried out, the Court is limited to determining whether the administrative decision was arbitrary, capricious, or an abuse of discretion. Carey v. Klutznick, 637 F.2d 834 (2d Cir., 1980); see Dunlop v. Bachowski, 421 U.S. 560, 567, 95 S. Ct. 1851, 1857, 44 L. Ed. 2d 377 (1975); Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 410, 91 S. Ct. 814, 820, 28 L. Ed. 2d 136 (1971); Borough of Bethel Park v. Stans, 449 F.2d 575 (3d Cir. 1971); Young v. Klutznick, 497 F. Supp. 1318 (E.D.Mich., 1980); City of Camden v. Plotkin, 466 F. Supp. 44, 51-53 (D.N.J.1978); West End Neighborhood Corp. v. Stans, 312 F. Supp. 1066 (D.D.C. 1970); Quon v. Stans, 309 F. Supp. 604 (N.D. Cal.1970). Bearing in mind that the Census Bureau's decisions enjoy a presumption of regularity and that the court may not substitute its judgment for that of the Bureau, the court's focus must be on the "reasonableness of or rational basis for" the methodology of the 1980 census and the manner in which it was executed. City of Camden v. Plotkin, 466 F. Supp. at 52-53. Thus, the court must assess whether the defendants acted within the scope of the authority granted to them by Congress, whether the decision was premised on a "`consideration of relevant factors and whether there has been a clear error of judgment.'" Id. at 53 (quoting Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. at 416, 91 S. Ct. at 823.)

The record clearly demonstrates that the Census Bureau did not abuse its discretion, or act in an arbitrary or capricious fashion with respect to the procedures planned for the 1980 census. There is no doubt that the Bureau acted within the scope of the authority granted to it by Congress in devising its plans for the 1980 decennial census. The procedures were designed in good faith for the purpose of enumerating the whole number of persons in each state as of April 1, 1980 with results to be made available to the President by December 31, 1980 for the purpose of apportionment. See 13 U.S.C. § 141. Furthermore, it is clear that the plans were developed only after a careful review of the many factors that must be considered in attempting a task as mammoth as counting the population of the United States. The Bureau's decisions were based in large measure on prior experience with the task. While the underlying plan for 1980 was quite similar to that utilized in prior censuses, the Bureau expanded the use of certain procedures such as the mailed questionnaire that had proven very effective in the 1970 census. It also developed several innovative procedures designed to improve coverage of the entire population such as the local review program and it planned procedures for enumerating the minority populations who from prior experience generally were differentially undercounted. Although there may be certain areas in which the Census Bureau's plans could have been improved further, the Court finds that the overall methodology for the 1980 decennial census was reasonable. As is evident from my earlier discussion of the procedures employed by the Bureau the plan which was set forth in numerous Census Bureau manuals was detailed, was realistic in its procedures for checking and rechecking, was reasonable as to the planned timing and sequence of procedures, and was rational with respect to the programs devised for coverage improvement. As stated by Dr. Hauser, the Census Bureau "has done very well in the methods they have adopted within the [budgetary constraints with which they ... operate." Plaintiffs Ex. 317, part III at 12-13.

Addressing next the implementation of the foregoing methodology in New York, the Court finds that the Bureau's conduct was on the whole unreasonable and at times irrational. First was the failure to commence census operations in a timely fashion in New York. Second was the use of grossly deficient Master Address Registers for certain areas of New York City, the use of poorly trained precanvass enumerators to complete the address registers, a task crucial in procuring an accurate count of the *431 population, the use of out-dated maps or the failure to provide maps, and the failure to correct the address registers following the advance Post Office checks.

Thus, from the outset, there were numerous instances in which Bureau procedures were cancelled, curtailed, or carried out in a shoddy manner because of time pressures. The more procedures were abbreviated because of time pressures, the more subsequent follow-up procedures became overburdened to the point where offices were closed with a substantial amount of follow-up work still to be done. Near chaos in some district offices was the ultimate result of the failure to permit the office to complete the procedures originally planned.

When problems at the district office level were identified to regional headquarters there often was no response or if there was a response, it frequently was a direction to just get the job done regardless of its quality. For example, the Bureau failed to respond to objections that the piece-rate pay structure did not permit many enumerators to earn a salary reflective of their effort and productivity in areas where enumeration was difficult. The failure to adopt an hourly-rate pay structure was a direct cause of the high turnover rate among enumerators as well as the type of falsification of questionnaires known as curbstoning as well as other types of fraud. Regional headquarters also failed to respond to requests for increased advertising of census operations generally and for paid advertising in minority news media despite the Bureau's awareness that special efforts were needed to avoid undercounting minority communities. It is clear that such conduct by the Bureau was unreasonable in light of the low mail return rates experienced in New York City.

The time pressures constantly imposed upon enumerators, crew leaders and district office managers also were unreasonable insofar as the Bureau knew that enumeration could not be completed within the time constraints placed upon personnel.

Another aspect of the Bureau's irrational conduct in implementing the plans for the 1980 census is the manner in which definitions were changed on a constant basis, making it impossible for enumerators to work with viable standards to guide them. Similar to this is the Bureau's conflicting directives concerning the point at which an office could close down with respect to the number of cases still outstanding. In some offices the figures changed from 2% to 5%, and in other offices a higher rate was permissible.

Understaffing at the enumerator level was pervasive in New York City. The Bureau nevertheless failed to take corrective action such as altering its pay scale in order to retain the personnel it did have and attract new personnel.

The lack of adequate training for enumerators involved in complex procedures, and a lack of prior supervisory experience in crew leaders and district office managers also was a pervasive problem. The Bureau, however, seemed unconcerned about the impact these factors would have on the quality of the resultant count.

As has been noted above, an undercount of the population occurred in the 1950, 1960, and 1970 censuses. The undercount in the past has been disproportionate as to Blacks and Hispanics. There is reason to believe that aliens, legal and especially illegal, are also undercounted in a disproportionate manner. The national undercount rate in 1970 was 2.5 percent. An estimate of the undercount for New York in 1970 is that 4.2% of the New York City population and 3.1 percent of the State population was not counted. The rates at which blacks and other minorities were undercounted in 1970 in New York State and City were disproportionately high as compared to the white population. See Plaintiff's Ex. 235 at ¶¶ 14-15. It is undisputed that a large portion of the New York City population consists of blacks, hispanics, other minorities, and legal and illegal aliens. Because of this, New York City is an area that from past experience is difficult to enumerate. Demographic changes during the past decade have exacerbated the difficulty *432 of taking the census in 1980. The inescapable conclusion from the foregoing is that unless the Census Bureau made extraordinary efforts to count the New York City population in 1980, there again would be a disproportionate undercount in New York City and in New York State.

Statistical evidence submitted by the parties confirms this conclusion. While defendant's claim that the national "measured undercount" approaches zero, independent studies, estimates, and preliminary results of the "Were You Counted" campaign reveal an undercount of about five or six percent for New York State and of about seven or eight percent in New York City. While these studies are not conclusive of the extent of the undercount in New York, common sense tells us that these estimates are more realistic than the Census Bureau's conclusion that no disproportionate undercount has occurred in this state and city, and that the "measured undercount" for the nation is about zero, a conclusion that even the Bureau believes is unreliable. Consequently, this Court finds that there has been a disproportionate undercount of the population of New York State and New York City as compared to the nation as a whole in the 1980 census.

Despite their knowledge that extraordinary efforts were necessary to count the population of New York City because it is difficult to enumerate, the Bureau as discussed above nevertheless failed to exert that effort in a manner that was reasonably calculated to overcome the inherent problems associated with census-taking in New York. It is clear that the unreasonable and sometimes irrational decisions made at national and regional headquarters further contributed to the underenumeration of the New York population in 1980.

The full extent of underenumeration in New York is unknown. The factors contributing to this lack of Bureau knowledge have been recited above and will not be repeated. Because of their failure to produce MAR's, FAR's, and vacant housing unit reports, the defendants were precluded from introducing evidence which may have permitted the court to determine what the true numbers for New York are. Because they introduced no evidence to rebut the plaintiff's showing of an undercount in New York that is disproportionate to the national average, they have failed to rebut the prima facie case established by plaintiffs.

While the Census Bureau has taken the position that it is unable to develop "statistically defensible" methods of adjustment, the Court finds that what is required here is not a scientifically perfect method for adjusting but a method that will reflect the true population of the city and the state more accurately than the unadjusted counts. The Court recognizes the difficulties posed by the task they must undertake. The task nevertheless must be accomplished.

The Bureau had planned to use three procedures to evaluate the accuracy of the 1980 census. They are demographic analysis, and the Post Enumeration Program, the CPS-IRS Match Study. The demographic method of analyzing the undercount provides estimates of net underenumeration for the total population and estimates of net census error for age, sex, and race groups. This data can be available by as early as January 1981. The Post Enumeration Program will provide estimates of the undercount for states, regions and SMSA's. This data can be available in September of 1981. The final procedure is the CPS-IRS Match. This procedure involves a comparison of Current Population Survey data with Internal Revenue Service records and provides estimates of differences in coverage for Blacks, Hispanics and all other groups. The results of this procedure can be available as early as February 1981. While the bureau has stated that the data produced by these studies is useful only for evaluation and not for adjustment, it appears to the Court that these techniques can and must be utilized by the Bureau to determine the extent of the undercount at national and sub-national levels so that adjustments may be made where necessary to *433 increase the accuracy of the unadjusted counts. This action is required to prevent an unwarranted dilution of the votes of New York City and New York State residents. The Constitution requires no less. See Carey v. Klutznick, 637 F.2d 834 at 838 (2d Cir. 1980); accord, Young v. Klutznick, 497 F. Supp. 1318 at 1333 (E.D.Mich. 1980).

The final issue that must be addressed is that of whether the statutory deadline of December 31, 1980 for the Bureau to report population figures to the President must be observed. See 13 U.S.C. § 141(b). Plaintiffs assert that the provision is unconstitutional insofar as it would result in the use of inaccurate population figures for apportionment. I do not find it necessary to reach that issue because the question can be decided as a matter of statutory construction. See Ashwander v. Tennessee Valley Authority, 297 U.S. 288, 346-48, 56 S. Ct. 466, 482-483, 80 L. Ed. 688 (1936) (Brandeis, J., concurring).

Statutory time limits are either discretionary or mandatory. A time limit is not mandatory unless it both requires action within a specified time period and prescribes a consequence for failure to act within that time. See Fort Worth National Corp. v. Federal Savings & Loan Insurance Corp., 469 F.2d 47, 58 (5th Cir. 1972); Diamond Match Co. v. United States, 181 F. Supp. 952, 958-59 (U.S. Customs Ct. 1960). It is apparent from the face of the statute that no sanctions are imposed if the Dec. 31 deadline is not met. It also is clear that the date set by the statute was provided for the purpose of orderly administration of the reapportionment process. To require the Bureau to report patently uncertain population figures to the President merely to meet the deadline, would defeat this goal. See Markham v. Cabell, 326 U.S. 404, 409, 66 S. Ct. 193, 195, 90 L. Ed. 165 (1945); Sorrells v. United States, 287 U.S. 435, 446, 53 S. Ct. 210, 214, 77 L. Ed. 413 (1932). Consequently, it is clear that the December 31 date is directory, not mandatory. See Carey v. Klutznick, 637 F.2d 834 at 837 (2d Cir. 1980).

In accordance with the above, I hold that there has been an undercount of the population of New York City and New York State that is disproportionate to the national average and that Census Bureau mismanagement in the implementation of the plan for the decennial census contributed to the undercount. I also hold that there are statistical methods available to the Bureau that will produce more accurate census data than the unadjusted counts and that the Bureau is required to utilize these methods to implement an adjustment. Finally, I find that the Bureau is not required to submit population totals to the President by December 31, 1980.

Judgment is rendered in favor of the plaintiffs:

1. Restraining the defendants from certifying population totals to the President on December 31, 1980;

2. Requiring the defendants to adjust population figures for 1980 for New York State and its subdivisions and New York City in particular in a reasonable and scientific manner to compensate for the disproportionate undercount;

3. Declaring that the use of reasonable and scientifically valid statistical, survey or sampling procedures to adjust census figures for the differential undercount is constitutionally permissible;

4. Declaring that the statutory deadline of December 31, 1980 for reporting census figures to the President contained in the Census Act is directory and not mandatory;

5. Directing the Census Bureau to report to this Court within thirty days from the date of this opinion its plan to effectuate the rulings of this Court.

No stay pending appeal shall be granted.

A judgment so providing shall be submitted by plaintiffs on notice forthwith.

SO ORDERED.

[1] Plaintiffs are Hugh L. Carey, Governor of the State of New York, and Edward I. Koch, Mayor of the City of New York, in both their official and individual capacities, several other citizens, voters and taxpayers residing within New York State, and the City and State of New York.

[2] Also named as defendants are Philip M. Klutznick, Secretary of Commerce, Vincent P. Barabba, Director of the Bureau of the Census, William Hill and Arthur G. Dukakis, Regional Directors of the Census Bureau for the New York and Boston Regions, respectively, Richard Bitzer, Acting Assistant Regional Director of the Census Bureau for the New York Region, Edmund L. Henshaw, Jr., Clerk of the United States House of Representatives, and Jimmy Carter, President of the United States.

[3] The process of geocoding involves entering a code indicating the city block, enumeration district, census tract, place, and other data of a similar nature for every address. In geocoding, "there is a computer program, and a data base known as the geographic base file, or GBF, and the addresses are compared in the computer to the GBF file."

[4] An edit failure occurred "[a]ny time the number of units reported by a respondent in a structure was greater than the number of units on the address register for that address." Structures that failed edit were assigned for recanvass. Tr. at 1139.

[5] Centralized offices were those in which the district office manager was a permanent Census Bureau employee. In New York City, seventeen of the twenty district offices were centralized. A centralized district generally is viewed as being more difficult to enumerate than a decentralized district. In decentralized districts, the district office manager was drawn from the local community. Tr. at 1213-14.

[6] Mission Night was a procedure designed to enumerate the populations of temporary residences whose rooms cost no more than $4.00 per night. This would include transient hotels, religious missions and other types of hostels. Tr. at 640.

[7] See Young v. Klutznick, 497 F. Supp. 1318 (E.D.Mich., 1980).

[8] In response to Judge Gilmore's direction to develop a "statistically defensible" method for adjusting the undercount the Bureau developed a definition of statistical defensibility. The criteria used by the Bureau to define a statistically defensible procedure are (1) appropriateness to use to be made of the results; (2) provision of measures of uncertainty; (3) listing and verification of assumptions; demonstration of robustness; (4) description of data sources, reliability and limitations; (5) reproductibility; (6) timeliness; (7) least cost for given degree of uncertainty. Defendants' Ex. 23 at 4-9.

[9] Defendants, in their post-trial brief assert again that plaintiffs lack standing to bring this lawsuit. They contend that while the Court, in determining defendants' motion to dismiss, was required to accept as true that New York State would be deprived of its rightful share of representatives if the undercount were not remedied, it may no longer do so after trial. Defendants argue that at trial plaintiffs did not establish any facts which would entitle them to maintain this suit and therefore, the Court should now dismiss for lack of standing. This argument is meritless and indicates to this Court that defendants fundamentally misconceive the standing issue. In analyzing a standing question, the court's focus should be on "whether the plaintiff [has alleged] that the challenged action has caused him injury in fact, economic or otherwise." Association of Data Processing Service Organizations, Inc. v. Camp, 397 U.S. 150, 152, 90 S. Ct. 827, 829, 25 L. Ed. 2d 184 (1970). As is evident from the foregoing language, the inquiry into standing is predicated on the allegations of the complaint. This assumes that the allegations are accepted as true, regardless of the stage at which the issue is raised. Thus, if a plaintiff ultimately fails to establish his claims at trial, this would result in a dismissal on the merits and not for lack of standing. In any event, defendants' challenge on this ground is moot in view of the Court's conclusion that plaintiffs have demonstrated that they will be injured by a failure to remedy the undercount in New York.