The opinion of the court was delivered by: WEINSTEIN
WEINSTEIN, District Judge.
A year and a half ago plaintiffs brought this action challenging the validity of section 131-a of the New York Social Services Law effective July 1, 1969. They alleged that it did not meet the standards set by section 402(a)(23) of the Social Security Act of 1935, as amended in 1968 (42 U.S.C. § 602(a)(23), referred to below as section 402), for participation by a state in the federally-funded Aid for Dependent Children (AFDC) Program.
Prior stages of the litigation have already been sufficiently described. See National Welfare Rights Organization v. Wyman, 304 F. Supp. 1346 (E.D.N.Y. 1969); Rosado v. Wyman, 304 F. Supp. 1350, 1354, 1356, (E.D.N.Y.), rev'd, 414 F.2d 170 (2d Cir. 1969), rev'd, 397 U.S. 397, 90 S. Ct. 1207, 25 L. Ed. 2d 442 (1970). For our present purposes it is enough to point out that the United States Supreme Court held that on July 1, 1969 defendants implemented AFDC schedules which "impermissibly lowered [the] standard of need by eliminating items that were included prior to the enactment of § 402." Rosado v. Wyman, 397 U.S. 397, 416, 90 S. Ct. 1207, 1219, 25 L. Ed. 2d 442 (1970).
The Supreme Court remanded the case to this Court to "review, taking into account the views of HEW [the Department of Health, Education and Welfare] should it care to offer its recommendations, any revised program adopted by the State, or, should New York choose not to submit a revamped program * * issue its order restraining the further use of federal monies." Id. at 421-422, 90 S. Ct. at 1222.
As a result of the litigation to date, it is apparent that since July 1969 New York has received hundreds of millions of dollars of federal money in violation of federal law. The violation resulted in illegally reducing payments to recipients of AFDC aid by tens of millions of dollars.
Effective June 1, 1970 New York State's AFDC program was revised. Defendants contend that this current program satisfies the requirement of section 402.
For the reasons indicated below the State is still in error. It must either comply with federal law or stop taking federal aid.
I. NEW YORK'S PRESENT AFDC PROGRAM
As in the past, the new State plan pays to AFDC recipients 100% of their need, as defined in the plan. In New York, therefore, the basic need schedule or standard of need serves two functions: It is a standard for determining eligibility to participate in the AFDC program and it represents the level of benefits that an AFDC recipient is to receive. The basic need schedules, effective June 1, 1970, are as follows (18 N.Y.C.R.R. 352.1 and 352.2):
CURRENT BASIC NEED SCHEDULE
1 2 3 4 5 6 7 Ea. add.
(New York City
suburban counties) $ 84 $ 134 $ 179 $ 231 $ 284 $ 329 $ 374 $ 45
(42 Northern counties) $ 69 $ 115 $ 161 $ 207 $ 253 $ 294 $ 335 $ 41
Western counties) $ 65 $ 111 $ 157 $ 203 $ 249 $ 290 $ 331 $ 41
Generally, the payment received for basic needs by an AFDC recipient of given family size will not correspond exactly to the above schedule. Once it is determined that a family has a "budgetary deficit," i.e., its income is not equal to or greater than the need schedule, the eligible family only receives payments from the State to meet the budgetary deficit.
In addition to the current basic need schedule there are schedules for the cost of rent and heating fuel and a very few other items of need such as household moving expenses. 18 N.Y.C.R.R. § 352.6(a). This new system results from a desirable consolidation of scores of special need items. See Rosado v. Wyman, 304 F. Supp. 1356, 1369 (E.D.N.Y. 1969) for an explanation of the advantages of this new flat-grant system. The schedules in effect on July 1, 1969 had also embodied a flat-grant system but the Supreme Court found that, in promulgating those schedules, the State had impermissibly eliminated some special need items.
During the year prior to July 1, 1969 New York State's AFDC program complied with federal law. During that year, as well as currently, the levels of benefits were designed "to fully make up budgetary deficits as defined by its standards of need." Rosado v. Wyman, 304 F. Supp. 1356, 1365 (E.D.N.Y.), rev'd, 414 F.2d 170 (2d Cir. 1969), rev'd, 397 U.S. 397, 90 S. Ct. 1207, 25 L. Ed. 2d 442 (1970). Posed is a simply stated question of fact: is the State's present standard of need lower than the one used in the base year, July 1, 1968 through June 30, 1969? Since the State has always claimed to pay 100% of its defined standard of need, the question can be rephrased to ask whether current payments under the schedules effective June 1, 1970 are no less than the amounts paid to like AFDC recipients in the base year. The simplicity of statement belies the difficulties encountered in making the required comparison.
II. STANDARD OF NEED COMPELLED BY SECTION 402
The Supreme Court has decided that section 402 does not preclude a State from redefining its method of determining need. It may consolidate and simplify its standard of need by eliminating certain "special needs" supplied after individual request and authorization, and provide instead for meeting both special needs and basic needs out of a uniform allowance. But it must satisfy the fundamental requirment that "all factors in the old equation [be] accounted for." Rosado v. Wyman, 397 U.S. 397, 419, 90 S. Ct. 1207, 1221, 25 L. Ed. 2d 442 (1970).
The obvious question is how New York State is to determine the amount to be included to meet the former special need items since by their very nature, they were provided to some but not all of the recipients with varying frequency and in varying combinations. To this the Supreme Court responded that the State might consolidate items on the basis of statistical averages -- so long as this represented a fair averaging -- and thereby arrive at a flat dollar amount for each family or person. Id.
At hearings conducted in June 1970 the defendants attempted to demonstrate that the revised schedules were in compliance with section 402 by showing the theory upon which the new schedules were devised. At that time HEW advised the Court that it could not -- using materials supplied by the State -- determine whether the revised AFDC schedules complied with section 402. The Court agreed with HEW's conclusion and ruled that defendants had not adequately justified the June 1, 1970 schedules. They had failed to demonstrate that the new schedules accounted for those items of special need eliminated on July 1, 1969 and not otherwise provided for under current regulations.
All parties, HEW and the Court agreed that since an AFDC recipient's need was determined much differently currently than in the base year no meaningful comparison could be made between need "schedules" for the two periods without further information. A comparison would have to be made, therefore, between current payments to individuals under the AFDC program and payments made to like individuals during the base year. The data, which was not then available but was required for any meaningful comparison between the base year and present payments, was the amount per recipient per month paid for "special need items" as defined in the AFDC program in effect during the base year. This data could not be obtained from a schedule because the special need items were individually tailored payments made upon individual application and verification. The problem was to convert these special need item payments -- by their nature not uniform throughout the population -- into a dollar value which would apply uniformly to all AFDC recipients for the base year. In the case of special need items, need and payment coincided; therefore the value obtained could be used as a substitute for need.
Neither defendants nor any local social service district had retained full records with respect to special need payments made during the base year. There were not even skeletal records for any area of the State other than New York City. Within the City of New York the problem was compounded by changes since the base year in record-keeping techniques, and by lack of computer records for critical periods.
An added complication in making comparisons is that the base year itself was not stable. A number of substantial adjustments were made in the AFDC program during that year. For instance, some special need items apparently were absorbed into the basic need grant. See the discussion of laundry at p. 1185, infra. In New York City there was substituted for some special need payments a flat payment of $100 per recipient per year that became the New York City cyclical grant.
All concerned concurred in the decision that a statistical sampling of New York City caseload should be undertaken in order to ascertain the average value per recipient per month of special need items included in the need standard prior to July 1, 1969 and for which provision is not specifically made by current regulations. Accepting the suggestion of defendants, the Court ruled that the appropriate base period was July 1, 1968 through June 30, 1969. Defendants, plaintiffs and HEW agreed that the results obtained from this sample would be treated as the uniform result for the entire State.
The Court is, in a sense, comparing base year apples with current year pears. Much of the computation consists of converting the apples into pears -- a task similar to that of historical recreations generally. See, e.g., H. Hughes, History as Art and as Science, 94 (1964) (" all history is contemporary in the sense that its presentation reflects the circumstances and attitudes of those who write it"). (emphasis in original). The Study is an attempt to construct a standard of need in the base year which is comparable to the current standard of need. In the process some almost arbitrary reconstructions must be made. For example, we have assumed that the basic grant for the base year was uniform over the twelve month period, although adjustments were made during the year. See, Rosado v. Wyman, 304 F. Supp. 1356, 1366 (E.D.N.Y. 1969). We do not, however, treat the New York City cyclical grant as being in force except during the last ten months of the year, when it was actually in effect (see discussion of clothing and household furnishings and repairs, p. 1187, infra).
Sampling has long been considered an acceptable method of determining the characteristics of a large universe. The United States Decennial Census of 1970, for example, is relying on sample reports to determine information on income, work experience and earnings, and housing characteristics of the population of the United States. For inter-census years, a small sample of only 50,000 households is used to represent the total population.
Such mathematical and statistical methods are well recognized by the courts as reliable and acceptable in determining adjudicative facts. See, e.g., Jones v. Georgia, 389 U.S. 24, 88 S. Ct. 4, 19 L. Ed. 2d 25 (1967) (finding of discrimination based entirely on jury selection statistics and absence of explanation); Hernandez v. Texas, 347 U.S. 475, 74 S. Ct. 667, 98 L. Ed. 866 (1954) (same); Maxwell v. Bishop, 398 F.2d 138, 141-148 (8th Cir. 1968), rev'd on other grounds, 398 U.S. 262, 90 S. Ct. 1578, 26 L. Ed. 2d 221 (1970) (racial discrimination in application of death penalty); Campbell v. Board of Education, 310 F. Supp. 94, 105 (E.D.N.Y. 1970) (election procedure); Gautreaux v. Chicago Housing Authority, 296 F. Supp. 907, 910-914 (N.D. Ill. 1969) (housing discrimination); Zippo Manufacturing Co. v. Rogers Imports, Inc., 216 F. Supp. 670 (S.D.N.Y. 1963) (unfair competition; sample of consumers); Iannucci v. Board of Supervisors, 20 N.Y. 2d 244, 251, 282 N.Y.S. 2d 502, 507, 229 N.E. 2d 195, 198 (1967) (reapportionment); Coordinating Committee for Multiple Litigation, Manual for Complex and Multidistrict Litigation, 2.612 (1970) (citing considerable case authority); Barksdale, The Use of Survey Research Findings as Legal Evidence (1957); Hart & McNaughton, Evidence and Inference in the Law, in Hayden Colloquium on Scientific Concept and Method 54-55 (Lerner ed. 1958); Ball, The Moment of Truth: Probability Theory and Standards of Proof, 14 Vand. L. Rev. 807, 813 (1961); Banzhaf, Weighted Voting Doesn't Work: A Mathematical Analysis, 19 Rutgers L. Rev. 317 (1964); Finkelstein, The Application of Statistical Decision Theory to Jury Discrimination Cases, 80 Harv. L. Rev. 338 (1967); Finkelstein and Fairley, A Bayesian Approach to Identification Evidence, 83 Harv. L. Rev. 389, 516-17 (1970); Kaplan, Decision Theory and the Factfinding Process, 20 Stan. L. Rev. 1065 (1968); Kingston, Probability and Legal Proceedings, 57 J. Crim. L.C. & P.S. 93 (1966); Liddle, Mathematical and Statistical Probability, 51 Iowa L. Rev. 849 (1966); Zeisal, The Uniqueness of Survey Evidence, 45 Cornell L.Q. 322 (1960). It was a principle recommendation of the prestigious committee which wrote the Manual for Complex and Multidistrict Litigation that "[scientifically] designed samples * * * meeting the tests of necessity and trustworthiness * * * [be used to] contribute materially to shortening the trial of the complex case." Manual for Complex and Multidistrict Litigation, supra, at 2.612.
Statisticians can tell us with some assurance what the reliability factors and probabilities are. Only the law can decide, as a matter of procedural and substantive policy, what probabilities will be required before the courts will change the status quo by granting a remedy. Thus, in deciding whether a sample is adequate, practical limits to fact finding precision must be considered. See, e.g., Zippo Manufacturing Co. v. Rogers Imports Inc., 216 F. Supp. 670, 681 (S.D.N.Y. 1963) (three samples of 500 each to cover all smokers in the country); Brown Shoe Co. v. United States, 370 U.S. 294, 340, n. 69, 82 S. Ct. 1502, 8 L. Ed. 2d 510 (1962). As Judge Feinberg pointed out in connection with the use of sampling techniques:
"Necessity * * * requires a comparison of the probative value of the survey with the evidence, if any, which as a practical matter could be used if the survey were excluded." Zippo Manufacturing Co. v. Rogers Imports, Inc., 216 F. Supp. 670, 683 (S.D.N.Y. 1963).
In the instant case where the question of the value of special needs arose after the fact, the State did not have adequate retained information or data at hand from which it could determine the amount of special need grants made during the relevant period, nor could any such information be readily produced. The only source of such information was a review of the individual case records of AFDC families. It was entirely impracticable to review all the case records. The monthly average number of cases in New York City from July 1968 through June 1969 was 186,629. Accordingly, a sample of such cases was the only feasible technique.
Two independently and randomly selected one percent samples of the "families" in New York City which received AFDC assistance during the base year were selected. Of the original 5,344 cases in the sample, sufficient information could be located for 3,343 cases. Though small, the State's and HEW's statisticians seem reasonably assured that the sample is a valid reflection of the universe.
Findings from the cases studied would be more representative had it been practicable to locate all the 5,344 records whose numbers had been drawn. In the short time available, and in the light of available resources, this was not possible. The statistics show that for the examined cases the average size of the recipient group and the average monthly payment are very close to these averages for the total caseload, thus indicating that the sample closely resembles the population. This conclusion is strengthened by data which shows that the distribution of recipient groups by size of group was sufficiently similar in the sample and in the population so that the sample cases could properly be considered, within accepted statistical confidence limits, as a fair random selection from the population.
For the sample observations derived from the 3,343 cases, the chances are better than 95 out of 100 that the average outlay per month for 20 special needs per person in the universe would lie in a range of +- $.15 about the figure of $.94 obtained from the sample, that is between $.79 and $1.09. The comparable ranges for the average needs for each family size are (according to HEW):
1. $ 1.47 +-.46
2. 2.74 +-.24
3. 2.71 +-.19
4. 3.57 +-.30
5. 4.17 +-.46
6. 6.10 +-.62
HEW, as amicus curiae, examined these statistics on the basis of the procedures outlined for the Study and field observations by its own staff. It found the Study results an acceptable indicator of what actually took place during the base year.
The Court recognizes that extrapolating to arrive at precise dollar figures for families of different sizes in various areas of the State, particularly where the sample is so small, presents opportunities for discrepancies and errors. There is, for example, little hard factual data before us to support a critical hypothesis that upstate "families" received special grants at the same rate as New York City "families." Nevertheless, the practical fact finding problems facing the Court require risks to be taken. The probability of injustice by failing to act or by delaying action to await more definitive studies is far greater than the probability of injustice by acting on the best obtainable incomplete data.
The results of the Study for each special need item may be thought of as an indication of average payment. A reasonable average value for each person or family in a group -- in which it is admitted that not all have an equal need for each specific component -- is obtained so long as the total amount of the payment for each item corresponds to the extent of the need for that item in the total caseload. This Court finds that the figures of average grants produced from the Study represent a fair averaging even though they involve rounding off, sample error, and other obstacles to precision. They are the basis of most of the calculations in this memorandum.
The State's contention that present AFDC need schedules in area SA-1 comply with section 402 -- that is, are at least equal to the need schedules in effect during the base ...