Ruth Johnson v. Henry C. White, Commissioner of Welfare, State of Connecticut

FRIENDLY, Circuit Judge: This class action by Connecticut beneficiaries of the Aid to Families with Dependent Children (AFDC) program, jointly funded by the Federal Government and the State, has been pending since September, 1971. It has required enormous effort by counsel for the plaintiffs and for the State, by HEW, and by Judge Blumenfeld, all of whom have labored conscientiously and cooperatively *1230 to achieve a just result. Although the history of this case is fully related in Judge Blumenfeld’s opinion, 353 F.Supp. 69 (D.Conn.1972), some of the factual material must be repeated in order to put into perspective the points we are asked to decide. In the Social Security Amendments of 1967, effective January 2, 1968, 81 Stat. 821,898, 42 U.S.C. § 602(a)(23), Congress added § 402(a)(23) to the Social Security Act. This stipulated that, in order to receive federal participatory funding, 1 a state’s plan for AFDC must provide that by July 1, 1969, the amounts used by the State to determine the needs of individuals will have been adjusted to reflect fully changes in living costs since such amounts were established, and any máximums that the State imposes on the amount of aid paid to families will have been proportionately adjusted. At that time, the State of Connecticut computed its “standard of need” by determining a separate budget for each AFDC-assisted unit; while this budget included certain standardized sums for the basic needs of food, clothing, personal incidentals, and household supplies, the sums for shelter, utilities, and various “special needs” were included either at the actual cost to the recipient unit or at that cost limited by a stated maximum. Payments were made on the basis of this standard of need as reduced by various sums representing income or resources otherwise .'available to the recipient unit. Following the Congressional mandate, the. State Commissioner of Welfare proceeded to adjust a large part of the standard of need to account for inflation. In the absence of any action reducing the percentage of need to be met, this correspondingly increased the level of payment as well. Consultation with officials at HEW led to further adjustments in various components of the standard. In 1971 the State decided to change from a system in which the needs of AFDC recipients were computed and paid on an individualized basis to a system in which, except for certain relatively unusual items, recipients would be paid on the basis of a “flat grant,” varying only by the number of persons in the assisted unit. This new system was to be known as the Connecticut Family Assistance Plan (CFAP). In order to determine the standard of need upon which the grant could be computed, the State, in the summer of 1971, conducted an elaborate statistical analysis, which will be discussed in detail at various points in this opinion. For the moment, it will suffice to say that, in broad terms, the State computed the standard of need for each size of assistance unit by averaging together a random sample of the sums budgeted for those units (before income disregards) under the prior assistance program. The budgets used had been constructed in the period from June 1, 1970 to May 31, 1971. The State planned, as of November 1, 1971, to implement a system in which the standard of need was constructed as just described, and the level of benefits would be equal to 85% of need. On October 28, 1971, following a hearing, the district judge issued a preliminary in *1231 junction preventing that implementation. 2 At a further hearing held a few days later, the parties and the court agreed that HEW should be invited to review the plaintiffs’ claims and to submit a brief as amicus curiae. HEW then conducted an extensive review of the State’s program; the State incorporated into its plan various changes suggested during this review, which resulted in yet further updating of some of the component parts of the new averaged standard of need. On April 3, 1972, HEW filed its brief endorsing the State’s program. Defendant then moved to dissolve the preliminary injunction, including in its moving papers a representation of the State’s willingness to raise the level of benefits to equal the standard of need, “barring unforeseen circumstances.” Following another hearing, the district court, in an opinion dated June 12, 1972, terminated the preliminary injunction and, with two exceptions not here relevant, ordered judgment entered for the defendant. Plaintiffs thereafter submitted two motions. One was entitled “Motion to Reopen, Set Aside, Alter and Amend Judgment, For Relief From Judgment, To Allow Further Evidence and Allow Reargument, For Rehearing and For Stay of the Execution of Judgment Pending Disposition of Motion.” The other was entitled “Motion to Amend and Supplement Findings of Fact and Conclusions of Law.” The motions were accompanied by various affidavits and proposed exhibits, indicating material which plaintiffs would introduce at the reopened hearing. At the parties’ request decision on these motions was deferred pending lengthy efforts at settlement. When these ultimately proved unproductive, the judge denied the motions and this appeal followed. I. Plaintiffs’ initial contention is that, by directing dismissal of the complaint, the judge transgressed his authority, and that we should reverse and remand on that ground alone. They do not contest that, just as a court may order a consolidation of the trial on the merits with “an application for a preliminary injunction,” F.R.Civ.P. 65(a)(2), so it may order consolidation with a motion to end one. Their point, rather, is that the judge did not give notice of his intention until after the hearing had been concluded. We agree that the judge should have done this either during the hearing or, in any event, before rendering his decision. Capital City Gas Co. v. Phillips Petroleum Co., 373 F.2d 128, 131 (2 Cir. 1967). However, in order to obtain a reversal for such an error, a party must show, not only surprise but “prejudice” in the sense of having other material evidence to introduce. Nationwide Amusements, Inc. v. Nattin, 452 F.2d 651, 652 (5 Cir. 1971) ; Eli Lilly & Co. v. Generix Drug Sales, Inc., 460 F.2d 1096, 1106-07 (5 Cir. 1972) ; 11 Wright & Miller, Federal Practice and Procedure § 2950 at 487-88 (1973). Apart from plaintiffs’ failure to make specific objection to the judge’s failure to give notice in either of their two post-judgment motions, our disposition of the case is such that plaintiffs will have no ground to complain of the error. So much of their proffered evidence as we deem relevant can be offered on the remand we are directing on other grounds. II. Turning to the merits, all of plaintiffs’ contentions center on the application of § 402(a)(23), which was given extensive consideration in Rosado v. Wyman, 397 *1232 U.S. 397, 90 S.Ct. 1207, 25 L.Ed.2d 442 (1970). However, since the parties are in sharp disagreement concerning the interpretation of that opinion, we find it convenient to deal at the outset with those attacks which do not require an in-depth analysis of it. Those attacks turn on Connecticut’s alleged failure to adjust the component elements of its standard of need “to reflect fully changes in living costs since such amounts were established.” The most basic challenge concerns the use, as a basis for updating, of the Consumer Price Index, United States Urban Average, which allegedly does not reflect inflation said to be more rapid in Connecticut than in the country generally. 3 In a letter to the states dated October 17, 1969, HEW implemented the updating requirement of § 402(a)(23) by establishing seven “Acceptable cost study methods.” These differed in the degree of the burden placed on the state administering agency. For example, while Method C allowed the state to “conduct a statewide cost study of the items of living,” Method B — the method applied by Connecticut in part and now questioned — allowed the states to rely on federally-generated statistics. It provided for the use of “the U. S. Department of Labor, Bureau of Labor Statistics, Consumer Price Index, for the appropriate region [to] determine the current index price for the applicable items of living.” In its amicus brief HEW interpreted this regulation as follows: “[I]t is the position of HEW that where a State such as Connecticut does not contain one of the 23 Standard Metropolitan Statistical areas for which there are individual city indexes then the use of the U. S. City Average Index will satisfy the requirement of using the C.P.I. for the ‘appropriate region.’ ” Plaintiffs have not presented — indeed in the absence of expensive cost studies could not present — definitive proof that the nationwide CPI — Urban Average Index failed to reflect changes in statewide Connecticut living costs. Their case at trial was that Connecticut should have used the cost of living indices for Boston and/or New York City, the two of the 23 Metropolitan Statistical areas nearest to Connecticut. We see no logic compelling that conclusion, particularly in view of HEW’s contrary position; see Rosado, 397 U.S. at 415, 90 S.Ct. 1207. Geographical proximity does not guarantee identity of economic circumstances. On appeal plaintiffs take a different tack, suggesting use of a Northeast regional index. While there is a controversy whether this was available at the time of Connecticut’s update, 4 we need not attempt to resolve it since plaintiffs did not urge use of any regional index upon the district court. All that is left of plaintiffs’ general attack on updating is the contention that even if they have suggested nothing better, or even if at the time of the update there was nothing better, use of the CPI — Urban Average Index was so manifestly inappropriate that Method B was unavailable and *1233 Connecticut was bound to engage in a state price study, despite HEW’s view to the contrary. We see no reason to think that § 402(a)(23) was intended to require statistical accuracy to the degree of denying to a state the use of common in-dices of inflation; accordingly, we reject this argument. 5 In addition to this general claim, plaintiffs also contest the updating of two specific items. The first is “Household furnishings,” an item previously made available on a “special need” basis. Plaintiffs contend that the 6.2% change made in updating that item from 1967 to 1969 is incorrect, and that the use of the correct sub-index from the CPI — Urban Average would show in the neighborhood of an 8% change. The evidence, however, indicated that while the CPI index for some of the items included in this broad category showed a change of 8% or more, the rate of inflation for other items was considerably lower; in addition, there was testimony that this category had been repriced, at least in part, through actual shopping surveys. HEW, in its brief, stated that it “consider[ed] the updating procedures to be proper.” The trial judge, although not mentioning this specific item by name, found that various components of the standard of need, including “certain special needs,” had been “sufficiently updated.” We think this issue, raised below, was fairly covered by this finding of the trial judge, who made a conscientious effort to address every issue tendered. There was evidence to support his conclusion, and plaintiffs do not now assert that they have any further data to submit on this particular issue; accordingly, we will not disturb his finding. Plaintiffs’ second specific challenge relates to the updating of the “personal incidentals” part of the basic welfare grant (under the pre-CFAP system) for food, clothing, personal incidentals, and household supplies (FCPH). The evidence showed the following: In 1967, a State Cost of Living Commission had reviewed various welfare standards, and had increased parts of the FCPH grant by a significant percentage. For the personal incidentals component, however, the Commission had set a figure which for adults and some children was the same figure as that which had been in use since 1963, and for other children was in fact somewhat lower. Following the adoption of § 402(a)(23), the state updated the personal incidentals item, using 1967 as the base date. Plaintiffs contend that the State failed to comply with the statutory requirement that items be updated to account for inflation “since such amounts were established.” The sum for personal incidentals, they contend, was “established” in 1963; defendants claim that it was “established” in 1967. We think, as did HEW and the trial court, that the State has the better of this argument. Connecticut was making an effort, even before the enactment of § 402(a)(23), to keep its standards in line with rising living costs, and plaintiffs do not contend, nor does the evidence show, that this effort was in any way a sham. Although it might be true to say that the Commission’s action in 1967 would not be in full compliance with the more stringent requirements of the later-enacted federal provision, that is quite different from saying that it was in any sense an attempt to circumvent § 402(a)(23) as later enacted. We think the district court was justified in following HEW’s lead in *1234 finding that the personal incidentals item had been sufficiently updated. III. We now turn to plaintiffs’ contentions relating to thé changes made by CFAP with respect ' to shelter allowances. Since this is the area where the profound differences between the parties over the meaning of Rosado are most important, some general discussion of that case may be useful by way of preface. We start from the proposition that, under the plain language of § 402(a)(23), if a state had previously “established” a standard of need fully conforming to the Constitution, the requirements of the Social Security Act, and HEW’s regulations thereunder, and had then adjusted the figures “to reflect fully changes in living costs since such amounts were established,” the result would be immune from successful attack on the ground that a more finely tuned analysis would show that the previous plan did not provide a sufficiently high standard of need. 6 The words of the statute, which Mr. Justice Harlan characterized as being, along with “those common-sense assumptions that must be made in determining direction without a compass,” a court’s “chief resources” in determining the meaning of § 402(a)(23), 397 U.S. at 412, 90 S.Ct. at 1218, do not permit any other conclusion. In addition, although the legislative history “reveals little except that we have before us a child born of the silent union of legislative compromise,” 397 U.S. at 412, 90 S.Ct. at 1218, a more expansive reading of the statute would be inconsistent with the fact that Congress refused to adopt language that arguably would have chartered more extensive judicial forays, see 397 U.S. at 409-10, 90 S.Ct. 1207. In opposition to this interpretation, plaintiffs rely on two passages in the Rosado opinion. One is the statement that a purpose of § 402(a)(23) was “to require States to face up realistically to the magnitude of the public assistance requirement and lay bare the extent to which their programs fall short of fulfilling actual need,” 397 U.S. at 412-13, 90 S.Ct. at 1218. The other is that a state “may not obscure the actual standard of need,” 397 U.S. at 413, 90 S.Ct. at 1218, and that while the statute “leaves the States free to effect downward adjustments in the level of benefits paid, it accomplishes within that framework the goal, however modest, of forcing a State to accept the political consequence of such a cutback and bringing to light the true extent to which actual assistance falls short of the minimum acceptable,” id. However, these passages must not be read in isolation. Earlier in the opinion, 397 U.S. at 408, 90 S.Ct. at 1216, Mr. Justice Harlan had noted that both as to the “standard of need” and as to the “level of benefits,” “Congress has always left to the States a great deal of discretion” — followed by a reference to King v. Smith, 392 U.S. 309, 318, 88 S.Ct. 2128, 20 L.Ed.2d 1118 (1968). 7 Later, after *1235 making the rather general statements on which plaintiffs rely, the opinion returned to the issue in the case, namely, the requirements of § 402(a)(23) in a situation where a state has not merely updated the figures established for need but at the same time has altered its plan. The holding was that while the amendment left the states free to do this within the limits of the Federal Constitution, the Social Security Act, and the regulations thereunder, a state could not use a changed plan to circumvent the requirement for updating needs. In what we take to be the core of the opinion, the Court said, 397 U.S. at 414r-15, 90 S.Ct. at 1219: It follows from what we fathom to be the congressional purpose that a State may not redefine its standard of need in such a way that it skirts the requirement of re-evaluating its existing standard. This would render the cost-of-living reappraisal a futile, hollow, and indeed, a deceptive gesture, and would avoid the consequences of increasing the numbers of those eligible and facing up to the failure to allocate sufficient funds to provide for them. Later the Court put the same thought thus, 397 U.S. at 419, 90 S.Ct. at 1221: Section 402(a)(23) invalidates any state program that substantially alters the content of the standard of need in such a way that it is less than it was prior to the enactment of § 402(a)(23), unless a State can demonstrate that the items formerly included no longer constituted part of the reality of existence for the majority of welfare recipients. What caused New York’s plan to founder was that its new program had eliminated the previous allowance for special grants (which was sometimes paid as a substitute flat grant of $100 per person) rather than averaging them out and adding that figure to the basic recurring grant, 397 U.S. at 416, 90 S.Ct. 1207. 8 When we now address the manner in which the shelter component of CFAP was constructed, it could be claimed that any averaging process with respect to shelter allowances is invalid. The cost of shelter can depend on factors such as locality, access to public housing, prior ownership of a dwelling, etc., beyond a recipient’s control. Some other states, such as New York, have not attempted to include shelter allowances within their flat-grant programs; see Boddie v. Wyman, 434 F.2d 1207, 1211-12 (2 Cir. 1970), affirmed, 402 U.S. 991, 91 S.Ct. 2168, 29 L.Ed.2d 157 (1971). Plaintiffs’ argument, however, is more modest; they claim only that the process of averaging is invalid when “statistically distinct populations” have been merged together, and that Connecticut violated that standard when it averaged together the budgets of assistance units which were the sole occupants of a housing unit with assistance units which shared quarters with non-recipients. Using statistical data plaintiffs showed, and the trial court found, that the budgeted needs of these two groups were sufficiently different to make them separate populations. However harshly this program might bear on those units for whom the averaged sum was lower than the previously *1236 determined budget for shelter, we do not think that § 402(a)(23) prevents averaging in this situation. This result follows from the Court’s indication that New York could have abolished its former system of special grants if only it had included them in the figures used to determine the basic recurring grants to all AFDC recipients; this action would, of course, have taken a portion of the grants away from those who needed them and given sums to those who did not. It follows also from the statement, 397 U.S. at 419, 90 S.Ct. at 1221: Providing all factors in the old equation are accounted for and fairly priced and providing the consolidation on a statistical basis reflects a fair averaging, a State may of course, consistently with § 402(a)(23) redefine its method for determining need. In any case, Connecticut itself has now moved to cure these inequities in CFAP. Legislation adopted in 1974 and 1975 has resulted in a new § 17-2(c) of Title 17 of the Connecticut General Statutes, which we reproduce in the margin. 9 As we read the statute, it strips the shelter component of the flat grant of any significance except for families whose need is no more than the flat grant but who would benefit if it were higher. 10 Plaintiffs are on firmer ground in their contentions that at least some of the previously established rental allowances were unlawful under other provisions of § 402(a) or HEW regulations. Their criticisms are directed at the four following situations: (1) When an AFDC child resided with a non-needy, non-legally liable relative, e. g., an aunt, the state budgeted $20 per month for one child, $30 for 2 children, and $40 for 3 or more children; (2) When an assistance unit as a whole resided with a stepfather, the state budgeted a prorata share of actual rent or of the maximum allowed rent; (3) When an assistance unit as a whole resided with a legally liable relative, or with a non-legally liable relative, or with a friend, the state also budgeted a prorata share of actual rent or of the maximum allowed rent; (4) Finally, in several instances the state budgeted no rent at all. To take the easiest case first, Connecticut cannot properly include in the base for CFAP shelter allowances sums that were determined under an unlawful attribution of income policy as delineated in Van Lare v. Hurley, 421 U.S. 338, 95 S.Ct. 1741, 44 L.Ed.2d 208 *1237 (1975). While § 402(a)(23) does not require revision of previously lawful plans to achieve a closer approach to perfection, it likewise does not permit continuation of an unlawful scheme simply because it was updated. See Roselli v. Affleck, supra, 508 F.2d at 1282. Van Lare would require an upward adjustment of the shelter allowances for those situations included in categories (2) and (3) where the assistance unit lived with a person not legally liable for the support of the children, although contributions actually made would be deductible from the grant. However, assuming that the sums are properly updated, see 353 F.Supp. at 83, it does not seem to us that Van Lare or its predecessors, King v. Smith, supra, 392 U.S. 309, 88 S.Ct. 2128, 20 L.Ed.2d 1118 and Lewis v. Martin, 397 U.S. 552, 90 S.Ct. 1282, 25 L.Ed.2d 561 (1970), condemn the budgets included in category (1). The most reasonable construction of this provision is not that it was an attempt by Connecticut to attribute income which in fact was not available to the dependent children, in violation of 45 C.F.R. § 233.90(a), but rather that it was an effort, not ruled out by any regulation that has been called to our attention, to make sure that relatives did not profit by charging outrageous rents to children who were in no position to protect themselves. For category (4), the record is unclear as to what circumstances led to budgeting certain families at a zero allowance for shelter. The State’s brief on appeal claims that these budgets were “except for assistance unit size one, caused in many cases by the recipients living in home-owned property not mortgaged.” However, plaintiffs’ post-trial motions contained offers of considerable factual material bearing on the zero-allowance cases, and it may be that plaintiffs can show that many or all of these cases in fact run afoul of Van Lare. Accordingly, this issue should remain open upon remand. We therefore remand the portion of this case relating to the shelter allowance component of CFAP for such introduction of evidence and ultimate adjustment of the standard as the district court may find appropriate in light of this opinion. 11 IV. Plaintiffs also make a number of attacks on the statistical accuracy of the way in which Connecticut moved from its previous plan to the flat grant program of CFAP. Accepting that a flat grant can be constructed from one based on individual budgets “providing the consolidation on a statistical basis reflects a fair averaging,” 397 U.S. at 419, 90 S.Ct. at 1221, plaintiffs assert that the proviso was not met. The results of the State’s statistical efforts are summarized in Stipulated Chart I, which we reproduce as an appendix to this opinion. Vertically, the *1238 chart is divided into assistance units 1 through 15, corresponding to the number of AFDC recipients in the aided family. Horizontally, the chart is divided into five categories: FCPH (the basic allowance for food, clothing, personal incidentals and household supplies); shelter; excess utilities; recurrent needs (i. e., recurrent “special” needs); and nonrecurrent needs (i. e., nonrecurrent “special” needs). Each category (except for “excess utilities,” which did not have to be updated because it was budgeted at actual cost) has an “updated” column, to account for those additional adjustments that had not already been included in the budgets drawn for the year 6/1/70-5/31/71. The updated columns (plus the excess utilities column) were added, for each numerical assistance unit, to yield an “updated standard of need” sum; it is this sum which forms the standard against which payment is calculated (after deduction of available income and addition of any needs not included in the composite standard). The challenges here under discussion go to the adequacy of the statistical techniques by which the factors for each size of assistance unit were computed. The first, and broadest, claim of the plaintiffs is that the sampling technique was insufficient. Plaintiffs do not contend that the sample, 3,079 out of approximately 27,000 families receiving AFDC assistance, was inadequate for all purposes. 12 Their claim, to quote the district court, is rather “that in certain components of need for certain assistance unit sizes, the sample size was too small to guarantee the selected level of confidence. The defendant’s confidence level, however, was for the average of budgeted needs as a whole, rather than for each component in the standard of need,” 353 F.Supp. at 75. In other words, the plaintiffs contend that each component part of the standard of need — apparently including each of the many special needs — should have been checked out to see if there were enough examples of budgeting for that item in the sample drawn; whereas defendants considered it sufficient if there were enough sample budgets from each level of assistance unit, regardless of how many times an individual item was budgeted. Without pretending to a degree of knowledge of statistics which we do not possess, it seems clear that plaintiffs’ suggested approach would be both more reliable and more arduous. 13 The question thus is whether the defendant’s method is sufficiently accurate that the additional expense of pursuing plaintiffs’ suggestion was not required. HEW was satisfied with what Connecticut had done. This is the kind of problem on which we join Mr. Justice Harlan both in viewing “with concern the escalating involvement of federal courts in this highly complicated area of welfare benefits,” 397 U.S. at 422, 90 S.Ct. at 1223, and in giving weight to the views of the expert agency charged with supervising the expenditure of federal funds, 397 U.S. at 415, 90 S.Ct. 1207, in the absence of clear demonstration of error. The more particularized claims are more troublesome. It will be noted that the figures for assistance units 10 through 15 in Chart I are peculiar in certain respects. On occasions the sums revealed by the survey do not fluctuate *1239 in any intelligible manner. Under shelter, for example, the survey showed: unit 10 $157.11 unit 11 144.61 unit 12 147.56 unit 13 164.05 unit 14 111.00 unit 15 220.35 While it may be reasonable to expect some flattening of the curve as the size of the units increases, there is no rational basis for thinking that the shelter need of a unit of 14 was a third less than that of a unit of 13 and only half that of a unit of 15. The explanation, of course, is the small size of the sample population in assistance units 10 to 15. According to the State’s figures there were only about 245 welfare families in those categories at the time of the survey, of which 39 were drawn into the sample actually used, distributed as follows: unit 10 21 cases 11 9 12 5 13 2 14 1 15 1 Any averaging of each unit, based on such a small number of cases, would give undue weight to peculiar circumstances. 14 Accordingly, the State did the following: First, for FCPH, expecting that there would be a definite correlation between family size and budgeted sum since items such as food would increase with the number of mouths to feed, the state averaged the FCPH budgets for each assistance unit size but apparently, in the course of this process, “smoothed out” the one clear statistical discrepancy, FCPH for unit size 12. Plaintiffs do not complain about this, except as regards an alleged arithmetical error discussed below. Second, for all the other items, which, it was thought, would give eccentric figures that did not correlate to the number of persons in the unit, the State amalgamated unit sizes 10 to 15, thus gaining a larger population from which an average could be drawn. Hence the identity of sums in all the other “updated” columns for units 10 to 15. HEW approved both the use of the individualized FCPH sums and the use of combined sums for the other categories. The district judge upheld this procedure on two grounds, 353 F.Supp. at 76-77 — first, because the small size oí the sample population meant that consolidation would lead to a more statistically accurate averaging; second, because of the “possible economies of scale at this level.” We do not find either point persuasive. Unless the second point were valid, the first has little weight. Merely increasing the size of the sample does not enhance accuracy if the populations are unlike; but the data gave little reason for thinking that economies of scale which had been conspicuously absent as the size of assistance units rose through the lower ranges would suddenly become evident at size 10. The burden of enlarging the samples for assistance units 10 to 15 from 39 to a larger number, even to the whole population of 245, would have been small. In sum, although the State considered variation in assistance unit size to be a crucial variable, it assumed without proof that it would not matter for most items in assistance units 10 to 15. Since the State’s own policy incorporated the opposite assumption, we think the State had the burden of proving, rather than assuming, that its treatment of these units did not fail to provide for needs which the previous assistance system had met. 15 *1240 The State says that, however this may be, rectification by adding to the sample is now impossible. While the various computer print-outs for CFAP apparently still exist, they would not show the figures for families not included in the original sample, and the State asserts that “it is impossible to get W-52T’s [i. e., budgets] at this time based on the old AFDC budgets in order to make up a new sample.” Assuming this to be true, the burden of making a field study of the actual needs of assistance units 10 to 15 at this time (if no other method is available) is hardly massive in view of the small number of families involved. We leave the fashioning of a remedy to the district court saying only that the present record does not support the flat grants, for other than FCPH, for assistance units 10 to 15 but that the remedy need not assure statistical perfection if a reasonable approximation can be achieved at substantially less cost. Plaintiffs also contend that two arithmetical errors were made in computing some of these unit sizes. The first contention is that an assistance unit two was accidentally fed into the computer as a unit ten, thereby erroneously reducing the average sum for unit 10. The second contention is that the sums shown on one of the unit twelve budgets were mispunched, resulting in a reduction of $407.79 in the FCPH component for that budget, and a lowering of the averaged budget for the unit as a whole. Neither of these contentions seems to have been made at trial, and only the second was included as part of the post-trial material. If we were not remanding on other grounds, we would scarcely do so for this. On the other hand, since a remand with respect to assistance units 10 to 15 is required in any event, plaintiffs may raise these issues as a part of it. The order of the district court is affirmed in part and reversed in part, and the cause is remanded for further proceedings consistent with this opinion. No costs. See Appendix on next page. *1241 Footnotes [1] The statute directs the Secretary of HEW to approve any state plan which fulfills the conditions specified in the twenty-four subdivisions of § 602(a) and an additional provision relating to residency requirements. 42 U.S.C. § 602(b). It also authorizes the Secretary, after reasonable notice and opportünity for a hearing, to suspend federal payments in aid of a plan upon a finding “that in the administration of the plan there is a failure to comply substantially with any provision required by section 602(a) of this title to be included in the plan.” 42 U.S.C. § 604(a)(2). Using this, and other parallel provisions, the Secretary convoked a “conformity hearing” in late 1970 to determine if Connecticut was in compliance with various requirements of the Social Security Act; the points here in dispute were not, however, there decided. We upheld the determination of HEW, that Connecticut was not in conformity, in Connecticut State Dept. of Public Welfare v. Dept. of HEW, 448 F.2d 209 (2 Cir. 1971). [2] Federal jurisdiction rested on a constitutional claim which was nonfrivolous at the time the complaint was filed. However, the court proceeded, without objection, to deal only with the federal statutory claim, on the basis of pendent jurisdiction. The constitutional claim turned out to be without merit in light of the supervening Supreme Court decision in Jefferson v. Hackney, 406 U.S. 535, 92 S.Ct. 1724, 32 L.Ed.2d 285 (1972), and was also mooted by action of the state in returning the level of benefits to 100% of need. [3] This challenge does not apply to those items, such as the rent allowance for those living in public housing, which were budgeted under the previous system at actual cost. Indeed, since the budgets used were drawn from the base period 6/1/70-5/31/71, the updating of such items exceeded the statutory requirement. Further, in addition to updating based on the CPI — Urban Average, some items were updated on the basis of actual pricing surveys, while for others the CPI figure was checked for validity against local prices. Perhaps the most important example of this last form of updating was the increase in the maximum sum allowed for private shelter, from $125 to $160, for the period from 2/62 (when the maximum had been established) to 7/69. [4] The State points to an article from the Department of Labor’s Monthly Labor Review for October, 1973, indicating that “a new set of consumer price indexes which measure price changes in urban areas grouped by regions” had been developed. Plaintiffs respond by pointing to Jackson v. Department of Public Welfare of State of Florida, 317 F.Supp. 1151, 1161 & n.9 (M.D.Fla.1970), which referred to Florida’s use of “the Consumer Price Index for the southeastern region of the United States.” [5] By Public Act 74-244 the Connecticut General Assembly amended § 17-2(a) of Title 17 of the Connecticut General Statutes so as to direct the Welfare Commissioner to compute, at least annually, “a redetermination and such revisions to all components of the standards of need for the several programs administered by the department so as to reflect changes in living costs using the current federal regional consumer price index.” This statute, however, does not moot this branch of the case because it appears that the figures originally used in determining CFAP remain as the base. [6] Section 402(a)(23) also requires proportional adjustment of “any máximums that the State imposes on the amount of aid paid to families.” Although some courts appear to have taken the term “máximums” to refer to ceilings placed on the cost of items included in the standard of need, see Houston Welfare Rights Org., Inc. v. Vowell, 391 F.Supp. 223, 228-30 (S.D.Tex.1975), the better view is that this separate statutory requirement was intended to cover dollar máximums placed on the level of benefits, resulting in payments at less than the standard of need. That interpretation is the more natural reading of the statutory language and best explains the rather brief consideration given to this clause in Rosado. See Utah Welfare Rights Org. v. Lindsay, 315 F.Supp. 294, 299-300 (D.Utah 1970); see also Dandridge v. Williams, 397 U.S. 471, 482-83 & n.15, 90 S.Ct. 1153, 25 L.Ed.2d 491 (1970). Since Connecticut does not impose flat dollar máximums on the level of payments, this part of the statute is not relevant to the issues here raised. [7] See also fn. 14 to Chief Justice Warren’s opinion in that case, 392 U.S. at 318, 88 S.Ct. 2128, quoting, inter alia, Representative Vinson’s explanation that “need is to be determined under the State law.” 79 Cong.Rec. 5471 (1935). [8] Our conclusion that an action challenging an updating required by § 402(a)(23) is not to be used for microscopic judicial scrutiny of previous state plans satisfactory to HEW seems to accord with New Jersey Welfare Rights Organization v. Cahill, 349 F.Supp. 501 (D.N.J.1972), affirmed, 483 F.2d 723 (3 Cir. 1973), and with Jackson v. Dept. of Public Welfare, 317 F.Supp. 1151 (M.D.Fla.1970). In Roselli v. Affleck, 508 F.2d 1277, 1281-82 (1 Cir. 1974), the court gave qualified approval to a district court’s rejection of a Rhode Island plan on the ground that the updated figures for shelter were not meeting actual need as the state professed to be doing. As we read the opinion, the court was not holding, as a matter of federal law, that state standards must reflect a judicially-determined “actual” need; rather, relying on the district court’s findings, the court was merely saying that the state could not, in going to a flat-grant system, change the basis for determining need from that which it had imposed on itself. [9] The welfare commissioner shall determine the regional shelter component in the aid to dependent children program as follows: He shall designate three regions within the state based on the current market value for the rental of private housing available to welfare recipients and shall establish a maximum payment in each such region which shall not exceed the mean rent in each region. The level of assistance payment to each recipient in that program shall be in an amount equal to the existing flat grant award for payment of rent to such recipient, except that if the actual rent paid by such recipient is greater than the existing flat grant award, such payment shall be adjusted by an amount equal to the difference between the existing flat grant award and the actual rent paid by the recipient up to the established maximum. [10] Since relief in this case can only be prospective, Rothstein v. Wyman, 467 F.2d 226 (2 Cir. 1972), cert. denied, 411 U.S. 921, 93 S.Ct. 1552, 36 L.Ed.2d 315 (1973), and since the effect of the statute is to make the level of benefits for shelter at least equal to actual rent (except where limited by “the established maximum”), it could be argued that the statute makes any claim against the computation of the shelter component of CFAP moot. However, some recipients, whose actual cost for shelter is less than the flat-grant sum, will still be receiving aid at the flat-grant rate. More important, a higher flat-grant sum will make some families eligible for AFDC aid who would otherwise be denied it, and eligibility under the categorical assistance program brings with it eligibility for several other programs as well. While increasing the number of persons eligible for these programs may not have been part of the Congress’ intent in passing § 402(a)(23), there is no denying that it is part of the provision’s effect. Compare Rosado, 397 U.S. at 413, 90 S.Ct. 1207, with Jefferson v. Hackney, 406 U.S. 535, 543-44, 92 S.Ct. 1724, 32 L.Ed.2d 285 (1972). Accordingly, although plaintiffs’ claims have lost much of their emotional force as a result of the 1974 and 1975 legislation, they have not been rendered moot. [11] One further shelter-allowance issue raised by the plaintiffs is that it was erroneous to include within the computation of CFAP budgets for rents which did not meet actual rental costs because the allowance for private housing was limited to $160 per month, plus 10% for furnished housing. The district judge rejected this contention, 353 F.Supp. at 77, on the basis of a separate housing survey conducted by the defendant which was alleged to have shown that, out of 463 cases randomly selected, in only 7, or about 2%, did the budgeted rent fail to meet actual need, and then only by an average of about $14 per month; the judge regarded this as de minimis. Plaintiffs, in their post-trial motions, indicated that they wished to tender new evidence assailing this survey and the resultant finding. Given our uncertainty as to the underlying factual situation, we hesitate to rule on the lawfulness of averaging together maximum and below-maximum payments where the maximum has been adjusted to account for inflation, as was here the case. See Houston Welfare Rights Org., Inc. v. Vowell, 391 F.Supp. 223, 238-30 (S.D.Tex.1975). Accordingly, we think it better to leave this issue open on remand, for consideration in light of the further evidence that may be introduced and the principles established in this opinion. [12] Before the trial court plaintiffs claimed that the potential population from which the sample had been drawn was in fact some 3,000 units smaller than it should have been. The district court dismissed the contention, 353 F.Supp. at 81, on the basis that there was no showing that the omission of these units had in any way skewed the results actually obtained. Neither in their post-trial motions nor before this court have plaintiffs indicated that there is any additional evidence to this effect. [13] The State’s statistical expert testified that, to meet plaintiffs’ proposed standard, it would be necessary to have a sample as large as about 96% of the entire AFDC population. One of the plaintiffs’ witnesses, although not willing to give a precise figure, agreed that the necessary sample might be very large. [14] The record does not establish how many families in the AFDC population would fall into each of the units from 10 to 15, although there was evidence that for size 15 there were only one, or perhaps two, recipient families. [15] We see no contradiction between our holding on this point and our ruling, supra, that the State could lawfully amalgamate the budgets of assistance units equal to housing units with unequal ones. The issue here is not the *1240 total merger of “statistically distinct populations,” but rather the unfounded assumption that assistance unit sizes can be merged for certain purposes while state policy continues to treat unit size as the crucial variable in its assistance plan as a whole.