Aspira of New York, Inc. v. Board of Education of City of New York

394 F. Supp. 1161, 1975 U.S. Dist. LEXIS 12183

Opinion

MEMORANDUM ON TESTING PROCEDURES

FRANKEL, District Judge.

Implementing the principles of Lau v. Nichols, 414 U.S. 563, 94 S.Ct. 786, 39 L.Ed.2d 1 (1974), the court’s consent decree dated August 29, 1974, states the plaintiff class of Hispanic children, whose “English language deficiency prevents them from effectively participating in the learning process and who can more effectively participate in Spanish”, shall receive a program including intensive training in English language skills, instruction in substantive courses in Spanish, and reinforcement of Spanish language skills. The decree goes on to provide in detail for a course of testing to (in effect) identify the members of the class, i. e., those whose English language difficulties prevent them from effectively participating in the learning process and who could better learn in Spanish, and then for the program of instruction these class members are to receive. The testing procedures for determining which students are eligible for the program have now generated some interesting and difficult questions on which the parties are in sharp disagreement.

The testing program formulated by defendant Board of Education begins with a group of tests called the “language assessment battery [L.A.B.]— English version.” In a first (and seemingly noncontroversial) step; this L.A.B. has been given to a sample population of English-speaking students whose performance was scored and who served as the “norming” group. Next, the same test has been given to all Hispanic students, from among whom will be selected those who will be entitled to the program of bilingual instruction. The third step comprises what has been called the “norming process.” Defendant Board, acting on the judgments of its experts, had determined before the court heard' oral argument yesterday that

(1) a Spanish-version L.A.B. would be given only to those Hispanic students whose scores fell below the 10th-percentile score of the norming group; 1 and

(2) from among those thus given the Spanish version, the bilingual program would then embrace the students who scored better on this version and were thus designated as being able to “more effectively participate in Spanish.”

Just prior to yesterday’s oral argument, defendants evolved a modification of the foregoing position: It is now proposed that the Spanish version of the L.A.B. will also be “normed” from a sample student population which is predominately Spanish-speaking. The position urged at oral argument is that the universe of Hispanic students taking the Spanish version would be ranked on the basis of percentiles taken from the Spanish-speaking sample, and that all Hispanic students scoring below the 10th percentile would be excluded from our class on the ground that they are shown in this fashion to be unable to participate more effectively in Spanish.

This last revision, along with other things, has been the subject of considerable discussion and some dispute. The central focus of controversy, however, is the defendants’ position that only Hispanic students scoring below the 10th percentile on the English version should be eligible at all for testing with the Spanish version of the L.A.B.

As their broadest position plaintiffs urge that there should be no cutoff at all — that every Spanish-surnamed student should receive the Spanish L.A.B. and be assigned to the bilingual program if he scores better on this than on the English version. The court has rejected this view primarily for laches; 2 most probably, it would be rejected in any event on the merits. '

We come then to the question as to where the cutoff should be. Defending' their judgment, defendants say “that all monolingual English children should be capable of effective participation in English language instruction” and that any Spanish-speaking or Spanish-surnamed student who is able to score above the bottom 10% of the English-speaking or norming group “may be assumed capable of effective participation in instruction in English.” 3 Plaintiffs have attacked the entire testing procedure and, having been rebuffed by the court in their attempt to have all Hispanic students tested, now seek to “test those 200,000 Hispanic children [out of an overall total of 300,000] receiving the lowest scores on the English version of the LAB,” 4 that being the number of Spanish version tests which have been printed. Plaintiffs’ experts have not suggested any specific percentile as a suitable cutoff point.

The most vivid point to emerge from all the argumentation is that we confront an enormous amount of speculation and uncertainty. Defendants’ leading expert, Dr. Anthony J. Polemeni, Director of the Office of Educational Evaluation of the New York City Board of Education, explaining the Board’s position by affidavit and informal presentations in open court, stresses persuasively that the Board has been called upon for a “pioneering” endeavor; that assessing comparative language skill in the fashion our decree requires has not heretofore been attempted on anything like the scale involved in this case; and that key requirements of information and analysis (for example, finding measures of equivalency or comparability between the Spanish and the English L. A.B. tests) remain unsupplied as the time to move speeds for all of us.

Without approaching confidence or certainty, the court is persuaded at this time, for at least this year’s efforts, that the 10% cutoff point may not be deemed consonant with the purposes of the decree. The premises upon which defendants offered this basic item include the assertion that “zero percentile could be a reasonable cutoff point because that is the point at which monolingual English speaking students are not effectively participating in the learning process.” 5 The quoted statement reflects an understanding of “effectively participating” which is not in keeping with the concept as it appears in the court’s decree. It should be evident that the group defendants correctly characterize as “monolingual English children” includes some unknown percentage — disturbed, deprived, culturally ghettoized— who do not effectively participate at all in the English spoken by their teachers and fellow students. The fact that these students could not be aided by a bilingual program — a matter emphasized in defendants’ contention that we must not do excessive “justice” for the plaintiffs herein — is not material to the court’s present problems. Our jurisdiction does not extend, in this or any case, to all the world’s ills. The case centers only upon students who may be suffering educational deprivation because they “do not understand English,” Lau v. Nichols, supra, at 566, 94 S.Ct. 786. Our interest at this juncture in monolingual English children, whatever their separate problems may be, is to compare their linguistic skills with those of Hispanic children so that the class of those entitled to the bilingual program under our decree may come to be defined with as much fairness and precision as the circumstances allow.

Hewing to this essential concern, the court is impressed that the 10% cutoff point is presented here without anything fairly to be deemed a rational basis. The court has been given sheets showing score distributions for the English sample population. The scores do not cluster notably above the 10th percentile. They tend to be dispersed rather widely below that point. There are no facts to explain the degree to which difficulties irrelevant to our concerns may account for the scores of students below the 10th percentile. There is, in short, no way of knowing how many of the English-speaking children in the lowest 10% are “effective participants” in any sense which could be deemed useful for our purposes.

Related to this vital gap in defendants’ submissions — and more critical, of course, in this case — is the absence of persuasive ground for believing that scores below the 10th percentile will fairly and sufficiently identify the Hispanic children “whose English language deficiency prevents them from effectively participating in the learning process x- # -x-.” The language just quoted is the first half of the definition of the children in plaintiff class; when that portion of the definition is satisfied, the task is to proceed in addition to see whether such children “can more effectively participate in Spanish * * *.” 6

In the standard procedures of administrative law, if time permitted thorough and orderly development, the court might remand the problem to the Board for study, analysis, and a better reasoned solution. But time is among the luxuries in shortest supply in this case. Our paramount obligation is to move toward the implementation of the decree, already approaching its first birthday, in the coming school year. Testing in Spanish is scheduled to begin May 30. The problem before us came on to be considered less than a week ago. While the court must not be profligate with the resources of a financially harried City, the mandate of the law forbids our defining plaintiff class with a stringency calculated to exclude substantial numbers who ought to be included.

With such considerations in view, the court concludes that the cutoff point should be determined now, accepting the certain imperfection guaranteed by the array of uncertainties attending this decision. As has been noted, the assertedly “ideal” view of plaintiffs — to test all Hispanic students in Spanish and give the bilingual program to all who do better in Spanish than in English — is not accepted. In addition to the reasons noted earlier, we may observe that the decree is not meant to enroll for bilingual instruction all who are more fluent in Spanish than in English. The setting and the goal remain a course of English-language instruction. So those who can now participate “effectively” in English are outside plaintiff class, whatever their relative fluency in Spanish may be.

We have examined with maximum feasible attention the distributions of scores in the English-speaking “norming” group. As a general matter, the scores scatter widely over the lowest 10th percentile and on up to the 20th percentile. There is, again describing the data generally, a sharp tendency to cluster at or above the 20th percentile.

Apart from what the score distributions show, it seems reasonable to assume, subject to what further experience may reveal, that a Hispanic student scoring better than a fifth of his English-speaking peers on the English-version L.A.B. has a level of proficiency enabling him to participate effectively in English-language instruction.

In sum, for the reasons thus adumbrated, the court holds that the Spanish L.A.B. will be given to all Hispanic students whose English L.A.B. scores fall below the 20th percentile score of the norming group. In other words, the students with such scores are to be deemed prima facie, in terms of the decree, those “whose English language deficiency prevents them from effectively participating in the learning process * * *»

Defendants’ worry about over-inclusion is not eliminated, but may be allayed in part, by the next step in defining the class. While this step is fraught with rough and unscientific assumptions, it is the most acceptable one the parties and the court have been able to evolve for the time being. As has been noted, defendants propose to “norm” the Spanish version L.A.B. by reference to a student sample population which is predominantly Spanish-speaking. The court approves this device in principle for the immediate needs of implementing the decree. Defendants have then proposed that a cutoff be employed again — namely, that the Hispanic students scoring below the cutoff in the Spanish battery be deemed incapable of participating effectively in Spanish, and therefore be relegated on this ground to English-language instruction outside the program given to the plaintiff class. The court does not adopt this proposal, but accepts a portion of the rough hypothesis upon which it rests. Specifically, it is ordered that those Hispanic students who take the Spanish-version L. A.B. (i.e., who fell below the 20th percentile on the English) will be included within the plaintiff class only if, in addition, they score at a higher percentile on the Spanish than they did on the English. For example, a student whose English score is at the 18th percentile will be within the plaintiff class if his or her Spanish score is at the 19th percentile or higher. A Spanish-version score at or below the 18th percentile would place the student outside the plaintiff class.

The crudity of this formulation is acknowledged on all sides. It is not possi ble to say with precise and certain meaning that an English-version score at a given percentile is similar to the same percentile score on the Spanish version. Certainly distinctions between students separated by a percentile will produce results that must seem capricious at the points of division. But we are merely a court, consigned to the drawing of lines, and we do the best we can. Thus far at least, nobody has proposed anything better. The percentile technique has the virtue, if not of identifying absolutely comparable levels of linguistic proficiency, of ranking people comparably by numerical standing within each of the linguistic scales. And all that we do today is surely open to improvement as the parties and their experts acquire — and may manage to infuse the court with — the wisdom available from further experience.

To summarize: Hispanic-surnamed students who scored below the 20th percentile on the L.A.B.-English will take the L.A.B-Spanish. Those whose scores on the latter exceed their scores on the former are to be in plaintiff class.

It is so ordered.

1 . For those not more comfortable than the court with such conceptions, this means, say, that if 10% of the norming group scored below 60 on the English L.A.B. and 90% above 60, the cutoff score for the Hispanic students will be 60, with those at or below this to receive the Spanish L.A.B.

2 . The broad argument was presented to Special Master Morris P. Glushien and decided adversely to plaintiffs on February 26, 1975. Plaintiffs made known to the court that they were not pleased with that ruling and might well desire review of it at some time. By letter dated March 24, 1975, the court assured plaintiffs that postponement of a review proceeding would not be deemed a “waiver” of the point. The postponement continued, however, for much too long, causing the subject to come on now in circumstances of needless and unacceptable crisis. The present proceeding begins with an order to show cause dated May 21, 1975. The testing of Spanish-surnamed children was set some time ago to begin on May 30, 1975. A three-day holiday weekend has intervened between the order to show cause and its return on May 27. The court’s own schedule, which is not wholly irrelevant, includes criminal trial commitments, involving a defendant in jail, which are not justly postponable. In sum, the idea of laches, as it affects both defendants and the court, seems pertinent. This is not meant, as plaintiffs’ counsel have seemed to infer, as criticism of their performance. The court recognizes, and appreciates, that they have been generally prompt and diligent throughout this taxing litigation. The court has also considered their judgment that the broad argument against any cutoff was not fully ripened until just now. But it is this last judgment the court finds mistaken and inequitable in its practical impact.

3 . Better from Dr. Anthony J. Polemeni to the court, May 1, 1975, p. 3.

4 . Affidavit of Herbert Teitelbaum, sworn to May 26,1975, ¶ 4.

5 . Affidavit of Dr. Polemeni sworn May 23, 1975, ¶ 13.

6 . The quotations are from the definition of the class in decretal paragraph 3 of the decree.