Download as pdf or txt
Download as pdf or txt
You are on page 1of 59

At Ex Parte Motion Office -of the Supreme Court of the State of New York, held in and for the

County of New York at the Courthouse thereof, 60 Centre Street, New York, New York, on the day of
,2013

PRESENT: HON. Justice of the Supreme Court SUPREME COURT OF THE STATE OF NEW YORK COUNTY OF NEW YORK ................................................................... X In the Matter of the Application of,

R.B. on behalf of his minor child L.B.; A.K. on behalf of his minor child S.K.; S.R. on behalf of
his minor child H.R.; L.W. on behalf of her minor child E.W. individually and on behalf of all others similarly situated, Petitioners, for Judgment pursuant to CPLR Art. 78 and common law claims, -againstTHE DEPARTMENT OF EDUCATION OF THE CITY OF NEW YORK flWa THE BOARD OF EDUCATION OF THE CITY SCHOOL DISTRICT OF THE CITY OF NEW YORK; DENNIS WALCOTT, as Chancellor of the Department of Education of the City of New York; GENTIAN FALSTROM, as Director of elementary enrollment of the Department of Education of the City of New York; ROBERT SANFT, as Director of the Office of Student Enrollment of the Department of Education of the City of New York;

Unsi

Order

MAY 16 2013

Unsigned Order to Show Cause


-I

ORDER TO SHOW CAUSE


Upon the attached Verified Petition, duly verified on May 15, 2013, the Affirmation of

Supreme Court Records OnLine Library - page 1 of 59

Stewart Lee Karlin, dated the 1Shday of May 2013, affidavit of Petitioners R.B. on behalf of his minor child L.B. sworn to on the 15'h day of May, 2013; A.K. on behalf of his minor child S.K., sworn to on the 15'hday of May, 2013; S.R. on behalf of his minor child H.R., sworn to on the10th day of May, 2013, and L.W. on behalf of her minor child E.W.,sworn to on the 14" day of May, 2013; Alexey Kuptsov, sworn to on the l S hday ofMay, 2013; Bogdan Grechuk, swornto onthe 13'h

day of May, 2013; Jonathan Goodman, sworn to on the 1Oth day of May, 20 13; Tom Alberts, sworn
to on the 13thday of May 2013; Vladimir Finkelstein, sworn to on the 14Ihday of May 2013; Vladislav Kargin, sworn to on the 14thday of May 20 13, Alexey Onatskiy, sworn to on the 16* day of May 2013; and upon the exhibits attached to the Affirmation of Stewart Lee Karlin, the memorandum of law in support of this order to show cause and upon all prior pleadings and proceedings, and sufficient cause having been alleged therefor, it is hereby, ORDERED that Respondents THE DEPARTMENT OF EDUCATION OF THE CITY OF NEW YORK fMa THE BOARD OF EDUCATION OF THE CITY SCHOOL DISTRICT OF THE CITY OFNEW YORK; DENNIS WALCOTT, as Chancellor ofthe Department of Education ofthe City of New York; GENTIAN FALSTROM, as Director of Elementary Enrollment of the Department of Education of the City of New York; ROBERT SANFT, as Director of the Office of Student Enrollment of the Department of Education of the City of New York; appear at a hearing and show cause at IAS Part

,Room

,of this Court, to be held at the Courthouse, 60


day of

Centre Street, New York, New York, 10007 on the

, 2013, at

o'clock in the (a.m. p.m.) or as soon as counsel may be heard why a preliminary injunction should not be made and entered pursuant to CPLR 630 1 et seq. ordering Respondents to enjoin selecting students for the Gifted and Talented Program using the current methodology and shortening Respondent's time to respond the Article 78 proceeding and why an ORDER should

Supreme Court Records OnLine Library - page 2 of 59

not be made

a.

Declaring pursuant to Article 78 of the Civil Practice Law and Rules that the notice provided by Respondents as to the change in the Gifted and Talented Program Selection Process was arbitrary and capricious and violated applicable laws;

b.

Declaring pursuant to Article 78 of the Civil Practice Law and Rules that the methodology for selection to the Gifted and Talented Program in its current form is arbitrary and capricious and violates Petitioners right to equal protection pursuant to the New York State Constitution;

c.

Declaring and Ordering pursuant to Article 78 of the Civil Practice Law and Rules that selection to the Gifted and Talented Program be completed pursuant to the procedure set forth in the G&T Handbook as originally published, prior to any amendment, with selection based on the students composite score and sibling priority only effecting selection of students in cases in which two students have the same composite score.

d.

Such additional relief as this Court may deem just and proper under the circumstances.

IT IS HEREBY FURTHER ORDERED that sufficient cause having been shown, the time
to answer or otherwise respond to the verified petition is May
20 13.

IT IS HEREBY FURTHER ORDERED that sufficient cause having been shown, therefore
pending the hearing of Petitioners application for a preliminary injunction pursuant to CPLR 6301 et seq. the Respondents are temporarily restrained from selectingstudents for the Gifted and Talented

Supreme Court Records OnLine Library - page 3 of 59

Program.

IT IS HEREBY FURTHER ORDERED that sufficient cause having been shown, that
Petitioners to protect the educational privacy rights of their minor children can use initials in the caption, Sufficient cause being alleged thereof, let personal service of a copy of this order on the New York City Department of Education and the Law Department of the City of New York, and the papers upon which this order is granted, upon the Attorney representing Respondents on or before the
~

day of May 2013 be deemed good and sufficient service. An affidavit or other proof of

service shall be presented to this Court on the return date directed in this order.

ITIS FURTHER ORDERED that if Respondents should file aresponse to the motion, it shall
file such response with the Clerk of this Court after service of a copy thereof upon Plaintiffs counsel
at his office at 9 Murray Street, Suite 4W, New York, NY 10007 on or before (a.m. p.m.) on the

, day of May 2013.

JUSTICE OF THE SUPREME COURT

Supreme Court Records OnLine Library - page 4 of 59

SUPREME COURT OF THE STATE OF NEW YORK COUNTY OF NEW YORK ................................................................... X In the Matter of the Application of, R.B. on behalf of his minor child L.B.; A.K. on behalf of his minor child S.K.; S.R. on behalf of his minor child H.R.; L.W. on behalf of her minor child E.W., individually and on behalf of all others similarly situated, Petitioners, for Judgment pursuant to CPLR Art. 78 and common law claims, -againstTHE DEPARTMENT OF EDUCATION OF THE CITY OF NEW YORK fWa THE BOARD OF EDUCATION OF THE CITY SCHOOL DISTRICT OF THE CITY OF NEW YORK; DENNIS WALCOTT, as Chancellor of the Department of Education of the City of New York; GENTIAN FALSTROM, as Director of Elementary Enrollment of the Department of Education of the City of New York; ROBERT SANFT, as Director of the Office of Student Enrollment of the Department of Education of the City of New York; Index No. :

VERIFIED PETITION

my 16 2013
a
---1

Petitioners by their attorney, Stewart Lee Karlin, allege the following: This special proceeding is brought to challenge respondents' conduct regarding the flawed admission process for the gifted and talent program by the New York City Department of Education that is arbitrary and capricious and denies Petitioners' equal protection under CPLR Article 78 and the Equal Protection Clause of the New York State Constitution.

Supreme Court Records OnLine Library - page 5 of 59

PARTIES

1.

Petitioner R.B. is the father of L.B., a child who resides in County of New York,

State New York. 2. Petitioner A.K. is the father of S.K., a child who resides in the County of Kings,

State of New York.

3.

Petitioner S.R., is the father of H.R., a child who resides in the County of Kings,

State of New York.


4.

Petitioner L.W., is the mother of E.W., a child who resides in the County of Kings,

State of New York.


5.

Upon information and belief, Petitioners claims for relief are brought on their own

behalf and on behalf of all those similarly situated. Respondents have acted or refused to act unlawfully, arbitrary and capriciously and in bad faith to the named Petitioners and to all those students seeking placement in the DOE Gifted and Talented Program who have scored highly on the G&T Test and are not provided placement to the program through the sibling priority provision.
6.

Respondent NEW YORK CITY DEPARTMENT OF EDUCATION fMa THE

BOARD OF EDUCATION OF THE CITY SCHOOL DISTRICT OF THE CITY OF NEW YORK (hereinafter DOE) was and still is a governmental agency organized and existing under the laws of the State of New York with its principal place of business located in the County of New York and is charged with responsibility of public education in the City of New York including providing education to the Petitioners.
7.

Respondent DENNIS WALCOTT (hereinafter Walcott) is the Chancellor of the

Department of Education and such is responsible for the operation of and is the chief executive officer of the New York City Department of Education fMa the Board of Education of the City
2

Supreme Court Records OnLine Library - page 6 of 59

School District of the City of New York.

8.

Respondent GENTIAN FALSTROM (hereinafter Falstrom) is the Director of

Elementary Enrollment and upon information and belief is in charge of the Gifted and Talented Program of the New York City Department of Education fWa the Board of Education of the City School District of the City of New York.

9.

Respondent ROBERT SANFT (hereinafter Sanft) is the Director of the Office of

Student Enrollment and upon information and belief overseesthe placement of students in the Gifted and Talented Program of the New York City Department of Education fMa the Board of Education of the City School District of the City of New York.
FACTUAL ALLEGATIONS COMMON TO ALL PETITIONERS1 Improper Notice of Changes to the G&T Selection Process Violated Petitioners RiPhts

11.

Since approximately 2007, The DOE offered to qualifying residents of New York

City the opportunity to attend special public elementary schools and classes. 12. The schools were meant to provide classes in which the students were selected from

the most academically gifted to enable those schools and classes to, inter alia, to teach more advanced subject matters and provide a more accelerated academic curriculum. 13. These new schools and classes are commonly referred to as Citywide Gifted &

Talented (G&T) Schools and these new classes are commonly referred to as District G&T classes. Herein after, the Citywide Gifted & Talented Schools will be referred to as Citywide G&T Schools and the District G&T classes will be referred to as the District G&T Classes.
14.

Currently, under the administration and control of the Respondents, the DOE

The factual allegations common to all Petitioners are derived from the affidavits of each Petitioner. Please see Exhibits 4,5,6,7 to the Affirmation of Stewart Lee Karlin.

Supreme Court Records OnLine Library - page 7 of 59

operates five Citywide G&T Schools and approximately 70 G&T Classes for K grade that operate with the normal district public schools of New York City. 15. Annually, the Respondents have published a Gifted & Talented Program

Handbook that provides the primary means of communication to the public on, inter alia, the admission criteria used by the Respondents to place children into the Citywide G&T Schools and District G&T Classes.
16.

In 2012, the Respondents cause to be published the Gifted & Talented Program

Handbook 2012-13 For students born in 2008 (The G&T Handbook) that provided detailed information about the Gifted & Talented Program including, inter alia, student assessment, assessment reliability and validity, scoring the test, placement eligibility for both District G&T Schools and Citywide G&T Classes. 17. By way of the G&T Handbook the Defendants made the following statements:

ImDortant Dates for the 2012-2013 Admission Process: -November 9, Deadline to submit Request for Testing (RFT) forms -October 15-November 9,20 12 Community Information Sessions -January 5,6, 12,13, 19,20,26,27; February 2 and 3,2013: Current DOE Pre-K Students and Non-public School Students G&T assessments administered at selected sites. -The Assessments: -The [Department of Education] will use two assessments that have been used in many districts across the country. The combination of verbal test items from the Otis-Lennon School Ability Test (OLSAT) and nonverbal test items from the Naglieri Nonverbal Ability Test (NNAT2) provide a thorough and balanced look at each childs intellectual abilities to identifj students who are eligible to apply for a G&T program. Tests are administered by NYC certified teachers who are trained by Pearson Assessments. -Assessment Reliability & Validity: - Each test item on the Verbal and Nonverbal Assessments has been statistically analyzed and evaluated for difficulty, reliability, fit, and effectiveness across each age group. All of the items were rigorously reviewed by educators, measurement specialists, and psychologists 4

Supreme Court Records OnLine Library - page 8 of 59

to ensure that the test is of high quality and without bias toward any subgroup, including gender and ethical/racial subgroups. The items have also be reviewed for clarity, appropriateness of content, accuracy of correct answers, plausibility of answer options, and appropriateness of vocabulary. -Both tests have been demonstrated to be reliable and valid assessments according to official studies, Reliability refers to the accuracy and precision of the test scores. Validity refers to the extent to which the test measures what it is intended to measure. -Scoring the Test -Calculating the Score -The scores of the assessment will be combined to generate a percentile rank. This year, the scores will also be combined to generate a composite score. Scores will be calculated based on the weight of each assessment and the students age. The verbal score will be weighted approximately 1/3 and the nonverbal score will be weighted approximately 2/3. -What is a percentile rank? - Percentile rank refers to a students relative standing in comparison to other students of the same age. A percentile is not the same as percent correct. - Percentile ranks are useful in showing the students standing within a group, but should not be used in describing differences between the scores of two or more students. Percentile rank is generated based on the childs age; each child is compared to others within the same three-month age band. Therefore, all percentile ranks show the students standing in comparison to other students of the same age, not grade. -What is a composite score? -A composite score is a numeric value within the range of 200 to 900 that provides a description of a students combined performance. Students within the same percentile rank may have different composite scores. -Placement Eliaibilitv -Eligibility for District G&T Programs: Your child must score in the 90 percentile of above. 1. Your child must live in New York City both at the time of RFT submission 2. and time of application. -Eligibility for Citywide G&T Programs: Your child must score in the 97thpercentile of above. 1. Your child must live in New York City both at the time of RFT submission 2. and time of application. -Completing the Application -Students who are entering kindergarten should participate in the kindergarten admission process.
5

Supreme Court Records OnLine Library - page 9 of 59

-Placement -G& T placements are based on composite test score, sibling priority, family preferences (ranked program choices), and the district to which you are zoned to elementary school (zoned district.) Students will only be placed in schools that have declared available seats. -There is no guarantee that a student will receive a placement offer to a G&T program. The number of eligible students typically exceeds the number of seats available at these programs. -Citywide Program Placement -Applicants with the highest composite scores will be placed first. (Emphasis supplied) Among students with the same score, siblings will be given first priority, then assignments will be made by lottery. -District Promam Placement -Each applicant eligible for placement will have priority for one or more district programs based upon the district to which they are zoned. Families may apply to any program in any district across the city, but applicants with priority to the district will be placed first. Among applicants from within the district, those with the highest composite scores will be placed first. Among students with the same score, siblings will be given first priority. -Placement ExceDtion Request -Siblings: -If a younger sibling does not receive an offer for his or her older siblings G&T program, you may file a [Placement Exception Request] for the younger sibling to attend the general education program offered in the same school. 18. During the period October 15,2012 to November 9, 1012 the Respondents held

Community Information Sessions at various public venues throughout the five boroughs of New York City, and made statements consistent with the sections of the G&T Handbook quoted in paragraph 16 above. 19. At the Community Infomation Session the Respondents through their agents

explained that the children would be placed in the Citywide G&T Schools in priority order determined by their composite score, and a sibling would only take priority if he/she had an equal composite score.
20.

Respondents agents also explained the children would be placed in the District G&T

classes in priority order determined by their composite score in combination with residency within
6

Supreme Court Records OnLine Library - page 10 of 59

the district school zone, such that a sibling would only take priority if both children lived in the same school zone and he/she had an equal composite score. 2 1.

In addition, the Respondents agents explained the benefit of registering your child

and requesting testing for their G&T assessments via Respondents website would cause the Department of Education to provide, via email, information about the G&T program pertinent to the students application and admission to the Citywide G&T and District G&T programs. Examples, such as receiving the childs test date location and the childs test scores via email were provided to encourage the parents of students to register their children via the website. 22. The admission placement procedures stated in the G&T Handbook was consistent

with Chancellors Regulation A-10 1 dated March 2,20 12 which controls the admission process of public schools in New York City. 23. 24. The deadline for all students to request testing was November 9,2012. On or before November 9,2012, Petitioners requested testing for their eligibility for

the Citywide G&T Schools and District G&T Classes. Each Petitioner, by way of their legal guardian and/or parent, made their request online via the Department of Education website, consistent with the instructions and paper application included within the G&T handbook. 25. On or before October 26,2012, the Respondents submitted an amendment to

Chancellors Regulation A- 101, regulating admissions into New York City public schools, inter

alia, at the elementary level, to the City Board in accordance with Education Law 2590-g.
26. On October 26,20 12, in accordance with Education Law 2590-g, the City Board was

required to make public and seek public comment relating to the proposed amendment by posting the required notice on the city boards official internet website. 27. Furthermore, Education Law 259O-g(8)(a)(i) requires that the notice include a
7

Supreme Court Records OnLine Library - page 11 of 59

description of the subject, purpose and subject ofthe proposed item under consideration. That notice failed to provide any information suggesting that the Respondents were seeking amendment of Chancellor Regulation A-101 to place children into the Citywide G&T Schools or District G&T Classes by any other method than described and outlined in the G&T handbook, except that the notice erroneouslysuggested that the Respondents were seeking to remove siblingpriority altogether where a sibling and non-sibling has the same composite score. 28. The October 26,2012, notice did not suggest that the Respondents suggested

amendments would give siblings more priority over non-siblings, and failed to mention any proposed changes that would suggest the Respondents would use the proposed amendment to not use composite scores to place students into the Citywide G&T Schools and District G&T Classes. Specifically, the posted notice included the following statement: -Description of the subject and purpose of the proposed item under consideration. Chancellors Regulation A-1 01 sets forth the policies concerning admission, discharge, and transfer of pupils in the New York City public school system. The prior version of the regulation, which was issued on March 2,2012, will be amended in the following respects: (1) sibling priority for G&T programs is changed to remove first priority for siblings and sibling priority for students with disabilities 29. The October 26,2012, public notice therefore states that the amendment being

sought to Chancellors Regulation A-1 0 1 would in fact remove any remaining priority in placement for siblings into the Citywide G&T Schools and District G&T Classes and nothing else.

30.

On or before November 5,2012, Respondents submitted a revised amendment to

Chancellors Regulation A-101 to the city board in accordance with Education Law 2590-g.
3 1.

On or about November 5,2012, in accordance with Education Law 2590-g, the City

Board posted a notice of the revised amendment being sought to Chancellors Regulation A-1 01 on the city boards official internet website.

Supreme Court Records OnLine Library - page 12 of 59

32.

On or before December 5,2012, Respondents submitted an revised amendment to

Chancellors Regulation A-101 to the city board in accordance with Education Law 2590-g. 33. On December 5,2012, in accordance with Education Law 2590-g, the City Board

posted a notice of the revised amendment being sought to Chancellors Regulation A-10 1 on the city boards official internet website.
34.

The December 5,20 12, notice failed to provide any information suggesting that the

Respondents were seeking amendment of Chancellor Regulation A- 101 to place children into the Citywide G&T Schools or District G&T Classes by any other method than described and outlined in the G&T handbook, except that the December 5, 2012, notice erroneously suggested that the Respondents were seeking to remove sibling priority altogether where a sibling and non-sibling has the same composite score. 35. The December 5,2012, notice did not suggest that the Respondents suggested

amendments would give siblings more priority over non-siblings, and failed to mention any proposed changes that would suggest the Respondents would use the proposed amendment to not use composite scores to place students into the Citywide G&T Schools and District G&T Classes. Specifically, the posted notice included the following statement:

I. Description of the subject and purpose of the proposed item under consideration: Chancellors Regulation A-1 01 sets forth the policies concerning admission, discharge, and transfer of pupils in the New York City public school system. The prior version of the regulation, which was issued on March 2,2012, will be amended in the following respects:
(1) sibling priority for G&T programs is changed to remove first priority for siblings and sibling priority for students with disabilities.
36.

The December 5,2012, public notice therefore states that the amendment being

sought to Chancellors Regulation A- 101 would in fact remove any remaining priority in placement for siblings into the Citywide G&T Schools and District G&T Classes. 9

Supreme Court Records OnLine Library - page 13 of 59

37.

However, contrary to the public notices, on information and belief, Respondents

expanded sibling priority in placing students into both Citywide G&T Schools and District G&T Classes, in effect deliberately sandbagging the children without sibling priority without notice.
38.

The revised amendment to Chancellors Regulation A-101 submitted to the City

Board in accordance with Education Law 2590-g, including Section 1I.A. entitled Sibling Priorities is silent as to the applicability of any sibling priority being applicable to admissions into the Citywide G&T Schools and District G&T Classes. Draft amended Section II.A.3. states: Siblings of students already pre-registered or enrolled at an elementary school at the time of application submission are given priority for admission into elementary school programs for which they meet the eligibility requirements as described below, subject to available seats. 39. Because the Citywide G&T School eligibility requirements, nor District G&T Class

eligibility requirements are described in the revised amended December 5, 20 12 Chancellors Regulation A-1 0 1, the most reasonable interpretation of the amended regulation (and the previous October 25 and November 5 pre-iterations) is that sibling priority is not applicable to the Citywide G&T School and District G&T Classes admissions. Such an interpretation is consistent with the October 25, November 5, and December 5 public notices disseminated by the city board.
40.

On December 20,2012, the City Board voted to approved the amended Chancellors

Regulation A-101 and it issued December 22, 201 1. At no time prior to the issuance of the amendments incorporated in the December 22,20 12 Chancellors Regulation A- 101, did the make any reasonable steps to notifl the students, their parents andor the guardians of the students registered to take the G&T test of their intention to use the amendments to Chancellor Regulation A-101 as a basis to apply a more expanded sibling priority criteria to the admission process for Citywide G&T Schools and District G&T Classes. Nor did the Respondents state that they intended 10

Supreme Court Records OnLine Library - page 14 of 59

to forego the use of composite scores to rank students for purposes of admission into Citywide G&T Schools and District G&T Classes. The Respondents, having the email addresses, residential addresses and/or other contact addresses of every student and their guardian andor parents registered to take the G&T test, could easily have provided clear and fair notice to those persons who would be most likely be adversely affected by the Respondents interpretationof the December 22 amended Chancellors Regulation A- 101. 41. The Respondents amended the G&T Handbook (Amended G&T Handbook)

without providing notice to Petitioners or the students, and their guardian and/or parents, registered to take the G&T test. Further to this date, Respondents have failed to update the foreign language versions of the G&T Handbook which are featured on the DOES website. 42. On or about April 2012, the Respondents notified students via email andor U.S.

Postal Service whether the results of the G&T test, including test scores for both the verbal and nonverbal test, and included both composite scores and percentile rank for each test and the students overall percentile ranking. 43.
Summary

The Respondents violated the notice provisions required under the law.

44.

Petitioners have brought this action on behalf of their minor children. Respondents have discriminated against Petitioners based on their lineage by

45.

exercising a discriminatory methodology which gives priority to siblings of children already enrolled in Defendants Gifted and Talented program (G&T), despite the fact that these childrens scores are lower than other children who do not have siblings in the program. 46. The placement policy that the New York City Department of Education (DOE)

uses to determine the eligibility and placement of children into the New York City G&T Program 11

Supreme Court Records OnLine Library - page 15 of 59

is arbitrary and capricious, in bad faith, and violates the purpose and meritocractic nature of the G&T Program because it inequitably reduces the placement chances of very well qualified students. It violates equal protection rights, irrespective of lineage. The methodology used reduces the chances of otherwise well qualified students without a sibling in a G&T placement, by substantial reducing the chances of admission for overwhelming qualified students while giving priority to much lower scoring-unqualified siblings of students already enrolled in the G&T Program. 47. In addition, the methodology used to calculate childrens percentile ranking to

determine a childs eligibility and placement for the G&T Program is deliberately and in bad faith fundamentally flawed in order for the unqualified sibling of a G&T student to be able to gain admission into the G & T program solely at the expense of well qualified students who being denied admission into the program because of the limited amount of seats. Further, using such a flawed methodology undermines the very purpose of the G&T program.
The DOE Gifted and Talented Program

48.

The determination of whether a student should be placed in a gifted education

program has a profound impact on the students. The G&T program provides students with exceptional abilities with differentiated instruction because there educational needs require it. The program provides enhanced educational opportunities that enable them to succeed in school and allows for educational opportunities in the future due to the accelerated pace of the program.
Petitioners Application Process
49.

Petitioners started the application process for the G&T program on or about

October 2012. At that time, they learned that due to excessive numbers of qualifling applicants in prior years, the DOE would move from a percentile ranking system to a new, more granular scoring system, based on composite scores. Petitioners also learned that there would be a new, more
12

Supreme Court Records OnLine Library - page 16 of 59

difficult section of the exam, called the NNAT2 module, that was introduced to further differentiate among top scoring students and designed to negate any advantages of test preparation. 50. Petitioners also learned that the rule which provided priority to siblings of

currently enrolled students (the prior year, siblings had guaranteed placement if they met a certain threshold, i.e., 97* percentile for the Citywide G&T Program, and 90* percentile for the District G&T Program) would be eliminated. Siblings would only have priority over nonsiblings if composite scores were the same. 5 1. Although the DOE is aware of the errors in the methodology of placing students

in the G&T program, it had informed the parents that because of fear of public backlash, it would revert back to the erroneous percentile ranking system instead of the composite scores that are universally accepted as the fairest and most accurate way of ranking childrens test scores.
52.

The DOE continues to mislead the public by suggesting that their published

percentile ranking system is a meaningful and accurate reflection of the childrens actual percentile ranking. However, the scoring and placement policy was changed in the middle of the testing process, with very poor communication. The G&T Handbook categorically stated that the children would be placed in order of composite scores (i.e., their actual scores that have been mathematically adjusted to take into account age differences) and that siblings would only take priority in the placement procedure if the composite scores of a sibling and non-sibling child were tied. The Fall 20 12 information sessions presented the same explanations. (Exhibit 1-B).

53.

Nevertheless, on or about December 19* 2012, Petitioners received an email from

DOE announcing that the sibling rule had been reinstated, at the request of families with siblings.
However, the email never gave any specifics stating that the DOE would not use composite scores to place children. Nor were Petitioners informed that this issue was being considered or that there

13

Supreme Court Records OnLine Library - page 17 of 59

would be public hearings to discuss the proposal to reinstate siblingpriority positive discrimination. Petitioners believed at this time that the change did not effect them. 54. According to the December 19* e-mail, the siblings would get priority over non-

siblings as long as they attained a score sufficient to place them in the 97'h percentile rank for the Citywide G&T Program and 90* percentile rank for the District G&T Program. At that time, and until well after all the testing had been completed, the G & T Handbook continued to present composite scores as the key mechanism to place children into the G&T Program. The article fiom the NY times interpreted the DOE email the same as Petitioners did in that composite scores would be used for non-siblings. (Exhibit 7-B ).
55.

Immediately after Petitioners were provided with their child's scores, the Schools

participating in the G&T Program held open houses to aid parent decisions on what order to rank the schools in order of preference. However, for the first time, while touring these schools parents, including Petitioners, learned that the composite scores were scrapped and that, because of the large number of children falling within the 99* percentile rank, after all siblings had been placed any remaining places would be conducted by lottery among all children in the 99thpercentile rank.
56.

Furthermore, on April 20,2013, Petitioners were informed that Pearson (the

company hired to administer the testing by Respondents) had made a scoring error and as a result the number of qualified applicants under the percentile ranking system would double, to the point where there would now be even more applicants than the prior year which used an easier module to the NNAT2 module.
57.

Specifically, based on recent statistics released by the DOE, of the 13,629

children testing for Kindergarten in the 2013 G&T Program, 5322 (39%) ranked in the 90th percentile, 2771 (20%) in the 97thpercentile and 1480 (1 1%) in the 9gthpercentile. 14

Supreme Court Records OnLine Library - page 18 of 59

58.

Now after scores have been readjusted following the first scoring scandal when

DOE announced that its vendor Pearson made several errors in the scoring assessment, the chances to get placement in a citywide program were dramatically reduced. 59. The question turns to be a question of fairness and mathematical accuracy. There

are multiple errors discovered over the course of the application process, indicating that Pearson methodology is not reliable and therefore DOE shall make this years placement based on valid composite scores. In fact, the DOE identified an additional error in the Gifted & Talented test results provided by Pearson on or about May 8,2013. The error impacted the qualification status of 146 test takers. Of these: 82 students who did not previously qualify for G&T now qualify for district programs; 64 students who previously qualified for district programs now also qualify for citywide programs. 60. In addition, 159 students percentile rank improved within their eligibility

category: 98 students remain eligible for district programs but will receive a higher percentile rank within the 90th-96th percentile range; 36 students remain eligible for citywide programs but will move from the 97th to the 98th percentile; 25 students remain eligible for citywide programs but will move from the 97 or 98th percentile to the 99th percentile. (See DOE Press Release attached to the Affirmation of Stewart Lee Karlin as Exhibit 7-C). 6 1. Chancellor Walcott stated the following However, the Department has

determined that when Pearson went through this process, they failed to check the test date data they used in the score calculation against the test date data indicated on students answer sheets and did not identify that a default test date had been incorrectly applied. This failure to complete the basic quality assurance checks Pearson confirmed that they had completed is deeply disturbing; for this reason the Department of Education is reviewing a variety of options including terminating 15

Supreme Court Records OnLine Library - page 19 of 59

Pearsons contract. (Exhibit 7-C). 62. The 5 Citywide G&T Program Schools only have approximately 300 seats

available this year before the siblings with priority take away seats from higher scoring non siblings. Accordingly, even if Petitioners were to rank all 5 schools (which is impractical given the locations), the lottery chance of each of their children (who are all in the 99thpercentile) is approximately 1 in 5 if no siblings take seats, but given the historical trends of sibling enrollment, the chances are lower. Unfortunately, the DOE has not yet published the number of siblings taking seats, but from
a recent tour of one of the 5 Citywide G&T Program schools, Brooklyn School of Inquiry, the parent

coordinator stated that they have at least 10 qualifying siblings that will take 18% of all available seats. Assuming that trend extends across all 5 Citywide G&T schools, Petitioners are relegated to
a 1 in 6 chance. However, due the recent change on May 8,2013 that includes additional students,

being taken, Petitioners chances have lowered even further. 63. Sadly, even admission into the District G&T Programs will be via lottery with

Petitioners children and others similarly situated chances not much greater than for admission to the Citywide G&T Program schools. Even though there are some District G&T Program classes available to Petitioners children and others similarly situated, because sibling priority takes effect percentile, it is therefore that much easier for siblingsto qualify for admission at the much lower 90th (they only need to be in the top 39% ofNew York City G&T Test takers - i.e., 61th percentile using the G&T Test results as a comparator). Thus, within the District G&T Program, the discrimination against non-siblings is more apparent. But since the sibling placements for District Programs have not been published, it is hard to provide an accurate figure of Petitioners lottery chances for a District G&T Program. However, it is clear the chances of each Petitioner are not good. 64. In trying to understand how such a large number of students were ranked in the 16

Supreme Court Records OnLine Library - page 20 of 59

99th percentile, Petitioners carefully reviewed and researched the methodology used by the DOE and

Pearson (the company contracted by the DOE to analyze and generate the test scores) to assess and score this years test for the G&T Program. Petitioners requested and reviewed the available manuals and contacted the DOE and Pearson representatives for details of the methodology that was being used.
65.

Among the aggrieved parents was Professor Alexey Kupstov who conducted

research to verify and validate his initial suspicion. Prof. Kuptsov was quickly able to confirm his suspicions and identifj and prove that the methodology used by DOE and Pearson to assess and score this years G&T test is fundamentally wrong. 66. To be sure they were correct, Prof. Kuptsov contacted a number of respected

professors of mathematics at various well respected educational institutionsin the United States and abroad and asked them for their independent analysis of the scoring methodology used by the DOE to calculate the percentile rankings they intend to use to place children into the G&T Program.

67.

Their experts conclusions were unanimous-the

scoring methodology and the

way the DOE uses that methodology to estimate the 99 percentile ranking is fundamentally wrong. Because those experts felt so passionately that the DOES methodology is wrong and recognizing the implications on the children who took the test by which many would be unfairly kept from placement in the G&T program, many of those experts offered to provide their assistance in this matter without any expectation of payment. Some of those experts have filed expert affidavits in support of the allegations being made in this petition.
68.

Further, before bringing this action, Petitioners alerted the DOE and Pearson to

the errors identified, and attempted to explain their concerns in the hope that they would be reasonable and use the composite scores of children to place children into the G&T Programs. Upon

17

Supreme Court Records OnLine Library - page 21 of 59

information and belief the placement letters for the G&T Program are due to be sent to the applicants in the middle of June 2013.

69.

Starting in the latter part of April, 2013, the Petitioners contacted the DOE

through its web portal; The Petitioners had direct email contact with Members of the Panel on Educational Policy (Judy Bergtraum, Kevin Diamond, Dmytro Fedkowskyj, Tino Hernandez, Jeanette Moy, Diane Peruggia, Ian Shapiro, Patrick Sullivan, Milton Williams Jr); direct contact

with ; The New York City Council, Committee on Education, Members (Fernando Cabrera,
Margaret S. Chin, Daniel Dromm, Helen D. Foster, Daniel R. Garodnick, David G. Greenfield, Vincent M. Ignizio, Robert Jackson (Chair), Karen Koslowitz, Jessica S. Lappin, Stephen T. Levin, Deborah L. Rose, Eric A. Ulrich, James Vacca, Albert Vann, Mark S. Weprin ;theoffice of Public Advocate Bill de Blasio; and email and telephone correspondence with Adina Lopatin, the Deputy Chief Academic Officer for Performance, George Benners and Shira Kessar, Office of Academic Policy. (E-mail exchanges attached to the Affirmation of Stewart Lee Karlin as Exhibit 4-B).
70.

No one the Petitioners reached out to about their concerns showed any illingness

to conduct an independent investigation of their calculations. Although this is of public concern affecting over 35,000 children in NYC who took the G&T test this year, the parents of the Petitioners received very little interest from the governmental agencies andor officials. 71. A letter to Chancellor of the DOE was sent twice to request a meeting in person to

discuss the issue, emails were sent to city councils members and the members of Panel of Education, and the Petitioners attended a forum on Public Education to get their voice heard, unfortunately all with little success.
72.

Filing this law suit is Petitioners last resort to stop the unfair and arbitrary

placement of children into the G&T Program this year and expose the highly flawed methodology 18

Supreme Court Records OnLine Library - page 22 of 59

supported by multiple experts as set forth in their expert affidavits, who have on public interest grounds agreed to provide their unbiased opinions without compensation. 73. Every parent in NYC, whose child took the G&T exam this year should know that

DOE mislead all exam takers and their families about the methodology of the scoring result and was resistant to step up and admit its mistakes and errors because of a potential political backlash at the expense of fairness and methodological rigor. 74. The fact that the DOE originally abandoned the percentile rank methodology in

favor of the composite score for placing children into the G&T Program is evidence that the DOE

a s flawed. However, the DOE because of must have suspected that the percentile methodology w
the political backlash from a group of parents, decided to revert back to the percentile rank placement system. 75. The DOEs current methodology favors siblings because they have a significant

advantage in that a sibling child scoring in the top 39% of the G&T Program test takers (equivalent to 61st percentile) will be guaranteed placement in a District G&T Program class, yet a child scoring perfect scores in the case of Petitioners, will be placed in a lottery with very little chances of admission. 76. The DOEs discrimination is plain - it awards siblings priority into the G&T

Program by guaranteeing siblings placement as long as they attain a fairly low threshold score (as low as 61 percentile rank compared to the children taking the same test for the G&T Program). Yet non siblings scoring perfect scores are not guaranteed admission, but are forced into a lottery where they have a 1 in 6 chance of being admitted into the G&T Program. 77. The DOEs G&T methodology has irreparably harmed Petitioners and others in

Petitioners situation who are discriminated against based on their lineage. The G& T methodology 19

Supreme Court Records OnLine Library - page 23 of 59

is imperative as it has a direct impact on each of the Petitioners educational fiitures and those of other children in their situation.
Facts Particular to R.B. on behalf of his minor child L.B.

78.

Petitioner R.B. is the father of L.B., a child who lives in New York, New York. On or about April, 2013, the G&T Program test results were announced and L.B.

79.

scored at the highest level. Out of the maximum composite scores of 160 and 150 for the NAT2 and OLSAT8 modules respectively, his child L.B. scored: 160 in the NNAT2 and 138 in the OLSAT8. L.B. scored the absolute maximum score on NNAT, I60 out of 160 points, and 140 out of 150(138 out of 150 after Pearson error) on the OLSAT portion of the test. Exhibit 5-A). 80. L.B. got this high score despite his language disadvantage; he is bilingual with
(Fkhbit5at714;

Romanian as his first language; he only became fluent in English at 3.5 years old and Romanian is not one of the alternative languages offered for the G&T test. Despite the nearly perfect score, it is entirely possible that, L.B., will not get a placement in any citywide or district program as L.B.s chances have been diminished to approximately a 1 in 6 chance. (Exhibit 5 at f[ 21). For the less competitive district G&T schools, a sibling scoring only in the 90th percentile (39% of all NY test takers) has priority over a top-scoring student. The number of 99 percentile applicants in District 3 where L.B. resides, is almost twice as large than the total number of seats, even before considering siblings. (Exhibit 5 at f 14). 8 1. Despite achieving the highest possible score no combined composite scores were

provided in the letter, instead the DOE letter merely provided a combined percentile rank for stating that L.B. was in the 99thpercentile rank. (Exhibit 5 at 715). 82. After the scores have been readjusted following the first and second scoring 20

Supreme Court Records OnLine Library - page 24 of 59

scandals, E.W.s chances to get placement in a citywide program were dramatically reduced. (Exhibit 5 at 717, 19). 83. L.B. has been aggrieved by the discriminatory and arbitrary methodology utilized

by the DOE. Due to its bias and discriminatory methodology, L.B. has suffered irreparable harm since he has missed deadlines to apply to other schools. 84. The DOES G&T methodology has irreparably harmed L.B. and others similarly

situated who were discriminated against based on their lineage. The G&T methodology is imperative as it has a direct impact on my L.B.s educational future as well as those children similarly situated.
Facts Particular to A.K. on behalf of his minor child S.K.

85. 86.

Petitioner A.K. is the father of S.K., a child who lives in Brooklyn, New York. On or about April, 2013, the G & T Program test results were announced and S.K.

scored at the highest level. Out of the maximum composite scores of 160 and 150 for the NNAT2 and OLSAT8 modules respectively, S.K. scored: 160 in theNNAT2 and 150 in the OLSAT8. These were perfect scores, (Exhibit 6 at 7 17; Exhibit 6-A). 87. Despite achieving the highest possible score, no combined composite scores were

provided in the letter, instead the DOE letter merely provided a combined percentile rank for stating that E.W. was in the 99fh percentile rank. (Exhibit 6 at 7 18). 88. After the scores have been readjusted following the first and second scoring

scandals, S.K.s chances to get placement in a citywide program were dramatically reduced. (Exhibit 6 at 7 20-22). 89. Had A.K. known that his daughter S.K. was going to be exposed to this

discriminatory treatment, he would not have chosen to expose her to the emotional stress of taking 21

Supreme Court Records OnLine Library - page 25 of 59

an assessment test. In addition, had A.K. not been mislead, he would have explored alternative school options for his minor child, which are now foreclosed because of application deadlines. Thus, as a result of the DOEs methodology S.K. has been irreparably harmed. (Exhibit 6 at T 7). 90. The major concern of A.K. is that S. K., will not be challenged enough in school.

She is already well beyond the K level and putting her in general education class will do irreparable harm to her. There is a great risk that she will lose interest in learning just because she will not be challenged in the classroom. A.K. a teacher himself has seen many examples of students who are not provided with appropriate enrichment opportunities. S.K. is the type of student requiring accelerated curriculum which the G&T Program is intended to serve yet due to the discriminatory and arbitrary methodology applied by the DOE S.K has been rreparably harmed. 91.

S.K. has been aggrieved by the discriminatory and arbitrary methodology utilized

by the DOE. Due to its bias and discriminatory methodology, S.K. has suffered irreparable harm since she has missed deadlines to apply to other schools. 92. The DOEs G&T methodology has irreparably harmed S.K. and others similarly

situated who were discriminated against based on their lineage. The G&T methodology is imperative as it has a direct impact on my S.W.s educational future as well as those children similarly situated.
Facts Particular to S.R. on behalf of his minor child H.R.
93.
94.

Petitioner S.R., is the father of H.R., a child who lives in Brooklyn, New York. H.R. who was four years old at the time took the approximately 1-1.5 hour G&T

Test at the end of January 20 13. (Exhibit 4 at 112). The test was carried out in a separate room to which S.R. was no permitted to observe, a very stressful situation for both S.R. and H.R. (Exhibit
4 at 7 13)

22

Supreme Court Records OnLine Library - page 26 of 59

95.

On or about April, 2013, the G&T Program test results were announced and H.R.

scored at the highest level. Out of the maximum composite scores of 160 and 150 for the NNAT2 and OLSAT8 modules respectively, my child H.R. scored: 160 in the NNAT2 and 144 in the OLSAT8. (Exhibit 4-A). 96. Despite achieving the highest possible score no combined composite scores were

provided in the letter, instead the DOE letter merely provided a combined percentile rank for stating that H.R.. was in the 99* percentile rank. (Exhibit 4 at 7 15). 97. After the scores have been readjusted following the first and second scoring

scandals, H.R.s chances to get placement in a citywide program were dramatically reduced. 99.H.R. has been aggrieved by the discriminatory and arbitrary methodology utilized by the DOE. Due to its bias and discriminatory methodology, H.R.. has suffered irreparable harm since she has missed deadlines to apply to other schools. (Exhibit 4 at 77 38-40). 98. The DOES G&T methodology has irreparably harmed H.R. and others similarly

situated who were discriminated against based on their lineage. The G&T methodology is imperative as it has a direct impact on my H.R.s educational future as well as those children similarly situated. (Exhibit 4 at T[T[38-40).
Facts Particular to L.W. on behalf of his minor child E.W.

99. 100.

Petitioner L.W., is the mother of E.W., a child who lives in Brooklyn, New York. To take the G&T Test, E.W. spent 80 minutes with an unfamiliar person, one on

one, without a parent or guardian present, something that L.W. would never have exposed her to if she knew that the process is based on a lottery, with her chances to receive a seat in the G&T Program as low as 15%. (Exhibit 7 at T[ 12). 101. On or about April 20 13 the G&T Program test results were announced and E. W. a 23

Supreme Court Records OnLine Library - page 27 of 59

trilingual child, scored at the highest level. Out of the maximum composite scores of 160 and 150 for the NNAT2 and OLSAT8 modules respectively, E.W. scored: 160 in the NNAT2 and 150 in the OLSAT8. They were perfect scores. (Score Report attached to the Affirmation of Stewart Lee Karlin as Exhibit 7-A). 102. Despite achieving the highest possible score no combined composite scores were

provided in the letter, instead the DOE letter merely provided a combined percentile rank for stating that E.W. was in the 99thpercentile rank. (Exhibit 7 at 7 14). 103. After the scores have been readjusted following the first and second scoring

scandals, E.W.s chances to get placement in a citywide program were dramatically reduced. (Exhibit 6 at T[ 16-18).

104. E.W. has been aggrieved by the discriminatory and arbitrary methodology utilized
by the DOE. Due to its bias and discriminatory methodology, E.W. has suffered irreparable harm since she has missed deadlines to apply to other schools.
105.

The DOES G&T methodology has irreparably harmed E.W. and others similarly

situated who were discriminated against based on their lineage. The G&T methodology is imperative as it has a direct impact on my E.W.s educational future as well as those children similarly situated.
Analysis of the DOE G&T Test bv Dr. Alexev Kuptsov

106.

Dr. Alexey Kuptsov is currently a Professor of Mathematics at the New York

University (WYU), Courant Institute, located in New York, New York and has extensive experience in mathematics through his education, teaching, consulting and applied mathematics research. (Exhibit 1 at lTT[ 2-10) 107. Dr. Kuptsov examined the method used by the DOE to rank students taking the
24

Supreme Court Records OnLine Library - page 28 of 59

New York City 20 13 Gifted & Talented test for purposes of determiningthe eligibilityof the student to be considered for admission to the New York City Schools that offer Gifted & Talented (G&T) programs. (Exhibit 1 at 7 12) 108. The percentile numbers generated by the methodology used by Respondents are

statistically flawed and so cannot be used to select and identify, with sufficient degree of accuracy, those students who are the more academically gifted and so should be given preference to school admissions into the New York City G&T programs. Moreover, because of this mathematical flaw in calculating percentile rank, the percentile rank is not a legitimate mechanism by which to rank the qualifying students for admission into the G&T Programs. (Exhibit 1 at 7 13) 109. The New York 2013 G&T test involves two parts. A nonverbal part, the NNAT2

test and a verbal part, the OLSAT8 test. Each student taking the New York 2013 G&T test receives a normalized score for each part of the test which, in simplistic terms, is the childs actual score that is manipulated by mathematical equation to account for age difference between the students at the time they took the test. This is an important consideration since a child that just over 4 years old would be taking the same test on the same day as a 5 year old. So every student receives a normalized score for the NNAT2 portion of the test and every student also receives a normalized score for the OLSAT8 portion of the test which are later combined with weights 2/3 and 1/3 to produce a composite score. There is no issue with and he do not question the validity of the calculation used to generate the normalized and composite scores of the students. (Exhibit 1 at 7 14)

110. Dr. Kuptsov also reviewed the publications of the Respondents provided to the
public before and at the time of the deadline for students to request testing for admission into the
New York City G&T program, in particular the G&T Handbook attached to the Affirmation of

25

Supreme Court Records OnLine Library - page 29 of 59

Stewart Lee Karlin as Exhibit 1-B. Those publications made clear that the Respondents would use composite scores to rank students for placement into the G&T program. Specifically, the Respondents stated that a composite score based on the NNAT2 and OLSATS portions of the test would be used, weighted such that approximately 2/3 would be based on the NNAT2 portion of the test and approximately1/3 would be based on the OLSATS portion of the test. (Exhibit 1 at 7 15) 111.

An excerpt from the G&T Handbook published by the DOE makes clear that

qualifying students for the G&T Programs were to be placed in order of composite test score and not percentile rankings as it states: Gifted & Talented placement are based on composite test score, sibling priority, family preferences (ranked program choices), and the district to which you are zoned to elementary school (zoned district). (Emphasis added) (Exhibit 1-B at pg. 7). 112. Dr. Kuptsov has found no reasonable basis for believing that the composite scores

generated by the DOE for the students that took this years G&T test are in any way inaccurate. Accordingly, Dr. Kuptsov believes to a reasonable degree of certainty that the composite scores would provide a fair and mathematically sound way of ranking students for admission into the New York City G&T programs and would be consistent with the placement criteria outlined by the DOE. (Exhibit 1 at 7 17)
1 13.

Significantly after the students had requested testing for admission into the New

York City G&T programs, the DOE decided to not use the students composite scores to rank students for placement into the G&T programs, but instead decided to use percentile ranks instead. (Exhibit 1 at 7 18) 114. A percentile rank of a score is universally accepted and understood to be the

percentage of scores in its frequency distribution that are the same or lower than it. For example,
26

Supreme Court Records OnLine Library - page 30 of 59

a test score that is greater than or equal to 75% of the scores of people taking the same test is said to be at the 75th percentile rank. (Exhibit 1 at T[ 19)
115.

When the DOE published the results of the G&T Test, having 11% of students

attaining a 99 percentile ranking, it was an indication that the Respondents methodology for calculation the percentile ranks were flawed. For example, for students seeking Kindergarten admission, approximately 11% of studentswere placed in the 99 percentile rank; approximately20% were placed in the 97 percentile rank; and 39% were placed in the 90 percentile rank. See Exhibit 1 at T[ 21 for the DOES published chart of the results. 1 16. Dr. Kuptsov reviewed the methodology by which the Respondents generated the

percentile rank of students that took the G&T Test and have found several flaws that render the resulting percentile rankings inaccurate and not suitable for: (1) determining the eligibility of students for G&T Programs; and (2) determining the rank by which qualifjring students should be placed into the G&T Programs. Any one error alone would render the Respondents reliance on the percentile rank erroneous. (Exhibit 1 at T[ 22) 117. The first error rendering the percentile ranking inaccurate is that the DOE has

misrepresented the meaning of percentile ranking as no national sample exists.


118.

Because the percentile rank of a student taking a test is a measure of that students

performance compared to all other students taking the same test, the only legitimate way of calculating a true percentile rank for a student is to compare that students score (e.g., normalized score) to the scores of all other students taking the same test. (Exhibit 1 at T[ 23) 1 19. For example, if 100 students took test A and one student received a score that

was greater than 90 other students, then that student would have a 90 percentile rank. If that same student instead received a test score that was greater than 97 other students, then that child would 27

Supreme Court Records OnLine Library - page 31 of 59

have a 97 percentile rank. A valid percentile rank does not indicate what score the student received, it only provides information about what percentage of students scored at the same or lower test score. (Exhibit 1 at 7 24) 120. But to obtain a meaningful and valid percentile rank, the students must take the

same test. So for example, if 50 students take test A and 50 other students take test B then the students taking test B cannot be used in the pool of students used to calculate the percentile rank of a student taking test A. It would clearly be wrong to calculate the percentile rank of a student that took test A based upon the scores of the combined students who took either test A or B and so a student who received a score higher than 90 students out of the 100 combined students should not be given a 90 percentile rank. The reason is that the same student may in fact have scored higher than all 50 of the students taking the different test B , but only scored higher than

40 students who took the same test A, so that student in fact received a 80 percentile rank, not a
90 percentile rank. (Exhibit 1 at 7 25)
121. The percentile rank can further be rendered invalid by failing to account for other

test differences (other than just the questions). For example, if test A is administered to a group of 50 students which involves two parts, A1 and A2 (group l), a second group of 50 students is administered test B which involves two parts, B1 and A2 (group 2), and a third group of 50 students is administered test C which involves two parts, C 1 and A 1 (group 3), even though there is some overlap of some parts of the tests, it is mathematically inappropriate and statistically wrong to combine all 150 students of groups 1,2 and 3 to calculate the percentile rank of a student who took test A. That is because the overall test is different, and the variations in the overall test will almost certainly introduce significant variations on how well the student will perform on each of the test modules (Al, A2, B1 and Cl). However, as Dr. Kuptsov discusses below, this is the method 28

Supreme Court Records OnLine Library - page 32 of 59

which the DOE has chosen to employ in using percentile rank for the 2013 G&T test. (Exhibit 1 at f 26) 122. Similarly, a student taking test A will be fresh when he takes part A1 of the test,

but a student taking test C will be tired and/or frustrated by the time the student answers the questions to part A1 of that test because the student has already had to answer the questions of the C1 portion of test C. (Exhibit 1 at f 27) 123. To avoid these factors that inherently introduce significant errors into the

percentile rankings, the only feasible and accurate way to calculate percentile ranking is by comparing student scores who have taken the exact same test. In the example above, to calculate an accurate percentile rank, a student taking test A would have his scores ranked in comparison only to other students that took test A, and you would refrain from using any test scores of students taking tests B and C. (Exhibit 1 at f 28) 124. Dr. Kuptsov reviewed the G&T Handbook, attached hereto as to the Affirmation

of Stewart Lee Karlin as Exhibit 1-B states that the DOE calculated the percentile rank based on a students relative standing in comparison to other students of the same age nationwide. The DOE determined three different percentile scores for each candidate (please see NYC Gifted and Talented Score Report FAQ for Parents, dated April 26,20 13, attached hereto as Exhibit 1-C), all of them are reported on the G&T test score reports, received by the students and their parents: 1) Domain Percentile Rank for verbal part of the test (OLSAT8) calculated based on norms published in 2003,

2) Domain Percentile Rank for nonverbal part of the test (NNAT2) calculated based on norms
published in 201 1, 3) Overall Percentile Rank calculated based on mathematical formulas and studentsperformance in OLSAT8 and NNAT2. While Domain Percentile Ranks are reported just for informational purposes, the Overall Percentile Rank is the only number that determines whether
29

Supreme Court Records OnLine Library - page 33 of 59

a student is eligible to apply for district and citywide G&T programs, and just as important, after the siblings with priority are placed, is the mechanism by which the Respondents will rank the remaining students for placement into the G&T program. (Exhibit 1 at 7 29)
125.

From the G&T Handbook, it is clear that the Respondents stated that they

calculated the percentile rank of the students by comparing the students score to other students of the same age nationwide. (Emphasis added). However, the Overall Percentile Rank does not represent students relative standing to other students nationwide. The Respondents have confirmed that there is no nationwide test, or any other test administered outside of the New York City G&T Test that is the same. This is misleading, and falsely implies that the same G&T test is administered in other parts of the country. (Exhibit 1 at 7 30) 126. For the reasons described above, the Respondents statement that their calculated

percentile rank is based on a students relative standing nationwide to be misleading and have no legitimacy based on sound mathematical and statistical analysis. As discussed above, one cannot generate a valid percentile rank of a student using test scores of students who took different tests without introducing significant errors. Accordingly, the only valid way to generate accurate percentile ranks for the students who took the G&T Test would be to use the test scores of only the students who took the same test, i.e., it would include only scores of students who took the NYC

G&T Test. (Exhibit 1 at 7 3 1)


127. The second error rendering the percentile ranking inaccurate is that the NYC

sample is fundamentally different fi-om national samples. 128. Faced with the fact that there is no nationwide sample for calculating the Overall Percentile Rank (percentile ranking), the DOE attempted to use some mathematical approximations in order to combine two independent results for NNAT-2 and OLSAT-8, based
30

Supreme Court Records OnLine Library - page 34 of 59

respectively on national data generated in two different years (from 2003 and 201 l), into a single number. (Exhibit 1 at 7 32) 129. The Respondents mixed numbers specific to national samples and numbers that

are specific to the G&T Test, Le., to the sample of NYC children. Such an approximation would be legitimate only if G&T Test scores in NYC closely resemble the national sample in terms of the shape of bell curve or, in mathematical terms, only if G&T Test and national scores are drawn fiom the same probability distribution. (Exhibit 1 at 7 34) 130. However, the published test results show a large disparity between the G&T Test

group and the national test group, thus rendering Respondents mathematical assumption invalid. (Exhibit 1 at fi 32-35). 131. The probability of a NYC student scoring the maximum score in both the NNAT-

2 and OLSAT-8 modules of the G&T Test. Applying simple probability formulas for the normal distribution, one can observe that on average, 1 child out of approximately 10,000 would be expected to score the maximum score in both NNAT-2 and OLSAT-8 modules. In New York City approximately 13,000students seeking kindergarten admission into the G&T Program took the G&T Test. Therefore, you expect on average between 1 and 2 children to score the absolute maximum (i.e., 13,000 + 10,000 = 1.3 students). Yet, both EW and SK, Petitioners in this action, received the maximum score in both the NNAT-2 and OLSAT-8 modules. Additionally, Dr. Kuptsov is aware of at least five children who have scored the maximum score. Given that this is know fiom just a small sample of test-takers, in reality the number of kindergarten children who scored the absolute maximum is very likely much higher. (Exhibit 1 at 7 37). 132. As already mentioned above, this calculation mixes up the data from different

samples both in terms of the students tested, years that those tests were administered and uses the 31

Supreme Court Records OnLine Library - page 35 of 59

covariance from the NYC score data, which is not a legitimate thing to do in view of the fact that NYC data is fundamentally different from the national data. (Exhibit 1 at 7 40) 133. Therefore, the statistical evidence supports his opinion that the Respondents approach has very likely lead to substantial distortions of the calculated percentile rankings and so the percentile rankings being relied on by Respondents to place children into the G&T Programs is likely wrong. (For a further mathematical analysis see Error 2 in Exhibit 1) 134. The third error rendering the percentile ranking inaccurate is the Respondents

have created a theoreticalnational sample with Gaussian copula methodology which is known for its limitations. 135. In the absence of a true nationwide sample, the Respondents create a

hypothetical national distribution by using the methodology described in Exhibit 1-B. In essence, they attempt to shoehorn various other tests administered outside of the G&T Test together, to artificially create a national test to resemble the G&T Test. The DOE collects the test scores of students nationwide that took either the NNAT-2 module or the OLSAT-8 module and, irrespective of what other modules were co-administered with either module, or in what order modules were administered in those tests, compare the scores of those students who took the G&T Test. (Exhibit 1 at 7 42) 136. The Respondents choose the simplest methodology, equivalent to the well-

known methodology known as the Gaussian Copula methodology. (Exhibit 1 at 7 43) 137. However, the Gaussian Copula methodology is well-known for being

particularly erroneous in estimating very high and very low percentile ranks, i.e., those appearing at the ends of the bell curve. These are also commonly referred to as tail events, where the precision is indispensable. So, when the Respondents methodology is used to, in essence, create

32

Supreme Court Records OnLine Library - page 36 of 59

a fake bell curve to represent a hypothetical national student population by which to compare the G&T Test scores, the inaccuracies in the results that are obtained will likely have very large

errors towards the ends of the bell curve. (Exhibit 1 at 7 44) 138. Because the admission for the G&T Programs is indeed dependent on the student

attaining a very high percentile rank, Le., 99 percentiles for Citywide G&T Programs (97 percentile for siblings), and because these 99 percentile rankings are at the very end of the calculated virtual bell curve, those rankings are expected to be associated with large errors. (Exhibit 1 at

7 45)

139. Experts who studied the Gaussian Copula model concur that it is poorly suited for
representation of tail events. Some experts who have published on this subject have explained this limitation of the Gaussian Copula methodology in the following way: The Gaussian Copula model lacks tail dependence, which means that tail events in the Gaussian copula are close to independent. Translating this in terms of the G&T Test, tail dependence is the likelihood of observing very high

or very low scores for both NNAT-2 and OLSAT-8 modules for a given student. (Exhibit 1 at f
46) 140.

It is very probable that a child who scored a high result on one of the modules,

say NNAT-2, would also perform well on the other module. The exact amount of dependence, often called correlation, between NNAT-2 and OLSAT-8 for the whole NYC G&T sample is not available to me, but in examining the test scores of the four Petitioners, the dependence is very high. For example, all the families I know where their child scored very high in the NNAT-2 module of the G&T Test also scored very high on the OLSAT-8 module of that test. Therefore, the tail independence for NNAT- 2 and OLSAT-8 scores which is implied by the usage of the Respondents methodology is likely to be wrong and lead to substantial errors in determining the hypothetical
9gthpercentile. (Exhibit 1 at 7 47)

33

Supreme Court Records OnLine Library - page 37 of 59

141.

To give a definitive answer of how much the use of the Gaussian Copula model

by the Respondents has distorted the percentile rankings cannot be determined by publically available information. However, the Gaussian Copula methodology has been shown to be a bad model to determine tail events such as 99 percentile rankings and so does likely result in a large error which significantly impacts the percentile rankings used by the Respondents to place children into the G&T program. (Exhibit 1 at f 48) 142. Therefore, the G&T Test results are indicative of and support Dr. Kuptsovs

opinions that the DOEs calculation of the percentile rankings were misrepresented to the public and contain significant errors, particularly with respect to the higher percentile rankings that will be used by the DOE to place children into the G&T Program. The DOEs methodology creates an arbitrary and artificial bell curve by which the G&T Test percentile rankings were generated, which has had the erroneous effect of considerably raising the percentile ranks of many students well above their true percentile rankings. For example, the G&T Test results indicate that: a. b. 143.
39% of students seeking Kindergarten G&T Programs attained the DOEs 90 percentile threshold for applying to a District G&T Program, and

10% attained the DOEs highest percentile ranking of 99 percentile. (Exhibit 1 at 7 49)
Because the Respondents methodology has wrongly inflated the percentile

ranks of many students who took the G&T Test, it would have also incorrectly elevated many sibling to the 97 percentile rank or 90 percentile rank, guaranteeing them placement into the Citywide and District G&T Programs respectively, unfairly precluding other truly eligible children fiom attending Citywide G&T Schools and District G&T classes. (Exhibit 1 at f 50)
144.

For the reasons set forth above, Dr. Kuptsov concludes that an accurate percentile

ranking cannot be calculated using the methodology used by the Respondents, and that the percentile 34

Supreme Court Records OnLine Library - page 38 of 59

ranking that the Respondents intend to use to place students into the G&T Programs is fundamentallyflawed. Moreover, the composite scores already generated by Respondents provides an accurate and fair basis on which to place those students into the G&T Program. 145. Thus the DOE has already compiled the data necessary to accurately select

students for placement in the G&T program and to do so would be a rather simple task. (Exhibit 1 at 7 51)
Analvsis of the DOE G&T Test bv Dr. Borrdan Grechuk

146.

Dr. Grechuk a Lecturer of Mathematics at the University of Leicester, with more

than ten years experience in mathematical research and having published many articles on topics of mathematical finance, risk measurement and optimization. (Exhibit 2 at 7 1; Exhibit 2-A) 147. Dr. Grechuks research and teaching experience has given him a theoretical and

empirical grounding for examining the methodology used by Respondents to evaluate and score the test results of the Gifted and Talented admission test for the 2013-2014 application period. 148. Upon review of the analysis and conclusions as set forth in the Affidavit of Dr. Alexey Kuptsov (Exhibit I), Dr. Grechuk concluded that Dr. Kuptsovs analysis and conclusions are valid and consistent with his own conclusions. Based on this Dr. Grechuk would recommend not to use percentile ranking and use composite scores instead for 2013 Gifted & Talented placements, as percentile ranking results could be substantially distorted because of its calculation methodology. (Exhibit 2 at fi 3).
149.

Additionally in Dr. Grechuks analysis, he determined that the methodology

makes a key assumption that the scores are approximately normally distributed. But from his experience, the distribution of students exam results is usually far from the normal, so this assumption looks unsubstantiated. In fact, taking a look at the results of 2013 Gifted and Talented 35

Supreme Court Records OnLine Library - page 39 of 59

Test (See Exhibit 1-A) one can see that the assumption could be easily rejected for the actual New York City test results. In particular, Dr. Grechuk saw, that 1,480 out of 13,629 children (which is more than 10%) tested at grade K have gotteb percentile rank of 99, which is maximally possible. With the normal distribution assumption made by the DOE, only about 1%, or 136 children would have percentile rank of 99. This difference is much more than could reasonably appear due to any statistical error.
150.

Dr. Grechuk in his analysis could not glean why the DOE chose not calculate

percentilerank directly from the test results of NYC children: if for example a childs composite score is 115.6, and 80% of other participants have lower score (thus 20% - higher score), then the rank is 80%. This would give the exact correct answer and does not require any assumption about the underlying distribution. The percentile ranking currently used by the DOE is fundamentally flawed because the distribution of scores is far from normal. 151. Another problem in the DOE methodology as presented by Dr. Grechuk is that

they round the percentile rating to the integer value. As a result, all 1,480 children mentioned above have an equal percentile rank 99, resulting in equal chance to be selected for the gifted programs, despite some of students having significantly lower composite scores. In Dr. Grechuks opinion, to a reasonable degree of certainty the system currently used by the DOE is fundamentally flawed. 152. The patently obvious method would be to use the composite scores for NYC

students only, because the sample base is more than large enough, and would result in a precise and accurate method of selecting children for placement in the DOE Gifted and Talented Program. Given the testing data already compiled by the DOE, this method would of calculation would be a rather simple task.
Analysis of the DOE G&T Test by Dr. Jonathan Goodman

36

Supreme Court Records OnLine Library - page 40 of 59

153. Dr. Jonathan Goodman, PH.D, is currently a Professor of Mathematics at the New York University (NYU), Courant Institute, in New York, New York. Dr. Goodman has extensive experience in mathematics through his education, teaching, consulting and applied mathematics research and has published articles relating to computational methods in statistics. (Exhibit 3-A, Exhibit 3 at yq 1-5).
154.

Dr. Goodman has examined the method used by the DOE to categorize the test

results of the students taking the 2013 Gifted and Talented test. He was provided with an email written by an expert with the New York City Department of Education (DOE) describing the methods the DOE uses to categorize test results, and the model used to justify the categorization methods. This email states, among other things, that the DOE uses a Gaussian (also called normal) model of the distribution of test scores. (Exhibit 3 at 7 8). 155. Dr. Goodman believes that this Gaussian model significantly underestimates the

number of students who will score significantly above and below average and thus is fundamentally flawed. Additionally he believse that it is possible to construct a more precise and accurate model of the distribution of test scores, and that such a more accurate model will predict far more students with test scores the DOE method characterizes as exceptional. (Exhibit 3 at 7 8) 156. The DOE has chosen to assign percentile ranks to test scores not relative to other children who took the test, but with respect to national norms. In Dr. Goodmans opinion this method of assigning percentile ranks is fundamentally flawed as the sample size of NYC students

who took the exam is more than large enough to create an accurate percentile ranking and would be
more precise and accurate given that it compares students who took the modules under the same conditions. For the DOE to use the method that they chose it is necessary to use a model of test score distribution. The DOE uses a Gaussian model for this purpose. (Exhibit 3 at 110)
37

Supreme Court Records OnLine Library - page 41 of 59

157. The assignment of percentile ranks to aggregate test scores on the basis of the Gaussian model is flawed. The Gaussian model has the well-known flaw that it often

underestimates the number of data with large Z-scores. Statisticians say that a distribution has fat tails if the number of data with high Z-scores is much larger than the Gaussian model predicts. Many common real-world distributionshave fat tails. Examples include daily stock market returns (there have been several 3 sigma events this year), insurance claim payouts, and test scores. The DOE uses its statistical analysis to choose the brightest children for Gifted and Talented programs. This implies that the DOE is primarily interested in children whose aggregate test scores have high Z-scores. This, in turn, implies that the DOE should pay particular attention to the fat tail difficulty. (Exhibit 3 at f 12) 158. Thus, in terms of the Gifted and Talented test, the DOE is seeking to identifjr

students for acceptance into the gifted and talented program, which are those students with the highest test scores and therefore those that appear at the tail end of the statistical curve applied by the DOE. The DOE is using a statistical model that, as shown above, is not accurate for the distinguishing between those events that occur at the tail ends. Given the limited number of seats in the Gifted and Talented Program, the DOE is seeking to distinguish students for selection that appear at the tail end of the curve, i.e. those with percentile ranks of 99% or 98%, yet are using a statistical model that is proven to be flawed for that purpose. The precision necessary to distinguish between the highest scores is not provided for by the DOE statistical model. (Exhibit 3 at 7 14) 159.

In Dr. Goodmans opinion, to a reasonable degree of certainty the system currently

used by the DOE is fundamentally flawed. The patently obvious method would be to use the composite scores for NYC students only, because the sample base is more than large enough, and would result in a precise and accurate method of selecting children for placement in the DOE Gifted

38

Supreme Court Records OnLine Library - page 42 of 59

and Talented Program. Given the testing data already compiled by the DOE, this method would of calculation would be a rather simple task. (Exhibit 3 at 7 15)

Analvsis of the DOE G&T Test bv Dr. Tom Alberts

160.

Dr. Alberts is a Postdoctoral Fellow in the Department of Mathematics at the

California Institute of Technology (Caltech). Dr. Alberts main focus of research is in probability theory and he has published articles relating to probabilitytheory which involves statistical analysis. (Exhibit 8 at 77 1-5; Exhibit 8-a)
161.

Dr. Alberts reviewed the affidavit of Alexey Kuptsov (Exhibit 1) in this matter and

i t h the concluded that his analysis is valid and that the conclusions he has come to are consistent w
same conclusions that Dr. Alberts reached. 162. In particular, Dr. Alberts re-emphasized the following fundamental errors in Defendants methodology: a. The use of an unrelated national sample of test scores in determining an Overall Percentile Rank is not sound methodology. The test scores in this national sample are not coming fi-om students taking the same test, which clearly indicates that it is a comparison of apples to oranges, and it is not even clear that a group of similaraged students from outside of NYC would have the same statistical properties as the students taking the G&T test from within NYC. In fact, the exact opposite seems to be true, as supported by the fact that the average NYC NNAT-2 score is more than one standard deviation larger than the average national NNAT-2 score. The use of Gaussian copula to model correlation between NNAT-2 scores and OLSAT-8 scores is likely to introduce fundamental errors. In gifted students one expects that there is an extremely high correlation between these two tests (i.e. a student who does extremely well on one is likely to do extremely well on the other), but Gaussian copula are notoriously bad for computing statistics for rare events (i.e. a student being gifted) in highly correlated situations. The simple fact that 39% of students taking the G&T test fell into the 90thpercentile
39

b.

c.

Supreme Court Records OnLine Library - page 43 of 59

(when the expected percentage is only 10%) and 10% of students fell into the 99h percentile (when the expected percentage is only 1%) is incontrovertible evidence that the methodology behind the percentile rankings is not accurate. Dr. Alberts suspects the combination of the national sample data with the NYC G&T data is the major source of error, but it is impossible to know without seeing the raw data. Regardless, the methodology is clearly overestimating the number of students that fall into any given percentile, and yet the current methodology treats all students that fall within a given percentile as equally gifted. This is entirely erroneous. (Exhibit 8 at 7 6) 163. Based on Dr. Alberts analysis he recommends not to use percentile ranking and use composite scores instead for 20 13 Gifted & Talented placements, as percentile ranking results could be substantially distorted because of its calculation methodology. (Exhibit 8 at 7 7)
Analvsis of the DOE G&T Test bv Dr. Vladimir Finkelstein

164.

Dr. Vladimir Finkelstein is an adjunct professor of mathematics at the New York

University (WYU), Courant Institute, in New York, New York. He has extensive experience in applied mathematics and statistics through my education, teaching, consulting and research, with a particular interest in mathematical modeling in finance. Dr. Finkelstein has published numerous articles relating to mathematical modeling, financial engineering and physics, and also is a consultant focusing on quantitative methods in finance. (Exhibit 9 at 77 1-6) 165. Dr. Finkelstein reviewed the analysis and conclusions presented in the affidavit of

Alexey Kuptsov (Exhibit 1) in this matter and he has concluded that Dr. Kuptsovs analysis and conclusions are valid and consisted with the conclusions Dr. Finkelstein reached. (Exhibit at 7 7) In particular, Dr. Alberts re-emphasized the following fundamental errors in Defendants methodology: a. The DOEs calculation methodology of percentile ranking mixes up the data from different national samples both in terms of the kids tested and years those tests were administered and the calculation of the standard deviation by the DOE results in iin almost doubling the number of children eligible for the lottery. The DOEs methodology uses the score averages corresponding to the national data, 40

b.

Supreme Court Records OnLine Library - page 44 of 59

C.

100, while all the kids taking these tests in NYC are probably in top 20% of the students. 5% of the kids tested in NYC should have been in the top 1% nationwide. In reality, as I understand, 11% ended up in the top 1% nationwide which indicates that the distribution is not even normal but has a bit of a fat tail. (Exhibit 9 at 7 7)

166.

Based on the above arguments Dr. Finkelstein would recommend not to use

percentile ranking and use composite scores instead for 2013 Gifted & Talented placements, as percentile ranking results could be substantially distorted because of its calculation methodology. (Exhibit 9 at 7 8)
Analvsis of the DOE G&T Test by Dr. Valdislav Kargin

167.

Dr. Vladislav Kargin is a former professor at the Stanford Department of

Mathematics currently researching in the field of probability theory, with particular interest in Random Matrices, Quantum Walks, and Markov Chains. In the past, he has researched in Math Finance, Game Theory, and Statistics. Dr. Kargin has had more than 25 of his papers published in peer-reviewed journals including top-field journals in probability and statistics. (Exhibit 10 at 77 1-6). 168. Dr. Kargin evaluated the procedure used to calculate the percentile ranking score

used by the New York City Department of Education as an eligibility criterion for its Gifted and Talented program. In particular he examined whether the score the way it is computed by the DOE represented a comparison with children tested nationwide. 169. Dr. Kargin opines that the percentile ranking score as computed by the DOE does not

represent a comparison with children tested nationwide and is meaningless. The percentile ranking score refers to the relative standing of a tested individual with respect to a comparison group. For example, if a child has the percentile score of 97, it means that 97 out of 100 children from a 41

Supreme Court Records OnLine Library - page 45 of 59

comparable group performed or would perform worse on the test. (Exhibit 10 at 7 8). 170. Dr. Kargin has noted that the percentile score used by DOE does not show the

ranking of a tested child with respect to the students tested by DOE. For example, suppose that a child has received a composite score which is worse than the score of 10% of the other students tested. The percentile ranking with respect to the group of all tested students should be less than

90%. However, a quick look at the data shows that the New York City percentile score of this
student would be 99. (Exhibit 10 at 7 9) 171. The reason for this puzzling feature is that the DOE intends to compare the performance of a given child with the performance of all children of the same age in the USA. (Exhibit 10 at 7 10) 172. As a result of this intention, the procedure used by the DOE for the calculation of the

percentile score relies heavily on the statistical theory. Indeed, it is not possible to test all children in the USA, and it appears that there is no representative nationwide sample of children which would took both tests simultaneously. Hence, an application of a statistical model is unavoidable. Unfortunately, the statistical model used by DOE relies on two critical assumptions which are invalid in the given situation. (Exhibit 10 at 11) 173. The first assumption is that the results of the test are normal and the second assumption is that the parameters of the model can be estimated from the New York City sample. (Exhibit 10 at 7 12) 174. Dr. Kargin briefly explained how the percentile score is computed by the DOE. First,

the number of questions answered is converted to normalized scores on both tests and the results are combined. Then, the nationwide average is subtracted and the result is divided by the nationwide standard deviation. The result is called the z-score. (The national-wide average score and standard

42

Supreme Court Records OnLine Library - page 46 of 59

deviation are combined fiom the corresponding quantities for the two component tests, OLSAT-8 and NNAT-2. Dr. Kargin proceeded to address the problems with this calculation). (Exhibit 10 at

713)
175. At the next and crucial step in the calculation, the z-scores are converted to the percentile ranking by using the table of the normal bell-shaped distribution. The crucial statistical assumptionthat underlies this procedure is that the distribution of the scores follows the normal bellshaped distribution. (Exhibit 10 at 7 14)
176.

The main idea of this procedure is that it is not necessary to know the score of every

child in the US population in order to calculate his or her percentile rank. If the distribution of scores is normal, then it is enough to know the average score and the standard deviation of the scores which can be estimated using either a representative sample of population or other statistical methods. (Exhibit 10 at 15) 177. For example, if the normality assumption is valid and we know that the z-score of a tested child is greater than 2 (that is, the childs normalized score exceeds the average nationwide score by more than two standard deviations), then it can concluded that approximately 97.5% of kids nationwide would perform worse on the exam. (Exhibit 10 at 7 16) 178. Since various kinds of data do satisfy the normality assumption, the procedure

outlined above is in widespread use in statistical practice. However, the cases when this assumption is violated are not infrequent and it is a standard statistical practice to check this assumption. Without its validity, the results of the procedure are meaningless. (Exhibit 10 at 7 17) 179. Dr. Kargin concludes that in this particular case, the assumption about the normality

of test scores is highly doubtful and the percentile ranking scores computed by DOE mean nothing. (Exhibit 10 at 7 18) 43

Supreme Court Records OnLine Library - page 47 of 59

180. Dr. Kargin then presented research which shows that the NNAT test (making up one of the modules of the G&T test) and and a test with good norms are administered to a group of children, NNAT will appear to identify more than 3 times as many gifted children as the properly normed test. (Exhibit 10 at 7 21) 181. A new set of norms was issued for the second edition of NNAT (NNAT2) which was

intendedto correct the problems with this test. However,the subsequentindependent research shows that the problem with normality persists even in the second edition of the test. (Exhibit 10 at 7 22) 182. In Dr. Kargins opinion this finding is consistent with the results of the academic

research that show that the NNAT in both editions over-estimates the proportion of gifted kids in a given sample. (Exhibit 10 at 725) 183. Another important issue addressed by Dr. Kargin is that of combining the results of two scores, NNAT2, and OLSAT8. (Exhibit 10 at 7 26) 184. The error is that the procedure assumes that the covariance estimated from the

sample of the children tested in New York City is a good estimate for the covariance estimated properly, by using a representative sample of children nationwide. The sample of the children tested for the Gifted and Talented programs is far from being a representative sample for the nationwide population. It consists of New York City children who were taken to the examination sites because their parents believed that their children have special abilities. (Exhibit 10 at 7 29)
185.

This is called the self-selection bias in statistical science. Estimating parameters

using a sample that suffers from a self-selection bias can be grossly erroneous. (Exhibit 10 at 7 30) There is no reason to believe that the estimate of the nationwide covariance based on the NYC sample is any better than the estimate of the nationwide average score. (Exhibit 10 at 7 32) In particular, using the lower correlation coeEcient means an additional over-estimation of the 44

Supreme Court Records OnLine Library - page 48 of 59

proportion of gifted children in the New York City sample relative to the nation-wide population. (Exhibit 10 at 7 33) 186. Based on his analysis Dr. Kargin believes that the percentile score calculated by

DOE cannot be characterized as the accurate estimate of the relative position of a tested child in the
nationwide population. Dr. Kargin would recommend not to use percentile ranking for 201 3 Gifted
& Talented placements, as percentile ranking results could be substantially distorted because of its

calculation methodology. (Exhibit 10 at 7 34)


Analvsis of the DOE G&T Test bv Dr. Alexev Onatskiy

187. Dr. Alexey Onatskiy is currently a Reader of Economics (a UK equivalent of the


associate professor) at the University of Cambridge, UK. with PhD degree in Economics from Harvard University. Dr. Onatskiys specialization in economics is Econometrics, which is statistics applied to economic data. Dr. Onatskiy has about twenty publications in refereed journals, including statistical ones. Dr. Onatskiy has also been a referee for numerous economic and statistical journals. (Exhibit 11 at 77 1-2) 188. Upon Dr. Onatskiys review of the analysis and conclusions as set forth in the

Affidavit of Dr. Alexey Kuptsov (Exhibit l), Dr. Onatskiy concluded that it is indeed highly unlikely that the nationwide joint distribution of the NNAT-2 and OLSAT-8 scores and the joint distribution of the NNAT-2 and OLSAT-8 scores conditional on tested children living in NYC are the same (see Error 2 of the Affidavit of Dr. Kuptsov attached to the Affirmation of Stewart Lee Karlin as Exhibit 1I). This fact indeed potentially invalidates the percentile calculation done by the Respondents. (Exhibit 12 at 7 3) 189. Furthermore, Dr. Onatskiy agreed with Error 3 of the Affidavit of Dr. Alexey

Kuptsov (Exhibit 1) in that using Gaussian copula for an analysis of the dependence at the
45

Supreme Court Records OnLine Library - page 49 of 59

extreme regions of a bivariate distribution, such as the joint distribution of the NNAT-2 and OLSAT-8 scores, may, in general, be misleading. (Exhibit 12 at f 4) 190. Dr. Onatskiy concluded that the methodology used for the current Gifted &

Talented placement is flawed and the percentile rankings made by Defendants may not have the intended interpretation. (Exhibit 12 at 7 5)
ResDondents have acted arbitrarily and capriciously and in bad faith

191.

It is well settled that the standard of review of an agency determination under

Article 78 is whether a decision was arbitrary and capricious. In reviewing an administrative agency determination, we must ascertain whether there is a rational basis for the action in question or whether it is arbitrary and capricious. Gilman v. N.Y. State Div. of Hous. & Cmty. Renewal, 99 N.Y.2d 144, 149 (2002). 192.
An action is arbitrary and capricious when it is taken without sound basis in

reason or regard to the facts. Pell v. Board of Educ., 34 N.Y.2d 222,23 1 (1974). 193. When reviewing administrative decisions in an Article 78 proceeding, courts must

set aside an agency action if there is no rational basis for the exercise of discretion or the action complained of is arbitraryand capricious. Pell v. Board of Educ., 34 N.Y.2d 222,231 (1974). 194. An agency action lis arbitrary and capricious when it is taken without sound basis in reason or regard to the facts. Peckham v. Calogero, 12 N.Y.3d 424,431 (2009). 195. Explanations or rationales supplied after the fact, in the context of litigation, are

insufficient to justify agency actions retroactively. See Scherbyn v. Wayne-Finger Lakes Bd. of

COOD. Educ. Servs., 77 N.Y.2d 753,759 (1991); see also Patrick J. Borchers & David L.
Markell, New York State Administrative Procedure & Practice 8.6 (1998)(noting that an agency is not free to invent post hoc rationalizations for its decisions).That is because

46

Supreme Court Records OnLine Library - page 50 of 59

"judicial review of an agency decision is limited to the reasons given by the agency in its decision. An agency cannot use its answer in a CPLR Article 78 proceeding as a substitute for providing a rational reason in its determination." Central N.Y. Coach Lines, Inc. v. Larocca, 120 A.D.2d 149, 152 (3d Dep't 1986). 196. The Court of Appeals has upheld the principle that a mathematical formula of grading must be rational and logically related to the ends for which the exam was administered. Bruno v. Lebow, 95 A.D.2d 731,464 N.Y.S.2d 141, afd, 60 N.Y.2d 826 (1983). 197. In cases where the existence of some empirical support is of particular necessity in the recalibration context because the regulation seeks to implement a mathematical formula to measure -Because the regulation seeks to measure a statistical correlation, the method must be tested to insure that the calculation bears a reasonable relation to reality. In the Matter of the Application of New York Association of Homes and Services for the 1996 WL 33476494 (Court of Appeals of New York) 198. [Alny dispute concerning the proper placement of a child in a particular education

program can best be resolved by seeking review of such professional educationaljudgment through the administrative process provided by statute. See Education Law 0 3 10, subd 7; Hoffman v. Board of Educ. of City ofN.Y., 49 N.Y.2d 121,127,424 N.Y.S.2d 376,400 N.E.2d 317. 199. Respondent have violated the law by failing to provide proper notice prior to

changing the admissions process and thus there actions are arbitrary and capricious. 200.

As a result of the foregoing, the Respondents have acted arbitrarily and

capriciously and in bad faith.


20 I.

Petitioners have complied with all conditions precedent to commencing this suit 47

Supreme Court Records OnLine Library - page 51 of 59

e
and have exhausted their administrative remedies.
202.

This is the first application for the relief requested herein.

AS AND FOR A SECOND CAUSE OF ACTION-EOUAL PROTECTION


ResDondents have violated Petitioners right to euual Drotection
201.

Petitioners repeat and reallege each and every paragraph as if fully set forth at length herein.

202.

Respondents, state actors acting under color of state law, have denied Petitioners

equal protection of the law when they revised the process for admission to the Citywide G&T Program because by their actions students with siblings within the Citywide G&T Schools and who have lesser composite and lesser percentile scores will have preferential treatment and will take priority in admission to the Citywide G&T program over Petitioners who have both higher composite and percentile scores on the G&T Test. 203. Further by Respondents actions, students with siblings within the District G&T

Classes that have lesser composite and lesser percentile scores will have preferential treatment and will take priority in admission to the District G&T Program over Petitioners who have both higher composite and percentile scores on the G&T Test.
204.

Respondents, by their actions, have violated Petitioners right to equal protection

of the laws as provided by the Equal Protection Clause of the New York State Constitution Article I Section 11.

WHEREFORE, Petitioners respectfully request a judgment against Respondents as


follows:
205.

Equitable relief against the Respondents including but not limited to the

following:

48

Supreme Court Records OnLine Library - page 52 of 59

a. b.

A stay of the Gifted and Talented Program Selection Process;

A declaration that the notice provided by Respondents as to the change in the Gifted and Talented Program Selection Process was arbitrary and capricious and violated applicable laws;
A declaration that the methodology for selection to the Gifted and Talented Program in its current form is arbitrary and capricious and violates Petitioners right to equal protection;
A declaration that selection to the Gifted and Talented Program be completed pursuant to the procedure set forth in the G&T Handbook as originally published, prior to any amendment, with selection based on the students composite score and sibling priority only effecting selection of students in cases in which two students have the same composite score.

c.

d.

e.

Such additional relief as this Court may deem just and proper under the circumstances.

Dated: New York, New York May 15,2013

Attorney for Petitioner 9 Murray Street, Suite 4W New York, NY 10007 (2 12) 792-9670

49

Supreme Court Records OnLine Library - page 53 of 59

VERIFICATION

STATE OF NEW YORK ) COUNTY OF NEW YORK)


ss.:

I, the undersigned, an attorney admitted to practice in law in the Courts of New York State,
state that I am the attorney of record for the Petitioner in the within action; I have read the foregoing, Petition, and know the contents thereof; the same is true to my own knowledge except
as to those matters therein alleged to be on information and belief, as to those matters I believe to

be true. The reason this verification is made by your affirmant and not by the Petitioners herein is that Petitioners A.K. on behalf of his minor child S.K., S.R. on behalf of his minor child H.R., and L.W, on behalf of her minor E.W. reside in a county other than the county where your affirmant maintains his law office. Additionally Petitioner R.B, on behalf of his minor child L.B., is traveling outside of the state. The grounds of your affirmants belief as to all matter not stated upon my own knowledge are as follows: Conversations with the Petitioner and information contained in the file. Dated: New York, New York May 15,2013

STEWAR~LEE KARLJN, ESQ.

50

Supreme Court Records OnLine Library - page 54 of 59

ATTORNEY CERTIFICATION

Pursuant to 22 NYCRR, Section 130-1.1.A the undersigned, an attorney duly admitted to practice law in the Courts of the State of New York, respectfully affirms the truth of the following statement under the penalties of perjury pursuant to the C.P.L.R.: The undersigned attorney hereby certifies that to the best of his knowledge, information and belief, formed after an inquiry reasonable under the circumstances, the presentation of the papers (s) or the contents therein are not frivolous as defined in Subsection C of Section 130-1.1, Dated: New York, New York May 15,2013
/e

, . A
,/

/~

STEWART LEE KARLIN, ESQ.

51

Supreme Court Records OnLine Library - page 55 of 59

SUPREME COURT OF THE STATE OF NEW YORK COUNTY OF NEW YORK

In the Matter of the Application of, R.B. on behalf of his minor child L.B.; A.K. on behalf of his minor child S.K.; S.R. on behalf of his minor child H.R.; L.W. on behalf of her minor child E.W., individually and on behalf of all others similarly situated, Petitioners,
for Judgment pursuant to CPLR Art. 78 and common law claims, -against-

AFFIRMATION OF STEWART LEE KARLIN Index No.:

THE DEPARTMENT OF EDUCATION OF THE CITY OF NEW YORK fMa THE BOARD OF EDUCATION OF THE CITY SCHOOL DISTRICT OF THE CITY OF NEW YORK; DENNIS WALCOTT, as Chancellor of the Department of Education of the City of New York; GENTIAN FALSTROM, as Director of elementary enrollment of the Department of Education of the City of New York; ROBERT SANFT, as Director of the Office of Student Enrollment of the Department of Education of the City of New York;
Respondents.

STEWART LEE KARLIN, an attorney duly admitted to practice law in the State of New York affirms under the penalty of perjury the following:
1.

I am the attorney for the Petitioners in the above-captioned matter and as such am personally familiar with the circumstances set forth below.

2.

This Affirmation is submitted in support of the Petition in the above-captioned action This is the first application for the relief requested herein.

3.

Supreme Court Records OnLine Library - page 56 of 59

4.

Petitioners annex hereto in support of the Petition the following original affidavits and true copies of the following exhibits: Exhibit 1: Affidavit of Dr. Alexey Kuptsov

Exhibit 1-A: Curriculum Vitae of Dr. Alexey Kuptsov Exhibit 1-B: Relevant Portions of the G&T Handbook as pwlished in Fall 20 12 Exhibit 1-C: NYC Gifted and Talented Score Report FAQ for Parents, dated April
26,2013

Exhibit 1-D: Calculations of Dr. Alexey Kuptsov Exhibit 2: Affidavit of Dr. Bogdan Grechuk

Exhibit 2-A: Curriculum Vitae of Dr. Bogdan Grechuk Exhibit 3: Affidavit of Dr. Jonathan Goodman

Exhibit 3-A: Curriculum Vitae of Dr. Jonathan Goodman Exhibit 4: Affidavit of S.R.

Exhibit 4-A: Score Report of H.R. Exhibit 4-B: E-mail exchange between parents of aggrieved children, the DOE and City Officials Exhibit 4-C: E-mail exchange between S.R. and the DOE Exhibit 5: Affidavit of R.B.

Exhibit 5-A: Score Report of L.B. Exhibit 6 : Affidavit of A.K.

Exhibit 6-A: Score Report of S.K. Exhibit 7: Affidavit of L.W.

Exhibit 7-A: Score Report of E.W.

Supreme Court Records OnLine Library - page 57 of 59

titled, A Policy Shift Exhibit 7-B: New York Times Article dated December 19th for the Gifted is Abandoned Exhibit 7-C Exhibit 8: DOE Press Release regarding Pearson scoring errors

Affidavit of Dr. Tom Alberts

Exhibit 8-A: Curriculum Vitae of Dr. Tom Alberts Exhibit 9: Affidavit of Dr. Vladimir Finkelstein

Exhibit 9-A: Curriculum Vitae of Dr. Vladimir Finkelstein Exhibit 10: Affidavit of Dr. Vladislav Kargin

Exhibit 10-A: Curriculum Vitae of Dr. Vladislav Kargin Exhibit 11: Affidavit of Dr. Alexey Onatskiy

Exhibit 11-A: Curriculum Vitae of Dr. Alexey Onatskiy Exhibit 12: Exhibit 13:

DOE Letter to Families Dated December 19,2012


DOE Notices Regarding Chancellors Regulation A-101

Exhibit 13-A: DOE Notice of October 16,2012 Exhibit 13-B: DOE Revised Notice Dated November 5,2012 Exhibit 13-C: DOE Revised Notice Dated December 5,2012 Exhibit 14:
5.

Relevant Portions of the Amended G&T Handbook

Aflknant has spoken to Thomas Crane, Chief of the General Litigation Section of the New York City Law Department who is responsible for this litigation on behalf of Respondents on May 14, 2013 regarding this litigation and plans to seek a TRO. On May 15, 2013, I further advised him that I am serving a true copy of all the papers on him by hand on May

16,2013 at 9:OO am and that I would appearing at the Ex Parte Office, Room 315 of the
New York County Supreme Court at about 1O:OO am seeking a temporary restraining order.

Supreme Court Records OnLine Library - page 58 of 59

Dated: New York, New York May 16,2013

e
STEWART LEE KARLIN

Supreme Court Records OnLine Library - page 59 of 59

You might also like