Citation
An assessment procedure for detecting giftedness using available data

Material Information

Title:
An assessment procedure for detecting giftedness using available data
Creator:
Schnell, Randy, 1955-
Publication Date:
Language:
English
Physical Description:
vi, 104 leaves : ; 28 cm.

Subjects

Subjects / Keywords:
Academically gifted students ( jstor )
Cognitive psychology ( jstor )
Educational administration ( jstor )
Intelligence quotient ( jstor )
Mathematics ( jstor )
Psychometrics ( jstor )
Schools ( jstor )
Screening tests ( jstor )
Stanford Binet test ( jstor )
Wechsler scales ( jstor )
Gifted children -- Identification ( lcsh )
Intelligence tests ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1987.
Bibliography:
Includes bibliographical references (leaves 92-103).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Randy Schnell.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
18420363 ( OCLC )
ocm18420363
001043393 ( ALEPH )

Downloads

This item has the following downloads:


Full Text











AN ASSESSMENT PROCEDURE FOR DETECTING
GIFTEDNESS USING AVAILABLE DATA

















BY

RANDY SCHNELL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA


1987
















ACKNOWLEDGEMENTS


I would like to thank Larry Loesch, Ph.D., for his

guidance and assistance as chairperson of my doctoral com-

mittee. Dr. Loesch graciously assumed the role of chair-

person during a period when my study was in disarray and

not progressing satisfactorily. I also wish to thank

Robert Jester, Ph.D., whose consultation assisted in

development of the framework of this study, and Janet

Larsen, Ed.D., who has supported my doctoral work since

1982 when she agreed to be my advisor. I would also like

to acknowledge Linda Crocker, Ph.D., whose patience, dili-

gence, and expertise in measurement were vital to the

study and greatly appreciated.

A number of other people who contributed to this

study also deserve recognition. These include John

Hilderbrand, Ph.D., Hugh Morehouse, Robert Haines, Ed.D.,

Grace Hutchinson, Michael Selby, Lois Rudloff, and Denise

Landau, Ph.D.

I wish to express special thanks to my parents and

other family members for their support.
















TABLE OF CONTENTS


Page


ACKNOWLEDGEMENTS . . .

ABSTRACT . . .


CHAPTER


INTRODUCTION .


Problem . .
Giftedness . .
Purpose . .
Research Questions .. ..
Definition of Terms .
Theoretical Rationale .
Need for This Study .
Overview of the Remainder of


the


Study


II LITERATURE REVIEW . .


Support for the Problem .
Group IQ Tests .
Intelligence Quotient Short


Forms


Intelligence Quotient Screening
Tests . .
Achievement Tests . .
Summary . .
Instruments Used in Study .
Wechsler Intelligence Scale for
Children-Revised (WISC-R) .
Slosson Intelligence Test SIT) .
Comprehensive Test of Basic Skills
(CTBS)/Test of Cognitive Skills
(TCS) . .
Stanford-Binet Intelligence Scale
(S-B) . .


iii














III


METHODOLOGY ....

Overview . .
Population and Sample .
Assessment Procedures .
Research Procedures .
Data Analysis .
Methodological Limitations .


RESULTS . .


. 62


Phase I--Item Selection 62
Phi Coefficients . 62
Index of Discrimination .. 66
Cutoff Scores . 68
Phase II--Cross Validation .. 70
Application of Cutoff Scores 70
Kappa Comparisons .. .74
Summary of Results .. 78
Summary of Results . .. 78


DISCUSSION . 80

Research Questions . .. .80
Phase I . 81
Common Items . 81
IRT Parameters . .. 82
Phase II . .. .. 83
Conclusions, Implications, and
Limitations . 85
Sampling . ... 85
Generalizability . .. 87
Screening Accuracy .. 88
Item Validity .. 88
Cutoffs . 90
Summation . 91


REFERENCES .


BIOGRAPHICAL SKETCH . .


104















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy



AN ASSESSMENT PROCEDURE FOR DETECTING
GIFTEDNESS USING AVAILABLE DATA


By

Randy Schnell


December 1987


Chairman: Larry Loesch, Ph.D.
Major Department: Counselor Education


The purpose of this study was to examine a new psy-

chometric screening procedure designed to discriminate

intellectually "gifted" seventh graders from other high

achieving seventh graders. All 179 students in the re-

search sample had been assessed to determine eligibility

in the Hillsborough County Public Schools "gifted" pro-

gram. An accurate and efficient screening procedure was

necessary in order to delete from time-consuming, formal

assessment those students who were unlikely to meet intel-

lectual criteria for gifted program placement. The Slos-

son Intelligence Test (SIT), previously used to screen

students referred to the gifted classes, had proved inef-

ficient for this purpose. A plethora of research on











methods used to identify gifted students revealed that

most methods were less than accurate.

The Comprehensive Test of Basic Skills and Test of

Cognitive Skills were item analyzed on a sample of 179

seventh grade subjects using two correlational procedures.

Results from the item analyses yielded 46 items from the

phi analysis and 14 from the index of discrimination anal-

ysis that discriminated gifted from not gifted students,

as classified by the WISC-R or Stanford Binet. Cross val-

idation was conducted on a second sample of 61 students to

determine (a) the accuracy of the new screening procedure

(NP) compared to the SIT and (b) at what cutoff points

either the NP or SIT was more accurate. In cross valida-

tion, total scores of both item subsets were compared with

SIT. The items obtained from the index of discrimination

analyses were found to be the best predictors of "gifted"

intelligence for the sample. There was some ambiguity re-

garding an optimal cutoff score; however, a score .33

standard deviations below the mean score was found to gen-

erally yield most accurate predictions.
















CHAPTER I
INTRODUCTION



The tremendous increase since 1970 in school programs

for the intellectually gifted has led most states to es-

tablish guidelines or requirements for program eligibility

(Karnes & Brown, 1979; Kolloff & Feldhusen, 1984). One

common requirement is for individual intellectual evalua-

tion of gifted program candidates (Chambers, Barron, &

Sprecher, 1980; Sternberg, 1982; Vernon, Adamson, & Ver-

non, 1977). Yarborough and Johnson (1983) reported the

use of individual intelligence test minimum scores as an

eligibility requirement in at least 73% of the nation's

gifted student programs. Although educators may recommend

and actually believe in the importance of a broader view

of giftedness, this more narrow definition (i.e., empha-

sizing IQ) is frequently employed because of pragmatic

concerns (Jenkins-Freidman, 1982). The "intelligence quo-

tient" has therefore emerged as the primary criterion for

gifted program eligibility (Barklay, Phillips, & Jones,

1983; Birch, 1984; Guilford, 1975; Karnes, Edwards, &

McCallum, 1986).

The individual intelligence testing requirement,

coupled with federal regulations mandating services for












children with special educational needs, has placed a bur-

den on local school districts to identify gifted students

through accurate referral procedures and efficient use of

testing personnel (Karnes & Brown, 1979). As a result,

school systems are currently faced with demands for intel-

lectual assessment of children which are often greater

than can be met by the qualified examiners available (Cro-

foot & Bennett, 1980; Fell & Fell, 1982). There is a

demand for screening procedures designed to facilitate the

referral process by maximizing the use of individual

testing time while minimizing errors in identification

(Jenkins-Friedman, 1982; Kramer, Markley, Shanks, &

Ryabik, 1983; Rust & Lose, 1980; Stephens & Gibson, 1963).

The Wechsler Intelligence Scale for Children--Revised

(WISC-R) and Stanford-Binet Intelligence Scale (S-B) are

the most widely used individual tests of children's intel-

ligence (Bryan & Bryan, 1975; Salvia & Ysseldyke, 1978;

Wikoff, 1978). Screening tests which estimate or predict

intelligence scores on the WISC-R and Stanford-Binet have

been studied for use among gifted school populations since

the 1950s (Pegnato & Birch, 1959; Sheldon & Manolakes,

1954).











Problem


The research problem addressed concerned the accurate

identification of intellectually gifted students from a

screening procedure prior to administration of an individ-

ual intelligence test. The expense of administering indi-

vidual intelligence tests has necessitated the screening

of potentially gifted students in an effort to delete from

formal testing those who probably do not possess gifted

intelligence (Rust & Lose, 1980; Stenson, 1982). Inaccu-

rate screening has resulted in not-gifted students receiv-

ing the time-consuming tests and in some gifted students

being excluded from testing.

Screening procedures such as the Slosson Intelligence

Test (Dirkes, Wessels, Quaforth, & Quenon, 1980; Grossman

& Johnson, 1983; Karnes & Brown, 1979; Rust & Lose, 1980),

the Ammons Quick Test (DeFilippis & Fulmar, 1980; Hirsch &

Hirsch, 1980), short forms of the WISC-R (Bersoff, 1971;

Elman, Blixt, & Sawicki, 1981; Karnes & Brown, 1981; Kil-

lian & Hughes, 1978; Kramer et al., 1983), Guilford's

Structure of the Intellect (Pearce, 1983), the Peabody

Picture Vocabulary Test (Mize, Smith, & Callaway, 1979;

Wright, 1983), and group IQ tests (Blosser, 1963; Grossman

& Johnson, 1983; Pegnato & Brich, 1959; Sheldon & Mano-

lakes, 1954) have all been shown to be more or less inac-

curate and/or inefficient for screening high ability











students. Chambers (1960) and Schena (1963) reported

somewhat more encouraging results with academic skill

measures as predictors of gifted intelligence. A more ac-

curate and efficient means for screening gifted students

needs to be found.


Giftedness


The characteristics associated with gifted

intelligence are almost as numerous as the students

themselves (Tuttle & Becker, 1980). Qualitative trait

differences between gifted and not-gifted children are

indicated frequently in the literature (Barrington, 1979;

Dirkes, 1981; Gensley, 1975; Male & Perrone, 1979; Ricca,

1984; Ryan, 1982; Stornberg, 1982); however, essential to

a discussion of intellectual giftedness in children is

their classification according to an IQ test cutoff score.

Gifted classifications by IQ cutoff scores are employed in

an attempt to objectify classification criteria and school

placement decisions. Classification by IQ score may mis-

leadingly imply that qualitative differences between

gifted and not-gifted children are necessarily demarcated

by an artificial cutoff (Braden, 1985). In this study

cutoff scores are used to quantify gifted intelligence ac-

cording to educational criteria and not to define learning

style, motivation, or other personality traits associated

with "giftedness."












Purpose


The primary purpose of this study was to determine

whether a large, group-administered achievement/ability

test battery possesses items that, in combination, yield a

score that accurately predicts gifted classification as

measured by the WISC-R or Stanford-Binet. A secondary

purpose of this research was to determine if the new

screening procedure (NP) classifies gifted and not-gifted

seventh graders more accurately than the Slosson Intelli-

gence Test (SIT) and at what cutoffs these classifications

are most accurate.

The new screening procedure developed Lor this study

consisted of a subset of items selected from norm refer-

enced, group-administered tests of academic aptitude and

achievement. The aptitude test used in this study was the

Test of Cognitive Skills (TCS) and the achievement battery

was the Comprehensive Test of Basic Skills (CTBS). The

TCS and CTBS have been normal on the same sample and are

typically administered in concurrent testing sessions.



Research Questions


The following research questions were addressed:

1. Can an accurate predictor of gifted IQ classifi-

cation on the WISC-R/S-B be derived from an instrument











composed of items on the CTBS and TCS in a situation in

which giftedness is viewed as a dichotomous variable?

2. Is the NP more accurate than the SIT in classi-

fying gifted and not-gifted seventh graders?

3. At what cutoff points) is the NP more accurate

than the SIT most accurate in classifying gifted and not-

gifted seventh graders?



Definition of Terms


Comprehensive Test of Basic Skills (CTBS). The CTBS

is a "series of norm-referenced, objective-based tests for

kindergarten through twelfth grade. The series is de-

signed to measure achievement in the basic skills commonly

found in state and district curricula" (CTB/McGraw-Hill,

1984, p. 1). At the junior high school levels the content

areas are reading, spelling, language, mathematics, refer-

ence skills, science, and social studies.

Full Scale IQ (FS-IQ). Full Scale IQ is the derived

intelligence quotient on the Wechsler Intelligence Scale

for Children--Revised and on the Stanford-Binet Intelli-

gence Scale.

General intelligence. General intelligence is a set

of general cognitive operations measured as the overall

ability required for success on IQ tests. General











intelligence is comprised ol those traits commonly

measured by IQ test items.

Gifted intelligence. Gifted intelligence is measured

as an intelligence quotient that falls in the Very Supe-

rior category or two standard deviations above the test

mean. For this research, gifted IQ=130 -1 SEM (where

SEM=3 IQ points) on the WISC-R, and IQ=132 -1 SEM (where

SEM=5 IQ points) on the Stanford-Binet, or IQ=127. (This

definition, which includes a SEM, is based on guidelines

from Hillsborough County Florida School District).

High-achievers. Students who, because of superior

academic performance, have been referred for gifted pro-

gram testing, but who are assessed as not intellectually

gifted are referred to as high-achievers.

Intelligence. Intelligence is "the aggregate global

capacity of the individual to act purposefully, to think

rationally, and to deal effectively with his environment"

(Wechsler, 1958, p. 7). Intelligence is comprised of in-

tellective factors such as "abstract reasoning, verbal,

spatial, numerical, and other factors" (Wechsler, 1950, p.

78), and nonintellective factors consisting of "capacities

and bents dependent upon temperament and personality which

are factors of personality itself" (Wechsler, 1950,

p. 78).

Intelligence quotient. An intelligence quotient is a

derived total or Full Scale score on an intelligence test.












Performance IQ (P-IQ). Performance IQ is a subscale

IQ score on the Wechsler Intelligence Scale for Children--

Revised that represents the examinee's nonverbal reasoning

or perceptual organization (Kaufman, 1975).

Reliability. Reliability is the squared population

correlation between the individual's obtained score and

the individual's hypothetical true score. Reliability is

"the proportion of true-score variance in scores on a par-

ticu.lar teL st at e he time it was taken" (JJnson, 1980, p.

260).

Slosson Intelligence Test (SIT). The SIT is an indi-

vidually administered intelligence test which requires

little specialized training to administer, only about 20

minutes to administer and score, and yields IQ scores

"which are close approximations to the Stanford-Binet IQ

[scores]" (Slosson & Jensen, 1982, p. 1).

Stanford-Binet Intelligence Scale (S-B). The S-B is

one of the two individually administered intelligence

tests used in this study to measure gifted intelligence.

(See page 41 for a detailed description.)

Test of Cognitive Skills (TCS). The TCS is "an abil-

ity test designed to assess a student's academic aptitude

and thereby predict the student's level of success in

school. Emphasis in TCS is placed on" problem solv-

ing, discovering relationships, evaluating, and











remembering" (CTB/McGraw-Hill, 1984, p. 1). The TCS is

administered concurrently with the CTBS.

Validity. Validity refers to the appropriateness of

inferences from test scores or other forms of assessment.

Validity deals with how faithfully the scores represent a

domain of skill, knowledge, or of a trait being measured

(American Psychological Association, 1985).

Verbal IQ (V-IQ). Verbal IQ is a subscale IQ score

on the Wechsler Intelligence Scale for Children--Revised

that represents the examine's verbal reasoning or verbal

comprehension (Kaufman, 1975).

Wechsler Intelligence Scale for Children--Revised

(WISC-R). The WISC-R is the individually administered

IQ test predominantly used in this study to measure gifted

intelligence. (See page 35 for a detailed description.)



Theoretical Rationale


Salvia and Ysseldyke (1978) have pointed out that

there is a hypothetical domain of items that may be used

to assess intelligence and that items may be drawn from

various sources. For example, WISC-R Information subtest

items are drawn from a domain of achievement oriented

items that measure specific content of learning acquired,

in large part, through formal education. This overlapping

of achievement and aptitude test item content has been











demonstrated empirically by Anastasi (1976). Her examina-

tion of the content of several current instruments classi-

fied as achievement and intelligence tests revealed simi-

larity in their content. Supporting this finding she

contended that it has long been known that IQ tests corre-

late about as highly with achievement tests as different

IQ tests correlate with each other. Further, one of the

most frequently employed means of validating IQ tests is

to compare them with measures of achievement.

In another attempt to show some common elements of

achievement and intelligence, Gronlund (1976) compared

factors measured by both reading readiness tests and IQ

tests. These elements included

1. visual discrimination--identifying similarities

and differences in words or pictures;

2. auditory discrimination--identifying similar-

ities and differences in spoken words;

3. verbal comprehension--demonstrating an under-

standing of the meaning of words, sentences, and

directions; and

4. copying--demonstrating skills in reproducing

geometric forms.

In his analysis of the WISC-R Verbal Scale, Kaufman

(1979) identified item content that reflects properties of

achievement tests. For example, the Information subtest











of the WISC-R measures acquired knowledge and is influ-

enced by outside reading and school learning. The Sim-

ilarities subtest was also found to be subject to reading

and vocabulary knowledge. Likewise, the other Verbal

Scale subtests were found to have strong components of

acquired knowledge.

In one of a very few studies in which IQ test item

content was related to specific academic skills, Washing-

ton, Engelmann, and Bereiter (1969) conducted an item

analysis of the Stanford-Binet Intelligence Scale and at-

tempted to construct an academic curriculum from it.

After the curriculum was presented to students an achieve-

ment test was administered. Resultss showed that the

prelearned S-B items were positively correlated with'post-

test achievement items for particular learning tasks. In

a second phase of the study no pretest was administered.

However, subsequent to the curriculum presentation and

post-test administration the Stanford-Binet was given.

The achievement test results were found to accurately pre-

dict S-B scores in terms of items responses. Results sug-

gested content validity across the IQ and achievement

measures.

A number of other researchers have investigated the

relationship between intelligence and achievement measures

and found them to be positively correlated (Hale, 1978;

liartlago & Steele, 1977; Reschley & Rcschley, 1979;












Schwarting & Schwarting, 1977) or to possess significant

overlap in their factor loadings (Carroll, 1966; Dean,

1977; Grossman & Johnson, 1982; Horn, 1970; Stewart & Mor-

ris, 1977; Undheim, 1976; Vernon, 1961, 1969; Wikoff,

1978).

The TCS, in terms of its title and stated purpose,

encompasses a construct closely related to the WISC-R and

Stanford-Binet in that all three were designed and have

been found to predict academic attainment (Hurrocks, 1964;

CTB/McGraw-Hill, 1984; Terman & Merrill, 1973; Wechsler,

1974). As described in its 1983 Technical Report, the TCS

subtests (Sequences, Analogies, Memory, and Verbal Reason-

ing) load on factors consistent with those described in

WISC-R and Stanford-Binet studies (Kaufman, 1979).

The TCS, according to its constructors, measures "a

number of cognitive abilities included in various

theories, however, like the WISC-R and Stanford-Binet, em-

phasis is placed on the kinds of reasoning and retention

skills necessary for school success" (CTB/McGraw-Hill,

1984, p. 5). The CTBS similarly measures a variety of

academic skill areas shown to be positively correlated

with the TCS. The TCS and CTBS together appear to assess

a theoretical construct common to the WISC-R and Stanford-

Binet. Therefore, the items chosen from the CTBS and TCS











should conform to that construct, thus allowing discrim-

ination of gifted from not-gifted students as do the WISC-

R or Stanford-Binet.

Support for the common item content of the CTBS, TCS,

WISC-R, and Stanford-Binet will be provided here by

illustrations of actual items found on these tests. Items

are categorized arbitrarily according to subtest classifi-

cation and to face commonalities. A brief description and

reference location will be given for those non-verbal test

items that cannot be readily reproduced in this format.

Items indicated are those designed for average and above

average seventh graders. Test items are printed in bold-

face.

1. WISC-R, Vocabulary-instructions: What does
mean?
rivalry

CTBS, Vocabulary (instructions: Choose the word or
phrase that means the same, ., as the underlined
word.)
their opponent
A. foe
B. employee
C. architect
D. assistant

2. WISC-R, Arithmetic
Tony bought a second hand bicycle for $28.00.
He paid 2/3 of what the bicycle cost new. How
much did it cost new?

CTBS, Mathematics Concepts and Applications
Homer's recipe will make 48 sugar cookies. He
made 3/4 of this recipe for a party. How many
cookies were for each of the 18 people at the
party?











3. Stanford-Binet, Arithmetic Reasoning
If a man's salary is $20 a week and he spends
$14 a week, how long will it take him to save
$300?

CTBS, Mathematics Concepts and Applications
To pay for groceries, Scott, Marvin, and Carol
each gave the clerk $1.35. The clerk gave them
$.45 In change. How much did the groceries
cost?
P. $1.80
G. $3.60
H. $4.05
J. $4.50

4. Stanford-Binet, Vocabulary
What does "Brunette" mean?

CTBS, Vocabulary (instructions: Choose the word or
phrase that means the same as the underlined
word.)
successful merchant
P. parade
G. business
H. customer
J. shopkeeper

5. WISC-R Picture Arrangement (instructions: ". .. I
want you to arrange these pictures in the right order
to tell a story that makes sense.")

TCS, Sequence (instructions: ". choose the part
that would continue the pattern or sequence.") Vari-
ous visual stimuli are presented such as letters,
numbers or geometric shapes.

6. WISC-R, Similarities
In what way are a telephone and a radio alike?

TCS, Verbal Reasoning (paraphrased instructions: The
words in the top and bottom rows are related in the
same way. Find the word that completes the bottom
row of words.)
radio electricity music
paper newspaper
F. ink
G. story
H. reporter
I. typewriter











7. Stanford-Binet, Problems of Fact
"An Indian who had come to town for the first
time in his life saw a boy riding along the
street. As the boy rode by the Indian said,
"The white boy is lazy; he walks sitting down!
What was the boy riding on that caused the In-
dian to say, 'He walks sitting down'?

TCS, Verbal Reasoning (instructions: find the true
statement)
All bicycles have gears.
Some bicycles have ten speeds.
Maria has a bicycle.
P. Maria likes her bicycle.
G. Maria's bicycle has gears.
H. Maria's bicycle goes too fast.
J. Maria's bicycle has ten speeds.

8. Stanford-Binet, Induction
(This is a sequential test in which paper is
folded and holes cut in it by the examiner. The
student must deduce a pattern to predict how
many holes will result from each cut.)

TCS, Sequences
(Students are presented sequential patterns that
are incomplete. The student must deduce the
pattern and predict the final pattern.)



Need for This Study


Teachers at the seventh-grade level typically experi-

ence more difficulty identifying potentially gifted stu-

dents than do teachers at lower grade levels (Schnell,

1982). Consequently, junior high school teachers refer a

larger proportion of not-gifted students for evaluation

than do teachers of elementary school students. This

phenomenon is believed to result from more limited contact

between individual teachers and students and from the fact

that the pool of potentially gifted students from which











junior high school teachers choose contains none of the

students who have been identified as gifted during

previous years. Junior high school teachers must base

their referrals on "the best of the rest." Clearly the

most advantageous time to identify students for a junior

high school curriculum for the gifted is when they enter

seventh grade.

The new procedure (NP) was developed by selecting

items from popular group tests administered at seventh-

grade, measuring both achievement and cognitive ability.

The rationale for choosing screening items from standar-

dized group tests is that this approach to student screen-

ing is both time- and cost-efficient. All students in the

population school district take the CTBS and TCS annually

and the screening data are readily available without addi-

tional tests or testing time being needed.

It was believed that of the 460 CTBS and TCS ques-

tions typically administered to seventh graders, there

existed a subset of items that would accurately predict

gifted IQ on the WISC-R or S-B. Because the number of

CTBS/TCS items is large, sampling of a wide range of

skills and abilities is possible. As the diversity of the

items increases, so does the general ability measured by

the total test (Kaufman, 1979).










The utility of finding a small pool of items that

correlates highly with gifted intelligence is that (a)

current forms of the CTBS/TCS selected items may be re-

tained for group administration to future gifted class

candidates, (b) if a gifted candidate is not present for

CTBS/TCS testing the entire test will not have to be

administered as a gifted screening, and (c) the procedure

of analyzing a test in this way might be useful for pre-

dicting intelligence (or other traits) among other popula-

tions, utilizing these or other tests.



Overview of the Remainder of the Study


The subsequent content of this study is divided into

four chapters. In Chapter II a review of related liter-

ature is presented. A description of the methodology used

for the research comprises Chapter III. Research results

are presented in Chapter IV and results are discussed in

Chapter V.















CHAPTER II
LITERATURE REVIEW



Support for the Problem


The psychometric screening of "gifted intelligence"

is beset by problems of not only time- and cost-efficiency

but of predictive accuracy. In the relevant literature it

is suggested that these problems exist for a variety of

screening methods and procedures. These problems are not

of recent origin, however. As early as 1959, Pegnato and

Birch found that sufficient psychological services were

rarely available to test all of the gifted class candi-

dates, thereby necessitating procedures for screening

prior to formal testing. Accordingly, the authors con-

ducted an investigation of the "relative efficiency and

effectiveness" (p. 300) of seven procedures for locating

gifted children in junior high schools: teacher ratings,

class rank, creative ability in art or music, student

council membership, superiority in mathematics, group

achievement, and group IQ. Seven hundred eighty-one met-

ropolitan school district students were selected for par-

ticipation in the study on the basis of high ratings in

one or more of the seven categories. All of the partic-

ipants received the Stanford-Binet. Scores on this












intelligence test were used as a criterion reference for

the designation of those children who were, indeed,

gifted. For each of the seven screening procedures,

effectiveness was judged by the percentage of gifted chil-

dren located; efficiency was defined as the ratio between

total number of gifted students and students predicted as

gifted. Of the 781 subjects, 91 (6.5% of the school popu-

lation) were judged to be gifted. These results indicated

that, among the seven methods, group IQ and achievement

tests were the better predictors, providing the best pos-

sible combination of effectiveness with efficiency. Other

methods, such as honor role inclusion, were fairly effec-

tive, but their efficiency was poor.

Even though Pegnato and Birch (1959) were largely un-

successful in finding an effective predictor of gifted IQ,

a substantial amount of research has focused on screening

the gifted since that time. The following sections of

this literature review are concentrated on four procedures

for gifted student identification that involve group IQ

tests, IQ short forms, IQ screening tests, and achievement

test scores.


Group IQ Tests

The largely unsuccessful attempt by Pegnato and Birch

(1959) to predict gifted intelligence using group IQ tests

was followed one year later by a similar study. Chambers











(1960) sought a screening instrument for use in a Michigan

school district. Using the IPAT (Cattell's test of gen-

eral intelligence), the California Test of Mental Matu-

rity, the SRA Primary Abilities Test, The Kuhlman-Anderson

Intelligence Test, and the WISC, Chambers tested 39 chil-

dren in grades three through six. For each screening

test, a cutoff was calculated above which all gifted stu-

dents (WISC IQ>124) would be identified. The accuracy of

each screening procedure was established at 100%, and the

efficiency was then determined based on the number of not-

gifted students misclassified by the screening procedure

a; (ift d. The rc;ult; rc.vc;ilt l d thilt t 'he S;A to(s;t nnd the

Kulhman-Anderson could be ranked respectively as the most

and least efficient, and that between 20% and 57% of the

students predicted as gifted were not.

Three years after Chambers' study, Blosser (1963)

tested 187 ninth graders on the Henmon-Nelson and Otis

group intelligence tests. The research sample had a mean

IQ of 120 on the Stanford-Binet with a range of 98 to 153.

The results indicated that of the 36 students predicted as

gifted by the Otis, only 13 (36%) were identified as such

by the Stanford-Binet. On the Henmon-Nelson 13 of 26 stu-

dents (57%) were correctly predicted as gifted. Because

19% of the gifted students were not identified by either

group test, both tests proved to be poor predictors of

giftedness.












More recently, Harrington (1982) also found that

group IQ tests tend to underestimate the IQs of many in-

tellectually gifted students. According to Harrington,

for every student identified as gifted on a group IQ test,

one gifted child is not referred. Harrington suggested

that the higher the ability level, the greater the dis-

crepancy between individual and group IQ scores. He also

found that a child's IQ may vary by as much as 30 points

between group and individual tests. Further, because

there may be a very small number oF items at the greater

difficulty levels on group tests, a child may have to per-

form perfectly to be predicted as gifted.


Intelligence Quotient Short Forms

So-called IQ short forms are comprised of abbreviated

versions of individually administered standardized intel-

ligence tests. Typically selected for short forms are

subtests of the WISC-R or items from the Stanford-Binet.

Short forms of the Wechsler Scales and Stanford-Binet have

been studied extensively (Birch, 1955; Carleton & Stacey,

1954; Enburg, Rowley, & Stone, 1961; Findley & Thompson,

1958; Grossman & Galvin, 1987; Meister & Kurko, 1951;

Nichols, 1962; Simpson & Bridges, 1959; Wright & Sandry,

1962; Wade, Phelps, & Falasco, 1986; Yakowitz & Armstrong,

1955; Zimet, Farley, & Dahlen, 1985). However, it is only

since 1978 that short forms have been relatively widely











studied as a method for screening potentially gifted stu-

dents.

An early attempt to predict gifted IQ using a short

form test was conducted by Thompson and Findley in 1962.

Finding that the Similarities (S), Information (I), Pic-

ture Arrangement (PA), Block Design (BD), and Picture Com-

pletion (PC) WISC subtests could be effectively used for

this purpose, Thompson and Findley published the Califor-

nia Abbreviated WISC for the Intellectually Gifted (CAW-

IQ) in 1966).

In their study, Killian and Hluyhes (1978) measured

the effectiveness of the Lyman short form (Lorr & Meister,

1942) and the Vocabulary-Block Design subtests of the

WISC-R dyad for predicting IQ on the Stanford-Binet and

WISC-R respectively. Subjects were 142 students between

5- and 15-years-old possessing a mean IQ of 125. Results

indicated a correlation of r=.92 between the WISC-R and V-

BD dyad whereas the Stanford-Binet and Lyman scores were

correlated at r=.78. Killian and Hughes did not present

results of the actual number of students correctly pre-

dicted as gifted. They did, however, indicate that 32% of

the students had short form/Full Scale IQ score discrep-

ancies of 6 points or more.

Employing a much larger sample of students than did

previous researchers, Karnes and Brown (1981) used











Silverstein's (1970) method of deriving "the best short

form combinations" (p. 169) to obtain an accurate gifted

IQ predictor. Silverstein's method takes subtest unreli-

ability into account when measuring predictive ability.

Nine hundred, forty-six gifted children ages 6.0 to 16.0

(X chronological age [CA] = 9.9) served as subjects.

Karnes and Brown found that the WISC Block Design subtest

was represented frequently in subtest combinations that

correlated with WISC Full Scale IQ. Supporting Killian

and Hughes' findings, the V-BD dyad was found to be the

most accurate for predicting gifted IQ. The use of sub-

test tetrads was found to be useful, increasing correla-

tion coefficients from .628 to .734. Again, actual accu-

racy ratios were not provided in the study.

Proceeding under the notion that, "since a short form

IQ test is composed entirely of some subset of questions

of items taken directly from a full-length IQ test, a

short form would seem to be an ideal predictor of full-

length IQ test performance" (p. 40), Dirks, Wessels, Qua-

forth, and Quenon (1980) compared various short form com-

binations with Full Scale IQ on the WISC-R. Subjects con-

sisted of 47 fourth graders with a mean IQ of 123 (range =

106 to 144). Twelve WISC-R subtest combinations were

studied. It was revealed that the short form combinations

of Similarities, Object Assembly and Vocabulary and S-OA










were each good predictors of Full Scale IQ. Although cor-

relations on the BD subtest were high, as shown in previ-

ous studies, they tended to predict an excessive number of

nongifted students as gifted. The S-OA dyad predicted 8

of 11 gifted students and 4 who were not. The S-OA-V

triad predicted 9 of 11 gifted students and 4 who were

not.

Utilizing the studies by Killian and Hughes (1978)

and Dirks et al. (1980), who noted that V-BD and S-OA

dyads, respectively, were the most effective in predicting

Full Scale IQ, Fell and Fell (1982) evaluated 92 WISC-R

protocols of children previously evaluated as gifted pro-

gram candidates. The students ranged in age from 6-0 to

11-7 (X age = 8.4) and possessed Full Scale IQs of 130 or

greater. Eleven subtest dyads were studied in terms of

frequency with which each produced an estimated IQ 2 130.

Greatest predictive accuracy was achieved using the S-V

and S-OA dyads. These correctly predicted 62% as gifted.

The I-BD dyad yielded prediction ratings of only 43%. Not

providing an exact number, the authors indicated that some

gifted children were overlooked. While results are con-

sistent with findings by Dirks et al., indicating that the

S-OA dyad is most effective, prediction accuracy was much

lower in this study.

In a fairly recent study, Kramer, Markley, Shanks,

and Ryabik (1983) utilized Thompson and Findley's (1966)











CAW-IQ on a sample of 73 children, ages 6-0 to 16-7 (X age

= 10-5). All subjects received the WISC-R and all were

analyzed in terms of the S, I, PA, BD, and PC subtest pat-

torn. Of the 48 students predicted as yifted, 39 were

predicted accurately; of 25 students predicted not to be

gifted, 21 were correctly described. This subtest short

form was considered to be a relatively accurate predictor

of gifted IQ.


Intelligence Quotient Screening Tests

Individually administered intelligence tests designed

to estimate mental ability, usually in 20 minutes or less,

have become a widely used procedure for screening gifted

intelligence. The Slosson Intelligence Test (Slosson &

Jensen, 1982) is one such screening test. It had been

adopted in the Hillsborough County Florida school district

for the purpose of screening the gifted. High correla-

tions between the SIT and the WISC-R or Stanford-Binet

have been reported in research findings (Lawrence & Ander-

son, 1979; Martin & Kidwell, 1977; Martin & Rudolph, 1972;

Mize, Smith, & Callaway, 1979; Ritter, Duffy, & Fischman,

1973; Slosson & Jensen, 1982; Stewart & Jones, 1976).

However, the few available studies conducted on gifted

samples have not supported the use of the SIT for screen-

ing.










In the previously discussed study (see Chapter I) by

Grossman and Johnson (1983), the Otis Lennon group IQ test

was found to be a better predictor of gifted achievement

than the SIT for high achieving students. Dirks, Wessels,

Quaforth, and Quenon (1980) also found the SIT to be a

poor predictor of gifted ability. They administered the

SIT to 47 academically talented fourth graders. The stu-

dents were also administered the WISC-R to determine their

actual IQ scores. Of the 11 students who were found to

possess gifted intelligence (IQ2130), only 8 were identi-

fied as such by the SIT. In addition, the SIT falsely

predicted gifted intelligence in 9 of the 38 nongifted

children. The researchers concluded that the SIT alone

should not be used to predict IQs of gifted children. The

SIT has also been found to significantly overestimate IQ

scores. Machen (1972) investigated the reliability and

concurrent validity of the SIT with the WISC, using 5

gifted children ages 9 through 11. The results revealed a

significant correlation between the two tests, though the

SIT tended to overestimate the WISC by at least one stan-

dard deviation. Additionally, the SIT has been shown to

underestimate IQ scores. Mize et al. (1979) found, in

their study of 207 students from all grade levels, that of

students with above average intelligence, 24% were overes-

timated and 24% were underestimated by 11 or more IQ

points on the SIT.











In 1979 Karnes and Brown further examined the ten-

dency of the SIT to over- or underestimate IQ scores. In

this study the validity of the SIT in relation to the

WISC-R was assessed for a group of 79 gifted children ages

6 through 12. A SIT-WISC-R correlation of r=.48 was cal-

culated; this coefficient was significant at the .001

level. The authors also computed a regression equation

with which to predict WISC-R IQ from the SIT. Results

indicated that at the lower ranges of SIT scores, it

tended to underestimate the WISC-R, while at the upper

ranges IQ was overestimated. Despite the high correlation

between the two tests, it is apparent that Karnes and

Brown were not confident in the SIT's predictive ability

because they recommended using two SEMs for the SIT when

screening gifted IQ to ensure that most gifted students

are identified. To obtain 95% accuracy, a cutoff score of

105 would have been necessary. However, Karnes and Brown

did not indicate how many nongifted students would have

been predicted as gifted using a cutoff this low.

In a similar study, presented at the Annual Meeting

of the Alabama Association of School Psychologists in

1983, Apple discussed the precision of the SIT in pre-

dicting WISC-R IQ of 61 gifted students ages 6 to 11.

Differences in scores were compared by use of independent

t-tests. The results supported the findings of Karnes and










Brown that at the lower SIT ranges WISC-R IQs were under-

estimated and that at the upper SIT ranges the WISC-R IQs

were overestimated. Apple concluded that valuable diag-

nostic information yielding a qualitative picture of the

child's strength is omitted when the SIT alone is used as

a screening indicator.

Whereas Karnes and Brown as well as Apple compared

SIT and WISC-R scores for youngsters already placed in

gifted classes, Rust and Lose (1980) attempted to accu-

rately screen potentially gifted students in first through

seventh grade. Based on teacher referrals and SIT scores

of 130 or above, 438 students were found eligible for

WISC-R evaluation. Of these, 132 were utilized in the re-

search sample. According to stepwise regression equa-

tions, the SIT was found to be a significant predictor of

Full Scale IQ. However, of the 132 students predicted,

only 61 achieved WISC-R IQs of 130 or above. Thus, set-

ting the SIT cutoff at 130 failed to screen out 54% of the

nongifted students. If a cutoff of 134 had been used, as

suggested by Karnes and Brown, 42 evaluations would have

been eliminated. However, of those 42, 12 would have been

gifted. Karnes and Brown noted that while there was a

high correlation between the SIT and WISC-R, there was a

great deal of variability with individual cases. It was

concluded that in all studies high error can be expected











when the SIT is used to predict WISC-R IQ among the

gifted.

Other IQ screening tests such as Guilford's Structure

of the Intellect Test (SOI) (Pearce, 1983; Stenson, 1982),

Ravens Progressive Matrixes (Pearce, 1983; Petty & Field,

1980), The Peabody Picture Vocabulary Test (Mize et al.,

1979; Pedriana & Bracken, 1982), and The Ammons Quick Test

(Joesting & Joesting, 1971; Kendall & Little, 1977; Nich-

olson, 1977) have been correlated with the WISC, WISC-R,

and the Stanford-Binet. In some instances significant

correlations have been found. However, very few studies

have been conducted with samples of gifted students. In

one such study, DeFilippis and Fulmer (1980) found that

the Ammons Quick Test underestimated WISC-R IQ for 99

first, fourth, and seventh graders with high ability.

In another study involving samples of gifted stu-

dents, Wright (1983) correlated WISC and Peabody Picture

Vocabulary Test (PPVT) scores of 35 students referred by

teachers for gifted program testing. A correlation of

r=.27 was calculated and it was found that nearly half of

those who scored two standard deviations above the PPVT

mean were not eligible for gifted program placement based

on WISC-R IQ scores. Wright recommended that the PPVT not

be used to screen gifted program candidates.

A third study was conducted by Stenson (1982) to de-

termine the concurrent validity of the Structure of












Intellect (SOI) Gifted Screener with the WISC-R. The sub-

jects were 3239 elementary school students. A multiple

correlation of r=.337 was significant at <.05; however,

only 11% of the variance in WISC-R scores was explained by

the Gifted Screener. No predictor variable contributed to

a significant multiple correlation coefficient when Full

Scale IQ or any combination of WISC-R subtests was used as

the criterion variable. Stenson concluded that the Gifted

Screener should not be used to predict WISC-R IQ for

gifted program prospects.

In 1985, Clarizio and Mehrens evaluated the technical

data manuals for the SOI to determine the test's value as

a screening test for gifted intelligence. It was con-

cluded that "the SOI model has severe psychometric limita-

tions" (p. 119). These limitations center around poor

reliability, inadequate normative data, and poor external

validity for many of the factors measured by the test.


Achievement Tests

In research introduced in Chapter I it was suggested

that achievement tests (CTBS) and cognitive ability tests

(TCS, WISC-R, S-B) measure much the same construct. In

the supportive literature were indications that distinc-

tions between achievement tests and cognitive ability

tests are often unclear. Correlational and factor

analytic studies lend credence to this contention.












Lennon (1978) has found that relationships between

intelligence and achievement tests are so strong as to

lead to the criticism that the two types of tests do not

measure anything different. Both tests measure what the

student has learned (Gronlund, 1976) and both tests pre-

dict future learning with similar degrees of success. IQ

tests and achievement tests differ in form but not neces-

sarily in content (Mercer, 1979).

In a series of studies in the late 1970s and 1980s

WISC-R IQs were correlated with achievement subtests of

the Wide Range Achievement Test (WRAT). Consistently high

correlations were found. Some of these early studies are

summarized in Table 1-1.

Also, in 1978, Stedman, Lawlis, Cortner, and Achten-

berg attempted to relate Kaufman's (1975) factor scores to

WRAT attainment in a population of 76 children, ages 6 to

13. Correlations were found to be positive and signif-

icant.

Yule, Gold, and Busch (1981) administered the WISC-R

and a battery of achievement tests to students at age 16

1/2. Achievement measures included tests of "sentence

reading," spelling, and arithmetic. WISC-R, Verbal IQ

shared 50% of the variance in reading, spelling, and

arithmetic. Correlations between Full Scale IQ and

achievement were as high as r=.80.













Table 2-1. Summary of Relationship between
WISC-R and WRAT.


WRAT
WISC-R Reading Spelling Arithmetic


Brooks (1977) N=30; 6-10 years
V..S. IQ 64a 55 74
P.S. IQ 71 70 71
F.S. IQ 70 65 76

Hartlage and Steele (1977) N=36;
Mean age = 7 yrs 9 months
V.S. IQ 75 35 76
P.S. IQ 54 33 67
F.S. IQ 68 35 76

Schwarting and Schwarting (1977) N=282; 6-16 years
(a) 6-11 yrs
V.S. IQ 68 61 69
P.S. IQ 63 60 69
F.S. IQ 72 65 75
(b) 12-16 yrs
V.S. IQ 74 69 66
P.S. IQ 40 34 55
F.S. IQ 62 56 66

Hale (1978) N=155; 6-16 years
V.S. IQ 54 49 64
P.S. IQ 29 26 44

Full Scale correlations not quoted.


aDecimal points omitted.



In 1982 a follow-up to the studies summarized in

Table 2-1 was conducted by Grossman and Johnson. In their

study, 77 students ages 6 to 16 were administered the

WISC-I and the WRAT. Factor scores were computed on two

of Kaufman's (1975) factors (Verbal Comprehension and












Freedom from Distractibility) and WRAT subtests. A multi-

ple regression analysis was computed wherein WISC-R factor

scores served as conjoint predictors and the WRAT standard

scores were employed as criterion variibl l's. Kresults in-

dicated a significant overall prediction of WRAT reading,

spelling and arithmetic by the two WISC-R factors.

Wright and Dappan (1982) assessed 250 students with a

mean age of nine years on the WISC-R and WRAT. Factor

analysis showed a common factor for subtests on both mea-

sures. Correlations between subtests from the two tests

were as high as r=.60 (on WISC-R, Arithmetic and WRAT,

Arithmetic). Some other subtests correlated at .40 to

.50.

Literature concerning the overlap of individual tests

of intelligence and tests of achievement include studies

in which the WISC-R and the Peabody Individual Achievement

Test (PIAT) (Dunn & Markwardt, 1970) were examined.

Wikoff (1978) factor analyzed the WISC-R along with the

PIAT for 180 referred children. Although the PIAT General

Information and Mathematics subtests loaded on factors

previously identified in the structure of the WISC-R, the

remaining subtests loaded on a separate factor, subse-

quently labeled Word Recognition. The results supported

the use of both instruments as sources of mutual but













supplementary information in the assessment of learning

problems.

Dean (1977) assessed the degree of redundancy between

the WISC-R and the PIAT using a canonical correlation

analysis with scores from 205 referred children. The re-

sults indicated that 65% of the functions of the PIAT

overlapped with the WISC-R and that 37% of the functions

of the WISC-R overlapped with the PIAT. The overlap was

attributed to common verbal-educational content. Dean

(1982) found a similar asymmetrical overlap between these

measures in samples of 100 Anglo and 100 Mexican-American

children. As in WikoLl s Lactor analysis, both ot Deans

analyses showed the PIAT subtests of reading and spelling

to offer the greatest degree of information not redundant

with the WISC-R.

Brock (1982), finding that a paucity of research

existed for factor analytic investigations of the WISC-R

in combinations with individual achievement tests, con-

ducted such a study. He factor analyzed the WISC-R, WRAT,

and PIAT for 183 male students in grades 3 through 6. An

attempt was made to determine the traits or common skills

measured by IQ and achievement tests when viewed concom-

itantly. Four factors emerged. One, a numerical factor,

was comprised of subtests from all three tests.












Moderately high correlations r=.40 to .50 were found be-

tween some of the other IQ and achievement subtests.

Stewart and Morris (1977) factor analyzed the WISC,

WAIS, WRAT, and CAT (California Achievement Test) for 182

students ranging in age from 11 to 18. A "substantial"

overlap of verbal intelligence and academic achievement

was found. Resulting factors conformed reasonably well to

those of Kaufman (1975). Subtests from each measure were

found to load on each of the IQ factors.

In a study wherein the abilities underlying reading

readiness were identified, Olsen and Rosen (1971) factor

analyzed three group reading tests and the WISC. Subjects

consisted of 218 first graders. The 35 subtests were cor-

related and the resulting matrices subjected to a prin-

cipal component analysis. Four common factors were re-

vealed. In one factor, reading comprehension loaded with

four WISC-R subtests. In another, "writing letters" cor-

related highly with WISC-R Vocabulary. In a third factor,

sentence writing and WISC-R Coding were included.

There have emerged two camps of thought on the issue

of reading skill acquisition and intelligence. On one

hand, in some research it has been suggested that reading

is a function of information processing or encoding skills

as opposed to being a primarily intellectual function. On

the other hand, in similarly focused factor analytic

investigations reading has been found to be highly loaded













on a general intelligence factor and a good predictor of

intelligence.

Researchers whose views represent the latter view-

point have supported the notion that reading ability is

highly correlated with general intelligence and is a good

predictor of intelligence. Jensen (1981) reported a cor-

relation of r = .68 between reading comprehension and Full

Scale IQ for a large sample of students. Other research-

ers have cited similarly high (.60-.70) correlations be-

tween reading and IQ for various samples of students in

grades K through 12 (Brooks, 1977; Hale, 1978; Hartlage &

Steele, 1977; Ryan, 1979; Wikoff, 1978; Yapp, 1977; Yule,

Gold, & Busch, 1981). In a literature review of 34

studies, Hammill and McNutt (1981) found a median correla-

tion of .75 between measures of intelligence and achieve-

ment.

Reasoning that the most efficient gifted screening

assessment would be significantly correlated with achieve-

ment if giftedness is defined as superior school-related

ability, Grossman and Johnson (1983) investigated the

Stanford Achievement Test. They found a significant cor-

relation with intelligence among 46 children with SIT IQs

above 120.

In another pertinent study, Schena (1963) found that

of 226 sixth and seventh graders who scored two or more











"levels" above the norm on the Metropolitan Reading Test,

61% scored above 130 on the Stanford-Binet. In his 1984

study, Sternberg found that IQ accounted for as much as

25% of the variance in scholastic performance.

In two of the few other studies in which achievement

was correlated with intelligence among superior students,

Mayfield '(1979) had 573 third graders evaluated in terms

of intelligence, achievement, creativity, and teacher per-

ception of IQ. Results yielded significant correlations

between intelligence and a wide range of achievement do-

mains among the student sample. Similarly, Karnes,

Edwards, and McCallum (1986) found a significant correla-

tion between total scores on the California Achievement

Test (CAT) and WISC-R Full Scale IQs of 41 gifted children

in grades four through six.

Thus the results of this body of literature comparing

intelligence with achievement appear conclusive, as in a

substantial number of studies it is suggested that the two

variables are fairly highly correlated.

Mallinson (1963) attempted to uncover a relationship

between intelligence and achievement in science and math.

The SRA achievement series and the SRA Primary Abilities

Test were given to secondary grade students. There was a

resulting correlation of r=.65 between verbal ability and

science (facts and principles). Verbal ability was also












found to have reasonably high correlations with factors of

arithmetic achievement.

In another study in which factors related to math

performance were investigated, Roach (1979) reported a

significant correlation (r=.80) between arithmetic

achievement and verbal IQ in third graders.

Correlation coefficients for CTBS and TCS subtests

(CTB/McGraw-Hill, 1984) were calculated using 2813 seventh

graders. Coefficients between .60 and .72 were not uncom-

mon. Correlations between the TCS Total Score and CTBS

subscales of Reading, Language, Math, Social Studies and

Science were .71, .71, .68, .65, .71 and .65, respec-

tively. The CTBS and TCS Total Batteries correlated at

r=.75. These high correlations suggest that both tests

may measure similar, though operationally distinguishable,

constructs. Support for the contention that these tests

represent similar constructs may also be found in the cor-

relations of TCS subscale scores with those scores of its

predecessor, the Short Form Test of Academic Aptitude

(SFTAA). The range of correlations was .55 to .82, not

dissimilar to those of the TCS and CTBS. The fact that

the average correlations were positive means that the sub-

scales must be measuring something in common (Jensen,

1980).

In summary, there is substantial evidence that

achievement test scores and IQ test scores are highly













correlated. The research also provides reason to believe

that a common factor underlies performance on both types

of tests.


Summary

In the literature relevant to the accuracy of various

procedures for screening intelligence among gifted stu-

dents, there have generally been mixed results. Some

encouraging findings have occurred on studies of achieve-

ment ratings and short-form IQ tests. Group IQ tests have

tended to underestimate the IQs of some gifted children,

though they generally predicted gifted IQ with moderate

accuracy. Much less effective means for predicting gifted

IQ are IQ screening tests.

The preceding literature has focused on the problems

of screening gifted intelligence among the school age pop-

ulations. Those procedures that have shown some success

have been inconsistent in their findings. Nearly all have

proved inefficient in terms of time and cost.



Instruments Used in Study


Wechsler Intelligence Scale for Children-Revised (WISC-R)

The WISC-R has been the most widely administered test

of children's intelligence (Bryan & Bryan, 1975; Grossman











& Galvin, 1987; Salvia & Ysseldyke, 1978; Vandiver & Van-

diver, 1979). In much of the research surrounding this

instrument it is suggested that it merits this distinc-

tion. Friedes (1978) described the standardization of the

WISC-R as "state of the art" (p. 232) and as meriting

"blue ribbons." In addition, he noted as praiseworthy the

high correlation coefficients between the WISC-R and the

Stanford-Binet.

Reliability coefficients of internal consistency for

the WISC-R Verbal, Performance, and Full Scale IQ scores

reported in the test manual were obtained by utilizing a

formula for computing reliability of a composite group of

tests (Wechsler, 1974). The average reliability coeffi-

cients across the range of age levels were V, r=.94; P,

r=.90; and FS, r=.96. The coefficients of individual sub-

tests based on split-half or test-retest methods ranged

from r=.77 to .86. Test-retest correlations for the Ver-

bal, Performance, and Full Scale IQs ranged from r=.90 to

.95 based on a 3-month interval between tests.

Factor analytic research by Kaufman (1979) has shown

that factors corresponding closely with the Verbal and

Performance Scales of the WISC-R exist. In 1980, Karnes

and Brown factor analyzed the WISC-R on 946 gifted stu-

dents ages 6.0 to 16.0. The resulting factors were

consistent with those found by Kaufman on the normal popu-

lation. Most verbal scale subtests had factor loadings in











Perceptual Organization. These studies strongly support

the validity of the WISC-R.


Slosson Intelligence Test

The SIT is an IQ test for children and adults de-

signed for use by either relatively untrained examiners or

qualified professionals. The SIT typically takes between

10 and 30 minutes to administer.

New norms (1982) represent a significant departure

from procedures previously employed (1961) for calculating

an IQ score. In norming the SIT, the Stanford-Binet was

used as the anchor test. Consistent with the 1974 revi-

sion of the Stanford-Binet, ratio IQs were abandoned in

favor of deviation IQs. Frequency distributions were cal-

culated for each of the 19 chronological age ranges on

both IQ scales. Then, utilizing a "modified table look-up

approach," appropriate IQs from the Stanford-Binet were

entered on the developing SIT tables. The mean IQ for the

SIT is 100 and the standard deviation is 16.

In the SIT manual, the authors present evidence to

persuade the reader that the revised SIT IQs are equiva-

lent to Stanford-Binet IQs. This task is undertaken, in

large part, by comparing previous (1961), less positively

correlated coefficients to newer data.

Slosson and Jensen (1982) stated that "the SIT is as

accurate as the Stanford-Binet in measuring a person's











intelligence when both instruments have been properly

administered" (p. 16) and further proposed that the SIT

qualifies as an alternate form "of the Stanford-Binet be-

cause the two tests possess equivalent means and standard

deviations" (p. 16). Based on their dubious assumption of

test equivalency between the revised SIT and the S-B, the

authors employed the Mean Absolute IQ Difference (MAD)

statistic to determine alternate form test reliability and

standard errors of measurement. The MAD procedure, which

is meant to be used only with equivalent forms of a test,

yields a statistic which is approximately equal to the

standard deviation times 8862. The authors did not indi-

cate the relative effects on the reliability and the SEMs

when the measures compared do not strictly meet the crite-

ria of alternate forms, as is apparent in this case. Nor

are there attempts to evaluate other kinds of test relia-

bility. It might be concluded, therefore, that the

authors' claim that "the SIT's reliability may be regarded

as not less than .95" (p. 136) should be interpreted with

caution.

Another claim made in the SIT manual is that the mean

difference between the Stanford-Binet and Slosson IQ

scores is less than one point, based on the sample of

1,109. The procedure by which this statistic was obtained

entails computing the means for IQ differences between the











two tests for three IQ levels across four age groups: be-

low 84, 84-116, and above 116. For example, at age 13-6

and above, mean differences between the Stanford-Binet and

SIT are -1.41, -1.10, and 2.62 at the three IQ levels. In

calculating the mean difference, negative means are added

to positive means resulting in a misleadingly low overall

mean difference. In this example the total mean differ-

ence for age 13-7 and above is -.67. However, if individ-

ual means had been summed in terms of nondirectional devi-

ation from zero, the mean difference would have been

approximately 1.7. With regard to the mean difference for

the entire sample, when the nondirectional procedure is

used the difference changes from -.04 to approximately

1.4. The mean scores are rendered even more difficult to

interpret because no standard errors for the means are

reported.

In spite of the apparent inconsistencies in the new

SIT manual, the revisions, particularly in its renorming,

represent considerable improvement in the test's validity

and reliability. These test improvements, along with the

ease of its administering and scoring, render the SIT a

test of considerable utility as an intellectual screening

procedure.











Comprehensive Test of Basic Skills (CTBS)/Test of
Cognitive Skills (TCS)

The CTBS (CTB/McGraw-Hill, 1984) and TCS (CTB/McGraw-

Hill, 1984) are the tests from which NP items were taken.

Psychometrically, these tests were well suited to this re-

search. The appropriateness of the CTBS and TCS for this

research are supported by several of their attributes,

some of which were discussed in Chapter I.

Items were chosen for both tests according to item

response theory (IRT) utilizing a three-parameter logistic

model. The items were chosen according to their ability

to (a) discriminate high ability traits from low ability

traits, (b) discriminate high ability students and low

ability students by matching item difficulty with student

total score, and (c) account for guessing as an influence

on score difficulty.

In terms of content validity, the CTBS is designed to

measure understanding of a broad range of concepts as

developed by various educational curricula. Test perfor-

mance reflects a student's skills in effective use of

information explicit in categories derived from Bloom's

taxonomy (Bloom, 1956). Item development specifications

were designed to ensure comprehensive coverage of the con-

tent and process categories.

The TCS is designed to measure an aptitude construct

that can be operationally distinguished from the











achievement construct of the CTBS, based on research con-

ducted at McGraw-Hill by Buchet (1974, cited in

TCB/McGraw-Hill, 1984). Empirical criteria for distin-

guishing between aptitude and achievement measures were

derived by the publishers.

Product moment correlations between the four subtests

of the TCS. were between r=.41 to r=.65. Coefficients be-

tween subtests and total score ranged from r=.72 to r=.85.

Therefore, it was suggested that all subtests measure gen-

eral intelligence but also measure independent factors. A

correlation coefficient between the CTBS and TCS of r=.78

was calculated on a sample of seventh graders.

Another attribute of the CTBS and TCS is the compre-

hensive sampling and norming standards applied. The norm-

ing samples contained approximately 250,000 students in

grades K-12 from public, Catholic, and other private

schools (CTB/McGraw-Hill, 1984). School districts were

randomly chosen from four geographic regions. Comprehen-

sive norming and standardization information is available

in the CTBS and TCS Technical Reports.

Internal reliability coefficients were calculated

according to the Kuder-Richardson formula 20. CTBS relia-

bility coefficients ranged from .30 to .96 on the 10 sub-

tests (CTB/McGraw-Hill, 1984). All subtests except Spell-

ing and Reference Skills had values at or above .90. On

the four TCS subtests reliability coefficients ranged from












.80 to .84. The TCS Technical Reports provide reliabil-

ities on SEMs for subtests based on number correct at each

grade level. Composite calculations for the total test

are not provided. Also reported are bias studies and

tables indicating how test biases are accounted for and

controlled.

In summary, the CTBS and TCS were well suited for

this study because of the sophisticated method utilized in

analyzing items and the tests' high validity and reliabil-

ity. Further, both tests employed sampling procedures

designed to provide norms for the entire U.S. school popu-

lation. Research has also been conducted to aid in reduc-

ing test bias for the CTDS and TCS.


Stanford-Binet Intelligence Scale (S-B)

The third revision of the Stanford-Binet (S-B), pub-

lished in 1960, remained unchanged in content and format

through 1985. A revised version of the S-B was published

in 1986. The 1960 version was constructed by combining

forms L and M of the 1937 scale and eliminating those

items considered obsolescent and by relocating items whose

difficulty level had altered during the intervening years.

The test was, however, restandardized in 1972. New norms

were derived from a sample of approximately 2,100 cases

during the 1971-72 school year. Children in the 1972 norm

group were chosen from 20,000 school age children in












grades 3 through 12 who were identified based on scores

from the Cognitive Abilities Test. The distribution of

scores in this subsample corresponded to the national dis-

tribution of the entire sample. The 1972 norms were be-

lieved to be based on a more representative sample than

previous norms (Terman & Merrill, 1973).

The reliability of the 1937 Stanford-Binet was deter-

mined by correlating IQs on forms L and M administered to

the standardization group within an interval of one week

or less. Such reliability coefficients are thus measures

of both short term temporal stability and equivalence

across the two item samples. In general, the Stanford-

Binet tends to be more reliable for older than for younger

age groups, and for lower than for higher IQs (Anastasi,

1976). Reliability coefficients range from .83 to .98.

The Stanford-Binet is considered a highly reliable test

with most coefficients for the various age and IQ levels

being over .90.

Validity ratings for the Stanford-Binet were obtained

from examination of the test content, from factor

analysis, and from correlations with achievement ratings.

An examination of the Stanford-Binet tasks indicates

assessment of a wide range of reasoning abilities. These

include tasks requiring hand-eye coordination, perceptual

discrimination, arithmetic reasoning, and verbal











reasoning. The most common type of test, especially of

the upper age levels, is that employing verbal content.

Data on criterion-related validity of the Stanford-

Binet have been obtained chiefly in terms of academic

achievement (Anastasi, 1976). Correlations between the

scale and school grades, teachers' ratings, and achieve-

ment test scores generally fall between .45 and .75. The

Stanford-Binet tends to correlate highly with performance

in nearly all academic courses, but predominantly with

verbal courses such as English and history. Correlations

with achievement test scores show the same pattern. The

rigorous standardization and ronorming of the Stanford

Binet, along with its high validity and reliability, indi-

cate that it was an appropriate IQ test for this study.















CHAPTER III
METHODOLOGY



Overview


In this research study a procedure was investigated

for analyzing a comprehensive, group-administered achieve-

ment and cognitive abilities test to determine whether an

item set can be derived that discriminates gifted from

high-achieving, but not-gifted, seventh graders. When

such an item set was derived, it was compared to a com-

monly used IQ screening test to assess the relative accu-

racies of each procedure in discriminating gifted from

not-gifted students in a second seventh grade population

sample.

This chapter is organized into the following sec-

tions: (a) Population and Sample, (b) Assessment Instru-

ments, (c) Research Procedures, (d) Data Analysis, and (e)

Methodological Limitations.



Population and Sample


The research sample of 179 students was drawn from a

population of seventh graders who had been tested on the

WISC-R or S-B for the "gifted program" in the Hillsborough











County (Florida) Public School District. Sampling was

conducted at the end of the 1984-85 school year. Nearly

all students had previously been administered the SIT. In

most instances, only those students who scored two stan-

dard deviations or more above the mean had been adminis-

tered the WISC-R or Stanford-Binet. All students in the

population also had current CTBS/TCS scores on file. Some

students in this population had met intellectual eligibil-

ity guidelines for the gifted program (on WISC-R or S-B

criteria) and some had not. All students were tested by

school psychologists during each of the three school years

under investigation.

Simple random sampling was conducted by the research-

er at the Hillsborough County School Board office in June

of 1986. Names of the seventh graders who were tested for

the gifted education program from September of 1982 and

June of 1985 were obtained from computer printouts con-

taining data for all students in the district who had been

tested by school psychologists. Students in the sample

pool were assigned a number, and numbers were selected

according to a random number table. Numbers were then

recorded and returned to the pool to ensure an equal

chance of selection for the remaining numbers. After 61

students were randomly selected for Phase II, the remain-

ing 118 students were assigned to Phase I.











Assessment Procedures

As previously discussed, the five assessment instru-

ments used for this research were the Wechsler

Intelligence Scale for Children-Revised (WISC-R), The

Stanford-Binet Intelligence Scale (S-B), the Slosson In-

telligence Test (SIT), the Comprehensive Test of Basic

Skills (CTBS),' and Test of Cognitive Skills (TCS).

Administration, scoring, and interpretation of the

WISC-R and Stanford-Binet were conducted by state certi-

fied school psychologists prior to the onset of this

study. All tests were individually administered and hand

scored using current norms. The WISC-R yields a Verbal

Scale IQ (representing verbal reasoning abilities), a Per-

formance Scale IQ (representing perceptual organization

and nonverbal reasoning), and a Full Scale IQ. Only the

Full Scale score, which represents total IQ, was used as a

measure of gifted intelligence. This procedure conformed

to school district guidelines. The Stanford-Binet yields

a total IQ score only. A cutoff score of 127 was used as

the gifted cutoff in the district. Students attaining a

Full Scale IQ of 127 or greater were considered to have

met the intellectual criterion for gifted program elig-

ibility. The IQ cutoff was chosen by the district as a

score that is two standard deviations above the test mean

(WISC-R IQ=130), minus one standard error of measurement

(three IQ points) (S-B IQ = 132 minus 5 IQ points). The












Slosson Intelligence Test was administered to students as

an IQ screening procedure by school guidance counselors or

curriculum specialists who typically had little formal

training in administration of individual intelligence

tests. SITs were given to students within one year prior

to WISC-R testing. SIT protocols were hand scored by the

test administrators. Under usual circumstances, a total

IQ of 135 was used as a screening cutoff. Children who

scored at or above this cutoff point were normally re-

ferred to the school psychologist for WISC-R evaluation.

There were some exceptions to this rule because occasion-

ally students who did not score at or above the cutoff

were referred. Generally, these students exhibited ex-

tremely high academic skills or other competencies that

compelled school personnel to refer them for formal test-

ing. The SIT yields a total IQ score. The 135 cutoff

score is two standard deviations above the test mean.

The CTBS and TCS were administered to students by

classroom teachers in group format according to standard-

ization procedures found in the teacher's manual. The

CTBS and TCS were designed to be easily administered

(Ahmann, 1972), and teachers have received little formal

training in their administration. The seventh-grade level

of the CTBS (Level H) yields subscale scoring in Reading

(two sections), Spelling, Language (two sections),











Mathematics (two sections), Reference Skills, Science, and

Social Studies. The Science and Social Studies subscales,

consisting of 86 items, were not administered in all

schools of the population school district; however, these

data were included in this research. In addition to sub-

scale scores, the CTBS yields an overall achievement index.

The TCS is comprised of four cognitive ability sub-

tests measuring competencies in Sequencing, Analogies,

Memory, and Verbal Reasoning. Derived scores are provided

for subtests and for the overall profile. Because in this

study responses to individual test items were analyzed,

derived subscale and total scores for the CTBS scores were

not utilized.



Research Procedures


In Phase I of this study items comprising the new

procedure (NP) were selected from the CTBS and TCS by con-

ducting item analyses of the performance of the 118 stu-

dents in the Phase I sample. Two sets of items were de-

lineated that, in general terms, were answered correctly

by the gifted students more frequently than by the not-

gifted students.

Gifted cutoff scores on the obtained NP items were

computed by subtracting fractions or multiples of standard

deviations from the mean NP score of the gifted group to











determine which cutoff points) most accurately predicted

giftedness. Newly referred students in the school setting

who scored at or above that point would be referred for

formal IQ testing. By adopting a cutoff score two stan-

dard deviations below the gifted mean, approximately 97%

of intellectually gifted students would be referred for IQ

testing based on research sample parameters. A cutoff

score incorporating a one standard deviation below the

mean cutoff would delete approximately 16% of the gifted

students from testing and gifted program eligibility.

However, a pitfall of including as broad a range of stu-

dents as permitted by the two standard deviations crite-

rion was that a relatively large number of not-gifted stu-

dents would also be referred for formal IQ testing, thus

reducing the accuracy of NP predictions of giftedness.

Adjustment of the cutoff point was desirable to ensure an

optimal ratio of gifted students accurately predicted to

not-gifted students inaccurately predicted.

After items for the NP were obtained and multiple

cutoff points was established, Phase II was begun. .In

Phase II, test scores based on selected NP items were com-

puted for the second sample of students. The NP was an-

alyzed in terms of its discrimination between the gifted

and not- gifted students on WISC-R. The accuracy of clas-

sification was then compared to accuracy of classification











obtained using students' SIT scores as a screening

procedure in predicting gifted IQ on the WISC-R.

Data collectors and recorders consisted of the re-

searcher and employees of the district school board.

School board employees working in the testing and evalua-

tion office obtained CTBS/TCS item responses from computer

data. WISC-R and SIT scores were obtained in a similar

manner by the researcher and research assistant.



Data Analysis


As discussed in Chapter I, two correlational analyses

were conducted in Phase I of this study to determine de-

sirable items for the NP. Phi coefficients, which are

designed to correlate two dichotomous variables, were com-

puted. The two dichotomous variables correlated were the

student's item response (correct or incorrect) and student

classification (gifted or not-gifted) on the WISC-R. The

criterion for deciding if an item was to be included was

significance at the .05 level. For the present sample,

this meant that any item with a phi coefficient greater

than .182, was selected for the NP screening test.

Phi is based on the proportions of cases passing and

failing an item in both the gifted and not-gifted crite-

rion groups. The phi coefficient is known to be biased











toward middle difficulty levels of test items. As previ-

ously discussed, the research (CTBS and TCS) items were

designed primarily to assess performance of medium diffi-

culty.

Test items were also analyzed according to the index

of discrimination (Ebel, 1965). The difference between

the percentage of gifted students and not-gifted students

passing each item provides an index of item validity that

can be interpreted independently of the size of the par-

ticular sample in which it was obtained (Anastasi, 1976).

The index of discrimination (D) has been shown to measure

item validity with equivalent accuracy to other more elab-

orate measures (Engelhart, 1965). Similar to the phi, D

values are biased in favor of items with intermediate dif-

ficulty levels. A coefficient of .20 or greater was used

as the criterion for selecting an item for the NP. Thus

in Phase I, two forms of the NP screening test were

created; one form, here designated as NP-phi, was based on

items selected using phi coefficients; the other form, NP-

D, was based on items selected using the index of discrim-

ination.

For Phase II, analyses using the coefficient Kappa

were conducted to test (a) whether the NP or SIT was more

accurate in classifying gifted seventh graders, and (b) at

what cutoff points either the NP or SIT was more accurate.

The Kappa analysis measured the proportion of correct (and












incorrect) classifications for NP-phi, NP-D, and the SIT

while adjusting for the percentage of correct classifica-

tions that could be expected on the basis of chance alone.

Kappa is a descriptive statistic and not a test of statis-

tical significance.

Using the Kappa statistic proportions of predictions

were compared at various cutoff points. Kappa adjusts for

predictions expected by chance alone by taking into

account both observed and expected proportion classifica-

tions (Cohen, 1960). For example, when considering K val-

ues for the NP, the numerator of K is regarded as the pro-

portion of students consistently classified by both the NP

and the WISC-R (observed) over and above the product of

the proportions of students classified individually by

each test (expected). The denominator of K is the maximum

possible increase in decision-consistency above chance

level, given the proportions classified by the two tests

independently.

The formula used to compute Kappa was


K = PC
1 PC


where P = proportion of consistent gifted and not-gifted

classifications for WISC-R and NP or WISC-R and

SIT













and Pc = proportion of gifted classifications for WISC-R

x proportion of gifted classic ications for NP

(or SIT) + proportion of not-gifted classifica-

tions for WISC-R x proportion of not-gifted

classifications for NP or SIT.



Methodological Limitations


Possibly the most severe methodological limitation of

this study concerns the appropriateness of the CTBS and

TCS for discriminating students who score in a restricted

range near the ceiling of the test. The CTBS/TCS tests

were deemed as appropriate for this research because the

data were readily available and, if usable, would preclude

students from taking a gifted screening test. Also, the

CTBS and TCS measure the wide range of skills and abil-

ities. However, because the CTBS and TCS were designed to

measure traits of the general population of students, they

were less sensitive to group differences in the extreme

ranges of ability and particularly at the ceiling level.

The location parameter indicated the ability level, in

scale score units, at which the item was most sensitive to

individual differences. Thus, test items were designed to

have their greatest sensitivity to individual differences

in the general range where most students taking the test











would score. Since on most of the items, students in the

research sample were expected to score well above the

location parameter, most of the items provided little dif-

ferentiation among these students with much higher scale

scores. However, in defense of these instruments for use

in this study, some items with extreme location parameters

were purposely included in the CTBS by its constructors.

These items were well above or below the range of perfor-

mance for which the test was designed. Item characteris-

tic curves indicate the existence of CTBS items that were

passed typically only by students with very high total

scores on the test (CTB/McGraw-Hill, 1984). IRT item lo-

cation parameters for the TCS had not been calculated,

though since the TCS was designed to predict achievement

for the general population, it might be assumed that loca-

tion parameter criteria would adhere to a similar ration-

ale.

A second limitation of this study concerns measure-

ment error due to the variable adherence to test standard-

ization by administrators of the CTBS, TCS and the SIT.

Because teachers and counselors receive varying degrees of

formal training on the importance of precise adherence to

test standards and on the influence of standardization de-

viations on test reliability, administrators' strict con-

formity to test standards was questionable. Because very

little training was provided for the CTBS/TCS and SIT,












misinterpretation of instructions or standards was pos-

sible as well.

A related source of potential measurement error

existed due to the group format by which the CTBS/TCS is

administered. As opposed to the WISC-R and SIT, adminis-

trators were restricted in their ability to closely mon-

itor individual students and control for such factors as

misinterpreted directions or acute physical or emotional

liabilities of students. Since regular CTBS/TCS testing

was conducted only once a year in the research school dis-

trict, efforts were made to test as many students as pos-

sible during that time. However, SIT and WISC-R testing

were more easily postponed to a later date if a situation

warranted such action.

Another limitation of the CTBS specific to this study

was that some students were not administered the Science

and Social Studies subtests in the school district, be-

cause they were optional and were administered only in

some schools. These subtests are comprised of 40 items

each and may contain items useful for discrimination of

gifted and not-gifted students. If deleted from analysis,

these items may detract from the overall accuracy of the

proposed new procedure in accomplishing its intended goal.

The administration of the WISC-R and SIT over a

3-year period suggests a question regarding score











equivalence. One might argue that since the items admin-

istered to a 9-year-old and an 11-year-old are different

and represent different test difficulty levels, the two

students are, in fact, being tested on different scales.

This argument implies that gifted intelligence for the 9-

year-old is not equivalent to gifted intelligence for the

11-year-old. -To the contrary, while imperfections in test

stability will cause some fluctuation in IQ over time, the

adoption of the deviation IQ (Wechsler, 1974) permits com-

parison of scores over age levels.
















CHAPTER IV
RESULTS



This chapter is presented in a format that sequen-

tially reflects the methodological progression of the

study. Phase I will be discussed first to present results

of two analyses used to select items for the new screening

procedure (NP). Next, Phase II cross validation results

are presented in terms of prediction accuracy of the NP-

phi, NP-D, and SIT in relation to cutoff scores.



Phase I--Item Selection


Phi Coefficients

The phi coefficient analysis of the 460 CTBS and TCS

items yielded 56 items that discriminated the 59 gifted

and 59 not-gifted seventh graders. This number of items

is over twice as many as would be expected by chance at

the .05 level of confidence. There were 13 items signif-

icant at the <.01 level and three significant at .001

(Table 4-1). As also shown in Table 4-1, all CTBS and TCS

subtests contributed items except for CTBS Reference

Skills and TCS Sequences.

Not all of the 56 significant phi analyses items were

retained for use in Phase II. In five cases the phi items











Table 4-1. Item Phi Values and Significance
Levels


Item No. Subtest Phi-Value Significance


16 CTBS-Voc. .2059 $.05
39 Voc. .3258 <.001
40 Voc. .1834 $.05
41 Voc. .2069 .05
49 Read. .1963 <.05
53 Read. .2172 <.05
61 Read. .2266 .05
65 Read. .1977 (.05
77 Read. .1963 .05
80 Read. .2502 <.01
83 Read. .2386 .01
85 Read. .2502 .01
109 Spell. .2645 (.01
113 Spell. .2285 1.01
120 Spell. .2199 $.01
122 Lang. .2652 $.01
127 Lang. .1842 $.05
129 Lang. .2069 .05
136 Lang. .2187 $.05
137 Lang. .2377 $.01
139 Lang. .2559 $.01
154 Lang. .2035 (.05
159 Lang. .1842 $.05
171 Lang. .1913 $.05
176 Lang. .2934 $.01
177 Lang. .2035 <.05
180 Lang. .2018 <.05
182 Lang. .1842 $.05
187 Lang. .1913 <.05
193 Lang. .1834 $.05
205 Math .2187 <.05
215 Math .3007 <.01
*216 Math .1905 $.05
220 Math .3245 $.001
230 Math .3564 $.001
236 Math .2652 $.01
264 Math .2161 $.05
266 Math .2331 (.05
270 Math .2146 $.05
271 Math .2806 <.01
273 Math .2784 $.01
276 Math .1885 $.05
277 Math .2188 <.05
(2)311 Sci. .3214 <.05












Continued.


Item No. Subtest Phi-Value Significance


*(2)343 Soc. St. .2812 $.05
(2)364 Soc. St. .3076 .05
(2)366 Soc. St. .3921 $.01
(2)378 Soc. St. .3076 $.05
407 TCS-Anal. .1977 .05
*421 Mem. .2148 (.05
425 Mem. .1916 (.05
429 Mem. .2035 $.05
434 Mem. .1835 <.05
*438 Mem. .2068 $.05
*440 Verb. Reas. .1858 $.05
457 Verb. Reas. .2168 $.05


* Item discriminates in favor of not-gifted.
2 Answered by less than 90% of sample.


Table 4-1.












discriminated in favor of the not-gifted students. That

is, not-gifted students responded correctly to the items

more frequently than the gifted students. Also deleted

from Phase II analyses were five items responded to by

only about half of the Phase I sample. These five items

were located in the Science and Social Studies subtests,

which were administered on an optional basis at the dis-

cretion of the various schools involved, They were de-

leted because the cutoff scores were set based on total

items administered to all students. Those items not ad-

ministered to all students were deleted so that all stu-

dents could potentially attain the maximum raw score.

Therefore, 47 phi items were retained for use in Phase II,

42 from the CTBS and 5 from the TCS.

The mean of the total scores for the 47 items se-

lected by the phi analyses was 39.37. The median and mode

were somewhat higher, 41.00 and 46.00, respectively. The

standard deviation was 7.06 and the range was 41. Most

students performed well on these items, with most respond-

ing correctly to nearly all of them. This homogeneity of

scores is reflected by the negative skew of the distribu-

tion. Even though scores appeared to concentrate near the

upper end of the distribution, the large range of scores

(i.e., 4 through 46) contributed to a standard deviation

of adequate size.












Index of Discrimination

The index of discrimination analyses, which measured

the difference between the percentage of gifted and not-

gifted students passing each item, yielded a subset of 24

items with D values of .20 or greater. As recommended by

Engelhart (1965), items possessing a D value of at least

20 are considered to show adequate discrimination. All

24 of the acceptable items were obtained from the CTBS.

In Table 4-2 the upper and lower values represent

percentage passing each item for the gifted and not-gifted

students, respectively. D values ranged from .221 through

.393. As might be expected, the highest percentage of

correct responding occurred with the beginning TCS subtest

items. This was because TCS items are ordered hierarch-

ically by subtest according to difficulty level.

In this analysis, also, an item was deleted if it

discriminated in favor of the not-gifted group or if it

was not administered to over 10% of the sample. In con-

trast to the phi analyses, a large proportion of items

(i.e., 10 or 42%) were deleted. Fourteen items (Table 4-

2) were retained from the CTBS.

In contrast with the phi data, results of scores

based on selected items from the D analyses revealed a

relatively normal distribution, with the mean of 10.02,

median of 10.02, and mode of 11.00 falling within a range

of one test item. The standard deviation was 2.62 and the













Table 4-2. Items with D Values
ponding Subtests


>.20 and Corres-


Item No. Subtest Upper x Lower x D


39
80
85
109
158
176
185
215
230
271
273
277
(2)281
(2)311
317
(2)330
(2)341
(2)343
351
(2)360
(2)364
(2)366
(2)377
(2)378


CTBS.
Read.
Read.
Spell
Lang.
Lang.
Lang.
Math
Math
Math
Math
Math
Math
Sci.
Sci.
Sci.
Soc.
Soc.
Soc.
Soc.
Soc.
Soc.
Soc.
Soc.


Voc.


St.
St.
St.
St.
St.
St.
St.
St.


.930
.912
.912
.931
.958
.897
.879
.842
.873
.948
.947
.737
.750
1.000
.937
.937
.875
.688
.937
.937
1.000
.938
.937
1.000


.667
.712
.712
.731
.867
.650
.667
.567
.552
.746
.746
.525
.538
.735
.727
.727
.667
.909
.727
.467
.435
.506
.727
.758


.263
.200
.200
.200
.208
.327
.212
.275
.323
.202
.201
.212
.212
.265
.210
.210
.208
-.221*
S.210
.240
.242
.393
.210
.242


*Item discriminates in favor of not-gifted.
2 answered correct by less than 90% of the sample.













range was 14. Unlike the phi distribution, the ceiling of

the D distribution was probably sufficiently high because

correct responding diminished beyond raw scores of 11.

Only 11 of 117 students received a raw score above 12,

while 65 received scores within one point of the mean.

Taking into account the small number of NP-D items (14),

the standard deviation of 2.6 is considered to be ade-

quately large for calculating cutoffs. The range of

scores covered both extremes of the distribution, and the

skew is not as great as might be expected given the gen-

erally restricted range of the sample at the upper ability

levels.

Consistent with previous research (e.g., Engelhart,

1965), there were commonalities between the phi and D an-

alyses results for this sample. Sixteen items, or 67% of

the D and 9% of the phi items, were selected from both

analyses. All common items were from the CTBS (on

achievement test) because the D analyses yielded no TCS

(cognitive ability) NP items. Similarly, the one D analy-

sis item found to discriminate in favor of the not-gifted

students also did so on the phi analysis.


Cutoff Scores

Cutoff scores for the phi, D, and SIT analyses were

computed in order to demarcate optimal cutting points for











differentiating the gifted from not-gifted students. Cut-

off scores were established such that a maximum number of

gifted and a minimum number of not-gifted students would

fall above the cutoff. Therefore, the true positive and

true negative findings were maximized while the false pos-

itives and false negatives were minimized. This procedure

required a "value judgment" by the rater as to the amount

of error which would be acceptable. As the cutoff is low-

ered to permit more gifted students to exceed it, increas-

ing numbers of not-gifted students would also exceed it,

thus increasing the possibility of false negatives. To

assist with this problem, multiple cutoff scores were cal-

culated so that the most desirable cutoffs could be deter-

mined. Cutoff scores were chosen according to fractions

of standard deviation units for the total score distribu-

tions on the SIT, the new test created by selecting items

with phi, and the new test created by selecting items with

D. The mean SIT IQ for the Phase I sample was 137.30 and

the standard deviation was 14.12. This mean IQ was ap-

proximately 37 points higher than that of the general pop-

ulation. It reflects the above average general intelli-

gence of the research sample. In Phase II, these standard

deviation cutoff scores calculated as standard deviations

from the Phase I sample, were applied to the cross valida-

tion sample to determine the relative accuracy of gifted

classifications using the NP-phi, NP-D, and SIT.














Phase II--Cross Validation


Application of Cutoff Scores

The first step in testing the NP items obtained in

Phase I was to calculate the numbers of NP items scored

correctly for.the Phase II sample and then to apply appro-

priate cutoff points to those scores. The Phase II sub-

sample consisted of 61 students. Proportions of correct

classifications among the NP-phi, NP-D and SIT items were

then analyzed to determine the test and cutoff that most

accurately differentiated the gifted and not-gifted stu-

dents.

Tables 4-3, 4-4, and 4-5 show the numbers of examin-

ees classified into prediction categories at various cut-

off points where the examine's true status was defined as

classification of gifted or not-gifted according to the

WISC-R. For all three classification procedures (NP-phi,

NP-D, and SIT) little variability existed between predic-

tion rates at some of the cutoffs. For example, classif-

ications at the +1.0 and -1.0 differed only slightly from

those at the +.66 and -.66 cutoffs respectively. A sim-

ilar result occurred for the +.50 cutoff compared to the

+.30 cutoff and for the -.50 cutoff compared to the -.30

cutoff.















Table 4-3.


Phase II: Number of Examinees in Each
Prediction Category by Cutoff/Phi


True True True False False False
Positive Negative Total Negative Positive Total


Cutoff

-1.00 SD 30 5 35 2 24 26

- .66 SD 30 5 35 2 24 26

- .50 SD 30 5 35 2 24 26

- .33 SD 30 8 38 2 21 23

0.00 SD 29 12 41 3 17 20

+ .33 SD 23 17 40 9 12 21

+ .50 SD 18 19 37 11 10 24

+ .66 SD 8 24 32 24 5 29

+1.00 SD 3 29 32 29 0 29


Total gifted = 32
Total not-gifted
















Table 4-4. Phase II: Number of Examinees in Each
Prediction Category by Cutoff/Index of
Discrimination


True True True False False False
Positive Negative Total Negative Positive Total


Cutoff

-1.00 SD 31 5 36 1 24 25

- .66 SD 31 9 40 1 20 21

- .50 SD 31 9 40 1 20 21

- .33 SD 31 10 41 1 19 20

0.00 SD 26 15 41 6 14 20

+ .33 SD 26 15 41 6 14 20

+ .50 SD 22 19 41 10 10 20

+ .66 SD 22 19 41 10 10 20

+1.00 SD 15 24 39 17 5 22















Table 4-5.


Phase II: Number of Examinees in Each
Prediction Category Cutoff/SIT


Cut- True True True False False False
off Positive Negative Total Negative Positive Total



-1.0 SD 32 6 38 1 22 23

- .66 SD 32 7 39 1 21 22

- .50 SD 31 7 38 2 21 23

- .33 SD 28 9 37 5 19 24

.00 SD 22 18 40 11 10 21

+ .33 SD 14 22 36 19 6 25

+ .50 SD 12 24 36 20 5 25

+ .66 SD 9 26 35 23 2 25

+1.0 SD 5 29 34 27 0 27












Table 4-6 was constructed to consolidate and clarify

the data on predictive accuracy, in that table, propor-

tions of correct and incorrect predictions are again

expressed as true positives (TP), true negatives (TN),

false positives (FP), and false negatives (FN). Deleted

were the cutoffs -1.0 SD, -.50 SD, +.50 SD, and -1.0 SD.

Inspection of Table 4.6 reveals that all three methods of

screening gifted students were approximately equivalent.

That is, all three found the greatest proportions of TP

predictions in .492 through .525 of the sample. Lowest TN

predictions occurred in .082 through .148 of the sample.

After the prediction rates were established for the

three tests and the prediction values were expressed as

proportions of the sample, analyses were conducted to de-

termine the proportions, if any, that were most accurate.

Coefficient Kappa analyses were calculated to assess the

relative gifted prediction accuracy of the NP-phi, NP-D,

and SIT.


Kappa Comparisons

In results shown in Table 4-7, P values correspond to

percentages of correct classifications (both true pos-

itives and true negatives). However, two equal classifi-

cation percentages (e.g., as observed for phi and D at 0.0

cutoff) may not have equal corresponding K-values because

the Kappa equation adjusts for chance correct predictions.















Table 4-6. Phase II: Proportions of Correct and
Incorrect Predictions for NP (phi),
NP (D), and SIT at Two Score Cutoffs


z- True True True False False False
Cutoff Positive Negative Total Negative Positive Total


Phi

- .66 SD .492 .082 .574 .033 .393 .426
- .33 SD .492 .131 .623 .033 .344 .377
.00 SD .475 .197 .672 .049 .229 .328
+ .33 SD .377 .279 .656 .164 .180 .344
+ .66 SD .131 .393 .525 .393 .082 .475

I)

- .66 SD .508 .148 .656 .016 .328 .344
- .33 SD .508 .164 .672 .016 .311 .328
.00 SD .426 .246 .672 .098 .230 .328
+ .33 SD .426 .246 .672 .098 .230 .328
+ .66 SD .361 .311 .672 .164 .164 .328

SIT

- .66 SD .525 .115 .639 .016 .344 .361
- .33 SD .459 .148 .607 .082 .311 .393
.00 SD .361 .295 .656 .180 .164 .344
+ .33 SD .270 .360 .590 .311 .082 .410
+ .66 SD .164 .426 .590 .361 .049 .410


values rounded to nearest


Note: All


.10 percentile.














Table 4-7. Percentage of Correct Classifications
and Kappa Values for NP (phi and D)
and SIT at Various Cutoff Scores



NP NP

Cutoff Phi-P* D-P* SIT-P* Phi-ki D-k1 SIT-k1


- .66 SD 57.4 65.6 63.9 .114 .289 .234

- .33 SD 62.3 67.2 60.7 .220 .324 .178

.00 SD 67.2 67,2 65.6 .328 .334 .309

+ .33 SD 65.6 67.2 59.0 .306 .320 .258

+ .66 SD 52.5 67,2 59.0 .076 .342 .204


*P = percentage
k = Kappa value












An example of the Kappa calculation of percentages of

correct classifications for NP-phi at the -.66 cutoff is

illustrated using four-fold tables (Figure 4-1). Values

in the equation, along with percentages of examinee clas-

sifications at other cutoffs, are located in Table 4-6.


P PC
K =
1 P


P = .492 + .082 = .574

PC = (.885)(.525) + (.475)(.115)

= .4646 + .0546

= .5192

Results of the Kappa analyses revealed a consistent

pattern of greater values for the NP-D than for the NP-phi

or the SIT at all cutoffs, suggesting the NP-D may be the

more accurate predictor of correct classification. The

greatest K value for NP-D occurs at the +.66 where the K

value of .342 indicates that there is a 34% improvement in

prediction accuracy over that expected by chance. The

greatest discrepancy between K values for NP-D and the SIT

appears at the -.33 cutoff where K = .324 for NP-D but

only .178 for the SIT.








78






IQ


Class
Gifted


Class
Gifted


phi



Class
Not-Gifted


.525


Class
Not-Gifted


.475


.574 .519
K =
1 .519


.114


Figure 4-1.


Computation Example of Kappa
analysis for True Phi Classi-
fications at -.66 SD Cutoff


.885


TP FP



.492 .393


FN TN




.033 .082


.055
.481











Summary of Results


In Phase I of this study 56 items from the CTBS/TCS

were identified using phi coefficients that significantly

differentiated gifted from not-gifted students. The index

of discrimination analyses yielded 24 items that success-

fully discriminated gifted from not-gifted. A large pro-

portion of significant D items were common to the phi

analysis results.

In Phase II, total scores for each subset of items

were computed using a cross validation sample of exam-

inees. The accuracy of classifications resulting from

each of the two new tests and the SIT were contrasted at

five different cutoff scores. Kappa analyses revealed

higher correct prediction values for the NP-D than for

either the NP-phi and the SIT at all cutoff levels.
















CHAPTER V
DISCUSSION



Research Questions


In order to answer the first research question, this

discussion focuses on issues related to NP item validity

from Phase I. Phase II NP items were selected from the

CTBS and TCS using correlational analyses. To answer the

remaining two research questions, the predictive accura-

cies of the NP-phi, NP-D, and SIT are discussed in terms

of cutoff score~s.

The following research questions were addressed in

this study.

1. Can an accurate predictor of gifted IQ classifi-

cation on the WISC-R/S-B be derived from an in-

strument composed of items on the CTBS and TCS

in a situation in which giftedness is viewed as

a dichotomous trait variable?

2. Is the NP more accurate than the SIT in classi-

fying gifted and not-gifted seventh graders?

3. At what cutoff points) is the NP more accurate

than the SIT in classifying gifted and not-

gifted seventh graders?













Phase I

In Phase I of the study it was suggested that some

items derived from the CTBS/TCS were valid for distin-

guishing between gifted and not-gifted seventh graders.

Evidence for this assertion included (a) the relatively

large number.of NP items that are common to both the phi

and D analyses, (b) the large number of items found to

discriminate between the two groups, and (c) the agreement

between CTBS and IRT location parameter estimates and NP

item statistics.

Common Items

The index of discrimination (D) analyses yielded

fewer discriminating items than did the phi analyses.

However, the validity of the D analyses is supported by

the overlap of items between the phi and D analyses. Of

the 14 items retained for the NP-D, 11 items were also in-

cluded in the phi NP scale. Overlap between phi and D

item discrimination for middle difficulty items was re-

ported by Engelhart (1965).

Support for the validity of the NP also was suggested

by the relatively large number of significant items, par-

ticularly on the phi analyses. In addition to the 56 sig-

nificant phi items initially computed at .05, another 29

were significant at the .10 level. Clearly an even larger

number of significant items would have resulted had the











sample size of 117 remained constant. However, the sample

size diminished below 75 on 99 items. Of these items, 20

had phi values that would have been significant at <.05

had the N been as high as 113.

IRT Parameters

A strong case for the NP item validity is made upon

investigation of item response theory item location param-

eters for the CTBS items. An investigation of the

response patterns of subjects on both item sets, in com-

parison with the response pattern on these items by the

original CTBS sample, suggest commonality in responding.

IRT'_item _location_paramneter. Item response theory

was utilized by the CTUS/TCS constructors for item selec-

tion. Item characteristic curves were plotted using a

three parameter logistic model involving (a) item dis-

crimination, (b) item location (or difficulty), and (c) a

"guessing" factor. A location parameter describes an

item's difficulty in terms of the student's ability level

(or latent trait). An item discriminates best for a stu-

dent whose ability level is near the item's location

parameter (or difficulty level). An item with a high

location parameter serves its function only for students

of high ability since low ability students would not be

expected to answer these items correctly.

Comparing response patterns. An assumption of item

response theory is the invariance of item parameters












(Anastasi, 1982; Baker 1984). That is, given certain con-

ditions, item parameters should be uniform among different

populations because individual items are assumed to meas-

ure the same trait in different populations. In this

case, the validity of the NP items would be supported if

NP items were common to CTBS items with high location

parameters. In fact, 41 of the original 65 NP items had

CTBS location parameters above the mean for the norming

sample (CTBS Technical Report, 1983). The proportion of

NP-D items to high location parameter items was even

greater (16 of 24, 67%).

In summary, the data support an affirmative answer to

the first research question. There exists a subset of

items that discriminated gifted and not-gifted seventh

graders who were all academic high achievers. In Phase

II, cross validation with a smaller but otherwise equiv-

alent sample was conducted to support these findings.


Phase II

Cross validation of Phase I findings was conducted in

Phase II by assessing the accuracy of NP items as compared

to the SIT in classifying the gifted and not-gifted stu-

dents and the cutoff score at which each procedure showed

greatest accuracy. These goals were accomplished by an-

alyzing proportions of correct and incorrect classifica-

tions using coefficient Kappa analyses.












An examination of total true classifications (i.e.,

correct classifications of gifted and not-gifted students)

revealed greatest prediction accuracy at the 0.0 SD cut-

off. At this cutoff, NP-D and NP-phi values were equal

(67.2) and slightly superior to the SIT (65.6). When per-

centage values were corrected for chance occurrences by

the Kappa analyses, NP-D was slightly more accurate at the

+.66 SD than at 0.0 SD, and NP-D items were superior to

NP-phi and the SIT. Furthermore, at the +.66 cutoff, the

Kappa value for the NP-D is .342 contrasted to much lower

values for NP-phi (.076) and SIT (.204).

In answering the second research question, data con-

sistently supported the NP as more accurate than the SIT

in classifying gifted and not-gifted students. Among the

two NP tests, NP-D was generally superior to the NP-phi.

The NP-D was clearly the better measure of true positive

and negative classifications based on Kappa analyses.

There is some ambiguity regarding the overall most

accurate cutoff for classifying students, rendering the

last research question more difficult to answer. There is

empirical support for the NP-D at both the -.33 and +.66

cutoffs as most accurate for classifying gifted and not-

gifted seventh graders in the sample. Kappa analyses sup-

port that finding in that the discrepancy between total











true classifications for the NP-D and the SIT was greatest

at the -.33 cutoff. Data supporting an NP-D cutoff at

+.66 may be less clear. Kappa analyses indicate that the

number of total true classifications at +.66 is superior

to the number at any other cutoff. The relative strength

of the NP-D in total true classifications was due to its

much greater accuracy in predicting true positives. The

SIT was more accurate in classifying true negatives. In

choosing a cutoff score for gifted IQ screening purposes,

specific classification priorities and goals must be taken

into account. For example, if the primary goal is to max-

imize total true classifications, the +.66 cutoff may be

desirable.



Conclusions, Implications, and Limitations


This study yields conclusions, implications, and lim-

itations germane to both pragmatic and theoretical issues.

In the following sections, the issues of sampling, gen-

eralizability, item validity, and cutoff scores are

discussed individually.


Sampling

The success of the NP in classifying two groups of

students, who in some respects appeared indistinguishable,

has some inherent implications. The gifted and not-gifted

children were similar in terms of achievement, IQ (in many












cases), and teacher perceptions of them as gifted. These

similarities suggest that the analysis results are robust

because the two groups were accurately differentiated in

spite of their likenesses.

The specific trait of the research sample as being

high academic achievers distinguishes this study from many

others and this distinction is essential to the utility of

the NP. In most of the related literature reviewed in

Chapter II, research samples were not as restricted in

range of IQ, rendering correct classification of gifted

and not-gifted more likely in those. In other studies,

many samples included students randomly selected from

general populations. Those students, excluded in this

study by preselection, were readily screened as not-gifted

in other studies, allowing for artificially inflated accu-

racy predictions.

A pertinent limitation to this study discussed in

Chapter III was the sampling procedure. Specifically,

some students were deleted from formal IQ testing, and

therefore, excluded from the sample because they scored

below the SIT screening cutoff of 132. However, one-sixth

(i.e., 11 of 61) Phase II students had SIT scores below

132, with some scores falling in the average and below

average range. This occurrence raises a question about

the criterion used for disqualifying students from IQ











testing based on their SIT scores. It is unclear how stu-

dents with SIT scores in the 80s were not disqualified

while others were. It seems that screening procedures

were not followed consistently at the school level. For-

tunately, even though there were relatively few students

with SIT IQs below 132, there were enough for data analy-

ses, and the range of scores was wide. In this case,

sampling error probably was not a confounding influence on

results because disqualification of the aforementioned

students seems to have occurred in a random, unsystematic

manner.


Generalizability

In large part, outcomes regarding screening accuracy,

item validity, and cutoff scores may not be generalizable

to other populations at this time. However, the success

of the research procedure (NP) is considered meaningful in

as much as the accuracy of the NP will be tested contin-

ually in its practical application on potentially gifted

students. In this way, the generalizability of screening

accuracy, item validity, and cutoff scores will be val-

idated on other populations.

In Chapter I the definition of intellectual gifted-

ness for this study was briefly discussed. This pragmatic

definition focuses only on IQ test scores that fall above











a particular cutoff score. Other, more theoretical def-

initions, may involve other qualitative differences.

Results of this study may not be generalizable to students

who are designated gifted using criteria other than their

IQs falling above the 96th percentile.


Screening Accuracy

Generalizability of research results would be en-

hanced by analyzing group test data for other restricted

populations to determine if subgroups may be successfully

discriminated using test items. Two preliminary steps in

such a process would be to (a) analyze test results of

other populations of potentially gifted students, such as

elementary or high school, on the CTBS/TCS, and (b) deter-

mine if other group achievement/cognitive ability tests

possess items that accurately classify potentially gifted

students. Further, researchers may seek to generalize

results on more divergent restricted populations. For

example, it may be useful to screen for mild mental retar-

dation among remedial students or for learning disabil-

ities among children with discrepant report card grades.

Item Validity

A somewhat unexpected result of this research was

that the CTBS items (which supposedly measure school

achievement) contributed much more to the NP than did the

TCS items (which are purported to measure intelligence).











Indeed, none of the NP-D items were obtained from the TCS.

The NP-phi test, which initially included eight TCS items,

was found to be generally less accurate in classifying

students than the NP-D. Essentially, the achievement

items had greater criterion related validity than did the

cognitive ability items for Full Scale IQ.

Ostensibly, the superior criterion validity of the

CTBS items is contrary to expectation, however a closer

examination suggests that this pattern may support rather

than refute the generalizability of research findings.

Examination of CTBS, TCS, and WISC-R subtests reveals that

the CTBS probably has more in common with the WISC-R than

does the TCS. For example, CTBS subtests such as Mathe-

matics, Science, Social Studies, and Vocabulary have What

appear to be direct correlates on the WISC-R. The rela-

tionship between TCS and WISC-R subtests appears more ob-

scure. This assertion is supported by Wurster (1985) who

compared overlap between the TCS and WISC-R with the SIT

and the WISC-R. She found that 87.8% of the SIT items

measured the same skills as the WISC-R Verbal Scale, how-

ever, "no items from the TCS appeared to measure any of

the skills that are measured by the 11 WISC-R subtests"

(p. 24).

The relationship of commonality between IQ test and

achievement test performance for this sample seems related











to two factors. First, it is likely that for high achiev-

ing students, "superior" intelligence is heavily loaded in

the verbal reasoning domains that are tapped to a greater

extent by the CTBS than by the TCS. The TCS, to a greater

degree than the CTBS, measures non-verbal skills or abil-

ities that may be less represented by high achieving stu-

dents' strengths' on the WISC-R or Stanford-Binet. Future

researchers might investigate this hypothesis by examining

the relationship between achievement and verbal vs. non-

verbal intelligence (as represented by these or similar

tests) for high achieving or gifted students.

A second explanation for the disproportionate repre-

sentation of achievement test items on the NP concerns the

theoretical rationale for this research. That is, the

content of achievement tests (e.g., CTBS) and IQ tests

(e.g., WISC-R) are often operationally indistinguishable.

This premise has been supported empirically by Anastasi

(1982), among many others (see Chapter II).

Cutoffs

Conclusions regarding the most desirable cutoff for

the NP (NP-D at -.33 or +.66) were discussed earlier in

the chapter. There are some other conclusions that may be

drawn regarding cutoff scores for the SIT with this popu-

lation. The mean SIT score for the Phase II sample was

137. This score is somewhat above what would be expected

yet consistent with research findings by Karnes and Brown












(1979) and Rust and Lose (1980) who found that the SIT

tends to overestimate high IQs on full length tests. In

their studies, those researchers recommended setting SIT

cutoff scores for gifted classification higher than two

standard deviations above the mean to offset this tendency

to overestimate the IQs of brighter students. The current

findings support those recommendations because the SIT was

most accurate in classifying true positive and false neg-

ative findings at over three standard deviations above the

mean (SIT IQ = 149).



Summation

In his later years Edwin R. Guthrie (1959) suggested

that research has no inherent value, but rather that its

value was gained from its practical applications. It is

in the spirit of that philosophy that this research may be

fully appreciated. Relatively accurate in its predictive

power and efficient in its method, the NP analysis will be

easily replicated on new student samples and test formats.

Thus, the NP may be best viewed as a procedure adaptable

to varying, yet specific, needs.















REFERENCES


Ahmann, J. S. (1972). The comprehensive test of basic
skills (a review). In O. K. Buros (Ed.), The seventh
mental measurement yearbook (pp. 614-615). Highland
Park, NJ: The Gryphon Press.

American Psychological Association (1985). Standards for
educational and psychological tests. Washington,
DC: American Psychological Association.

Anastasi, A. (1976). Psychological testing. New York:
MacMillan publishing.

Apple, D. (1983, November). Screening gifted children: A
comparison of the SIT and WISC-R. Paper presented at
the annual meeting of the Alabama Association of
School Psychologists, Guntersville, AL.

Barklay, J. E., Phillips, G., & Jones, T. (1983). Devel-
oping a predictive index for giftedness. Measurement
and Evaluation in Guidance, 16, 25-35.

Barrington, B. L. (1979). In the name of education. In N.
Colangelo and R. T. Zarram (Eds.), New voices in
counseling the gifted (pp. 65-70). Dubuque, Iowa:
Kendall/Hunt.

Bersoff, D. N. (1971). Short forms of individual intelli-
gence tests for children: Review and critique. Jour-
nal of School Psychology, 9, 310-320.

Birch, J. W. (1955). The utility of short forms of the
Stanford Binet tests of intelligence with mentally
retarded children. American Journal of Mental Defi-
ciency, 59, 462-484.

Birch, J. W. (1984). Is any identification procedure
necessary? Gifted Child Quarterly, 28, 157-161.

Bloom, B. S. (1956). Taxonomy of educational objectives
handbook I: The cognitive domain. New York: David
McKay Co.











Blosser, G. H. (1963). Group intelligence tests as
screening devices in locating gifted and superior
students in ninth grade. Exceptional Children, 29,
282-286.

Braden, J. P. (1985). A modest proposal: Using probabil-
ities of special-education eligibility instead of
cutting scores. Unpublished manuscript.

Brock, H. (1982). Factor structure of intellectual and
achievement measures for learning disabled children.
Psychology in the Schools, 19, 297-304.

Brooks, C. R. (1977). WISC, WISC-R, S-B L-M, WRAT: Rela-
tionships and trends among children ages six to ten
referred for psychological evaluation. Psychology in
the Schools, 14, 30-33.

Bryan, J. R., & Bryan, T. H. (1975). Understanding learn-
ing disabilities. Sherman Oaks, CA: Alfred Publish-
ing Co.

Burket, G. R. (1974). Empirical criteria for distinguish-
ing and validating aptitude and achievement
measures. In D. A. Green (Ed.), The aptitude-
achievement distinction (pp. 35-51). Monterey, CA:
CTB/McGraw-Hill.

Carleton, F. O., & Stacey, C. L. (1954). Evaluation of
selected short forms of the Wechsler Intelligence
Scale for Children. Journal of Clinical Psychology,
10, 258-261.

Carroll, J. B. (1966). Factors of verbal achievement. In
A. Anastasi (Ed.), Testing problems in perspective
(pp. 406-413). Washington, D.C. American Council on
Education.

Chambers, J. A. (1960). Preliminary screening methods in
the identification of intellectually superior chil-
dren. Exceptional Children, 26, 145-150.

Chambers, J. A., Barron, F., & Sprecher, J. W. (1980).
Identifying gifted Mexican-American students. Gifted
Child Quarterly, 24, 123-128.

Clarizio, H. F., & Meherens, W. A. (1985). Psychometric
limitations of Guilford's structure-of-intellect
model for identification and programming of the
gifted. Gifted Child Quarterly, 29, 113-119.










Cohen, J. A. (1960). A coefficient of agreement for nom-
inal scales. Educational and Psychological Measure-
ment, 20, 37-46.

Crofoot, M. J., & Bennett, T. S. (1980). A comparison of
three screening tests and the WISC-R in special
education evaluations. Psychology in the Schools,
17, 474-478.

CTB/McGraw-Hill. (1984). The Comprehensive Test of Basic
Skills: Technical report. Monterey, CA: CTB/McGraw-
Hill.

CTB/McGraw-Hill. (1984). The Test of Cognitive Skills:
Technical report. Monterey, CA: CTB/McGraw-Hill.

Dean, R. S. (1977). Canonical analysis of a jangle fal-
lacy. Multivariate Experimental Clinical Research,
3, 17-20.

Dean, R. S. (1982). Intelligence-achievement discrep-
ancies in diagnosing pediatric learning disabil-
ities. Clinical Neuropsychology, 4, 58-62.

DeFilippis, N. A., & Fulmar, K. (1980). Effects at age
and IQ level on the validity of one short intelli-
gence test used for screening purposes. Educational
and Psychological Measurement, 40, 543-545.

Dirkes, M. A. (1981). Only the gifted can do. Educational
Horizons, 59, 138-143.

Dirks, J., Wessels, K., Quaforth, J., & Quenon, B.
(1980). Can short form WISC-R tests identify chil-
dren with high Full Scale IQ? Psychology in the
Schools, 17, 40-46.

Dunn, L. M., & Markwardt, F. C. (1970). Peabody Individ-
ual Achievement Test. Circle Pines, MN: American
Guidance Service.

Ebel, R. L. (1965). Measuring educational achievement.
Englewood Cliffs, NJ: Prentice-Hall.

Elman, L., Blixt, S., & Sawicki, R. (1981). The develop-
ment of cutoff scores on a WISC-R in the multidimen-
sional assessment of gifted children. Psychology in
the Schools, 43, 426, 428.




Full Text
26
In the previously discussed study (see Chapter I) by
Grossman and Johnson (1983), the Otis Lennon group IQ test
was found to be a better predictor of gifted achievement
than the SIT for high achieving students. Dirks, Wessels,
Quaforth, and Quenon (1980) also found the SIT to be a
poor predictor of gifted ability. They administered the
SIT to 47 academically talented fourth graders. The stu
dents were also administered the WISC-R to determine their
actual IQ scores. Of the 11 students who were found to
possess gifted intelligence (IQ>130), only 8 were identi
fied as such by the SIT. In addition, the SIT falsely
predicted gifted intelligence in 9 of the 38 nongifted
children. The researchers concluded that the SIT alone
should not be used to predict IQs of gifted children. The
SIT has also been found to significantly overestimate IQ
scores. Machen (1972) investigated the reliability and
concurrent validity of the SIT with the WISC, using 5
gifted children ages 9 through 11. The results revealed a
significant correlation between the two tests, though the
SIT tended to overestimate the WISC by at least one stan
dard deviation. Additionally, the SIT has been shown to
underestimate IQ scores. Mize et al. (1979) found, in
their study of 207 students from all grade levels, that of
students with above average intelligence, 24% were overes
timated and 24% were underestimated by 11 or more IQ
points on the SIT.


IllMETHODOLOGY 49
Overview 49
Population and Sample 49
Assessment Procedures 51
Research Procedures 53
Data Analysis 55
Methodological Limitations 58
IV RESULTS 62
Phase I--Item Selection 62
Phi Coefficients 62
Index of Discrimination 66
Cutoff Scores 68
Phase II--Cross Validation 70
Application of Cutoff Scores .... 70
Kappa Comparisons 7 4
Summary of Results 7 8
Summary of Results 78
V DISCUSSION 80
Research Questions 80
Phase I 81
Common Items 81
IRT Parameters 82
Phase II 83
Conclusions, Implications, and
Limitations 85
Sampling 85
Generalizabi1ity 87
Screening Accuracy 88
Item Validity 88
Cutoffs 90
Summation 91
REFERENCES 9 2
BIOGRAPHICAL SKETCH 104
IV


100
Ryan, G. T. (1979). The influence of readability on crit
ical reading comprehension of secondary social
studies students. Dissertation Abstracts Interna
tional 39, 6684 (University Microfilms No. 79-
11,017)
Ryan, G. T. (1982). An analysis of motivational factors
of gifted and non-gifted learners. Paper presented
at the annual meeting of the American Educational
Research Association, New York.
Salvia, J., &Ysseldyke, J. E. (1978). Assessment in spe
cial and remedial education. Boston: Houghton Miff
lin Co.
Schena, R. A. (1963). A search for talented pupils.
Journal of Experimental Education, 32, 27-41.
Schnell, R. E. (1982). Is there a correlation between
intelligence and reading achievement for students in
grades three through eight who were referred to the
gifted program in Hendry County Florida. Unpublished
manuscript.
Schwarting, F. G., & Schwarting, K. R. (1977). The rela
tionship of the WISC-R and WRAT: A study based upon
a selected population. Psychology in the Schools,
14, 431-433.
Sheldon, W. D., & Manolakes, G. (1954). A comparison of
the Stanford-Binet revised form L, and the Califor
nia Test of Mental Maturity (S-Form). Journal of
Educational Psychology, 45 499-504 .
Silverstein, . B. (1970). Reappraisal of the validity of
a short form of Wechsler's Scales. Psychological
Reports, 26, 559-561.
Simpson, W. H., & Bridges, C. C. (1959). A short..form of
the Wechsler Intelligence Scale for Children. Jour
nal of Clinical Psychology, 15, 414.
Slosson, R. J. (1961). Slosson Intelligence Test for
Children and Adults. East Aurora, NY: Slosson Educa
tional Publications.
Slosson, R. J., & Jensen, J. A. (1982). Slosson Intelli
gence Test (SIT) norms tables application and devel
opment East Aurora, NY: Slosson Educational Publi
cations .


58
and Pc = proportion of gifted classifications for WISC-R
x proportion of gifted classifications for NP
(or SIT) + proportion of not-gifted classifica
tions for WISC-R x proportion of not-gifted
classifications for NP or SIT.
Methodological Limitations
Possibly the most severe methodological limitation of
this study concerns the appropriateness of the CTBS and
TCS for discriminating students who score in a restricted
range near the ceiling of the test. The CTBS/TCS tests
were deemed as appropriate for this research because the
data were readily available and, if usable, would preclude
students from taking a gifted screening test. Also, the
CTBS and TCS measure the wide range of skills and abil
ities. However, because the CTBS and TCS were designed to
measure traits of the general population of students, they
were less sensitive to group differences in the extreme
ranges of ability and particularly at the ceiling level.
The location parameter indicated the ability level, in
scale score units, at which the item was most sensitive to
individual differences. Thus, test items were designed to
have their greatest sensitivity to individual differences
in the general range where most students taking the test


70
Phase II--Cross Validation
Application of Cutoff Scores
The first step in testing the NP items obtained in
Phase I was to calculate the numbers of NP items scored
correctly for.the Phase II sample and then to apply appro
priate cutoff points to those scores. The Phase II sub
sample consisted of 61 students. Proportions of correct
classifications among the NP-phi, NP-D and SIT items were
then analyzed to determine the test and cutoff that most
accurately differentiated the gifted and not-gifted stu
dents .
Tables 4-3, 4-4, and 4-5 show the numbers of examin
ees classified into prediction categories at various cut
off points where the examinee's true status was defined as
classification of gifted or not-gifted according to the
WISC-R. For all three classification procedures (NP-phi,
NP-D, and SIT) little variability existed between predic
tion rates at some of the cutoffs. For example, classif
ications at the +1.0 and -1.0 differed only slightly from
those at the +.66 and -.66 cutoffs respectively. A sim
ilar result occurred for the +.50 cutoff compared to the
+.30 cutoff and for the -.50 cutoff compared to the -.30
cutoff .


REFERENCES
Ahmann, J. S. (1972). The comprehensive test of basic
skills (a review). In O. K. Buros (Ed.), The seventh
mental measurement yearbook (pp. 614-615). Highland
Bark, NJ: The Gryphon Press.
American Psychological Association (1985). Standards for
educational and psychological tests. Washington,
DC: American Psychological Association.
Anastasi, A. (1976). Psychological testing. New York:
MacMillan Publishing.
Apple, D. (1983, November). Screening gifted children: A
comparison of the SIT and WISC-R. Paper presented at
the annual meeting of the Alabama Association of
School Psychologists, Guntersvi1 le, AL.
Barklay, J. E., Phillips, G., & Jones, T. (1983). Devel
oping a predictive index for giftedness. Measurement
and Evaluation in Guidance, ^6 2 5-35.
Barrington, B. L. (1979). In the name of education. In N.
Colangelo and R. T. Zarram (Eds.), New voices in
counseling the gifted (pp. 65-70). Dubuque, Iowa:
Kendal 1/Hunt.
Bersoff, D. N. (1971). Short forms of individual intelli
gence tests for children: Review and critique. Jour
nal of School Psychology, 9, 310-320.
Birch, J. W. (1955). The utility of short forms of the
Stanford Binet tests of intelligence with mentally
retarded children. American Journal of Mental Defi
ciency 59, 462-484.
Birch, J. W. (1984). Is any identification procedure
necessary? Gifted Child Quarterly, 28, 157-161 .
Bloom, B. S. (1956). Taxonomy of educational objectives
handbook I: The cognitive domain. New York: David
McKay Co.
92


27
In 1979 Karnes and Brown further examined the ten
dency of the SIT to over- or underestimate IQ scores. In
this study the validity of the SIT in relation to the
WISC-R was assessed for a group of 79 gifted children ages
6 through 12. A SIT-WISC-R correlation of r=.48 was cal
culated; this coefficient was significant at the .001
level. The. authors also computed a regression equation
with which to predict WISC-R IQ from the SIT. Results
indicated that at the lower ranges of SIT scores, it
tended to underestimate the WISC-R, while at the upper
ranges IQ was overestimated. Despite the high correlation
between the two tests, it is apparent that Karnes and
Brown were not confident in the SIT's predictive ability
because they recommended using two SEMs for the SIT when
screening gifted IQ to ensure that most gifted students
are identified. To obtain 95% accuracy, a cutoff score of
105 would have been necessary. However, Karnes and Brown
did not indicate how many nongifted students would have
been predicted as gifted using a cutoff this low.
In a similar study, presented at the Annual Meeting
of the Alabama Association of School Psychologists in
1983, Apple discussed the precision of the SIT in pre
dicting WISC-R IQ of 61 gifted students ages 6 to 11.
Differences in scores were compared by use of independent
t-tests. The results supported the findings of Karnes and


55
obtained using students' SIT scores as a screening
procedure in predicting gifted IQ on the WISC-R.
Data collectors and recorders consisted of the re
searcher and employees of the district school board.
School board employees working in the testing and evalua
tion office obtained CTBS/TCS item responses from computer
data. WISC-R and SIT scores were obtained in a similar
manner by the researcher and research assistant.
Data Analysis
As discussed in Chapter I, two correlational analyses
were conducted in Phase I of this study to determine de
sirable items for the NP. Phi coefficients, which are
designed to correlate two dichotomous variables, were com
puted. The two dichotomous variables correlated were the
student's item response (correct or incorrect) and student
classification (gifted or not-gifted) on the WISC-R. The
criterion for deciding if an item was to be included was
significance at the .05 level. For the present sample,
this meant that any item with a phi coefficient greater
than .182, was selected for the NP screening test.
Phi is based on the proportions of cases passing and
failing an item in both the gifted and not-gifted crite
rion groups. The phi coefficient is known to be biased


29
when the SIT is used to predict WISC-R IQ among the
gifted.
Other IQ screening tests such as Guilford's Structure
of the Intellect Test (SOI) (Pearce, 1983; Stenson, 1982),
Ravens Progressive Matrixes (Pearce, 1983; Petty & Field,
1980), The Peabody Picture Vocabulary Test (Mize et al.,
1979; Pedriana & Bracken, 1982), and The Ammons Quick Test
(Joesting & Joesting, 1971; Kendall & Little, 1977; Nich
olson, 1 977 ) have been correlated with the WISC, WISC-R,
and the Stanford-Binet. In some instances significant
correlations have been found. However, very few studies
have been conducted with samples of gifted students. In
one such study, DeFilippis and Fulmer (1980) found that
the Ammons Quick Test underestimated WISC-R IQ for 99
first, fourth, and seventh graders with high ability.
In another study involving samples of gifted stu
dents, Wright (1983) correlated WISC and Peabody Picture
Vocabulary Test (PPVT) scores of 35 students referred by
teachers for gifted program testing, A correlation of
r=.27 was calculated and it was found that nearly half of
those who scored two standard deviations above the PPVT
mean were not eligible for gifted program placement based
on WISC-R IQ scores. Wright recommended that the PPVT not
be used to screen gifted program candidates.
A third study was conducted by Stenson (1982) to de
termine the concurrent validity of the Structure of


65
discriminated in favor of the not-gifted students. That
is, not-gifted students responded correctly to the items
more frequently than the gifted students. Also deleted
from Phase II analyses were five items responded to by
only about half of the Phase I sample. These five items
were located in the Science and Social Studies subtests,
which were administered on an optional basis at the dis
cretion of the various schools involved. They were de
leted because the cutoff scores were set based on total
items administered to all students. Those items not ad
ministered to all students were deleted so that all stu
dents could potentially attain the maximum raw score.
Therefore, 47 phi items were retained for use in Phase II,
42 from the CTBS and 5 from the TCS.
The mean of the total scores for the 47 items se
lected by the phi analyses was 39.37. The median and mode
were somewhat higher, 41.00 and 46.00, respectively. The
standard deviation was 7.06 and the range was 41. Most
students performed well on these items, with most respond
ing correctly to nearly all of them. This homogeneity of
scores is reflected by the negative skew of the distribu
tion. Even though scores appeared to concentrate near the
upper end of the distribution, the large range of scores
(i.e., 4 through 46) contributed to a standard deviation
of adequate size.


37
"levels" above the norm on the Metropolitan Reading Test,
61% scored above 130 on the Stanford-Binet. In his 1984
study, Sternberg found that IQ accounted for as much as
25% of the variance in scholastic performance.
In two of the few other studies in which achievement
was correlated with intelligence among superior students,
Mayfield (1979) had 573 third graders evaluated in terms
of intelligence, achievement, creativity, and teacher per
ception of IQ. Results yielded significant correlations
between intelligence and a wide range of achievement do
mains among the student sample. Similarly, Karnes,
Edwards, and McCallum (1986) found a significant correla
tion between total scores on the California Achievement
Test (CAT) and WISC-R Full Scale IQs of 41 gifted children
in grades four through six.
Thus the results of this body of literature comparing
intelligence with achievement appear conclusive, as in a
substantial number of studies it is suggested that the two
variables are fairly highly correlated.
Mallinson (1963) attempted to uncover a relationship
between intelligence and achievement in science and math.
The SRA achievement series and the SRA Primary Abilities
Test were given to secondary grade students. There was a
resulting correlation of r=.65 between verbal ability and
science (facts and principles). Verbal ability was also


9 4
Cohen, J. A. (1960). A coefficient of agreement for nom
inal scales. Educational and Psychological Measure
ment 20 37-46 .
Crofoot, M. J., & Bennett, T. S. (1980). A comparison of
three screening tests and the WISC-R in special
education evaluations. Psychology in the Schools,
1/7, 474-478 .
CTB/McGraw-Hi11. (1984). The Comprehensive Test of Basic
Skills: Technical report. Monterey, CA: CTB/McGraw-
Hil 1.
CTB/McGraw-Hi11. (1984). The Test of Cognitive Skills:
Technical report. Monterey, CA: CTB/McGraw-Hi11.
Dean, R. S. (1977). Canonical analysis of a jangle fal
lacy. Multivariate Experimental Clinical Research,
3, 17-20.
Dean, R. S. (1982). Intelligence-achievement discrep
ancies in diagnosing pediatric learning disabil
ities. Clinical Neuropsychology, 4, 58-62.
DeFilippis, N. A., & Fulmar, K. (1980). Effects at age
and IQ level on the validity of one short intelli
gence test used for screening purposes. Educational
and Psychological Measurement, 40 543-545.
Dirkes, M. A. (1981). Only the gifted can do. Educational
Horizons, 59, 138-143.
Dirks, J., Wessels, K., Quaforth, J., & Quenon, B.
(1980). Can short form WISC-R tests identify chil
dren with high Full Scale IQ? Psychology in the
Schools, 17, 40-46.
Dunn, L. M., & Markwardt, F. C. (1970). Peabody Individ
ual Achievement Test. Circle Pines, MN: American
Guidance Service.
Ebel, R. L. (1965). Measuring educational achievement.
Englewood Cliffs, NJ: Prentice-Hall.
Elman, L., Blixt, S., & Sawicki, R. (1981). The develop
ment of cutoff scores on a WISC-R in the multidimen
sional assessment of gifted children. Psychology in
the Schools, 43, 426, 428.


28
Brown that at the lower SIT ranges WISC-R IQs were under
estimated and that at the upper SIT ranges the WISC-R IQs
were overestimated. Apple concluded that valuable diag
nostic information yielding a qualitative picture of the
child's strength is omitted when the SIT alone is used as
a screening indicator.
Whereas Karnes and Brown as well as Apple compared
SIT and WISC-R scores for youngsters already placed in
gifted classes, Rust and Lose (1980) attempted to accu
rately screen potentially gifted students in first through
seventh grade. Based on teacher referrals and SIT scores
of 130 or above, 438 students were found eligible for
WISC-R evaluation. Of these, 132 were utilized in the re
search sample. According to stepwise regression equa
tions, the SIT was found to be a significant predictor of
Full Scale IQ. However, of the 132 students predicted,
only 61 achieved WISC-R IQs of 130 or above. Thus, set
ting the SIT cutoff at 130 failed to screen out 54% of the
nongifted students. If a cutoff of 134 had been used, as
suggested by Karnes and Brown, 42 evaluations would have
been eliminated. However, of those 42, 12 would have been
gifted. Karnes and Brown noted that while there was a
high correlation between the SIT and WISC-R, there was a
great deal of variability with individual cases. It was
concluded that in all studies high error can be expected


101
Stedman, J. N., Lawlis, G. F., Cortner, R. H., & Achten-
berg, D. (1978). Relationships between WISC-R fac
tors, Wide Range Achievement Test scores, and visual
motor maturation in children referred for psycholog
ical evaluation. Journal of Consulting and Clinical
Psychology, 46, 869-872.
Stenson, C. M. (1982). Note on concurrent validity of
structure of intellect gifted screener with Wechsler
Intelligence Scale for Children-Revised. Psycholog
ical Reports, 50, 552.
Sternberg, R. J. (1982). Lies we live by: Misconceptions
of test in identifying the gifted. Gifted Child
Quarterly, 26, 157-161.
Stewart, K. D., & Jones, E. C. (1976). Validity of the
Slosson Intelligence Test: A ten year review. Psy
chology in the Schools, 13, 372-380.
Stewart, D. W., & Morris, L. (1977). Intelligence and
academic achievement in a clinical adolescent popu
lation. Psychology in the Schools, 14, 513-518.
Terman, L. M., & Merrill, M. A. (1973). Stanford-Binet
Intelligence Scale. Boston: Houghton Mifflin.
Thompson, J., & Findley, C. J. (1962). The validation of
an abbreviated Wechsler Intelligence Scale for Chil
dren for use with the educable mentally retarded.
Educational and Psychological Measurement, 22, 539-
542 .
Tuttle, F. B., & Becker, L. A. (1980). Characteristics and
identification of gifted and talented students. Wash
ington: National Educational Association.
Undheim, J. O. (1976). Ability structure in 10-11-year-
old children and the theory of fluid and cry§tali2ed
intelligence. Journal of Educational Psychology, 4,
411-423.
Vandiver, P. C. & Vandiver, S. S. (1979). A "nonbiased
assessment" of intelligence testing. The Educational
Forum, 44, 97-108.
Vernon, P. E. (1961). The structure of human abilities.
London: Methuen.


15
7. Stanford-Binet, Problems of Fact
"An Indian who had come to town for the first
time in his life saw a boy riding along the
street- As the boy rode by the Indian said,
"The white boy is lazy; he walks sitting down!
What was the boy riding on that caused the In
dian to say, 'He walks sitting down'?
TCS, Verbal Reasoning (instructions: find the true
statement)
All bicycles have gears.
Some bicycles have ten speeds.
Maria has a bicycle.
F. Maria likes her bicycle.
G. Maria's bicycle has gears.
H. Maria's bicycle goes too fast.
J. Maria's bicycle has ten speeds.
8. Stanford-Binet, Induction
(This is a sequential test in which paper is
folded and holes cut in it by the examiner. The
student must deduce a pattern to predict how
many holes will result from each cut.)
TCS, Sequences
(Students are presented sequential patterns that
are incomplete. The student must deduce the
pattern and predict the final pattern.)
Need for This Study
Teachers at the seventh-grade level typically experi
ence more difficulty identifying potentially gifted stu
dents than do teachers at lower grade levels (Schnell,
1982). Consequently, junior high school teachers refer a
larger proportion of not-gifted students for evaluation
than do teachers of elementary school students. This
phenomenon is believed to result from more limited contact
between individual teachers and students and from the fact
that the pool of potentially gifted students from which


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
This dissertation was submitted to the Graduate Faculty of
the College of Education and to the Graduate School and
was accepted as partial fulfillment of the requirements
for the degree of Doctor of Philosophy.
August 1987
Dean, College of Education
Dean, Graduate School


2
children with special educational needs, has placed a bur
den on local school districts to identify gifted students
through accurate referral procedures and efficient use of
testing personnel (Karnes & Brown, 1979). As a result,
school systems are currently faced with demands for intel
lectual assessment of children which are often greater
than can be met by the qualified examiners available (Cro-
foot & Bennett, 1980; Fell & Fell, 1982). There is a
demand for screening procedures designed to facilitate the
referral process by maximizing the use of individual
testing time while minimizing errors in identification
(Jenkins-Friedman, 1982; Kramer, Markley, Shanks, &
Ryabik, 1983; Rust & Lose, 1980; Stephens & Gibson, 1963).
The Wechsler Intelligence Scale for Children--Revised
(WISC-R) and Stanford-Binet Intelligence Scale (S-B) are
the most widely used individual tests of children's intel
ligence (Bryan & Bryan, 1975; Salvia & Ysseldyke, 1978;
Wikoff, 1978). Screening tests which estimate or predict
intelligence scores on the WISC-R and Stanford-Binet have
been studied for use among gifted school populations since
the 1950s (Pegnato k Birch, 1959; Sheldon b Manolakes,
1954 ) .


11
of the WISC-R measures acquired knowledge and is influ
enced by outside reading and school learning. The Sim
ilarities subtest was also found to be subject to reading
and vocabulary knowledge. Likewise, the other Verbal
Scale subtests were found to have strong components of
acquired knowledge.
In one of a very few studies in which IQ test item
content was related to specific academic skills, Washing
ton, Engelmann, and Bereiter (1969) conducted an item
analysis of the Stanford-Binet Intelligence Scale and at
tempted to construct an academic curriculum from it.
After the curriculum was presented to students an achieve
ment test was administered. Results showed that the
prelearned S-B items were positively correlated with post-
test achievement items for particular learning tasks. In
a second phase of the study no pretest was administered.
However, subsequent to the curriculum presentation and
post-test administration the Stanford-Binet was given.
The achievement test results were found to accurately pre
dict S-B scores in terms of items responses. Results sug
gested content validity across the IQ and achievement
measures.
A number of other researchers have investigated the
relationship between intelligence and achievement measures
and found them to be positively correlated (Hale, 1978;
Hartlage i. Steele, 1 977 ; Reschley & Reschley, 1979;


10
demonstrated empirically by Anastasi (1976). Her examina
tion of the content of several current instruments classi
fied as achievement and intelligence tests revealed simi
larity in their content. Supporting this finding she
contended that it has long been known that IQ tests corre
late about as highly with achievement tests as different
IQ tests correlate with each other. Further, one of the
most frequently employed means of validating IQ tests is
to compare them with measures of achievement.
In another attempt to show some common elements of
achievement and intelligence, Gronlund (1976) compared
factors measured by both reading readiness tests and IQ
tests. These elements included
1. visual discrimination--identifying similarities
and differences in words or pictures;
2. auditory di serimination--identifying similar
ities and differences in spoken words;
3. verbal comprehension--demonstrating an under
standing of the meaning of words, sentences, and
directions; and
4. copying--demonstrating skills in reproducing
geometric forms.
In his analysis of the WISC-R Verbal Scale, Kaufman
(1979) identified item content that reflects properties of
achievement tests. For example, the Information subtest


5
Purpose
The primary purpose of this study was to determine
whether a large, group-administered achievement/abi1ity
test battery possesses items that, in combination, yield a
score that accurately predicts gifted classification as
measured by the WISC-R or Stanford-Binet. A secondary
purpose of this research was to determine if the new
screening procedure (NP) classifies gifted and not-gifted
seventh graders more accurately than the Slosson Intelli
gence Test (SIT) and at what cutoffs these classifications
are most accurate.
The new screening procedure developed Lor this study
consisted of a subset of items selected from norm refer
enced, group-administered tests of academic aptitude and
achievement. The aptitude test used in this study was the
Test of Cognitive Skills (TCS) and the achievement battery
was the Comprehensive Test of Basic Skills (CTBS). The
TCS and CTBS have been normal on the same sample and are
typically administered in concurrent testing sessions.
Research Questions
The following research questions were addressed:
1. Can an accurate predictor of gifted IQ classifi
cation on the WISC-R/S-B be derived from an instrument


79
Summary of Results
In Phase I of this study 56 items from the CTBS/TCS
were identified using phi coefficients that significantly
differentiated gifted from not-giffed students. The index
of discrimination analyses yielded 24 items that success
fully discriminated gifted from not-gifted. A large pro
portion of' significant D items were common to the phi
analysis results.
In Phase II, total scores for each subset of items
were computed using a cross validation sample of exam
inees. The accuracy of classifications resulting from
each of the two new tests and the SIT were contrasted at
five different cutoff scores. Kappa analyses revealed
higher correct prediction values for the NP-D than for
either the NP-phi and the SIT at all cutoff levels.


81
Phase I
In Phase I of the study it was suggested that some
items derived from the CTBS/TCS were valid for distin
guishing between gifted and not-gifted seventh graders.
Evidence for this assertion included (a) the relatively
large number.of NP items that are common to both the phi
and D analyses, (b) the large number of items found to
discriminate between the two groups, and (c) the agreement
between CTBS and IRT location parameter estimates and NP
item statistics.
Common Items
The index of discrimination (D) analyses yielded
fewer discriminating items than did the phi analyses.
However, the validity of the D analyses is supported by
the overlap of items between the phi and D analyses. Of
the 14 items retained for the NP-D, 11 items were also in
cluded in the phi NP scale. Overlap between phi and D
item discrimination for middle difficulty items was re
ported by Engelhart (1965).
Support for the validity of the NP also was suggested
by the relatively large number of significant items, par
ticularly on the phi analyses. In addition to the 56 sig
nificant phi items initially computed at <¡.05, another 29
were significant at the .10 level. Clearly an even larger
number of significant items would have resulted had the


96
Hale, R. L. (1978). The WISC-R as a predictor of WRAT
performance. Psychology in the Schools, 15, 172-175.
Hammill, D., & McNutt, G. (1981). The correlates of
reading. Austin, TX: Pro-Ed.
Harrington, R. G. (1982). Standardized testing may be
hazardous to the educational progress of intellec
tually gifted children. Education, 103, 112-117.
Hartlage, L. C., & Steele, C. T. (1977). WISC and WISC-R
correlates of academic achievement. Psychology in
the Schools, 14, 15-18.
Hirsch, F. J., & Hirsch, S. J. (1980). The Quick Test as
a screening device for gifted students. Psychology
in the Schools, 17, 37-46.
Horn, J. L. (1970). Factor analysis with variables of
different metric. Educational and Psychological Mea
surement, 29 753-762 .
Hurrocks, J. E. (1964). Assessment of behavior, the meth
odology and content of psychological measurement.
Columbus, OH: Charles E. Merrill Publishing.
Jenkins-Friedman, R. (1982). Myth: Cosmetic use of-mul
tiple situation criteria. Gifted Child Quarterly,
26, 24-26.
Jensen, A. R. (1980). Bias in mental testing. New York:
McMillan.
Jensen, A. R. (1981). Straight talk about mental tests.
New York: Free Press.
Joesting, J., & Joesting, R. (1971). The Quick Test as a
screening device in a welfare setting. Psychological
Reports, 29, 1289-1290.
Karnes, F. A., & Brown, K. E. (1979). Comparison of the
SIT with the WISC-R for gifted students. Psychology
in the Schools, 16, 478-482.
Karnes, F. A., & Brown, K. E. (1980). Factor analysis of
the WISC-R for the gifted. Journal of Educational
Psychology, 72, 197-199.
Karnes, F. A., & Brown, K. E. (1981). A short form of the
WISC-R for gifted students. Psychology in the
Schools, 18, 169-173.


90
to two factors. First, it is likely that for high achiev
ing students, "superior" intelligence is heavily loaded in
the verbal reasoning domains that are tapped to a greater
extent by the CTBS than by the TCS. The TCS, to a greater
degree than the CTBS, measures non-verbal skills or abil
ities that may be less represented by high achieving stu
dents' strengths' on the WISC-R or Stanford-Binet. Future
researchers might investigate this hypothesis by examining
the relationship between achievement and verbal vs. non
verbal intelligence (as represented by these or similar
tests) for high achieving or gifted students.
A second explanation for the disproportionate repre
sentation of achievement test items on the NP concerns the
theoretical rationale for this research. That is, the
content of achievement tests (e.g., CTBS) and IQ tests
(e.g., WISC-R) are often operationally indistinguishable.
This premise has been supported empirically by Anastasi
(1982), among many others (see Chapter II).
Cutoff s
Conclusions regarding the most desirable cutoff for
the NP (NP-D at -.33 or +.66) were discussed earlier in
the chapter. There are some other conclusions that may be
drawn regarding cutoff scores for the SIT with this popu
lation. The mean SIT score for the Phase II sample was
137. This score is somewhat above what would be expected
yet consistent with research findings by Karnes and Brown


AN ASSESSMENT PROCEDURE FOR DETECTING
GIFTEDNESS USING AVAILABLE DATA
BY
RANDY SCHNELL
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1987


59
would score. Since on most of the items, students in the
research sample were expected to score well above the
location parameter, most of the items provided little dif
ferentiation among these students with much higher scale
scores. However, in defense of these instruments for use
in this study, some items with extreme location parameters
were purposely included in the CTBS by its constructors.
These items were well above or below the range of perfor
mance for which the test was designed. Item characteris
tic curves indicate the existence of CTBS items that were
passed typically only by students with very high total
scores on the test (CTB/McGraw-Hil1, 1984). IRT item lo
cation parameters for the TCS had not been calculated,
though since the TCS was designed to predict achievement
for the general population, it might be assumed that loca
tion parameter criteria would adhere to a similar ration
ale.
A second limitation of this study concerns measure
ment error due to the variable adherence to test standard
ization by administrators of the CTBS, TCS and the SIT.
Because teachers and counselors receive varying degrees of
formal training on the importance of precise adherence to
test standards and on the influence of standardization de
viations on test reliability, administrators' strict con
formity to test standards was questionable. Because very
little training was provided for the CTBS/TCS and SIT,


19
intelligence test were used as a criterion reference for
the designation of those children who were, indeed,
gifted. For each of the seven screening procedures,
effectiveness was judged by the percentage of gifted chil
dren located; efficiency was defined as the ratio between
total number of gifted students and students predicted as
gifted. Of the 781 subjects, 91 (6.5' oi the school popu
lation) were judged to be gifted. These results indicated
that, among the seven methods, group IQ and achievement
tests were the better predictors, providing the best pos
sible combination of effectiveness with efficiency. Other
methods, such as honor role inclusion, were fairly effec
tive, but their efficiency was poor.
Even though Pegnato and Birch (1959) were largely un
successful in finding an effective predictor of gifted IQ,
a substantial amount of research has focused on screening
the gifted since that time. The following sections of
this literature review are concentrated on four procedures
for gifted student identification that involve group IQ
tests, IQ short forms, IQ screening tests, and achievement
test scores.
Group IQ Tests
The largely unsuccessful attempt by Pegnato and Birch
(1959) to predict gifted intelligence using group IQ tests
was followed one year later by a similar study. Chambers


48
reasoning. The most common type of test, especially of
the upper age levels, is that employing verbal content.
Data on criterion-related validity of the Stanford-
Binet have been obtained chiefly in terms of academic
achievement (Anastasi, 1976). Correlations between the
scale and school grades, teachers' ratings, and achieve
ment test scores generally fall between .45 and .75. The
Stanford-Binet tends to correlate highly with performance
in nearly all academic courses, but predominantly with
verbal courses such as English and history. Correlations
with achievement test scores show the same pattern. The
rigorous standardization and ronorming of the Stanford
Binet, along with its high validity and reliability, indi
cate that it was an appropriate IQ test for this study.


14
3. Stanford-Binet, Arithmetic Reasoning
If a man's salary is $20 a week and he spends
$14 a week, how long will it take him to save
$300?
CTBS, Mathematics Concepts and Applications
To pay for groceries, Scott, Marvin, and Carol
each gave the clerk $1.35. The clerk gave them
$.45 in change. How much did the groceries
cost?
P. $1.80
G. $3.60
H. $4.05
J. $4.50
4. Stanford-Binet, Vocabulary
What does "Brunette" mean?
CTBS, Vocabulary (instructions: Choose the word or
phrase that means the same ... as the underlined
word.)
successful merchant
P. parade
G. business
H. customer
J. shopkeeper
5. WISC-R Picture Arrangement (instructions: "... I
want you to arrange these pictures in the right order
to tell a story that makes sense.")
TCS, Sequence (instructions: ". . choose the part
that would continue the pattern or sequence.") Vari
ous visual stimuli are presented such as letters,
numbers or geometric shapes.
6. WISC-R, Similarities
In what way are a telephone and a radio alike?
TCS, Verbal Reasoning (paraphrased instructions:
words in the top and bottom rows are related in the
same way. Find the word that completes the bottom
row of words.)
radio electricity music
paper newspaper
F. ink
G. story
H. reporter
I. typewriter
The


60
misinterpretation of instructions or standards was pos
sible as well.
A related source of potential measurement error
existed due to the group format by which the CTBS/TCS is
administered. As opposed to the WISC-R and SIT, adminis
trators were restricted in their ability to closely mon
itor individual students and control for such factors as
misinterpreted directions or acute physical or emotional
liabilities of students. Since regular CTBS/TCS testing
was conducted only once a year in the research school dis
trict, efforts were made to test as many students as pos
sible during that time. However, SIT and WISC-R testing
were more easily postponed to a later date if a situation
warranted such action.
Another limitation of the CTBS specific to this study
was that some students were not administered the Science
and Social Studies subtests in the school district, be
cause they were optional and were administered only in
some schools. These subtests are comprised of 40 items
each and may contain items useful for discrimination of
gifted and not-gifted students. If deleted from analysis,
these items may detract from the overall accuracy of the
proposed new procedure in accomplishing its intended goal.
The administration of the WISC-R and SIT over a
3-year period suggests a question regarding score


17
The utility of finding a small pool of items that
correlates highly with gifted intelligence is that (a)
current forms of the CTBS/TCS selected items may be re
tained for group administration to future gifted class
candidates, (b) if a gifted candidate is not present for
CTBS/TCS testing the entire test will not have to be
administered as a gifted screening, and (c) the procedure
of analyzing a test in this way might be useful for pre
dicting intelligence (or other traits) among other popula
tions, utilizing these or other tests.
Overview of the Remainder of the Study
The subsequent content of this study is divided into
four chapters. In Chapter II a review of related liter
ature is presented. A description of the methodology used
for the research comprises Chapter III. Research results
are presented in Chapter IV and results are discussed in
Chapter V.


46
.80 to .84. The TCS Technical Reports provide reliabil
ities on SEMs for subtests based on number correct at each
grade level. Composite calculations for the total test
are not provided. Also reported are bias studies and
tables indicating how test biases are accounted for and
control led.
In summary, the CTBS and TCS were well suited for
this study because of the sophisticated method utilized in
analyzing items and the tests' high validity and reliabil
ity. Further, both tests employed sampling procedures
designed to provide norms for the entire U.S. school popu
lation. Research has also been conducted to aid in reduc
ing test bias for the CTBS and TCS.
Stanford-Binet Intelligence Scale (S-B)
The third revision of the Stanford-Binet (S-B), pub
lished in 1960, remained unchanged in content and format
through 1985. A revised version of the S-B was published
in 1986. The 1960 version was constructed by combining
forms L and M of the 1937 scale and eliminating those
items considered obsolescent and by relocating items whose
difficulty level had altered during the intervening years.
The test was, however, restandardized in 1972. New norms
were derived from a sample of approximately 2,100 cases
during the 1971-72 school year. Children in the 1972 norm
group were chosen from 20,000 school age children in


36
on a general intelligence factor and a good predictor of
intel1igence.
Researchers whose views represent the latter view
point have supported the notion that reading ability is
highly correlated with general intelligence and is a good
predictor of intelligence. Jensen (1981) reported a cor
relation of r = .68 between reading comprehension and Full
Scale IQ for a large sample of students. Other research
ers have cited similarly high (.60-.70) correlations be
tween reading and IQ for various samples of students in
grades K through 12 (Brooks, 1977; Hale, 1978; Hartlage &
Steele, 1977; Ryan, 1979; Wikoff, 1978; Yapp, 1977; Yule,
Gold, & Busch, 1981). In a literature review of 34
studies, Hammill and McNutt (1981) found a median correla
tion of .75 between measures of intelligence and achieve
ment .
Reasoning that the most efficient gifted screening
assessment would be significantly correlated with achieve
ment if giftedness is defined as superior school-related
ability, Grossman and Johnson (1983) investigated the
Stanford Achievement Test. They found a significant cor
relation with intelligence among 46 children with SIT IQs
above 120.
In another pertinent study, Schena (1963) found that
of 226 sixth and seventh graders who scored two or more


31
Lennon (1978) has found that relationships between
intelligence and achievement tests are so strong as to
lead to the criticism that the two types of tests do not
measure anything different. Both tests measure what the
student has learned (Gronlund, 1976) and both tests pre
dict future learning with similar degrees of success. IQ
tests and achievement tests differ in form but not neces
sarily in content (Mercer, 1979).
In a series of studies in the late 1970s and 1980s
WISC-R IQs were correlated with achievement subtests of
the Wide Range Achievement Test (WRAT). Consistently high
correlations were found. Some of these early studies are
summarized in Table 1-1.
Also, in 1978, Stedman, Lawlis, Cortner, and Achten-
berg attempted to relate Kaufman's (1975) factor scores to
WRAT attainment in a population of 76 children, ages 6 to
13. Correlations were found to be positive and signif
icant .
Yule, Gold, and Busch (1981) administered the WISC-R
and a battery of achievement tests to students at age 16
1/2. Achievement measures included tests of "sentence
reading," spelling, and arithmetic. WISC-R, Verbal IQ
shared 50% of the variance in reading, spelling, and
arithmetic. Correlations between Full Scale IQ and
achievement were as high as r=.80.


35
Moderately high correlations r=.40 to .50 were found be
tween some of the other IQ and achievement subtests.
Stewart and Morris (1977) factor analyzed the WISC,
WAIS, WRAT, and CAT (California Achievement Test) for 182
students ranging in age from 11 to 18. A "substantial"
overlap of verbal intelligence and academic achievement
was found. Resulting factors conformed reasonably well to
those of Kaufman ( 1975). Subtests from each measure were
found to load on each of the IQ factors.
In a study wherein the abilities underlying reading
readiness were identified, Olsen and Rosen (1971) factor
analyzed three group reading tests and the WISC. Subjects
consisted of 218 first graders. The 35 subtests were cor
related and the resulting matrices subjected to a prin
cipal component analysis. Four common factors were re
vealed. In one factor, reading comprehension loaded with
four WISC-R subtests. In another, "writing letters" cor
related highly with WISC-R Vocabulary. In a third factor,
sentence writing and WISC-R Coding were included.
There have emerged two camps of thought on the issue
of reading skill acquisition and intelligence. On one
hand, in some research it has been suggested that reading
is a function of information processing or encoding skills
as opposed to being a primarily intellectual function. On
the other hand, in similarly focused factor analytic
investigations reading has been found to be highly loaded


57
incorrect) classifications for NP-phi, NP-D, and the SIT
while adjusting for the percentage of correct classifica
tions that could be expected on the basis of chance alone.
Kappa is a descriptive statistic and not a test of statis
tical significance.
Using the Kappa statistic proportions of predictions
were compared at various cutoff points. Kappa adjusts for
predictions expected by chance alone by taking into
account both observed and expected proportion classifica
tions (Cohen, 1960). For example, when considering K val
ues for the NP, the numerator of K is regarded as the pro
portion of students consistently classified by both the NP
and the WISC-R (observed) over and above the product of
the proportions of students classified individually by
each test (expected). The denominator of K is the maximum
possible increase in decision-consistency above chance
level, given the proportions classified by the two tests
independently.
The formula used to compute Kappa was
where P = proportion of consistent gifted and not-gifted
classifications for WISC-R and NP or WISC-R and
SIT


ACKNOWLEDGEMENTS
I would like to thank Larry Loesch, Ph.D., for his
guidance and assistance as chairperson of my doctoral com
mittee. Dr. Loesch graciously assumed the role of chair
person during a period when my study was in disarray and
not progressing satisfactorily. I also wish to thank
Robert Jester, Ph.D., whose consultation assisted in
development of the framework of this study, and Janet
Larsen, Ed.D., who has supported my doctoral work since
1982 when she agreed to be my advisor. I would also like
to acknowledge Linda Crocker, Ph.D., whose patience, dili
gence, and expertise in measurement were vital to the
study and greatly appreciated.
A number of other people who contributed to this
study also deserve recognition. These include John
Hilderbrand, Ph.D., Hugh Morehouse, Robert Haines, Ed.D.,
Grace Hutchinson, Michael Selby, Lois Rudloff, and Denise
Landau, Ph.D.
I wish to express special thanks to my parents and
other family members for their support.
11


74
Table 4-6 was constructed to consolidate and clarify
the data on predictive accuracy, in that table, propor
tions of correct and incorrect predictions are again
expressed as true positives (TP), true negatives (TN),
false positives (FP), and false negatives (FN). Deleted
were the cutoffs -1.0 SD, -.50 SD, +.50 SD, and -1.0 SD.
Inspection of Table 4.6 reveals that all three methods of
screening gifted students were approximately equivalent.
That is, all three found the greatest proportions of TP
predictions in .492 through .525 of the sample. Lowest TN
predictions occurred in .082 through .148 of the sample.
After the prediction rates were established for the
three tests and the prediction values were expressed as
proportions of the sample, analyses were conducted to de
termine the proportions, if any, that were most accurate.
Coefficient Kappa analyses were calculated to assess the
relative gifted prediction accuracy of the NP-phi, NP-D,
and SIT.
Kappa Comparisons
In results shown in Table 4-7, P values correspond to
percentages of correct classifications (both true pos
itives and true negatives). However, two equal classifi
cation percentages (e.g., as observed for phi and D at 0.0
cutoff) may not have equal corresponding K-values because
the Kappa equation adjusts for chance correct predictions.



PAGE 1

$1 $66(660(17 352&('85( )25 '(7(&7,1* *,)7('1(66 86,1* $9$,/$%/( '$7$ %< 5$1'< 6&+1(// $ ',66(57$7,21 35(6(17(' 72 7+( *5$'8$7( 6&+22/ 2) 7+( 81,9(56,7< 2) )/25,'$ ,1 3$57,$/ )8/),//0(17 2) 7+( 5(48,5(0(176 )25 7+( '(*5(( 2) '2&725 2) 3+,/2623+< 81,9(56,7< 2) )/25,'$

PAGE 2

$&.12:/('*(0(176 ZRXOG OLNH WR WKDQN /DUU\ /RHVFK 3K' IRU KLV JXLGDQFH DQG DVVLVWDQFH DV FKDLUSHUVRQ RI P\ GRFWRUDO FRPn PLWWHH 'U /RHVFK JUDFLRXVO\ DVVXPHG WKH UROH RI FKDLUn SHUVRQ GXULQJ D SHULRG ZKHQ P\ VWXG\ ZDV LQ GLVDUUD\ DQG QRW SURJUHVVLQJ VDWLVIDFWRULO\ DOVR ZLVK WR WKDQN 5REHUW -HVWHU 3K' ZKRVH FRQVXOWDWLRQ DVVLVWHG LQ GHYHORSPHQW RI WKH IUDPHZRUN RI WKLV VWXG\ DQG -DQHW /DUVHQ (G' ZKR KDV VXSSRUWHG P\ GRFWRUDO ZRUN VLQFH ZKHQ VKH DJUHHG WR EH P\ DGYLVRU ZRXOG DOVR OLNH WR DFNQRZOHGJH /LQGD &URFNHU 3K' ZKRVH SDWLHQFH GLOLn JHQFH DQG H[SHUWLVH LQ PHDVXUHPHQW ZHUH YLWDO WR WKH VWXG\ DQG JUHDWO\ DSSUHFLDWHG $ QXPEHU RI RWKHU SHRSOH ZKR FRQWULEXWHG WR WKLV VWXG\ DOVR GHVHUYH UHFRJQLWLRQ 7KHVH LQFOXGH -RKQ +LOGHUEUDQG 3K' +XJK 0RUHKRXVH 5REHUW +DLQHV (G' *UDFH +XWFKLQVRQ 0LFKDHO 6HOE\ /RLV 5XGORII DQG 'HQLVH /DQGDX 3K' ZLVK WR H[SUHVV VSHFLDO WKDQNV WR P\ SDUHQWV DQG RWKHU IDPLO\ PHPEHUV IRU WKHLU VXSSRUW

PAGE 3

7$%/( 2) &217(176 3DJH $&.12:/('*(0(176 L L $%675$&7 9 &+$37(5 ,1752'8&7,21 3UREOHP *LIWHGQHVV 3XUSRVH 5HVHDUFK 4XHVWLRQV 'HILQLWLRQ RI 7HUPV 7KHRUHWLFDO 5DWLRQDOH 1HHG IRU 7KLV 6WXG\ 2YHUYLHZ RI WKH 5HPDLQGHU RI WKH 6WXG\ ,, /,7(5$785( 5(9,(: 6XSSRUW IRU WKH 3UREOHP *URXS ,4 7HVWV ,QWHOOLJHQFH 4XRWLHQW 6KRUW )RUPV ,QWHOOLJHQFH 4XRWLHQW 6FUHHQLQJ 7HVWV $FKLHYHPHQW 7HVWV 6XPPDU\ ,QVWUXPHQWV 8VHG LQ 6WXG\ :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG :,6&5f 6ORVVRQ ,QWHOOLJHQFH 7HVW 6,7f &RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV &7%6f7HVW RI &RJQLWLYH 6NLOOV 7&6f 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH 6%f LLL

PAGE 4

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

PAGE 5

$EVWUDFW RI 'LVVHUWDWLRQ 3UHVHQWHG WR WKH *UDGXDWH 6FKRRO RI WKH 8QLYHUVLW\ RI )ORULGD LQ 3DUWLDO )XOILOOPHQW RI WKH 5HTXLUHPHQWV IRU WKH 'HJUHH RI 'RFWRU RI 3KLORVRSK\ $1 $66(660(17 352&('85( )25 '(7(&7,1* *,)7('1(66 86,1* $9$,/$%/( '$7$ %\ 5DQG\ 6FKQHOO 'HFHPEHU &KDLUPDQ /DUU\ /RHVFK 3K' 0DMRU 'HSDUWPHQW &RXQVHORU (GXFDWLRQ 7KH SXUSRVH RI WKLV VWXG\ ZDV WR H[DPLQH D QHZ SV\n FKRPHWULF VFUHHQLQJ SURFHGXUH GHVLJQHG WR GLVFULPLQDWH LQWHOOHFWXDOO\ JLIWHG VHYHQWK JUDGHUV IURP RWKHU KLJK DFKLHYLQJ VHYHQWK JUDGHUV $OO VWXGHQWV LQ WKH UHn VHDUFK VDPSOH KDG EHHQ DVVHVVHG WR GHWHUPLQH HOLJLELOLW\ LQ WKH +LOOVERURXJK &RXQW\ 3XEOLF 6FKRROV JLIWHG SURn JUDP $Q DFFXUDWH DQG HIILFLHQW VFUHHQLQJ SURFHGXUH ZDV QHFHVVDU\ LQ RUGHU WR GHOHWH IURP WLPHFRQVXPLQJ IRUPDO DVVHVVPHQW WKRVH VWXGHQWV ZKR ZHUH XQOLNHO\ WR PHHW LQWHOn OHFWXDO FULWHULD IRU JLIWHG SURJUDP SODFHPHQW 7KH 6ORV VRQ ,QWHOOLJHQFH 7HVW 6,7f SUHYLRXVO\ XVHG WR VFUHHQ VWXGHQWV UHIHUUHG WR WKH JLIWHG FODVVHV KDG SURYHG LQHIn ILFLHQW IRU WKLV SXUSRVH $ SOHWKRUD RI UHVHDUFK RQ Y

PAGE 6

PHWKRGV XVHG WR LGHQWLI\ JLIWHG VWXGHQWV UHYHDOHG WKDW PRVW PHWKRGV ZHUH OHVV WKDQ DFFXUDWH 7KH &RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV DQG 7HVW RI &RJQLWLYH 6NLOOV ZHUH LWHP DQDO\]HG RQ D VDPSOH RI VHYHQWK JUDGH VXEMHFWV XVLQJ WZR FRUUHODWLRQDO SURFHGXUHV 5HVXOWV IURP WKH LWHP DQDO\VHV \LHOGHG LWHPV IURP WKH SKL DQDO\VLV DQG IURP WKH LQGH[ RI GLVFULPLQDWLRQ DQDOn \VLV WKDW GLVFULPLQDWHG JLIWHG IURP QRW JLIWHG VWXGHQWV DV FODVVLILHG E\ WKH :,6&5 RU 6WDQIRUG %LQHW &URVV YDOn LGDWLRQ ZDV FRQGXFWHG RQ D VHFRQG VDPSOH RI VWXGHQWV WR GHWHUPLQH Df WKH DFFXUDF\ RI WKH QHZ VFUHHQLQJ SURFHGXUH 13f FRPSDUHG WR WKH 6,7 DQG Ef DW ZKDW FXWRII SRLQWV HLWKHU WKH 13 RU 6,7 ZDV PRUH DFFXUDWH ,Q FURVV YDOLGDn WLRQ WRWDO VFRUHV RI ERWK LWHP VXEVHWV ZHUH FRPSDUHG ZLWK 6,7 7KH LWHPV REWDLQHG IURP WKH LQGH[ RI GLVFULPLQDWLRQ DQDO\VHV ZHUH IRXQG WR EH WKH EHVW SUHGLFWRUV RI JLIWHG LQWHOOLJHQFH IRU WKH VDPSOH 7KHUH ZDV VRPH DPELJXLW\ UHn JDUGLQJ DQ RSWLPDO FXWRII VFRUH KRZHYHU D VFRUH VWDQGDUG GHYLDWLRQV EHORZ WKH PHDQ VFRUH ZDV IRXQG WR JHQn HUDOO\ \LHOG PRVW DFFXUDWH SUHGLFWLRQV YL

PAGE 7

&+$37(5 ,1752'8&7,21 7KH WUHPHQGRXV LQFUHDVH VLQFH LQ VFKRRO SURJUDPV IRU WKH LQWHOOHFWXDOO\ JLIWHG KDV OHG PRVW VWDWHV WR HVn WDEOLVK JXLGHOLQHV RU UHTXLUHPHQWV IRU SURJUDP HOLJLELOLW\ .DUQHV t %URZQ .ROORII t )HOGKXVHQ f 2QH FRPPRQ UHTXLUHPHQW LV IRU LQGLYLGXDO LQWHOOHFWXDO HYDOXDn WLRQ RI JLIWHG SURJUDP FDQGLGDWHV &KDPEHUV %DUURQ t 6SUHHKHU 6WHUQEHUJ 9HUQRQ $GDPVRQ N 9HUn QRQ f
PAGE 8

FKLOGUHQ ZLWK VSHFLDO HGXFDWLRQDO QHHGV KDV SODFHG D EXUn GHQ RQ ORFDO VFKRRO GLVWULFWV WR LGHQWLI\ JLIWHG VWXGHQWV WKURXJK DFFXUDWH UHIHUUDO SURFHGXUHV DQG HIILFLHQW XVH RI WHVWLQJ SHUVRQQHO .DUQHV t %URZQ f $V D UHVXOW VFKRRO V\VWHPV DUH FXUUHQWO\ IDFHG ZLWK GHPDQGV IRU LQWHOn OHFWXDO DVVHVVPHQW RI FKLOGUHQ ZKLFK DUH RIWHQ JUHDWHU WKDQ FDQ EH PHW E\ WKH TXDOLILHG H[DPLQHUV DYDLODEOH &UR IRRW t %HQQHWW )HOO t )HOO f 7KHUH LV D GHPDQG IRU VFUHHQLQJ SURFHGXUHV GHVLJQHG WR IDFLOLWDWH WKH UHIHUUDO SURFHVV E\ PD[LPL]LQJ WKH XVH RI LQGLYLGXDO WHVWLQJ WLPH ZKLOH PLQLPL]LQJ HUURUV LQ LGHQWLILFDWLRQ -HQNLQV)ULHGPDQ .UDPHU 0DUNOH\ 6KDQNV t 5\DELN 5XVW t /RVH 6WHSKHQV t *LEVRQ f 7KH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG :,6&5f DQG 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH 6%f DUH WKH PRVW ZLGHO\ XVHG LQGLYLGXDO WHVWV RI FKLOGUHQnV LQWHOn OLJHQFH %U\DQ t %U\DQ 6DOYLD t
PAGE 9

3UREOHP 7KH UHVHDUFK SUREOHP DGGUHVVHG FRQFHUQHG WKH DFFXUDWH LGHQWLILFDWLRQ RI LQWHOOHFWXDOO\ JLIWHG VWXGHQWV IURP D VFUHHQLQJ SURFHGXUH SULRU WR DGPLQLVWUDWLRQ RI DQ LQGLYLGn XDO LQWHOOLJHQFH WHVW 7KH H[SHQVH RI DGPLQLVWHULQJ LQGLn YLGXDO LQWHOOLJHQFH WHVWV KDV QHFHVVLWDWHG WKH VFUHHQLQJ RI SRWHQWLDOO\ JLIWHG VWXGHQWV LQ DQ HIIRUW WR GHOHWH IURP IRUPDO WHVWLQJ WKRVH ZKR SUREDEO\ GR QRW SRVVHVV JLIWHG LQWHOOLJHQFH 5XVW t /RVH 6WHQVRQ f ,QDFFXn UDWH VFUHHQLQJ KDV UHVXOWHG LQ QRWJLIWHG VWXGHQWV UHFHLYn LQJ WKH WLPHFRQVXPLQJ WHVWV DQG LQ VRPH JLIWHG VWXGHQWV EHLQJ H[FOXGHG IURP WHVWLQJ 6FUHHQLQJ SURFHGXUHV VXFK DV WKH 6ORVVRQ ,QWHOOLJHQFH 7HVW 'LUNHV :HVVHOV 4XDIRUWK t 4XHQRQ *URVVPDQ t -RKKVRQ .DUQHV t %URZQ 5XVW t /RVH f WKH $PPRQV 4XLFN 7HVW 'H)LOLSSLV t )XOPDU +LUVFK t +LUVFK f VKRUW IRUPV RI WKH :,6&5 %HUVRII (OPDQ %OL[W t 6DZLFNL .DUQHV t %URZQ .LOn OLDQ t +XJKHV .UDPHU HW DO f *XLOIRUGnV 6WUXFWXUH RI WKH ,QWHOOHFW 3HDUFH f WKH 3HDERG\ 3LFWXUH 9RFDEXODU\ 7HVW 0L]H 6PLWK t &DOODZD\ :ULJKW f DQG JURXS ,4 WHVWV %ORVVHU *URVVPDQ t -RKQVRQ 3HJQDWR t %ULFK 6KHOGRQ t 0DQR ODNHV f KDYH DOO EHHQ VKRZQ WR EH PRUH RU OHVV LQDFn FXUDWH DQGRU LQHIILFLHQW IRU VFUHHQLQJ KLJK DELOLW\

PAGE 10

VWXGHQWV &KDPEHUV f DQG 6FKHQD f UHSRUWHG VRPHZKDW PRUH HQFRXUDJLQJ UHVXOWV ZLWK DFDGHPLF VNLOO PHDVXUHV DV SUHGLFWRUV RI JLIWHG LQWHOOLJHQFH $ PRUH DFn FXUDWH DQG HIILFLHQW PHDQV IRU VFUHHQLQJ JLIWHG VWXGHQWV QHHGV WR EH IRXQG *LIWHGQHVV 7KH FKDUDFWHULVWLFV DVVRFLDWHG ZLWK JLIWHG LQWHOOLJHQFH DUH DOPRVW DV QXPHURXV DV WKH VWXGHQWV WKHPVHOYHV 7XWWOH t %HFNHU f 4XDOLWDWLYH WUDLW GLIIHUHQFHV EHWZHHQ JLIWHG DQG QRWJLIWHG FKLOGUHQ DUH LQGLFDWHG IUHTXHQWO\ LQ WKH OLWHUDWXUH %DUULQJWRQ 'LUNHV *HQVOH\ 0DOH t 3HUURQH 5LFFD 5\DQ 6WHUQEHUJ f KRZHYHU HVVHQWLDO WR D GLVFXVVLRQ RI LQWHOOHFWXDO JLIWHGQHVV LQ FKLOGUHQ LV WKHLU FODVVLILFDWLRQ DFFRUGLQJ WR DQ ,4 WHVW FXWRII VFRUH *LIWHG FODVVLILFDWLRQV E\ ,4 FXWRII VFRUHV DUH HPSOR\HG LQ DQ DWWHPSW WR REMHFWLI\ FODVVLILFDWLRQ FULWHULD DQG VFKRRO SODFHPHQW GHFLVLRQV &ODVVLILFDWLRQ E\ ,4 VFRUH PD\ PLVn OHDGLQJO\ LPSO\ WKDW TXDOLWDWLYH GLIIHUHQFHV EHWZHHQ JLIWHG DQG QRWJLIWHG FKLOGUHQ DUH QHFHVVDULO\ GHPDUFDWHG E\ DQ DUWLILFLDO FXWRII %UDGHQ f ,Q WKLV VWXG\ FXWRII VFRUHV DUH XVHG WR TXDQWLI\ JLIWHG LQWHOOLJHQFH DFn FRUGLQJ WR HGXFDWLRQDO FULWHULD DQG QRW WR GHILQH OHDUQLQJ VW\OH PRWLYDWLRQ RU RWKHU SHUVRQDOLW\ WUDLWV DVVRFLDWHG ZLWK JLIWHGQHVV

PAGE 11

3XUSRVH 7KH SULPDU\ SXUSRVH RI WKLV VWXG\ ZDV WR GHWHUPLQH ZKHWKHU D ODUJH JURXSDGPLQLVWHUHG DFKLHYHPHQWDELLW\ WHVW EDWWHU\ SRVVHVVHV LWHPV WKDW LQ FRPELQDWLRQ \LHOG D VFRUH WKDW DFFXUDWHO\ SUHGLFWV JLIWHG FODVVLILFDWLRQ DV PHDVXUHG E\ WKH :,6&5 RU 6WDQIRUG%LQHW $ VHFRQGDU\ SXUSRVH RI WKLV UHVHDUFK ZDV WR GHWHUPLQH LI WKH QHZ VFUHHQLQJ SURFHGXUH 13f FODVVLILHV JLIWHG DQG QRWJLIWHG VHYHQWK JUDGHUV PRUH DFFXUDWHO\ WKDQ WKH 6ORVVRQ ,QWHOOLn JHQFH 7HVW 6,7f DQG DW ZKDW FXWRIIV WKHVH FODVVLILFDWLRQV DUH PRVW DFFXUDWH 7KH QHZ VFUHHQLQJ SURFHGXUH GHYHORSHG /RU WKLV VWXG\ FRQVLVWHG RI D VXEVHW RI LWHPV VHOHFWHG IURP QRUP UHIHUn HQFHG JURXSDGPLQLVWHUHG WHVWV RI DFDGHPLF DSWLWXGH DQG DFKLHYHPHQW 7KH DSWLWXGH WHVW XVHG LQ WKLV VWXG\ ZDV WKH 7HVW RI &RJQLWLYH 6NLOOV 7&6f DQG WKH DFKLHYHPHQW EDWWHU\ ZDV WKH &RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV &7%6f 7KH 7&6 DQG &7%6 KDYH EHHQ QRUPDO RQ WKH VDPH VDPSOH DQG DUH W\SLFDOO\ DGPLQLVWHUHG LQ FRQFXUUHQW WHVWLQJ VHVVLRQV 5HVHDUFK 4XHVWLRQV 7KH IROORZLQJ UHVHDUFK TXHVWLRQV ZHUH DGGUHVVHG &DQ DQ DFFXUDWH SUHGLFWRU RI JLIWHG ,4 FODVVLILn FDWLRQ RQ WKH :,6&56% EH GHULYHG IURP DQ LQVWUXPHQW

PAGE 12

FRPSRVHG RI LWHPV RQ WKH &7%6 DQG 7&6 LQ D VLWXDWLRQ LQ ZKLFK JLIWHGQHVV LV YLHZHG DV D GLFKRWRPRXV YDULDEOH" ,V WKH 13 PRUH DFFXUDWH WKDQ WKH 6,7 LQ FODVVLn I\LQJ JLIWHG DQG QRWJLIWHG VHYHQWK JUDGHUV" $W ZKDW FXWRII SRLQWVf LV WKH 13 PRUH DFFXUDWH WKDQ WKH 6,7 PRVW DFFXUDWH LQ FODVVLI\LQJ JLIWHG DQG QRW JLIWHG VHYHQWK JUDGHUV" 'HILQLWLRQ RI 7HUPV &RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV &7%6f 7KH &7%6 LV D VHULHV RI QRUPUHIHUHQFHG REMHFWLYHEDVHG WHVWV IRU NLQGHUJDUWHQ WKURXJK WZHOIWK JUDGH 7KH VHULHV LV GHn VLJQHG WR PHDVXUH DFKLHYHPHQW LQ WKH EDVLF VNLOOV FRPPRQO\ IRXQG LQ VWDWH DQG GLVWULFW FXUULFXOD &7%0F*UDZ+L S f $W WKH MXQLRU KLJK VFKRRO OHYHOV WKH FRQWHQW DUHDV DUH UHDGLQJ VSHOOLQJ ODQJXDJH PDWKHPDWLFV UHIHUn HQFH VNLOOV VFLHQFH DQG VRFLDO VWXGLHV )XOO 6FDOH ,4 )6,4f )XOO 6FDOH ,4 LV WKH GHULYHG LQWHOOLJHQFH TXRWLHQW RQ WKH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG DQG RQ WKH 6WDQIRUG%LQHW ,QWHOOLn JHQFH 6FDOH *HQHUDO LQWHOOLJHQFH *HQHUDO LQWHOOLJHQFH LV D VHW RI JHQHUDO FRJQLWLYH RSHUDWLRQV PHDVXUHG DV WKH RYHUDOO DELOLW\ UHTXLUHG IRU VXFFHVV RQ ,4 WHVWV *HQHUDO

PAGE 13

LQWHOOLJHQFH LV FRPSULVHG RO WKRVH WUDLWV FRPPRQO\ PHDVXUHG E\ ,4 WHVW LWHPV *LIWHG LQWHOOLJHQFH *LIWHG LQWHOOLJHQFH LV PHDVXUHG DV DQ LQWHOOLJHQFH TXRWLHQW WKDW IDOOV LQ WKH 9HU\ 6XSHn ULRU FDWHJRU\ RU WZR VWDQGDUG GHYLDWLRQV DERYH WKH WHVW PHDQ )RU WKLV UHVHDUFK JLIWHG ,4 6(0 ZKHUH 6(0 ,4 SRLQWVf RQ WKH :,6&5 DQG ,4 6(0 ZKHUH 6(0 ,4 SRLQWVf RQ WKH 6WDQIRUG%LQHW RU ,4 7KLV GHILQLWLRQ ZKLFK LQFOXGHV D 6(0 LV EDVHG RQ JXLGHOLQHV IURP +LOOVERURXJK &RXQW\ )ORULGD 6FKRRO 'LVWULFWf +LJKDFKLHYHUV 6WXGHQWV ZKR EHFDXVH RI VXSHULRU DFDGHPLF SHUIRUPDQFH KDYH EHHQ UHIHUUHG IRU JLIWHG SURn JUDP WHVWLQJ EXW ZKR DUH DVVHVVHG DV QRW LQWHOOHFWXDOO\ JLIWHG DUH UHIHUUHG WR DV KLJKDFKLHYHUV ,QWHOLJHQFH ,QWHOOLJHQFH LV WKH DJJUHJDWH JOREDO FDSDFLW\ RI WKH LQGLYLGXDO WR DFW SXUSRVHIXOO\ WR WKLQN UDWLRQDOO\ DQG WR GHDO HIIHFWLYHO\ ZLWK KLV HQYLURQPHQW :HFKVOHU S f ,QWHOOLJHQFH LV FRPSULVHG RI LQn WHOOHFWLYH IDFWRUV VXFK DV DEVWUDFW UHDVRQLQJ YHUEDO VSDWLDO QXPHULFDO DQG RWKHU IDFWRUV :HFKVOHU S f DQG QRQLQWHOHFWLYH IDFWRUV FRQVLVWLQJ RI FDSDFLWLHV DQG EHQWV GHSHQGHQW XSRQ WHPSHUDPHQW DQG SHUVRQDOLW\ ZKLFK DUH IDFWRUV RI SHUVRQDOLW\ LWVHOI :HFKVOHU S f ,QWHOOLJHQFH TXRWLHQW $Q LQWHOOLJHQFH TXRWLHQW LV D GHULYHG WRWDO RU )XOO 6FDOH VFRUH RQ DQ LQWHOOLJHQFH WHVW

PAGE 14

3HUIRUPDQFH ,4 3,4f 3HUIRUPDQFH ,4 LV D VXEVFDOH ,4 VFRUH RQ WKH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ 5HYLVHG WKDW UHSUHVHQWV WKH H[DPLQHHnV QRQYHUEDO UHDVRQLQJ RU SHUFHSWXDO RUJDQL]DWLRQ .DXIPDQ f 5HOLDELOLW\ 5HOLDELOLW\ LV WKH VTXDUHG SRSXODWLRQ FRUUHODWLRQ EHWZHHQ WKH LQGLYLGXDOnV REWDLQHG VFRUH DQG WKH LQGLYLGXDOnV K\SRWKHWLFDO WUXH VFRUH 5HOLDELOLW\ LV WKH SURSRUWLRQ RI WUXHVFRUH YDULDQFH LQ VFRUHV RQ D SDUn WLFXODU WHVW DW W KH WLPH LW ZDV WDNHQ -HQVHQ S f 6ORVVRQ ,QWHOOLJHQFH 7HVW 6,7f 7KH 6,7 LV DQ LQGLn YLGXDOO\ DGPLQLVWHUHG LQWHOOLJHQFH WHVW ZKLFK UHTXLUHV OLWWOH VSHFLDOL]HG WUDLQLQJ WR DGPLQLVWHU RQO\ DERXW PLQXWHV WR DGPLQLVWHU DQG VFRUH DQG \LHOGV ,4 VFRUHV ZKLFK DUH FORVH DSSUR[LPDWLRQV WR WKH 6WDQIRUG%LQHW ,4 >VFRUHV@ 6ORVVRQ t -HQVHQ S f 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH 6%f 7KH 6% LV RQH RI WKH WZR LQGLYLGXDOO\ DGPLQLVWHUHG LQWHOOLJHQFH WHVWV XVHG LQ WKLV VWXG\ WR PHDVXUH JLIWHG LQWHOOLJHQFH 6HH SDJH IRU D GHWDLOHG GHVFULSWLRQf 7HVW RI &RJQLWLYH 6NLOOV 7&6f 7KH 7&6 LV DQ DELOn LW\ WHVW GHVLJQHG WR DVVHVV D VWXGHQWnV DFDGHPLF DSWLWXGH DQG WKHUHE\ SUHGLFW WKH VWXGHQWnV OHYHO RI VXFFHVV LQ VFKRRO (PSKDVLV LQ 7&6 LV SODFHG RQ SUREOHP VROYn LQJ GLVFRYHULQJ UHODWLRQVKLSV HYDOXDWLQJ DQG

PAGE 15

UHPHPEHULQJ &7%0F*UDZ+L S f 7KH 7&6 LV DGPLQLVWHUHG FRQFXUUHQWO\ ZLWK WKH &7%6 9DOLGLW\ 9DOLGLW\ UHIHUV WR WKH DSSURSULDWHQHVV RI LQIHUHQFHV IURP WHVW VFRUHV RU RWKHU IRUPV RI DVVHVVPHQW 9DOLGLW\ GHDOV ZLWK KRZ IDLWKIXOO\ WKH VFRUHV UHSUHVHQW D GRPDLQ RI VNLOO NQRZOHGJH RU RI D WUDLW EHLQJ PHDVXUHG $PHULFDQ 3V\FKRORJLFDO $VVRFLDWLRQ f 9HUEDO ,4 9 ,4f 9HUEDO ,4 LV D VXEVFDOH ,4 VFRUH RQ WKH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG WKDW UHSUHVHQWV WKH H[DPLQHHnV YHUEDO UHDVRQLQJ RU YHUEDO FRPSUHKHQVLRQ .DXIPDQ f :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG :6&5f 7KH :,6&5 LV WKH LQGLYLGXDOO\ DGPLQLVWHUHG ,4 WHVW SUHGRPLQDQWO\ XVHG LQ WKLV VWXG\ WR PHDVXUH JLIWHG LQWHOOLJHQFH 6HH SDJH IRU D GHWDLOHG GHVFULSWLRQf 7KHRUHWLFDO 5DWLRQDOH 6DOYLD DQG
PAGE 16

GHPRQVWUDWHG HPSLULFDOO\ E\ $QDVWDVL f +HU H[DPLQDn WLRQ RI WKH FRQWHQW RI VHYHUDO FXUUHQW LQVWUXPHQWV FODVVLn ILHG DV DFKLHYHPHQW DQG LQWHOOLJHQFH WHVWV UHYHDOHG VLPLn ODULW\ LQ WKHLU FRQWHQW 6XSSRUWLQJ WKLV ILQGLQJ VKH FRQWHQGHG WKDW LW KDV ORQJ EHHQ NQRZQ WKDW ,4 WHVWV FRUUHn ODWH DERXW DV KLJKO\ ZLWK DFKLHYHPHQW WHVWV DV GLIIHUHQW ,4 WHVWV FRUUHODWH ZLWK HDFK RWKHU )XUWKHU RQH RI WKH PRVW IUHTXHQWO\ HPSOR\HG PHDQV RI YDOLGDWLQJ ,4 WHVWV LV WR FRPSDUH WKHP ZLWK PHDVXUHV RI DFKLHYHPHQW ,Q DQRWKHU DWWHPSW WR VKRZ VRPH FRPPRQ HOHPHQWV RI DFKLHYHPHQW DQG LQWHOOLJHQFH *URQOXQG f FRPSDUHG IDFWRUV PHDVXUHG E\ ERWK UHDGLQJ UHDGLQHVV WHVWV DQG ,4 WHVWV 7KHVH HOHPHQWV LQFOXGHG YLVXDO GLVFULPLQDWLRQLGHQWLI\LQJ VLPLODULWLHV DQG GLIIHUHQFHV LQ ZRUGV RU SLFWXUHV DXGLWRU\ GL VHULPLQDWLRQLGHQWLI\LQJ VLPLODUn LWLHV DQG GLIIHUHQFHV LQ VSRNHQ ZRUGV YHUEDO FRPSUHKHQVLRQGHPRQVWUDWLQJ DQ XQGHUn VWDQGLQJ RI WKH PHDQLQJ RI ZRUGV VHQWHQFHV DQG GLUHFWLRQV DQG FRS\LQJGHPRQVWUDWLQJ VNLOOV LQ UHSURGXFLQJ JHRPHWULF IRUPV ,Q KLV DQDO\VLV RI WKH :,6&5 9HUEDO 6FDOH .DXIPDQ f LGHQWLILHG LWHP FRQWHQW WKDW UHIOHFWV SURSHUWLHV RI DFKLHYHPHQW WHVWV )RU H[DPSOH WKH ,QIRUPDWLRQ VXEWHVW

PAGE 17

RI WKH :,6&5 PHDVXUHV DFTXLUHG NQRZOHGJH DQG LV LQIOXn HQFHG E\ RXWVLGH UHDGLQJ DQG VFKRRO OHDUQLQJ 7KH 6LPn LODULWLHV VXEWHVW ZDV DOVR IRXQG WR EH VXEMHFW WR UHDGLQJ DQG YRFDEXODU\ NQRZOHGJH /LNHZLVH WKH RWKHU 9HUEDO 6FDOH VXEWHVWV ZHUH IRXQG WR KDYH VWURQJ FRPSRQHQWV RI DFTXLUHG NQRZOHGJH ,Q RQH RI D YHU\ IHZ VWXGLHV LQ ZKLFK ,4 WHVW LWHP FRQWHQW ZDV UHODWHG WR VSHFLILF DFDGHPLF VNLOOV :DVKLQJn WRQ (QJHOPDQQ DQG %HUHLWHU f FRQGXFWHG DQ LWHP DQDO\VLV RI WKH 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH DQG DWn WHPSWHG WR FRQVWUXFW DQ DFDGHPLF FXUULFXOXP IURP LW $IWHU WKH FXUULFXOXP ZDV SUHVHQWHG WR VWXGHQWV DQ DFKLHYHn PHQW WHVW ZDV DGPLQLVWHUHG 5HVXOWV VKRZHG WKDW WKH SUHOHDUQHG 6% LWHPV ZHUH SRVLWLYHO\ FRUUHODWHG ZLWK SRVW WHVW DFKLHYHPHQW LWHPV IRU SDUWLFXODU OHDUQLQJ WDVNV ,Q D VHFRQG SKDVH RI WKH VWXG\ QR SUHWHVW ZDV DGPLQLVWHUHG +RZHYHU VXEVHTXHQW WR WKH FXUULFXOXP SUHVHQWDWLRQ DQG SRVWWHVW DGPLQLVWUDWLRQ WKH 6WDQIRUG%LQHW ZDV JLYHQ 7KH DFKLHYHPHQW WHVW UHVXOWV ZHUH IRXQG WR DFFXUDWHO\ SUHn GLFW 6% VFRUHV LQ WHUPV RI LWHPV UHVSRQVHV 5HVXOWV VXJn JHVWHG FRQWHQW YDOLGLW\ DFURVV WKH ,4 DQG DFKLHYHPHQW PHDVXUHV $ QXPEHU RI RWKHU UHVHDUFKHUV KDYH LQYHVWLJDWHG WKH UHODWLRQVKLS EHWZHHQ LQWHOOLJHQFH DQG DFKLHYHPHQW PHDVXUHV DQG IRXQG WKHP WR EH SRVLWLYHO\ FRUUHODWHG +DOH +DUWODJH L 6WHHOH 5HVFKOH\ t 5HVFKOH\

PAGE 18

6FKZDUWLQJ t 6FKZDUWLQJ f RU WR SRVVHVV VLJQLILFDQW RYHUODS LQ WKHLU IDFWRU ORDGLQJV &DUUROO 'HDQ *URVVPDQ t -RKQVRQ +RUQ 6WHZDUW t 0RUn ULV 8QGKHLP 9HUQRQ :LNRII f 7KH 7&6 LQ WHUPV RI LWV WLWOH DQG VWDWHG SXUSRVH HQFRPSDVVHV D FRQVWUXFW FORVHO\ UHODWHG WR WKH :,6&5 DQG 6WDQIRUG%LQHW LQ WKDW DOO WKUHH ZHUH GHVLJQHG DQG KDYH EHHQ IRXQG WR SUHGLFW DFDGHPLF DWWDLQPHQW +XUURFNV &7%0F*UDZ+L 7HUPDQ t 0HUULOO :HFKVOHU f $V GHVFULEHG LQ LWV 7HFKQLFDO 5HSRUW WKH 7&6 VXEWHVWV 6HTXHQFHV $QDORJLHV 0HPRU\ DQG 9HUEDO 5HDVRQn LQJf ORDG RQ IDFWRUV FRQVLVWHQW ZLWK WKRVH GHVFULEHG LQ :,6&5 DQG 6WDQIRUG%LQHW VWXGLHV .DXIPDQ f 7KH 7&6 DFFRUGLQJ WR LWV FRQVWUXFWRUV PHDVXUHV D QXPEHU RI FRJQLWLYH DELOLWLHV LQFOXGHG LQ YDULRXV WKHRULHV KRZHYHU OLNH WKH :,6&5 DQG 6WDQIRUG%LQHW HPn SKDVLV LV SODFHG RQ WKH NLQGV RI UHDVRQLQJ DQG UHWHQWLRQ VNLOOV QHFHVVDU\ IRU VFKRRO VXFFHVV &7%0F*UDZ+L S f 7KH &7%6 VLPLODUO\ PHDVXUHV D YDULHW\ RI DFDGHPLF VNLOO DUHDV VKRZQ WR EH SRVLWLYHO\ FRUUHODWHG ZLWK WKH 7&6 7KH 7&6 DQG &7%6 WRJHWKHU DSSHDU WR DVVHVV D WKHRUHWLFDO FRQVWUXFW FRPPRQ WR WKH :,6&5 DQG 6WDQIRUG %LQHW 7KHUHIRUH WKH LWHPV FKRVHQ IURP WKH &7%6 DQG 7&6

PAGE 19

VKRXOG FRQIRUP WR WKDW FRQVWUXFW WKXV DOORZLQJ GLVFULPn LQDWLRQ RI JLIWHG IURP QRWJLIWHG VWXGHQWV DV GR WKH :,6& 5 RU 6WDQIRUG%LQHW 6XSSRUW IRU WKH FRPPRQ LWHP FRQWHQW RI WKH &7%6 7&6 :,6&5 DQG 6WDQIRUG%LQHW ZLOO EH SURYLGHG KHUH E\ LOOXVWUDWLRQV RI DFWXDO LWHPV IRXQG RQ WKHVH WHVWV ,WHPV DUH FDWHJRUL]HG DUELWUDULO\ DFFRUGLQJ WR VXEWHVW FODVVLILn FDWLRQ DQG WR IDFH FRPPRQDOLWLHV $ EULHI GHVFULSWLRQ DQG UHIHUHQFH ORFDWLRQ ZLOO EH JLYHQ IRU WKRVH QRQYHUEDO WHVW LWHPV WKDW FDQQRW EH UHDGLO\ UHSURGXFHG LQ WKLV IRUPDW ,WHPV LQGLFDWHG DUH WKRVH GHVLJQHG IRU DYHUDJH DQG DERYH DYHUDJH VHYHQWK JUDGHUV 7HVW LWHPV DUH SULQWHG LQ EROGn IDFH :,6&5 9RFDEXODU\LQVWUXFWLRQV :KDW GRHV f PHDQ" ULYDOU\ &7%6 9RFDEXODU\ LQVWUXFWLRQV &KRRVH WKH ZRUG RU SKUDVH WKDW PHDQV WKH VDPH DV WKH XQGHUOLQHG ZRUG f WKHLU RSSRQHQW $ IRH % HPSOR\HH & DUFKLWHFW DVVLVWDQW :,6&5 $ULWKPHWLF 7RQ\ ERXJKW D VHFRQG KDQG ELF\FOH IRU +H SDLG RI ZKDW WKH ELF\FOH FRVW QHZ +RZ PXFK GLG LW FRVW QHZ" &7%6 0DWKHPDWLFV &RQFHSWV DQG $SSOLFDWLRQV +RPHUnV UHFLSH ZLOO PDNH VXJDU FRRNLHV +H PDGH RI WKLV UHFLSH IRU D SDUW\ +RZ PDQ\ FRRNLHV ZHUH IRU HDFK RI WKH SHRSOH DW WKH SDUW\"

PAGE 20

6WDQIRUG%LQHW $ULWKPHWLF 5HDVRQLQJ ,I D PDQnV VDODU\ LV D ZHHN DQG KH VSHQGV D ZHHN KRZ ORQJ ZLOO LW WDNH KLP WR VDYH &7%6 0DWKHPDWLFV &RQFHSWV DQG $SSOLFDWLRQV 7R SD\ IRU JURFHULHV 6FRWW 0DUYLQ DQG &DURO HDFK JDYH WKH FOHUN 7KH FOHUN JDYH WKHP LQ FKDQJH +RZ PXFK GLG WKH JURFHULHV FRVW" 3 + 6WDQIRUG%LQHW 9RFDEXODU\ :KDW GRHV %UXQHWWH PHDQ" &7%6 9RFDEXODU\ LQVWUXFWLRQV &KRRVH WKH ZRUG RU SKUDVH WKDW PHDQV WKH VDPH DV WKH XQGHUOLQHG ZRUGf VXFFHVVIXO PHUFKDQW 3 SDUDGH EXVLQHVV + FXVWRPHU VKRSNHHSHU :,6&5 3LFWXUH $UUDQJHPHQW LQVWUXFWLRQV ZDQW \RX WR DUUDQJH WKHVH SLFWXUHV LQ WKH ULJKW RUGHU WR WHOO D VWRU\ WKDW PDNHV VHQVHf 7&6 6HTXHQFH LQVWUXFWLRQV FKRRVH WKH SDUW WKDW ZRXOG FRQWLQXH WKH SDWWHUQ RU VHTXHQFHf 9DULn RXV YLVXDO VWLPXOL DUH SUHVHQWHG VXFK DV OHWWHUV QXPEHUV RU JHRPHWULF VKDSHV :,6&5 6LPLODULWLHV ,Q ZKDW ZD\ DUH D WHOHSKRQH DQG D UDGLR DOLNH" 7&6 9HUEDO 5HDVRQLQJ SDUDSKUDVHG LQVWUXFWLRQV ZRUGV LQ WKH WRS DQG ERWWRP URZV DUH UHODWHG LQ WKH VDPH ZD\ )LQG WKH ZRUG WKDW FRPSOHWHV WKH ERWWRP URZ RI ZRUGVf UDGLR HOHFWULFLW\ PXVLF SDSHU QHZVSDSHU ) LQN VWRU\ + UHSRUWHU W\SHZULWHU 7KH

PAGE 21

6WDQIRUG%LQHW 3UREOHPV RI )DFW $Q ,QGLDQ ZKR KDG FRPH WR WRZQ IRU WKH ILUVW WLPH LQ KLV OLIH VDZ D ER\ ULGLQJ DORQJ WKH VWUHHW $V WKH ER\ URGH E\ WKH ,QGLDQ VDLG 7KH ZKLWH ER\ LV OD]\ KH ZDONV VLWWLQJ GRZQ :KDW ZDV WKH ER\ ULGLQJ RQ WKDW FDXVHG WKH ,Qn GLDQ WR VD\ n+H ZDONV VLWWLQJ GRZQn" 7&6 9HUEDO 5HDVRQLQJ LQVWUXFWLRQV ILQG WKH WUXH VWDWHPHQWf $OO ELF\FOHV KDYH JHDUV 6RPH ELF\FOHV KDYH WHQ VSHHGV 0DULD KDV D ELF\FOH ) 0DULD OLNHV KHU ELF\FOH 0DULDnV ELF\FOH KDV JHDUV + 0DULDnV ELF\FOH JRHV WRR IDVW 0DULDnV ELF\FOH KDV WHQ VSHHGV 6WDQIRUG%LQHW ,QGXFWLRQ 7KLV LV D VHTXHQWLDO WHVW LQ ZKLFK SDSHU LV IROGHG DQG KROHV FXW LQ LW E\ WKH H[DPLQHU 7KH VWXGHQW PXVW GHGXFH D SDWWHUQ WR SUHGLFW KRZ PDQ\ KROHV ZLOO UHVXOW IURP HDFK FXWf 7&6 6HTXHQFHV 6WXGHQWV DUH SUHVHQWHG VHTXHQWLDO SDWWHUQV WKDW DUH LQFRPSOHWH 7KH VWXGHQW PXVW GHGXFH WKH SDWWHUQ DQG SUHGLFW WKH ILQDO SDWWHUQf 1HHG IRU 7KLV 6WXG\ 7HDFKHUV DW WKH VHYHQWKJUDGH OHYHO W\SLFDOO\ H[SHULn HQFH PRUH GLIILFXOW\ LGHQWLI\LQJ SRWHQWLDOO\ JLIWHG VWXn GHQWV WKDQ GR WHDFKHUV DW ORZHU JUDGH OHYHOV 6FKQHOO f &RQVHTXHQWO\ MXQLRU KLJK VFKRRO WHDFKHUV UHIHU D ODUJHU SURSRUWLRQ RI QRWJLIWHG VWXGHQWV IRU HYDOXDWLRQ WKDQ GR WHDFKHUV RI HOHPHQWDU\ VFKRRO VWXGHQWV 7KLV SKHQRPHQRQ LV EHOLHYHG WR UHVXOW IURP PRUH OLPLWHG FRQWDFW EHWZHHQ LQGLYLGXDO WHDFKHUV DQG VWXGHQWV DQG IURP WKH IDFW WKDW WKH SRRO RI SRWHQWLDOO\ JLIWHG VWXGHQWV IURP ZKLFK

PAGE 22

MXQLRU KLJK VFKRRO WHDFKHUV FKRRVH FRQWDLQV QRQH RI WKH VWXGHQWV ZKR KDYH EHHQ LGHQWLILHG DV JLIWHG GXULQJ SUHYLRXV \HDUV -XQLRU KLJK VFKRRO WHDFKHUV PXVW EDVH WKHLU UHIHUUDOV RQ WKH EHVW RI WKH UHVW &OHDUO\ WKH PRVW DGYDQWDJHRXV WLPH WR LGHQWLI\ VWXGHQWV IRU D MXQLRU KLJK VFKRRO FXUULFXOXP IRU WKH JLIWHG LV ZKHQ WKH\ HQWHU VHYHQWK JUGGH 7KH QHZ SURFHGXUH 13f ZDV GHYHORSHG E\ VHOHFWLQJ LWHPV IURP SRSXODU JURXS WHVWV DGPLQLVWHUHG DW VHYHQWK JUDGH PHDVXULQJ ERWK DFKLHYHPHQW DQG FRJQLWLYH DELOLW\ 7KH UDWLRQDOH IRU FKRRVLQJ VFUHHQLQJ LWHPV IURP VWDQGDUn GL]HG JURXS WHVWV LV WKDW WKLV DSSURDFK WR VWXGHQW VFUHHQn LQJ LV ERWK WLPH DQG FRVWHIILFLHQW $OO VWXGHQWV LQ WKH SRSXODWLRQ VFKRRO GLVWULFW WDNH WKH &7%6 DQG 7&6 DQQXDOO\ DQG WKH VFUHHQLQJ GDWD DUH UHDGLO\ DYDLODEOH ZLWKRXW DGGLn WLRQDO WHVWV RU WHVWLQJ WLPH EHLQJ QHHGHG ,W ZDV EHOLHYHG WKDW RI WKH &7%6 DQG 7&6 TXHVn WLRQV W\SLFDOO\ DGPLQLVWHUHG WR VHYHQWK JUDGHUV WKHUH H[LVWHG D VXEVHW RI LWHPV WKDW ZRXOG DFFXUDWHO\ SUHGLFW JLIWHG ,4 RQ WKH :,6&5 RU 6% %HFDXVH WKH QXPEHU RI &7%67&6 LWHPV LV ODUJH VDPSOLQJ RI D ZLGH UDQJH RI VNLOOV DQG DELOLWLHV LV SRVVLEOH $V WKH GLYHUVLW\ RI WKH LWHPV LQFUHDVHV VR GRHV WKH JHQHUDO DELOLW\ PHDVXUHG E\ WKH WRWDO WHVW .DXIPDQ f

PAGE 23

7KH XWLOLW\ RI ILQGLQJ D VPDOO SRRO RI LWHPV WKDW FRUUHODWHV KLJKO\ ZLWK JLIWHG LQWHOOLJHQFH LV WKDW Df FXUUHQW IRUPV RI WKH &7%67&6 VHOHFWHG LWHPV PD\ EH UHn WDLQHG IRU JURXS DGPLQLVWUDWLRQ WR IXWXUH JLIWHG FODVV FDQGLGDWHV Ef LI D JLIWHG FDQGLGDWH LV QRW SUHVHQW IRU &7%67&6 WHVWLQJ WKH HQWLUH WHVW ZLOO QRW KDYH WR EH DGPLQLVWHUHG DV D JLIWHG VFUHHQLQJ DQG Ff WKH SURFHGXUH RI DQDO\]LQJ D WHVW LQ WKLV ZD\ PLJKW EH XVHIXO IRU SUHn GLFWLQJ LQWHOOLJHQFH RU RWKHU WUDLWVf DPRQJ RWKHU SRSXODn WLRQV XWLOL]LQJ WKHVH RU RWKHU WHVWV 2YHUYLHZ RI WKH 5HPDLQGHU RI WKH 6WXG\ 7KH VXEVHTXHQW FRQWHQW RI WKLV VWXG\ LV GLYLGHG LQWR IRXU FKDSWHUV ,Q &KDSWHU ,, D UHYLHZ RI UHODWHG OLWHUn DWXUH LV SUHVHQWHG $ GHVFULSWLRQ RI WKH PHWKRGRORJ\ XVHG IRU WKH UHVHDUFK FRPSULVHV &KDSWHU ,,, 5HVHDUFK UHVXOWV DUH SUHVHQWHG LQ &KDSWHU ,9 DQG UHVXOWV DUH GLVFXVVHG LQ &KDSWHU 9

PAGE 24

&+$37(5 ,, /,7(5$785( 5(9,(: 6XSSRUW IRU WKH 3UREOHP 7KH SV\FKRPHWULF VFUHHQLQJ RI JLIWHG LQWHOOLJHQFH LV EHVHW E\ SUREOHPV RI QRW RQO\ WLPH DQG FRVWHIILFLHQF\ EXW RI SUHGLFWLYH DFFXUDF\ ,Q WKH UHOHYDQW OLWHUDWXUH LW LV VXJJHVWHG WKDW WKHVH SUREOHPV H[LVW IRU D YDULHW\ RI VFUHHQLQJ PHWKRGV DQG SURFHGXUHV 7KHVH SUREOHPV DUH QRW RI UHFHQW RULJLQ KRZHYHU $V HDUO\ DV 3HJQDWR DQG %LUFK IRXQG WKDW VXIILFLHQW SV\FKRORJLFDO VHUYLFHV ZHUH UDUHO\ DYDLODEOH WR WHVW DOO RI WKH JLIWHG FODVV FDQGLn GDWHV WKHUHE\ QHFHVVLWDWLQJ SURFHGXUHV IRU VFUHHQLQJ SULRU WR IRUPDO WHVWLQJ $FFRUGLQJO\ WKH DXWKRUV FRQn GXFWHG DQ LQYHVWLJDWLRQ RI WKH UHODWLYH HIILFLHQF\ DQG HIIHFWLYHQHVV S f RI VHYHQ SURFHGXUHV IRU ORFDWLQJ JLIWHG FKLOGUHQ LQ MXQLRU KLJK VFKRROV WHDFKHU UDWLQJV FODVV UDQN FUHDWLYH DELOLW\ LQ DUW RU PXVLF VWXGHQW FRXQFLO PHPEHUVKLS VXSHULRULW\ LQ PDWKHPDWLFV JURXS DFKLHYHPHQW DQG JURXS ,4 6HYHQ KXQGUHG HLJKW\RQH PHWn URSROLWDQ VFKRRO GLVWULFW VWXGHQWV ZHUH VHOHFWHG IRU SDUn WLFLSDWLRQ LQ WKH VWXG\ RQ WKH EDVLV RI KLJK UDWLQJV LQ RQH RU PRUH RI WKH VHYHQ FDWHJRULHV $OO RI WKH SDUWLFn LSDQWV UHFHLYHG WKH 6WDQIRUG%LQHW 6FRUHV RQ WKLV

PAGE 25

LQWHOOLJHQFH WHVW ZHUH XVHG DV D FULWHULRQ UHIHUHQFH IRU WKH GHVLJQDWLRQ RI WKRVH FKLOGUHQ ZKR ZHUH LQGHHG JLIWHG )RU HDFK RI WKH VHYHQ VFUHHQLQJ SURFHGXUHV HIIHFWLYHQHVV ZDV MXGJHG E\ WKH SHUFHQWDJH RI JLIWHG FKLOn GUHQ ORFDWHG HIILFLHQF\ ZDV GHILQHG DV WKH UDWLR EHWZHHQ WRWDO QXPEHU RI JLIWHG VWXGHQWV DQG VWXGHQWV SUHGLFWHG DV JLIWHG 2I WKH VXEMHFWV n RL WKH VFKRRO SRSXn ODWLRQf ZHUH MXGJHG WR EH JLIWHG 7KHVH UHVXOWV LQGLFDWHG WKDW DPRQJ WKH VHYHQ PHWKRGV JURXS ,4 DQG DFKLHYHPHQW WHVWV ZHUH WKH EHWWHU SUHGLFWRUV SURYLGLQJ WKH EHVW SRVn VLEOH FRPELQDWLRQ RI HIIHFWLYHQHVV ZLWK HIILFLHQF\ 2WKHU PHWKRGV VXFK DV KRQRU UROH LQFOXVLRQ ZHUH IDLUO\ HIIHFn WLYH EXW WKHLU HIILFLHQF\ ZDV SRRU (YHQ WKRXJK 3HJQDWR DQG %LUFK f ZHUH ODUJHO\ XQn VXFFHVVIXO LQ ILQGLQJ DQ HIIHFWLYH SUHGLFWRU RI JLIWHG ,4 D VXEVWDQWLDO DPRXQW RI UHVHDUFK KDV IRFXVHG RQ VFUHHQLQJ WKH JLIWHG VLQFH WKDW WLPH 7KH IROORZLQJ VHFWLRQV RI WKLV OLWHUDWXUH UHYLHZ DUH FRQFHQWUDWHG RQ IRXU SURFHGXUHV IRU JLIWHG VWXGHQW LGHQWLILFDWLRQ WKDW LQYROYH JURXS ,4 WHVWV ,4 VKRUW IRUPV ,4 VFUHHQLQJ WHVWV DQG DFKLHYHPHQW WHVW VFRUHV *URXS ,4 7HVWV 7KH ODUJHO\ XQVXFFHVVIXO DWWHPSW E\ 3HJQDWR DQG %LUFK f WR SUHGLFW JLIWHG LQWHOOLJHQFH XVLQJ JURXS ,4 WHVWV ZDV IROORZHG RQH \HDU ODWHU E\ D VLPLODU VWXG\ &KDPEHUV

PAGE 26

f VRXJKW D VFUHHQLQJ LQVWUXPHQW IRU XVH LQ D 0LFKLJDQ VFKRRO GLVWULFW 8VLQJ WKH ,3$7 &DWWHOOnV WHVW RI JHQn HUDO LQWHOOLJHQFHf WKH &DOLIRUQLD 7HVW RI 0HQWDO 0DWXn ULW\ WKH 65$ 3ULPDU\ $ELOLWLHV 7HVW 7KH .XKOPDQ$QGHUVRQ ,QWHOOLJHQFH 7HVW DQG WKH :,6& &KDPEHUV WHVWHG FKLOn GUHQ LQ JUDGHV WKUHH WKURXJK VL[ )RU HDFK VFUHHQLQJ WHVW D FXWRII ZDV FDOFXODWHG DERYH ZKLFK DOO JLIWHG VWXn GHQWV :,6& ,4!f ZRXOG EH LGHQWLILHG 7KH DFFXUDF\ RI HDFK VFUHHQLQJ SURFHGXUH ZDV HVWDEOLVKHG DW b DQG WKH HIILFLHQF\ ZDV WKHQ GHWHUPLQHG EDVHG RQ WKH QXPEHU RI QRW JLIWHG VWXGHQWV PLVFODVVLILHG E\ WKH VFUHHQLQJ SURFHGXUH QV JLIWHG 7KH UHVXOWV UHYHDOHG WKDW WKH 6.$ WHVW DQG WKH .XKPDQ$QGHUVRQ FRXOG EH UDQNHG UHVSHFWLYHO\ DV WKH PRVW DQG OHDVW HIILFLHQW DQG WKDW EHWZHHQ b DQG b RI WKH VWXGHQWV SUHGLFWHG DV JLIWHG ZHUH QRW 7KUHH \HDUV DIWHU &KDPEHUVn VWXG\ %ORVVHU f WHVWHG QLQWK JUDGHUV RQ WKH +HQPRQ1H VRQ DQG 2WLV JURXS LQWHOOLJHQFH WHVWV 7KH UHVHDUFK VDPSOH KDG D PHDQ ,4 RI RQ WKH 6WDQIRUG%LQHW ZLWK D UDQJH RI WR 7KH UHVXOWV LQGLFDWHG WKDW RI WKH VWXGHQWV SUHGLFWHG DV JLIWHG E\ WKH 2WLV RQO\ bf ZHUH LGHQWLILHG DV VXFK E\ WKH 6WDQIRUG%LQHW 2Q WKH +HQPRQ1HOVRQ RI VWXn GHQWV bf ZHUH FRUUHFWO\ SUHGLFWHG DV JLIWHG %HFDXVH b RI WKH JLIWHG VWXGHQWV ZHUH QRW LGHQWLILHG E\ HLWKHU JURXS WHVW ERWK WHVWV SURYHG WR EH SRRU SUHGLFWRUV RI JLIWHGQHVV

PAGE 27

0RUH UHFHQWO\ +DUULQJWRQ f DOVR IRXQG WKDW JURXS ,4 WHVWV WHQG WR XQGHUHVWLPDWH WKH ,4V RI PDQ\ LQn WHOOHFWXDOO\ JLIWHG VWXGHQWV $FFRUGLQJ WR +DUULQJWRQ IRU HYHU\ VWXGHQW LGHQWLILHG DV JLIWHG RQ D JURXS ,4 WHVW RQH JLIWHG FKLOG LV QRW UHIHUUHG +DUULQJWRQ VXJJHVWHG WKDW WKH KLJKHU WKH DELOLW\ OHYHO WKH JUHDWHU WKH GLVn FUHSDQF\ EHWZHHQ LQGLYLGXDO DQG JURXS ,4 VFRUHV +H DOVR IRXQG WKDW D FKLOGnV ,4 PD\ YDU\ E\ DV PXFK DV SRLQWV EHWZHHQ JURXS DQG LQGLYLGXDO WHVWV IXUWKHU EHFDXVH WKHUH PD\ EH D YHU\ VPDOO QXPEHU RI LWHPV DW WKH JUHDWHU GLIILFXOW\ OHYHOV RQ JURXS WHVWV D FKLOG PD\ KDYH WR SHUn IRUP SHUIHFWO\ WR EH SUHGLFWHG DV JLIWHG ,QWHOOLJHQFH 4XRWLHQW 6KRUW )RUPV 6RFDOOHG ,4 VKRUW IRUPV DUH FRPSULVHG RI DEEUHYLDWHG YHUVLRQV RI LQGLYLGXDOO\ DGPLQLVWHUHG VWDQGDUGL]HG LQWHOn OLJHQFH WHVWV 7\SLFDOO\ VHOHFWHG IRU VKRUW IRUPV DUH VXEWHVWV RI WKH :,6&5 RU LWHPV IURP WKH 6WDQIRUG%LQHW 6KRUW IRUPV RI WKH :HFKVOHU 6FDOHV DQG 6WDQIRUG%LQHW KDYH EHHQ VWXGLHG H[WHQVLYHO\ %LUFK &DUOHWRQ t 6WDFH\ (QEXUJ 5RZOH\ t 6WRQH )LQGOH\ t 7KRPSVRQ *URVVPDQ t *DOYLQ 0HLVWHU t .XUNR 1LFKROV 6LPSVRQ t %ULGJHV :ULJKW t 6DQGU\ :DGH 3KHOSV t )DODVFR
PAGE 28

VWXGLHG DV D PHWKRG IRU VFUHHQLQJ SRWHQWLDOO\ JLIWHG VWXn GHQWV $Q HDUO\ DWWHPSW WR SUHGLFW JLIWHG ,4 XVLQJ D VKRUW IRUP WHVW ZDV FRQGXFWHG E\ 7KRPSVRQ DQG )LQGOH\ LQ )LQGLQJ WKDW WKH 6LPLODULWLHV 6f ,QIRUPDWLRQ ,f 3LFn WXUH $UUDQJHPHQW 3$f %ORFN 'HVLJQ %'f DQG 3LFWXUH &RPn SOHWLRQ 3&f :,6& VXEWHVWV FRXOG EH HIIHFWLYHO\ XVHG IRU WKLV SXUSRVH 7KRPSVRQ DQG )LQGOH\ SXEOLVKHG WKH &DOLIRUn QLD $EEUHYLDWHG :,6& IRU WKH ,QWHOOHFWXDOO\ *LIWHG &$: ,4f LQ f ,Q WKHLU VWXG\ .LOOLDQ DQG +XJKHV f PHDVXUHG WKH HIIHFWLYHQHVV RI WKH /\PDQ VKRUW IRUP /RUU t 0HLVWHU f DQG WKH 9RFDEXODU\%ORFN 'HVLJQ VXEWHVWV RI WKH :,6&5 G\DG IRU SUHGLFWLQJ ,4 RQ WKH 6WDQIRUG%LQHW DQG :,6&5 UHVSHFWLYHO\ 6XEMHFWV ZHUH VWXGHQWV EHWZHHQ DQG \HDUVROG SRVVHVVLQJ D PHDQ ,4 RI 5HVXOWV LQGLFDWHG D FRUUHODWLRQ RI U EHWZHHQ WKH :,6&5 DQG 9 %' G\DG ZKHUHDV WKH 6WDQIRUG%LQHW DQG /\PDQ VFRUHV ZHUH FRUUHODWHG DW U .LOOLDQ DQG +XJKHV GLG QRW SUHVHQW UHVXOWV RI WKH DFWXDO QXPEHU RI VWXGHQWV FRUUHFWO\ SUHn GLFWHG DV JLIWHG 7KH\ GLG KRZHYHU LQGLFDWH WKDW b RI WKH VWXGHQWV KDG VKRUW IRULQ3fXOO 6FDOH ,4 VFRUH GLVFUHSn DQFLHV RI SRLQWV RU PRUH (PSOR\LQJ D PXFK ODUJHU VDPSOH RI VWXGHQWV WKDQ GLG SUHYLRXV UHVHDUFKHUV .DUQHV DQG %URZQ f XVHG

PAGE 29

6LYHUVWHLQnV f PHWKRG RI GHULYLQJ WKH EHVW VKRUW IRUP FRPELQDWLRQV S f WR REWDLQ DQ DFFXUDWH JLIWHG ,4 SUHGLFWRU 6LOYHUVWHLQnV PHWKRG WDNHV VXEWHVW XQUHOLn DELOLW\ LQWR DFFRXQW ZKHQ PHDVXULQJ SUHGLFWLYH DELOLW\ 1LQH KXQGUHG IRUW\VL[ JLIWHG FKLOGUHQ DJHV WR ; FKURQRORJLFDO DJH >&$@ f VHUYHG DV VXEMHFWV .DUQHV DQG %URZQ IRXQG WKDW WKH :,6& %ORFN 'HVLJQ VXEWHVW ZDV UHSUHVHQWHG IUHTXHQWO\ LQ VXEWHVW FRPELQDWLRQV WKDW FRUUHODWHG ZLWK :,6& )XOO 6FDOH ,4 6XSSRUWLQJ .LOOLDQ DQG +XJKHVn ILQGLQJV WKH 9%' G\DG ZDV IRXQG WR EH WKH PRVW DFFXUDWH IRU SUHGLFWLQJ JLIWHG ,4 7KH XVH RI VXEn WHVW WHWUDGV ZDV IRXQG WR EH XVHIXO LQFUHDVLQJ FRUUHODn WLRQ FRHIILFLHQWV IURP WR $JDLQ DFWXDO DFFXn UDF\ UDWLRV ZHUH QRW SURYLGHG LQ WKH VWXG\ 3URFHHGLQJ XQGHU WKH QRWLRQ WKDW VLQFH D VKRUW IRUP ,4 WHVW LV FRPSRVHG HQWLUHO\ RI VRPH VXEVHW RI TXHVWLRQV RI LWHPV WDNHQ GLUHFWO\ IURP D IXOOOHQJWK ,4 WHVW D VKRUW IRUP ZRXOG VHHP WR EH DQ LGHDO SUHGLFWRU RI IXOO OHQJWK ,4 WHVW SHUIRUPDQFH S f 'LUNV :HVVHOV 4XD IRUWK DQG 4XHQRQ f FRPSDUHG YDULRXV VKRUW IRUP FRPn ELQDWLRQV ZLWK )XOO 6FDOH ,4 RQ WKH :,6&5 6XEMHFWV FRQn VLVWHG RI IRXUWK JUDGHUV ZLWK D PHDQ ,4 RI UDQJH WR f 7ZHOYH :,6&5 VXEWHVW FRPELQDWLRQV ZHUH VWXGLHG ,W ZDV UHYHDOHG WKDW WKH VKRUW IRUP FRPELQDWLRQV RI 6LPLODULWLHV 2EMHFW $VVHPEO\ DQG 9RFDEXODU\ DQG 62$

PAGE 30

ZHUH HDFK JRRG SUHGLFWRUV RI )XOO 6FDOH ,4 $OWKRXJK FRUn UHODWLRQV RQ WKH %' VXEWHVW ZHUH KLJK DV VKRZQ LQ SUHYLn RXV VWXGLHV WKH\ WHQGHG WR SUHGLFW DQ H[FHVVLYH QXPEHU RI QRQJLIWHG VWXGHQWV DV JLIWHG 7KH 6a2$ G\DG SUHGLFWHG RI JLIWHG VWXGHQWV DQG ZKR ZHUH QRW 7KH 62$9 WULDG SUHGLFWHG RI JLIWHG VWXGHQWV DQG ZKR ZHUH QRW 8WLOL]LQJ WKH VWXGLHV E\ .LOOLDQ DQG +XJKHV f DQG 'LUNV HW DO f ZKR QRWHG WKDW 9%' DQG 62$ G\DGV UHVSHFWLYHO\ ZHUH WKH PRVW HIIHFWLYH LQ SUHGLFWLQJ )XOO 6FDOH ,4 )HOO DQG )HOO f HYDOXDWHG :,6&5 SURWRFROV RI FKLOGUHQ SUHYLRXVO\ HYDOXDWHG DV JLIWHG SURn JUDP FDQGLGDWRV 7KH VWXGHQWV UDQJHG LQ DJH IURP WR ; DJH f DQG SRVVHVVHG )XOO 6FDOH ,4V RI RU JUHDWHU (OHYHQ VXEWHVW G\DGV ZHUH VWXGLHG LQ WHUPV RI IUHTXHQF\ ZLWK ZKLFK HDFK SURGXFHG DQ HVWLPDWHG ,4 *UHDWHVW SUHGLFWLYH DFFXUDF\ ZDV DFKLHYHG XVLQJ WKH 69 DQG 62$ G\DGV 7KHVH FRUUHFWO\ SUHGLFWHG b DV JLIWHG 7KH ,%' G\DG \LHOGHG SUHGLFWLRQ UDWLQJV RI RQO\ b 1RW SURYLGLQJ DQ H[DFW QXPEHU WKH DXWKRUV LQGLFDWHG WKDW VRPH JLIWHG FKLOGUHQ ZHUH RYHUORRNHG :KLOH UHVXOWV DUH FRQn VLVWHQW ZLWK ILQGLQJV E\ 'LUNV HW DO LQGLFDWLQJ WKDW WKH 62$ G\DG LV PRVW HIIHFWLYH SUHGLFWLRQ DFFXUDF\ ZDV PXFK ORZHU LQ WKLV VWXG\ ,Q D IDLUO\ UHFHQW VWXG\ .UDPHU 0DUNOH\ 6KDQNV DQG 5\DELN f XWLOL]HG 7KRPSVRQ DQG )LQGOH\nV f

PAGE 31

&$:,4 RQ D VDPSOH RI FKLOGUHQ DJHV WR ; DJH f $OO VXEMHFWV UHFHLYHG WKH :,6&5 DQG DOO ZHUH DQDO\]HG LQ WHUPV RI WKH 6 3$ %' DQG 3& VXEWHVW SDWn WHUQ 2I WKH VWXGHQWV SUHGLFWHG DV JLIWHG ZHUH SUHGLFWHG DFFXUDWHO\ RI VWXGHQWV SUHGLFWHG QRW WR EH JLIWHG ZHUH FRUUHFWO\ GHVFULEHG 7KLV VXEWHVW VKRUW IRUP ZDV FRQVLGHUHG WR EH D UHODWLYHO\ DFFXUDWH SUHGLFWRU RI JLIWHG ,4 ,QWHOOLJHQFH 4XRWLHQW 6FUHHQLQJ 7HVWV ,QGLYLGXDOO\ DGPLQLVWHUHG LQWHOOLJHQFH WHVWV GHVLJQHG WR HVWLPDWH PHQWDO DELOLW\ XVXDOO\ LQ PLQXWHV RU OHVV KDYH EHFRPH D ZLGHO\ XVHG SURFHGXUH IRU VFUHHQLQJ JLIWHG LQWHOOLJHQFH 7KH 6ORVVRQ ,QWHOOLJHQFH 7HVW 6ORVVRQ t -HQVHQ f LV RQH VXFK VFUHHQLQJ WHVW ,W KDG EHHQ DGRSWHG LQ WKH +LOOVERURXJK &RXQW\ )ORULGD VFKRRO GLVWULFW IRU WKH SXUSRVH RI VFUHHQLQJ WKH JLIWHG +LJK FRUUHODn WLRQV EHWZHHQ WKH 6,7 DQG WKH :,6&5 RU 6WDQIRUG%LQHW KDYH EHHQ UHSRUWHG LQ UHVHDUFK ILQGLQJV /DZUHQFH t $QGHUn VRQ 0DUWLQ t .LGZHOO 0DUWLQ t 5XGROSK 0L]H 6PLWK t &DOODZD\ 5LWWHU 'XII\ t )LVFKPDQ 6ORVVRQ t -HQVHQ 6WHZDUW t -RQHV f +RZHYHU WKH IHZ DYDLODEOH VWXGLHV FRQGXFWHG RQ JLIWHG VDPSOHV KDYH QRW VXSSRUWHG WKH XVH RI WKH 6,7 IRU VFUHHQn LQJ

PAGE 32

,Q WKH SUHYLRXVO\ GLVFXVVHG VWXG\ VHH &KDSWHU ,f E\ *URVVPDQ DQG -RKQVRQ f WKH 2WLV /HQQRQ JURXS ,4 WHVW ZDV IRXQG WR EH D EHWWHU SUHGLFWRU RI JLIWHG DFKLHYHPHQW WKDQ WKH 6,7 IRU KLJK DFKLHYLQJ VWXGHQWV 'LUNV :HVVHOV 4XDIRUWK DQG 4XHQRQ f DOVR IRXQG WKH 6,7 WR EH D SRRU SUHGLFWRU RI JLIWHG DELOLW\ 7KH\ DGPLQLVWHUHG WKH 6,7 WR DFDGHPLFDOO\ WDOHQWHG IRXUWK JUDGHUV 7KH VWXn GHQWV ZHUH DOVR DGPLQLVWHUHG WKH :,6&5 WR GHWHUPLQH WKHLU DFWXDO ,4 VFRUHV 2I WKH VWXGHQWV ZKR ZHUH IRXQG WR SRVVHVV JLIWHG LQWHOOLJHQFH ,4!f RQO\ ZHUH LGHQWLn ILHG DV VXFK E\ WKH 6,7 ,Q DGGLWLRQ WKH 6,7 IDOVHO\ SUHGLFWHG JLIWHG LQWHOOLJHQFH LQ RI WKH QRQJLIWHG FKLOGUHQ 7KH UHVHDUFKHUV FRQFOXGHG WKDW WKH 6,7 DORQH VKRXOG QRW EH XVHG WR SUHGLFW ,4V RI JLIWHG FKLOGUHQ 7KH 6,7 KDV DOVR EHHQ IRXQG WR VLJQLILFDQWO\ RYHUHVWLPDWH ,4 VFRUHV 0DFKHQ f LQYHVWLJDWHG WKH UHOLDELOLW\ DQG FRQFXUUHQW YDOLGLW\ RI WKH 6,7 ZLWK WKH :,6& XVLQJ JLIWHG FKLOGUHQ DJHV WKURXJK 7KH UHVXOWV UHYHDOHG D VLJQLILFDQW FRUUHODWLRQ EHWZHHQ WKH WZR WHVWV WKRXJK WKH 6,7 WHQGHG WR RYHUHVWLPDWH WKH :,6& E\ DW OHDVW RQH VWDQn GDUG GHYLDWLRQ $GGLWLRQDOO\ WKH 6,7 KDV EHHQ VKRZQ WR XQGHUHVWLPDWH ,4 VFRUHV 0L]H HW DO f IRXQG LQ WKHLU VWXG\ RI VWXGHQWV IURP DOO JUDGH OHYHOV WKDW RI VWXGHQWV ZLWK DERYH DYHUDJH LQWHOOLJHQFH b ZHUH RYHUHVn WLPDWHG DQG b ZHUH XQGHUHVWLPDWHG E\ RU PRUH ,4 SRLQWV RQ WKH 6,7

PAGE 33

,Q .DUQHV DQG %URZQ IXUWKHU H[DPLQHG WKH WHQn GHQF\ RI WKH 6,7 WR RYHU RU XQGHUHVWLPDWH ,4 VFRUHV ,Q WKLV VWXG\ WKH YDOLGLW\ RI WKH 6,7 LQ UHODWLRQ WR WKH :,6&5 ZDV DVVHVVHG IRU D JURXS RI JLIWHG FKLOGUHQ DJHV WKURXJK $ 6,7:,6&5 FRUUHODWLRQ RI U ZDV FDOn FXODWHG WKLV FRHIILFLHQW ZDV VLJQLILFDQW DW WKH OHYHO 7KH DXWKRUV DOVR FRPSXWHG D UHJUHVVLRQ HTXDWLRQ ZLWK ZKLFK WR SUHGLFW :,6&5 ,4 IURP WKH 6,7 5HVXOWV LQGLFDWHG WKDW DW WKH ORZHU UDQJHV RI 6,7 VFRUHV LW WHQGHG WR XQGHUHVWLPDWH WKH :,6&5 ZKLOH DW WKH XSSHU UDQJHV ,4 ZDV RYHUHVWLPDWHG 'HVSLWH WKH KLJK FRUUHODWLRQ EHWZHHQ WKH WZR WHVWV LW LV DSSDUHQW WKDW .DUQHV DQG %URZQ ZHUH QRW FRQILGHQW LQ WKH 6,7nV SUHGLFWLYH DELOLW\ EHFDXVH WKH\ UHFRPPHQGHG XVLQJ WZR 6(0V IRU WKH 6,7 ZKHQ VFUHHQLQJ JLIWHG ,4 WR HQVXUH WKDW PRVW JLIWHG VWXGHQWV DUH LGHQWLILHG 7R REWDLQ b DFFXUDF\ D FXWRII VFRUH RI ZRXOG KDYH EHHQ QHFHVVDU\ +RZHYHU .DUQHV DQG %URZQ GLG QRW LQGLFDWH KRZ PDQ\ QRQJLIWHG VWXGHQWV ZRXOG KDYH EHHQ SUHGLFWHG DV JLIWHG XVLQJ D FXWRII WKLV ORZ ,Q D VLPLODU VWXG\ SUHVHQWHG DW WKH $QQXDO 0HHWLQJ RI WKH $ODEDPD $VVRFLDWLRQ RI 6FKRRO 3V\FKRORJLVWV LQ $SSOH GLVFXVVHG WKH SUHFLVLRQ RI WKH 6,7 LQ SUHn GLFWLQJ :,6&5 ,4 RI JLIWHG VWXGHQWV DJHV WR 'LIIHUHQFHV LQ VFRUHV ZHUH FRPSDUHG E\ XVH RI LQGHSHQGHQW WWHVWV 7KH UHVXOWV VXSSRUWHG WKH ILQGLQJV RI .DUQHV DQG

PAGE 34

%URZQ WKDW DW WKH ORZHU 6,7 UDQJHV :,6&5 ,4V ZHUH XQGHUn HVWLPDWHG DQG WKDW DW WKH XSSHU 6,7 UDQJHV WKH :,6&5 ,4V ZHUH RYHUHVWLPDWHG $SSOH FRQFOXGHG WKDW YDOXDEOH GLDJn QRVWLF LQIRUPDWLRQ \LHOGLQJ D TXDOLWDWLYH SLFWXUH RI WKH FKLOGnV VWUHQJWK LV RPLWWHG ZKHQ WKH 6,7 DORQH LV XVHG DV D VFUHHQLQJ LQGLFDWRU :KHUHDV .DUQHV DQG %URZQ DV ZHOO DV $SSOH FRPSDUHG 6,7 DQG :,6&5 VFRUHV IRU \RXQJVWHUV DOUHDG\ SODFHG LQ JLIWHG FODVVHV 5XVW DQG /RVH f DWWHPSWHG WR DFFXn UDWHO\ VFUHHQ SRWHQWLDOO\ JLIWHG VWXGHQWV LQ ILUVW WKURXJK VHYHQWK JUDGH %DVHG RQ WHDFKHU UHIHUUDOV DQG 6,7 VFRUHV RI RU DERYH VWXGHQWV ZHUH IRXQG HOLJLEOH IRU :,6&5 HYDOXDWLRQ 2I WKHVH ZHUH XWLOL]HG LQ WKH UHn VHDUFK VDPSOH $FFRUGLQJ WR VWHSZLVH UHJUHVVLRQ HTXDn WLRQV WKH 6,7 ZDV IRXQG WR EH D VLJQLILFDQW SUHGLFWRU RI )XOO 6FDOH ,4 +RZHYHU RI WKH VWXGHQWV SUHGLFWHG RQO\ DFKLHYHG :,6&5 ,4V RI RU DERYH 7KXV VHWn WLQJ WKH 6,7 FXWRII DW IDLOHG WR VFUHHQ RXW b RI WKH QRQJLIWHG VWXGHQWV ,I D FXWRII RI KDG EHHQ XVHG DV VXJJHVWHG E\ .DUQHV DQG %URZQ HYDOXDWLRQV ZRXOG KDYH EHHQ HOLPLQDWHG +RZHYHU RI WKRVH ZRXOG KDYH EHHQ JLIWHG .DUQHV DQG %URZQ QRWHG WKDW ZKLOH WKHUH ZDV D KLJK FRUUHODWLRQ EHWZHHQ WKH 6,7 DQG :,6&5 WKHUH ZDV D JUHDW GHDO RI YDULDELOLW\ ZLWK LQGLYLGXDO FDVHV ,W ZDV FRQFOXGHG WKDW LQ DOO VWXGLHV KLJK HUURU FDQ EH H[SHFWHG

PAGE 35

ZKHQ WKH 6,7 LV XVHG WR SUHGLFW :,6&5 ,4 DPRQJ WKH JLIWHG 2WKHU ,4 VFUHHQLQJ WHVWV VXFK DV *XLOIRUGnV 6WUXFWXUH RI WKH ,QWHOOHFW 7HVW 62,f 3HDUFH 6WHQVRQ f 5DYHQV 3URJUHVVLYH 0DWUL[HV 3HDUFH 3HWW\ t )LHOG f 7KH 3HDERG\ 3LFWXUH 9RFDEXODU\ 7HVW 0L]H HW DO 3HGULDQD t %UDFNHQ f DQG 7KH $PPRQV 4XLFN 7HVW -RHVWLQJ t -RHVWLQJ .HQGDOO t /LWWOH 1LFKn ROVRQ f KDYH EHHQ FRUUHODWHG ZLWK WKH :,6& :,6&5 DQG WKH 6WDQIRUG%LQHW ,Q VRPH LQVWDQFHV VLJQLILFDQW FRUUHODWLRQV KDYH EHHQ IRXQG +RZHYHU YHU\ IHZ VWXGLHV KDYH EHHQ FRQGXFWHG ZLWK VDPSOHV RI JLIWHG VWXGHQWV ,Q RQH VXFK VWXG\ 'H)LOLSSLV DQG )XOPHU f IRXQG WKDW WKH $PPRQV 4XLFN 7HVW XQGHUHVWLPDWHG :,6&5 ,4 IRU ILUVW IRXUWK DQG VHYHQWK JUDGHUV ZLWK KLJK DELOLW\ ,Q DQRWKHU VWXG\ LQYROYLQJ VDPSOHV RI JLIWHG VWXn GHQWV :ULJKW f FRUUHODWHG :,6& DQG 3HDERG\ 3LFWXUH 9RFDEXODU\ 7HVW 3397f VFRUHV RI VWXGHQWV UHIHUUHG E\ WHDFKHUV IRU JLIWHG SURJUDP WHVWLQJ $ FRUUHODWLRQ RI U ZDV FDOFXODWHG DQG LW ZDV IRXQG WKDW QHDUO\ KDOI RI WKRVH ZKR VFRUHG WZR VWDQGDUG GHYLDWLRQV DERYH WKH 3397 PHDQ ZHUH QRW HOLJLEOH IRU JLIWHG SURJUDP SODFHPHQW EDVHG RQ :,6&5 ,4 VFRUHV :ULJKW UHFRPPHQGHG WKDW WKH 3397 QRW EH XVHG WR VFUHHQ JLIWHG SURJUDP FDQGLGDWHV $ WKLUG VWXG\ ZDV FRQGXFWHG E\ 6WHQVRQ f WR GHn WHUPLQH WKH FRQFXUUHQW YDOLGLW\ RI WKH 6WUXFWXUH RI

PAGE 36

,QWHOOHFW 62,f *LIWHG 6FUHHQHU ZLWK WKH :,6&5 7KH VXEn MHFWV ZHUH HOHPHQWDU\ VFKRRO VWXGHQWV $ PXOWLSOH FRUUHODWLRQ RI U ZDV VLJQLILFDQW DW KRZHYHU RQO\ b RI WKH YDULDQFH LQ :,6&5 VFRUHV ZDV H[SODLQHG E\ WKH *LIWHG 6FUHHQHU 1R SUHGLFWRU YDULDEOH FRQWULEXWHG WR D VLJQLILFDQW PXOWLSOH FRUUHODWLRQ FRHIILFLHQW ZKHQ )XOO 6FDOH ,4 RU DQ\ FRPELQDWLRQ RI :,6&5 VXEWHVWV ZDV XVHG DV WKH FULWHULRQ YDULDEOH 6WHQVRQ FRQFOXGHG WKDW WKH *LIWHG 6FUHHQHU VKRXOG QRW EH XVHG WR SUHGLFW :,6&5 ,4 IRU JLIWHG SURJUDP SURVSHFWV ,Q &ODUL]LR DQG 0HKUHQV HYDOXDWHG WKH WHFKQLFDO GDWD PDQXDOV IRU WKH 62, WR GHWHUPLQH WKH WHVWnV YDOXH DV D VFUHHQLQJ WHVW IRU JLIWHG LQWHOOLJHQFH ,W ZDV FRQn FOXGHG WKDW WKH 62, PRGHO KDV VHYHUH SV\FKRPHWULF OLPLWDn WLRQV S f 7KHVH OLPLWDWLRQV FHQWHU DURXQG SRRU UHOLDELOLW\ LQDGHTXDWH QRUPDWLYH GDWD DQG SRRU H[WHUQDO YDOLGLW\ IRU PDQ\ RI WKH IDFWRUV PHDVXUHG E\ WKH WHVW $FKLHYHPHQW 7HVWV ,Q UHVHDUFK LQWURGXFHG LQ &KDSWHU LW ZDV VXJJHVWHG WKDW DFKLHYHPHQW WHVWV &7%6f DQG FRJQLWLYH DELOLW\ WHVWV 7&6 :,6&5 6%f PHDVXUH PXFK WKH VDPH FRQVWUXFW ,Q WKH VXSSRUWLYH OLWHUDWXUH ZHUH LQGLFDWLRQV WKDW GLVWLQFn WLRQV EHWZHHQ DFKLHYHPHQW WHVWV DQG FRJQLWLYH DELOLW\ WHVWV DUH RIWHQ XQFOHDU &RUUHODWLRQDO DQG IDFWRU DQDO\WLF VWXGLHV OHQG FUHGHQFH WR WKLV FRQWHQWLRQ

PAGE 37

/HQQRQ f KDV IRXQG WKDW UHODWLRQVKLSV EHWZHHQ LQWHOOLJHQFH DQG DFKLHYHPHQW WHVWV DUH VR VWURQJ DV WR OHDG WR WKH FULWLFLVP WKDW WKH WZR W\SHV RI WHVWV GR QRW PHDVXUH DQ\WKLQJ GLIIHUHQW %RWK WHVWV PHDVXUH ZKDW WKH VWXGHQW KDV OHDUQHG *URQOXQG f DQG ERWK WHVWV SUHn GLFW IXWXUH OHDUQLQJ ZLWK VLPLODU GHJUHHV RI VXFFHVV ,4 WHVWV DQG DFKLHYHPHQW WHVWV GLIIHU LQ IRUP EXW QRW QHFHVn VDULO\ LQ FRQWHQW 0HUFHU f ,Q D VHULHV RI VWXGLHV LQ WKH ODWH V DQG V :,6&5 ,4V ZHUH FRUUHODWHG ZLWK DFKLHYHPHQW VXEWHVWV RI WKH :LGH 5DQJH $FKLHYHPHQW 7HVW :5$7f &RQVLVWHQWO\ KLJK FRUUHODWLRQV ZHUH IRXQG 6RPH RI WKHVH HDUO\ VWXGLHV DUH VXPPDUL]HG LQ 7DEOH $OVR LQ 6WHGPDQ /DZOLV &RUWQHU DQG $FKWHQ EHUJ DWWHPSWHG WR UHODWH .DXIPDQnV f IDFWRU VFRUHV WR :5$7 DWWDLQPHQW LQ D SRSXODWLRQ RI FKLOGUHQ DJHV WR &RUUHODWLRQV ZHUH IRXQG WR EH SRVLWLYH DQG VLJQLIn LFDQW
PAGE 38

7DEOH 6XPPDU\ RI 5HODWLRQVKLS EHWZHHQ :,6&5 DQG :5$7 :5$7 :,6&5 5HDGLQJ 6SHOOLQJ $ULWKPHWLF %URRNV f 1 \HDUV 96 ,4 G 36 ,4 )6 ,4 +DUWODJH DQG 6WHHOH f 1 0HDQ DJH \UV PRQWKV 96 ,4 36 ,4 )6 ,4 6FKZDUWLQJ DQG 6FKZDUWLQJ f 1 \HDUV Df \UV 96 ,4 36 ,4 )6 ,4 Ef \UV r 96 ,4 36 ,4 )6 ,4 +DOH f 1 \HDUV 96 ,4 36 ,4 )XOO 6FDOH FRUUHODWLRQV QRW TXRWHG D'HFLPDO SRLQWV RPLWWHG ,Q D IROORZXS WR WKH VWXGLHV VXPPDUL]HG LQ 7DEOH ZDV FRQGXFWHG E\ *URVVPDQ DQG -RKQVRQ ,Q WKHLU VWXG\ VWXGHQWV DJHV WR ZHUH DGPLQLVWHUHG WKH :,6&5 DQG WKH :+$7 )DFWRU VFRUHV ZHUH FRPSXWHG RQ WZR RI .DXIPDQnV f IDFWRUV 9HUEDO &RPSUHKHQVLRQ DQG

PAGE 39

)UHHGRP IURP 'LVWUDFWLELLW\f DQG :5$7 VXEWHVWV $ PXOWLn SOH UHJUHVVLRQ DQDO\VLV ZDV FRPSXWHG ZKHUHLQ :,6&5 IDFWRU VFRUHV VHUYHG DV FRQMRLQW SUHGLFWRUV DQG WKH :5$7 VWDQGDUG VFRUHV ZHUH HPSOR\HG DV FULWHULRQ YDULDEOHV 5HVXOWV LQn GLFDWHG D VLJQLILFDQW RYHUDOO SUHGLFWLRQ RI :5$7 UHDGLQJ VSHOOLQJ DQG DULWKPHWLF E\ WKH WZR :,6&5 IDFWRUV :ULJKW DQG 'DSSDQ f DVVHVVHG VWXGHQWV ZLWK D PHDQ DJH RI QLQH \HDUV RQ WKH :,6&5 DQG :5$7 )DFWRU DQDO\VLV VKRZHG D FRPPRQ IDFWRU IRU VXEWHVWV RQ ERWK PHDn VXUHV &RUUHODWLRQV EHWZHHQ VXEWHVWV IURP WKH WZR WHVWV ZHUH DV KLJK DV U RQ :,6&5 $ULWKPHWLF DQG :5$7 $ULWKPHWLFf 6RPH RWKHU VXEWHVWV FRUUHODWHG DW WR /LWHUDWXUH FRQFHUQLQJ WKH RYHUODS RI LQGLYLGXDO WHVWV RI LQWHOOLJHQFH DQG WHVWV RI DFKLHYHPHQW LQFOXGH VWXGLHV LQ ZKLFK WKH :,6&5 DQG WKH 3HDERG\ ,QGLYLGXDO $FKLHYHPHQW 7HVW 3,$7f 'XQQ t 0DUNZDUGW f ZHUH H[DPLQHG :LNRII f IDFWRU DQDO\]HG WKH :,6&5 DORQJ ZLWK WKH 3,$7 IRU UHIHUUHG FKLOGUHQ $OWKRXJK WKH 3,$7 *HQHUDO ,QIRUPDWLRQ DQG 0DWKHPDWLFV VXEWHVWV ORDGHG RQ IDFWRUV SUHYLRXVO\ LGHQWLILHG LQ WKH VWUXFWXUH RI WKH :,6&5 WKH UHPDLQLQJ VXEWHVWV ORDGHG RQ D VHSDUDWH IDFWRU VXEVHn TXHQWO\ ODEHOHG :RUG 5HFRJQLWLRQ 7KH UHVXOWV VXSSRUWHG WKH XVH RI ERWK LQVWUXPHQWV DV VRXUFHV RI PXWXDO EXW

PAGE 40

VXSSOHPHQWDU\ LQIRUPDWLRQ LQ WKH DVVHVVPHQW RI OHDUQLQJ SUREOHPV 'HDQ f DVVHVVHG WKH GHJUHH RI UHGXQGDQF\ EHWZHHQ WKH :6&5 DQG WKH 3,$7 XVLQJ D FDQRQLFDO FRUUHODWLRQ DQDO\VLV ZLWK VFRUHV IURP UHIHUUHG FKLOGUHQ 7KH UHn VXOWV LQGLFDWHG WKDW b RI WKH IXQFWLRQV RI WKH 3,$7 RYHUODSSHG ZLWK WKH :,6&5 DQG WKDW b RI WKH IXQFWLRQV RI WKH :,6&5 RYHUODSSHG ZLWK WKH 3,$7 7KH RYHUODS ZDV DWWULEXWHG WR FRPPRQ YHUEDHGXFDWLRQD FRQWHQW 'HDQ f IRXQG D VLPLODU DV\PPHWULFDO RYHUODS EHWZHHQ WKHVH PHDVXUHV LQ VDPSOHV RI $QJOR DQG 0H[LFDQ$PHULFDQ FKLOGUHQ $V LQ :LNROO V IDFWRU DQDO\VLV ERWK RI 'HDQ V DQDO\VHV VKRZHG WKH 3,$7 VXEWHVWV RI UHDGLQJ DQG VSHOOLQJ WR RIIHU WKH JUHDWHVW GHJUHH RI LQIRUPDWLRQ QRW UHGXQGDQW ZLWK WKH :,6&5 %URFN f ILQGLQJ WKDW D SDXFLW\ RI UHVHDUFK H[LVWHG IRU IDFWRU DQDO\WLF LQYHVWLJDWLRQV RI WKH :,6&5 LQ FRPELQDWLRQV ZLWK LQGLYLGXDO DFKLHYHPHQW WHVWV FRQn GXFWHG VXFK D VWXG\ +H IDFWRU DQDO\]HG WKH :,6&5 :5$7 DQG 3,$7 IRU PDOH VWXGHQWV LQ JUDGHV WKURXJK $Q DWWHPSW ZDV PDGH WR GHWHUPLQH WKH WUDLWV RU FRPPRQ VNLOOV PHDVXUHG E\ ,4 DQG DFKLHYHPHQW WHVWV ZKHQ YLHZHG FRQFRPn LWDQWO\ )RXU IDFWRUV HPHUJHG 2QH D QXPHULFDO IDFWRU ZDV FRPSULVHG RI VXEWHVWV IURP DOO WKUHH WHVWV

PAGE 41

0RGHUDWHO\ KLJK FRUUHODWLRQV U WR ZHUH IRXQG EHn WZHHQ VRPH RI WKH RWKHU ,4 DQG DFKLHYHPHQW VXEWHVWV 6WHZDUW DQG 0RUULV f IDFWRU DQDO\]HG WKH :,6& :$,6 :5$7 DQG &$7 &DOLIRUQLD $FKLHYHPHQW 7HVWf IRU VWXGHQWV UDQJLQJ LQ DJH IURP WR $ VXEVWDQWLDO RYHUODS RI YHUEDO LQWHOOLJHQFH DQG DFDGHPLF DFKLHYHPHQW ZDV IRXQG 5HVXOWLQJ IDFWRUV FRQIRUPHG UHDVRQDEO\ ZHOO WR WKRVH RI .DXIPDQ f 6XEWHVWV IURP HDFK PHDVXUH ZHUH IRXQG WR ORDG RQ HDFK RI WKH ,4 IDFWRUV ,Q D VWXG\ ZKHUHLQ WKH DELOLWLHV XQGHUO\LQJ UHDGLQJ UHDGLQHVV ZHUH LGHQWLILHG 2OVHQ DQG 5RVHQ f IDFWRU DQDO\]HG WKUHH JURXS UHDGLQJ WHVWV DQG WKH :,6& 6XEMHFWV FRQVLVWHG RI ILUVW JUDGHUV 7KH VXEWHVWV ZHUH FRUn UHODWHG DQG WKH UHVXOWLQJ PDWULFHV VXEMHFWHG WR D SULQn FLSDO FRPSRQHQW DQDO\VLV )RXU FRPPRQ IDFWRUV ZHUH UHn YHDOHG ,Q RQH IDFWRU UHDGLQJ FRPSUHKHQVLRQ ORDGHG ZLWK IRXU :,6&5 VXEWHVWV ,Q DQRWKHU ZULWLQJ OHWWHUV FRUn UHODWHG KLJKO\ ZLWK :,6&5 9RFDEXODU\ ,Q D WKLUG IDFWRU VHQWHQFH ZULWLQJ DQG :,6&5 &RGLQJ ZHUH LQFOXGHG 7KHUH KDYH HPHUJHG WZR FDPSV RI WKRXJKW RQ WKH LVVXH RI UHDGLQJ VNLOO DFTXLVLWLRQ DQG LQWHOOLJHQFH 2Q RQH KDQG LQ VRPH UHVHDUFK LW KDV EHHQ VXJJHVWHG WKDW UHDGLQJ LV D IXQFWLRQ RI LQIRUPDWLRQ SURFHVVLQJ RU HQFRGLQJ VNLOOV DV RSSRVHG WR EHLQJ D SULPDULO\ LQWHOOHFWXDO IXQFWLRQ 2Q WKH RWKHU KDQG LQ VLPLODUO\ IRFXVHG IDFWRU DQDO\WLF LQYHVWLJDWLRQV UHDGLQJ KDV EHHQ IRXQG WR EH KLJKO\ ORDGHG

PAGE 42

RQ D JHQHUDO LQWHOOLJHQFH IDFWRU DQG D JRRG SUHGLFWRU RI LQWHOLJHQFH 5HVHDUFKHUV ZKRVH YLHZV UHSUHVHQW WKH ODWWHU YLHZn SRLQW KDYH VXSSRUWHG WKH QRWLRQ WKDW UHDGLQJ DELOLW\ LV KLJKO\ FRUUHODWHG ZLWK JHQHUDO LQWHOOLJHQFH DQG LV D JRRG SUHGLFWRU RI LQWHOOLJHQFH -HQVHQ f UHSRUWHG D FRUn UHODWLRQ RI U EHWZHHQ UHDGLQJ FRPSUHKHQVLRQ DQG )XOO 6FDOH ,4 IRU D ODUJH VDPSOH RI VWXGHQWV 2WKHU UHVHDUFKn HUV KDYH FLWHG VLPLODUO\ KLJK f FRUUHODWLRQV EHn WZHHQ UHDGLQJ DQG ,4 IRU YDULRXV VDPSOHV RI VWXGHQWV LQ JUDGHV WKURXJK %URRNV +DOH +DUWODJH t 6WHHOH 5\DQ :LNRII
PAGE 43

OHYHOV DERYH WKH QRUP RQ WKH 0HWURSROLWDQ 5HDGLQJ 7HVW b VFRUHG DERYH RQ WKH 6WDQIRUG%LQHW ,Q KLV VWXG\ 6WHUQEHUJ IRXQG WKDW ,4 DFFRXQWHG IRU DV PXFK DV b RI WKH YDULDQFH LQ VFKRODVWLF SHUIRUPDQFH ,Q WZR RI WKH IHZ RWKHU VWXGLHV LQ ZKLFK DFKLHYHPHQW ZDV FRUUHODWHG ZLWK LQWHOOLJHQFH DPRQJ VXSHULRU VWXGHQWV 0D\ILHOG f KDG WKLUG JUDGHUV HYDOXDWHG LQ WHUPV RI LQWHOOLJHQFH DFKLHYHPHQW FUHDWLYLW\ DQG WHDFKHU SHUn FHSWLRQ RI ,4 5HVXOWV \LHOGHG VLJQLILFDQW FRUUHODWLRQV EHWZHHQ LQWHOOLJHQFH DQG D ZLGH UDQJH RI DFKLHYHPHQW GRn PDLQV DPRQJ WKH VWXGHQW VDPSOH 6LPLODUO\ .DUQHV (GZDUGV DQG 0F&DOOXP f IRXQG D VLJQLILFDQW FRUUHODn WLRQ EHWZHHQ WRWDO VFRUHV RQ WKH &DOLIRUQLD $FKLHYHPHQW 7HVW &$7f DQG :,6&5 )XOO 6FDOH ,4V RI JLIWHG FKLOGUHQ LQ JUDGHV IRXU WKURXJK VL[ 7KXV WKH UHVXOWV RI WKLV ERG\ RI OLWHUDWXUH FRPSDULQJ LQWHOOLJHQFH ZLWK DFKLHYHPHQW DSSHDU FRQFOXVLYH DV LQ D VXEVWDQWLDO QXPEHU RI VWXGLHV LW LV VXJJHVWHG WKDW WKH WZR YDULDEOHV DUH IDLUO\ KLJKO\ FRUUHODWHG 0DOOLQVRQ f DWWHPSWHG WR XQFRYHU D UHODWLRQVKLS EHWZHHQ LQWHOOLJHQFH DQG DFKLHYHPHQW LQ VFLHQFH DQG PDWK 7KH 65$ DFKLHYHPHQW VHULHV DQG WKH 65$ 3ULPDU\ $ELOLWLHV 7HVW ZHUH JLYHQ WR VHFRQGDU\ JUDGH VWXGHQWV 7KHUH ZDV D UHVXOWLQJ FRUUHODWLRQ RI U EHWZHHQ YHUEDO DELOLW\ DQG VFLHQFH IDFWV DQG SULQFLSOHVf 9HUEDO DELOLW\ ZDV DOVR

PAGE 44

IRXQG WR KDYH UHDVRQDEO\ KLJK FRUUHODWLRQV ZLWK IDFWRUV RI DULWKPHWLF DFKLHYHPHQW ,Q DQRWKHU VWXG\ LQ ZKLFK IDFWRUV UHODWHG WR PDWK SHUIRUPDQFH ZHUH LQYHVWLJDWHG 5RDFK f UHSRUWHG D VLJQLILFDQW FRUUHODWLRQ U f EHWZHHQ DULWKPHWLF DFKLHYHPHQW DQG YHUEDO ,4 LQ WKLUG JUDGHUV &RUUHODWLRQ FRHIILFLHQWV IRU &7%6 DQG 7&6 VXEWHVWV &7%0F*UDZ+L f ZHUH FDOFXODWHG XVLQJ VHYHQWK JUDGHUV &RHIILFLHQWV EHWZHHQ DQG ZHUH QRW XQFRPn PRQ &RUUHODWLRQV EHWZHHQ WKH 7&6 7RWDO 6FRUH DQG &7%6 VXEVFDOHV RI 5HDGLQJ /DQJXDJH 0DWK 6RFLDO 6WXGLHV DQG 6FLHQFH ZHUH DQG UHVSHFn WLYHO\ 7KH &7%6 DQG 7&6 7RWDO %DWWHULHV FRUUHODWHG DW U 7KHVH KLJK FRUUHODWLRQV VXJJHVW WKDW ERWK WHVWV PD\ PHDVXUH VLPLODU WKRXJK RSHUDWLRQDOO\ GLVWLQJXLVKDEOH FRQVWUXFWV 6XSSRUW IRU WKH FRQWHQWLRQ WKDW WKHVH WHVWV UHSUHVHQW VLPLODU FRQVWUXFWV PD\ DOVR EH IRXQG LQ WKH FRUn UHODWLRQV RI 7&6 VXEVFDOH VFRUHV ZLWK WKRVH VFRUHV RI LWV SUHGHFHVVRU WKH 6KRUW )RUP 7HVW RI $FDGHPLF $SWLWXGH 6)7$$f 7KH UDQJH RI FRUUHODWLRQV ZDV WR QRW GLVVLPLODU WR WKRVH RI WKH 7&6 DQG &7%6 7KH IDFW WKDW WKH DYHUDJH FRUUHODWLRQV ZHUH SRVLWLYH PHDQV WKDW WKH VXEn VFDOHV PXVW EH PHDVXULQJ VRPHWKLQJ LQ FRPPRQ -HQVHQ f ,Q VXPPDU\ WKHUH LV VXEVWDQWLDO HYLGHQFH WKDW DFKLHYHPHQW WHVW VFRUHV DQG ,4 WHVW VFRUHV DUH KLJKO\

PAGE 45

FRUUHODWHG 7KH UHVHDUFK DOVR SURYLGHV UHDVRQ WR EHOLHYH WKDW D FRPPRQ IDFWRU XQGHUOLHV SHUIRUPDQFH RQ ERWK W\SHV RI WHVWV 6XPPDU\ ,Q WKH OLWHUDWXUH UHOHYDQW WR WKH DFFXUDF\ RI YDULRXV SURFHGXUHV IRU VFUHHQLQJ LQWHOOLJHQFH DPRQJ JLIWHG VWXn GHQWV WKHUH KDYH JHQHUDOO\ EHHQ PL[HG UHVXOWV 6RPH HQFRXUDJLQJ ILQGLQJV KDYH RFFXUUHG RQ VWXGLHV RI DFKLHYHn PHQW UDWLQJV DQG VKRUWIRUP ,4 WHVWV *URXS ,4 WHVWV KDYH WHQGHG WR XQGHUHVWLPDWH WKH ,4V RI VRPH JLIWHG FKLOGUHQ WKRXJK WKH\ JHQHUDOO\ SUHGLFWHG JLIWHG ,4 ZLWK PRGHUDWH DFFXUDF\ 0XFK OHVV HIIHFWLYH PHDQV IRU SUHGLFWLQJ JLIWHG ,4 DUH ,4 VFUHHQLQJ WHVWV 7KH SUHFHGLQJ OLWHUDWXUH KDV IRFXVHG RQ WKH SUREOHPV RI VFUHHQLQJ JLIWHG LQWHOOLJHQFH DPRQJ WKH VFKRRO DJH SRSn XODWLRQV 7KRVH SURFHGXUHV WKDW KDYH VKRZQ VRPH VXFFHVV KDYH EHHQ LQFRQVLVWHQW LQ WKHLU ILQGLQJV 1HDUO\ DOO KDYH SURYHG LQHIILFLHQW LQ WHUPV RI WLPH DQG FRVW ,QVWUXPHQWV 8VHG LQ 6WXG\ :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG :,6&5f 7KH :,6&5 KDV EHHQ WKH PRVW ZLGHO\ DGPLQLVWHUHG WHVW RI FKLOGUHQnV LQWHOOLJHQFH %U\DQ t %U\DQ *URVVPDQ

PAGE 46

t *DOYLQ 6DOYLD t
PAGE 47

3HUFHSWXDO 2UJDQL]DWLRQ 7KHVH VWXGLHV VWURQJO\ VXSSRUW WKH YDOLGLW\ RI WKH :,6&5 6ORVVRQ ,QWHOOLJHQFH 7HVW 7KH 6,7 LV DQ ,4 WHVW IRU FKLOGUHQ DQG DGXOWV GHn VLJQHG IRU XVH E\ HLWKHU UHODWLYHO\ XQWUDLQHG H[DPLQHUV RU TXDOLILHG SURIHVVLRQDOV 7KH 6,7 W\SLFDOO\ WDNHV EHWZHHQ DQG PLQXWHV WR DGPLQLVWHU 1HZ QRUPV f UHSUHVHQW D VLJQLILFDQW GHSDUWXUH IURP SURFHGXUHV SUHYLRXVO\ HPSOR\HG f IRU FDOFXODWLQJ DQ ,4 VFRUH ,Q QRUPLQJ WKH 6,7 WKH 6WDQIRUG%LQHW ZDV XVHG DV WKH DQFKRU WHVW &RQVLVWHQW ZLWK WKH UHYLn VLRQ RI WKH 6WDQIRUG%LQHW UDWLR ,4V ZHUH DEDQGRQHG LQ IDYRU RI GHYLDWLRQ ,4V )UHTXHQF\ GLVWULEXWLRQV ZHUH FDOn FXODWHG IRU HDFK RI WKH FKURQRORJLFDO DJH UDQJHV RQ ERWK ,4 VFDOHV 7KHQ XWLOL]LQJ D PRGLILHG WDEOH ORRNXS DSSURDFK DSSURSULDWH ,4V IURP WKH 6WDQIRUG%LQHW ZHUH HQWHUHG RQ WKH GHYHORSLQJ 6,7 WDEOHV 7KH PHDQ ,4 IRU WKH 6,7 LV DQG WKH VWDQGDUG GHYLDWLRQ LV ,Q WKH 6,7 PDQXDO WKH DXWKRUV SUHVHQW HYLGHQFH WR SHUVXDGH WKH UHDGHU WKDW WKH UHYLVHG 6,7 ,4V DUH HTXLYDn OHQW WR 6WDQI RUG%LQHW ,4V 7KLV WDVN LV XQGHUWDNHQ LQ ODUJH SDUW E\ FRPSDULQJ SUHYLRXV f OHVV SRVLWLYHO\ FRUUHODWHG FRHIILFLHQWV WR QHZHU GDWD 6ORVVRQ DQG -HQVHQ f VWDWHG WKDW WKH 6,7 LV DV DFFXUDWH DV WKH 6WDQIRUG%LQHW LQ PHDVXULQJ D SHUVRQnV

PAGE 48

LQWHOOLJHQFH ZKHQ ERWK LQVWUXPHQWV KDYH EHHQ SURSHUO\ DGPLQLVWHUHG S f DQG IXUWKHU SURSRVHG WKDW WKH 6,7 TXDOLILHV DV DQ DOWHUQDWH IRUP RI WKH 6WDQIRUG%LQHW EHn FDXVH WKH WZR WHVWV SRVVHVV HTXLYDOHQW PHDQV DQG VWDQGDUG GHYLDWLRQV S f %DVHG RQ WKHLU GXELRXV DVVXPSWLRQ RI WHVW HTXLYDOHQF\ EHWZHHQ WKH UHYLVHG 6,7 DQG WKH 6% WKH DXWKRUV HPSOR\HG WKH 0HDQ $EVROXWH ,4 'LIIHUHQFH 0$'f VWDWLVWLF WR GHWHUPLQH DOWHUQDWH IRUP WHVW UHOLDELOLW\ DQG VWDQGDUG HUURUV RI PHDVXUHPHQW 7KH 0$' SURFHGXUH ZKLFK LV PHDQW WR EH XVHG RQO\ ZLWK HTXLYDOHQW IRUPV RI D WHVW \LHOGV D VWDWLVWLF ZKLFK LV DSSUR[LPDWHO\ HTXDO WR WKH VWDQGDUG GHYLDWLRQ WLPHV 7KH DXWKRUV GLG QRW LQGLn FDWH WKH UHODWLYH HIIHFWV RQ WKH UHOLDELOLW\ DQG WKH 6(0V ZKHQ WKH PHDVXUHV FRPSDUHG GR QRW VWULFWO\ PHHW WKH &ULWHn ULD RI DOWHUQDWH IRUPV DV LV DSSDUHQW LQ WKLV FDVH 1RU DUH WKHUH DWWHPSWV WR HYDOXDWH RWKHU NLQGV RI WHVW UHOLDn ELOLW\ ,W PLJKW EH FRQFOXGHG WKHUHIRUH WKDW WKH DXWKRUVn FODLP WKDW WKH 6,7nV UHOLDELOLW\ PD\ EH UHJDUGHG DV QRW OHVV WKDQ S f VKRXOG EH LQWHUSUHWHG ZLWK FDXWLRQ $QRWKHU FODLP PDGH LQ WKH 6,7 PDQXDO LV WKDW WKH PHDQ GLIIHUHQFH EHWZHHQ WKH 6WDQIRUG%LQHW DQG 6ORVVRQ ,4 VFRUHV LV OHVV WKDQ RQH SRLQW EDVHG RQ WKH VDPSOH RI 7KH SURFHGXUH E\ ZKLFK WKLV VWDWLVWLF ZDV REWDLQHG HQWDLOV FRPSXWLQJ WKH PHDQV IRU ,4 GLIIHUHQFHV EHWZHHQ WKH

PAGE 49

WZR WHVWV eRU WKUHH ,4 OHYHOV DFURVV IRXU DJH JURXSV EHn ORZ DQG DERYH )RU H[DPSOH DW DJH DQG DERYH PHDQ GLIIHUHQFHV EHWZHHQ WKH 6WDQIRUG%LQHW DQG 6,7 DUH DQG DW WKH WKUHH ,4 OHYHOV ,Q FDOFXODWLQJ WKH PHDQ GLIIHUHQFH QHJDWLYH PHDQV DUH DGGHG WR SRVLWLYH PHDQV UHVXOWLQJ LQ D PLVOHDGLQJO\ ORZ RYHUDOO PHDQ GLIIHUHQFH ,Q WKLV H[DPSOH WKH WRWDO PHDQ GLIIHUn HQFH IRU DJH DQG DERYH LV +RZHYHU LI LQGLYLGn XDO PHDQV KDG EHHQ VXPPHG LQ WHUPV RI QRQGLUHFWLRQD GHYLn DWLRQ IURP ]HUR WKH PHDQ GLIIHUHQFH ZRXOG KDYH EHHQ DSSUR[LPDWHO\ :LWK UHJDUG WR WKH PHDQ GLIIHUHQFH IRU WKH HQWLUH VDPSOH ZKHQ WKH QRQGLUHFWLRQD SURFHGXUH LV XVHG WKH GLIIHUHQFH FKDQJHV IURP WR DSSUR[LPDWHO\ 7KH PHDQ VFRUHV DUH UHQGHUHG HYHQ PRUH GLIILFXOW WR LQWHUSUHW EHFDXVH QR VWDQGDUG HUURUV IRU WKH PHDQV DUH UHSRUWHG ,Q VSLWH RI WKH DSSDUHQW LQFRQVLVWHQFLHV LQ WKH QHZ 6,7 PDQXDO WKH UHYLVLRQV SDUWLFXODUO\ LQ LWV UHQRUPLQJ UHSUHVHQW FRQVLGHUDEOH LPSURYHPHQW LQ WKH WHVWnV YDOLGLW\ DQG UHOLDELOLW\ 7KHVH WHVW LPSURYHPHQWV DORQJ ZLWK WKH HDVH RI LWV DGPLQLVWHULQJ DQG VFRULQJ UHQGHU WKH 6,7 D WHVW RI FRQVLGHUDEOH XWLOLW\ DV DQ LQWHOOHFWXDO VFUHHQLQJ SURFHGXUH

PAGE 50

&RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV &7%6f7HVW RI &RJQLWLYH 6NLOOV 7&6f 7KH &7%6 &7%0F*UDZ+L f DQG 7&6 &7%0F*UDZ +LOO f DUH WKH WHVWV IURP ZKLFK 13 LWHPV ZHUH WDNHQ 3V\FKRPHWULFDOO\ WKHVH WHVWV ZHUH ZHOO VXLWHG WR WKLV UHn VHDUFK 7KH DSSURSULDWHQHVV RI WKH &7%6 DQG 7&6 IRU WKLV UHVHDUFK DUH VXSSRUWHG E\ VHYHUDO RI WKHLU DWWULEXWHV VRPH RI ZKLFK ZHUH GLVFXVVHG LQ &KDSWHU ,WHPV ZHUH FKRVHQ IRU ERWK WHVWV DFFRUGLQJ WR LWHP UHVSRQVH WKHRU\ ,57f XWLOL]LQJ D WKUHHSDUDPHWHU ORJLVWLF PRGHO 7KH LWHPV ZHUH FKRVHQ DFFRUGLQJ WR WKHLU DELOLW\ WR Df GLVFULPLQDWH KLJK DELOLW\ WUDLWV IURP ORZ DELOLW\ WUDLWV Ef GLVFULPLQDWH KLJK DELOLW\ VWXGHQWV DQG ORZ DELOLW\ VWXGHQWV E\ PDWFKLQJ LWHP GLIILFXOW\ ZLWK VWXGHQW WRWDO VFRUH DQG Ff DFFRXQW IRU JXHVVLQJ DV DQ LQIOXHQFH RQ VFRUH GLIILFXOW\ ,Q WHUPV RI FRQWHQW YDOLGLW\ WKH &7%6 LV GHVLJQHG WR PHDVXUH XQGHUVWDQGLQJ RI D EURDG UJQJH RI FRQFHSWV DV GHYHORSHG E\ YDULRXV HGXFDWLRQDO FXUULFXOD 7HVW SHUIRUn PDQFH UHIOHFWV D VWXGHQWnV VNLOOV LQ HIIHFWLYH XVH RI LQIRUPDWLRQ H[SOLFLW LQ FDWHJRULHV GHULYHG IURP %ORRPnV WD[RQRP\ %ORRP f ,WHP GHYHORSPHQW VSHFLILFDWLRQV ZHUH GHVLJQHG WR HQVXUH FRPSUHKHQVLYH FRYHUDJH RI WKH FRQn WHQW DQG SURFHVV FDWHJRULHV 7KH 7&6 LV GHVLJQHG WR PHDVXUH DQ DSWLWXGH FRQVWUXFW WKDW FDQ EH RSHUDWLRQDOO\ GLVWLQJXLVKHG IURP WKH

PAGE 51

DFKLHYHPHQW FRQVWUXFW RI WKH &7%6 EDVHG RQ UHVHDUFK FRQn GXFWHG DW 0F*UDZ+LOO E\ %XFKHW FLWHG LQ 7&%0F*UDZ+L f (PSLULFDO FULWHULD IRU GLVWLQn JXLVKLQJ EHWZHHQ DSWLWXGH DQG DFKLHYHPHQW PHDVXUHV ZHUH GHULYHG E\ WKH SXEOLVKFUV 3URGXFW PRPHQW FRUUHODWLRQV EHWZHHQ WKH IRXU VXEWHVWV RI WKH 7&6 ZHUH EHWZHHQ U WR U &RHIILFLHQWV EHn WZHHQ VXEWHVWV DQG WRWDO VFRUH UDQJHG IURP U WR U 7KHUHIRUH LW ZDV VXJJHVWHG WKDW DOO VXEWHVWV PHDVXUH JHQn HUDO LQWHOOLJHQFH EXW DOVR PHDVXUH LQGHSHQGHQW IDFWRUV $ FRUUHODWLRQ FRHIILFLHQW EHWZHHQ WKH &7%6 DQG 7&6 RI U ZDV FDOFXODWHG RQ D VDPSOH RI VHYHQWK JUDGHUV $QRWKHU DWWULEXWH RI WKH &7%6 DQG 7&6 LV WKH FRPSUHn KHQVLYH VDPSOLQJ DQG QRUPLQJ VWDQGDUGV DSSOLHG 7KH IRUPn LQJ VDPSOHV FRQWDLQHG DSSUR[LPDWHO\ VWXGHQWV LQ JUDGHV IURP SXEOLF &DWKROLF DQG RWKHU SULYDWH VFKRROV &7%0F*UDZ+L f 6FKRRO GLVWULFWV ZHUH UDQGRPO\ FKRVHQ IURP IRXU JHRJUDSKLF UHJLRQV &RPSUHKHQn VLYH QRUPLQJ DQG VWDQGDUGL]DWLRQ LQIRUPDWLRQ LV DYDLODEOH LQ WKH &7%6 DQG 7&6 7HFKQLFDO 5HSRUWV ,QWHUQDO UHOLDELOLW\ FRHIILFLHQWV ZHUH FDOFXODWHG DFFRUGLQJ WR WKH .XGHU5LFKDUGVRQ IRUPXOD &7%6 UHOLDn ELOLW\ FRHIILFLHQWV UDQJHG IURP WR RQ WKH VXEn WHVWV &7%0F*UDZ+LO f $OO VXEWHVWV H[FHSW 6SHOOn LQJ DQG 5HIHUHQFH 6NLOOV KDG YDOXHV DW RU DERYH 2Q WKH IRXU 7&6 VXEWHVWV UHOLDELOLW\ FRHIILFLHQWV UDQJHG IURP

PAGE 52

WR 7KH 7&6 7HFKQLFDO 5HSRUWV SURYLGH UHOLDELOn LWLHV RQ 6(0V IRU VXEWHVWV EDVHG RQ QXPEHU FRUUHFW DW HDFK JUDGH OHYHO &RPSRVLWH FDOFXODWLRQV IRU WKH WRWDO WHVW DUH QRW SURYLGHG $OVR UHSRUWHG DUH ELDV VWXGLHV DQG WDEOHV LQGLFDWLQJ KRZ WHVW ELDVHV DUH DFFRXQWHG IRU DQG FRQWURO OHG ,Q VXPPDU\ WKH &7%6 DQG 7&6 ZHUH ZHOO VXLWHG IRU WKLV VWXG\ EHFDXVH RI WKH VRSKLVWLFDWHG PHWKRG XWLOL]HG LQ DQDO\]LQJ LWHPV DQG WKH WHVWVn KLJK YDOLGLW\ DQG UHOLDELOn LW\ )XUWKHU ERWK WHVWV HPSOR\HG VDPSOLQJ SURFHGXUHV GHVLJQHG WR SURYLGH QRUPV IRU WKH HQWLUH 86 VFKRRO SRSXn ODWLRQ 5HVHDUFK KDV DOVR EHHQ FRQGXFWHG WR DLG LQ UHGXFn LQJ WHVW ELDV IRU WKH &7%6 DQG 7&6 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH 6%f 7KH WKLUG UHYLVLRQ RI WKH 6WDQIRUG%LQHW 6%f SXEn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

PAGE 53

JUDGHV WKURXJK ZKR ZHUH LGHQWLILHG EDVHG RQ VFRUHV IURP WKH &RJQLWLYH $ELOLWLHV 7HVW 7KH GLVWULEXWLRQ RI VFRUHV LQ WKLV VXEVDPSOH FRUUHVSRQGHG WR WKH QDWLRQDO GLVn WULEXWLRQ RI WKH HQWLUH VDPSOH 7KH QRUPV ZHUH EHn OLHYHG WR EH EDVHG RQ D PRUH UHSUHVHQWDWLYH VDPSOH WKDQ SUHYLRXV QRUPV 7HUPDQ t 0HUULOO f 7KH UHOLDELOLW\ RI WKH 6WDQIRUG%LQHW ZDV GHWHUn PLQHG E\ FRUUHODWLQJ ,4V RQ IRUPV / DQG 0 DGPLQLVWHUHG WR WKH VWDQGDUGL]DWLRQ JURXS ZLWKLQ DQ LQWHUYDO RI RQH ZHHN RU OHVV 6XFK UHOLDELOLW\ FRHIILFLHQWV DUH WKXV PHDVXUHV RI ERWK VKRUW WHUP WHPSRUDO VWDELOLW\ DQG HTXLYDOHQFH DFURVV WKH WZR LWHP VDPSOHV ,Q JHQHUDO WKH 6WDQIRUG %LQHW WHQGV WR EH PRUH UHOLDEOH IRU ROGHU WKDQ IRU \RXQJHU DJH JURXSV DQG IRU ORZHU WKDQ IRU KLJKHU ,4V $QDVWDVL f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

PAGE 54

UHDVRQLQJ 7KH PRVW FRPPRQ W\SH RI WHVW HVSHFLDOO\ RI WKH XSSHU DJH OHYHOV LV WKDW HPSOR\LQJ YHUEDO FRQWHQW 'DWD RQ FULWHULRQUHODWHG YDOLGLW\ RI WKH 6WDQIRUG %LQHW KDYH EHHQ REWDLQHG FKLHIO\ LQ WHUPV RI DFDGHPLF DFKLHYHPHQW $QDVWDVL f &RUUHODWLRQV EHWZHHQ WKH VFDOH DQG VFKRRO JUDGHV WHDFKHUVn UDWLQJV DQG DFKLHYHn PHQW WHVW VFRUHV JHQHUDOO\ IDOO EHWZHHQ DQG 7KH 6WDQIRUG%LQHW WHQGV WR FRUUHODWH KLJKO\ ZLWK SHUIRUPDQFH LQ QHDUO\ DOO DFDGHPLF FRXUVHV EXW SUHGRPLQDQWO\ ZLWK YHUEDO FRXUVHV VXFK DV (QJOLVK DQG KLVWRU\ &RUUHODWLRQV ZLWK DFKLHYHPHQW WHVW VFRUHV VKRZ WKH VDPH SDWWHUQ 7KH ULJRURXV VWDQGDUGL]DWLRQ DQG URQRUPLQJ RI WKH 6WDQIRUG %LQHW DORQJ ZLWK LWV KLJK YDOLGLW\ DQG UHOLDELOLW\ LQGLn FDWH WKDW LW ZDV DQ DSSURSULDWH ,4 WHVW IRU WKLV VWXG\

PAGE 55

&+$37(5 ,,, 0(7+2'2/2*< 2YHUYLHZ ,Q WKLV UHVHDUFK VWXG\ D SURFHGXUH ZDV LQYHVWLJDWHG IRU DQDO\]LQJ D FRPSUHKHQVLYH JURXSDGPLQLVWHUHG DFKLHYHn PHQW DQG FRJQLWLYH DELOLWLHV WHVW WR GHWHUPLQH ZKHWKHU DQ LWHP VHW FDQ EH GHULYHG WKDW GLVFULPLQDWHV JLIWHG IURP KLJKDFKLHYLQJ EXW QRWJLIWHG VHYHQWK JUDGHUV :KHQ VXFK DQ LWHP VHW ZDV GHULYHG LW ZDV FRPSDUHG WR D FRPn PRQO\ XVHG ,4 VFUHHQLQJ WHVW WR DVVHVV WKH UHODWLYH DFFXn UDFLHV RI HDFK SURFHGXUH LQ GLVFULPLQDWLQJ JLIWHG IURP QRWJLIWHG VWXGHQWV LQ D VHFRQG VHYHQWK JUDGH SRSXODWLRQ VDPSOH 7KLV FKDSWHU LV RUJDQL]HG LQWR WKH IROORZLQJ VHFn WLRQV Df 3RSXODWLRQ DQG 6DPSOH Ef $VVHVVPHQW ,QVWUXn PHQWV Ff 5HVHDUFK 3URFHGXUHV Gf 'DWD $QDO\VLV DQG Hf 0HWKRGRORJLFDO /LPLWDWLRQV 3RSXODWLRQ DQG 6DPSOH 7KH UHVHDUFK VDPSOH RI VWXGHQWV ZDV GUDZQ IURP D SRSXODWLRQ RI VHYHQWK JUDGHUV ZKR KDG EHHQ WHVWHG RQ WKH :,6&5 RU 6% IRU WKH JLIWHG SURJUDP LQ WKH +LOOVERURXJK

PAGE 56

&RXQW\ )ORULGDf 3XEOLF 6FKRRO 'LVWULFW 6DPSOLQJ ZDV FRQGXFWHG DW WKH HQG RI WKH VFKRRO \HDU 1HDUO\ DOO VWXGHQWV KDG SUHYLRXVO\ EHHQ DGPLQLVWHUHG WKH 6,7 ,Q PRVW LQVWDQFHV RQO\ WKRVH VWXGHQWV ZKR VFRUHG WZR VWDQn GDUG GHYLDWLRQV RU PRUH DERYH WKH PHDQ KDG EHHQ DGPLQLVn WHUHG WKH :,6&5 RU 6WDQIRUG%LQHW $OO VWXGHQWV LQ WKH SRSXODWLRQ DOVR KDG FXUUHQW &7%67&6 VFRUHV RQ ILOH 6RPH VWXGHQWV LQ WKLV SRSXODWLRQ KDG PHW LQWHOOHFWXDO HOLJLELOn LW\ JXLGHOLQHV IRU WKH JLIWHG SURJUDP RQ :,6&5 RU 6% FULWHULDf DQG VRPH KDG QRW $OO VWXGHQWV ZHUH WHVWHG E\ VFKRRO SV\FKRORJLVWV GXULQJ HDFK RI WKH WKUHH VFKRRO \HDUV XQGHU LQYHVWLJDWLRQ 6LPSOH UDQGRP VDPSOLQJ ZDV FRQGXFWHG E\ WKH UHVHDUFKn HU DW WKH +LOOVERURXJK &RXQW\ 6FKRRO %RDUG RIILFH LQ -XQH RI 1DPHV RI WKH VHYHQWK JUDGHUV ZKR ZHUH WHVWHG IRU WKH JLIWHG HGXFDWLRQ SURJUDP IURP 6HSWHPEHU RI DQG -XQH RI ZHUH REWDLQHG IURP FRPSXWHU SULQWRXWV FRQn WDLQLQJ GDWD IRU DOO VWXGHQWV LQ WKH GLVWULFW ZKR KDG EHHQ WHVWHG E\ VFKRRO SV\FKRORJLVWV 6WXGHQWV LQ WKH VDPSOH SRRO ZHUH DVVLJQHG D QXPEHU DQG QXPEHUV ZHUH VHOHFWHG DFFRUGLQJ WR D UDQGRP QXPEHU WDEOH 1XPEHUV ZHUH WKHQ UHFRUGHG DQG UHWXUQHG WR WKH SRRO WR HQVXUH DQ HTXDO FKDQFH RI VHOHFWLRQ IRU WKH UHPDLQLQJ QXPEHUV $IWHU VWXGHQWV ZHUH UDQGRPO\ VHOHFWHG IRU 3KDVH ,, WKH UHPDLQn LQJ VWXGHQWV ZHUH DVVLJQHG WR 3KDVH ,

PAGE 57

$VVHVVPHQW 3URFHGXUHV $V SUHYLRXVO\ GLVFXVVHG WKH ILYH DVVHVVPHQW LQVWUXn PHQWV XVHG IRU WKLV UHVHDUFK ZHUH WKH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG :,6&5f 7KH 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH 6a%f WKH 6ORVVRQ ,Qn WHOOLJHQFH 7HVW 6,7f WKH &RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV &7%6f f DQG 7HVW RI &RJQLWLYH 6NLOOV 7&6f $GPLQLVWUDWLRQ VFRULQJ DQG LQWHUSUHWDWLRQ RI WKH :,6&5 DQG 6WDQIRUG%LQHW ZHUH FRQGXFWHG E\ VWDWH FHUWLn ILHG VFKRRO SV\FKRORJLVWV SULRU WR WKH RQVHW RI WKLV VWXG\ $OO WHVWV ZHUH LQGLYLGXDOO\ DGPLQLVWHUHG DQG KDQG VFRUHG XVLQJ FXUUHQW QRUPV 7KH :,6&5 \LHOGV D 9HUEDO 6FDOH ,4 UHSUHVHQWLQJ YHUEDO UHDVRQLQJ DELOLWLHVf D 3HUn IRUPDQFH 6FDOH ,4 UHSUHVHQWLQJ SHUFHSWXDO RUJDQL]DWLRQ DQG QRQYHUEDO UHDVRQLQJf DQG D )XOO 6FDOH ,4 2QO\ WKH )XOO 6FDOH VFRUH ZKLFK UHSUHVHQWV WRWDO ,4 ZDV XVHG DV D PHDVXUH RI JLIWHG LQWHOOLJHQFH 7KLV SURFHGXUH FRQIRUPHG WR VFKRRO GLVWULFW JXLGHOLQHV 7KH 6WDQIRUG%LQHW \LHOGV D WRWDO ,4 VFRUH RQO\ $ FXWRII VFRUH RI ZDV XVHG DV WKH JLIWHG FXWRII LQ WKH GLVWULFW 6WXGHQWV DWWDLQLQJ D )XOO 6FDOH ,4 RI RU JUHDWHU ZHUH FRQVLGHUHG WR KDYH PHW WKH LQWHOOHFWXDO FULWHULRQ IRU JLIWHG SURJUDP HOLJn LELOLW\ 7KH ,4 FXWRII ZDV FKRVHQ E\ WKH GLVWULFW DV D VFRUH WKDW LV WZR VWDQGDUG GHYLDWLRQV DERYH WKH WHVW PHDQ :,6&5 ,4 f PLQXV RQH VWDQGDUG HUURU RI PHDVXUHPHQW WKUHH ,4 SRLQWVf 6% ,4 PLQXV ,4 SRLQWVf 7KH

PAGE 58

6ORVVRQ ,QWHOOLJHQFH 7HVW ZDV DGPLQLVWHUHG WR VWXGHQWV DV DQ ,4 VFUHHQLQJ SURFHGXUH E\ VFKRRO JXLGDQFH FRXQVHORUV RU FXUULFXOXP VSHFLDOLVWV ZKR W\SLFDOO\ KDG OLWWOH IRUPDO WUDLQLQJ LQ DGPLQLVWUDWLRQ RI LQGLYLGXDO LQWHOOLJHQFH WHVWV 6,7V ZHUH JLYHQ WR VWXGHQWV ZLWKLQ RQH \HDU SULRU WR :,6&5 WHVWLQJ 6,7 SURWRFROV ZHUH KDQG VFRUHG E\ WKH WHVW DGPLQLVWUDWRUV 8QGHU XVXDO FLUFXPVWDQFHV D WRWDO ,4 RI ZDV XVHG DV D VFUHHQLQJ FXWRII &KLOGUHQ ZKR VFRUHG DW RU DERYH WKLV FXWRII SRLQW ZHUH QRUPDOO\ UHn IHUUHG WR WKH VFKRRO SV\FKRORJLVW IRU :,6&5 HYDOXDWLRQ 7KHUH ZHUH VRPH H[FHSWLRQV WR WKLV UXOH EHFDXVH RFFDVLRQn DOO\ VWXGHQWV ZKR GLG QRW VFRUH DW RU DERYH WKH FXWRII ZHUH UHIHUUHG *HQHUDOO\ WKHVH VWXGHQWV H[KLELWHG H[n WUHPHO\ KLJK DFDGHPLF VNLOOV RU RWKHU FRPSHWHQFLHV WKDW FRPSHOOHG VFKRRO SHUVRQQHO WR UHIHU WKHP IRU IRUPDO WHVWn LQJ 7KH 6,7 \LHOGV D WRWDO ,4 VFRUH 7KH FXWRII VFRUH LV WZR VWDQGDUG GHYLDWLRQV DERYH WKH WHVW PHDQ 7KH &7%6 DQG 7&6 ZHUH DGPLQLVWHUHG WR VWXGHQWV E\ FODVVURRP WHDFKHUV LQ JURXS IRUPDW DFFRUGLQJ WR VWDQGDUGn L]DWLRQ SURFHGXUHV IRXQG LQ WKH WHDFKHUnV PDQXDO 7KH &7%6 DQG 7&6 ZHUH GHVLJQHG WR EH HDVLO\ DGPLQLVWHUHG $KPDQQ f DQG WHDFKHUV KDYH UHFHLYHG OLWWOH IRUPDO WUDLQLQJ LQ WKHLU DGPLQLVWUDWLRQ 7KH VHYHQWKJUDGH OHYHO RI WKH &7%6 /HYHO +f \LHOGV VXEVFDOH VFRULQJ LQ 5HDGLQJ WZR VHFWLRQVf 6SHOOLQJ /DQJXDJH WZR VHFWLRQVf

PAGE 59

0DWKHPDWLFV WZR VHFWLRQVf 5HIHUHQFH 6NLOOV 6FLHQFH DQG 6RFLDO 6WXGLHV 7KH 6FLHQFH DQG 6RFLDO 6WXGLHV VXEVFDOHV FRQVLVWLQJ RI LWHPV ZHUH QRW DGPLQLVWHUHG LQ DOO VFKRROV RI WKH SRSXODWLRQ VFKRRO GLVWULFW KRZHYHU WKHVH GDWD ZHUH LQFOXGHG LQ WKLV UHVHDUFK ,Q DGGLWLRQ WR VXEn VFDOH VFRUHV WKH &7%6 \LHOGV DQ RYHUDOO DFKLHYHPHQW LQGH[ 7KH 7&6 LV FRPSULVHG RI IRXU FRJQLWLYH DELOLW\ VXEn WHVWV PHDVXULQJ FRPSHWHQFLHV LQ 6HTXHQFLQJ $QDORJLHV 0HPRU\ DQG 9HUEDO 5HDVRQLQJ 'HULYHG VFRUHV DUH SURYLGHG IRU VXEWHVWV DQG IRU WKH RYHUDOO SURILOH %HFDXVH LQ WKLV VWXG\ UHVSRQVHV WR LQGLYLGXDO WHVW LWHPV ZHUH DQDO\]HG GHULYHG VXEVFDOH DQG WRWDO VFRUHV IRU WKH &7%6 VFRUHV ZHUH QRW XWLOL]HG 5HVHDUFK 3URFHGXUHV ,Q 3KDVH RI WKLV VWXG\ LWHPV FRPSULVLQJ WKH QHZ SURFHGXUH 13f ZHUH VHOHFWHG IURP WKH &7%6 DQG 7&6 E\ FRQn GXFWLQJ LWHP DQDO\VHV RI WKH SHUIRUPDQFH RI WKH VWXn GHQWV LQ WKH 3KDVH VDPSOH 7ZR VHWV RI LWHPV ZHUH GHn OLQHDWHG WKDW LQ JHQHUDO WHUPV ZHUH DQVZHUHG FRUUHFWO\ E\ WKH JLIWHG VWXGHQWV PRUH IUHTXHQWO\ WKDQ E\ WKH QRW JLIWHG VWXGHQWV *LIWHG FXWRII VFRUHV RQ WKH REWDLQHG 13 LWHPV ZHUH FRPSXWHG E\ VXEWUDFWLQJ IUDFWLRQV RU PXOWLSOHV RI VWDQGDUG GHYLDWLRQV IURP WKH PHDQ 13 VFRUH RI WKH JLIWHG JURXS WR

PAGE 60

GHWHUPLQH ZKLFK FXWRII SRLQWVf PRVW DFFXUDWHO\ SUHGLFWHG JLIWHGQHVV 1HZO\ UHIHUUHG VWXGHQWV LQ WKH VFKRRO VHWWLQJ ZKR VFRUHG DW RU DERYH WKDW SRLQW ZRXOG EH UHIHUUHG IRU IRUPDO ,4 WHVWLQJ %\ DGRSWLQJ D FXWRII VFRUH WZR VWDQn GDUG GHYLDWLRQV EHORZ WKH JLIWHG PHDQ DSSUR[LPDWHO\ b RI LQWHOOHFWXDOO\ JLIWHG VWXGHQWV ZRXOG EH UHIHUUHG IRU ,4 WHVWLQJ EDVHG RQ UHVHDUFK VDPSOH SDUDPHWHUV $ FXWRII VFRUH LQFRUSRUDWLQJ D RQH VWDQGDUG GHYLDWLRQ EHORZ WKH PHDQ FXWRII ZRXOG GHOHWH DSSUR[LPDWHO\ b RI WKH JLIWHG VWXGHQWV IURP WHVWLQJ DQG JLIWHG SURJUDP HOLJLELOLW\ +RZHYHU D SLWIDOO RI LQFOXGLQJ DV EURDG D UDQJH RI VWXn GHQWV DV SHUPLWWHG E\ WKH WZR VWDQGDUG GHYLDWLRQV FULWHn ULRQ ZDV WKDW D UHODWLYHO\ ODUJH QXPEHU RI QRWJLIWHG VWXn GHQWV ZRXOG DOVR EH UHIHUUHG IRU IRUPDO ,4 WHVWLQJ WKXV UHGXFLQJ WKH DFFXUDF\ RI 13 SUHGLFWLRQV RI JLIWHGQHVV $GMXVWPHQW RI WKH FXWRII SRLQW ZDV GHVLUDEOH WR HQVXUH DQ RSWLPDO UDWLR RI JLIWHG VWXGHQWV DFFXUDWHO\ SUHGLFWHG WR QRWJLIWHG VWXGHQWV LQDFFXUDWHO\ SUHGLFWHG $IWHU LWHPV IRU WKH 13 ZHUH REWDLQHG DQG PXOWLSOH FXWRII SRLQWV ZDV HVWDEOLVKHG 3KDVH ,, ZDV EHJXQ ,Q 3KDVH ,, WHVW VFRUHV EDVHG RQ VHOHFWHG 13 LWHPV ZHUH FRPn SXWHG IRU WKH VHFRQG VDPSOH RI VWXGHQWV 7KH 13 ZDV DQn DO\]HG LQ WHUPV RI LWV GLVFULPLQDWLRQ EHWZHHQ WKH JLIWHG DQG QRW JLIWHG VWXGHQWV RQ :,6&5 7KH DFFXUDF\ RI FODVn VLILFDWLRQ ZDV WKHQ FRPSDUHG WR DFFXUDF\ RI FODVVLILFDWLRQ

PAGE 61

REWDLQHG XVLQJ VWXGHQWVn 6,7 VFRUHV DV D VFUHHQLQJ SURFHGXUH LQ SUHGLFWLQJ JLIWHG ,4 RQ WKH :,6&5 'DWD FROOHFWRUV DQG UHFRUGHUV FRQVLVWHG RI WKH UHn VHDUFKHU DQG HPSOR\HHV RI WKH GLVWULFW VFKRRO ERDUG 6FKRRO ERDUG HPSOR\HHV ZRUNLQJ LQ WKH WHVWLQJ DQG HYDOXDn WLRQ RIILFH REWDLQHG &7%67&6 LWHP UHVSRQVHV IURP FRPSXWHU GDWD :,6&5 DQG 6,7 VFRUHV ZHUH REWDLQHG LQ D VLPLODU PDQQHU E\ WKH UHVHDUFKHU DQG UHVHDUFK DVVLVWDQW 'DWD $QDO\VLV $V GLVFXVVHG LQ &KDSWHU WZR FRUUHODWLRQDO DQDO\VHV ZHUH FRQGXFWHG LQ 3KDVH RI WKLV VWXG\ WR GHWHUPLQH GHn VLUDEOH LWHPV IRU WKH 13 3KL FRHIILFLHQWV ZKLFK DUH GHVLJQHG WR FRUUHODWH WZR GLFKRWRPRXV YDULDEOHV ZHUH FRPn SXWHG 7KH WZR GLFKRWRPRXV YDULDEOHV FRUUHODWHG ZHUH WKH VWXGHQWnV LWHP UHVSRQVH FRUUHFW RU LQFRUUHFWf DQG VWXGHQW FODVVLILFDWLRQ JLIWHG RU QRWJLIWHGf RQ WKH :,6&5 7KH FULWHULRQ IRU GHFLGLQJ LI DQ LWHP ZDV WR EH LQFOXGHG ZDV VLJQLILFDQFH DW WKH OHYHO )RU WKH SUHVHQW VDPSOH WKLV PHDQW WKDW DQ\ LWHP ZLWK D SKL FRHIILFLHQW JUHDWHU WKDQ ZDV VHOHFWHG IRU WKH 13 VFUHHQLQJ WHVW 3KL LV EDVHG RQ WKH SURSRUWLRQV RI FDVHV SDVVLQJ DQG IDLOLQJ DQ LWHP LQ ERWK WKH JLIWHG DQG QRWJLIWHG FULWHn ULRQ JURXSV 7KH SKL FRHIILFLHQW LV NQRZQ WR EH ELDVHG

PAGE 62

WRZDUG PLGGOH GLIILFXOW\ OHYHOV RI WHVW LWHPV $V SUHYLn RXVO\ GLVFXVVHG WKH UHVHDUFK &7%6 DQG 7&6f LWHPV ZHUH GHVLJQHG SULPDULO\ WR DVVHVV SHUIRUPDQFH RI PHGLXP GLIILn FXOW\ 7HVW LWHPV ZHUH DOVR DQDO\]HG DFFRUGLQJ WR WKH LQGH[ RI GLVFULPLQDWLRQ (EHO f 7KH GLIIHUHQFH EHWZHHQ WKH SHUFHQWDJH RI JLIWHG VWXGHQWV DQG QRWJLIWHG VWXGHQWV SDVVLQJ HDFK LWHP SURYLGHV DQ LQGH[ RI LWHP YDOLGLW\ WKDW FDQ EH LQWHUSUHWHG LQGHSHQGHQWO\ RI WKH VL]H RI WKH SDUn WLFXODU VDPSOH LQ ZKLFK LW ZDV REWDLQHG $QDVWDVL f 7KH LQGH[ RI GLVFULPLQDWLRQ 'f KDV EHHQ VKRZQ WR PHDVXUH LWHP YDOLGLW\ ZLWK HTXLYDOHQW DFFXUDF\ WR RWKHU PRUH HODEn RUDWH PHDVXUHV (QJHOKDUW f 6LPLODU WR WKH SKL YDOXHV DUH ELDVHG LQ IDYRU RI LWHPV ZLWK LQWHUPHGLDWH GLIn ILFXOW\ OHYHOV $ FRHIILFLHQW RI RU JUHDWHU ZDV XVHG DV WKH FULWHULRQ IRU VHOHFWLQJ DQ LWHP IRU WKH 13 7KXV LQ 3KDVH WZR IRUPV RI WKH 13 VFUHHQLQJ WHVW ZHUH FUHDWHG RQH IRUP KHUH GHVLJQDWHG DV 13SKL ZDV EDVHG RQ LWHPV VHOHFWHG XVLQJ SKL FRHIILFLHQWV WKH RWKHU IRUP 13 ZDV EDVHG RQ LWHPV VHOHFWHG XVLQJ WKH LQGH[ RI GLVFULPn LQDWLRQ )RU 3KDVH ,, DQDO\VHV XVLQJ WKH FRHIILFLHQW .DSSD ZHUH FRQGXFWHG WR WHVW Df ZKHWKHU WKH 13 RU 6,7 ZDV PRUH DFFXUDWH LQ FODVVLI\LQJ JLIWHG VHYHQWK JUDGHUV DQG Ef DW ZKDW FXWRII SRLQWV HLWKHU WKH 13 RU 6,7 ZDV PRUH DFFXUDWH 7KH .DSSD DQDO\VLV PHDVXUHG WKH SURSRUWLRQ RI FRUUHFW DQG

PAGE 63

LQFRUUHFWf FODVVLILFDWLRQV IRU 13SKL 13' DQG WKH 6,7 ZKLOH DGMXVWLQJ IRU WKH SHUFHQWDJH RI FRUUHFW FODVVLILFDn WLRQV WKDW FRXOG EH H[SHFWHG RQ WKH EDVLV RI FKDQFH DORQH .DSSD LV D GHVFULSWLYH VWDWLVWLF DQG QRW D WHVW RI VWDWLVn WLFDO VLJQLILFDQFH 8VLQJ WKH .DSSD VWDWLVWLF SURSRUWLRQV RI SUHGLFWLRQV ZHUH FRPSDUHG DW YDULRXV FXWRII SRLQWV .DSSD DGMXVWV IRU SUHGLFWLRQV H[SHFWHG E\ FKDQFH DORQH E\ WDNLQJ LQWR DFFRXQW ERWK REVHUYHG DQG H[SHFWHG SURSRUWLRQ FODVVLILFDn WLRQV &RKHQ f )RU H[DPSOH ZKHQ FRQVLGHULQJ YDOn XHV IRU WKH 13 WKH QXPHUDWRU RI LV UHJDUGHG DV WKH SURn SRUWLRQ RI VWXGHQWV FRQVLVWHQWO\ FODVVLILHG E\ ERWK WKH 13 DQG WKH :,6&5 REVHUYHGf RYHU DQG DERYH WKH SURGXFW RI WKH SURSRUWLRQV RI VWXGHQWV FODVVLILHG LQGLYLGXDOO\ E\ HDFK WHVW H[SHFWHGf 7KH GHQRPLQDWRU RI LV WKH PD[LPXP SRVVLEOH LQFUHDVH LQ GHFLVLRQFRQVLVWHQF\ DERYH FKDQFH OHYHO JLYHQ WKH SURSRUWLRQV FODVVLILHG E\ WKH WZR WHVWV LQGHSHQGHQWO\ 7KH IRUPXOD XVHG WR FRPSXWH .DSSD ZDV ZKHUH 3 SURSRUWLRQ RI FRQVLVWHQW JLIWHG DQG QRWJLIWHG FODVVLILFDWLRQV IRU :,6&5 DQG 13 RU :,6&5 DQG 6,7

PAGE 64

DQG 3F SURSRUWLRQ RI JLIWHG FODVVLILFDWLRQV IRU :,6&5 [ SURSRUWLRQ RI JLIWHG FODVVLILFDWLRQV IRU 13 RU 6,7f SURSRUWLRQ RI QRWJLIWHG FODVVLILFDn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n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

PAGE 65

ZRXOG VFRUH 6LQFH RQ PRVW RI WKH LWHPV VWXGHQWV LQ WKH UHVHDUFK VDPSOH ZHUH H[SHFWHG WR VFRUH ZHOO DERYH WKH ORFDWLRQ SDUDPHWHU PRVW RI WKH LWHPV SURYLGHG OLWWOH GLIn IHUHQWLDWLRQ DPRQJ WKHVH VWXGHQWV ZLWK PXFK KLJKHU VFDOH VFRUHV +RZHYHU LQ GHIHQVH RI WKHVH LQVWUXPHQWV IRU XVH LQ WKLV VWXG\ VRPH LWHPV ZLWK H[WUHPH ORFDWLRQ SDUDPHWHUV ZHUH SXUSRVHO\ LQFOXGHG LQ WKH &7%6 E\ LWV FRQVWUXFWRUV 7KHVH LWHPV ZHUH ZHOO DERYH RU EHORZ WKH UDQJH RI SHUIRUn PDQFH IRU ZKLFK WKH WHVW ZDV GHVLJQHG ,WHP FKDUDFWHULVn WLF FXUYHV LQGLFDWH WKH H[LVWHQFH RI &7%6 LWHPV WKDW ZHUH SDVVHG W\SLFDOO\ RQO\ E\ VWXGHQWV ZLWK YHU\ KLJK WRWDO VFRUHV RQ WKH WHVW &7%0F*UDZ+LO f ,57 LWHP ORn FDWLRQ SDUDPHWHUV IRU WKH 7&6 KDG QRW EHHQ FDOFXODWHG WKRXJK VLQFH WKH 7&6 ZDV GHVLJQHG WR SUHGLFW DFKLHYHPHQW IRU WKH JHQHUDO SRSXODWLRQ LW PLJKW EH DVVXPHG WKDW ORFDn WLRQ SDUDPHWHU FULWHULD ZRXOG DGKHUH WR D VLPLODU UDWLRQn DOH $ VHFRQG OLPLWDWLRQ RI WKLV VWXG\ FRQFHUQV PHDVXUHn PHQW HUURU GXH WR WKH YDULDEOH DGKHUHQFH WR WHVW VWDQGDUGn L]DWLRQ E\ DGPLQLVWUDWRUV RI WKH &7%6 7&6 DQG WKH 6,7 %HFDXVH WHDFKHUV DQG FRXQVHORUV UHFHLYH YDU\LQJ GHJUHHV RI IRUPDO WUDLQLQJ RQ WKH LPSRUWDQFH RI SUHFLVH DGKHUHQFH WR WHVW VWDQGDUGV DQG RQ WKH LQIOXHQFH RI VWDQGDUGL]DWLRQ GHn YLDWLRQV RQ WHVW UHOLDELOLW\ DGPLQLVWUDWRUVn VWULFW FRQn IRUPLW\ WR WHVW VWDQGDUGV ZDV TXHVWLRQDEOH %HFDXVH YHU\ OLWWOH WUDLQLQJ ZDV SURYLGHG IRU WKH &7%67&6 DQG 6,7

PAGE 66

PLVLQWHUSUHWDWLRQ RI LQVWUXFWLRQV RU VWDQGDUGV ZDV SRVn VLEOH DV ZHOO $ UHODWHG VRXUFH RI SRWHQWLDO PHDVXUHPHQW HUURU H[LVWHG GXH WR WKH JURXS IRUPDW E\ ZKLFK WKH &7%67&6 LV DGPLQLVWHUHG $V RSSRVHG WR WKH :,6&5 DQG 6,7 DGPLQLVn WUDWRUV ZHUH UHVWULFWHG LQ WKHLU DELOLW\ WR FORVHO\ PRQn LWRU LQGLYLGXDO VWXGHQWV DQG FRQWURO IRU VXFK IDFWRUV DV PLVLQWHUSUHWHG GLUHFWLRQV RU DFXWH SK\VLFDO RU HPRWLRQDO OLDELOLWLHV RI VWXGHQWV 6LQFH UHJXODU &7%67&6 WHVWLQJ ZDV FRQGXFWHG RQO\ RQFH D \HDU LQ WKH UHVHDUFK VFKRRO GLVn WULFW HIIRUWV ZHUH PDGH WR WHVW DV PDQ\ VWXGHQWV DV SRVn VLEOH GXULQJ WKDW WLPH +RZHYHU 6,7 DQG :,6&5 WHVWLQJ ZHUH PRUH HDVLO\ SRVWSRQHG WR D ODWHU GDWH LI D VLWXDWLRQ ZDUUDQWHG VXFK DFWLRQ $QRWKHU OLPLWDWLRQ RI WKH &7%6 VSHFLILF WR WKLV VWXG\ ZDV WKDW VRPH VWXGHQWV ZHUH QRW DGPLQLVWHUHG WKH 6FLHQFH DQG 6RFLDO 6WXGLHV VXEWHVWV LQ WKH VFKRRO GLVWULFW EHn FDXVH WKH\ ZHUH RSWLRQDO DQG ZHUH DGPLQLVWHUHG RQO\ LQ VRPH VFKRROV 7KHVH VXEWHVWV DUH FRPSULVHG RI LWHPV HDFK DQG PD\ FRQWDLQ LWHPV XVHIXO IRU GLVFULPLQDWLRQ RI JLIWHG DQG QRWJLIWHG VWXGHQWV ,I GHOHWHG IURP DQDO\VLV WKHVH LWHPV PD\ GHWUDFW IURP WKH RYHUDOO DFFXUDF\ RI WKH SURSRVHG QHZ SURFHGXUH LQ DFFRPSOLVKLQJ LWV LQWHQGHG JRDO 7KH DGPLQLVWUDWLRQ RI WKH :,6&5 DQG 6,7 RYHU D \HDU SHULRG VXJJHVWV D TXHVWLRQ UHJDUGLQJ VFRUH

PAGE 67

HTXLYDOHQFH 2QH PLJKW DUJXH WKDW VLQFH WKH LWHPV DGPLQn LVWHUHG WR D \HDUROG DQG DQ \HDUROG DUH GLIIHUHQW DQG UHSUHVHQW GLIIHUHQW WHVW GLIILFXOW\ OHYHOV WKH WZR VWXGHQWV DUH LQ IDFW EHLQJ WHVWHG RQ GLIIHUHQW VFDOHV 7KLV DUJXPHQW LPSOLHV WKDW JLIWHG LQWHOOLJHQFH IRU WKH \HDUROG LV QRW HTXLYDOHQW WR JLIWHG LQWHOOLJHQFH IRU WKH \HDUROG n7R WKH FRQWUDU\ ZKLOH LPSHUIHFWLRQV LQ WHVW VWDELOLW\ ZLOO FDXVH VRPH IOXFWXDWLRQ LQ ,4 RYHU WLPH WKH DGRSWLRQ RI WKH GHYLDWLRQ ,4 :HFKVOHU f SHUPLWV FRPn SDULVRQ RI VFRUHV RYHU DJH OHYHOV

PAGE 68

&+$37(5 ,9 5(68/76 7KLV FKDSWHU LV SUHVHQWHG LQ D IRUPDW WKDW VHTXHQn WLDOO\ UHIOHFWV WKH PHWKRGRORJLFDO SURJUHVVLRQ RI WKH VWXG\ 3KDVH ZLOO EH GLVFXVVHG ILUVW WR SUHVHQW UHVXOWV RI WZR DQDO\VHV XVHG WR VHOHFW LWHPV IRU WKH QHZ VFUHHQLQJ SURFHGXUH 13f 1H[W 3KDVH ,, FURVV YDOLGDWLRQ UHVXOWV DUH SUHVHQWHG LQ WHUPV RI SUHGLFWLRQ DFFXUDF\ RI WKH 13 SKL 13' DQG 6,7 LQ UHODWLRQ WR FXWRII VFRUHV 3KDVH f§ ,WHP 6HOHFWLRQ 3KL &RHIILFLHQWV 7KH SKL FRHIILFLHQW DQDO\VLV RI WKH &7%6 DQG 7&6 LWHPV \LHOGHG LWHPV WKDW GLVFULPLQDWHG WKH JLIWHG DQG QRWJLIWHG VHYHQWK JUDGHUV 7KLV QXPEHU RI LWHPV LV RYHU WZLFH DV PDQ\ DV ZRXOG EH H[SHFWHG E\ FKDQFH DW WKH e OHYHO RI FRQILGHQFH 7KHUH ZHUH LWHPV VLJQLIn LFDQW DW WKH e OHYHO DQG WKUHH VLJQLILFDQW DW e 7DEOH f $V DOVR VKRZQ LQ 7DEOH DOO &7%6 DQG 7&6 VXEWHVWV FRQWULEXWHG LWHPV H[FHSW IRU &7%6 5HIHUHQFH 6NLOOV DQG 7&6 6HTXHQFHV 1RW DOO RI WKH VLJQLILFDQW SKL DQDO\VHV LWHPV ZHUH UHWDLQHG IRU XVH LQ 3KDVH ,, ,Q ILYH FDVHV WKH SKL LWHPV

PAGE 69

7DEOH ,WHP 3KL 9DOXHV DQG 6LJQLILFDQFH /HYHOV ,WHP 1R 6XEWHVW 3KL9DOXH 6LJQLILFDQFH &7%69RF e 9RF 9RF e 9RF e 5HDG e 5HDG e 5HDG e 5HDG e 5HDG e 5HDG e 5HDG e 5HDG e 6SHOO e 6SHO e 6SHOO e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e /DQJ e 0DWK e 0DWK f r 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e 0DWK e f 6FL e

PAGE 70

7DEOH &RQWLQXHG ,WHP 1R 6XEWHVW 3KL9DOXH 6LJQLILFDQFH rf 6RF 6W f 6RF 6W f 6RF 6W f 6RF 6W 7&6$QDO r 0HP 0HP 0HP 0HP r 0HP r 9HUE 5HDV 9HUE 5HDV r ,WHP GLVFULPLQDWHV LQ IDYRU RI QRWJLIWHG $QVZHUHG E\ OHVV WKDQ b RI VDPSOH

PAGE 71

GLVFULPLQDWHG LQ IDYRU RI WKH QRWJLIWHG VWXGHQWV 7KDW LV QRWJLIWHG VWXGHQWV UHVSRQGHG FRUUHFWO\ WR WKH LWHPV PRUH IUHTXHQWO\ WKDQ WKH JLIWHG VWXGHQWV $OVR GHOHWHG IURP 3KDVH ,, DQDO\VHV ZHUH ILYH LWHPV UHVSRQGHG WR E\ RQO\ DERXW KDOI RI WKH 3KDVH VDPSOH 7KHVH ILYH LWHPV ZHUH ORFDWHG LQ WKH 6FLHQFH DQG 6RFLDO 6WXGLHV VXEWHVWV ZKLFK ZHUH DGPLQLVWHUHG RQ DQ RSWLRQDO EDVLV DW WKH GLVn FUHWLRQ RI WKH YDULRXV VFKRROV LQYROYHG 7KH\ ZHUH GHn OHWHG EHFDXVH WKH FXWRII VFRUHV ZHUH VHW EDVHG RQ WRWDO LWHPV DGPLQLVWHUHG WR DOO VWXGHQWV 7KRVH LWHPV QRW DGn PLQLVWHUHG WR DOO VWXGHQWV ZHUH GHOHWHG VR WKDW DOO VWXn GHQWV FRXOG SRWHQWLDOO\ DWWDLQ WKH PD[LPXP UDZ VFRUH 7KHUHIRUH SKL LWHPV ZHUH UHWDLQHG IRU XVH LQ 3KDVH ,, IURP WKH &7%6 DQG IURP WKH 7&6 7KH PHDQ RI WKH WRWDO VFRUHV IRU WKH LWHPV VHn OHFWHG E\ WKH SKL DQDO\VHV ZDV 7KH PHGLDQ DQG PRGH ZHUH VRPHZKDW KLJKHU DQG UHVSHFWLYHO\ 7KH VWDQGDUG GHYLDWLRQ ZDV DQG WKH UDQJH ZDV 0RVW VWXGHQWV SHUIRUPHG ZHOO RQ WKHVH LWHPV ZLWK PRVW UHVSRQGn LQJ FRUUHFWO\ WR QHDUO\ DOO RI WKHP 7KLV KRPRJHQHLW\ RI VFRUHV LV UHIOHFWHG E\ WKH QHJDWLYH VNHZ RI WKH GLVWULEXn WLRQ (YHQ WKRXJK VFRUHV DSSHDUHG WR FRQFHQWUDWH QHDU WKH XSSHU HQG RI WKH GLVWULEXWLRQ WKH ODUJH UDQJH RI VFRUHV LH WKURXJK f FRQWULEXWHG WR D VWDQGDUG GHYLDWLRQ RI DGHTXDWH VL]H

PAGE 72

,QGH[ RI 'LVFULPLQDWLRQ 7KH LQGH[ RI GLVFULPLQDWLRQ DQDO\VHV ZKLFK PHDVXUHG WKH GLIIHUHQFH EHWZHHQ WKH SHUFHQWDJH RI JLIWHG DQG QRW JLIWHG VWXGHQWV SDVVLQJ HDFK LWHP \LHOGHG D VXEVHW RI LWHPV ZLWK YDOXHV RI RU JUHDWHU $V UHFRPPHQGHG E\ (QJHOKDUW f LWHPV SRVVHVVLQJ D YDOXH RI DW OHDVW DUH FRQVLGHUHG WR VKRZ DGHTXDWH GLVFULPLQDWLRQ $OO RI WKH DFFHSWDEOH LWHPV ZHUH REWDLQHG IURP WKH &7%6 ,Q 7DEOH WKH XSSHU DQG ORZHU YDOXHV UHSUHVHQW SHUFHQWDJH SDVVLQJ HDFK LWHP IRU WKH JLIWHG DQG QRWJLIWHG VWXGHQWV UHVSHFWLYHO\ YDOXHV UDQJHG IURP WKURXJK $V PLJKW EH H[SHFWHG WKH KLJKHVW SHUFHQWDJH RI FRUUHFW UHVSRQGLQJ RFFXUUHG ZLWK WKH EHJLQQLQJ 7&6 VXEWHVW LWHPV 7KLV ZDV EHFDXVH 7&6 LWHPV DUH RUGHUHG KLHUDUFKn LFDOO\ E\ VXEWHVW DFFRUGLQJ WR GLIILFXOW\ OHYHO ,Q WKLV DQDO\VLV DOVR DQ LWHP ZDV GHOHWHG LI LW GLVFULPLQDWHG LQ IDYRU RI WKH QRWJLIWHG JURXS RU LI LW ZDV QRW DGPLQLVWHUHG WR RYHU b RI WKH VDPSOH ,Q FRQn WUDVW WR WKH SKL DQDO\VHV D ODUJH SURSRUWLRQ RI LWHPV LH RU bf ZHUH GHOHWHG )RXUWHHQ LWHPV 7DEOH f ZHUH UHWDLQHG IURP WKH &7%6 ,Q FRQWUDVW ZLWK WKH SKL GDWD UHVXOWV RI VFRUHV EDVHG RQ VHOHFWHG LWHPV IURP WKH DQDO\VHV UHYHDOHG D UHODWLYHO\ QRUPDO GLVWULEXWLRQ ZLWK WKH PHDQ RI PHGLDQ RI DQG PRGH RI IDOOLQJ ZLWKLQ D UDQJH RI RQH WHVW LWHP 7KH VWDQGDUG GHYLDWLRQ ZDV DQG WKH

PAGE 73

7DEOH ,WHPV ZLWK 9DOXHV DQG &RUUHVn SRQGLQJ 6XEWHVWV ,WHP 1R 6XEWHVW 8SSHU [ /RZHU [ &7%6 9RF 5HDG 5HDG 6SHOO /DQJ /DQJ /DQJ 0DWK 0DWK 0DWK 0DWK 0DWK f 0DWK f 6FL 6FL f 6FL f 6RF 6W f 6RF 6W 6RF 6W f 6RF 6W f 6RF 6W f 6RF 6W f 6RF 6W f 6RF 6W r,WHP GLVFULPLQDWHV LQ IDYRU RI QRWJLIWHG DQVZHUHG FRUUHFW E\ OHVV WKDQ b RI WKH VDPSOH

PAGE 74

UDQJH ZDV 8QOLNH WKH SKL GLVWULEXWLRQ WKH FHLOLQJ RI WKH GLVWULEXWLRQ ZDV SUREDEO\ VXIILFLHQWO\ KLJK EHFDXVH FRUUHFW UHVSRQGLQJ GLPLQLVKHG EH\RQG UDZ VFRUHV RI 2QO\ RI VWXGHQWV UHFHLYHG D UDZ VFRUH DERYH ZKLOH UHFHLYHG VFRUHV ZLWKLQ RQH SRLQW RI WKH PHDQ 7DNLQJ LQWR DFFRXQW WKH VPDOO QXPEHU RI 13' LWHPV f WKH VWDQGDUG GHYLDWLRQ RI LV FRQVLGHUHG WR EH DGHn TXDWHO\ ODUJH IRU FDOFXODWLQJ FXWRIIV 7KH UDQJH RI VFRUHV FRYHUHG ERWK H[WUHPHV RI WKH GLVWULEXWLRQ DQG WKH VNHZ LV QRW DV JUHDW DV PLJKW EH H[SHFWHG JLYHQ WKH JHQn HUDOO\ UHVWULFWHG UDQJH RI WKH VDPSOH DW WKH XSSHU DELOLW\ HYHOV &RQVLVWHQW ZLWK SUHYLRXV UHVHDUFK HJ (QJHOKDUW f WKHUH ZHUH FRPPRQDOLWLHV EHWZHHQ WKH SKL DQG DQn DO\VHV UHVXOWV IRU WKLV VDPSOH 6L[WHHQ LWHPV RU b RI WKH DQG b RI WKH SKL LWHPV ZHUH VHOHFWHG IURP ERWK DQDO\VHV $OO FRPPRQ LWHPV ZHUH IURP WKH &7%6 RQ DFKLHYHPHQW WHVWf EHFDXVH WKH DQDO\VHV \LHOGHG QR 7&6 FRJQLWLYH DELOLW\f 13 LWHPV 6LPLODUO\ WKH RQH DQDO\n VLV LWHP IRXQG WR GLVFULPLQDWH LQ IDYRU RI WKH QRWJLIWHG VWXGHQWV DOVR GLG VR RQ WKH SKL DQDO\VLV &XWRII 6FRUHV &XWRII VFRUHV IRU WKH SKL DQG 6,7 DQDO\VHV ZHUH FRPSXWHG LQ RUGHU WR GHPDUFDWH RSWLPDO FXWWLQJ SRLQWV IRU

PAGE 75

GLIIHUHQWLDWLQJ WKH JLIWHG IURP QRWJLIWHG VWXGHQWV &XWn RII VFRUHV ZHUH HVWDEOLVKHG VXFK WKDW D PD[LPXP QXPEHU RI JLIWHG DQG D PLQLPXP QXPEHU RI QRWJLIWHG VWXGHQWV ZRXOG IDOO DERYH WKH FXWRII 7KHUHIRUH WKH WUXH SRVLWLYH DQG WUXH QHJDWLYH ILQGLQJV ZHUH PD[LPL]HG ZKLOH WKH IDOVH SRVn LWLYHV DQG IDOVH QHJDWLYHV ZHUH PLQLPL]HG 7KLV SURFHGXUH UHTXLUHG D YDOXH MXGJPHQW E\ WKH UDWHU DV WR WKH DPRXQW RI HUURU ZKLFK ZRXOG EH DFFHSWDEOH $V WKH FXWRII LV ORZn HUHG WR SHUPLW PRUH JLIWHG VWXGHQWV WR H[FHHG LW LQFUHDVn LQJ QXPEHUV RI QRWJLIWHG VWXGHQWV ZRXOG DOVR H[FHHG LW WKXV LQFUHDVLQJ WKH SRVVLELOLW\ RI IDOVH QHJDWLYHV 7R DVVLVW ZLWK WKLV SUREOHP PXOWLSOH FXWRII VFRUHV ZHUH FDOn FXODWHG VR WKDW WKH PRVW GHVLUDEOH FXWRIIV FRXOG EH GHWHUn PLQHG &XWRII VFRUHV ZHUH FKRVHQ DFFRUGLQJ WR IUDFWLRQV RI VWDQGDUG GHYLDWLRQ XQLWV IRU WKH WRWDO VFRUH GLVWULEXn WLRQV RQ WKH 6,7 WKH QHZ WHVW FUHDWHG E\ VHOHFWLQJ LWHPV ZLWK SKL DQG WKH QHZ WHVW FUHDWHG E\ VHOHFWLQJ LWHPV ZLWK 7KH PHDQ 6,7 ,4 IRU WKH 3KDVH VDPSOH ZDV DQG WKH VWDQGDUG GHYLDWLRQ ZDV 7KLV PHDQ ,4 ZDV DSn SUR[LPDWHO\ SRLQWV KLJKHU WKDQ WKDW RI WKH JHQHUDO SRSn XODWLRQ ,W UHIOHFWV WKH DERYH DYHUDJH JHQHUDO LQWHOOLn JHQFH RI WKH UHVHDUFK VDPSOH ,Q 3KDVH ,, WKHVH VWDQGDUG GHYLDWLRQ FXWRII VFRUHV FDOFXODWHG DV VWDQGDUG GHYLDWLRQV IURP WKH 3KDVH VDPSOH ZHUH DSSOLHG WR WKH FURVV YDOLGDn WLRQ VDPSOH WR GHWHUPLQH WKH UHODWLYH DFFXUDF\ RI JLIWHG FODVVLILFDWLRQV XVLQJ WKH 13SKL 13' DQG 6,7

PAGE 76

3KDVH ,,&URVV 9DOLGDWLRQ $SSOLFDWLRQ RI &XWRII 6FRUHV 7KH ILUVW VWHS LQ WHVWLQJ WKH 13 LWHPV REWDLQHG LQ 3KDVH ZDV WR FDOFXODWH WKH QXPEHUV RI 13 LWHPV VFRUHG FRUUHFWO\ IRUWKH 3KDVH ,, VDPSOH DQG WKHQ WR DSSO\ DSSURn SULDWH FXWRII SRLQWV WR WKRVH VFRUHV 7KH 3KDVH ,, VXEn VDPSOH FRQVLVWHG RI VWXGHQWV 3URSRUWLRQV RI FRUUHFW FODVVLILFDWLRQV DPRQJ WKH 13SKL 13' DQG 6,7 LWHPV ZHUH WKHQ DQDO\]HG WR GHWHUPLQH WKH WHVW DQG FXWRII WKDW PRVW DFFXUDWHO\ GLIIHUHQWLDWHG WKH JLIWHG DQG QRWJLIWHG VWXn GHQWV 7DEOHV DQG VKRZ WKH QXPEHUV RI H[DPLQn HHV FODVVLILHG LQWR SUHGLFWLRQ FDWHJRULHV DW YDULRXV FXWn RII SRLQWV ZKHUH WKH H[DPLQHHnV WUXH VWDWXV ZDV GHILQHG DV FODVVLILFDWLRQ RI JLIWHG RU QRWJLIWHG DFFRUGLQJ WR WKH :,6&5 )RU DOO WKUHH FODVVLILFDWLRQ SURFHGXUHV 13SKL 13' DQG 6,7f OLWWOH YDULDELOLW\ H[LVWHG EHWZHHQ SUHGLFn WLRQ UDWHV DW VRPH RI WKH FXWRIIV )RU H[DPSOH FODVVLIn LFDWLRQV DW WKH DQG GLIIHUHG RQO\ VOLJKWO\ IURP WKRVH DW WKH DQG FXWRIIV UHVSHFWLYHO\ $ VLPn LODU UHVXOW RFFXUUHG IRU WKH FXWRII FRPSDUHG WR WKH FXWRII DQG IRU WKH FXWRII FRPSDUHG WR WKH FXWRII

PAGE 77

7DEOH 3KDVH ,, 1XPEHU RI ([DPLQHHV LQ (DFK 3UHGLFWLRQ &DWHJRU\ E\ &XWReI3KL 7UXH 3RVLWLYH 7UXH 1HJDWLYH 7UXH 7RWDO )DOVH 1HJDWLYH )DOVH 3RVLWLYH )DOVH 7RWDO &XWRII ‘ 6' 6' 6' 6' 6' 6' 6' 6' 6' 7RWDO JLIWHG 7RWDO QRWJLIWHG

PAGE 78

7DEOH 3KDVH ,, 1XPEHU RI ([DPLQHHV LQ (DFK 3UHGLFWLRQ &DWHJRU\ E\ &XWRII,QGH[ RI 'LVFULPLQDWLRQ 7UXH 3RVLWLYH 7UXH 1HJDWLYH 7UXH 7RWD )DOVH 1HJDWLYH )DOVH 3RVLWLYH )DOVH 7RWDO &XWRII 6' 6' 6' 6' 6' 6' 6' 6' 6'

PAGE 79

7DEOH 3KDVH ,, 1XPEHU RI ([DPLQHHV LQ (DFK 3UHGLFWLRQ &DWHJRU\ &XWRII6,7 &XWn RII 7UXH 3RVLWLYH 7UXH 1HJDWLYH 7UXH 7RWDO )DOVH 1HJDWLYH )DOVH 3RVLWLYH )DOVH 7RWDO 6' 6' 6' 6' 6' 6' 6' 6' 6'

PAGE 80

7DEOH ZDV FRQVWUXFWHG WR FRQVROLGDWH DQG FODULI\ WKH GDWD RQ SUHGLFWLYH DFFXUDF\ LQ WKDW WDEOH SURSRUn WLRQV RI FRUUHFW DQG LQFRUUHFW SUHGLFWLRQV DUH DJDLQ H[SUHVVHG DV WUXH SRVLWLYHV 73f WUXH QHJDWLYHV 71f IDOVH SRVLWLYHV )3f DQG IDOVH QHJDWLYHV )1f 'HOHWHG ZHUH WKH FXWRIIV 6' 6' 6' DQG 6' ,QVSHFWLRQ RI 7DEOH UHYHDOV WKDW DOO WKUHH PHWKRGV RI VFUHHQLQJ JLIWHG VWXGHQWV ZHUH DSSUR[LPDWHO\ HTXLYDOHQW 7KDW LV DOO WKUHH IRXQG WKH JUHDWHVW SURSRUWLRQV RI 73 SUHGLFWLRQV LQ WKURXJK RI WKH VDPSOH /RZHVW 71 SUHGLFWLRQV RFFXUUHG LQ WKURXJK RI WKH VDPSOH $IWHU WKH SUHGLFWLRQ UDWHV ZHUH HVWDEOLVKHG IRU WKH WKUHH WHVWV DQG WKH SUHGLFWLRQ YDOXHV ZHUH H[SUHVVHG DV SURSRUWLRQV RI WKH VDPSOH DQDO\VHV ZHUH FRQGXFWHG WR GHn WHUPLQH WKH SURSRUWLRQV LI DQ\ WKDW ZHUH PRVW DFFXUDWH &RHIILFLHQW .DSSD DQDO\VHV ZHUH FDOFXODWHG WR DVVHVV WKH UHODWLYH JLIWHG SUHGLFWLRQ DFFXUDF\ RI WKH 13SKL 13' DQG 6,7 .DSSD &RPSDULVRQV ,Q UHVXOWV VKRZQ LQ 7DEOH 3 YDOXHV FRUUHVSRQG WR SHUFHQWDJHV RI FRUUHFW FODVVLILFDWLRQV ERWK WUXH SRVn LWLYHV DQG WUXH QHJDWLYHVf +RZHYHU WZR HTXDO FODVVLILn FDWLRQ SHUFHQWDJHV HJ DV REVHUYHG IRU SKL DQG DW FXWRIIf PD\ QRW KDYH HTXDO FRUUHVSRQGLQJ .YDOXHV EHFDXVH WKH .DSSD HTXDWLRQ DGMXVWV IRU FKDQFH FRUUHFW SUHGLFWLRQV

PAGE 81

7DEOH 3KDVH ,, 3URSRUWLRQV RI &RUUHFW DQG ,QFRUUHFW 3UHGLFWLRQV IRU 13 SKLf 13 'f DQG 6,7 DW 7ZR 6FRUH &XWRIIV ] &XWRII @ 7UXH 3RVLWLYH 7UXH 1HJDWLYH 7UXH 7RWD )DOVH 1HJDWLYH )DOVH 3RVLWLYH )DOVH 7RWDO 3KL 6' 6' 6' 6' /f 6' 6' 6' 6' 6' 6' n 6,7 6' 6' 6' 6' 6' 1RWH $OO YDOXHV URXQGHG WR QHDUHVW SHUFHQWLOH

PAGE 82

7DEOH 3HUFHQWDJH RI &RUUHFW &ODVVLILFDWLRQV DQG .DSSD 9DOXHV IRU 13 SKL DQG 'f DQG 6,7 DW 9DULRXV &XWRII 6FRUHV &XWRII 13 6,73r 13 6,7N 3KL3r '3r 3KLNA 'N 6' 6' 6' 6' 6' r3 SHUFHQWDJH LN .DSSD YDOXH

PAGE 83

$Q H[DPSOH RI WKH .DSSD FDOFXODWLRQ RI SHUFHQWDJHV RI FRUUHFW FODVVLILFDWLRQV IRU 13SKL DW WKH FXWRII LV LOOXVWUDWHG XVLQJ IRXUIROG WDEOHV )LJXUH f 9DOXHV LQ WKH HTXDWLRQ DORQJ ZLWK SHUFHQWDJHV RI H[DPLQHH FODVn VLILFDWLRQV DW RWKHU FXWRIIV DUH ORFDWHG LQ 7DEOH 3 3F f f f f 5HVXOWV RI WKH .DSSD DQDO\VHV UHYHDOHG D FRQVLVWHQW SDWWHUQ RI JUHDWHU YDOXHV IRU WKH 13' WKDQ IRU WKH 13SKL RU WKH 6,7 DW DOO FXWRIIV VXJJHVWLQJ WKH 13' PD\ EH WKH PRUH DFFXUDWH SUHGLFWRU RI FRUUHFW FODVVLILFDWLRQ 7KH JUHDWHVW YDOXH IRU 13' RFFXUV DW WKH ZKHUH WKH YDOXH RI LQGLFDWHV WKDW WKHUH LV D b LPSURYHPHQW LQ SUHGLFWLRQ DFFXUDF\ RYHU WKDW H[SHFWHG E\ FKDQFH 7KH JUHDWHVW GLVFUHSDQF\ EHWZHHQ YDOXHV IRU 13' DQG WKH 6,7 DSSHDUV DW WKH FXWRII ZKHUH IRU 13' EXW RQO\ IRU WKH 6,7

PAGE 84

,4 &ODVV &ODVV *LIWHG 1RW*LIWHG &ODVV *LIWHG f SKL &ODVV 1RW*LIWHG 73 )3 )1 71 77V7 )LJXUH &RPSXWDWLRQ ([DPSOH RI .DSSD DQDO\VLV IRU 7UXH 3KL &ODVVLn ILFDWLRQV DW 6' &XWRII

PAGE 85

6XPPDU\ RI 5HVXOWV ,Q 3KDVH RI WKLV VWXG\ LWHPV IURP WKH &7%67&6 ZHUH LGHQWLILHG XVLQJ SKL FRHIILFLHQWV WKDW VLJQLILFDQWO\ GLIIHUHQWLDWHG JLIWHG IURP QRWJLIIHG VWXGHQWV 7KH LQGH[ RI GLVFULPLQDWLRQ DQDO\VHV \LHOGHG LWHPV WKDW VXFFHVVn IXOO\ GLVFULPLQDWHG JLIWHG IURP QRWJLIWHG $ ODUJH SURn SRUWLRQ RIn VLJQLILFDQW LWHPV ZHUH FRPPRQ WR WKH SKL DQDO\VLV UHVXOWV ,Q 3KDVH ,, WRWDO VFRUHV IRU HDFK VXEVHW RI LWHPV ZHUH FRPSXWHG XVLQJ D FURVV YDOLGDWLRQ VDPSOH RI H[DPn LQHHV 7KH DFFXUDF\ RI FODVVLILFDWLRQV UHVXOWLQJ IURP HDFK RI WKH WZR QHZ WHVWV DQG WKH 6,7 ZHUH FRQWUDVWHG DW ILYH GLIIHUHQW FXWRII VFRUHV .DSSD DQDO\VHV UHYHDOHG KLJKHU FRUUHFW SUHGLFWLRQ YDOXHV IRU WKH 13' WKDQ IRU HLWKHU WKH 13SKL DQG WKH 6,7 DW DOO FXWRII OHYHOV

PAGE 86

&+$37(5 9 ',6&866,21 5HVHDUFK 4XHVWLRQV ,Q RUGHU WR DQVZHU WKH ILUVW UHVHDUFK TXHVWLRQ WKLV GLVFXVVLRQ IRFXVHV RQ LVVXHV UHODWHG WR 13 LWHP YDOLGLW\ IURP 3KDVH 3KDVH ,, 13 LWHPV ZHUH VHOHFWHG IURP WKH &7%6 DQG 7&6 XVLQJ FRUUHODWLRQDO DQDO\VHV 7R DQVZHU WKH UHPDLQLQJ WZR UHVHDUFK TXHVWLRQV WKH SUHGLFWLYH DFFXUDn FLHV RI WKH 13SKL 13' DQG 6,7 DUH GLVFXVVHG LQ WHUPV RI FXWRII VFRUHV 7KH IROORZLQJ UHVHDUFK TXHVWLRQV ZHUH DGGUHVVHG LQ WKLV VWXG\ &DQ DQ DFFXUDWH SUHGLFWRU RI JLIWHG ,4 FODVVLILn FDWLRQ RQ WKH :,6&56% EH GHULYHG IURP DQ LQn VWUXPHQW FRPSRVHG RI LWHPV RQ WKH &7%6 DQG 7&6 LQ D VLWXDWLRQ LQ ZKLFK JLIWHGQHVV LV YLHZHG DV D GLFKRWRPRXV WUDLW YDULDEOH" ,V WKH 13 PRUH DFFXUDWH WKDQ WKH 6,7 LQ FODVVLn I\LQJ JLIWHG DQG QRWJLIWHG VHYHQWK JUDGHUV" $W ZKDW FXWRII SRLQWVf LV WKH 13 PRUH DFFXUDWH WKDQ WKH 6,7 LQ FODVVLI\LQJ JLIWHG DQG QRW JLIWHG VHYHQWK JUDGHUV"

PAGE 87

3KDVH ,Q 3KDVH RI WKH VWXG\ LW ZDV VXJJHVWHG WKDW VRPH LWHPV GHULYHG IURP WKH &7%67&6 ZHUH YDOLG IRU GLVWLQn JXLVKLQJ EHWZHHQ JLIWHG DQG QRWJLIWHG VHYHQWK JUDGHUV (YLGHQFH IRU WKLV DVVHUWLRQ LQFOXGHG Df WKH UHODWLYHO\ ODUJH QXPEHURI 13 LWHPV WKDW DUH FRPPRQ WR ERWK WKH SKL DQG DQDO\VHV Ef WKH ODUJH QXPEHU RI LWHPV IRXQG WR GLVFULPLQDWH EHWZHHQ WKH WZR JURXSV DQG Ff WKH DJUHHPHQW EHWZHHQ &7%6 DQG ,57 ORFDWLRQ SDUDPHWHU HVWLPDWHV DQG 13 LWHP VWDWLVWLFV &RPPRQ ,WHPV 7KH LQGH[ RI GLVFULPLQDWLRQ 'f DQDO\VHV \LHOGHG IHZHU GLVFULPLQDWLQJ LWHPV WKDQ GLG WKH SKL DQDO\VHV +RZHYHU WKH YDOLGLW\ RI WKH DQDO\VHV LV VXSSRUWHG E\ WKH RYHUODS RI LWHPV EHWZHHQ WKH SKL DQG DQDO\VHV 2I WKH LWHPV UHWDLQHG IRU WKH 13' LWHPV ZHUH DOVR LQn FOXGHG LQ WKH SKL 13 VFDOH 2YHUODS EHWZHHQ SKL DQG LWHP GLVFULPLQDWLRQ IRU PLGGOH GLIILFXOW\ LWHPV ZDV UHn SRUWHG E\ (QJHOKDUW f 6XSSRUW IRU WKH YDOLGLW\ RI WKH 13 DOVR ZDV VXJJHVWHG E\ WKH UHODWLYHO\ ODUJH QXPEHU RI VLJQLILFDQW LWHPV SDUn WLFXODUO\ RQ WKH SKL DQDO\VHV ,Q DGGLWLRQ WR WKH VLJn QLILFDQW SKL LWHPV LQLWLDOO\ FRPSXWHG DW c DQRWKHU ZHUH VLJQLILFDQW DW WKH OHYHO &OHDUO\ DQ HYHQ ODUJHU QXPEHU RI VLJQLILFDQW LWHPV ZRXOG KDYH UHVXOWHG KDG WKH

PAGE 88

VDPSOH VL]H RI UHPDLQHG FRQVWDQW +RZHYHU WKH VDPSOH VL]H GLPLQLVKHG EHORZ RQ LWHPV 2I WKHVH LWHPV KDG SKL YDOXHV WKDW ZRXOG KDYH EHHQ VLJQLILFDQW DW KDG WKH 1 EHHQ DV KLJK DV ,57 3DUDPHWHUV $ VWURQJ FDVH IRU WKH 13 LWHP YDOLGLW\ LV PDGH XSRQ LQYHVWLJDWLRQ RI LWHP UHVSRQVH WKHRU\ LWHP ORFDWLRQ SDUDPn HWHUV IRU WKH &7%6 LWHPV $Q LQYHVWLJDWLRQ RI WKH UHVSRQVH SDWWHUQV RI VXEMHFWV RQ ERWK LWHP VHWV LQ FRPn SDULVRQ ZLWK WKH UHVSRQVH SDWWHUQ RQ WKHVH LWHPV E\ WKH RULJLQDO &7%6 VDPSOH VXJJHVW FRPPRQDOLW\ LQ UHVSRQGLQJ ,57 LWHP ORFDWLRQ SDUDPHWHU ,WHP UHVSRQVH WKHRU\ ZDV XWLOL]HG E\ WKH &7%67&6 FRQVWUXFWRUV IRU LWHP VHOHFn WLRQ ,WHP FKDUDFWHULVWLF FXUYHV ZHUH SORWWHG XVLQJ D WKUHH SDUDPHWHU ORJLVWLF PRGHO LQYROYLQJ Df LWHP GLVn FULPLQDWLRQ Ef LWHP ORFDWLRQ RU GLIILFXOW\f DQG Ff D JXHVVLQJ IDFWRU $ ORFDWLRQ SDUDPHWHU GHVFULEHV DQ LWHPnV GLIILFXOW\ LQ WHUPV RI WKH VWXGHQWnV DELOLW\ OHYHO RU ODWHQW WUDLWf $Q LWHP GLVFULPLQDWHV EHVW IRU D VWXn GHQW ZKRVH DELOLW\ OHYHO LV QHDU WKH LWHPnV ORFDWLRQ SDUDPHWHU RU GLIILFXOW\ OHYHOf $Q LWHP ZLWK D KLJK ORFDWLRQ SDUDPHWHU VHUYHV LWV IXQFWLRQ RQO\ IRU VWXGHQWV RI KLJK DELOLW\ VLQFH ORZ DELOLW\ VWXGHQWV ZRXOG QRW EH H[SHFWHG WR DQVZHU WKHVH LWHPV FRUUHFWO\ &RPSDULQJ UHVSRQVH SDWWHUQV $Q DVVXPSWLRQ RI LWHP UHVSRQVH WKHRU\ LV WKH LQYDULDQFH RI LWHP SDUDPHWHUV

PAGE 89

$QDVWDVL %DNHU f 7KDW LV JLYHQ FHUWDLQ FRQn GLWLRQV LWHP SDUDPHWHUV VKRXOG EH XQLIRUP DPRQJ GLIIHUHQW SRSXODWLRQV EHFDXVH LQGLYLGXDO LWHPV DUH DVVXPHG WR PHDVn XUH WKH VDPH WUDLW LQ GLIIHUHQW SRSXODWLRQV ,Q WKLV FDVH WKH YDOLGLW\ RI WKH 13 LWHPV ZRXOG EH VXSSRUWHG LI 13 LWHPV ZHUH FRPPRQ WR &7%6 LWHPV ZLWK KLJK ORFDWLRQ SDUDPHWHUV ,Q IDFW RI WKH RULJLQDO 13 LWHPV KDG &7%6 ORFDWLRQ SDUDPHWHUV DERYH WKH PHDQ IRU WKH QRUPLQJ VDPSOH &7%6 7HFKQLFDO 5HSRUW f 7KH SURSRUWLRQ RI 13' LWHPV WR KLJK ORFDWLRQ SDUDPHWHU LWHPV ZDV HYHQ JUHDWHU RI bf ,Q VXPPDU\ WKH GDWD VXSSRUW DQ DIILUPDWLYH DQVZHU WR WKH ILUVW UHVHDUFK TXHVWLRQ 7KHUH H[LVWV D VXEVHW RI LWHPV WKDW GLVFULPLQDWHG JLIWHG DQG QRWJLIWHG VHYHQWK JUDGHUV ZKR ZHUH DOO DFDGHPLF KLJK DFKLHYHUV ,Q 3KDVH ,, FURVV YDOLGDWLRQ ZLWK D VPDOOHU EXW RWKHUZLVH HTXLYn DOHQW VDPSOH ZDV FRQGXFWHG WR VXSSRUW WKHVH ILQGLQJV 3KDVH ,, &URVV YDOLGDWLRQ RI 3KDVH ILQGLQJV ZDV FRQGXFWHG LQ 3KDVH ,, E\ DVVHVVLQJ WKH DFFXUDF\ RI 13 LWHPV DV FRPSDUHG WR WKH 6,7 LQ FODVVLI\LQJ WKH JLIWHG DQG QRWJLIWHG VWXn GHQWV DQG WKH FXWRII VFRUH DW ZKLFK HDFK SURFHGXUH VKRZHG JUHDWHVW DFFXUDF\ 7KHVH JRDOV ZHUH DFFRPSOLVKHG E\ DQn DO\]LQJ SURSRUWLRQV RI FRUUHFW DQG LQFRUUHFW FODVVLILFDn WLRQV XVLQJ FRHIILFLHQW .DSSD DQDO\VHV

PAGE 90

$Q H[DPLQDWLRQ RI WRWDO WUXH FODVVLILFDWLRQV LH FRUUHFW FODVVLILFDWLRQV RI JLIWHG DQG QRWJLIWHG VWXGHQWVf UHYHDOHG JUHDWHVW SUHGLFWLRQ DFFXUDF\ DW WKH 6' FXWn RII $W WKLV FXWRII 13' DQG 13SKL YDOXHV ZHUH HTXDO f DQG VOLJKWO\ VXSHULRU WR WKH 6,7 f :KHQ SHUn FHQWDJH YDOXHV ZHUH FRUUHFWHG IRU FKDQFH RFFXUUHQFHV E\ WKH .DSSD DQDO\VHV 13' ZDV VOLJKWO\ PRUH DFFXUDWH DW WKH 6' WKDQ DW 6' DQG 13' LWHPV ZHUH VXSHULRU WR 13SKL DQG WKH 6,7 )XUWKHUPRUH DW WKH FXWRII WKH .DSSD YDOXH IRU WKH 13' LV FRQWUDVWHG WR PXFK ORZHU YDOXHV IRU 13SKL f DQG 6,7 f ,Q DQVZHULQJ WKH VHFRQG UHVHDUFK TXHVWLRQ GDWD FRQn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n SRUW WKDW ILQGLQJ LQ WKDW WKH GLVFUHSDQF\ EHWZHHQ WRWDO

PAGE 91

WUXH FODVVLILFDWLRQV IRU WKH 13' DQG WKH 6,7 ZDV JUHDWHVW DW WKH FXWRII 'DWD VXSSRUWLQJ DQ 13' FXWRII DW PD\ EH OHVV FOHDU .DSSD DQDO\VHV LQGLFDWH WKDW WKH QXPEHU RI WRWDO WUXH FODVVLILFDWLRQV DW LV VXSHULRU WR WKH QXPEHU DW DQ\ RWKHU FXWRII 7KH UHODWLYH VWUHQJWK RI WKH 13' LQ WRWDO WUXH FODVVLILFDWLRQV ZDV GXH WR LWV PXFK JUHDWHU DFFXUDF\ LQ SUHGLFWLQJ WUXH SRVLWLYHV 7KH 6,7 ZDV PRUH DFFXUDWH LQ FODVVLI\LQJ WUXH QHJDWLYHV ,Q FKRRVLQJ D FXWRII VFRUH IRU JLIWHG ,4 VFUHHQLQJ SXUSRVHV VSHFLILF FODVVLILFDWLRQ SULRULWLHV DQG JRDOV PXVW EH WDNHQ LQWR DFFRXQW )RU H[DPSOH LI WKH SULPDU\ JRDO LV WR PD[n LPL]H WRWDO WUXH FODVVLILFDWLRQV WKH FXWRII PD\ EH GHVLUDEOH &RQFOXVLRQV ,PSOLFDWLRQV DQG /LPLWDWLRQV 7KLV VWXG\ \LHOGV FRQFOXVLRQV LPSOLFDWLRQV DQG OLPn LWDWLRQV JHUPDQH WR ERWK SUDJPDWLF DQG WKHRUHWLFDO LVVXHV ,Q WKH IROORZLQJ VHFWLRQV WKH LVVXHV RI VDPSOLQJ JHQn HUD L]DELLW\ LWHP YDOLGLW\ DQG FXWRII VFRUHV DUH GLVFXVVHG LQGLYLGXDOO\ 6DPSOLQJ 7KH VXFFHVV RI WKH 13 LQ FODVVLI\LQJ WZR JURXSV RI VWXGHQWV ZKR LQ VRPH UHVSHFWV DSSHDUHG LQGLVWLQJXLVKDEOH KDV VRPH LQKHUHQW LPSOLFDWLRQV 7KH JLIWHG DQG QRWJLIWHG FKLOGUHQ ZHUH VLPLODU LQ WHUPV RI DFKLHYHPHQW ,4 LQ PDQ\

PAGE 92

FDVHVf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n UDF\ SUHGLFWLRQV $ SHUWLQHQW OLPLWDWLRQ WR WKLV VWXG\ GLVFXVVHG LQ &KDSWHU ,,, ZDV WKH VDPSOLQJ SURFHGXUH 6SHFLILFDOO\ VRPH VWXGHQWV ZHUH GHOHWHG IURP IRUPDO ,4 WHVWLQJ DQG WKHUHIRUH H[FOXGHG IURP WKH VDPSOH EHFDXVH WKH\ VFRUHG EHORZ WKH 6,7 VFUHHQLQJ FXWRII RI +RZHYHU RQHVL[WK LH RI f 3KDVH ,, VWXGHQWV KDG 6,7 VFRUHV EHORZ ZLWK VRPH VFRUHV IDOOLQJ LQ WKH DYHUDJH DQG EHORZ DYHUDJH UDQJH 7KLV RFFXUUHQFH UDLVHV D TXHVWLRQ DERXW WKH FULWHULRQ XVHG IRU GLVTXDOLI\LQJ VWXGHQWV IURP ,4

PAGE 93

WHVWLQJ EDVHG RQ WKHLU 6,7 VFRUHV ,W LV XQFOHDU KRZ VWXn GHQWV ZLWK 6,7 VFRUHV LQ WKH V ZHUH QRW GLVTXDOLILHG ZKLOH RWKHUV ZHUH ,W VHHPV WKDW VFUHHQLQJ SURFHGXUHV ZHUH QRW IROORZHG FRQVLVWHQWO\ DW WKH VFKRRO OHYHO )RUn WXQDWHO\ HYHQ WKRXJK WKHUH ZHUH UHODWLYHO\ IHZ VWXGHQWV ZLWK 6,7 ,4V EHORZ WKHUH ZHUH HQRXJK IRU GDWD DQDO\n VHV DQG WKH 7DQJH RI VFRUHV ZDV ZLGH ,Q WKLV FDVH VDPSOLQJ HUURU SUREDEO\ ZDV QRW D FRQIRXQGLQJ LQIOXHQFH RQ UHVXOWV EHFDXVH GLVTXDOLILFDWLRQ RI WKH DIRUHPHQWLRQHG VWXGHQWV VHHPV WR KDYH RFFXUUHG LQ D UDQGRP XQV\VWHPDWLF PDQQHU *HQHUDOL]DELOLW\ ,Q ODUJH SDUW RXWFRPHV UHJDUGLQJ VFUHHQLQJ DFFXUDF\ LWHP YDOLGLW\ DQG FXWRII VFRUHV PD\ QRW EH JHQHUDOL]DEOH WR RWKHU SRSXODWLRQV DW WKLV WLPH +RZHYHU WKH VXFFHVV RI WKH UHVHDUFK SURFHGXUH 13f LV FRQVLGHUHG PHDQLQJIXO LQ DV PXFK DV WKH DFFXUDF\ RI WKH 13 ZLOO EH WHVWHG FRQWLQn XDOO\ LQ LWV SUDFWLFDO DSSOLFDWLRQ RQ SRWHQWLDOO\ JLIWHG VWXGHQWV ,Q WKLV ZD\ WKH JHQHUDOL]DELOLW\ RI VFUHHQLQJ DFFXUDF\ LWHP YDOLGLW\ DQG FXWRII VFRUHV ZLOO EH YDOn LGDWHG RQ RWKHU SRSXODWLRQV ,Q &KDSWHU WKH GHILQLWLRQ RI LQWHOOHFWXDO JLIWHGn QHVV IRU WKLV VWXG\ ZDV EULHIO\ GLVFXVVHG 7KLV SUDJPDWLF GHILQLWLRQ IRFXVHV RQO\ RQ ,4 WHVW VFRUHV WKDW IDOO DERYH

PAGE 94

D SDUWLFXODU FXWRII VFRUH 2WKHU PRUH WKHRUHWLFDO GHIn LQLWLRQV PD\ LQYROYH RWKHU TXDOLWDWLYH GLIIHUHQFHV 5HVXOWV RI WKLV VWXG\ PD\ QRW EH JHQHUDOL]DEOH WR VWXGHQWV ZKR DUH GHVLJQDWHG JLIWHG XVLQJ FULWHULD RWKHU WKDQ WKHLU ,4V IDOOLQJ DERYH WKH WK SHUFHQWLOH 6FUHHQLQJ $FFXUDF\ *HQHUD L]DEL LW\ RI UHVHDUFK UHVXOWV ZRXOG EH HQn KDQFHG E\ DQDO\]LQJ JURXS WHVW GDWD IRU RWKHU UHVWULFWHG SRSXODWLRQV WR GHWHUPLQH LI VXEJURXSV PD\ EH VXFFHVVIXOO\ GLVFULPLQDWHG XVLQJ WHVW LWHPV 7ZR SUHOLPLQDU\ VWHSV LQ VXFK D SURFHVV ZRXOG EH WR Df DQDO\]H WHVW UHVXOWV RI RWKHU SRSXODWLRQV RI SRWHQWLDOO\ JLIWHG VWXGHQWV VXFK DV HOHPHQWDU\ RU KLJK VFKRRO RQ WKH &7%67&6 DQG Ef GHWHUn PLQH LI RWKHU JURXS DFKLHYHPHQWFRJQLWLYH DELOLW\ WHVWV SRVVHVV LWHPV WKDW DFFXUDWHO\ FODVVLI\ SRWHQWLDOO\ JLIWHG VWXGHQWV )XUWKHU UHVHDUFKHUV PD\ VHHN WR JHQHUDOL]H UHVXOWV RQ PRUH GLYHUJHQW UHVWULFWHG SRSXODWLRQV )RU H[DPSOH LW PD\ EH XVHIXO WR VFUHHQ IRU PLOG PHQWDO UHWDUn GDWLRQ DPRQJ UHPHGLDO VWXGHQWV RU IRU OHDUQLQJ GLVDELOn LWLHV DPRQJ FKLOGUHQ ZLWK GLVFUHSDQW UHSRUW FDUG JUDGHV ,WHP 9DOLGLW\ $ VRPHZKDW XQH[SHFWHG UHVXOW RI WKLV UHVHDUFK ZDV WKDW WKH &7%6 LWHPV ZKLFK VXSSRVHGO\ PHDVXUH VFKRRO DFKLHYHPHQWf FRQWULEXWHG PXFK PRUH WR WKH 13 WKDQ GLG WKH 7&6 LWHPV ZKLFK DUH SXUSRUWHG WR PHDVXUH LQWHOOLJHQFHf

PAGE 95

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n PDWLFV 6FLHQFH 6RFLDO 6WXGLHV DQG 9RFDEXODU\ KDYH ZKDW DSSHDU WR EH GLUHFW FRUUHODWHV RQ WKH :,6&5 7KH UHODn WLRQVKLS EHWZHHQ 7&6 DQG :,6&5 VXEWHVWV DSSHDUV PRUH REn VFXUH 7KLV DVVHUWLRQ LV VXSSRUWHG E\ :XUVWHU f ZKR FRPSDUHG RYHUODS EHWZHHQ WKH 7&6 DQG :,6&5 ZLWK WKH 6,7 DQG WKH :,6&5 6KH IRXQG WKDW b RI WKH 6,7 LWHPV PHDVXUHG WKH VDPH VNLOOV DV WKH :,6&5 9HUEDO 6FDOH KRZn HYHU QR LWHPV IURP WKH 7&6 DSSHDUHG WR PHDVXUH DQ\ RI WKH VNLOOV WKDW DUH PHDVXUHG E\ WKH :,6&5 VXEWHVWV S f 7KH UHODWLRQVKLS RI FRPPRQDOLW\ EHWZHHQ ,4 WHVW DQG DFKLHYHPHQW WHVW SHUIRUPDQFH IRU WKLV VDPSOH VHHPV UHODWHG

PAGE 96

WR WZR IDFWRUV )LUVW LW LV OLNHO\ WKDW IRU KLJK DFKLHYn LQJ VWXGHQWV VXSHULRU LQWHOOLJHQFH LV KHDYLO\ ORDGHG LQ WKH YHUEDO UHDVRQLQJ GRPDLQV WKDW DUH WDSSHG WR D JUHDWHU H[WHQW E\ WKH &7%6 WKDQ E\ WKH 7&6 7KH 7&6 WR D JUHDWHU GHJUHH WKDQ WKH &7%6 PHDVXUHV QRQYHUEDO VNLOOV RU DELOn LWLHV WKDW PD\ EH OHVV UHSUHVHQWHG E\ KLJK DFKLHYLQJ VWXn GHQWVn VWUHQJWKVn RQ WKH :,6&5 RU 6WDQIRUG%LQHW )XWXUH UHVHDUFKHUV PLJKW LQYHVWLJDWH WKLV K\SRWKHVLV E\ H[DPLQLQJ WKH UHODWLRQVKLS EHWZHHQ DFKLHYHPHQW DQG YHUEDO YV QRQn YHUEDO LQWHOOLJHQFH DV UHSUHVHQWHG E\ WKHVH RU VLPLODU WHVWVf IRU KLJK DFKLHYLQJ RU JLIWHG VWXGHQWV $ VHFRQG H[SODQDWLRQ IRU WKH GLVSURSRUWLRQDWH UHSUHn VHQWDWLRQ RI DFKLHYHPHQW WHVW LWHPV RQ WKH 13 FRQFHUQV WKH WKHRUHWLFDO UDWLRQDOH IRU WKLV UHVHDUFK 7KDW LV WKH FRQWHQW RI DFKLHYHPHQW WHVWV HJ &7%6f DQG ,4 WHVWV HJ :,6&5f DUH RIWHQ RSHUDWLRQDOO\ LQGLVWLQJXLVKDEOH 7KLV SUHPLVH KDV EHHQ VXSSRUWHG HPSLULFDOO\ E\ $QDVWDVL f DPRQJ PDQ\ RWKHUV VHH &KDSWHU ,,f &XWRII V &RQFOXVLRQV UHJDUGLQJ WKH PRVW GHVLUDEOH FXWRII IRU WKH 13 13' DW RU f ZHUH GLVFXVVHG HDUOLHU LQ WKH FKDSWHU 7KHUH DUH VRPH RWKHU FRQFOXVLRQV WKDW PD\ EH GUDZQ UHJDUGLQJ FXWRII VFRUHV IRU WKH 6,7 ZLWK WKLV SRSXn ODWLRQ 7KH PHDQ 6,7 VFRUH IRU WKH 3KDVH ,, VDPSOH ZDV 7KLV VFRUH LV VRPHZKDW DERYH ZKDW ZRXOG EH H[SHFWHG \HW FRQVLVWHQW ZLWK UHVHDUFK ILQGLQJV E\ .DUQHV DQG %URZQ

PAGE 97

f DQG 5XVW DQG /RVH f ZKR IRXQG WKDW WKH 6,7 WHQGV WR RYHUHVWLPDWH KLJK ,4V RQ IXOO OHQJWK WHVWV ,Q WKHLU VWXGLHV WKRVH UHVHDUFKHUV UHFRPPHQGHG VHWWLQJ 6,7 FXWRII VFRUHV IRU JLIWHG FODVVLILFDWLRQ KLJKHU WKDQ WZR VWDQGDUG GHYLDWLRQV DERYH WKH PHDQ WR RIIVHW WKLV WHQGHQF\ WR RYHUHVWLPDWH WKH ,4V RI EULJKWHU VWXGHQWV 7KH FXUUHQW ILQGLQJV VXSSRUW WKRVH UHFRPPHQGDWLRQV EHFDXVH WKH 6,7 ZDV PRVW DFFXUDWH LQ FODVVLI\LQJ WUXH SRVLWLYH DQG IDOVH QHJn DWLYH ILQGLQJV DW RYHU WKUHH VWDQGDUG GHYLDWLRQV DERYH WKH PHDQ 6,7 ,4 f 6XPPDWLRQ ,Q KLV ODWHU \HDUV (GZLQ 5 *XWKULH f VXJJHVWHG WKDW UHVHDUFK KDV QR LQKHUHQW YDOXH EXW UDWKHU WKDW LWV YDOXH ZDV JDLQHG IURP LWV SUDFWLFDO DSSOLFDWLRQV ,W LV LQ WKH VSLULW RI WKDW SKLORVRSK\ WKDW WKLV UHVHDUFK PD\ EH IXOO\ DSSUHFLDWHG 5HODWLYHO\ DFFXUDWH LQ LWV SUHGLFWLYH SRZHU DQG HIILFLHQW LQ LWV PHWKRG WKH 13 DQDO\VLV ZLOO EH HDVLO\ UHSOLFDWHG RQ QHZ VWXGHQW VDPSOHV DQG WHVW IRUPDWV 7KXV WKH 13 PD\ EH EHVW YLHZHG DV D SURFHGXUH DGDSWDEOH WR YDU\LQJ \HW VSHFLILF QHHGV

PAGE 98

5()(5(1&(6 $KPDQQ 6 f 7KH FRPSUHKHQVLYH WHVW RI EDVLF VNLOOV D UHYLHZf ,Q 2 %XURV (Gf 7KH VHYHQWK PHQWDO PHDVXUHPHQW \HDUERRN SS f +LJKODQG %DUN 17KH *U\SKRQ 3UHVV $PHULFDQ 3V\FKRORJLFDO $VVRFLDWLRQ f 6WDQGDUGV IRU HGXFDWLRQDO DQG SV\FKRORJLFDO WHVWV :DVKLQJWRQ '& $PHULFDQ 3V\FKRORJLFDO $VVRFLDWLRQ $QDVWDVL $ f 3V\FKRORJLFDO WHVWLQJ 1HZ
PAGE 99

%ORVVHU + f *URXS LQWHOOLJHQFH WHVWV DV VFUHHQLQJ GHYLFHV LQ ORFDWLQJ JLIWHG DQG VXSHULRU VWXGHQWV LQ QLQWK JUDGH ([FHSWLRQDO &KLOGUHQ %UDGHQ 3 f $ PRGHVW SURSRVDO 8VLQJ SUREDELOn LWLHV RI VSHFLDOHGXFDWLRQ HOLJLELOLW\ LQVWHDG RI FXWWLQJ VFRUHV 8QSXEOLVKHG PDQXVFULSW %URFN + f )DFWRU VWUXFWXUH RI LQWHOOHFWXDO DQG DFKLHYHPHQW PHDVXUHV IRU OHDUQLQJ GLVDEOHG FKLOGUHQ 3V\FKRORJ\ LQ WKH 6FKRROV BOMf %URRNV & 5 f :,6& :,6&5 6% /0 :5$7 5HODn WLRQVKLSV DQG WUHQGV DPRQJ FKLOGUHQ DJHV VL[ WR WHQ UHIHUUHG IRU SV\FKRORJLFDO HYDOXDWLRQ 3V\FKRORJ\ LQ WKH 6FKRROV %U\DQ 5 t %U\DQ 7 + f 8QGHUVWDQGLQJ OHDUQn LQJ GLVDELOLWLHV 6KHUPDQ 2DNV &$ $OIUHG 3XEOLVKn LQJ &R %XUNHW 5 f (PSLULFDO FULWHULD IRU GLVWLQJXLVKn LQJ DQG YDOLGDWLQJ DSWLWXGH DQG DFKLHYHPHQW PHDVXUHV ,Q $ *UHHQ (Gf 7KH DSWLWXGH DFKLHYHPHQW GLVWLQFWLRQ SS f 0RQWHUH\ &$ &7%0F* UDZ+LOO &DUOHWRQ ) t 6WDFH\ & / f (YDOXDWLRQ RI VHOHFWHG VKRUW IRUPV RI WKH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ -RXUQDO RI &OLQLFDO 3V\FKRORJ\ &DUUROO % f )DFWRUV RI YHUEDO DFKLHYHPHQW ,Q $ $QDVWDVL (Gf 7HVWLQJ SUREOHPV LQ SHUVSHFWLYH SS f :DVKLQJWRQ '& $PHULFDQ &RXQFLO RQ (GXFDWLRQ &KDPEHUV $ f 3UHOLPLQDU\ VFUHHQLQJ PHWKRGV LQ WKH LGHQWLILFDWLRQ RI LQWHOOHFWXDOO\ VXSHULRU FKLOn GUHQ ([FHSWLRQDO &KLOGUHQ &KDPEHUV $ %DUURQ ) t 6SUHFKHU : f ,GHQWLI\LQJ JLIWHG 0H[LFDQ$PHULFDQ VWXGHQWV *LIWHG &KLOG 4XDUWHUO\ &ODUL]LR + ) t 0HKHUHQV : $ f 3V\FKRPHWULF OLPLWDWLRQV RI *XLOIRUGnV VWUXFWXUHRILQWHOHFW PRGHO IRU LGHQWLILFDWLRQ DQG SURJUDPPLQJ RI WKH JLIWHG *LIWHG &KLOG 4XDUWHUO\

PAGE 100

&RKHQ $ f $ FRHIILFLHQW RI DJUHHPHQW IRU QRPn LQDO VFDOHV (GXFDWLRQDO DQG 3V\FKRORJLFDO 0HDVXUHn PHQW &URIRRW 0 t %HQQHWW 7 6 f $ FRPSDULVRQ RI WKUHH VFUHHQLQJ WHVWV DQG WKH :,6&5 LQ VSHFLDO HGXFDWLRQ HYDOXDWLRQV 3V\FKRORJ\ LQ WKH 6FKRROV &7%0F*UDZ+L f 7KH &RPSUHKHQVLYH 7HVW RI %DVLF 6NLOOV 7HFKQLFDO UHSRUW 0RQWHUH\ &$ &7%0F*UDZ +LO &7%0F*UDZ+L f 7KH 7HVW RI &RJQLWLYH 6NLOOV 7HFKQLFDO UHSRUW 0RQWHUH\ &$ &7%0F*UDZ+L 'HDQ 5 6 f &DQRQLFDO DQDO\VLV RI D MDQJOH IDOn ODF\ 0XOWLYDULDWH ([SHULPHQWDO &OLQLFDO 5HVHDUFK 'HDQ 5 6 f ,QWHOOLJHQFHDFKLHYHPHQW GLVFUHSn DQFLHV LQ GLDJQRVLQJ SHGLDWULF OHDUQLQJ GLVDELOn LWLHV &OLQLFDO 1HXURSV\FKRORJ\ 'H)LOLSSLV 1 $ t )XOPDU f (IIHFWV DW DJH DQG ,4 OHYHO RQ WKH YDOLGLW\ RI RQH VKRUW LQWHOOLn JHQFH WHVW XVHG IRU VFUHHQLQJ SXUSRVHV (GXFDWLRQDO DQG 3V\FKRORJLFDO 0HDVXUHPHQW 'LUNHV 0 $ f 2QO\ WKH JLIWHG FDQ GR (GXFDWLRQDO +RUL]RQV 'LUNV :HVVHOV 4XDIRUWK t 4XHQRQ % f &DQ VKRUW IRUP :,6&5 WHVWV LGHQWLI\ FKLOn GUHQ ZLWK KLJK )XOO 6FDOH ,4" 3V\FKRORJ\ LQ WKH 6FKRROV 'XQQ / 0 t 0DUNZDUGW ) & f 3HDERG\ ,QGLYLGn XDO $FKLHYHPHQW 7HVW &LUFOH 3LQHV 01 $PHULFDQ *XLGDQFH 6HUYLFH (EHO 5 / f 0HDVXULQJ HGXFDWLRQDO DFKLHYHPHQW (QJOHZRRG &OLIIV 13UHQWLFH+DOO (OPDQ / %OL[W 6 t 6DZLFNL 5 f 7KH GHYHORSn PHQW RI FXWRII VFRUHV RQ D :,6&5 LQ WKH PXOWLGLPHQn VLRQDO DVVHVVPHQW RI JLIWHG FKLOGUHQ 3V\FKRORJ\ LQ WKH 6FKRROV

PAGE 101

(QEXUJ 5 5RZOH\ 9 1 t 6WRQH % f 6KRUW IRUPV RI WKH :,6& IRU XVH ZLWK HPRWLRQDOO\ GLVWXUEHG FKLOGUHQ -RXUQDO RI &OLQLFDO 3V\FKRORJ\ (QJHOKDUW 0 f $ FRPSDULVRQ RI VHYHUDO LWHP GLVFULPLQDWLRQ LQGLFHV -RXUQDO RI (GXFDWLRQDO 0HDn VXUHPHQW )HOO / EH )HOO 6 6 f (IIHFWLYHQHVV RI :6&5 VKRUWIRUPV LQ VFUHHQLQJ JLIWHG FKLOGUHQ 3V\FKRORJn LFDO 5HSRUWV )LQGOH\ & -f t 7KRPSVRQ 0 f $Q DEEUHYLDWHG :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ IRU XVH ZLWK HGXFDEOH PHQWDOO\ UHWDUGHG $PHULFDQ -RXUQDO RI 0HQWDO 'HILFLHQF\ )ULHGHV f 5HYLHZ RI WKH :,6&5 ,Q 2. %XURV (Gf 7KH HLJKWK PHQWDO PHDVXUHPHQW \HDUERRN SS f +LJKODQG 3DUN 17KH *U\SKRQ 3UHVV *HQVOH\ f 3DUHQW SHUVSHFWLYHV RU WKH FXULRVLW\ TXRWLHQW *LIWHG &KLOG 4XDUWHUO\ *URQOXQG 1 ( f 0HDVXUHPHQW DQG HYDOXDWLRQ LQ WHDFKLQJ 1HZ
PAGE 102

+DOH 5 / f 7KH :,6&5 DV D SUHGLFWRU RI :5$7 SHUIRUPDQFH 3V\FKRORJ\ LQ WKH 6FKRROV +DPPLOO t 0F1XWW f 7KH FRUUHODWHV RI UHDGLQJ $XVWLQ 7; 3UR(G +DUULQJWRQ 5 f 6WDQGDUGL]HG WHVWLQJ PD\ EH KD]DUGRXV WR WKH HGXFDWLRQDO SURJUHVV RI LQWHOOHFn WXDOO\ JLIWHG FKLOGUHQ (GXFDWLRQ +DUWODJH / & t 6WHHOH & 7 f :,6& DQG :,6&5 FRUUHODWHV RI DFDGHPLF DFKLHYHPHQW 3V\FKRORJ\ LQ WKH 6FKRROV +LUVFK ) t +LUVFK 6 f 7KH 4XLFN 7HVW DV D VFUHHQLQJ GHYLFH IRU JLIWHG VWXGHQWV 3V\FKRORJ\ LQ WKH 6FKRROV +RUQ / f )DFWRU DQDO\VLV ZLWK YDULDEOHV RI GLIIHUHQW PHWULF (GXFDWLRQDO DQG 3V\FKRORJLFDO 0HDn VXUHPHQW +XUURFNV ( f $VVHVVPHQW RI EHKDYLRU WKH PHWKn RGRORJ\ DQG FRQWHQW RI SV\FKRORJLFDO PHDVXUHPHQW &ROXPEXV 2+ &KDUOHV ( 0HUULOO 3XEOLVKLQJ -HQNLQV)ULHGPDQ 5 f 0\WK &RVPHWLF XVH RIPXOn WLSOH VLWXDWLRQ FULWHULD *LIWHG &KLOG 4XDUWHUO\ -HQVHQ $ 5 f %LDV LQ PHQWDO WHVWLQJ 1HZ
PAGE 103

.DUQHV ) $ (GZDUGV 5 3 t 0F&DOOXP 5 6 f 1RUPDWLYH DFKLHYHPHQW DVVHVVPHQW RI JLIWHG FKLOGUHQ &RPSDULQJ WKH .$%& :5$7 DQG &$7 3V\FKRORJ\ LQ WKH 6FKRROV .DXIPDQ $ 6 f )DFWRU DQDO\VLV RI WKH :,6&5 DW HOHYHQ DJH OHYHOV EHWZHHQ DQG \HDUV -RXUQDO RI &RQVXOWLQJ DQG &OLQLFDO 3V\FKRORJ\ B .DXIPDQ $ 6 f ,QWHOOLJHQFH WHVWLQJ ZLWK WKH :,6&5 +HZ
PAGE 104

0DOH 5 $ t 3HUURQH 3 f ,GHQWLI\LQJ WDOHQW DQG JLIWHGQHVV 5RHSHU 5HYLHZ 0DOOLQVRQ * f $Q DQDO\VLV RI WKH IDFWRUV UHn ODWHG WR WKH PRWLYDWLRQ LQG DFKLHYHPHQW RI VWXGHQWV LQ VFLHQFH FRXUVHV LQ WKH MXQLRU DQG VHQLRU KLJK ILQDO UHSRUW 5HSRUW 1R &53f .DODPD]RR 0, :HVWHUQ 0LFKLJDQ 8QLYHUVLW\ 6FKRRO RI *UDGXDWH 6WXGLHV (5,& 'RFXPHQW 5HSURGXFWLRQ 6HUYLFH 1R (' f 0DUWLQ -' t .LGZHOO & f ,QWHUFRUUHODWLRQV RI WKH :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ 5HYLVHG WKH 6ORVVRQ ,QWHOOLJHQFH 7HVW DQG WKH 1DWLRQDO (GXFDWLRQDO 'HYHORSPHQWDO 7HVW (GXFDWLRQD DQG 3V\FKRORJLFDO 0HDVXUHPHQW 0DUWLQ t 5XGROSK / f &RUUHODWHV RI WKH :HFKVOHU $GXOW ,QWHOOLJHQFH 6FDOH WKH 6ORVVRQ ,Qn WHOOLJHQFH 7HVW $&7 VFRUHV DQG JUDGH SRLQW DYHUn DJHV (GXFDWLRQDO DQG 3V\FKRORJLFDO 0HDVXUHPHQW 0D\ILHOG % f 7HDFKHU SHUFHSWLRQ RI FUHDWLYLW\ LQWHOOLJHQFH DQG DFKLHYHPHQW *LIWHG &KLOG 4XDU WHUO\ 0F1HPDU 4 f 3V\FKRORJLFDO VWDWLVWLFV UG (Gf 1HZ
PAGE 105

1LFKROVRQ & /f &RUUHODWLRQV EHWZHHQ WKH 4XLFN 7HVW DQG WKH :FFKVOFU ,QWHOOLJHQFH 6FDOH IRU &KLO GUHQ5HYLVHG 3V\FKRORJLFDO 5HSRUWV 2OVHQ $ 9 t 5RVHQ & / )HEUXDU\f ([SORUDn WLRQ RI WKH VWUXFWXUH RI VHOHFWHG UHDGLQJ UHDGLQHVV WHVWV 3DSHU SUHVHQWHG DW WKH PHHWLQJ RI WKH $PHUn LFDQ (GXFDWLRQDO 5HVHDUFK $VVRFLDWLRQ 1HZ
PAGE 106

5\DQ 7 f 7KH LQIOXHQFH RI UHDGDELOLW\ RQ FULWn LFDO UHDGLQJ FRPSUHKHQVLRQ RI VHFRQGDU\ VRFLDO VWXGLHV VWXGHQWV 'LVVHUWDWLRQ $EVWUDFWV ,QWHUQDn WLRQDO 8QLYHUVLW\ 0LFURILOPV 1R f 5\DQ 7 f $Q DQDO\VLV RI PRWLYDWLRQDO IDFWRUV RI JLIWHG DQG QRQJLIWHG OHDUQHUV 3DSHU SUHVHQWHG DW WKH DQQXDO PHHWLQJ RI WKH $PHULFDQ (GXFDWLRQDO 5HVHDUFK $VVRFLDWLRQ 1HZ
PAGE 107

6WHGPDQ 1 /DZOLV ) &RUWQHU 5 + t $FKWHQ EHUJ f 5HODWLRQVKLSV EHWZHHQ :,6&5 IDFn WRUV :LGH 5DQJH $FKLHYHPHQW 7HVW VFRUHV DQG YLVXDO PRWRU PDWXUDWLRQ LQ FKLOGUHQ UHIHUUHG IRU SV\FKRORJn LFDO HYDOXDWLRQ -RXUQDO RI &RQVXOWLQJ DQG &OLQLFDO 3V\FKRORJ\ 6WHQVRQ & 0 f 1RWH RQ FRQFXUUHQW YDOLGLW\ RI VWUXFWXUH RI LQWHOOHFW JLIWHG VFUHHQHU ZLWK :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOGUHQ5HYLVHG 3V\FKRORJn LFDO 5HSRUWV 6WHUQEHUJ 5 f /LHV ZH OLYH E\ 0LVFRQFHSWLRQV RI WHVW LQ LGHQWLI\LQJ WKH JLIWHG *LIWHG &KLOG 4XDUWHUO\ 6WHZDUW t -RQHV ( & f 9DOLGLW\ RI WKH 6ORVVRQ ,QWHOOLJHQFH 7HVW $ WHQ \HDU UHYLHZ 3V\n FKRORJ\ LQ WKH 6FKRROV 6WHZDUW : t 0RUULV / f ,QWHOOLJHQFH DQG DFDGHPLF DFKLHYHPHQW LQ D FOLQLFDO DGROHVFHQW SRSXn ODWLRQ 3V\FKRORJ\ LQ WKH 6FKRROV 7HUPDQ / 0 t 0HUULOO 0 $ f 6WDQIRUG%LQHW ,QWHOOLJHQFH 6FDOH %RVWRQ +RXJKWRQ 0LIIOLQ 7KRPSVRQ t )LQGOH\ & f 7KH YDOLGDWLRQ RI DQ DEEUHYLDWHG :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLOn GUHQ IRU XVH ZLWK WKH HGXFDEOH PHQWDOO\ UHWDUGHG (GXFDWLRQDO DQG 3V\FKRORJLFDO 0HDVXUHPHQW 7XWWOH ) % t %HFNHU / $ f &KDUDFWHULVWLFV DQG LGHQWLILFDWLRQ RI JLIWHG DQG WDOHQWHG VWXGHQWV :DVKn LQJWRQ 1DWLRQDO (GXFDWLRQDO $VVRFLDWLRQ 8QGKHLP 2 f $ELOLW\ VWUXFWXUH LQ \HDU ROG FKLOGUHQ DQG WKH WKHRU\ RI IOXLG DQG FU\iWDOLHG LQWHOOLJHQFH -RXUQDO RI (GXFDWLRQDO 3V\FKRORJ\ 9DQGLYHU 3 & t 9DQGLYHU 6 6 f $ QRQELDVHG DVVHVVPHQW RI LQWHOOLJHQFH WHVWLQJ 7KH (GXFDWLRQDO )RUXP 9HUQRQ 3 ( f 7KH VWUXFWXUH RI KXPDQ DELOLWLHV /RQGRQ 0HWKXHQ

PAGE 108

9HUQRQ 3 / $GDPVRQ t 9HUQRQ ) f 7KH SV\FKRORJ\ DQG HGXFDWLRQ RI JLIWHG FKLOGUHQ %RXOn GHU &2 :HVWYLHZ 3UHVV :DGH / 3KHOSV / t )DODVFR 6 f 8VH RI DQ DEEUHYLDWHG YHUVLRQ RI WKH :,6&5 ZLWK OHDUQLQJ GLVn DEOHG FKLOGUHQ 3V\FKRORJ\ LQ WKH 6FKRROV :DVKLQJWRQ ( (QJHOPDQQ 6 t %HUHLWHU & f $FKLHYHPHQW FRPSRQHQWV RI 6WDQIRUG%LQHW SHUIRU WLLFWLLRH B (5,& 'RFXPHQW 5HSURGXFWLRQ 6HUYLFH 1R (' f :HFKVOHU f &RJQLWLYH FRQDWLYH DQG QRQLQWHO OHFWLYH LQWHOOLJHQFH $PHULFDQ 3V\FKRORJLVW :HFKVOHU f 7KH PHDVXUHPHQW DQG DSSUDLVDO RI DGXOW LQWHOOLJHQFH %DOWLPRUH :LOOLDPV DQG :LONLQV :HFKVOHU f :HFKVOHU ,QWHOOLJHQFH 6FDOH IRU &KLGUHQ5HYLVHG 1HZ
PAGE 109


PAGE 110

%,2*5$3+,&$/ 6.(7&+ 5DQG\ (YDQ 6FKQHOO ZDV ERUQ LQ &OHYHODQG 2KLR RQ -DQXDU\ 5HDUHG LQ +ROO\ZRRG )ORULGD IURP DJH VHYHQ KHDWWDLQHG KLV %$ LQ SV\FKRORJ\ ZLWK D PLQRU LQ HGXFDWLRQ IURP )ORULGD $WODQWLF 8QLYHUVLW\ LQ 3XUn VXLQJ D ILHOG RI JUDGXDWH VWXGLHV WKDW HQFRPSDVVHG ERWK SV\FKRORJ\ DQG HGXFDWLRQ 5DQG\ DWWDLQHG KLV (G6 f DQG 3K' $XJXVW f IURP WKH 8QLYHUVLW\ RI )ORULGD 3URIHVVLRQDOO\ 5DQG\ KDV ZRUNHG DV D VFKRRO SV\FKROn RJLVW LQ ERWK XUEDQ DQG UXUDO VHWWLQJV LQ )ORULGD +H LV FXUUHQWO\ ZRUNLQJ DW 0HPSKLV &LW\ 6FKRROV 0HQWDO +HDOWK &HQWHU LQ 7HQQHVVHH ZKHUH KH LV FRRUGLQDWRU RI WKH $GROHVn FHQW 6H[ 2IIHQGHU 3URJUDP RI WKH 0HPSKLV 6KHOE\ &RXQW\ 6H[ $EXVH 3URMHFW

PAGE 111

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

PAGE 112

, FHUWLI\ WKDW KDYH UHDG WKLV VWXG\ DQG WKDW LQ P\ RSLQLRQ LW FRQIRUPV WR DFFHSWDEOH VWDQGDUGV RI VFKRODUO\ SUHVHQWDWLRQ DQG LV IXOO\ DGHTXDWH LQ VFRSH DQG TXDOLW\ DV D GLVVHUWDWLRQ IRU WKH GHJUHH RI 'RFWRU RI 3KLORVRSK\ 7KLV GLVVHUWDWLRQ ZDV VXEPLWWHG WR WKH *UDGXDWH )DFXOW\ RI WKH &ROOHJH RI (GXFDWLRQ DQG WR WKH *UDGXDWH 6FKRRO DQG ZDV DFFHSWHG DV SDUWLDO IXOILOOPHQW RI WKH UHTXLUHPHQWV IRU WKH GHJUHH RI 'RFWRU RI 3KLORVRSK\ $XJXVW 'HDQ &ROOHJH RI (GXFDWLRQ 'HDQ *UDGXDWH 6FKRRO

PAGE 113

81,9(56,7< 2) )/25,'$


42
intelligence when both instruments have been properly
administered" (p. 16) and further proposed that the SIT
qualifies as an alternate form "of the Stanford-Binet be
cause the two tests possess equivalent means and standard
deviations" (p. 16). Based on their dubious assumption of
test equivalency between the revised SIT and the S-B, the
authors employed the Mean Absolute IQ Difference (MAD)
statistic to determine alternate form test reliability and
standard errors of measurement. The MAD procedure, which
is meant to be used only with equivalent forms of a test,
yields a statistic which is approximately equal to the
standard deviation times 8862. The authors did not indi
cate the relative effects on the reliability and the SEMs
when the measures compared do not strictly meet the Crite
ria of alternate forms, as is apparent in this case. Nor
are there attempts to evaluate other kinds of test relia
bility. It might be concluded, therefore, that the
authors' claim that "the SIT's reliability may be regarded
as not less than .95" (p. 136) should be interpreted with
caution.
Another claim made in the SIT manual is that the mean
difference between the Stanford-Binet and Slosson IQ
scores is less than one point, based on the sample of
1,109. The procedure by which this statistic was obtained
entails computing the means for IQ differences between the


20
(1960) sought a screening instrument for use in a Michigan
school district. Using the IPAT (Cattell's test of gen
eral intelligence), the California Test of Mental Matu
rity, the SRA Primary Abilities Test, The Kuhlman-Anderson
Intelligence Test, and the WISC, Chambers tested 39 chil
dren in grades three through six. For each screening
test, a cutoff was calculated above which all gifted stu
dents (WISC IQ>124) would be identified. The accuracy of
each screening procedure was established at 100%, and the
efficiency was then determined based on the number of not-
gifted students misclassified by the screening procedure
ns gifted. The results revealed that the SKA test and the
Ku1hman-Anderson could be ranked respectively as the most
and least efficient, and that between 20% and 57% of the
students predicted as gifted were not.
Three years after Chambers' study, Blosser (1963)
tested 187 ninth graders on the Henmon-Ne1 son and Otis
group intelligence tests. The research sample had a mean
IQ of 120 on the Stanford-Binet with a range of 98 to 153.
The results indicated that of the 36 students predicted as
gifted by the Otis, only 13 (36%) were identified as such
by the Stanford-Binet. On the Henmon-Nelson 13 of 26 stu
dents (57%) were correctly predicted as gifted. Because
19% of the gifted students were not identified by either
group test, both tests proved to be poor predictors of
giftedness.


8
Performance IQ (P-IQ). Performance IQ is a subscale
IQ score on the Wechsler Intelligence Scale for Children--
Revised that represents the examinee's nonverbal reasoning
or perceptual organization (Kaufman, 1975).
Reliability. Reliability is the squared population
correlation between the individual's obtained score and
the individual's hypothetical true score. Reliability is
"the proportion of true-score variance in scores on a par
ticular test at t he time it was taken" (Jensen, 1 980, p.
260) .
Slosson Intelligence Test (SIT). The SIT is an indi
vidually administered intelligence test which requires
little specialized training to administer, only about 20
minutes to administer and score, and yields IQ scores
"which are close approximations to the Stanford-Binet IQ
[scores]" (Slosson & Jensen, 1982, p. 1).
Stanford-Binet Intelligence Scale (S-B). The S-B is
one of the two individually administered intelligence
tests used in this study to measure gifted intelligence.
(See page 41 for a detailed description.)
Test of Cognitive Skills (TCS). The TCS is "an abil
ity test designed to assess a student's academic aptitude
and thereby predict the student's level of success in
school. Emphasis in TCS is placed on" . problem solv
ing, discovering relationships, evaluating, and


87
testing based on their SIT scores. It is unclear how stu
dents with SIT scores in the 80s were not disqualified
while others were. It seems that screening procedures
were not followed consistently at the school level. For
tunately, even though there were relatively few students
with SIT IQs below 132, there were enough for data analy
ses, and the Tange of scores was wide. In this case,
sampling error probably was not a confounding influence on
results because disqualification of the aforementioned
students seems to have occurred in a random, unsystematic
manner.
Generalizability
In large part, outcomes regarding screening accuracy,
item validity, and cutoff scores may not be generalizable
to other populations at this time. However, the success
of the research procedure (NP) is considered meaningful in
as much as the accuracy of the NP will be tested contin
ually in its practical application on potentially gifted
students. In this way, the generalizability of screening
accuracy, item validity, and cutoff scores will be val
idated on other populations.
In Chapter I the definition of intellectual gifted
ness for this study was briefly discussed. This pragmatic
definition focuses only on IQ test scores that fall above


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy
AN ASSESSMENT PROCEDURE FOR DETECTING
GIFTEDNESS USING AVAILABLE DATA
By
Randy Schnell
December 1987
Chairman: Larry Loesch, Ph.D.
Major Department: Counselor Education
The purpose of this study was to examine a new psy
chometric screening procedure designed to discriminate
intellectually "gifted" seventh graders from other high
achieving seventh graders. All 179 students in the re
search sample had been assessed to determine eligibility
in the Hillsborough County Public Schools "gifted" pro
gram. An accurate and efficient screening procedure was
necessary in order to delete from time-consuming, formal
assessment those students who were unlikely to meet intel
lectual criteria for gifted program placement. The Slos-
son Intelligence Test (SIT), previously used to screen
students referred to the gifted classes, had proved inef
ficient for this purpose. A plethora of research on
v


44
Comprehensive Test of Basic Skills (CTBS)/Test of
Cognitive Skills (TCS)
The CTBS (CTB/McGraw-Hi 11, 1984) and TCS (CTB/McGraw-
Hill, 1984) are the tests from which NP items were taken.
Psychometrically, these tests were well suited to this re
search. The appropriateness of the CTBS and TCS for this
research are supported by several of their attributes,
some of which were discussed in Chapter I.
Items were chosen for both tests according to item
response theory (IRT) utilizing a three-parameter logistic
model. The items were chosen according to their ability
to (a) discriminate high ability traits from low ability
traits, (b) discriminate high ability students and low
ability students by matching item difficulty with student
total score, and (c) account for guessing as an influence
on score difficulty.
In terms of content validity, the CTBS is designed to
measure understanding of a broad rgnge of concepts as
developed by various educational curricula. Test perfor
mance reflects a student's skills in effective use of
information explicit in categories derived from Bloom's
taxonomy (Bloom, 1956). Item development specifications
were designed to ensure comprehensive coverage of the con
tent and process categories.
The TCS is designed to measure an aptitude construct
that can be operationally distinguished from the


TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS i i
ABSTRACT V
CHAPTER
I INTRODUCTION 1
Problem 3
Giftedness 4
Purpose 5
Research Questions 5
Definition of Terms 6
Theoretical Rationale 9
Need for This Study 15
Overview of the Remainder of the Study 17
II LITERATURE REVIEW 18
Support for the Problem 18
Group IQ Tests 19
Intelligence Quotient Short Forms 21
Intelligence Quotient Screening
Tests 25
Achievement Tests 30
Summary 39
Instruments Used in Study 39
Wechsler Intelligence Scale for
Children-Revised (WISC-R) ... 39
Slosson Intelligence Test SIT) . 41
Comprehensive Test of Basic Skills
(CTBS)/Test of Cognitive Skills
(TCS) 4 4
Stanford-Binet Intelligence Scale
(S-B)
iii
46


CHAPTER III
METHODOLOGY
Overview
In this research study a procedure was investigated
for analyzing a comprehensive, group-administered achieve
ment and cognitive abilities test to determine whether an
item set can be derived that discriminates gifted from
high-achieving, but not-gifted, seventh graders. When
such an item set was derived, it was compared to a com
monly used IQ screening test to assess the relative accu
racies of each procedure in discriminating gifted from
not-gifted students in a second seventh grade population
sample.
This chapter is organized into the following sec
tions: (a) Population and Sample, (b) Assessment Instru
ments, (c) Research Procedures, (d) Data Analysis, and (e)
Methodological Limitations.
Population and Sample
The research sample of 179 students was drawn from a
population of seventh graders who had been tested on the
WISC-R or S-B for the "gifted program" in the Hillsborough
49


9
remembering" (CTB/McGraw-Hi 1 1 1 984 p. 1). The TCS is
administered concurrently with the CTBS.
Validity. Validity refers to the appropriateness of
inferences from test scores or other forms of assessment.
Validity deals with how faithfully the scores represent a
domain of skill, knowledge, or of a trait being measured
(American Psychological Association, 1985).
Verbal IQ (V- IQ). Verbal IQ is a subscale IQ score
on the Wechsler Intelligence Scale for Children--Revised
that represents the examinee's verbal reasoning or verbal
comprehension (Kaufman, 1975).
Wechsler Intelligence Scale for Children--Revised
(W1SC-R). The WISC-R is the individually administered
IQ test predominantly used in this study to measure gifted
intelligence. (See page 35 for a detailed description.)
Theoretical Rationale
Salvia and Ysseldyke (1978) have pointed out that
there is a hypothetical domain of items that may be used
to assess intelligence and that items may be drawn from
various sources. For example, WISC-R Information subtest
items are drawn from a domain of achievement oriented
items that measure specific content of learning acquired,
in large part, through formal education. This overlapping
of achievement and aptitude test item content has been


69
differentiating the gifted from not-gifted students. Cut
off scores were established such that a maximum number of
gifted and a minimum number of not-gifted students would
fall above the cutoff. Therefore, the true positive and
true negative findings were maximized while the false pos
itives and false negatives were minimized. This procedure
required a "value judgment" by the rater as to the amount
of error which would be acceptable. As the cutoff is low
ered to permit more gifted students to exceed it, increas
ing numbers of not-gifted students would also exceed it,
thus increasing the possibility of false negatives. To
assist with this problem, multiple cutoff scores were cal
culated so that the most desirable cutoffs could be deter
mined. Cutoff scores were chosen according to fractions
of standard deviation units for the total score distribu
tions on the SIT, the new test created by selecting items
with phi, and the new test created by selecting items with
D. The mean SIT IQ for the Phase I sample was 1 37.30 and
the standard deviation was 14.12. This mean IQ was ap
proximately 37 points higher than that of the general pop
ulation. It reflects the above average general intelli
gence of the research sample. In Phase II, these standard
deviation cutoff scores calculated as standard deviations
from the Phase I sample, were applied to the cross valida
tion sample to determine the relative accuracy of gifted
classifications using the NP-phi, NP-D, and SIT.


51
Assessment Procedures
As previously discussed, the five assessment instru
ments used for this research were the Wechsler
Intelligence Scale for Children-Revised (WISC-R), The
Stanford-Binet Intelligence Scale (S~B), the Slosson In
telligence Test (SIT), the Comprehensive Test of Basic
Skills (CTBS) , and Test of Cognitive Skills (TCS).
Administration, scoring, and interpretation of the
WISC-R and Stanford-Binet were conducted by state certi
fied school psychologists prior to the onset of this
study. All tests were individually administered and hand
scored using current norms. The WISC-R yields a Verbal
Scale IQ (representing verbal reasoning abilities), a Per
formance Scale IQ (representing perceptual organization
and nonverbal reasoning), and a Full Scale IQ. Only the
Full Scale score, which represents total IQ, was used as a
measure of gifted intelligence. This procedure conformed
to school district guidelines. The Stanford-Binet yields
a total IQ score only. A cutoff score of 127 was used as
the gifted cutoff in the district. Students attaining a
Full Scale IQ of 127 or greater were considered to have
met the intellectual criterion for gifted program elig
ibility. The IQ cutoff was chosen by the district as a
score that is two standard deviations above the test mean
(WISC-R IQ=130), minus one standard error of measurement
(three IQ points) (S-B IQ = 132 minus 5 IQ points). The


45
achievement construct of the CTBS, based on research con
ducted at McGraw-Hill by Buchet (1974, cited in
TCB/McGraw-Hi 1 1 1984). Empirical criteria for distin
guishing between aptitude and achievement measures were
derived by the publishcrs.
Product moment correlations between the four subtests
of the TCS. were between r = .41 to r = .65. Coefficients be
tween subtests and total score ranged from r-.12 to r=.85.
Therefore, it was suggested that all subtests measure gen
eral intelligence but also measure independent factors. A
correlation coefficient between the CTBS and TCS of r=.78
was calculated on a sample of seventh graders.
Another attribute of the CTBS and TCS is the compre
hensive sampling and norming standards applied. The form
ing samples contained approximately 250,000 students in
grades K-12 from public, Catholic, and other private
schools (CTB/McGraw-Hi 1 1 1984). School districts were
randomly chosen from four geographic regions. Comprehen
sive norming and standardization information is available
in the CTBS and TCS Technical Reports.
Internal reliability coefficients were calculated
according to the Kuder-Richardson formula 20. CTBS relia
bility coefficients ranged from .30 to .96 on the 10 sub
tests (CTB/McGraw-Hil1, 1984). All subtests except Spell
ing and Reference Skills had values at or above .90. On
the four TCS subtests reliability coefficients ranged from


16
junior high school teachers choose contains none of the
students who have been identified as gifted during
previous years. Junior high school teachers must base
their referrals on "the best of the rest." Clearly the
most advantageous time to identify students for a junior
high school curriculum for the gifted is when they enter
seventh grdde.
The new procedure (NP) was developed by selecting
items from popular group tests administered at seventh-
grade, measuring both achievement and cognitive ability.
The rationale for choosing screening items from standar
dized group tests is that this approach to student screen
ing is both time- and cost-efficient. All students in the
population school district take the CTBS and TCS annually
and the screening data are readily available without addi
tional tests or testing time being needed.
It was believed that of the 460 CTBS and TCS ques
tions typically administered to seventh graders, there
existed a subset of items that would accurately predict
gifted IQ on the WISC-R or S-B. Because the number of
CTBS/TCS items is large, sampling of a wide range of
skills and abilities is possible. As the diversity of the
items increases, so does the general ability measured by
the total test (Kaufman, 1979).


82
sample size of 117 remained constant. However, the sample
size diminished below 75 on 99 items. Of these items, 20
had phi values that would have been significant at <.05
had the N been as high as 113.
IRT Parameters
A strong case for the NP item validity is made upon
investigation of item response theory item location param
eters for the CTBS items. An investigation of the
response patterns of subjects on both item sets, in com
parison with the response pattern on these items by the
original CTBS sample, suggest commonality in responding.
IRT item location parameter. Item response theory
was utilized by the CTBS/TCS constructors for item selec
tion. Item characteristic curves were plotted using a
three parameter logistic model involving (a) item dis
crimination, (b) item location (or difficulty), and (c) a
"guessing" factor. A location parameter describes an
item's difficulty in terms of the student's ability level
(or latent trait). An item discriminates best for a stu
dent whose ability level is near the item's location
parameter (or difficulty level). An item with a high
location parameter serves its function only for students
of high ability since low ability students would not be
expected to answer these items correctly.
Comparing response patterns. An assumption of item
response theory is the invariance of item parameters


54
determine which cutoff point(s) most accurately predicted
giftedness. Newly referred students in the school setting
who scored at or above that point would be referred for
formal IQ testing. By adopting a cutoff score two stan
dard deviations below the gifted mean, approximately 97%
of intellectually gifted students would be referred for IQ
testing based on research sample parameters. A cutoff
score incorporating a one standard deviation below the
mean cutoff would delete approximately 16% of the gifted
students from testing and gifted program eligibility.
However, a pitfall of including as broad a range of stu
dents as permitted by the two standard deviations crite
rion was that a relatively large number of not-gifted stu
dents would also be referred for formal IQ testing,- thus
reducing the accuracy of NP predictions of giftedness.
Adjustment of the cutoff point was desirable to ensure an
optimal ratio of gifted students accurately predicted to
not-gifted students inaccurately predicted.
After items for the NP were obtained and multiple
cutoff points was established, Phase II was begun. In
Phase II, test scores based on selected NP items were com
puted for the second sample of students. The NP was an
alyzed in terms of its discrimination between the gifted
and not- gifted students on WISC-R. The accuracy of clas
sification was then compared to accuracy of classification


40
& Galvin, 1987; Salvia & Ysseldyke, 1978; Vandiver & Van
diver, 1979). In much of the research surrounding this
instrument it is suggested that it merits this distinc
tion. Friedes (1978) described the standardization of the
WISC-R as "state of the art" (p. 232) and as meriting
"blue ribbons." In addition, he noted as praiseworthy the
high correlation coefficients between the WISC-R and the
Stanford-Binet.
Reliability coefficients of internal consistency for
the WISC-R Verbal, Performance, and Full Scale IQ scores
reported in the test manual were obtained by utilizing a
formula for computing reliability of a composite group of
tests (Wechsler, 1974). The average reliability coeffi
cients across the range of age levels were V, r=.94; P,
r=.90; and FS, r=.96. The coefficients of individual sub
tests based on split-half or test-retest methods ranged
from r-.ll to .86. Test-retest correlations for the Ver
bal, Performance, and Full Scale IQs ranged from r=.90 to
.95 based on a 3-month interval between tests.
Factor analytic research by Kaufman (1979) has shown
that factors corresponding closely with the Verbal and
Performance Scales of the WISC-R exist. In 1980, Karnes
and Brown factor analyzed the WISC-R on 946 gifted stu
dents ages 6.0 to 16.0. The resulting factors were
consistent with those found by Kaufman on the normal popu
lation. Most verbal scale subtests had factor loadings in


6
composed of items on the CTBS and TCS in a situation in
which giftedness is viewed as a dichotomous variable?
2. Is the NP more accurate than the SIT in classi
fying gifted and not-gifted seventh graders?
3. At what cutoff point(s) is the NP more accurate
than the SIT most accurate in classifying gifted and not-
gifted seventh graders?
Definition of Terms
Comprehensive Test of Basic Skills (CTBS). The CTBS
is a "series of norm-referenced, objective-based tests for
kindergarten through twelfth grade. The series is de
signed to measure achievement in the basic skills commonly
found in state and district curricula" (CTB/McGraw-Hi11,
1984, p. 1). At the junior high school levels the content
areas are reading, spelling, language, mathematics, refer
ence skills, science, and social studies.
Full Scale IQ (FS-IQ). Full Scale IQ is the derived
intelligence quotient on the Wechsler Intelligence Scale
for Children--Revised and on the Stanford-Binet Intelli
gence Scale.
General intelligence. General intelligence is a set
of general cognitive operations measured as the overall
ability required for success on IQ tests. General


85
true classifications for the NP-D and the SIT was greatest
at the -.33 cutoff. Data supporting an NP-D cutoff at
+.66 may be less clear. Kappa analyses indicate that the
number of total true classifications at +.66 is superior
to the number at any other cutoff. The relative strength
of the NP-D in total true classifications was due to its
much greater accuracy in predicting true positives. The
SIT was more accurate in classifying true negatives. In
choosing a cutoff score for gifted IQ screening purposes,
specific classification priorities and goals must be taken
into account. For example, if the primary goal is to max
imize total true classifications, the +.66 cutoff may be
desirable.
Conclusions, Implications, and Limitations
This study yields conclusions, implications, and lim
itations germane to both pragmatic and theoretical issues.
In the following sections, the issues of sampling, gen
era 1izabi1ity, item validity, and cutoff scores are
discussed individually.
Sampling
The success of the NP in classifying two groups of
students, who in some respects appeared indistinguishable,
has some inherent implications. The gifted and not-gifted
children were similar in terms of achievement, IQ (in many


CHAPTER IV
RESULTS
This chapter is presented in a format that sequen
tially reflects the methodological progression of the
study. Phase I will be discussed first to present results
of two analyses used to select items for the new screening
procedure (NP). Next, Phase II cross validation results
are presented in terms of prediction accuracy of the NP-
phi, NP-D, and SIT in relation to cutoff scores.
Phase I Item Selection
Phi Coefficients
The phi coefficient analysis of the 460 CTBS and TCS
items yielded 56 items that discriminated the 59 gifted
and 59 not-gifted seventh graders. This number of items
is over twice as many as would be expected by chance at
the £.05 level of confidence. There were 13 items signif
icant at the £.01 level and three significant at £.001
(Table 4-1). As also shown in Table 4-1, all CTBS and TCS
subtests contributed items except for CTBS Reference
Skills and TCS Sequences.
Not all of the 56 significant phi analyses items were
retained for use in Phase II. In five cases the phi items
62


32
Table 2-1. Summary of Relationship between
WISC-R and WRAT.
WRAT
WISC-R
Reading
Spelling
Arithmetic
Brooks (1977) N=
30; 6-10 years
V..S.
IQ
6 4d
55
74
P.S.
IQ
71
70
71
F.S.
IQ
70
65
76
Hartlage and Steele (1977)
N = 3 6 ;
Mean age = 7 yrs 9 months
V.S.
IQ
75
35
76
P.S.
IQ
54
33
67
F.S.
IQ
68
35
76
Schwarting and Schwarting
(1977) N=282;
6-16 years
(a) 6-11 yrs
V.S.
IQ
68
61
69
P.S.
IQ
63
60
69
F.S.
IQ
72
65
75
(b) 12-16 yrs
*
V.S.
IQ
74
69
66
P.S.
IQ
40
34
55
F.S.
IQ
62
56
66
Hale (1978) N=155; 6-16 years
V.S.
IQ
54
49
64
P.S.
IQ
29
26
44
Full Scale correlations not quoted.
aDecimal points omitted.
In 1982 a follow-up to the studies summarized in
Table 2-1 was conducted by Grossman and Johnson. In their
study, 77 students ages 6 to 16 were administered the
WISC-R and the WHAT. Factor scores were computed on two
of Kaufman's (1975) factors (Verbal Comprehension and


30
Intellect (SOI) Gifted Screener with the WISC-R. The sub
jects were 3239 elementary school students. A multiple
correlation of r=.337 was significant at <.05; however,
only 11% of the variance in WISC-R scores was explained by
the Gifted Screener. No predictor variable contributed to
a significant multiple correlation coefficient when Full
Scale IQ or any combination of WISC-R subtests was used as
the criterion variable. Stenson concluded that the Gifted
Screener should not be used to predict WISC-R IQ for
gifted program prospects.
In 1985, Clarizio and Mehrens evaluated the technical
data manuals for the SOI to determine the test's value as
a screening test for gifted intelligence. It was con
cluded that "the SOI model has severe psychometric limita
tions" (p. 119). These limitations center around poor
reliability, inadequate normative data, and poor external
validity for many of the factors measured by the test.
Achievement Tests
In research introduced in Chapter I it was suggested
that achievement tests (CTBS) and cognitive ability tests
(TCS, WISC-R, S-B) measure much the same construct. In
the supportive literature were indications that distinc
tions between achievement tests and cognitive ability
tests are often unclear. Correlational and factor
analytic studies lend credence to this contention.


7
intelligence is comprised ol those traits commonly
measured by IQ test items.
Gifted intelligence. Gifted intelligence is measured
as an intelligence quotient that falls in the Very Supe
rior category or two standard deviations above the test
mean. For this research, gifted IQ=130 -1 SEM (where
SEM = 3 IQ points) on the WISC-R, and IQ = 132 -1 SEM (where
SEM = 5 IQ points) on the Stanford-Binet, or IQ=127. (This
definition, which includes a SEM, is based on guidelines
from Hillsborough County Florida School District).
High-achievers. Students who, because of superior
academic performance, have been referred for gifted pro
gram testing, but who are assessed as not intellectually
gifted are referred to as high-achievers.
Intel1igence. Intelligence is "the aggregate global
capacity of the individual to act purposefully, to think
rationally, and to deal effectively with his environment"
(Wechsler, 1958, p. 7). Intelligence is comprised of in
tellective factors such as "abstract reasoning, verbal,
spatial, numerical, and other factors" (Wechsler, 1950, p.
78), and nonintel1ective factors consisting of "capacities
and bents dependent upon temperament and personality which
are . factors of personality itself" (Wechsler, 1950,
p. 78).
Intelligence quotient. An intelligence quotient is a
derived total or Full Scale score on an intelligence test.


61
equivalence. One might argue that since the items admin
istered to a 9-year-old and an 11-year-old are different
and represent different test difficulty levels, the two
students are, in fact, being tested on different scales.
This argument implies that gifted intelligence for the 9-
year-old is not equivalent to gifted intelligence for the
11-year-old. 'To the contrary, while imperfections in test
stability will cause some fluctuation in IQ over time, the
adoption of the deviation IQ (Wechsler, 1974) permits com
parison of scores over age levels.


89
Indeed, none of the NP-D items were obtained from the TCS.
The NP-phi test, which initially included eight TCS items,
was found to be generally less accurate in classifying
students than the NP-D. Essentially, the achievement
items had greater criterion related validity than did the
cognitive ability items for Full Scale IQ.
Ostensibly, the superior criterion validity of the
CTBS items is contrary to expectation, however a closer
examination suggests that this pattern may support rather
than refute the genera 1 izabi1ity of research findings.
Examination of CTBS, TCS, and WISC-R subtests reveals that
the CTBS probably has more in common with the WISC-R than
does the TCS. For example, CTBS subtests such as Mathe
matics, Science, Social Studies, and Vocabulary have what
appear to be direct correlates on the WISC-R. The rela
tionship between TCS and WISC-R subtests appears more ob
scure. This assertion is supported by Wurster (1985) who
compared overlap between the TCS and WISC-R with the SIT
and the WISC-R. She found that 87.8% of the SIT items
measured the same skills as the WISC-R Verbal Scale, how
ever, "no items from the TCS appeared to measure any of
the skills that are measured by the 11 WISC-R subtests"
(p. 24).
The relationship of commonality between IQ test and
achievement test performance for this sample seems related


22
studied as a method for screening potentially gifted stu
dents .
An early attempt to predict gifted IQ using a short
form test was conducted by Thompson and Findley in 1962.
Finding that the Similarities (S), Information (I), Pic
ture Arrangement (PA), Block Design (BD), and Picture Com
pletion (PC) WISC subtests could be effectively used for
this purpose, Thompson and Findley published the Califor
nia Abbreviated WISC for the Intellectually Gifted (CAW-
IQ) in 1966).
In their study, Killian and Hughes (1978) measured
the effectiveness of the Lyman short form (Lorr & Meister,
1942) and the Vocabulary-Block Design subtests of the
WISC-R dyad for predicting IQ on the Stanford-Binet and
WISC-R respectively. Subjects were 142 students between
5- and 15-years-old possessing a mean IQ of 125. Results
indicated a correlation of r=.92 between the WISC-R and V-
BD dyad whereas the Stanford-Binet and Lyman scores were
correlated at r=.78. Killian and Hughes did not present
results of the actual number of students correctly pre
dicted as gifted. They did, however, indicate that 32% of
the students had short forin/Pull Scale IQ score discrep
ancies of 6 points or more.
Employing a much larger sample of students than did
previous researchers, Karnes and Brown (1981) used


97
Karnes, F. A., Edwards, R. P., & McCallum, R. S. (1986).
Normative achievement assessment of gifted children:
Comparing the K-ABC, WRAT, and CAT. Psychology in
the Schools, 23, 346-352.
Kaufman, A. S. (1975). Factor analysis of the WISC-R at
eleven age levels between 6 1/2 and 16 1/2 years.
Journal of Consulting and Clinical Psychology, 4_3,
135-147.
Kaufman, A. S. (1979). Intelligence testing with the
WISC-R. Hew York: John Wiley and Sons.
Kendall, P. C., & Little, V. L. (1977). Correspondence of
brief intelligence measures to the Wechsler scales
of delinquents. Journal of Consulting and Clinical
Psychology, 45, 660-666 .
Killian, J. B., & Hughes, C. D. (1978). A comparison of
short forms of the Wechsler Intelligence Sale for
Children-Revised in the screening of gifted refer
rals. Gifted Child Quarterly, 22, 111-115.
Kolloff, P. B., & Feldhusen, J. F. (1984). The effects of
enrichment on self concept and creative thinking.
Gifted Child Quarterly, 28, 53-57.
Kramer, J. J., Markley, R. P., Shanks, K., & Ryabik, J.
E. (1983). The seductive nature of WISC-R short
forms: An analysis with gifted referrals. Psychology
in the Schools, 20 137-141 .
Lawrence, D., & Anderson, H. N. (1979). A comparison of
the Slasson Intelligence Test and the WISC-R with
elementary school children. Psychology in the
Schools, 16, 361-364.
Lennon, R. T. (1978, March). Perspective on intelligence
testing. Address presented at the National Council on
Measurement in Education, Toronto.
Lorr, M., & Meister, R. K. (1942). The optimum use of
test data. Educational and Psychological Measure-


CHAPTER II
LITERATURE REVIEW
Support for the Problem
The psychometric screening of "gifted intelligence"
is beset by problems of not only time- and cost-efficiency
but of predictive accuracy. In the relevant literature it
is suggested that these problems exist for a variety of
screening methods and procedures. These problems are not
of recent origin, however. As early as 1959, Pegnato and
Birch found that sufficient psychological services were
rarely available to test all of the gifted class candi
dates, thereby necessitating procedures for screening
prior to formal testing. Accordingly, the authors con
ducted an investigation of the "relative efficiency and
effectiveness" (p. 300) of seven procedures for locating
gifted children in junior high schools: teacher ratings,
class rank, creative ability in art or music, student
council membership, superiority in mathematics, group
achievement, and group IQ. Seven hundred eighty-one met
ropolitan school district students were selected for par
ticipation in the study on the basis of high ratings in
one or more of the seven categories. All of the partic
ipants received the Stanford-Binet. Scores on this
1 8


84
An examination of total true classifications (i.e.,
correct classifications of gifted and not-gifted students)
revealed greatest prediction accuracy at the 0.0 SD cut
off. At this cutoff, NP-D and NP-phi values were equal
(67.2) and slightly superior to the SIT (65.6). When per
centage values were corrected for chance occurrences by
the Kappa analyses, NP-D was slightly more accurate at the
+.66 SD than at 0.0 SD, and NP-D items were superior to
NP-phi and the SIT. Furthermore, at the +.66 cutoff, the
Kappa value for the NP-D is .342 contrasted to much lower
values for NP-phi (.076) and SIT (.204).
In answering the second research question, data con
sistently supported the NP as more accurate than the SIT
in classifying gifted and not-gifted students. Among the
two NP tests, NP-D was generally superior to the NP-phi.
The NP-D was clearly the better measure of true positive
and negative classifications based on Kappa analyses.
There is some ambiguity regarding the overall most
accurate cutoff for classifying students, rendering the
last research question more difficult to answer. There is
empirical support for the NP-D at both the -.33 and +.66
cutoffs as most accurate for classifying gifted and not-
gifted seventh graders in the sample. Kappa analyses sup
port that finding in that the discrepancy between total


23
Si1verstein's (1970) method of deriving "the best short
form combinations" (p. 169) to obtain an accurate gifted
IQ predictor. Silverstein's method takes subtest unreli
ability into account when measuring predictive ability.
Nine hundred, forty-six gifted children ages 6.0 to 16.0
(X chronological age [CA] = 9.9) served as subjects.
Karnes and Brown found that the WISC Block Design subtest
was represented frequently in subtest combinations that
correlated with WISC Full Scale IQ. Supporting Killian
and Hughes' findings, the V-BD dyad was found to be the
most accurate for predicting gifted IQ. The use of sub
test tetrads was found to be useful, increasing correla
tion coefficients from .628 to .734. Again, actual accu
racy ratios were not provided in the study.
Proceeding under the notion that, "since a short form
IQ test is composed entirely of some subset of questions
of items taken directly from a full-length IQ test, a
short form would seem to be an ideal predictor of full-
length IQ test performance" (p. 40), Dirks, Wessels, Qua-
forth, and Quenon (1980) compared various short form com
binations with Full Scale IQ on the WISC-R. Subjects con
sisted of 47 fourth graders with a mean IQ of 123 (range =
106 to 144). Twelve WISC-R subtest combinations were
studied. It was revealed that the short form combinations
of Similarities, Object Assembly and Vocabulary and S-OA


93
Blosser, G. H. (1963). Group intelligence tests as
screening devices in locating gifted and superior
students in ninth grade. Exceptional Children, 29,
282-286.
Braden, J. P. (1985). A modest proposal: Using probabil
ities of special-education eligibility instead of
cutting scores. Unpublished manuscript.
Brock, H. (1982). Factor structure of intellectual and
achievement measures for learning disabled children.
Psychology in the Schools, _lj), 297-304.
Brooks, C. R. (1977). WISC, WISC-R, S-B L-M, WRAT: Rela
tionships and trends among children ages six to ten
referred for psychological evaluation. Psychology in
the Schools, 14, 30-33.
Bryan, J. R., & Bryan, T. H. (1975). Understanding learn
ing disabilities. Sherman Oaks, CA: Alfred Publish
ing Co.
Burket, G. R. (1974). Empirical criteria for distinguish
ing and validating aptitude and achievement
measures. In D. A. Green (Ed.), The aptitude-
achievement distinction (pp. 35-51). Monterey, CA:
CTB/McG raw-Hill.
Carleton, F. 0., & Stacey, C. L. (1954). Evaluation of
selected short forms of the Wechsler Intelligence
Scale for Children. Journal of Clinical Psychology,
10, 258-261.
Carroll, J. B. (1966). Factors of verbal achievement. In
A. Anastasi (Ed.), Testing problems in perspective
(pp. 406-413). Washington, D.C. American Council on
Education.
Chambers, J. A. (1960). Preliminary screening methods in
the identification of intellectually superior chil
dren. Exceptional Children, 26, 145-150.
Chambers, J. A., Barron, F., & Sprecher, J. W. (1980).
Identifying gifted Mexican-American students. Gifted
Child Quarterly, 24, 123-128.
Clarizio, H. F., & Meherens, W. A. (1985). Psychometric
limitations of Guilford's structure-of-intel1ect
model for identification and programming of the
gifted. Gifted Child Quarterly, 29, 113-119.


methods used to identify gifted students revealed that
most methods were less than accurate.
The Comprehensive Test of Basic Skills and Test of
Cognitive Skills were item analyzed on a sample of 179
seventh grade subjects using two correlational procedures.
Results from the item analyses yielded 46 items from the
phi analysis and 14 from the index of discrimination anal
ysis that discriminated gifted from not gifted students,
as classified by the WISC-R or Stanford Binet. Cross val
idation was conducted on a second sample of 61 students to
determine (a) the accuracy of the new screening procedure
(NP) compared to the SIT and (b) at what cutoff points
either the NP or SIT was more accurate. In cross valida
tion, total scores of both item subsets were compared with
SIT. The items obtained from the index of discrimination
analyses were found to be the best predictors of "gifted"
intelligence for the sample. There was some ambiguity re
garding an optimal cutoff score; however, a score .33
standard deviations below the mean score was found to gen
erally yield most accurate predictions.
vi


103
Yarborough, G. H., & Johnson, R. A. (1983). Identifying
the gifted: A theory-practice gap. Gifted Child
Quarterly, 27, 135-138.
Yule, W. Gold, R. P., Busch, C. (1981). WISC-R corre
lates of academic attainment at 16 1/2 years. Brit
ish Journal of Educational Psychology, 51, 237-240.
Zimet, S. G., Farley, G. K., & Dahlen, N. W. (1985). An
abbreviated form of the WISC-R for use with emotion
ally disturbed children. Psychology in the Schols,
72, 19-22.


83
(Anastasi, 1982; Baker 1984). That is, given certain con
ditions, item parameters should be uniform among different
populations because individual items are assumed to meas
ure the same trait in different populations. In this
case, the validity of the NP items would be supported if
NP items were common to CTBS items with high location
parameters. In fact, 41 of the original 65 NP items had
CTBS location parameters above the mean for the norming
sample (CTBS Technical Report, 1983). The proportion of
NP-D items to high location parameter items was even
greater (16 of 24 67%).
In summary, the data support an affirmative answer to
the first research question. There exists a subset of
items that discriminated gifted and not-gifted seventh
graders who were all academic high achievers. In Phase
II, cross validation with a smaller but otherwise equiv
alent sample was conducted to support these findings.
Phase II
Cross validation of Phase I findings was conducted in
Phase II by assessing the accuracy of NP items as compared
to the SIT in classifying the gifted and not-gifted stu
dents and the cutoff score at which each procedure showed
greatest accuracy. These goals were accomplished by an
alyzing proportions of correct and incorrect classifica
tions using coefficient Kappa analyses.


UNIVERSITY OF FLORIDA
3 1262 08554 6736


99
Nicholson, C. L.(1977). Correlations between the Quick
Test and the Wcchslcr Intelligence Scale for Chil-
dren-Revised. Psychological Reports, 4 0, 523-526.
Olsen, A. V. & Rosen, C. L. (1971, February). Explora
tion of the structure of selected reading readiness
tests. Paper presented at the meeting of the Amer
ican Educational Research Association, New York.
Pearce, N. (1983). A comparison of the WISC-R, Raven's
Standard Progressive Matrices, and Meeker's SOI-
Screening Form for gifted. Gifted Child Quarterly,
27, 13-18.
Pedriana, A. J., & Bracken, B. A. (1982). Performance of
gifted children on the PPVT and PPVT-R. Psychology
in the Schools, 15, 183-185.
Pegnato, C. W., & Birch, J. W. (1959). Locating gifted
children in junior high schools--A comparison of
methods. Exceptional Children, 25, 300-304.
Petty, M. F., & Field, C. J. (1980). Fluctuations in
mental test scores. Educational Research, 22, 198-
202 .
Reschley, D. J., * Reschley, J. E. ( 1979 ). Validity of
WISC-R factor scopes in predicting achievement and
attention for four socio-cultural groups. Journal of
School Psychology, 17, 355-361.
Ricca, J. (1984). Learning styles and preferred instruc
tional strategies of gifted students. Gifted Child
Quarterly, 12, 121-126.
Ritter, D., Duffy, J., & Fischman, R. (1973). Comparabil
ity of Slosson and S-B estimates of intelligence.
Journal of School Psychology, 3, 224-227.
Roach, P. A. (1979). The effects of conceptual style
preference, related cognitive variables and sex on
achievement in mathematics. British Journal of Edu
cational Psychology, 49, 79-82.
Rust, J. D., & Lose, B. D. (1980). Screening for gifted
ness with the Slosson and the Scale for Rating
Behavioral Characteristics of Superior Students.
Psychology in the Schools, 17, 446-451.


correlated. The research also provides reason to believe
that a common factor underlies performance on both types
of tests.
Summary
In the literature relevant to the accuracy of various
procedures for screening intelligence among gifted stu
dents, there have generally been mixed results. Some
encouraging findings have occurred on studies of achieve
ment ratings and short-form IQ tests. Group IQ tests have
tended to underestimate the IQs of some gifted children,
though they generally predicted gifted IQ with moderate
accuracy. Much less effective means for predicting gifted
IQ are IQ screening tests.
The preceding literature has focused on the problems
of screening gifted intelligence among the school age pop
ulations. Those procedures that have shown some success
have been inconsistent in their findings. Nearly all have
proved inefficient in terms of time and cost.
Instruments Used in Study
Wechsler Intelligence Scale for Children-Revised (WISC-R)
The WISC-R has been the most widely administered test
of children's intelligence (Bryan & Bryan, 1975; Grossman


33
Freedom from Distractibi1ity) and WRAT subtests. A multi
ple regression analysis was computed wherein WISC-R factor
scores served as conjoint predictors and the WRAT standard
scores were employed as criterion variables. Results in
dicated a significant overall prediction of WRAT reading,
spelling and arithmetic by the two WISC-R factors.
Wright and Dappan (1982) assessed 250 students with a
mean age of nine years on the WISC-R and WRAT. Factor
analysis showed a common factor for subtests on both mea
sures. Correlations between subtests from the two tests
were as high as r=.60 (on WISC-R, Arithmetic and WRAT,
Arithmetic). Some other subtests correlated at .40 to
.50.
Literature concerning the overlap of individual tests
of intelligence and tests of achievement include studies
in which the WISC-R and the Peabody Individual Achievement
Test (PIAT) (Dunn & Markwardt, 1970) were examined.
Wikoff (1978) factor analyzed the WISC-R along with the
PIAT for 180 referred children. Although the PIAT General
Information and Mathematics subtests loaded on factors
previously identified in the structure of the WISC-R, the
remaining subtests loaded on a separate factor, subse
quently labeled Word Recognition. The results supported
the use of both instruments as sources of mutual but


75
Table 4-6. Phase II: Proportions of Correct and
Incorrect Predictions for NP (phi),
NP (D), and SIT at Two Score Cutoffs
z -
Cutoff ]
True
Positive
True
Negative
True
Tota 1
False
Negative
False
Positive
False
Total
Phi
- .66
SD
.492
.082
. 574
.033
. 393
.426
- .33
SD
.492
.131
.623
.033
. 344
.377
.00
SD
.475
.197
.672
.049
. 229
. 328
+ .33
SD
.377
.279
.656
. 164
.180
.344
+ .66
L)
- .66
SD
.131
. 393
. 525
. 393
.082
.475
SD
.508
.148
.656
.016
. 328
. 344
- .33
SD
.508
.164
.672
.016
.311
.328
.00
SD
.426
.246
.672
.098
.230
. 328
+ .33
SD
.426
.246
. 672
.098
.230
.328
+ .66
SD
.361
.311
.672
. 164
. 164 '
.328
SIT
- .66
SD
. 525
.115
.639
.016
.344
. 361
- .33
SD
. 459
. 148
.607
.082
.311
.393
.00
SD
. 361
. 295
.656
.180
. 164
. 344
+ .33
SD
.270
.360
.590
.311
.082
.410
+ .66
SD
.164
.426
.590
. 361
.049
.410
Note: All values rounded to nearest .10 percentile.


43
two tests £or three IQ levels across four age groups: be
low 84 84-1 1 6, and above 116. For example, at age 13-6
and above, mean differences between the Stanford-Binet and
SIT are -1.41, -1.10, and 2.62 at the three IQ levels. In
calculating the mean difference, negative means are added
to positive means resulting in a misleadingly low overall
mean difference. In this example the total mean differ
ence for age 13-7 and above is -.67. However, if individ
ual means had been summed in terms of nondirectiona1 devi
ation from zero, the mean difference would have been
approximately 1.7. With regard to the mean difference for
the entire sample, when the nondirectiona 1 procedure is
used the difference changes from -.04 to approximately
1.4. The mean scores are rendered even more difficult to
interpret because no standard errors for the means are
reported.
In spite of the apparent inconsistencies in the new
SIT manual, the revisions, particularly in its renorming,
represent considerable improvement in the test's validity
and reliability. These test improvements, along with the
ease of its administering and scoring, render the SIT a
test of considerable utility as an intellectual screening
procedure.


98
Male, R. A., & Perrone, P. (1979 ). Identifying talent and
giftedness. Roeper Review, 2, 5-11.
Mallinson, G. G. (1963). An analysis of the factors re
lated to the motivation ind achievement of students
in science courses in the junior and senior high,
final report (Report No. CRP-503). Kalamazoo, MI:
Western Michigan University School of Graduate
Studies. (ERIC Document Reproduction Service No. ED
002889 )
Martin, J.D., & Kidwell, J. C. (1977). Intercorrelations
of the Wechsler Intelligence Scale for Children-
Revised, the Slosson Intelligence Test, and the
National Educational Developmental Test. Educationa1
and Psychological Measurement, 37, 1117-1120.
Martin, J. D., & Rudolph, L. (1972), Correlates of the
Wechsler Adult Intelligence Scale, the Slosson In
telligence Test, ACT scores and grade point aver
ages. Educational and Psychological Measurement, 32,
459-462.
Mayfield, B. (1979). Teacher perception of creativity,
intelligence and achievement. Gifted Child Quar-
terly, 23, 812-817.
McNemar, Q. (1962). Psychological statistics (3rd Ed.).
New York: John Wiley.
Meister, R. K., & Kurko, V. K. (1951). An evaluation of a
short administration of the revised Stanford-Binet
Intelligence Examination. Educational and Psycholog
ical Measurement, 11, 489-493.
Mercer, J. R. (1979). In defense of racially and cultu
rally non-discriminatory assessment. School Psychol
ogy Digest, 8, 89-115.
Mize, J. M., Smith, J. W., & Callaway, B. (1979). Compar
ison of reading disabled childrens scores on the
WISC-R, Peabody Picture Vocabulary Test and Slosson
Intelligence Test. Psychology in the Schools, 16,
356-358.
Nichols, J. E.(1962). Brief forms of the Wechsler Intel
ligence scales for Research. Journal of Clinical
Psychology, 18, 167.


34
supplementary information in the assessment of learning
problems.
Dean (1977) assessed the degree of redundancy between
the W1SC-R and the PIAT using a canonical correlation
analysis with scores from 205 referred children. The re
sults indicated that 65% of the functions of the PIAT
overlapped with the WISC-R and that 37% of the functions
of the WISC-R overlapped with the PIAT. The overlap was
attributed to common verba1-educationa1 content. Dean
(1982) found a similar asymmetrical overlap between these
measures in samples of 100 Anglo and 100 Mexican-American
children. As in Wikoll s factor analysis, both of Dean s
analyses showed the PIAT subtests of reading and spelling
to offer the greatest degree of information not redundant
with the WISC-R.
Brock (1982), finding that a paucity of research
existed for factor analytic investigations of the WISC-R
in combinations with individual achievement tests, con
ducted such a study. He factor analyzed the WISC-R, WRAT,
and PIAT for 183 male students in grades 3 through 6. An
attempt was made to determine the traits or common skills
measured by IQ and achievement tests when viewed concom
itantly. Four factors emerged. One, a numerical factor,
was comprised of subtests from all three tests.


13
should conform to that construct, thus allowing discrim
ination of gifted from not-gifted students as do the WISC-
R or Stanford-Binet.
Support for the common item content of the CTBS, TCS,
WISC-R, and Stanford-Binet will be provided here by
illustrations of actual items found on these tests. Items
are categorized arbitrarily according to subtest classifi
cation and to face commonalities. A brief description and
reference location will be given for those non-verbal test
items that cannot be readily reproduced in this format.
Items indicated are those designed for average and above
average seventh graders. Test items are printed in bold
face .
1. WISC-R, Vocabulary-instructions: What does
mean?
rivalry
CTBS, Vocabulary (instructions: Choose the word or
phrase that means the same, . ., as the underlined
word. )
their opponent
A. foe
B. employee
C. architect
D. assistant
2. WISC-R Arithmetic
Tony bought a second hand bicycle for $28.00.
He paid 2/3 of what the bicycle cost new. How
much did it cost new?
CTBS, Mathematics Concepts and Applications
Homer's recipe will make 48 sugar cookies. He
made 3/4 of this recipe for a party. How many
cookies were for each of the 18 people at the
party?


52
Slosson Intelligence Test was administered to students as
an IQ screening procedure by school guidance counselors or
curriculum specialists who typically had little formal
training in administration of individual intelligence
tests. SITs were given to students within one year prior
to WISC-R testing. SIT protocols were hand scored by the
test administrators. Under usual circumstances, a total
IQ of 135 was used as a screening cutoff. Children who
scored at or above this cutoff point were normally re
ferred to the school psychologist for WISC-R evaluation.
There were some exceptions to this rule because occasion
ally students who did not score at or above the cutoff
were referred. Generally, these students exhibited ex
tremely high academic skills or other competencies that
compelled school personnel to refer them for formal test
ing. The SIT yields a total IQ score. The 135 cutoff
score is two standard deviations above the test mean.
The CTBS and TCS were administered to students by
classroom teachers in group format according to standard
ization procedures found in the teacher's manual. The
CTBS and TCS were designed to be easily administered
(Ahmann, 1972), and teachers have received little formal
training in their administration. The seventh-grade level
of the CTBS (Level H) yields subscale scoring in Reading
(two sections), Spelling, Language (two sections),


Table 4-1. Continued.
Item No.
Subtest
Phi-Value
Significance
*(2)343
Soc. St.
. 2812
$.05
(2)364
Soc. St.
. 3076
$.05
(2)366
Soc. St.
. 3921
$.01
( 2 ) 378
Soc. St.
. 3076
$.05
407
TCS-Anal.
.1977
$.05
*421
Mem.
.2148
$.05
425
Mem.
.1916
$.05
429
Mem.
. 2035
$.05
434
Mem.
.1835
< .05
*438
Mem.
. 2068
$.05
*440
Verb. Reas.
.1858
$.05
457
Verb. Reas.
. 2168
$.05
* Item discriminates in favor of not-gifted
2 Answered by less than 90% of sample.


86
cases), and teacher perceptions of them as gifted. These
similarities suggest that the analysis results are robust
because the two groups were accurately differentiated in
spite of their likenesses.
The specific trait of the research sample as being
high academic achievers distinguishes this study from many
others and this distinction is essential to the utility of
the NP. In most of the related literature reviewed in
Chapter II, research samples were not as restricted in
range of IQ, rendering correct classification of gifted
and not-gifted more likely in those. In other studies,
many samples included students randomly selected from
general populations. Those students, excluded in this
study by preselection, were readily screened as not-gifted
in other studies, allowing for artificially inflated accu
racy predictions.
A pertinent limitation to this study discussed in
Chapter III was the sampling procedure. Specifically,
some students were deleted from formal IQ testing, and
therefore, excluded from the sample because they scored
below the SIT screening cutoff of 132. However, one-sixth
(i.e., 11 of 61) Phase II students had SIT scores below
132, with some scores falling in the average and below
average range. This occurrence raises a question about
the criterion used for disqualifying students from IQ


91
(1979) and Rust and Lose (1980) who found that the SIT
tends to overestimate high IQs on full length tests. In
their studies, those researchers recommended setting SIT
cutoff scores for gifted classification higher than two
standard deviations above the mean to offset this tendency
to overestimate the IQs of brighter students. The current
findings support those recommendations because the SIT was
most accurate in classifying true positive and false neg
ative findings at over three standard deviations above the
mean (SIT IQ = 149).
Summation
In his later years Edwin R. Guthrie (1959) suggested
that research has no inherent value, but rather that its
value was gained from its practical applications. It is
in the spirit of that philosophy that this research may be
fully appreciated. Relatively accurate in its predictive
power and efficient in its method, the NP analysis will be
easily replicated on new student samples and test formats.
Thus, the NP may be best viewed as a procedure adaptable
to varying, yet specific, needs.


88
a particular cutoff score. Other, more theoretical def
initions, may involve other qualitative differences.
Results of this study may not be generalizable to students
who are designated gifted using criteria other than their
IQs falling above the 96th percentile.
Screening Accuracy
Genera1 izabi1 ity of research results would be en
hanced by analyzing group test data for other restricted
populations to determine if subgroups may be successfully
discriminated using test items. Two preliminary steps in
such a process would be to (a) analyze test results of
other populations of potentially gifted students, such as
elementary or high school, on the CTBS/TCS, and (b) deter
mine if other group achievement/cognitive ability tests
possess items that accurately classify potentially gifted
students. Further, researchers may seek to generalize
results on more divergent restricted populations. For
example, it may be useful to screen for mild mental retar
dation among remedial students or for learning disabil
ities among children with discrepant report card grades.
Item Validity
A somewhat unexpected result of this research was
that the CTBS items (which supposedly measure school
achievement) contributed much more to the NP than did the
TCS items (which are purported to measure intelligence).


41
Perceptual Organization. These studies strongly support
the validity of the WISC-R.
Slosson Intelligence Test
The SIT is an IQ test for children and adults de
signed for use by either relatively untrained examiners or
qualified professionals. The SIT typically takes between
10 and 30 minutes to administer.
New norms (1982) represent a significant departure
from procedures previously employed (1961) for calculating
an IQ score. In norming the SIT, the Stanford-Binet was
used as the anchor test. Consistent with the 1974 revi
sion of the Stanford-Binet, ratio IQs were abandoned in
favor of deviation IQs. Frequency distributions were cal
culated for each of the 19 chronological age ranges on
both IQ scales. Then, utilizing a "modified table look-up
approach," appropriate IQs from the Stanford-Binet were
entered on the developing SIT tables. The mean IQ for the
SIT is 100 and the standard deviation is 16.
In the SIT manual, the authors present evidence to
persuade the reader that the revised SIT IQs are equiva
lent to Stanf ord-Binet IQs. This task is undertaken, in
large part, by comparing previous (1961), less positively
correlated coefficients to newer data,
Slosson and Jensen (1982) stated that "the SIT is as
accurate as the Stanford-Binet in measuring a person's


50
County (Florida) Public School District. Sampling was
conducted at the end of the 1984-85 school year. Nearly
all students had previously been administered the SIT. In
most instances, only those students who scored two stan
dard deviations or more above the mean had been adminis
tered the WISC-R or Stanford-Binet. All students in the
population also had current CTBS/TCS scores on file. Some
students in this population had met intellectual eligibil
ity guidelines for the gifted program (on WISC-R or S-B
criteria) and some had not. All students were tested by
school psychologists during each of the three school years
under investigation.
Simple random sampling was conducted by the research
er at the Hillsborough County School Board office in June
of 1986. Names of the seventh graders who were tested for
the gifted education program from September of 1982 and
June of 1985 were obtained from computer printouts con
taining data for all students in the district who had been
tested by school psychologists. Students in the sample
pool were assigned a number, and numbers were selected
according to a random number table. Numbers were then
recorded and returned to the pool to ensure an equal
chance of selection for the remaining numbers. After 61
students were randomly selected for Phase II, the remain
ing 118 students were assigned to Phase I.


21
More recently, Harrington (1982) also found that
group IQ tests tend to underestimate the IQs of many in
tellectually gifted students. According to Harrington,
for every student identified as gifted on a group IQ test,
one gifted child is not referred. Harrington suggested
that the higher the ability level, the greater the dis
crepancy between individual and group IQ scores. He also
found that a child's IQ may vary by as much as 30 points
between group and individual tests. further, because
there may be a very small number of items at the greater
difficulty levels on group tests, a child may have to per
form perfectly to be predicted as gifted.
Intelligence Quotient Short Forms
So-called IQ short forms are comprised of abbreviated
versions of individually administered standardized intel
ligence tests. Typically selected for short forms are
subtests of the WISC-R or items from the Stanford-Binet.
Short forms of the Wechsler Scales and Stanford-Binet have
been studied extensively (Birch, 1955; Carleton & Stacey,
1954; Enburg, Rowley, & Stone, 1961; Findley & Thompson,
1958; Grossman & Galvin, 1987; Meister & Kurko, 1951;
Nichols, 1962; Simpson & Bridges, 1959; Wright & Sandry,
1962; Wade, Phelps, & Falasco, 1986; Yakowitz & Armstrong,
1955; Zimet, Farley, & Dahlen, 1985). However, it is only
since 1978 that short forms have been relatively widely


24
were each good predictors of Full Scale IQ. Although cor
relations on the BD subtest were high, as shown in previ
ous studies, they tended to predict an excessive number of
nongifted students as gifted. The S~OA dyad predicted 8
of 11 gifted students and 4 who were not. The S-OA-V
triad predicted 9 of 11 gifted students and 4 who were
not.
Utilizing the studies by Killian and Hughes (1978)
and Dirks et al. (1980), who noted that V-BD and S-OA
dyads, respectively, were the most effective in predicting
Full Scale IQ, Fell and Fell (1982) evaluated 92 WISC-R
protocols of children previously evaluated as gifted pro
gram candidatos. The students ranged in age from 6-0 to
11-7 (X age = 8.4) and possessed Full Scale IQs of 130 or
greater. Eleven subtest dyads were studied in terms of
frequency with which each produced an estimated IQ > 130.
Greatest predictive accuracy was achieved using the S-V
and S-OA dyads. These correctly predicted 62% as gifted.
The I-BD dyad yielded prediction ratings of only 43%. Not
providing an exact number, the authors indicated that some
gifted children were overlooked. While results are con
sistent with findings by Dirks et al., indicating that the
S-OA dyad is most effective, prediction accuracy was much
lower in this study.
In a fairly recent study, Kramer, Markley, Shanks,
and Ryabik (1983) utilized Thompson and Findley's (1966)


76
Table 4-7. Percentage of Correct Classifications
and Kappa Values for NP (phi and D)
and SIT at Various Cutoff Scores
Cutoff
NP
SIT-P*
NP
SIT-k1
Phi-P*
D-P*
Phi-k^
D-k1
- .66
SD
57.4
65.6
63.9
.114
.289
.234
- -33
SD
62.3
67.2
60.7
.220
.324
.178
.00
SD
67.2
67.2
65.6
. 328
.334
.309
+ .33
SD
65.6
67.2
59.0
.306
.320
.258
+ .66
SD
52.5
67.2
59.0
.076
. 342
.204
*P = percentage
ik = Kappa value


56
toward middle difficulty levels of test items. As previ
ously discussed, the research (CTBS and TCS) items were
designed primarily to assess performance of medium diffi
culty.
Test items were also analyzed according to the index
of discrimination (Ebel, 1965). The difference between
the percentage of gifted students and not-gifted students
passing each item provides an index of item validity that
can be interpreted independently of the size of the par
ticular sample in which it was obtained (Anastasi, 1976).
The index of discrimination (D) has been shown to measure
item validity with equivalent accuracy to other more elab
orate measures (Engelhart, 1965). Similar to the phi, D
values are biased in favor of items with intermediate dif
ficulty levels. A coefficient of .20 or greater was used
as the criterion for selecting an item for the NP. Thus
in Phase I, two forms of the NP screening test were
created; one form, here designated as NP-phi, was based on
items selected using phi coefficients; the other form, NP-
D, was based on items selected using the index of discrim
ination .
For Phase II, analyses using the coefficient Kappa
were conducted to test (a) whether the NP or SIT was more
accurate in classifying gifted seventh graders, and (b) at
what cutoff points either the NP or SIT was more accurate.
The Kappa analysis measured the proportion of correct (and


47
grades 3 through 12 who were identified based on scores
from the Cognitive Abilities Test. The distribution of
scores in this subsample corresponded to the national dis
tribution of the entire sample. The 1972 norms were be
lieved to be based on a more representative sample than
previous norms (Terman & Merrill, 1973).
The reliability of the 1937 Stanford-Binet was deter
mined by correlating IQs on forms L and M administered to
the standardization group within an interval of one week
or less. Such reliability coefficients are thus measures
of both short term temporal stability and equivalence
across the two item samples. In general, the Stanford-
Binet tends to be more reliable for older than for younger
age groups, and for lower than for higher IQs (Anastasi,
1976). Reliability coefficients range from .83 to .98.
The Stanford-Binet is considered a highly reliable test
with most coefficients for the various age and IQ levels
being over .90.
Validity ratings for the Stanford-Binet were obtained
from examination of the test content, from factor
analysis, and from correlations with achievement ratings.
An examination of the Stanford-Binet tasks indicates
assessment of a wide range of reasoning abilities. These
include tasks requiring hand-eye coordination, perceptual
discrimination, arithmetic reasoning, and verbal


73
Table 4-5. Phase II: Number of Examinees in Each
Prediction Category Cutoff/SIT
Cut
off
True
Positive
True
Negative
True
Total
False
Negative
False
Positive
False
Total
-1.0
SD
32
6
38
1
22
23
- .66
SD
32
7
39
1
21
22
- .50
SD
31
7
38
2
21
23
- .33
SD
28
9
37
5
19
24
.00
SD
22
18
40
11
10
21
+ .33
SD
14
22
36
19
6
25
+ .50
SD
12
24
36
20
5
25
+ .66
SD
9
26
35
23
2
25
+ 1.0
SD
5
29
34
27
0
27


25
CAW-IQ on a sample of 73 children, ages 6-0 to 16-7 (X age
= 10-5). All subjects received the WISC-R and all were
analyzed in terms of the S, I, PA, BD, and PC subtest pat
tern. Of the 48 students predicted as gifted, 3 9 were
predicted accurately; of 25 students predicted not to be
gifted, 21 were correctly described. This subtest short
form was considered to be a relatively accurate predictor
of gifted IQ.
Intelligence Quotient Screening Tests
Individually administered intelligence tests designed
to estimate mental ability, usually in 20 minutes or less,
have become a widely used procedure for screening gifted
intelligence. The Slosson Intelligence Test (Slosson &
Jensen, 1982) is one such screening test. It had been
adopted in the Hillsborough County Florida school district
for the purpose of screening the gifted. High correla
tions between the SIT and the WISC-R or Stanford-Binet
have been reported in research findings (Lawrence & Ander
son, 1979; Martin & Kidwell, 1977; Martin & Rudolph, 1972;
Mize, Smith, & Callaway, 1979; Ritter, Duffy, & Fischman,
1973; Slosson & Jensen, 1982; Stewart & Jones, 1976).
However, the few available studies conducted on gifted
samples have not supported the use of the SIT for screen
ing .


BIOGRAPHICAL SKETCH
Randy Evan Schnell was born in Cleveland, Ohio, on
January 24, 1955. Reared in Hollywood, Florida, from age
seven, he.attained his B.A. in psychology with a minor in
education from Florida Atlantic University in 1977. Pur
suing a field of graduate studies that encompassed both
psychology and education, Randy attained his Ed.S. (1983)
and Ph.D. (August, 1987) from the University of Florida.
Professionally, Randy has worked as a school psychol
ogist in both urban and rural settings in Florida. He is
currently working at Memphis City Schools Mental Health
Center in Tennessee where he is coordinator of the Adoles
cent Sex Offender Program of the Memphis, Shelby County
Sex Abuse Project.
104


71
Table 4-3. Phase II: Number of Examinees in Each
Prediction Category by Cuto£f/Phi
True
Positive
True
Negative
True
Total
False
Negative
False
Positive
False
Total
Cutoff

-1.00
SD
30
5
35
2
24
26
- .66
SD
30
5
35
2
24
26
- .50
SD
30
5
35
2
24
26
- .33
SD
30
8
38
2
21
23
0.00
SD
29
12
41
3
17
20
+ .33
SD
23
17
40
9
12
21
+ .50
SD
18
19
37
11
10
24
+ .66
SD
8
24
32
24
5
29
+ 1.00
SD
3
29
32
29
0
29
Total gifted = 32
Total not-gifted = 29


72
Table 4-4. Phase II: Number of Examinees in Each
Prediction Category by Cutoff/Index of
Discrimination
True
Positive
True
Negative
True
Tota 1
False
Negative
False
Positive
False
Total
Cutoff
-1.00
SD
31
5
36
1
24
25
- .66
SD
31
9
40
1
20
21
- .50
SD
31
9
40
1
20
21
- .33
SD
31
10
41
1
19
20
0.00
SD
26
15
41
6
14
20
+ .33
SD
26
15
41
6
14
20
+ .50
SD
22
19
41
10
10
20
+ .66
SD
22
19
41
10
10
20
+ 1.00
SD
15
24
39
17
5
22


102
Vernon, P. L., Adamson, G., & Vernon, D. F. (1977). The
psychology and education of gifted children. Boul
der, CO: Westview Press.
Wade, D. L., Phelps, L., & Falasco, S. (1986). Use of an
abbreviated version of the WISC-R with learning dis
abled children. Psychology in the Schools, 23, 353-
356.
Washington, E. D., Engelmann, S., & Bereiter, C. (1969).
Achievement components of Stanford-Binet perfor-
tiictiioe .(ERIC Document Reproduction Service No. ED
056771)
Wechsler, D. (1950). Cognitive, conative, and non-intel-
lective intelligence. American Psychologist, 5, 78-
83.
Wechsler, D. (1958). The measurement and appraisal of
adult intelligence. Baltimore: Williams and Wilkins.
Wechsler, D. (1974). Wechsler Intelligence Scale for
Chi1dren-Revised. New York: The Psychological Corp.
Wikoff, R. C. (1978). The WISC-R as a predictor of
achievement. Psychology in the Schools, 16, 364-366.
Wright, D. (1983). Effectiveness of the PPVT-R for
screening gifted students. Psychology in the
Schools, 20 25-26 .
Wright, D., & Dappen, L. (1982). Factor analysis at the
WISC-R and the WRAT with a referral population. Jour
nal of School Psychology, 20 306-312 .
Wright, B. W., & Sandry, M. (1962). A short form of the
Wechsler Intelligence Scale for Children. Journal of
Clinical Psychology, 18, 1966.
Wurster, J. A. (1985). Gifted screening: The Slosson In
telligence Test versus the Test of Cognitive Skills.
Unpublished manuscript.
Yakowitz, J. M., & Armstrong, R. G. (1955). Validity of
short forms of the Wechsler Intelligence Scale for
Children (WISC). Journal of Clinical Psychology, 11,
275-277.
Yapp, K. 0. (1977) Relationships between amount of read
ing activity and reading achievement. Reading World,
17, 23-29.


68
range was 14. Unlike the phi distribution, the ceiling of
the D distribution was probably sufficiently high because
correct responding diminished beyond raw scores of 11.
Only 11 of 117 students received a raw score above 12,
while 65 received scores within one point of the mean.
Taking into account the small number of NP-D items (14),
the standard deviation of 2.6 is considered to be ade
quately large for calculating cutoffs. The range of
scores covered both extremes of the distribution, and the
skew is not as great as might be expected given the gen
erally restricted range of the sample at the upper ability
1evels.
Consistent with previous research (e.g., Engelhart,
1965), there were commonalities between the phi and D an
alyses results for this sample. Sixteen items, or 67% of
the D and 9% of the phi items, were selected from both
analyses. All common items were from the CTBS (on
achievement test) because the D analyses yielded no TCS
(cognitive ability) NP items. Similarly, the one D analy
sis item found to discriminate in favor of the not-gifted
students also did so on the phi analysis.
Cutoff Scores
Cutoff scores for the phi, D, and SIT analyses were
computed in order to demarcate optimal cutting points for


95
Enburg, R., Rowley, V. N., & Stone, B. (1961). Short
forms of the WISC for use with emotionally disturbed
children. Journal of Clinical Psychology, 17, 280.
Engelhart, M. D. (1965). A comparison of several item
discrimination indices. Journal of Educational Mea
surement 2, 69-76.
Fell, L., be Fell, S. S. ( 1982 ). Effectiveness of W1SC-R
short-forms in screening gifted children. Psycholog
ical Reports, 51, 1017-1018.
Findley, C. J. & Thompson, J. M. (1958). An abbreviated
Wechsler Intelligence Scale for Children for use
with educable mentally retarded. American Journal of
Mental Deficiency, 63, 473-480.
Friedes, D. (1978). Review of the WISC-R. In O.K. Buros
(Ed.), The eighth mental measurement yearbook (pp.
414-422). Highland Park, NJ: The Gryphon Press.
Gensley, J. (1979). Parent perspectives or the curiosity
quotient. Gifted Child Quarterly, 23, 723-724.
Gronlund, N. E. (1976). Measurement and evaluation in
teaching. New York: McMillan Publishing Co.
Grossman, F. M., & Galvin, G. A. (1987). Clinically and
theoretically derived WISC-R subtest regroupings:
Predicting academic achievement in a referral popu
lation. Psychology in the Schools, 24, 105-108.
Grossman, F. M., & Johnson, K. M. (1982). WISC-R factor
scores as predictors of WRAT performance: A multi
variate analysis. Psychology in the Schools, 19,
465-468.
Grossman, F. M., & Johnson, K. M. (1983). Validity of the
Slosson and Otis-Lennon in predicting achievement of
gifted students. Educational and Psychological Mea
surement 4 3 617-622 .
Guilford, J. p. (1975). Varieties of creative giftedness,
their measurement and development. Gifted Child
Quarterly, 19, 107-121.
Guthrie, E. R. (1959). Association by contiguity. In S.
Koch (Ed.), Psychology, a study of science (pp. 158-
195). New York: McGraw-Hill.


67
Table 4-2. Items with D Values >.20 and Corres
ponding Subtests
Item No.
Subtest
Upper x
Lower x
D
39
CTBS. Voc.
.930
.667
.263
80
Read.
.912
.712
.200
85
Read.
.912
.712
.200
109
Spell
.931
.731
.200
158
Lang.
.958
.867
.208
176
Lang.
.897
.650
.327
185
Lang .
.879
.667
.212
215
Math
. 842
. 567
.275
230
Math
.873
.552
.323
271
Math
.948
.746
.202
273
Math
.947
.746
.201
277
Math
.737
.525
.212
(2)281
Math
.750
.538
.212
(2)311
Sci.
1.000
.735
.265
317
Sci .
.937
.727
.210
(2)330
Sci.
.937
.727
.210
(2)341
Soc. St.
.875
.667
.208
(2)343
Soc. St.
.688
.909
-.221
351
Soc. St.
.937
.727
. .210
(2)360
Soc. St.
.937
.467
. 240
(2)364
Soc. St.
1.000
.435
.242
(2)366
Soc. St.
.938
. 506
.393
(2)377
Soc. St.
.937
.727
.210
(2)378
Soc. St.
1.000
.758
.242
*Item discriminates in favor of not-gifted.
2 answered correct by less than 90% of the sample.


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Loesch; Chairman
of Counselor Education
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Linda M. Crocker
Professor of Foundations of Education
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Jamps H/ Pitts
Assistant Professor of Counselor
Education


CHAPTER V
DISCUSSION
Research Questions
In order to answer the first research question, this
discussion focuses on issues related to NP item validity
from Phase I. Phase II NP items were selected from the
CTBS and TCS using correlational analyses. To answer the
remaining two research questions, the predictive accura
cies of the NP-phi, NP-D, and SIT are discussed in terms
of cutoff scores.
The following research questions were addressed in
this study.
1. Can an accurate predictor of gifted IQ classifi
cation on the WISC-R/S-B be derived from an in
strument composed of items on the CTBS and TCS
in a situation in which giftedness is viewed as
a dichotomous trait variable?
2. Is the NP more accurate than the SIT in classi
fying gifted and not-gifted seventh graders?
3. At what cutoff point(s) is the NP more accurate
than the SIT in classifying gifted and not-
gifted seventh graders?
80


4
students. Chambers (1960) and Schena (1963) reported
somewhat more encouraging results with academic skill
measures as predictors of gifted intelligence. A more ac
curate and efficient means for screening gifted students
needs to be found.
Giftedness
The characteristics associated with gifted
intelligence are almost as numerous as the students
themselves (Tuttle & Becker, 1980). Qualitative trait
differences between gifted and not-gifted children are
indicated frequently in the literature (Barrington, 1979;
Dirkes, 1981; Gensley, 1975; Male & Perrone, 1979; Ricca,
1984; Ryan, 1982; Sternberg, 1982); however, essential to
a discussion of intellectual giftedness in children is
their classification according to an IQ test cutoff score.
Gifted classifications by IQ cutoff scores are employed in
an attempt to objectify classification criteria and school
placement decisions. Classification by IQ score may mis
leadingly imply that qualitative differences between
gifted and not-gifted children are necessarily demarcated
by an artificial cutoff (Braden, 1985). In this study
cutoff scores are used to quantify gifted intelligence ac
cording to educational criteria and not to define learning
style, motivation, or other personality traits associated
with "giftedness."


38
found to have reasonably high correlations with factors of
arithmetic achievement.
In another study in which factors related to math
performance were investigated, Roach (1979) reported a
significant correlation (r=.80) between arithmetic
achievement and verbal IQ in third graders.
Correlation coefficients for CTBS and TCS subtests
(CTB/McGraw-Hi11, 1984) were calculated using 2813 seventh
graders. Coefficients between .60 and .72 were not uncom
mon. Correlations between the TCS Total Score and CTBS
subscales of Reading, Language, Math, Social Studies and
Science were .71, .71, .68, .65, .71 and .65, respec
tively. The CTBS and TCS Total Batteries correlated at
r=.75. These high correlations suggest that both tests
may measure similar, though operationally distinguishable,
constructs. Support for the contention that these tests
represent similar constructs may also be found in the cor
relations of TCS subscale scores with those scores of its
predecessor, the Short Form Test of Academic Aptitude
(SFTAA). The range of correlations was .55 to .82, not
dissimilar to those of the TCS and CTBS. The fact that
the average correlations were positive means that the sub
scales must be measuring something in common (Jensen,
1980 ) .
In summary, there is substantial evidence that
achievement test scores and IQ test scores are highly


77
An example of the Kappa calculation of percentages of
correct classifications for NP-phi at the -.66 cutoff is
illustrated using four-fold tables (Figure 4-1). Values
in the equation, along with percentages of examinee clas
sifications at other cutoffs, are located in Table 4-6.
P = .492 + .082 = .574
Pc = ( .885 )(. 525) + ( .475)( 115)
= .4646 + .0546
= .5192
Results of the Kappa analyses revealed a consistent
pattern of greater values for the NP-D than for the NP-phi
or the SIT at all cutoffs, suggesting the NP-D may be the
more accurate predictor of correct classification. The
greatest K value for NP-D occurs at the +.66 where the K
value of .342 indicates that there is a 34% improvement in
prediction accuracy over that expected by chance. The
greatest discrepancy between K values for NP-D and the SIT
appears at the -.33 cutoff where K = .324 for NP-D but
only .178 for the SIT.


53
Mathematics (two sections), Reference Skills, Science, and
Social Studies. The Science and Social Studies subscales,
consisting of 86 items, were not administered in all
schools of the population school district; however, these
data were included in this research. In addition to sub
scale scores, the CTBS yields an overall achievement index.
The TCS is comprised of four cognitive ability sub
tests measuring competencies in Sequencing, Analogies,
Memory, and Verbal Reasoning. Derived scores are provided
for subtests and for the overall profile. Because in this
study responses to individual test items were analyzed,
derived subscale and total scores for the CTBS scores were
not utilized.
Research Procedures
In Phase I of this study items comprising the new
procedure (NP) were selected from the CTBS and TCS by con
ducting item analyses of the performance of the 118 stu
dents in the Phase I sample. Two sets of items were de
lineated that, in general terms, were answered correctly
by the gifted students more frequently than by the not-
gifted students.
Gifted cutoff scores on the obtained NP items were
computed by subtracting fractions or multiples of standard
deviations from the mean NP score of the gifted group to


78
IQ
Class Class
Gifted Not-Gifted
Class
Gifted
phi
Class
Not-Gifted
.525 .475
TP
FP
.492
.393
FN
TN
.033
.082
.885
.115
.574 .519
1 .519
.055
TTsT
.114
Figure 4-1. Computation Example of Kappa
analysis for True Phi Classi
fications at -.66 SD Cutoff


63
Table 4-1. Item Phi Values and Significance
Levels
Item No.
Subtest
Phi-Value
Significance
16
CTBS-Voc.
. 2059
£.05
39
Voc.
. 3258
<.001
40
Voc.
.1834
£.05
41
Voc.
. 2069
£.05
49
Read.
.1963
£.05
53
Read.
. 2172
£.05
61
Read.
. 2266
£.05
65
Read.
. 1977
£.05
77
Read .
.1963
£.05
80
Read.
. 2502
£.01
83
Read.
. 2386
£.01
85
Read.
. 2502
£.01
109
Spell.
.2645
£.01
113
Spel1.
. 2285
£.01
120
Spell.
.2199
£.01
122
Lang.
. 2652
£.01
127
Lang.
. 1842
£.05
129
Lang.
. 2069
£.05
136
Lang.
.2187
£.05
137
Lang.
. 2377
£.01
139
Lang .
.2559
£.01
154
Lang .
. 2035
£.05
159
Lang .
. 1842
£.05
171
Lang.
.1913
£.05
176
Lang.
. 2934
£.01
177
Lang.
. 2035
£.05
180
Lang.
. 2018
£.05
182
Lang.
. 1842
£.05
187
Lang .
.1913
£.05
193
Lang.
. 1834
£.05
205
Math
.2187
£ .05
215
Math
. 3007
<01
*216
Math
.1905
£.05
220
Math
. 3245
£.001
230
Math
. 3564
£.001
236
Math
. 2652
£.01
264
Math
.2161
£.05
266
Math
. 2331
£.05
270
Math
.2146
£.05
271
Math
. 2806
£.01
273
Math
.2784
£.01
276
Math
.1885
£.05
277
Math
.2188
£ .05
(2)311
Sci.
. 3214
£.05


66
Index of Discrimination
The index of discrimination analyses, which measured
the difference between the percentage of gifted and not-
gifted students passing each item, yielded a subset of 24
items with D values of .20 or greater. As recommended by
Engelhart (1965), items possessing a D value of at least
20 are considered to show adequate discrimination. All
24 of the acceptable items were obtained from the CTBS.
In Table 4-2 the upper and lower values represent
percentage passing each item for the gifted and not-gifted
students, respectively. D values ranged from .221 through
.393. As might be expected, the highest percentage of
correct responding occurred with the beginning TCS subtest
items. This was because TCS items are ordered hierarch
ically by subtest according to difficulty level.
In this analysis, also, an item was deleted if it
discriminated in favor of the not-gifted group or if it
was not administered to over 10% of the sample. In con
trast to the phi analyses, a large proportion of items
(i.e., 10 or 42%) were deleted. Fourteen items (Table 4-
2) were retained from the CTBS.
In contrast with the phi data, results of scores
based on selected items from the D analyses revealed a
relatively normal distribution, with the mean of 10.02,
median of 10.02, and mode of 11.00 falling within a range
of one test item. The standard deviation was 2.62 and the


CHAPTER I
INTRODUCTION
The tremendous increase since 1970 in school programs
for the intellectually gifted has led most states to es
tablish guidelines or requirements for program eligibility
(Karnes & Brown, 1979; Kolloff & Feldhusen, 1984). One
common requirement is for individual intellectual evalua
tion of gifted program candidates (Chambers, Barron, &
Spreeher, 1980; Sternberg, 1982; Vernon, Adamson, k Ver
non, 1977). Yarborough and Johnson (1983) reported the
use of individual intelligence test minimum scores as an
eligibility requirement in at least 73% of the nation's
gifted student programs. Although educators may recommend
and actually believe in the importance of a broader view
of giftedness, this more narrow definition (i.e., empha
sizing IQ) is frequently employed because of pragmatic
concerns (Jenkins-Freidman, 1982). The "intelligence quo
tient" has therefore emerged as the primary criterion for
gifted program eligibility (Barklay, Phillips, & Jones,
1983; Birch, 1984; Guilford, 1975; Karnes, Edwards, &
McCallum, 1986).
The individual intelligence testing requirement,
coupled with federal regulations mandating services for
1


3
Problem
The research problem addressed concerned the accurate
identification of intellectually gifted students from a
screening procedure prior to administration of an individ
ual intelligence test. The expense of administering indi
vidual intelligence tests has necessitated the screening
of potentially gifted students in an effort to delete from
formal testing those who probably do not possess gifted
intelligence (Rust & Lose, 1980; Stenson, 1982). Inaccu
rate screening has resulted in not-gifted students receiv
ing the time-consuming tests and in some gifted students
being excluded from testing.
Screening procedures such as the Slosson Intelligence
Test (Dirkes, Wessels, Quaforth, & Quenon, 1980; Grossman
& Johhson, 1983; Karnes & Brown, 1979; Rust & Lose, 1980),
the Ammons Quick Test (DeFilippis & Fulmar, 1980; Hirsch &
Hirsch, 1980), short forms of the WISC-R (Bersoff, 1971;
Elman, Blixt, & Sawicki, 1981; Karnes & Brown, 1981; Kil
lian & Hughes, 1978; Kramer et al., 1983), Guilford's
Structure of the Intellect (Pearce, 1983), the Peabody
Picture Vocabulary Test (Mize, Smith, & Callaway, 1979;
Wright, 1983), and group IQ tests (Blosser, 1963; Grossman
& Johnson, 1983; Pegnato & Brich, 1959; Sheldon & Mano-
lakes, 1954) have all been shown to be more or less inac
curate and/or inefficient for screening high ability


12
Schwarting & Schwarting, 1977) or to possess significant
overlap in their factor loadings (Carroll, 1966; Dean,
1977; Grossman & Johnson, 1982; Horn, 1970; Stewart & Mor
ris, 1977; Undheim, 1976; Vernon, 1961, 1969; Wikoff,
1978) .
The TCS, in terms of its title and stated purpose,
encompasses a construct closely related to the WISC-R and
Stanford-Binet in that all three were designed and have
been found to predict academic attainment (Hurrocks, 1964;
CTB/McGraw-Hi11, 1984; Terman & Merrill, 1973; Wechsler,
1974). As described in its 1983 Technical Report, the TCS
subtests (Sequences, Analogies, Memory, and Verbal Reason
ing) load on factors consistent with those described in
WISC-R and Stanford-Binet studies (Kaufman, 1979).
The TCS, according to its constructors, measures "a
number of cognitive abilities included in various
theories, however, like the WISC-R and Stanford-Binet, em
phasis is placed on the kinds of reasoning and retention
skills necessary for school success" (CTB/McGraw-Hi11,
1984, p. 5). The CTBS similarly measures a variety of
academic skill areas shown to be positively correlated
with the TCS. The TCS and CTBS together appear to assess
a theoretical construct common to the WISC-R and Stanford-
Binet. Therefore, the items chosen from the CTBS and TCS