OVERDUE FINES ARE 25¢ PER DAY
PER ITEM

Return to book drop to remove
this checkout from your record.

 

 

 

AN EVALUATION OF CERTAIN MEASURES OF.APTITUDE AND ACHIEVEMENT
IN THE PREDICTION OF SCHOLASTIC SUCCESS

By

James Najeeb Jacobs

AN ABSTRACT

Submitted to the School of Advanced.Graduate Studies of.Michigan
State University of Agriculture and.App1ied Sciences
in partial fulfillment of the requirements
for the degree of

DOCTOR OF EDUCATION

Department of Foundations of Education

Year 1957

Approved

 

@JM

PROBLEM. The problem of this study was to investigate the single
and combined value of seven aptitude and achievement tests in predicting
general high school academic success, as well as success in eight high
school subject areas, 1.6. English, social studies, science, mathematics,

industrial arts, home economics, business education and foreign language.

PREDICTOR LES_T_S_ LIE CRITERIA. The predictor tests included the
following: (1) two measures of achievement, the Mathematics Proficiency
and the English Hoficiency Tests, (2) four "special" aptitude measures
taken from the Differential Aptitude Battery, i.e. Verbal Reasoning,
Numerical Ability, Mechanical Reasoning, and Language Usage, and (3) a
general scholastic aptitude test, the Terman-McNemar Test of Mental
Ability.

Two criteria for scholastic success in subject areas were used.
They were grade point averages and the four subject area measures of the
Essential High School Content Battery. Grade point averages were com-
puted by averaging the final course marks in at least three courses
within a subject area. Composite scores of these two measures served as
criteria for general scholastic success.

Whereas the predictor tests were administered while the students
were in the eighth grade, the criterion measures were secured at least
three years later. Thus, the test validity established in this study is

of the predictive type.

SUBJECTS. Six senior high schools in the Cincinnati Public
School system were originally surveyed. Three of these six schools were

finally singled out for further study because of their uniform marking

practices. The three schools used were identified through the use of an
analysis of covariance of grade point averages adjusted for mean school
levels of scholastic ability. The subjects thus derived consisted of a
total of 595 senior students made up of 266 boys and 329 girls. Only
students for whom complete test records were available were included in

this study.

METHODS AND PROCEDURES. The relationships between the predictor
tests and the criteria were established by Pearson product-moment correl-
ations. The methods of multiple correlation and regression were used to
ascertain the combined predictive power of the test predictors. The
significance of difference between mean predictor test performance of

various sub-groups was tested by means of Student's "t" ratio.

FINDINGS. The general findings of this study may be enumerated
as follows:

1. The Mathematics Proficiency Test proved to be the best pre-
dictor of total grade point average for boys and girls; the zero order
correlations being .657 and .716, respectiveTy. The Terman-McNemar Test
correlated highest with the composite scores of the EHSCB for both boys
and girls; the correlations were .803 and .858, respectively.

2. On the whole, the Mathematics Proficiency Test and the Terman-
McNemar Test showed the highest correlations with grade point averages
in.various subject areas. The correlations ranged from .36h to .690.
The TermaoncNemar was, in general, the best predictor of the EHSCB
sub-tests. The correlations ranged from..67h to .818.

3. Although many differences proved to be non-significant, girls

were a more predictable group than boys in terms of the criteria and

predictors used.

h. Since verbal tests correlated more highly than "number" tests
with the EHSCB criteria, and the "number" tests correlated more highly
with grade point averages, it seems evident the differences in these two
types of criteria lie in the different aspects of "intelligence" measured
by them.

5. There were no statistically significant differences in the
multiple correlations derived from.the best combination of two predictors
and a combination of all eight predictors.

6. On the whole, the best predictor tests in one subject area
were found to be the best predictors in other subject areas. Therefore,
the existence of special abilities, which would.be needed for deter-
mining choice of subject area majors, did not seem to differentiate
subject area groups. ‘

7. The mean predictor test differences between those students
majoring in any one subject area and those not majoring in that area
were statistically significant in eightybfour out of eighty-eight cases.
These differences are likely accounted for'by the different combinations
of subject area majors which the more and less capable students tend to

elect.

AN EVALUATION OF CERTAIN MEASURES OF APTITUDE AND ACHIEVEMENT

IN THE PREDICTION OF SCHOLASTIC SUCCESS

 

A'Dissertation
Presented to
the Faculty of the School of Advanced Graduate Studies of
Michigan.State University of

Agriculture and Applied Sciences

 

In Partial Fulfillment
of the Requirements for the Degree

Doctor of Education

Department of Foundations of Education

 

by
James Najeeb Jacobs

June 1957

ACKNOMEDGMENT

This investigation could not have been accomplished without the
cooperation and assistance of a number of persons. Acknowledgment is
made to Mr. Robert P. Curry, Assistant Superintendent, Department of
Instruction, and to Miss Joan Bollenbacher, Supervisor of Appraisal
Services, Cincinnati Public Schools. Thanks are also expressed to the
staff of the Division of Appraisal Services whose help and encouragement
made this study possible.

Sincere thanks are also expressed to my doctoral committee,

Dr. Harry SundJaall, Dr. James Karslake, and Dr. Raymond Hatch. A very
special mention of appreciation must be made to my wife, whose patience,

skill and persistence, was instrumental in the completion of this study.

TABLE OF CONTENTS

CHAPTER

I} FORMULATION.AND DEFINITION OF THE PROBLEM . . . . . . . .
IntrOduCtion o o o o o o o o o o o o o o o o o o o o o 0

General purpose and statement of the problem

General purpose . . . . . . . . . . . .
Statement of the problem . . . . . . . .
The significance of the study . . .‘. . .
‘Scope and limitations of the study . . . .
Statement of objectives and hypotheses . .
Definitions of terms used

Organization of the study . . . . . . .
II. REVIEW OF THE LITERATURE . . . . . . . .
The prediction of general academic success

Non-intellectual factors . . ... . . . .

InteneCtual faCtOI‘S c o o o o o o o o o o o 0

Studies relating to prediction in subject areas

Summary 0 o o o o o a o o o o o o o o o o o o 0

III. THE CRITERIA AND SOURCES OF DATA . . . . . . . . .

The criteria 0 o o o o o o o o o o o o o o o o o o 0

Prediction in specific subjects versus prediction of

SUbjeCt-matter areas 0 o o o o o o o o o o o

The courses constituting a subject-matter area

Method of determining grade point averages . .

The comparability of criteria between schools

iii

PAGE

CD -4 fir 5? k0 k» F4 l4

12
13
1h
15
15
2h
31
149
52
52

52
Sh
S7
58

CHAPTER
Sources of data . . . . . . . . . . . . . . . .
Source of grade point averages . . . . . . .
Review of tests used . . . . . . . . . . . .
Iv. METHODSANDIROCEDURES
The types of tests used in this study . . . . .
Treatmentofrawdata ............

Statistical methods and.procedures . . . . . .

V. THE DIRECT AND MULTIPLE PREDICTION OF GENERAL HIGH SCHOOL

ACADEMICSUCCESS

The direct prediction of total grade point average and the

composite score of the Essential High.School Content

Battery 0 o o o o o o o o o o o o o o o o o o o .

The multiple prediction of total grade point average and

composite score of the EHSCB . . . . . . . . . . . . .

VI. THE PREDICTION OF ACADEMIC SUCCESS IN EIGHT HIGH SCHOOL

SUBJECT MAS O O O O O O O O O O O O O O O C O O O O O

the

The direct prediction of subject area grade point averages

and the Essential High School Content Battery criteria .

Multiple prediction of grade point averages and the

Essential High.School Content Battery criteria

The significance of difference between subject area majors

VII. SUMMARY, CONCLUSIONS, AND IMPLICATIONS FOR FURTHER RESEARCH .

smary . O O O O O O O O O O 0 O O O O O O O O O O O 0 O

COHClUSionS o o o o o o o o o o o o o o o o o o o o o o o 0

Implications for further researCh o o o o o o o o o o o o 0

iv

PAGE

611
65
6S
78
78
80

81

89

89

98

10h

10h

121
128
137
137
139
1h3

CHAPTER PAGE
BIBLIOGRAPHY . . . . . . . . . . . . . th
APPENDIX A. Doolittle solutions to the beta coefficients . . 150
APPENDIX B. Solution of regression coefficients . . . . 166
APPENDIX C. Nomographs for the prediction of grade point

averages. . . . . . . . . . . 175
APPENDIX D. Regression equations for the prediction of boys

and girls grade point averages . . . . . 191

TABLE
1.

2.

3.

h.
5.

7.

9.
10.

ll.

12.

13.

LIST OF TABLES

Summary of the Correlations between the Holzinger-Crowder
Uni-Factor Tests and Teachers' Marks in certain High
School Subject Areas . . . . . . . . . .

Correlations of Four Multiple Aptitude Tests with Teachers'
Marks in Five Subject Areas . . . . . . . .

DAT Subtests Providing the Highest Median Correlation with
Each of the Eight High School Subject Areas for Boys and
Girls 0 o O O o o o o o o o o o a

Subject Matter Areas Defined by Their Constituent Courses .

Total Grade Point Average and ACE Means for the Six
Cincinnati Public High Schools . . . . . . .

Results from an Analysis of Covariance of Total Grade Point
Averages.Adjusting these Means for Levels of Scholastic
Ability in Five Cincinnati High Schools . . . . .

Analysis of Covariance of Total and.Aubject.Area Grade
Point Averages Adjusting these Means for Levels of
Scholastic Ability in Three Cincinnati High Schools . .

Split-half and Alternate Form Reliability Coefficients, by
Separate Grade Levels . . . . . . . . . .

Types of Scores Used for’Each of the Predictor Tests . .

Correlations Between the Predictor Variables and Total
Grade Point Average, and the Significance of’Difference
of the Correlations Between Boys and.Girls. . . . .

Means, Standard.Deviations, and the Significance of Difference
Between Means for the Total Grade Point Average and the
Predictor Variables Between Boys and.Girls . . . .

Correlations Between the Predictor Variables and the Composite
Score of the Essential High School Content Battery Together
with the Significance of'Difference of the Correlations
Between Boys and.Girls . . . . . . . . .

Inter-Correlations Among the Predictor Variables Used in the
Multiple Regression Analysis of Total Grade Point Averages

PAGE

36

378

no
55

59

60

62

77
8O

91

93

96

99

.
.

llllll|l1|lll||

. o a A .
.,

Q

a

9

w

.

. u

\ _ I

. Q .

. ‘ Q. I

TABLE
11;.

15.

16.

17.

18.

19.

20.

22.

23.

2h.

25.

Solution of the Regression Coefficients for the Prediction
of Total Grade Point Averages of Boys and Girls . . .

Correlations Between the Predictor Variables and Grade
Point Averages in Each of the Subject-Matter Areas
and Sex 0 o o o o o o o o o o o o

z-Ratios Indicating the Significance of Difference Between
Correlations for Boys and Girls in Each Subject Area in
Which both sexes Have Majored . . . . . . .

A Summary of the Three Best Predictors of Grade Point
Averages in Each Subject Area and Each Sex . . . .

A Comparison of the Average DAT Validity Coefficients
Established in this Study and Those Derived from the
Data Presented in the DAT Manual, by Sex, and Subject
Area, for the DAT Predictors Used . . . . . .

Correlations Between the Predictor Variables and the Four
Subtestsrof the Essential High School Content Battery for
Boys and Girls 0 o o o o o o o o o o

A Comparison of the Average Correlations Between the
Predictors and the EHSCB Tests and Grade Point Averages
by Sex and Subject Area . . . . . . . . .

A Summary of the Three Best Predictors of the Essential High
School Content Battery for Boys and Girls . . . .

Correlations Between the EHSCB Criteria and Grade Point
Averages in Corresponding Subject Areas, by Sex . . .

A Summary of the Multiple Correlation Coefficients with the
Standard Errors, Using Eight Test Variables as Predictors
of Grade Point Averages in Each Subject Area and Sex .

The Multiple Correlations and Standard Errors of the Best
Two Predictor Combinations, the Total Predictor
Combination, and the Significance of Difference Between
These Two Multiple Correlations for Each Subject Area and

5 ex 0 o o o o o o o o o o o o a

Multiple Correlations and Standard Errors of the Two-
Predictor Combinations in Predicting Scores on the
Essential High School Content Battery for Boys and Girls

PAGE

100

106

108

111

113

115

116

118

120

1211

127

129

TABLE PAGE

26. The Mean Scores and "t" Ratios on the Predictor Tests

Made by Boys Majors and Non-Majors in Each of Five
SUbjeCt Areas 0 o o o o o o o o o o 131

27. The Mean Scores and "t" Ratios on the Predictor Tests

Made by Girl Majors and Non-Majors in Each of Six
SUbjeCt Areas 0 o o o o o o o o o 0 13h

viii

L.

CHAPTER I
FORMULATION AND DEFINITION OF THE PROBLEM
I. INTRODUCTION

In the Cincinnati Public Schools, group testing is the responsi-
bility of the Division of Appraisal Services. This Division gives over
one hundred thousand tests yearly in an attempt to evaluate pupils for
various purposes. Among the tests given each year as a part of the
regular testing program are tests of achievement and intelligence. Peri-
odically, however, other tests are given on an experimental basis to
determine their value for specific purposes. All intelligence tests
given by this Division are administered by trained examiners whose sole
function is test administration. Test score variability due to vari-
ations in test administration are, therefore, minimal.

One of the many reasons for administering a testing program is to
evaluate pupil performance at a given time in order to provide data that
will make it possible to predict performance in future situations. In
the Cincinnati Public Schools, curricular, administrative, and instruc-
tional decisions are often based on information of this kind. It is,
therefore, imerative that the instruments used actually do the job they
are intended to do. This aspect of a test is called validity; when the
test is called on to predict future performance, it must have predictive
validity.

The administrators of a testing program constantly face the
problem of selecting the particular tests which will provide best the

informtion they are seeking. Mam tests may possess a degree of
validity for specific purposes, but few, if any, possess validity
which can be applied in a variety of situations. The problem of the
test administrator is to determine which tests perform the prescribed
function best. This problem is encountered not only in the selection of
new tests, but also in terms of re-evaluating the testing instruments
which are in current use in the testing program. This re-evaluation or
confirmation of validity must be done periodically because often the
characteristics of a (test) population may alter significantly. Conse-
quently, a test that may have been valid for a past pepulation may be
rather ineffective in accomplishing its function on a different popu-
lation. It is not sufficient to make armchair speculations concerning
the predictive validity of tests; nor can one fully accept validity as
it is demonstrated in test manuals, since validity is highly specific to
the population on whom it was originally obtained. Test administrators
are conmitted, therefore, to eacperimentally determining the validity of
a test before incorporating it into the testing program itself.

During the school year 1952-53, the current senior class
(1956-57) in all six of Cincinnati's high schools was given the Terman-
McNemar Test of Mental Ability and. an English and Mathematics Pro-
ficiency Test as a regular part of the Division's testing program. In
addition, when in the eighth grade, this group also was given four
subtests of the Differential Aptitude Battery, 1.9. Verbal Reasoning,
Numerical Ability, Mechanical Reasoning, and Language Usage, for deter-
mining experimentally their value in the Division's testing. program.
These specific subtests were used because a review of other studies and

some past experience has shown these subtests to be of greatest value
for the purpose for which they were to be used, i.e. the prediction of
academic achievement. Using these Differential Aptitude Tests together
with the Terman-McNemar Test of Mental Ability, and the English and
Mathematics Proficiency Tests, an attempt was made to predict the
achievement of pupils at the high school level.

The junior high school pupil anticipating his entrance into a
senior high school has new decisions to make concerning his educational
future. With the growing conplexities in curricula, one of the funda-
mental decisions he must make is what subjects shall be his area of
concentration. It is the responsibility of the school to provide educa-
tional information to students in an attenpt to aid them in making
suitable educational plans at the high school level.

To define what is meant by "suitable educational plans ," is
indeed a difficult task. It is a problem with new facets, each
requiring its'own analysis and evaluation. Most of these facets, do,
necessarily, require subjective analysis and interpretation since they
deal with attitudes, values, home environment, personality factors, etc.
Among the most significant objective areas of evaluation, however, are

those derived from standardized tests.
II. GENERAL PURPOSE AND STATMT OF THE PROBLEM

General m. If the premise is accepted that one of the more
important aspects of suitable educational planning is the student's
attainment of scholastic success in whatever subject-matter area he
chooses (relative to his capabilities), it would be agreed that the

h
prediction of student performance on standardized tests would be helpful

in educational planning.

With this promise in mind, the general purpose of this study is
to evaluate certain aptitude and achievement tests. The resulting educa-
tional information then will be used to aid junior high school coun-
selors in helping students to select high school majors in subject areas
in which they are likely to achieve success.

Statement 9; the problem. The problem of this stucbr is to
evaluate singly, and in combination, the ability of seven measures of
aptitude and achievement to predict academic success in eight subject
areas at the high school level, i.e. English, social studies, science,
mathematics, foreign language, industrial arts, home economics, and
business education, in addition to predicting the general academic success

of high school students.
III. THE SIGNIFICANCE OF THE STUDY

The significance of this study rests primarily on an attempt to
overcome certain limitations of similar studies conducted previously and
to determine the predictive validity of the instruments used on a parti-
cular population in the Cincinnati Public Schools.

Although.many studies have been conducted in an attempt to
predict academic success at the high school level, most of these
studies are of limited value in situations outside the one in which the
study was conducted. It is difficult and often dangerous to generalize
regarding the results of any study of this type to other school systems

(and.even schools within a given system unless one can demonstrate the

comparability of the schools involved) because of the differences that
may exist in the characteristics of the papulation. Each school is
unique. There are varying eaphases in the curricula, different methods
of teaching, different personnel and course content, and, of course,
varying abilities of the student population itself. These differences
emphasize the lack of general validity of an instrument and the need
for gecific validity. In discussing the use of tests in guidance,
McDaniel states, "Test meanings are ultra centered, tenporal, and
situational. Test data reported in a manual apply, within limits, to
the time, place, and sample studied. "1

Marv earlier prediction studies have dealt with prediction in a
specific subject; thus, the degree of prediction and relatedness of the
criterion and predictor in that subject area could not be compared with
others using the same predictors. This stuck is designed to make such
comarisons possible.

Authors of may prediction studies have secured criterion
measures only for the period in which the predictors were used. The
resulting data collected for the purpose of prognosis may thus be con-
sidered concurrent validity and not predictive validity. In its
Technical Recommendations f9}; Achievement 2212;, the Committee on Test
Standards for the AERA and NCMUE states:

The former (concurrent validity) answers the question, "Mth what

degree of accuracy can the test scores replace the scores on an
existing criterion?” the latter (predictive validity) answers the

13.13. McDaniel, "The Use of Tests in Guidance," California
Guidance Newsletter, 3:3, November, 191:9.

question, "With what degree of accuracy can the test scores esti-

mate the scores on the griterion that the test subjects would

achieve sometime later?

\ In those studies in which predictive measures have been secured
prior to the criterion measures, the length of time intervening has
often been too short; thus, long range prediction cannot be inferred
safely, and the usefulness of such prediction is of limited value.
Furthermore, the number of cases in the sample used is often insuffi-
cient for a valid conclusion. In this study, the predictors were admin-
istered from three to three and one-fourth years before the criterion
measures were secured, and the total sample included 595 students.

Because of the changes occurring within am given school system,
i.e. changes in curriculum, physical facilities, teaching personnel,
student population, etc. , it is necessary to periodically re-check the
effectiveness of prediction in the given situation. As previously
stated, this is one of the purposes. for conducting this study.

In reviewing prediction studies of academic success, it was
noticed also that few studies report differential prediction between
the sexes, even though sex differences in looming show that boys and
girls acquire knowledge and skills selectively.3 If these differences
are ignored, it may lead to erroneous conclusions in interpretation.

Furthermore, the estimation of variance of test scores among high

 

ZConittee on Test Standards for the mm and NCMUE, Technical
Recomendations for Achievement Tests (Washington, D.C. : Nat ona
uca on ssocimfon, I555), p.17.

 

3Alexander 0. Wesman, "Separation of Sex Groups in Test
Reporting," Journal g_f_ Educational menolog, to: 223, April, 19h9.

school students shows sex differences, the girls being a more variable
group due to their generally lower dropout rate.h Combining groups,
therefore, would result in spurious measures of relationship.

Due to differences in grading practices among schools, the
comarability of schools should be demonstrated before combining
students from different schools into a single sample for study. When
this is not done, it is impossible to know whether the criterion
measures have the same significance or meaning. For example, it would
not be known whether an "A” in one school indicates the same standard
as an ”A“ from another school.

In this study, through the use of an analysis of covariance
technique, comparable schools were identified and selected. Finally,
one of the more significant aspects of this study is an attenpt at
multiple prediction of the criterion, using intelligence, achievement
and aptitude tests as combined predictors. In so doing, a relative
ranking of the predictors results. From this one can select the best
predictors and determine which combination has the relatively highest

predictive power.
IV. SCOPE AND LDETATIONS OF THE STUDY

The sasple studied was drawn from the twelfth grade classes of
three Cincinnati Public High Schools. As such, generalizations resulting
from this study must be limited to this student population.

 

“Clifford P. Archer, "Student Mortality," En clo edia of
Educational Research (New York: The Macmillan Comparw, I555) , 571158.

Since only those students for whom complete test records were
available were included in this study, another element of selectivity
is present. The degree to which this selectivity has affected the
representativeness of the population is not known.

Other limitations which may or may not prevail are those con-
cerned with the tenability of certain asmtions which are required
for certain statistical tests and analyses. For purposes of correlation
analyses, for example , assumptions regarding the linearity of regression
are made. In addition, the assumption of homoscedasticity is made.
Normal distributions of the traits being measured also are assured.

Other limitations are those associated with methodology and the
instruments used. Limitations lie in the selection of the samle, its
number, and its cross-sectional nature. Although the validity of the
tests used in this study is the problem under scrutixw, the reliability
of the instruments place limitations on their usefulness as predictors
of academic success.

One of the severest limitations of this study has to do with the
questionable validity of grade point averages and achievement test
scores as measures of scholastic success. For purposes of this study,
it is necessary to define scholastic success operationally in terms of
the criteria used.

V. MAW OF OBJECTIVE AND HIPOTHEES

Objectives. In an atteapt to pursue the general purpose and
problem of this study, the specific objectives represent an attemt to
answer the following questions:

1. Which of the aptitude and achievement measures in this study
correlates highest with general academic success as measured by total
grade point average and the composite score of the Essential High School
Content Battery?

2. Which of the aptitude and achievement tests correlates highest
with scholastic success in each of eight subject areas as measured by
grade point averages in these subject areas and the areas measured by
the Essential High School Content Battery?

3. Are there sex differences in terms of predicting the criteria
used as measures of scholastic success?

h. In terms of their relation to the aptitude and achievement
predictor variables, what are the differences between the grade point
average and the Essential High School Content Battery criteria?

5. Are there significant differences in predicting grade point
averages between the multiple correlations derived from a combination
of all predictor variables and those derived through a combination of
two of these predictors?

6. With a knowledge of the individual predictor test correlations
with the grade point average criteria, is it possible to isolate certain
abilities which are needed for success in a given subject area?

7..Are there significant differences in performance on the
aptitude and achievement predictors between those students majoring
in any one subject area and those who are not majoring in that subject
area?

In addition, and perhaps most important, it is an objective of

this study to aid counselors in using the educational information

10
obtained from this study. Too often, the findings of educational
research are merely a matter of academic interest and do not result in
action. Probably such information is used infrequently because studies
of this type generally result in correlation coefficients indicating
the degree to which prediction can be made. Conputing the correlation
coefficient may be a necessary step, but too often it has the peculiar
quality of being meaningless, misunderstood, or misused, even among
those who use tests frequently.

An objective of this study, therefore, is to aid junior high
school counselors, who may wish to utilize the information secured in
this stuck, in interpreting the phenomenon of regression through the
use of nomographs. These monographs will indicate the best prediction
of the criterion through its regression on the predictor variables, in
the form of an easily understood, graphical presentation.

ntheses. Assuming that academic achievement is due largely to
a generalized verbal factor, and assuming further that the Terman-
Hcllemar Intelligence Test is a valid measure of this verbal factor, it
is hypothesised that this test will be the best all-around predictor of
academic success. It is likely that in the process of averaging marks
from different subjects within a subject area, the existence of special
abilities would be obscured and in their place would merge an aspect
of the "general" intelligence factor.

This general factor, however, is likely to play a more inortant
role in the prediction of the Essential High School Content Battery
criteria, since the latter criteria are probably more dependent upon
verbal skills than are the characteristics upon which school marks are

based.

Since twelfth grade girls probably represent a more heterogeneous
population than do twelfth grade boys, it would.be expected that the
aptitude and achievement predictors would correlate higher with both
criteria for girls than they would for boys.

Due to the more highly verbal nature of the Essential High
School Content Battery criteria over the grade point average criteria
it would be expected that much of the difference between these two
criteria would be in terms of the degrees to which the verbally loaded
measures differentially correlate with them.

Boys would be expected to perform higher on the "number" tests,
the Mechanical Reasoning and the Terman-McNemar intelligence test,
while girls probably would exceed.boys on the Spelling, Sentences,
English.Proficiency, and Verbal Reasoning tests. This hypothesis is
based on the findings of numerous studies showing that on the whole, boys
exceed girls on quantitative measures while girls generally exceed.boys
on linguistic measures. Even though the TermanAMcNemar Test is verbally
oriented, it is nevertheless, a general aptitude test, and since it is
expected that the remaining twelfth grade boys represent a more homo-
geneous and academically select group than do girls it, therefore, would
be expected that they would perform higher on this test.

Finally, it is hypothesized that there will be significant
differences in mean performance on the aptitude and achievement
predictors when students majoring in a subject area are compared to those
not majoring in that subject area. This hypothesis results from the

notion that on the whole, the brighter students tend to enter the college

12
preparaton curriculum, i.e. mathematics, science, and foreign language,
while the less capable students tend to select other subject areas for

majors.
VI. DEFINITIONS OF TERIVB EB!)

Subject-matter area 2; subject area. A group of subjects or

courses which, by virtue of their content, are similar. Examples of
subject areas are science, mathematics, English, etc.

Sub act 23 m. These refer to the entities which make up
subject areas. Examples of subjects are biology, physics, algebra, etc.

m. Marks are teachers' ratings of a student's performance in
a subject. The terms "marks" and. ”grades" are used interchangeably.

M 33393 23:22 average. This represents an arithmetic'mean
of all marks given in full-year subjects during the student's four years
of high school. It does not include courses such as physical education
or health.

93393 M averag . This represents the arithmetic mean of three
or more marks given within a subject area in three or more full-year
subjects.

£21.23. A major consists of three or more subjects taken within
a subject area. The students in this study are required to comlete
three majors for high school graduation.

Criteria. In this study, the measures used as standards of
scholastic success are grade point averages and an achievement test,
the Essential High School Content Battery.

Predictor test (variables). These refer to the seven aptitude and

13
achievement tests administered to the pupils in eighth grade for
purposes of predicting academic success in high school.

Scholastic 23 academic success. This expression is defined
operationally in terms of the degree to which a student attains high
scores on the two criteria used in this study.

VII. ORGANIZATION OF'THE STUDY

In Chapter II an attempt is made to review pertinent studies
relating to the prediction of general scholastic success through the
use of ”non-intellectual" and "intellectual" variables. Secondly,
Chapter II contains a review of studies relating to the prediction of
scholastic success in particular subject areas using instruments such
as aptitude batteries, general scholastic aptitude tests, and achievement
tests as predictors of success. Chapter III provides an explanation of
the criteria and sources of data including a review of tests used.
The methods and procedures used in this study will be discussed in
Chapter IV. Chapters V'and‘Vijresent the findings and analysis of the
prediction of general academic success and subject area academic success,
respectively. Chapter VII contains the summary, conclusions, and impli-

cations for further research.

CHAPTER II
REVIEW OF THE LITERATURE

In reviewing rhe literature pertaining to the prediction of high
school academic success, one is impressed with the tremendous amount of
effort put forth in its prediction, at least in terms of the number of
investigations carried out. The greatest bulk of these studies seem to
have been carried out between 1920 and.l9h0 with a drop off in the past
fifteen years. Perhaps the incidence of prediction studies dealing with
academic success is on the upswing with the relatively recent appear-
ance of well constructed multiple aptitude test batteries.

Practically all the prediction studies reviewed used the Pearson
product-moment correlation coefficient as the index of relationship
between the predictor used and the criterion. As for the criteria used,
the school grade and the results of standardized achievement tests
stand far ahead in use. Indeed, one is hard pressed to find other
criteria as easily available and more generally accepted as these.

The literature contains studies which run the gamut of subject-
matter performance to be predicted. However, one topic which has re-
ceived much attention is the prediction of general academic success.
Two approaches may be identified; one approach dealing with
"intellectual" factors as predictors, and another dealing with "non-
intellectual" factors as predictors. These two approaches, in the
prediction of general academic success, will be dealt with in the first
section of this chapter.

The second section of this chapter will attempt to review those

15
investigations dealing with the prediction of success in specific
subject areas. In the latter section, an attempt will be made to review
studies which have utilized achievement tests and general and special
aptitude tests including some of the well known batteries. This

section will be followed by a summary of Chapter II.
I. THE HEDICTION OF GENERAL ACADEMIC SUCCESS

Log-intellectual factors. It is a well known fact that no matter
how reliable and valid a general intelligence test may be, it does not
account for all the variation in academic achievement. In an effort to
account for this variation, investigators have naturally turned their
attentions to the so called 'non-intellectual"factors of personality.

In isolating some of the personality factors which relate to
achievement, meal used a thirteen iten personality rating scale. The
subjects were 230 students constituting a graduating class in a mid-
western high school. The procedure was to have each student evaluated
on the personality scale by each teacher having the student in class at
the time of the study. The personality factors purportedly measured by
the rating scale were 3 sociability, attractiveness, nervousness, popu-
larity, punctuality, courtesy, cooperation, persistence, honesty, connon
sense, sincerity, dependability, and general attitude toward school. In
addition, the Otis Intelligence scores were available having been admin-
istered to the subjects during their 9-3 grade level. The criterion for

 

l'Viola Ames, ”Factors Related to High School Achievement,"
Journal 2; Educational momma, 3h:229-236, April, 191:3.

16
scholastic success was the four year high school average of all grades
received. Direct correlations with the criterion showed sociability,
attractiveness, and popularity to show no significant relationship.
Persistence, comon sense, and dependability correlated .60, .52, and
.57, respectively, with the criterion. Other correlations with the
criterion were as follows: punctuality, .h7; cooperation, .15; honesty,
.hl; sincerity, .h9; nervousness, .28; courtesy, .32; and general ati-
tude toward school, .57. None of these thirteen traits correlated highly
with the Otis. The multiple correlation derived using a combination
of these traits with the Otis was .72. The correlation between the Otis
and the average grade criterion was .534. By using the factor analysis
technique, it was found the total of fifteen variables measured two
factors; ability to succeed socially and ability to conform to school
situations. It was concluded also that social success was not related
to scholastic achievement, but that the ability to conform to school
situations was related to scholastic achievement.

This study certainly demonstrated the imortance of this type of
approach, however, it is questionable whether the latter study could be
described as predictive in nature. The teachers who rated the subjects
did so when the. subjects were in their twelfth grade and in the teachers'
classes. It is very likely, therefore, that the ratings were influenced-
by the past achievement of the subjects. The study may describe, however,
the traits upon which teachers base the grades they give.

The use of non-intellectual factors as predictive measures of
success would naturally lead investigators to the use of standardized
personality tests as well as other instruments. One such study was

17
carried out by Cough.2

Gough selected two criterion samples of twenty-seven each from
a total group of 231 high school seniors. Pairs were matched on the
basis of Otis I.Q. and sex and then split according to their three year
high school honor point ratio (HPR). The two grows were then compared
in performance on the Minnesota Multaphasic Personality Inventory, the
Security-Insecurity Inventory, the Otis and Pintner intelligence tests,
and socio-economic status as measured by the Sims Score Cards. The
resulting analyses revealed that none of these variables showed a
statistically significant difference between the two grows.

Through an item analysis of the Minnesota Multaphasic Person-
ality Inventory, a grow of items found to differentiate achievers
(called the Ac Scale) was found to correlate .h3 with three year HPR.
The correlation between the Otis Intelligence Test and HPR was .62.
Other variables were also included in an attempt to predict HPR. Cor-
relations of the various scales of the MMPI ranged from .35 to -.21.
The Sims Score Cards showed a correlation of .25 with HR. The Cooper-
ative English Test correlated .72 with HPR. The highest degree of
multiple correlation using non-intellectual factors was .514. This
consisted of a combination of the Ac Scale and the St (Status) Scale
of the WI.

Another stucv which attewted to relate high school achievement

with personality variables as named by a standardized test is

 

2Harrison G. Cough, "Factors Relating to the Academic Achievement
of High School Students ," Journal of Educational Echelofl, 140265-78,
February, 19119.

18

reported by Hinkelman.3

The criterion used for high school success was the Hydra-Rush
High.School Progress Test for English, social studies, mathematics,
science, and total achievement ratings. The instrument used for eval-
uating certain personality characteristics was the Johnson Temperment
Analysis. The data from.this study were derived from.thirty recently
graduated high school students. The personality traits purportedly
measured.by this instrument are nine bi-polar traits: nervous-composed,
depressive-gay hearted, active-quiet, cordial-cold, sympathetic-hard.
'boiled, subjective-objective, aggressive-submissive, critical-appreciative,
and self-master-impulsive.

The results of this study showed that three traits seemed to
have the strongest relation to achievement, i.e. ”objective", "cowosed",
and "selfemastery". Three other traits, "appreciative", "submissive",
and "active", also yielded statistically significant coefficients.
Hinkelman concludes, "Recognition of the relatinn of personality factors
to school achievement can make a vital contribution toward more accurate
prediction of success."h

Other personality tests which have been used as predictors of
success in high school are the Berneuter Personality Inventory and the
Bell Adjustment Inventory. A study involving the Berneuter Personality

 

3Emmet Arthur Hinkelman, "Relation of Certain Personalit

variables to High school Achievement,” School Review, 60532-53 ,
December, 1952.

bag. 1:. 53h.

19
Inventory as it relates to the prediction of academic success is
reported by Hemzek.5 With data on 92 and 99 sophomore boys and girls,
Hemzek investigated the direct and differential predictive possibil-
ities for the N, S, and D scores of the Bernreuter with honor point
averages in several academic areas. The resulting direct and differ-
ential coefficients of correlation were so low as to be negligible. The
results of this study confirmed the conclusion that Finch and Nezmaek6
came to with a similar stucbv of the Bernreuter for the prediction of
total scholastic achievement in high school. The authors concluded,
"The data at hand furnished no evidence that the Bernreuter inventory
is measuring am traits that contribute in am inportant degree to

successful achievement in high school.»7
Super8 has summarized the research done on the Bernreuter and,

in general, also found the trend for relationships between grades and
Bernreuter scores to be negligible. Super9 also reports that studies
with the Bell Adjustment Inventory toward the prediction of grades

 

sClaude L. Hemzek, “The Value of the Bernreuter Personality
Inventory for Direct and Differential Prediction of Academic Success
as Measured by Teachers' Marks ," Journal 25'. Applied Echeloa,
22:576-586, December, 1938.

6F.H. Finch and Claude L. Nemzek, ”The Relationship of the
Bernreuter Personality Inventory to Scholastic Achievement and
Intelligence," School 23d Society, 365914-596, November, 1932.

71hid., p.596.

8Donald E. Swer, "The Bernreuter Personality Inventory: A
Review of Research," szchelogical Bulletin, 39 :9h-125 , March, 191:2.

9Donald a. Super, #22? Vocational Fitness (New York:
Harper and Brothers, 19h8 , p. .

20
have been negative.

Although it seems apparent that personality tests of various
types do have some value in terms of their relationships with school
success, the greatest weakness seems to be in the lack of sufficiently
high reliability of the instruments themselves. Furthermore, most of
the personality instruments on the market today have not been standard-
ized with a view toward use in a typical educational setting. Marv
such inventories have been partially or wholly standardized and vali-
dated on special, atypical groups such as clinical cases.

There are mam non-intellectual variables which could be related
with achievement of academic success outside of those comonly measured
by personality inventories. Three investigations of this type will now
be reviewed.

The first is an investigation carried out by Curtis and
Hensek.1° In this study, the authors investigated the relationships
between certain unsettled hone conditions and academic success. Six
types of broken home conditions were studied; loss of father by death,
loss of father by divorce or separation, unwloynent of father, loss
of mother by death, loss of mother by divorce or separation, and
employment of mother outside the home. From the school records, fifty
pupils were singled out for investigation in each of the six hone
conditions mentioned above. This group of 300 pupils was matched with
300 other pupils from normal homes on the basis of age, grade in school,

 

10Erta Agnes Curtis and Claude L. Nanak, ”The Relation of
Certain Unsettled Home Conditions to the Academic Success of High
School Pails,” Journal '23.: Social Echelon, 9:hl9-h35, November, 1938.

21
sex and.nationality. Honor point averages based on teachers' marks were
used as the criterion of academic success. The data from this study
showed that the achievement of pupils from.broken homes was signifi-
cantly inferior to that of pupils from normal homes. A further study of
other factors such as amount of absence and tardiness, number of
sisters and brothers, language spoken in the home, amount of outside
employment, etc., failed to indicate any relation with the differential
achievement observed.

A second study attempted to reveal the relationships between
social-class and sex differences with high school achievement. Heimann
and.Schenkll selected a group of 11h sophomore students at random from
the Wisconsin Counseling Study. This group was categorized into two
social classes, Class III and Class IV, according to Hollingshead's
classification (Class III being the higher social class). School
achievement was indicated by the four'year high school average for each
student. Although students from four schools were involved, the authors
used a normalizing and standardizing procedure on marks for each subject
to account for school differences in marking. This is unique and an
apparently sound approach to the problem. Through a study of the mean
differences, it was found the average achievement of the higher social
class was significantly greater than that of the lower social
class. It was also found that girls (from both classes combined)

achieved significantly higher than boys (from both classes combined).

 

11Robert A. Heimann and Quentin F. Schenk, "Relations of Social
Class and.Sex‘Differences in High.School Achievement," School Review,
62:213-221, April, 1951.. ' ""'"" "'"""

22
Among their conclusions the authors state, "Clinical evidences of
differences in individual performance warn of the danger of over-
generalizatinn of group data in relation to social class and sex dif-
ferences in achievement."12

Studying bays and girls separately, Nemmele attempted to deter-
mine the value of certain non-intellectual factors for the direct and
differential prediction of academic success. He studied the predictive
values of chronological age, the amount of education of the father and
mother, and the occupational status of the father. The criteria uses
for academic success were honor point averages in mathematics, English,
languages, and art and vocational courses. Nemzek concluded that the
aforementioned factors were of negligible value for purposes of pre-
diction of academic success as measured by honor point average.

.Among the numerous and varied factors that make for academic
success, motivation is generally considered to be highly important. Since
interest is a function of motivation, it is to be eXpected that the
literature would contain many studies relating interest to achievement.
One of the perturbing factors which influences the expected high degree
of relationship between interests and achievement is that frequently
students of high ability can achieve well in subjects even though they
may lack intrinsic value in the subject. Such is particularly the case

for college bound students who find it necessary to take certain subjects

 

lzIbide, pa 220s

l3Claude Nemzek, "Value of Certain Factors for Direct and Differ-

ential Prediction of Academic Success," Jogggal‘gf'Social P cholo ,
12:21-30, August, 19h0.

23
to qualify for college admission even though they lack interest in the
subject. With this type of individual in the typical cross-sectional
analysis, the correlations found between interest and achievement are
naturally low. Such has been the case in most studies attempting to
relate interest with achievement. One interesting approach to the solu-
tion of this difficulty was contrived by Thorndike.1h Later this same
approach was adapted for use by Prandsen and Sessions.15 The approach
sisply eliminated the cross-sectional disturbances by finding the
correlation between interest and achievement on an individual basis.
Thus each subject ranks his interests in order and this is correlated
with the ranked achievements in various subjects. Using this procedure,
Frandsen and Sessions related the nine interest scales of the Kuder
Preference Record to the achievement of 187 high school seniors in
subjects which seemed to match the Kuder interest categories. They
found a median intra-individual correlation of .27 batman patterns of
Kuder interests and achievement. The median correlations between self-
rated interests and rank order of school achievement was found to be .51.

Townsendl6 studied the relationships between Strong's scales
and scores on objective tests of school achievement made by 50 to 100

 

11m. Thorndike, "Interests and Abilities," Journal 9; Applied
Pacholog, 28:13-52, April, 192m.

15Arden N. Frandsen and Alwyn D. Sessions, "Interest and School
Achievement ," Educational and Echelogical Measurement, 13:9h-101,

Spring 9 1953 e

15A. Townsend, "Achievement and Interest Ratings for Ind endent
School Boys ," Educational Record Bulletin, 143:1;9-5’4, January, 19 .

2h

students in private schools. The only significant relationships found
were those between mathematics and science teacher-chemistry (r-.36),
accountant-chemistry (r-.31), CPA-chemistry (r-.h2), and mathematician-
geometry (r-.31).

In reviewing the studies relating interests and achievement,
one is impressed with the wide range of results and the apparent dis-
crepancies reported even in similar subject~matter fields. This is, at
least in part, due to the varying interest patterns of students from|
one school to another and/or the narrow range of interests exhibited by
many high school students. Certainly the lack of reliable criteria is
frequently the cause for low relationship. It is unfortunate that pro-
portionately more time and concern is not spent in securing reliable
and valid criteria.

Since the present study is concerned with the prediction of
academic success through the use of tests measuring "intellectual"

factors, our attention will now be turned to this topic.

Intellectual factors in the prediction 9}; g‘ eneral academic
suggess. Of the numerous studies carried out in the prediction of
general academic success, the use of intelligence tests as predictors
far surpasses the use of any other single predictor. Studies of this
type, however, were much more frequent in latter decades than they are
now.

The great bulk of prediction studies of general academic success

 

léIbide , p eh9-She

25
have been carried out on the college level. Those relating to predicticl

at the high school level are very frequently found as part of a study
generally dealing with prediction in specific subject areas. The latter
topic will be more fully eiqilored in the following section.

To demonstrate the comparability of different mental tests to
the problem of predicting academic success in high school, Jordan 17
applied four mental tests to the same group of children. He found the
following correlations with average grades in all subjects:

Otis Intelligence .1450
Miller Mental Ability . h76
Terman Group .1192

Apparently the results derived from various mental tests do not differ
significantly.

The lack of high correlation between the typical group test of
intelligence and success in high school could be attributed to the
relative lack of stability of these scores over those derived from
individually administered intelligence tests. A comparison of the results
of these two types of tests can be made in an investigation carried out

18 His results showed a correlation of J48? between the

by mmre
Stanford-Binet and average marks given over a two and one-half year
period of time, and .586 with school marks averaged over a one year

period. The respective correlations derived from the Arm Alpha Group

 

17m. Jordan, "Correlations of Four Intelligence Tests With
Grades ," Journal 2; Educational Echolog, 13:1419-h29, October, 1922.

18William Proctor, "Psychological Tests as Means of Measuring
the Probable Success of High School Pupils ," Journal of Educational
Research, 1:258-270, April, 1920.

26
Examination were .h13 and .3143. Apparently, the individually administered
intelligence test proved to be a slightly better predictor of grades
than did the group test.

Although the major purpose of the present study is to predict
success in high school, the ultimate purpose is to be useful in helping
students to choose appropriate high school majors. A study dealing with
the selection of subject fields as they relate to intelligence was
carried out by Powers.” In this study, Powers divided the students into
quartiles on the basis of the Otis Intelligence Test. He found the
highest quartile students tended to select, for the most part, advanced
mathmtics com-see, and to follow consequtively the following subject
fields in decreasing order: Latin, science, modern language, manual
training and mechanical drawing, history, commercial subjects, and
domestic art. The first quartile students tended to select those subject
fields in an met reverse frequency. The author concludes, "Students
possessing superior intelligence are attracted to those subjects which
make larger demands on intellectual capacity and lesser demands on
manual dexterity. "20

It would seem quite dangerous to generalize these results from
school to school because of the varying values placed on different
curricula. It is conceivable, for instance, for certain schools to

attract the brighter youngsters in the manual arts if this type of skill

 

l98.12. Powers, "Intelligence as a Factor in the Selection of High
School Subjects," School Review, 30:55, June, 1922.

2°Ibid.

27
is highly valued in the conmunity.

Pintner21 in summarizing the relationships between intelligence
test scores and high school marks found that the coefficients ranged
from .28 to .60 most of them being greater than .140.

In another sumnary by Ross and Hooks22 the range of correlations,
derived from a study of thirteen different mental tests by twelve
different authors, was between .12 and .69 with a median of .118.

Other studies tend to confirm these general results. Nemzek23
found, for example, correlations between intelligence tests and high
school scholarship to range from .hOl to .502 for boys and from .1495
to .606 for girls.

Embreezh undertook a study in an attenpt to determine whether
the predictive efficiency of certain measm'es differed with various
levels of intelligence. His subjects were 271 high school graduates,
each of whom had complete records from the eighth to twolfth grades.
High school achievement was measured by the students' boner point ratio
in the tenth, eleventh, and twelfth grades in senior-high school. Three
independent variables were used as predictors. These were ninth grade

 

21Rudolf Pintner, Intell once Testing Methods 29.2 Results
(New York: Henry Holt and onpany, , p. .

220.0. Ross and um. Hooks, "How shall We Predict High School
Achievement ,” Journal 2;: Educational Research, 22:18h-19S, October, 1930.

23Claude L. Nemzek, ”The Value of Certain Factors for the Direct
and Differential Prediction of Academic Success,” Journal 2;
merimental Education, 7:199-202, March, 1939. a

21*Royal B. Embree, Jr., "Prediction of Senior High school
Success at Various Levels of Intelligence ," Journal 2; Educational

«‘rf

28
honor point ratios, a measure of intelligence, and age at entering
the ninth grade. The subjects were divided into three LQ. groupings;
90-109, 110-129, and 130 and above. Zero order correlations between
each of the predictors and honor point ratios were cowuted for each of
the three grows. It was found that no significant differences existed
amng the three groups in terms of their predictability with the inde-
pendent variables used. By combining all three grows, the following
relationships were uncovered. A correlation of .853 existed between
ninth grade honor point ratio and the criterion. The 1.0. variable
(based on the median of five standard intelligence tests) correlated to
the extent of .596 with the criterion while age at entering high school
correlated «21:1; with the criterion. By combining the three independent
variables, a multiple correlation of .893 was established with the
criterion. The relationship between ninth grade honor point ratios and
the average senior high school honor point ratio, partialing out the
effects of I.Q. , produced a correlation of .823, indicating the inde-
pendence of this relationship with I.Q. The relative unimportance of
the age factor is evidenced by the fact that the multiple correlation
with the criterion using only the 1.0. and the ninth grade honor point
ratio was .891.

The degree of relationship shown in this study is extremely high
considering, especially, the fact that the subjects represent a rather
restricted grow even when the three LQ. groupings were combined. The
mean 1.0. of the total grows was 119.57 with a standard deviation of
12.38.

A notable aspect of this study is the fact that it represents

29
true prediction since the predictor measures were made before the
criterion measures were secured. The extremely high correlation between
ninth grade HPR and the average senior high school HPR is consistent
with other findings which have shown that past achievement in school
is one of the best predictors of future achievement.

Kelly,25 for instance, correlated the marks given in grades four,
five, six, and seven with marks given in the first year of high school.
Beginning with the fourth grade and continuing through the seventh
grade, he found the following correlations: .62h, .531, .728, .719.

The eighth grade arithmetic average, eighth grade English
average, ninth grade foreign language average and the 1.0. were used as
predictors of the total high school average in a stucnr by Dodes.26 Two
grows were studied; one grow coming from a junior high school and one
coming from an eight year elementary school. The correlation reported
for the elementary grow between the criterion and 1.0. was .115, while
the English average correlated .18 and the arithmetic average .36. For
the junior high school grow, the correlations with the criterion were
as follows; 1.0. .37, English average .50, arithmetic average .50, and
language average .62. Using the best two predictors, LC). and language
average, a multiple correlation of .77 was attained.

A correlation of .71 between average elementary school marks and

 

25131.. Kelly, Educational Guidance (New York Teachers' College
Contributions to Education, No. 71, New York, 19114), Poll6.

261.11. Dodes, "Prediction of High School Success," h Points,
31:5-elh, November, 19149.

30
and average high school marks was reported by Miles.27Another investi-
gator, Ross ,28 reported a correlation of .60 between average elementary
school marks and first year high school average. These investigations
substantiate the high predictive power of past achievement to future
achievement.

A study using a variety of predictors of high school academic
success was carried out by Tozer.29 In this» stuw, Tozer studied 132
students in grades nine through twelve in an attempt to predict the
average of all high school grades. The instruments used were 1) the
Terman Grow Test of Mental Ability, 2) the Cross English Test, 3) the
Sims Score Card for Socio~economic Status, and h) the New York Rating
Scale for School Habits. Correlation coefficients relating each of
these variables with grade point average were cowuted as well as the
multiple regression equation. The correlations Tozer found with seventy-
six sophomores and freshmen were: .75 with the Terman, .63 with the
Cross mglish Test, .09 with Socio-economic Status, and .81 with
ratings of school habits.

With fifty-six senior and Junior students in the same high

 

2714.12. Miles, A Cog arisen of mementagz and High School Grades

(University of Iowa, Studies in SdE'c'ation, Vol. 1, No. 1, Iowa City,
Iowa, 1911), p.22.

283.0. Ross, The Relation Between Grade School Record and Hi h
School Achievement (NE:- Teachers' College, ContHFEFIons 5' Education,
mm, ew or , 1925), p.70. -

299.13. Tozer, “A Statistical Prediction of High School Success
for see of Educational Guidance ," Journal 2f. Educational Research,

31
school, the correlations with the variables in the same order were .65 ,
.70, .22, and .88 respectively. These results tend to indicate the value
of intelligence and the importance of good school habits to school
success. For the grow of ninth and tenth graders, the multiple correl-
ation using a combination of all the variables was .881, while the
multiple correlation for the eleventh and twelfth graders was found to
be .91. Tozer concluded that the results of the study tended to show
that if a counselor had an accurate rating for school habits as well as
the rating of the intelligence level of a student, the counselor would
be materially aided in his guidance insofar as advising the individual
to take certain work in the regular academic curriculum.

The high correlation derived from the rating of school habits is
less impressive when one realizes the ratings were made by the very
teachers who were the source of the criterion measures (grades). It
would have been preferable to obtain an outside measure of this
variable. Furthermore, this study cannot be considered as predictive in
nature since the variables were obtained at the same time as the
criterion measure. Thus, validity is of the concurrent type rather than

the predictive type.
II. STUDIES RELATING TO PREDICTION IN SUBJECT AREAS

In recent years, the instruments given the most attention in the1
prediction studies of high school success are the multiple aptitude
batteries. Several of these studies will be reviewed now in an attempt
to cover some of the better known batteries as they relate to academicJ

success in subject areas.

32

One battery of aptitude tests which has received much attention
in the field of education is the Primary Mental Abilities (PMA)
battery. One of the basic differences between this battery and the
Differential Aptitude Battery (four subtests of which are used in this
stuck) is that, although the two are based on factor analytic proce-
dures, the DAT was constructed with a view toward educational use (thus
the Spelling and Sentences subtests) while the Primary Mental Abilities
Tests are oriented more toward factorial purity.

For purposes of stucb'ing the relationships between the Primary
Mental Abilities Tests and achievement in various fields, Shaw30 admin-
istered the mm to a group of 591 ninth grade students in two schools
in Iowa. The PMA Tests, consisting of the following subtests: Verbal-
Meaning, Hard-Fluency, Reasoning, Memory, Number, and Space, were
correlated with the following measures of achievement; the Iowa Tests
of Educational Development, the Cooperative Reading Test, Reading
Couprehension and an experimental reading test. Thirteen measures of
achievement were thus derived. The ranges of zero order coefficients
obtained for each subtest of the PMA are as follows: Verbal-Meaning,
.1401: to .793; Hard-Fluency, .161 to .1119; Reasoning, .197 to .562;
Memory, .116 to .287; Number, .090 to .hBh; and Space, .061 to .389.

Using the composite score of the Iowa Tests as a measure of
general academic success, the Verbal-Meaning Test correlated .793,

 

3°Duane C. Shay, "A Study of the Relationships Between the
Thurstone Primary Mental Abilities and High School Achievement ,”
Journal 9; Educational ggcmloa, ho:239-2h9, April, 191:9.

33
while the Reasoning Test correlated .501; Number, .359 3 Word-Fluency,
.355; Space, .350; and Manory, .238.

The highest relationship is seen to exist between the Verbal-
Meaning Test and the criterion. The multiple correlation of the PMA to
the Iowa composite score criterion was .8211. The difference between
this correlation and the Verbal-Meaning Test alone (.7910 is not
appreciable.

To say that the PMA Tests are factorially pure is obviously a
matter of degree. Although the inter-correlations of the subtests were
not presented in the aforementioned study, it is clear they must be high
because of the small increase in multiple correlation over the correl-
ation derived from the best single test and the criterion. It will be
noted that in no instance was the term "prediction” used either in the
study itself, or the writer's reporting of this article. The FHA and
the achievement measures were given concurrently; thus no true pre-
diction was made. Rather, the relationships reported siwly reflect the
degree to which the PMA could be used as substitutes for the achieve-
ment measures. It is unfortunate that the PMA Tests have not been as
well validated as some of the other batteries of aptitude tests.

One battery of uni-factor tests which was intensively investi-
gated in a stuw by Mitchell31 is the Holzinger-Crowder Uni-Factor
Tests. The problem of the study was to determine the extent to which

 

3lBlythe C. Mitchell, "The Relation of High School Achievement
to the Abilities Measured by the Holzinger-Crowder Uni-Factor Tests, "
Educational and. Echelogical Measurement, l5:h87-90, Fall, 1955.

3h
this battery would serve as predictors of high school achievement. The

battery furnishes separate measures of Verbal, Spatial, Numerical and
Reasoning ability.

The criterion measures were both achievement test results and
teachers' marks. The grows studied represented students from fourteen
different comunities, although each set of validity coefficients was
based upon the results for a single grade in one community. On the
whole, the correlations reported in Mitchell's study represent con-
current validities, since in most cases both-predictor and criterion
measures were secured within several weeks of one another.

Eight achievement tests were used as criteria for the prediction
of grades in mathematics. The range of correlations established for the
Verbal Test in predicting mathematics achievement was between .104 and
- .6h; the Spatial Test correlations ranged from .28 to .h8; the Numer-
ical Test from .31; to .76 and the Reasoning Test between .141: and .67.

In the area of science, five standardized tests were used as
criteria. The range of correlations for each of the predictors is as
follows; Verbal, .60 to .75; Spatial, .17 to .39; Nmnerical, .30 to .117;
and Reasoning, .146 to .61.

In social studies, four achievement tests were used as criteria.
The range of correlations for the Verbal Test was .58 to .65; Spatial,
' .18 to .37: Nmnerical, .26 to .39; and Reasoning, .hl to .h6.

Language arts was measured through the use of six different
achievement tests. The range of correlations for the Verbal Test was
.51 to .80; Spatial, .16 to .39; Numerical, .21 to .58; and Reasoning,
.h3 to .72.

35

USing the median score of the Essential High School Content
Battery as a measure of the total achievement, the Verbal Test correl-
ated .77; Spatial, .31; Numerical, .50; and.Reasoning, .58.

USing the teachers' marks as the criterion of school achievement,
the correlations between the predictor tests and twenty-seven subjects
were cowuted. For purposes of sumnarization, the subjects that were
clearly defined in.a subject area were combined. The ranges of correl-
ations together with the respective medians are presented in Table 1.

An inspection of Table 1 shows that on the whole, the prediction
of achievement in the tool subject areas is higher than in other areas.
It will be noted also that on the whole, prediction of achievement test
results is considerably higher than the prediction of teachers' grades.
The Verbal Test seems to be one of the best all around.predictors of
school achievement, and.probably represents a substantial portion of
what is commonly called general intelligence.

In addition to the zero order correlations, Mitchell also
reported multiple correlations using the combined four predictor tests
in predicting scores on various standardized tests. Since the Essential
High School Content Battery is used in the present study, the multiple
correlations he reports with each.of its subtests are of interest. They

are as follows:

§§§§§' Ehltiplegfi
Mathematics .723
English .780
Science ’ .771

Social Studies .620

TABLE].

36

SW OF THE CWTIONS BETWEEN THE HOLZINGER-CROWDER UNI-FACTOR
TESTS AND TEACHES' MARKS IN CERTAIN HIGH SCHOOL SUBJECT AREAS32

 

 

 

E

Subject Area Median N Range of r's Median r Best Predictor
English 361 .16 to .58 .16 Verbal

History 353 .33 to .60 .16 Verbal
Mathematics 96 .15 to .55 .112 Bees. and Verb.
Science 1811 .31: to .57 .119 Verbal
Home‘Economics 38 .10 to .56 .36 Reasoning
Business Education 78 -.01 to .61 .33 Nulnerical
Industrial Arts h8 -.06 to .85 .13 Spatial

 

321cm” p. 89.

37
It will be noted the highest prediction is English, followed by

science, mathematics and social studies. It will be of interest to come
pare these results with those obtained in the present study.

Segel33 attempted to determine the validity of an aptitude
battery by administering the Multiple Aptitude Test to a representative
school population for purposes of secondary school guidance work. This
battery included measures of Wbrd Fluency, Language Fluency, Mathe-
matical Reasoning, Spatial.Re1ationships and Mechanical Reasoning. The
battery was adapted from aptitude tests in the War Department.

One of the methods used for investigating the validity of this
battery was an approach similar to the one used in this study, i.e. to
determine its power to predict success in high school subject areas.
Although the tests comprising the battery do not exactly coincide with
the ones used in this study, there is enough similarity for comparative
purposes. Four tests were thus singled out for comparative purposes.A
partial reproduction of the table Segel reported, showing the correl-
ation of the four tests with grades in five subject areas, is shown in
Table 2.

It is clear through an inspection of Table 2 that most of the
correlations probably do not represent relationships greater than
chance expectation.

In the subject areas in which the respective tests seem to have

 

33navid Segel, "The Validity of a Multiple Aptitude Test at the
Secondary Level," Educational gaginychological Measurement, 7:695—705,

37a

TABIE 2

CORREIATIONS OF FOUR MULTIPLE API‘ITUDE TESTS MTR
TEACHERS' MARKS IN FIVE SUBJECT AREAS3

 

 

 

Subject Area N Mechanical Word Language Mathematical
Aptitude Fluency Usage Rea soning
Industrial Arts 87 J48 .09 .12 . 26
Foreign Language 78 -.Oh . 20 .51. .32
Social Studies 112 .25 .30 .10 . 23
English 120 .02 .149 .32 .11
Mathematics 10h . 20 .19 . 30 .6 2

 

31*Ibid., p. 703.

38
the greatest face validity, however, predictions are fairly high.
Industrial arts correlated .148 with the Mechanical Aptitude lest; foreign
language correlated .51. with the Language Usage Test; English correlated
.119 with the kbrd Fluency Test; and mathematics correlated .62 with the
Mathematics Reasoning Test.

In his conclusion, Segel stated that this study supports the
Impothesis that a multiple aptitude test of this type is of value for
differential diagnosis and prognostic work.

Milking35 used the Primary Mental Abilities and the Differential j
Aptitude Tests in predicting success in high school subject areas. For
comparative purposes, he selected three tests from each battery which
measured similar abilities, i.e. the verbal, number, and spatial tests.
Computing separate correlations for 139 girls and 128‘boys between the
three tests and teachers' grades, the following general results were
obtained. '

1. For both boys and girls, the DAT number test correlated highest
with grades in English (.55 and .58 for boys and girls, respectively),
and grades in science (.69 for both boys and girls). Mathematics grades
were best predicted by the DAT verbal test for boys (.66) and the DAT
number test for girls (.67). Marks in home economics and industrial
arts showed no significant relationship to any of the tests.

2. The reported correlations were generally higher for girls than

they were for boys; however, regardless of sex, all the tests proved to

 

35 William D. Wolking, "Predicting Academic Achievement with the
Differential Aptitude and the Primary Mental Abilities Tests,” Journal

25. A2222 W: 39:115-118. April. 1955.

39
be most valid for predicting grades in science.

3. The number test of the DAT was the best over-all predictor of
academic success; its correlations with all subject areas except home
economic and industrial arts being .55 or higher.

If. The DAT verbal test proved to be the second best over-all pre-
dictor of academic success, while the PMA verbal test was the third best

Ever-all predictor of success.

5. For the most part, the Differential Aptitude tests proved to be
superior to the Primary Mental Abilities tests in terms of their
relationship with marks in the various subject areas.

I" 6. The study indicates some potential for the predictinn of academic
success in general, but throws some doubt on the immediate usefulness
of the various subtests as differential predictors of success in various
Lsubject areas.

There probably has been no other test or battery of tests for
which more validation data has been supplied than the Differential
Aptitude tests. Thousands of correlation coefficients have been reported
indicating the relation of these tests to school marks, achievement
tests, and other criteria of success. Fortunately, the authors of this
battery have sunmarized the results of numerous studies in their manual
by providing median correlations for boys and girls in most of the
subject areas. In subject areas in which median correlations were not
reported, they were computed by the writer. The DAT subtest, providing
the highest median correlation with each of the subject areas of interest
in this 817W, is shown in Table 3 for boys and girls separately.

Table 3 reveals a number of interesting facts. Except in the

to

TABLEB

DAT ems momma m HIGHEST mom CORRELATION
ma EACH OF THE RIGHT HIGH SCHOO SUBJECT AREAS
FOR BOYS AND GlRIS3

 

 

Sub] ect Area §_e_oc_ Best Predictor
Test Median r
English Boys Sentences .50
Girls Sentences .53
Social Studies Boys Verbal Reasoning .h8
Girls Verbal Reasoning .52
Science Boys Verbal Reasoning .5 h
Girls Verbal Reasoning .55
Mathematics Boys Numerical Ability .h?
Girls Numerical Ability .5 2
Foreign Language Boys Sentences .51
Girls Numerical Ability in
Industrial. Arts* Boys Numerical Ability . 28
Home Economics Girls Numerical Ability .32
Business Girls Sentences .39
Education

 

*Includes the DAT manual areas of industrial arts, mechanical drawing,
shop, and woodworking.

35Bennett, Seashore and Wesnan, $.93} , p.hO-Sl.

bl
area of foreign language, higher relationships between the tests and
marks are seen to exist for girls than for boys. Secondly, the tool
subject areas (and foreign language) are seen to relate higher to the
predictors than do the remaining three subject areas, i.e. industrial
arts, home economics and business education. The Sentences, Verbal
Reasoning, and Numerical Ability Tests are seen to be among the best
three predictors of school marks.

The Differential Aptitude Tests have also been studied in terms
of their relation to achievement test scores as criteria of success in
subject areas. Although the DAT manual reports the results of numerous
correlational studies with various standardized tests, the results of
studies using the Essential High School Content Battery (ERSCB) as
criteria are of particular interest, since this instrument is used as
an alternate criterion in the present study.

The Numerical Test of the DAT proved to show the highest rela-
tionship to the Mathematics Test of the EHSCB for both boys and girls,
the correlations being .66 and .56 respectively. The Science Test was
best predicted by the DAT Verbal Reasoning Test with correlations of
.65 for boys and .62 for girls. The Verbal Reasoning Test also correl-
ated highest with the Social Studies Test to the extent of .57 for boys
and .58 for girls. The highest correlation in predicting the English
scores of girls was shown by the Sentences Test (.66), while for boys,
the Verbal Reasoning Test proved to be the best predictor (.65). The
composite score of the mm was predicted highest by the Verbal
Reasoning Test for boys (.75) and by the Sentences Test for girls(.67).

It is seen that the three DAT Tests showing the highest median

h2

correlations with school marks; the Verbal Reasoning, Numerical
Ability, and Sentences Tests also are the same three subtests which
correlate highest with the EHSCB. From this, one might speculate that
few courses or subject areas require special abilities, as such, but
rather require various degrees of a generalized factor.
In investigating a problem of transfer of training, wesman36
administered tests measuring achievement and intelligence. The intel-
ligence test measured verbal, numerical, and spatial abilities and the
achievement test covered social studies, natural science, mathematics,
literature, reading comprehension, contemporary affairs, and a foreign
language. He obtained an average correlation of .h85 between verbal
ability test and the various measures of scholastic achievement. The
average coefficient between number ability and the same measures was
.35 while the average correlation derived from the Spatial test was .285.

Holzinger andSwineford37 investigated the relation of two
bi-factors to achievement in several subjects. They reported a multiple
correlation of .768 between a general mental factor on the one hand and
the American Council Cooperative Plane Geometry Test scores on the

other. Zero order coefficients, reported in such subjects as English,

biology, foreign language, chemistry, history, Shop and crafts, and

 

36Alexander G. wesman,_A Stugy_ of Transfer of Training (New York
Teachers' College, Contributions to Education, No. 909, New York, l9h5),
p025.

37Francis Swineford and Karl J. Holzinger, A Stqu_ in Factor
Analysis: The Reliability_ of Bi-Factors and Their Relation_ to ather
Measures (fififiyerSity of Chicago, Supplementary'Educational— Monographs,
NO. €350

h3
drawing ranged from .219 to .586 for the "general mental factor" and

from -.003 to .682 for the spatial factor.

General intelligence tests also have been used by various inves-
tigators for prediction of success in subject areas. One such study was
conducted by Ohlson.38 In this study, Ohlson investigated the Terman
Group Test scores of 200 boys and 306 girls to ascertain what correl-
ation, if amt, existed between the mental ability of the students and
the marks they received in high school. The Terman Test was given
during the students' last year of high school. The correlation between
the Terman Group scores and the average marks received by the total
group of 506 pupils was .38. The highest correlation was seen to exist
with marks in English; the correlation being .h5. The mathematics and
science departments, with about the same number of students, showed
lower correlations, which was also true of the foreign language
department; being .33, .31, and .214 respectively. In the vocational
department, commercial, home economics and art, and manual arts, the
correlation between the Terman Test and school marks was very slight,
being .18, .12, .15 respectively. Marks in history correlated to the
extent of .37 with the Terman Test.

R833 and Hooks39 smmnarized a group of correlations relating
intelligence tests and achievement in high school subjects, such as
English, Latin, and mathematics. The coefficients they reported ranged

 

38David Ohlson, "School Marks and Intelligence ,*' Educational
Administration 55g Sgervision, 13:90.102, February, l92’h

39c.c. Ross and um. Hooks, "How Shall We Predict High School
Achievement?” Journal 2f Educational Research, 22:18h-95, March, 1930.

14h
from..18 to .72 with a median of .39.

Eurich and Cainho in summarizing the results of correlational
studies between intelligence and high school achievement as reflected
in grades, calculated the median correlations of more than 300 coef-
ficients reported in various studies. In the area of science, the median
correlation was found to be .hh; in mathematics .37; in foreign
language .33; and in history and English, .h5. The median correlations
reported between intelligence tests and achievement test results were:
.h5 in science .hl in mathematics; .h6 in foreign language; and .27 in
history and English. The authors state, "Although the coefficients
occassionally fell in the lower .70's, the summary indicates clearly
that intelligence tests cannot be depended upon with any high degree of

accuracy for predicting achievement in specific subjects."l‘l
Aaronh2 also has summarized a number of investigations attempting

prediction of high school achievement. Although the correlations
reported relate to subjects rather than subject areas, the results are
worth reporting. Her summary includes the median correlations estab-
lished for intelligence tests in.predicting success (as indicated by
teachers' marks) in high school algebra, plane geometry, Spanish,
biology, physics, and chemistry. These correlations were found to be

hoAlvin C. Enrich, and Leo F. Cain, "Prognosis in Secondary
gghools £1" Englclopedia 9; Educational Research, 191:1 Edition ,pp. 8M;-
9, 19 . """"“""""“'

h11bid., p.8h6.

hZSadie Aaron, "The Predictive value of Cumulative Test Results,"
(Doctor's Thesis, Stanford University, California, 19h6), p.227.

h5
.h8 for algebra, .hh for geometry, .35 for Spanish, .51 for biology, .53

for physics, and .29 for chemistry.

Many studies have also been conducted in which the prediction was
focused on one subject area only. Several of these studies will now be
reviewed with an emphasis on those subject areas in which comprehensive
reviews have not been made.

Prescotth3 reported a study in which he attempted to determine
the effectiveness of the Turse Clerical Aptitudes Test in predicting
success in commercial subjects. Since this subject area includes
subjects in business education, an area considered in the present study,
the results will be of interest.

The Turse Clerical Aptitudes Test includes separate measures of
Verbal Skills, Number Skills, Learning Ability, Clerical Speed, Clerical
Accuracy, and General Clerical Aptitude. The criterion measures
included teachers' marks and achievement tests. The subjects were
students entered in the commercial curricula at two large high schools.

Correlations between the Verbal Skills, Number Skills, Clerical
Speed, and General Clerical Aptitude and.various achievement tests
ranged from .32 to .68 with a median of .58, while the range of correl-
ations reported with teachers' marks for the same four predictors was
.36 to .70 with a median of .51.

An investigation was undertaken by Limphh to select a battery of

 

hBGeorge A. Prescott, "Prediction of Achievement in Commercial

Subjects," Educatio al egginychological Measurement, 15:h91-h92,
Winter, 195 .

th.E. Limp, "A tbrk in Commercial Prognosis," Journal‘gf
Educational Research, l6zh6-56, June, 1927.

146
tests which would predict the ability of high school students to succeed
in typewriting and stenography. Fbrty tests were administered including
intelligence, will-temperament, motor, and secretarial skills tests as
well as tests of attention, perception, speed and coordination of
reaction, and ability to follow directions. The subjects of this study
were 118 beginning students in typewriting and shorthand. The criterion
of success in typewriting and shorthand was a combination of weekly
speed tests, average rankings by teachers, and a semester grade. The
highest correlation beheeen the predicted scores and the criterion
scores was .61 for shorthand and .62 for typewriting. His findings
showed that secretarial aptitude can be predicted to a fairly high
degree.

Pilliteleyl“S also undertook a study to determine the ability of
certain standardized tests to predict secretarial success. The subjects
were 108 students in the Packard School of New York City. TWO criteria
of success were used; the completion of the course in shorthand, and
the time taken to finish the course. Students were advanced as readily
as they progressed. The tests administered were the Army Group ham-
ination (Alpha), the Hoke Pragmatic Test of Stenographic Ability, the
kbodworth-House Mental Hygiene Inventory, and the Sims Socio-Economic
Rating Scale.

The significant findings of Whiteley's study were: 1) There was
a definite negative relation between the time it took to finish the

 

b58arah S. unteley, "Predicting Stenographic Success Through
Prognostic Tests," The Balance Sheet, 18:2h2-hh, March, 1932.

117
course and the scores obtained on the Armr Alpha; 2) the Sims Rating
Scale failed to descriminate between students who finished the course
and those dropping out; It) the Hoke prognostic Test of Stenographic
Ability proved to be the best single predictor of success in this course
as measured by the completion of the course and the time taken to finish
the course. The Hoke Test did not, however, differentiate between
graduates and drop-outs.

In the subject area of mathematics, Lee and Hughes"‘6 studied 329
students taking algebra and geometry. Teachers' marks and achievement
tests were used as criteria of success. The predictors used in their
investigation included the Lee Test of Algebraic Ability (and Geometry
Aptitude), the Hughes Trait Rating Scale, the Kuhlman-Anderson and
Terman Group intelligence tests, and teachers' ratings of mathematical
ability made two weeks after the students hadientered the courses.

The aptitude tests gave the best single prediction with achieve-
ment test scores in algebra (.62) and geometry (.63), followed by the
Kuhlnan-Anderson Test which predicted algebra and geometry achievemnt
to the extent of .56 and .51; ,respectively. The best predictors of
teachers" marks proved to be their own ratings at the beginning of the
courses. The correlations were .59 and .h2 for algebra and geometry,
respectively. Trait ratings were found to be much more important in
predicting marksthan they were in predicting achievement test results.

On the whole, the order of merit for predicting achievement in

 

d

146.1. Murray Lee and W. Hardin Hughes, "Predicting Success in
Algebra and Geometry," School Review, hams-96, March, 1931;.

118
mathematics seems to be: 1) good prognostic tests, 2) mathematics marks
for the previous year, 3) intelligence quotient, h) mental age, 5)
achievement tests in arithemetic and algebra, and 6) average marks in
previous years.)47

Many studies have been conducted for purposes of predicting
success in foreign language. Seagoe,)48 for example, studied 120 students
in the seventh grade in an attempt to predict their achievement in
foreign language over a three year period. The predictor tests used
included the Terman Group Test, KuhlmanaAnderson Test, Otis Intermediate
Test, the New Stanford Achievement Test, the Luria-Orleans Modern
Language Prognosis, the Stenquist Mechanical Aptitude Test and the
Orleans Algebra Prognosis Tests. The mathematics tests were included to
determine the comparative relationship to, or independence of, the
foreign language prediction. Certain sections of the Stanford Achieve-
ment Test and the Mechanical Aptitude Tests were included to explore the
relation of fereign language achievement to such theoretically unrelated
factors as scientific and.practical ability.

The median correlations with course records were as follows:
language prognosis .73; algebra, .h6; reading achievement, .h9; intel-
ligence tests, .53; arithmetic achievement, .SO; physiology achievement,
.hO; and.Stenquist Mechanical, -.11. Reading achievement seemed to be

less valid than either the general intelligence or language prognosis

 

1‘7H.R. Douglass, "Special Methods on High School Level: Mathema-
matics," Review‘gfiEducational Research, 2:7-20, 81-82, February, 1932.

heM.V’. Seagoe, "Prediction of.Achievement in Foreign Language,"
Journal 2;: Applied P cholo , 22:632-6h0, December, 1938.

, ’49
13051530

Most of the studies of prediction of foreign language achievement
have been sumarized by Kaulfers.h9 Some of the median correlations he
reports with various measures are .600 with prognosis tests, .h9 with
achievement in algebra, .h6 with achievement in English, .385 with
achievement in reading, .356 with intelligence tests, .1614 with achieve-
ment in arithmeitc, and «2).; with chronological age. The great range of
correlations reported, varying from low negative to nearly perfect
positive correlation for a single characteristic, is notewortlv.

III. SUM’IARI

It is apparent that in the vast majority of studies reported in
the literature, correlational techniques are the most common methods of
showing the relationships between various predictors and criteria of
achievement. For the most part, the correlations range from .hO to .60
with a few reaching the .70's. The prediction of general academic
success seems to show correlations of about the same magnitude as those
shown in predicting success in subject areas and specific subjects.
Although the correlations are sufficiently high to make them useful in
studying groups, they are not sufficiently high to warrant their use
on individuals in a counseling situation - at least when'considered
alone. Individual predictions of success in high school, based on tests
or other measures, can be considered only a small segment of the total
picture that is needed in aiding students to make wise selections of

 

1‘9 Walter Vincent Kaulfers, "Present Status of Prognosis in
Foreign Language,” School Review, 39:585-596. June, 1931.

50
subject areas at the high school level.

F is is the case in most studies of prediction, there are two
fundamental considerations. They are the reliability and validity of
the criterion, and the reliability and validity of the predictive
measure. For the most part, the reliability of tests, as predictors is
sufficiently high to warrant their use. The reliability of school marks,
however, has been shown repeatedly to be low. The validity of marks also
may be seriously questioned as indicators of success in school.

The use of achievement tests as criteria of success may have
their advantages as more reliable measures, but their validity in
specific situations is difficult to ascertain. Certainly the successful
outcomes of courses of instruction cannot be measured totally by paper
and pencil tests. The more intangible outcomes, however, may be and
usually are, reflected in teachers' marks. It is seen, therefore, that
both types of criteria of school success have their assets and limita-
tions.

The prediction of academic success of girls is generally of a
higher magnitude than that of boys, although the magnitudes of correl-
ations reported in subject areas are too wide in range to make it
possible to rank subject areas in order of their predictability. On the
whole, however, the typical tool subject areas, i.e. English, mathe-
matics, science and social studies, seem to be susceptible to higher
prediction than the vocational subject areas such as industrial arts,

Lhome economics and business education.

The use of personality tests as predictors of academic success

have shown widely diversified results. Such is the case with other

51
non-intellectual variables. On the whole, however, the results of pre-
dicting success with non-intellectual variables has shown lower measures
of relationship when compared to the results derived from the use of
intellectual measures. The multiple prediction of success in school,
using a combination of intellectual and non-intellectual variables as
predictors, has proven to be a fruitful approach because of their rela-
tively low inter-correlations.

The next chapter describes the criteria and sources of data which
were used in the evaluation of the seven tests used in this study as
predictors of high school achievement.

CHAPTER III
THE CRITERIA AND SOURCES OF DATA
I. THE CRITERIA

Prediction in gecific subjects my prediction of subject-
m 23%. Since standardized tests began to be used, hundreds of
studies have been made in an attempt to predict academic success at
all educational levels including, of course, the high school level.
Host of these studies deal with the prediction of success in certain
specific subjects such as biology, algebra, Spanish, etc. The
assunption is made that there may be differential prediction among
specific subjects. This assimption finds some support in the fact that
students often do not achieve the same degree of success in one subject
as they do in another. These differentials are the result of many
factors such as interest, aptitude, past success in the subject area,
and teacher differences. The results of such studies, however, are too
often of limited guidance value. The typical eighth grader is not as
much interested in his success in a particular subject as he is in his
possible success in various areas of academic study. For example, the
more fundamental decision will be based on whether one should major in
science rather than whether one should pursue chemistry, physics, etc.

This approach makes more sense from at least two points of view.
First, in most high schools, a certain number of high school majors
(similar to those defined in this stuchr) must be selected as part of
the graduation requirements. From this viewpoint, the junior high school

53
student entering high school must make decisions in terms of the

selection of subject areas to constitute his majors.

Secondly, it is common knowledge that one of the most important
factors in the academic success of a student is the conpleocity of
attitudes, interests and motivations which make up his ”pm-perspective"
of the subject matter. On this basis, it is maintained that a student's
perspective is more oriented toward a subject field or area rather than
toward specific subjects, primarily because he has some notion through
past experience as to the nature of most subject areas. For example, the
typical junior high school student has some idea concerning the nature
of science, mathematics, social studies, and industrial arts because in
marry cases he has had some contact with these areas in past cm'rimlla.
Within the area of science, however, he may have no notion as to what
biology is.

Thus, one of the basic assumptions to the approach of this stch
is that students tend to select specific subjects within the areas in
which they feel they have a desirable perspective. It follows, therefore,
that in using this lblistic approach, prediction of academic success in
subject areas will prove more useful.

Two criteria serve as a basis for this stuck - grade point aver-
ages and scores on the Essential High School Content Battery. The former
serve as the major criterion since the major purpose of this study is to
predict the high school grades in subject matter areas. The latter
criterion is more supplementary and is used for cooperative purposes
and as a check on the validity of grades. It should be noted, however,
that although prediction is to be attempted in eight high school subject

Sh
matter areas, i.e. business education, English, foreign language, home

economics, industrial arts, mathematics, science, and social studies,

the EHSCB serves as a check on only four of the areas (English, mathe-
matics, social studies, and science). In addition, the total score on

the ‘03 will serve as a supplementary criterion for the total grade

point average, which is the criterion for general academic success.

Much has been written concerning the use of the grade point
average as a criterion for success in course work. Many weaknesses are
evident in this criterion. Among these weaknesses are those with
reference to the un-reliability of grades; teacher, school and system
wide differences in grading practices; and the lack of ability to
measure accurately varying objectives, content, and educational out-
comes. It is argued, then, that success in education. cannot be reflected
in a school grade. To some extent, the presence of these limitations
cannot be denied; however, when all is said and done, it still remains
a fact that students, parents, teachers, administrators and business and
industry rely to a considerable content upon school grades as a reflec-
tion of a student's academic success. No other defense for the use of
this criterion will be made. It seems obvious that the important fact
to be remembered is that the criteria for success 539. grades and to
accept the results in light of this fact, with due cognizance of their

Llimitations.

The courses constitutm .a_ subject-matter 533g. Before the pro-
cedure for determining grade point averages can be described, it is
necessary to define first the courses making up a subject-matter area.
For this purpose, the Proggam 2; Studies of the Cincinnati Public

55
Schools was used. In this Program pf Studies, ten subject-matter areas
are delineated, two of which (Music and Art) are not included in this
study because of the small numbers of students selecting these areas for
high school majors.

Since a major is defined as aw combination of three or more
units in a subject-matter area, it is obvious that a major may be many
different combinations of courses within an area. Since the approach of
this stuw is to determine whether '"holistic" prediction can be made,
it is not necessary to know what exact combination of courses is
involved. To indicate the courses most commonly used as majors in each
subject-matter area, and the courses which constitute each of the eight
subject areas, the following table is presented.

mush

SUBJECT MATTE AREAS DEFINED BY
THEIR CONSTITUENT COURSES

 

 

yam Area Constituent Courses-H-

 

Business‘Education Typewriting 1*, 11*
hbrld Geography
Business Arithmetic
Consumer Education
Shorthand 1*
Bookkeeping 1,11
Salesmanship and.Advertising
Secretarial Practice

with Shorthand 11*

Office Practice*

Foreign Language French 1*, 11*, III, IV
German 1*, 11*, III, IV
Spanish 1*, 11*, III, IV
Latin 1*, 11*, III, IV

English
(English I, II, 111

required of all
students)

Home Economics

IndustrialaArts
(majors about equally
distributed.among all
courses)

Mathematics
(one unit of Mathematics
required of all students)

Science
(one unit required
of all students)

Social Studies
(two units required
of all students)

56

TABLE h (continued)

English 1*, 11*, 111*, IV
Dramatics

Public Speaking
Journalism

Advanced Speech

Debating

Home Economics 1*, 11*, III-x-
Consumer Education

Electricity 1, II

Metalwork I, 11

hbodwork I, 11

Mechanical Drawing 1, II, 111, IV
Graphic Arts 1, 11

General mathematics

Plane Geometry*

Business Arithmetic

Algebra*

Consumer Mathematics

Mathematics 111* (primarily
Advanced Algebra)

Mathematics IV* (primarily Solid
Geometry and Trig.)

General Science
Biology*
Botany

Zoology
Chemistry*
Physics*

Physiologyt

world History*

world Geography
American History*
Economics and Sociolo
(each.met one semeste§§
American Problems*

 

*Indicates those courses most commonly chosen to constitute majors.

**Each.course here presented carries one high school unit of credit and
meets five periods per week for the full year of school.

57

Method pf determining grade mint averages. All course grades
were converted to the typical four-point scale, ranging from 0.0 to h.0,

 

or from the lowest I'F'" to the highest "A", respectively. The final
course grades from two of the six schools originally studied, recorded
grades in this manner so that no conversion was necessary. In three of
the remaining high schools, grades were recorded on a thirty-two point
scale, this score representing the sum of semester grades. The con-
version to a four-point scale sinply involved dividing the score by
eight. The remaining schools recorded grades as A, B, C, D, F. This
system was converted by ascribing four points for an A, three points
for a B, two points for a C, one point for a D, and zero points for an
F.

After the scores in all six schools were converted to the common
four-point scale, grade point averages in majors were calculated by
dividing the sum of all courses taken in a subject area (the minimum,
of course, being three) by the number of courses taken. In the great
majority of cases, this average was based on the grades in three
courses. Some averages were based on grades in four courses, and very
rarely did five courses comprise a major.

The total grade point average, which is used as the criterion
for general academic success, was derived by computing the average
grades in all one unit courses whether the course was a part of a major
or not. This average did not include such courses as Physical Education
or Health since these courses carry only one-half unit of credit. In
general, the total grade point average was based on between fifteen

and eighteen course grades.

58

The comarabilitz 2; criteria between schools. As stated
previously, the original subjects of this study came from all six compre-
hensive high schools of the Cincinnati Public Schools. Since there is a
significant variability in the nature of the student population
attending these schools, a serious question arises. Do grade point
averages reflect a similar standard from school to school? In other
word, does an ”A" given in School "X" carry the same significance as an
"A" given in Schoot "1"? Obviously, before the subjects in each school
could be pooled and treated as a single population, it would be neces-
sary to answer this question.

The reasoning used in determining the answer to this question
was as follows: if a direct analysis of the significance of difference
between grade point average means in the criteria among schools were
used, the fact would be overlooked that there are individual differences
in the capability of youngsters comprising a school. In other words,
by using an analysis of variance, for example, suppose it was found
that significant differences existed in the criteria between the
schools. Suppose, however, that although School "X" did have a signifi-
cantly higher grade point average than school "I", it also had a higher
level of scholastic ability. Then, one would 2393915. a difference in the
criterion scores merely on the basis of differences in initial levels
of ability. In such a case, if no differences were found in criterion
scores, than one could conclude that there probably age differences in
grading practices between the schools.

To approach the problem in this light, the method of analysis of

covariance was used. Using this method, the criterion means were

59
adjusted statistically, relative to the levels of ability of the groups
involved. The instrument used to estimate the ability levels of the
schools was the American Council on Education Paychological Examination
(l9h8 high school edition). This instrument was administered in Feb-
ruary, 1956 when the subjects were high school juniors.

Befbre proceeding with this analysis, however, there was reason
to believe that one of the high schools would not conform to the grading
practices in the other schools. This school is a college preparatory
school, admitting only students with an 1.0. of 110 or above, with high
past achievement records, and with the recommendation of the principal
of the school previously attended. This fact is shown in the following
table where a comparison with other schools can be made.

The college preparatory school just mentioned is seen as School 6
in Table 5. It will be noticed that with the mean ACE score of 11h, it

TABLE 5

TOTAL GRADE POINT AVERAGE AND ACE MEANS FOR THE
SIX CINCINNATI PUBLIC HIGH SCHOOLS

....

O a race 'vln ‘

 

School N Average Means ACE Means
1 168 2.22 77.5
2 117 2.16 h5.9
3 221 2.66 91.6
h 206 2.32 83.1
5 182 2.36 89.3
6 %§; 2.h0 l .0
Totals 1 Total mean 2733' . 6

 

would be necessary.for this school to give an average grade point
average of well over 3.5 to be consistent with the grading practices in
the other schools relative to the ability levels of the student body. In

6O
addition, this school does not offer "non-academic" courses such as
Industrial Arts, Home Economics, and Business Education from which a
student can select a major. For these reasons, School 6 was imrnediately
excluded from further analysis.

A subsequent analysis of the total grade point averages in the
remaining five schools gave the results shown in Table 5. Random samples
of fifty students from each school were used in this analysis. Samples
were drawn by use of a table of random numbers. The level of signi-
ficance arbitrarily selected was at the five per cent level of
confidence.

The F ratio in the above analysis indicates significant
differences in the criterion means even when adjustments are made for
levels of ability as measured by the ACE. By using a series of designs
such as the one above for not only total grade point average but also
for grade point averages in the subject areas, and by withdrawing those

TABIE 6

RESULTS FROM AN ANALYSIS OF COVARIANCE OF TOTAL GRADE
POINT AVERAGES ADJUSTING THESE MEANS Fm
LEVELS OF SCHOLASTIC ABILITY IN
FIVE CINCINNATI HIGH SCHOOLS

 

 

"" ' "' "'""" ' "' """" ' ' "' ""'"""De""gree's"""—_""'"—T""d'nee er-
Source of Sum of Squares of of Mean F at 5%
Variation Errors of Estimate Freedom Square Ratio Level
Total $7.99 2&8
Within

groups 57.89 2M1 .237
Adjusted

means 10.10 )4 2.525 10.65 2.1a

schools whose contribution to between variance was largest, three
schools (schools 1, 3, 1;) finally were selected in which grade point

61
averages were not significantly different (with the exception of one
subject area). The results of the analyses among the three remaining
schools are shown in Table 7, by subject area. In general, random
samples of the scores of fifty students from each school were used in
the analyses. In three areas, however, (Language, Home Economics, and
Industrial Arts) the number of students selecting these majors was less
than fifty. In such cases, the lowest number of scores represented by
any one school was used, and a similar number of scores was chosen
randomly from the rmnaining twc schools. In one case, Business Educa-
tion, a random sample of sixty scores from each school was used for the
analysis.

By inspection of Table 7, only one of the analyses met the
standard of significance set, namely grade point averages in mathematics.
A11 of the remaining analyses confirmed the null hypothesis. Why grade
point averages in mathematics were significantly different is not known.
Perhaps the particularly harsh or lenient marking practices of one
teacher are responsible for the difference. Because of this discrepancy,
the pooled subjects in only two schools (school 1 and h) were used for
the prediction of grades in mathematics. The F-ratio derived from the
analysis of covariance of these two schools was 2.21, while that needed
for significance at the five per cent level of confidence- was 3.91;. The
degrees of freedom for this evaluation were 1 and 97.

It is an interesting fact that on the whole, the marking practices
among these three schools, in their respective subject areas, is
strikingly similar when due adjustment is made for levels of scholastic
ability. It is worth;' of note that these three schools are among the

62
TABLE 7
ANALYSIS OF CUVARIANCE OF TOTAL AND SUBJECT AREA GRADE POINT AVERAGES

ADJUSTING THESE MEANS FOR LEVELS OF SCHOLASTIC
ABILITY IN THIEE CINCINNATI HIGH SCHOOLS

 

Source of Sum of Squares of of Mean F at 5%
Variation Errors of Estimate Freedom Square Ratio Level

Total Grade Point Average

Within
groups 32.82 lh6 . 225
Adjusted
means .81; 2 .1120 1.86 3.06
Business Education Gra de Point Average
Total 22.00 178
Within -
grows 21.35 176 .121
Adjusted
means .65 2 .325 2.68 3.05
Em; ish Grade Point Average
Within '
groups 53.97 11.6 .369
Adj usted
means 1083 2 .915 20118 30%

Law Grade Point Average

Total 25 . 32 61
Within
groups 214.112 59 .1113
Adjusted
means ' .90 2 .1150 1.08 3.15

Home Economics Grade Point Average

Total 11.32 3h

Within

groups 10.111 32 .311
Adjusted

means .90 2 .1450 1.08 3.15

TABLE 7 (continued)

Industrial Arts Grade Point Average

Total 20.59 58
Within

groups 20.32 56
Adjusted

means 1.27 2

Mathematics Grade Point Average

Total 69.hl 1&8
Within

groups 57.82 lh6
Adjusted

means 11.59 2‘

Science Grade Point Average

Total h6.77 lh8
Within

groups hh.9h 1&6
Adjusted

means 1.83 2

Social Studies Grade Point Average

Total h6.99 1&8
Within

groups h5.95 lh6
Adjusted

means 1.0h 2

.362
.135

.389
5.795

.307
.915

.31h
.520

.36

1h.8h

2.98

1.67

63

3.15

3.06

oldest schools in the Cincinnati system. It is likely that the teaching

personnel comprising these schools represent the older and.more well-

established teachers in the system. If this be the case, it may well be

suspected that their similarity is due to the longer periods of inter-

action between the teachers in these schools. Informal exchanges of

:marking practices may have led to the homogeneity observed.

0n the basis of these analyses then, the pooled twelfth grade

students from.these three separate high schools comprise the subjects

'used in the remainder of this investigation. The total number thus

61:
derived.was 595 students made up of 329 girls, and 266 boys.

II. SOURCES OF DATA

The method of selecting the high schools used in this study was
described in the preceding section of this chapter. The total number of
students comprising the senior classes of these schools, however, was
not included in this study. The method of selecting the experimental
group was simple; only students who had taken.gll’of the standardized
tests used in this study were included. This final group consisted of a
total of 595 students.

Source of test data. The testing instruments used were adminis-
tered to the students comprising this study on the following dates:

1. Differential Aptitude Test Sub-tests - February, 1953

2. (Metropolitan) English Proficiency Test - March, 1953

3. (Metropolitan) Mathematics Proficiency Test - March, 1953

h. TermanéMcNemar Test of Mental Ability - September, 1953

5. American Council on Education Psychological Examination
February, 1956

6. Essential High.School Content Battery - May, 1956

With the exception of the English and.Mathematics Proficiency
Tests, all tests were administered by trained examiners from the Division
of Appraisal Services, Department of Instruction, Cincinnati Public
Schools. The tests were scored, checked, and recorded by personnel
trained for this purpose. Summary sheets of test score data were then
typed and sent to the respective schools. It is from these data sheets,

that the test data were obtained. All test scores were sent back to the

65
schools except the Differential Aptitude Test results which were
considered experimental in nature.

m a; grade m averages. After all the test data were
recorded on large tabulation sheets, students who had not taken all of
the tests were immediately rejected for further study. The names of the
remaining group were then used to look up the course marks for each
student. The grades were obtained from the office records of the
respective schools. These office records include not only the course
grades but also indicate the student's high school majors. In some
instances, students were completing (or had completed) a major which was
not recorded on the office records. For this reason, care was taken to
peruse the courses taken for further identification of student majors.
When three or more one-unit courses were found in any of the subject
matter areas studied, the grades from these courses were recorded. The
results of averaging these and reducing them (if necessary) to a four-
point scale represent the final grade point averages used.

m 9; 2222?. Ego-d. The tests used in this study not only
include predictor tests but also an achievement battery, the Essential
High School Content Battery, used as an alternate criterion of
scholastic success, and a scholastic aptitude test, the American Council
on Education Psychological Examination, used as a basis for; the
covariance analysis described in the previous section. The predictor
tests are as follows: four sub-tests from the Differential Aptitude
Battery, i.e. Verbal Reasoning, Numerical Ability, Mechanical Reasoning,
and Language Usage, (made up of two sub-tests, Spelling and Sentences);

the Terman-McNemar Test of Mental Ability; and an English and

66

Mathematics Proficiency Test. The predictor tests thus include measures
of special and general aptitude as well as two measures of past accomp-
lishment.

A more detailed description of each of the tests listed above
will now be made.
A. 1133 Differential Aptitude £32532

1. General moss. The general purpose of this battery is to

"provide an integrated, scientific, and well standardized

procedure for measuring the abilities of boys and girls in grades

eight through twalve for purposes of educational and vocational
guidance."1
2. Description. These tests were administered during January and
February, 1953 to all pupils in grade eight of the Cincinnati
Public Schools. The pupils of this class now represent the current
(1956-57) senior class. The description of each test given to
this class is as follows:

1133:1331 Reasona‘ g: This test is composed of simple analogies.
The words used inthe items come from history, geography,
literature, science, and other content areas. The items are
intended to sample the student's knowledge and his ability to
abstract and generalize relationships inherent in that
knowledge.

 

1George K. Bennett, Harold G. Seashore, Alexander G. Wesman,
Differential Aptitude Tests, Manual-Second Edition (New York: The
Psychological Corporation, 1952), p.1.

67
Numerical Ability: The items on this test are designed to
test understanding of numerical relationships and facility in
handling numerical concepts. The problems are of the type
usually called ”arithmetic computation" rather than the
"arithmetic reasoning" type. The items were set up in this
manner to avoid the language elements of the usual arithmetic
reasoning problem, in which reading ability may play a signi-
ficant role. Actual try-out of the test in its preliminary
form, however, demonstrated that the items are so constructed
that the measurement of reasoning ability is not sacrificed

by the use of the computation type.

Mechanical Reasoning This test is essentially a new form of
the series of Mechanical Comprehension Tests used widely by
industry and the military. Each item consists of a pictorially
presented mechanical situation together with a simply worded
question. Care was taken to present items in terms of simple,
frequently encountered mechanisms that do not resemble test-
book illustrations or require special knowledge. It should be
noted that the authors of the Differential Aptitude Test
Battery consider the Mechanical Reasoning scores of less
educational and vocational significance for girls than for

boys .

M: The spelling words were selected from the lists in
the Gate's.Spell-i_.pg Difficulties _i_.p_ 2876 Words, then further

selected for their prominence in every day vocabulary.

3-2312

68
Sentences: This section of the test is designed to measure
the student's ability to distinguish between good and bad
grammar, punctuation and word usage. It should be noted that
Sentences and Spelling are more nearly achievement tests than
any of the others. Their chief reason for being included in
the battery is that it is believed that they represent basic

skills necessary in many vocational pursuits.

3. Reliability. The authors of these tests present an elaborate
array of statistical data including numerous reliability coef-
ficients. Since reliability is a function of the group on whom it
was established, the authors present separate coefficients for
each sex by grade level. These reliability coefficients appear
sufficiently high to accept the long range consistency of the

scores 0

h. Validity. It is better to speak of the validities of the
Differential Aptitude Test since the number of validity coef-
ficients 13 momentous, being derived from a great variety of
situations using varying criteria. The particular types of
validity, with which this study is concemed, have been summarized
in the chapter reviewing the literature as prediction of course

grades and prediction of achievement test results.

TermannMcNemar Tests pf dental Ability
1. General m. It is the general purpose of this test to
attempt to measure those aspects of intelligence which are

69
considered verbal in nature. It does not profess to measure

”performance” or "qualitative" aspects of intelligence.

2. Description. This test represents a revision and restandard-
ization of the Terman Group Test of Mental Ability. The test
consists of 162 items arranged in seven sub-tests; Information,
Synonyms, logical Selection, Classification, Analogies, Opposites
and Best Answer. Because the arithmetical and numerical sub-
tests used in the original forms have been taken out, the test is
primarily a general verbal intelligence test. Since the number
of items in each sub-test is small, no separate norms are pre-
sented. Data are available to interpret the resulting total raw
score in terms of normalized standard scores, mental ages,

_. percentile ranks, and "deviation 1.0.“. The deviation 1.0. was
the particular score used in this study. It is simply the differ-
ences between the obtained standard score and the average

standard score for other individuals of the same age.

3 W. Three methods of determining the reliability of

this instrument were employed. The split-half method produced a
coefficient of .96 when determined on 279 cases in grades seven
through nine. The alternate form method showed a coefficient of
.95, being computed on 239 cases in grades seven through nine.

The probable error of measurement of this test is about 2.2

standard score points.

1;. Validity. According to the manual for this test, "the best

7O
evidence of the validity of the Terman Test is to be found in its
successful use over the period of yeaits since the test was first
used."2 One type of validity evidence is presented, however,
through the careful and comprehensive item analysis done on the
test. Items were chosen which successfully differentiated groups
of different maturity levels (as indicated by grade level). In
addition, an internal type of validity is evidenced by an average
item-test tetrachoric correlation of .53.

C. 1113 M1313 Proficiengy Test.
1. Generala m. The authors of the achievement battery of

which this test is a part assert, ”The separate subject-matter
tests conprising these batteries provide reliable measures of
individual achievement." They say further that a major use of
the tests is, "to determine the achievement level of each pupil
in each subject... To provide an objective and reliable basis for

classification and growing for instructional purposes."3

2. W. This test is a special edition published for the
Cincinnati Public Schools by the World Book Compaq. It is
composed of three sub-tests from the Metropolitan Achievement

 

2Lewis M. Terman and Quinn McNemar, "Construction of the Tests ,"
Terman-McNemar Test pf. Mental Ability (NW York: The World Book Compaw,

19 2 , p030

3Richard 1). Allen, Harold H. Bixler, et a1., "Content of the
Series,” Metro olitan Achievement Tests Intermediate. and Advanced
Arithmetic ests ew York: The World ook Company, 1957), p.I.

71
Tests, Advanced Battery, Partial: Form R. These sub-tests are
Reading, Vocabulary, and Spelling. The reading sub-test consists
of 52 items which attenpt to measure "paragraph meaning" and
"word meaning.“ The vocabulary test consists of 55 items which
in general require the student to mark a word meaning the same as
a key word. The Spelling Test consists of 50 items which are read
to the examinee. The sum of the raw scores of these sub-tests

represents the English Proficiency Test score used in this study.

3. Reliability. The manual for the Metropolitan Achievement Tests
reports the following split-half reliability coefficients corrected
by the Spearman-Brown formula: Reading - .937, Vocabulary - .9211,
and Spelling - .9113. These coefficients were computed on raw

scores from 280 seventh graders.

h. Validity. The type of validity associated with this test

(as well as other sub-tests of the battery) is often termed
”curricular" or "content" validity. The items are representative
of courses of study, textbooks, and the opinions of experts in the
field.

D. 1.113 Mathematics Proficiengy 1'33;
1. General m. The purpose of this test is similar to that
described for the English Proficiency Test except, of course,
that this instrument attenpts to measure achievement in arith-
metic.

2. Descrgtion. This test is also a sub-test of the Metropolitan

72
Achievement Battery, Advanced Arithmetic Test, Form B. The name
"Mathematics Proficiency" is a local term, since the same
mathematics achievement test is not given every year. This name
will be used throughout this study. The test consists of two
parts, Arithmetic Fundamentals and Arithmatic Problems. The
former consists of 57 items which measure essentially computa-
tional skills, while the latter consists of 33 items commonly
described as "story" or "word" problems. The sum of the two raw
scores of these sub-tests represents the Mathematics Proficiency

scores used in this study.

3. Reliability. The manual gives the following corrected split-
half reliability coefficients conputed from the raw scores of 280
seventh graders: Arithmetic Fundamentals, .9114 and Arithmetic

mblems ’ 0879 e

h. Validi y. The validity of this test is similar to the type
described for the English Proficiency Test.
E. The American Council on Education chhological Examination (19h?
mamam "' *-
1. General mg. The purpose of this test is to measure the
learning ability or scholastic aptitude of students in grades
nine through twelve.h

2.Descrgtion. The American Council on Education Psychological

 

hAmerican Council on Education, Manual of Directions, Tables of
Equivalent Scores and Percentile Ranks Wetan, New Jersey).

73
Examination is composed of four sub-tests: Same-Opposite,
Completion, Arithmetic and Number Series. The composite score of
the first two sub-tests represents the linguistic score, while the
latter two sub-tests represent the quantitative score. The differ-
entiation between the quantitative and linguistic abilities was
the result of factor analysis demonstrating these two basic
factors. The total score for the entire test indicates general
scholastic ability. For use in this study, the.American Council
on Education Test served as a basis for adjusting criterion
means for varying levels of scholastic ability among schools, in
order to determine the comparability of grade point averages from
school to school. The combined total of the L and Q raw scores

served for this purpose.

3. Reliability. The reliability of this test is estimated.by its
correlation with an equated form of the.American Council on
Education Examination and the test itself. This adaptation of a
test-retest procedure gave reliability coefficients of .89h and
.931 for ninth grade (N 302) and twelfth grade (N 26h) popu.
lations respectively. The corresponding standard errors of

measurements are 6.1h and 6.33.

h. Validi yp The validity of this test is based primarily on the
relevance of the material to scholastic aptitude and the simi-
larity'of test content to others which have been validated in

various school systems.

7h
F. 11113 Essential Egg School Content Batten (EHSCB)

1. General m. In their manual of directions, the authors of
this battery of tests state that this battery is ,"a conprehensive
battery of high school achievement tests coveringiin a single
booklet, four basic areas-u-mathematics, science, social studies,
and English. The battery is designed for use as a survey-type
instrument from the end of the ninth through the end of the
Welfth grade."5 A further purpose is to measure the students'

growth and development in the four areas mentioned above.

2. The Essential High School Content Battery, for purposes of
this study, serves as a supplementary criterion of success in
each of the areas of subject matter it attempts to measure. The
total score on the battery also serves as a swplementary
criterion for total grade point average or general academic
success. This battery was administered at the end of grade eleven
to the current senior class. Since this battery is composed of
four separate sub-tests, each one will be described in turn. The
table of norms provides for the direct conversion of raw scores

into standard scores for each of the sub-tests.

Mathematics - This test samples arithmetic skills, general

mathematics, algebra , geometry and to some degree

 

5David P. Harry and halter N. Durost, "Manual of Directions"
Essential H h School Content Batter-y (New York: The World Book

Comm, l9 1 , p.1.

‘75
trigonometry and advanced algebra. The author describes the
emphasis as being on “content having social utility, and on

understandings rather than manipulative skills."6 The Sub-
test is composed of sixtyhsix items being divided into eight

sub-parts as follows: fundamental skills in computation,
vocabulary and concepts, understanding of functional rela-
tionships, application of mathematics to life problems,
interpretation of’mathematical graphs, knowledge of’mathe-
matical facts and formulas, interpretation of data in

tabular form, and knowledge of important theorems.

Science - The science sub-test is made up of three parts: Part
A measures functional knowledge of factual material, Part B
measures the understanding and application of scientific
principles and concepts, and Part C measures the application
and understanding of methods of science. The items in the
above parts tap content in both the physical and biological

sciences. The total sub-test consists of seventy items.

§22;3;_Studiesa- The content areas covered by this sub-test
include.imerican History, Wbrld History, Civics and.Government,
Economics and Problems of Democracy. On the whole, however, it
measures factual knowledge in the field of social studies.

This sub-test has a total of ninety items being distributed

over ten subeparts as follows: acquaintance with

 

61mm, p.2.

76
contributions of famous Americans, understanding of current
social and political problems, understanding of vocabulary of
social studies, knowledge of civic information, growth of
Amarican democracy, knowledge and understanding of global
geography, knowledge of contributions of world leaders,
understanding of international relationships, knowledge of
the sequence of events in United States history, and know-

ledge of world events.

M - This sub-test includes measurement in the following
areas: understanding of the written language, precision in

the use of English, acquaintance with literary works, and
knowledge of reference sources. The sections of the sub-

test measuring these areas are as follows: reading for
information, vocabulary, business definitions, use of refer-
ences, literature acquaintance, language usage, capitalization
and punctuatiOn and spelling. The total sub-test contains

120 items.

3. Reliability. The reliability of the E.H.S.C.B. has been
indicated in three ways: by use of the split-half method, the
alternate forms method and by use of the standard error of
measurement. A partial reproduction of the reliability table
(6a) in the E.H.S.C.B. manual (for grade 11 only) is shown in

Table 8.

h. Validi I: The manual states evidence of the validity of the

77

E.H§S.C.B. in essentially two ways. The first type is commonly

called ”curricular" validity. The authors made intensive analyses

of typical offerings in the various areas of subject matter as

well as analyses of textbooks for the determination of test

content. The second type of validity is called item validity.

Presumably, itemptest correlations were used as indices of item

val idity e

SPLIT-HALF AND ALTERNATE FORM RELIABILITY
COEFFICIENTS, BI SEPIRATE GRADE LEWEUS

TABLE 8

 

 

Mathematics 101
Science 268
Social Studies 151
English 181

Total Battery 113

.87

.93
.89
.90
.90
.95

2.7

 

CHAPTER IV
METHODS AND PROCEDURES

I. THE TYPES OF TESTS USED IN THE STUDY

The types of instruments used in the prediction of attributes
depends, of course, upon the nature of the attributes themselves.Among
the types of tests used most commonly in the prediction of academic
achievement are so-called intelligence, achievement, and aptitude tests.
In a recent publication of the Test Service Bulletin, however, the
authors point out that discrimination among these three types of tests
cannot be on the basis of content or process since they are basically
similar in all three types of tests. The authors state further that in
terms of differentiating these types...

A logical candidate would seem to be function. What are we trying
to accomplish with the test scores? How are the results to be used?
hhat inferences are to be drawn concerning the examinee? If a
test's function is to record present or past accomplishment, what
is measured.may be called achievement. If we wish to make inferences
concerning future learning, what is measured is thought of as apti-
tude. One kind of aptitude test, usually some combination of verbal
and.mumerical and/or abstract reasoning measures, is sometimes
called an intelligence test; more properly, in educational settings,
it is called a scholastic aptitude test.

Since the purpose of this study is to evaluate certain tests for

their ability to predict academic success, the complexity of the pre-
dicted attributes warrants the evlauation of the three types of instru-

ments, since the prediction of future achievement is dependent not only-

 

lerold.G. Seashore, "Aptitude, Intelligence, and Achievement,”
Test Service BulletinI No. 51 New'York: The Psychological Corporation,
19g6): p.I. .

79
upon academic aptitude but also upon past accomplishment. Rather than
make armchair speculations as to the nature of the mental abilities
needed for success in each of the subject areas, it is better to secure
empirical evidence.

For this study, two measures of achievement (past accomplishment)
are used, i.e. the English.Proficiency Test and the Mathematics Profi-
ciency Test. The verbal Reasoning, Numerical Ability, Mechanical
Reasoning, Spelling, and.Sentences Tests of the Differential Aptitude
Battery represent five measures of "special” abilities or aptitudes.

The measure of general scholastic aptitude used in the study is the
TermanéMCNemar Tests of Mental Ability;

In an attempt to discover the mental abilities needed for success
in the various subject areas, it is recognized that paper and.pencil
tests measuring mental traits have a high degree of inter-correlation.
This fact acts against the possibility of isolating certain mental
traits needed for success. For a test to make a worth while contribution
to a testing program, it must either do a better job of performing the
functions than another test, or add to the performance of the function.
Before a test can add something which is not already being measured by
another test, it must obviously be independent of any high.relationship
with the existing test. Although one of the major purposes of this study
is to determine which tests show the highest relationships with the
criteria, it also is intended to isolate the differential abilities, if
any, that are being measured.by the seven tests used and that are needed

also for success in various subject areas.

80
II. TREATMENT OF RAW DATA

Before proceeding with an explanation of how the raw data were
treated, it may be well to state what form the raw data were in before
treatment. The scores used in this study were in the same form as they
were found and recorded from the original records. The types of scores
used for each of the tests are seen in Table 9.

TABLE 9

TYPES OF SCORES USED FOR EACH
OF THE PREDICTOR TESTS

 

Test Types of Scores
Differential Aptitude Tests Raw Score
English Proficiency Test Raw Score
Mathematics Proficiency Test Raw Score
Terman-McNemar Tests of Mental Ability I.Q. Score
A.C.E. Psychological Examination Raw Score
Essential High School Content Battery' Standard Score

 

After the raw data had been recorded on large tabulation sheets,
and each student given a four digit code number for personal and school
identification, the International Business Machines Service Bureau was
commissioned to punch IBM cards for each student with the appropriate
information from the original data sheets. All card punching was
verified.

With the generous help of the Applied.Science Division of the

International Business Machines Corporation, it was decided to utilize

81
a "ready-made" library program which would provide the means, standard
deviations, and inter-correlations between all the variables used. The
IBM data were then gang punched into the format necessary for this pro-
gram. The actual processing of the data was done by IBM's Magnetic Drum
Data Processing Machine, commonly called the 650.

Initially, the students comprising this study were separated by
sex. The two resulting groups were then processed for the procurement of
the data necessary for the prediction of total grade point average and
the grade point average in English (since the total group of students
must have had a major in English). From these two groups, social studies
majors then were sorted out and the necessary data again computed for
this group. Each major was sorted out in turn, until all majors had been
processed. It should be mentioned that the data thus derived are
extremely accurate. The library program used provided for checks which
indicate errors which may occur. The final data could not be punched

out of the 650 until the detected errors (if any) were corrected.
III. STATISTICAL METHODS AND PROCEDURES

2,1513 coefficient 21; correlation. The index of relationship used
in this study is the Pearson product-moment coefficient of correlation.
As such, it is well to take cognizance of the statement and illustration
given by Guilford:

A correlation is always relative to the situation under which it
is obtained, and its size does not represent any absolute natural or
cosmic fact. To speak of the correlation between intelligence and
scholarship ,is absurd. One needs to say which intelligence, measured
under what circumstances, in what population, and to say what kind
of scholarship, measured by what instruments ,or judged by what
standards. Always, the coefficient of correlation is purely relative

82
to the circumstances under which it was obtained and should be
interpreted in the liggt of those circumstances; never certainly,
in arm absolute sense.

As applied to this study, it must be remembered that the group
used in this study is a selected grow, by virtue of the fact that they
have attained the status of high school senior. This automatically
eliminates the great heterogeneity one finds at the eighth grade level,
which of course, is the point at which the prediction of this study is
being attempted. In addition to this fact, further selection takes
place on the basis of high school majors. Each group thus isolated
represents a much more homogeneous grow than the total grow. It is to
be expected, therefore, that correlations found within sub-grows will
be of a lower magnitude than those found for the total grow.

Methods 2;: achieving multiple prediction. All of the direct
predictions of success were computed using zero order correlations.

In addition to the simple correlations between the predictors and the
criteria, multiple predictions were also made which resulted in the
multiple correlation coefficient and the multiple regression equation.
The method used for determining the regression weights was that
described by Thorndike.3 This is an abbreviated Doolittle solution. The

standard partial regression weights were checked for accuracy by means

of the following equation:

B12r2n 3131311 Bligh-1...... 1‘11; 14

 

2J.P. Guilford, Fundamental Statistics in szcholog and Education
(New York: McGraw-Hill Book Company,l952, , p.220.

3Robert L. Thorndike, Personnel Selection (New York: John Wiley
and Sons, Inc. , l9h9), pp. 335-339.

bGuilford, 93. 22513,, p. 393.

83
where 1 represents the criterion; 2,3,h, etc., represent the predictor
variables; and "n" represents the last predictor variable. After this
was done, the "b" coefficients were computed along with the constant
"a", and finally the complete regression equation was written.

The sets of inter-correlations among the predictors used in all
the multiple regression analyses in this study were those derived from
the total group of boys and the total group of girls. Thus the sets of
inter-correlations found_g£tgr’the grpups had been selected by major,
were not used. Once selection has occured, the inter-correlations would
decrease in magnitude because of the greater degree of homogeneity; The
significance of this fact is great, since it means the interbcorrelations
among the variables are at their maximum because of the greater degree
of heterogeneity; As such, the magnitude of the multiple correlations
are necessarily less than they would be if the "selected" group inter-
correlations were utilized. — L

The reason for this decision is simply that prediction is being
attempted at the eighth grade level where the group of students are as
yet un-differentiated as to the selection of their high school majors.
To use the inter-correlations of the predictors derived.from.the
selected groups would presume the eighth grade population to be differ-
entiated already with respect to high school major which, of course, is
not the case.

By inspection of the products of the beta weights of each
predictor variable with its corresponding correlation with the criterion,
the two predictors of grade point averages with the highest.product

(and.lowest inter-correlations) were used together to predict the

8h
criterion. The multiple correlation for three variables was computed by

use of the following equation:

2 rf2+rf3-2r12 r13 r23 (5)

R

 

l - r53

The latter equation, based on the two best predictors of the
criterion, was written primarily on the grounds of practicality.
Counselors would find it indeed difficult to manipulate the longer
regression equation for each counselee especially when he is respon-
sible for helping a large number of students. To simplify further
prediction, a nomograph was constructed to represent .the regression and
provide the prediction of grade point average through simple reading of
the graph.

Multiple prediction of the alternate criterion, i.e. the
Essential High School Content Battery, was achieved by selecting the
two predictors with the highest correlation with the criterion and the
lowest inter-correlation. The method previously described for computing
multiple R for three variables was used. The resulting equations for
the prediction of boys' and girls' composite score on the EHSCB were
graphed along with the prediction of total grade point averages as
criteria of general academic success.

Methods 2; differentiatlgg' 33033. As stated previously, the
primary method of differentiating groups was on the basis of sex and
high school major. The major reasons for differentiating grows by sex
are first, boys and girls tend to elect certain course selections

 

5Ibid. , p. 393.

85
differentially. (Secondly, since validity is specific to the criterion,
and since the criteria may be'different between the sexes, it is neces-
sary to separate sex grows. Finally, it is likely the various abilities
needed for success would be different for boys and girls.6

The purpose of differentiating grows constituting subject areas
is to determine whether the abilities needed for success in one subject
area are different from those needed for success in another subject area.
In other words, can differential prediction be made? If there is no
differential prediction, i.e. the abilities needed for success are
similar in all subject areas, than it would be possible to achieve the
optimal prediction using those predictors which measure the abilities
common to all subject areas. If this were the case, it would not mean
that differential prediction among subject areas is not possible.
Certain required abilities may be common to all subject areas, but there
also may be other abilities not measured in this study which could
possibly differentiate subject area grows.

For purposes of comparison, it is necessary not only to
differentiate grows by sex and subject area, but it is also necessary
to differentiate the successful student from the unsuccessful student
within any given subject area. This would be done to determine quanti-
tatively which abilities the successful grow possessed to a greater
extent than the unsuccessful grow. Since there is no absolute measure

of succeSs in arm single course or grow of courses, its total

 

6Alexander Wesman, "Separation of Sex Grows in Test Reporting ,"
Journal 2; Educational P cholo , no: 228, April, 1919.

86
measurement is impossible. Arbitrary standards such as grades, however,
can be set w and "success" and "failure" must be interpreted in light
of the arbitrary standards which grades reflect.

Even within the realm of grades, what constitutes success or
failure? Does the achievement of an "A" represent success, or will a
"B" suffice? Here again, operational definition is necessary because
teachers' grades depend, among other things, on the range of talent
fowd in the classroom. For example, in a class of very capable pwils,
the average pwil (who may receive a grade of "0") would be swerior
to new members of another class whose over-all caliber may be lower.

Another standard for success could be defined simply as passing
a course. From this point of view, arw student achieving senior status
with the required number of majors must be a success in his majors;
that is, he must have passed certain courses in a subject area in order
to fulfill the requirements for a major. One could assume, therefore,
that the shrdents majoring in any given subject area possess the
abilities needed for success in that area to a greater extent than do
students not majoring in that area. This statement is predicated on the
assumption that there is a unique combination of- abilities needed for
success in each subject area. If this is found to be unwarranted,
insofar as the abilities measured by the (predictors are concerned, and
that rather, success is based on a general common factor or factors,
then one would suspect that students who possess a large amount of the
common factors could also be successful in a subject area they did not
choose to major in. Even if the assumption is found to be warranted,

one cannot know whether a student would have been successful in a major

87
had he selected it.

With these limitations in view, one method of determining differ-
ences is to compare the performance on each predictor variable for each
grow comprising a major with the remaining grow. Science majors, for
example, were compared with non-science majors. Statistical tests for
the significance of difference betwwn means have been made between the
two grows for each predictor variable.

It cannot be over emphasized, that the latter approach is pri-
marily one of description rather than prediction. Suppose, for example,
the science majors were found to have a significantly higher 1.62. than
the non-science majors. From this, it does not follow that to be suc-
cessful as a science major one must have a high I.Q. or to have a high
I.Q. means that one would be successful as a science major. Evidence of
this type is indirect and inferential, since, to reiterate, one does not
know whether the non-majors in a subject area would have been successful
had they majored in that area.

9_t_1_r_e_r_; statistical procedures used in mg 93. Three other
statistical procedures used bear mentioning. The first involves the
procedure for determining the significance of difference between
correlation coefficients. When this was done, the coefficients were
converted to their respective ”z" ftmctions. The standard error of the
difference between the "2's" was then computed and divided into the
difference of the two ”z'hfunctions.

The second procedure connnonly used was in averaging correlation
coefficients. This was done by first converting each correlation into

its amropriate "2" function, averaging the "2's" , and re-converting the

88
average "2" to the correlation coefficient.

The third technique used was an application of the well-known
Student's "t" test. This test was used for determining the significance
of difference between means. Since the groups on whom comparisons were
made were independent, no correction for correlated.means was needed.

Chapter V and VI will now present the analysis and interpretation

of the findings of this study.

CHAPTERV

THE DIRECT AND MULTIPLE PREDICTION OF GENERAL-
HIGH SCHOOL‘ACADEMIC SUCCESS

I. THE DIRECT PREDICTION OF TOTAL GRADE POINT AVERAGE AND THE

COMPOSITE SCORE OF THE ESSENTIAL HIGH SCHOOL CONTENT BATTERY

The prediction of general academic success using total grade
point average as the criterion is a highly complicated matter. The cri-
terion itself represents a combination of many varied courses taught by
many different teachers whose standards in grading more than likely lack
uniformity. As a matter of fact, even though teachers of different
subjects made a concerted effort to grade uniformily, it is doubtful
whether aw great degree of uniformity could be achieved. The varying
types of subject matter being dealt with, the varying objectives of the
courses, and the various abilities needed to achieve success contribute
to this lack of uniformity.

Even in the light of the aforementioned weaknesses of using
total grade point averages as criteria of general academic success, it
is an important factor for evaluation. Hediction of general academic
success may result in the early identification of potential drop-outs.

Mitchelll found, for example, that high school pwils scoring in the
middle fifth of the class on an LC. test administered upon entering

high school, have three times as mam chances of remaining in high

school until they finish as one who scores in the lower fifth; pupils

lClaude Mitchell, "Prognostic Value of Intelligence Tests ,"
Journal 2; Educational Research, 28:577-581, April, 1935.

90
in the upper fifth had twenty-one times as many chances of remaining
as those in the lower fifth.

High school administrators also use total grade point averages
for determining class rank, which in turn is a consideration used by
colleges and universities for entrance. The Essential High School
Content Battery composite score represents a more objective and uniform
criterion of academic success in the tool subjects.

Predictive information of these criteria may also prove useful
in the guidance process. If very low prognostication is made in a given
case, for example, a counselor may swport a student's view of leaving
school and entering into a vocation where his chances for success are
greater.

M prediction 2; 39.15.31... grade mil}: average. The degree of
relationship between each aptitude and achievement test predictor used
and the total grade point average is seen in Table 10. It will be
noticed that separate predictions are made for each sex. This type of
analysis will be used in all of the foregoing predictions.

Considering the complexity of the criterion, the relationships
shown in Table-10 are rather high, particularly between the Mathematics
Proficiency, Numerical Ability, and the Terman-McNemar predictors with
the criterion. For both boys and girls, these three tests correlated
highest with the total grade point average criterion. For boys, the
correlations ranged from .310 to .657 while for girls, the range was
between .167 and .716. For both boys and girls, the Mechanical Reasoning
Test provided the lowest correlation with the criterion while the

Terman-McNemar Intelligence Test provided the highest correlation with

CORRELATIONS BETWEEN THE PREDICTOR VARIABLES AND TOTAL
GRADE POINT AVERAGE, AND THE SIGNIFICANCE OF
DIFFERENCE OF THE CORRELATIONS BETWEEN
BOYS AND GIRLS

TABLE 10

91

 

 

Predictor Variable Boys Girls z
( =266) (N=322)

Verbal Reasoning ‘ .510 .596 1.1;?3

Numerical Ability .613 .686 1.).L61

Mechanical Reasoning .310 .1137 1.751

Spelling . A38 . S 82 2 . 2917*
Sentences . Shh . 588 . 809

Terman-McNemar Intelligence .582 .661 1.582

English Proficiency .539 .638 1.859

Mathematics Proficiency .657 .716 1.365

 

t6 Significant at the five per cent level of confidence.

92
the criteria.

Another important aspect of Table 10 is the consistently higher
relationships existing between the predictor test variables and the
criterion variable for girls than for boys. In no case are the correl-
ations higher for boys. In only one case, however, that of the Spelling
Test, is the correlation significantly higher than chance would allow.
In addition, it will be noticed that a ranking of these "validity"
coefficients for girls and boys results in a very similar ordering.
Since these data were arrived at independently, the latter fact lends
to an acceptance of the statistics as reliable measures of the
relationships.

In regard to the higher relationships among girls, a very
probable explanation lies in the fact that the boys represent a much
more homogeneous group than do the girls. Since correlation is a
function of group variability, the correlation would be expected to be
higher among girls than among boys. This is probably due, at least in
part, to the greater number of drop-outs among boys than among girls.
Since these drop-outs largely are among the boys of’lower academic
caliber,2 the remaining group is not only relatively homogeneous but
also represents the academically better students. These tendencies are
shown in Table 11, where the means and standard deviations together with
the significance of difference between the means for boys and girls, are
shown.

 

2Lee J. Cronbach. Educational ngchology (New'York: Harcourt,
Brace and Company, 195h5, P. 193.

TABLE 11

MEANS, STANDARD DEVIATIONS, AND THE SIGNIFICANCE OF DIFFERENCE

BETWEEN MEANS FOR THE TOTAL GRADE POINT AVERAGE AND THE
PREDICTOR VARIABLES BETWEEN BOYS AND GIRLS

93

 

 

Variable Sex N Mean t-ratio Standard
Deviation
Verbal Boys 266 16 .39 .67 7 . 5 3
Reasoning Girls 329 15.96 7.99
Numerical Boys 266 16.00 2.53% 6.66
Ability Girls 329 lb. 59 6 . 88
Mechanical Boys 266 35.38 l3.38** 11.59
Reasoning Girls 329 23.11 10.h9
Spelling Boys 266 25.hh h.3l** 21.03
Sentences Boys 266 19.9h h.21** 12.78
Girls 329 2h.h7 13.36
Terman-McNemar Boys 266 109.98 1.95 13.hh
Intelligence Girls 329 107.72 1h.75
English qu3 266 100.95 1.65 22.08
Proficiency Girls 329 10h.05 23.71
Mathematics Boys 266 58.hh 2.80** 17.00
Proficiency Girls 329 5h.h2 ., 17.85
{Petal Grade Boys 266 2.2h 2.53% .60
Point Average Girls 329 2 .37 . 65

k

* Significant at the five per cent level of confidence.
** Significant at the one per cent level of confidence.

9h
Inspection of Table 11 shows that with the exception of the

Mechanical Reasoning Test, which would be expected to be more variable
for boys than for girls by virtue of the nature of the test itself, the
girls tend to be a more variable grow.

The mean test scores of boys exceeds that of girls for the Verbal
Reasoning, Numerical Ability, Mechanical Reasoning, Terman-McNemar Intel-
ligence, and the Mathematics Proficiency Tests. The differences between
the Verbal Reasoning and Terman-McNemar Intelligence Tests, however,
were not statistically significant. The mean differences favoring the
girls were found on the Spelling, Sentences and English Proficiency
Tests. Of these, only the English Proficiency Test means proved to be
not significantly different from zero. In general, it seems the test
predictors in which boys score higher may be classified as quantitative
abilities while those predictors on which girls score higher are appar-
ently of a more linguistic nature.

It is interesting to note that although the Mathematics
Proficiency Test is the best single predictor of general academic
success, and although boys achieved higher on this variable than did the
girls, nevertheless the girls achieved a significantly higher mean total
grade point average than did the boys. This difference is seen to be
significant at the five per cent level of confidence. This discrepancy
suggests that grades are based not only won the nature of academic
abilities, but also upon other aspects of pwil behavior probably cen-
tering around that area often designated as behavior. The fact that boys
tend to be "problem" children more than girls may possibly be a factor

in their lower grades. Another explanation lies in the nature of the

95
criterion itself. Since the total grade point average represents an

average of all one-unit courses taken by the student in high school, it
obviously'becomes dependent upon‘ the type of curricula the students
enter. Thus, if boys enter into more demanding curricula than girls, this
could be reflected in the lower mean grade point average.

22.3353 prediction 3f the Essential §_i_g_h m Content Batteg
Criterion. The probable biases in grading practices suggested in the
latter section can be overcome through the use of an objective measure
of achievement, i.e. the composite score of the Essential High School
Content Battery (EHSCB). The differences between these two criteria
should not be overlooked. Whereas the total grade point average is an
over-all average of all one-unit courses taken in high school, the
composite score of the EHSCB represents an overball average score of
performance on each of its sub-tests, i.e. mathematics, science, social
studies and English. The total grade point average is, therefore, a much
more complex and inclusive composite of performance than the EHSCB
criterion. The latter is more a composite of the ”fundamental"or "tool"
academic subjects.

The correlations of the test predictors with this criterion are
shown in Table 12 along with the "2" ratio indicating the significance
of difference of the correlations between boys and girls.

Again it is seen without exception, that the correlations are
higher among girls than among boys. In only two instances, however, are
the differences significant, namely for the Spelling and the Terman
predictors. The best three predictors are seen to be the Terman, English

Proficiency, and Mathematics Proficiency Tests for both boys and girls.

TABLE 12

96

CORRELATIONS BETWEEN THE PREDICTOR VARIABLES AND THE COMPOSITE

SCORE OF THE ESSENTIAL HIGH SCHOOL CONTENT BATTERY TOGETHER
‘WITH THE SIGNIFICANCE OF DIFFERENCE OF THE CORRELATIONS
BETWEEN BOYS AND GIRIS

 

 

 

Predictor Variable Boys Girls z
@266) (N=329)

Verbal Reasoning .700 .7hl 1.002
Numerical Ability .6hh . 703 1. 328
Mechanical Reasoning .LSS .557 1.630
Spelling .562 .670 2.1h9*
Sentences .662 .687 .5h3
Terman-McNemar Intelligence .803 .858 2.17h*
English Proficiency' .732 .788 1.715
Mathematics Proficiency .709 .750 1.139

 

* Significant at the five per cent level of confidence.

97
It is interesting to note the fact that the best two predictors of

total grade point average (the Terman and the Mathematics Proficiency
Tests) are also two of the best three predictors of the EHSCB criterion.
In comparing the prediction of both criteria, it is seen that the
English Proficiency Test has replaced the Numerical Ability Test as

the third best predictor. Even in'view of this fact, the data suggest
that the EHSCB criterion is fairLy consistent with the total grade point
average criterion. This idea is partially'borne out by the correlations
between these two criteria; for boys, r=.7th.028 and for girls
r=.7653.023. Considering the differences in these two criteria, the
relationship is remarkably high.

For boys, the correlations range from .hSS to .803, while for
girls the correlations range from .557 to .858. For both boys and girls,
the highest correlation with the criterion was provided by the Terman-
McNemar Test and the lowest correlation was provided by the Mechanical
Reasoning Test. On the whole, it seems that the Differential.Aptitude
Test predictors are relatively better predictors of the EHSCB criterion
than they are with the grade point average criterion.

The relationships between the test predictors and the EHSCB
criterion are seen to be higher than those for the total grade point
average. This probably is due to the greater objectivity and reliability
of a test score over a grade point average. It also is indicated that a
test is more likely to correlate highly with another test by virtue of
their common limitations of sampling performance than it is with
teachers' ratings as reflected in school marks.

The mean composite score of the EHSCB criterion for boys and girls

98
was 125.h6, respectively; The mean difference produced a
t-ratio of 5.81, indicating a significant difference at below the one
per cent level of confidence. Noting that the mean total grade point
average for boys (2.2h) was significantly lower than that for girls
(2.37), a discrepancy is again apparent between the two criteria. This
may be explained.partially, however, by the fact that half of the EHSCB
criterion consists of the science and mathematics sub-tests. Since'bqys
tend to select these subject areas more than do girls, it is to be
expected that they would achieve higher on this criterion.
II. THE MULTIPLE PREDICTION OF TOTAL GRADE POINT
AVERAGE AND THE COMPOSITE SCORE OF THE EHSCB

In an attempt to determine the highest degree of prediction of
total grade point averages with the test predictors used, the methods of
multiple correlation and regression are utilized. The tables of inter-
correlations for boys and girls, necessary for this type of analysis, are
seen in.Table 13. As mentioned in CHAPTER IV - METHODOLOGY, the procedure
used for determining the standard partial regression coefficients was
the abbreviated.Doolittle method. The Doolittle solutions of the beta
coefficients are recorded in Appendix A. Table 1h shows the solutions of
the regression coefficients for boys and girls. The meaning of the

symbols heading each column in Table lh is as follows:

Column Meaning
k indicates each predictor variable.
Blk the beta weights for each k variable, 1 representing the

dependent criterion variable

rlk the correlation coefficient between each predictor
variable k and the criterion 1.

99

 

 

 

40m. avamfiofimonm nmflawqm
mem. 9mm. mestozlqmeeme
04m. wmo. mmo. mmoCmpnmm
4mm. mmo. mmm. 3H0. mafiHHmmm
own. mam. Hon. mmm. mod. wnflnommmm Hmownmnemz
Haw. mam. mzm. mmm. Hma. mam. hpflaflnd HmOHAmEdz
3mm. mmw. New. Hep. mwm. meg. mmm. mnflcommmm Hmnao>
mNom
Nmm. .hocmfiOHMOhm nmfldwcm
wmw. mmw. emEoZoEIQMEAmB
Nmo. so». new. moocopqcm
Sm . ewe . mac . 30 . mfieaoem
m3. 8:. ram. NE. E. mficcooom H3228:
mm». 80. m3. mam. Jam. in. 323a Hoofloeez
So. 3.. wow. amt. cam. 3m. 20. mfieeooom Horace
mHhfiU
.MOHm .monm HmEmZOZ. .mMmm .Hfl£¢ mmdnmwhmb
.npmz SmHHwnm lamenme .9Gmm .HHQO .nomz .Edz HOQOfiUmhm puma

 

 

mmommm>< 82Hom mm¢mu

H4908 mo mHmMH<z< ZOHmmmmwmm mgmHBADE Ema 2H
ammo mmqm¢Hm¢> MOBUHQHEM mme UZOZ¢ mZOHH¢AHMMOOLmMBZH

mH mqm4H

 

 

 

 

QNOJo U m
005m.m nae mueoe.
IE I NMflmmmmc
mmom. . m4.:m eoHo. mono. Noam. 0He. omem. .eoom .eooz
memo. 00.:0H mooo.- memo. 0e00.- omo. mmao.u .eeom seafloor
mega. .. 2..on mwoo. mmqo. mmmd. How. :08. SEmZOvHIcmseme
momo. . no.0m mmoo. moeo. mewo. 00m. zero. moocooeom
oome. . mm.mm emoo. momo. N000. «am. Home. mcaeeoom
030. .. Sumo. mooo. 030. mmoo. NS. 2.00. .moom .rooz
moan. . mm.ee memo. memo. omme. 000. 000m. aoaeaoa .202
$40. 3.3” mmoo... mmmo. OHNO... 0mm. mmmo... .mmmm HwnamS
memo
owmm. n m
2.30 u 5 TS».
bet“ a . mammal
meos . a: 0m omeo. ammo. ammo. emo. Hoam. .eoom .eooz
meme. me.ooe ma00.- memo. mmmo.- emm. 4500.- .oeom eoaawcm
0mmo.a- 0m.00e emoo. Nero. Nome. mom. ooHN. oeso2o2.coeooe
05H. .. 20.3 $8. 0.50. weoo. 3m. Ema. mooeoeeom
mono. . 4a.mm maoo. 0000. mmmo. one. somo. mcaeaoom
memo. om.mm 0000.- memo. mmoo.- oam. 00H0.- .ooom .eooa
05mm. . 00.0H Hoeo. Nomo. meoe. meo. cope. aeaeaoa .eez
ooeo. . mm.oH Heoo. meeo. N500. mam. omeo. .ooom Honoo>
mNom
oHpoeaee
in, 321v x: can. cadmium fine cam .38.“me

 

 

mafia 92¢. mMom mo goggdw BzHOnH ago H4908 mo
ZOHBOHQME BE mom mazmHonmmoo ZOHmmmmomm BE. mo ZOHHDAom

JH BBQ.

 

 

lOl
Blkrlk the product of the same two columns.

the ratio of the standard deviation of the criterion

1/ k to each of the predictor variable standard deviations.

blk the "b" coefficients of each variable (the product of
the Elk and 1 /1< columns).

Mk the means of each k predictor variable.

(“Mk)blk the product of the (-) Mk and blk columns.

The multiple correlation coefficients are seen in the last row of
each table. For boys and girls, these multiple correlation coefficients
are seen to be .723 and .767, respectively. The standard errors of these
correlations are .030 and .023 respectively. It is seen that the
degree of prediction for girls is higher than for boys, although this
difference did not prove to be statistically significant (z=.88). Both
multiple correlations represent a rather high degree of prediction of the
total grade point average. From Table 1h, the final regression equations
for predicting total grade point average may be written. For boys, the
equation is as follows: total grade point average (X31)=.0011X2+.0161X3

+.0006xh+.0015x5+.0023x6+.0088x7+.0003x8+.0107x9+.14078. (Standard error

.of multiple estimate t.hlh.) For girls, the equation is: total grade point
average (xGl)=-.0029x2+.0213x3+.ooosxh+.0039x5+.0023x6+.0088x7-.0003x8
+.0107X9+.h078. (Standard error of multiple estimate t.hl6). In the above
equations, the meaning of the X subscripts refer to the test predictor
variables in the same order as shown in Table lh.
hath reference to Table 1h, it will be noted that the Terman-
McNemar Intelligence Test and the Mathematics Proficiency Test account
:for'3h.56 per cent of the total 52.35 per cent of the variance accounted

for through multiple regression.eBy using the equation for computing

102
multiple R with three variables, the following correlations with the
criterion were obtained for boys and girls respectively;ZR=.703i.031,
and .7h2t.025. It might be noted at this time that although the Numer-
ical Ability Test for girls had a higher weighting in the total
regression equation than did the TermanéMcNemar Test, its correlation with
the Mathematics Proficiency Test proved to be too high t0 make those two
tests the best predictive combination.

By comparing the two multiple correlations for boys, it will be
noted that the multiple R based on all eight variables (.723) accounts
for only 2.85 per cent more variance than does the simple combination
of the Terman-McNemar and Mathematics Proficiency Tests (.703). For
girls, the difference between .767 and .7h2 indicates that the former
coefficient accounts for only 3.70 per cent more variance than the
latter. When tested for significance of difference, neither pair of
multiple correlations proved to be statistically different.

It is obvious than, that it is more economical to use the
regression equation based on the two predictor variables mentioned. By
solving for the regression coefficients of these two variables in.pre-
dicting total grade point average, the resulting regression equation was
graphically represented by use of a nomograph. These nomographs for boys
and girls are shown as Figures 1 and 2 in Appendix B. By connecting the
two appropriate points found of the predictor test scales, the point at
'which the line crosses the middle axis indicates the best prediction of
'total grade point average.

For purposes of predicting the composite score of the EHSCB on

1319 same nomograph scales as those used above, the regression equations

103
using the Terman-McNemar and.Mathematics Proficiency Test toward pre-
dicting the EHSCB criterion were written and graphed accordingly;

The multiple correlations of these two test predictors with the
EHSCB composite score were $583-$016 and .875*.013 for boys and girls,
respectively. This degree of relationship is amazingly high, even though
it is based on the correlation between test scores. In fact it represents
a degree of relationship which often cannot be achieved in some test-
retest reliabilities. The differences between these two correlations

did not prove to be statistically significant.

CHAPTER VI

THE PREDICTION OF ACADEMIC SUCCESS IN EIGHT
HIGH SCHOOL SUBJECT AREAS

The prediction of achievement in each of the subject areas repre-
sents, primarily, an attempt to provide educational information necessary
to aid in the selection of high school majors which are appropriate to
the abilities of the student. Information of this type also can be used
to section students in various classes if this is consistent with the
school's philosophy. In addition, it could be used to identify potential
failures which may, for example, result in recommendations for attendance
in special classes of a remedial nature. This would be particularly the
case in areas such as English and mathematics. Subjects such as these,
which frequently are required for graduation, must be pursued by a
student even though he may not have the motivation and/or the ability
necessary to succeed in them.

I. THE DIRECT PREDICTION OF SUBJECT AREA GRADE POINT

AVERAGES AND THE ESSENTIAL HIGH SCHOOL CONTENT
BATTERY CRITERIA

Direct prediction'gf grade point averages igieight subject areas.
It should be remembered that the grade point averages used as the criteria

for success, represent average marks established in a subject area in
‘which the student has taken three or more units of credit. Therefore,
the ensuing correlations represent prediction over at least a three year
Period of time since all of the predictors were administered prior to
the students' beginning his high school work.

105

Direct correlations of each test predictor variable with each
high school subject area, within which students have selected majors,
are seen in Table 15. Since all students are required to obtain three
units of credit in English, all students in the study have majors in this
area. The next most selected subject area from which majors were sel-
ected is social studies with 6h.2% of the boys and 63.5% of the girls
selecting this area for majors. There is little doubt that at least
one reason for this rather high.percentage is the fact that all students
are required to obtain a minor, i.e. two units of credit, in social
studies as a graduation requirement. Thus, with the addition of only
one more subject in the area, the student can complete a major (three
of which, including English, are required for graduation).

The next most selected subject area for girls is business edu-
cation (56.8%), followed by science (h1.5%), mathematics* (21.5%),
foreign language (18.8%), and home economics (lh.8%).

The next most selected subject area for boys is mathematics*
(67.2%) followed by science (65.0%), industrial arts (27.8%), and foreign
language (12.h%).

The most striking aspect of Table 15 is the consistently higher -
relationships between the test predictors and grade point averages among
girls than among bays. This phenomenon was also noted in the prediction

of total grade point averages. The only exception to this, noted in

 

* These percentages are based on 71 girls and 179 boys who actualry
majored in mathematics in the three schools studied, Due to the necessity
for ommitting the mathematics students from one school (because of its
lack of uniformity with the other two schools) the figures in Table 15
indicate the numbers of mathematics majors in the two remaining schools.

106

.oocmoflcho mo Hobma pace Hog mbflm one pm each Scam pcmammwfle hapnmoawwawflm poz *

 

 

H~0. 004. 004. H~m. 004. 000. 000. 000. ~0H nHaHo .00 .nsm
004. 000. HHO. 0H4. s0Hm. s~0H. 004. 000. 04 aHaHo .0000 0200
40m. 000. ~00. ~00. mom. *40H. ammo. 00m. 4~ 0000 none .naocH
000. ~H0. 000. 004. ~00. .0Hm. 000. H04. 00 nHaHo owosmcsH
HH4. 04o.- ammo.- *0Ho. ammo. s0Hm. 004. *H00. 00 seem smHoaee
H~0. ~H0. 000. 000. 400. 000. 400. 000. H0 eHnH0 aaHese
0H0. 004. 004. 004. H04. H~N. 004. 0~m. ~oH seem -oeesa
000. 000. H00. 400. 000. 004. N00. o~0. ~0H anHo

~00. 000. H40. H00. 000. 004. ~00. 004. 0~H seem oecoHom
mm0. 000. 000. H00. 400. 000. 400. 400. 000 nHoHo noHeepm
000. 000. 000. H04. H00. 00H. 004. 004. H~H seem HaHoom
000. ~40. ~00. 400. o~0. HH4. H00. 000. 0mm thH0

000. 4H0. 000. 004. 044. 00H. 004. 004. 00m 0000 soHHmsm
.Heai .Hesm .0.H .acom .HHoao .ooom .HHha .nsom 2 saw sons
.000: .000 cheese .000: .eaz .0aoe econham

 

 

 

 

 

 

i

 

 

 

Nam 92¢ m<mm< mmaadzleomwmbm mmH ho mo¢m 2H
mmo<mm>¢ 92H0m mm<mu 92¢ mmqm<Hm¢> moeonumm mma zmmgemm mZOHH<Ammmoo

mH mqmda

10?
those subject areas in which both boys and girls have majored, is in the
case of the Mechanical Reasoning Test as it relates to science grades.
In this case only is the degree of relation higher for boys than for
girls. The difference, however, is not statistically significant as
seen in Table 16. In the latter table, the correlations found in Table
15 were tested for significance of difference for each test predictor
and subject area in which comparisons could be made between boys and
girls.

Table 16 indicates that the test predictors in general, show the
greatest amount of differential prediction, due to sex, in the subject
areas of English and foreign language. In the case of English, only
two of the eight test predictors (Verbal Reasoning and Spelling) failed
to show a significant difference between the sexes.

In the area of foreign language, the Spelling, Sentences, Terman-
McNemar, and English Proficiency Tests showed significant differential
prediction. It is interesting to note that these predictors are all
measures of verbal or linguistic facilities. By referring back to Table
15, it will be seen that these same tests failed to prove significantly
different from zero correlation with the foreign language achievement
boys. For girls, the case is quite the contrary. In fact, all’of the
predictor tests show a substantial correlation with foreign language
achievement among girls with the exception, perhaps, of the Mechanical
Reasoning Test.

In only two other cases, namely, the Mechanical Reasoning Test
in social studies and the Spelling Test in the science area, did the

differences in the correlations between boys and girls prove to be

108

.mQCmUHmcoo mo Hm>mH pace Hem mco one we commemmmflp Mo mocmOHmemHm mmmeHocH **
.momoofimmoo mo HmbmH pace Hog o>flm one no moamHmMMHb mo cosmoHMHame mmemowocH *

 

 

,. . .. mmmpmqmq
4H. ss0~.m **00.m *~m.m ssHN.0 00. 00. 0~.H. sMHoaee
00. H0.H 04.H 00.H -.H 00.H 00.H 00.H noHeasosooz
o0. 00.H 04.H Hm. a~m.m Hm. 00. 00. oeaoHom

. noaeapm

o4.H m0. 00.H H0. m~.H ss00.0 00. m0. HsHoom
$40.0 #000 ism-HA #0410 00.H $005 $00.0 00.H enHHwam
.moam .monm .O.H .mmmm - .HHQ< .mmmm coed
.aeoz .000 sashes .ocom .HHoam .aooz .ssz .0aoe econham

 

ammowmz abdm mmxmm mHom mOHm3.zH mmmd
Homwmbm mo<m 2H mQMHU Qz¢ mwom mom mZOHa<Ammmoo
ZMMBBmm mozmmmmmHm mo moz¢0HmquHm mmH UZHB¢UHQZH mOHeamIN

0H mqmda

109
statistically significant.

Referring to Table 16 as a whole, twelve out of the forty com-
parisons proved to be statistically significantly different. It is
not to be denied, however, that although these differences did not
prove to be significant in many cases, there is obviously a tendency
for girls to be a more predictable group with the type of tests and
criteria used in this study. In this, sheer numbers and cumulative
probabilities reflect a genuine difference in predictability rather
than single tests of statistical significance taken alone.

.A combination of two factors may help to explain the phenomenon.
The first involves the greater restriction of range (seen in Table 11
of CHAPTER V) among boys than among girls. A plausible explanation for
the greater restriction of range was put forth in that chapter where it
was stated that a higher percentage of boys than girls drop out of school;
these drop-outs representing, on the whole, students of lower academic
ability. Thus, the boys remaining in the twelfth grade represent a more
homogeneous and able group than do the girls as a group. The greater
restriction of range would,of course, discourage high correlation.

A second factor possibly operating to cause this phenomenon is
the probable greater degree of objectivity excercised in marking girls.
Due to their greater conformance and compliance, it would be expected
that fewer extraneous variables would be allowed to influence the marks
given. Boys, on the other hand, would more likely be graded down, not
necessarily because of poorer achievement but because of factors such as
poor attitude, aggressiveness, non-conformance in the school situation,

etc. If this were the case, it would lend toward lower measures of

110
relationship between grades and the test predictors.

To summarize the best predictors of each subject area, Table 17
is presented. An examination of this table will reveal that in nine out
of thirteen cases, the Mathematics Proficiency Test is the best single
predictor of grade point averages. This instrument is one of the best
three predictors in every instance except in the case of predicting
home economics grade point averages. The Terman-McNemar, Numerical
Ability, Spelling and English Proficiency Tests proved to be the best
single predictors in the subject areas of social studies (girls), foreign
language (boys and girls), and home economics (girls), respectively.

It seems peculiar that a Mathematics achievement test should be
among the best predictors in areas such as English, social studies,
and foreign language, since in terms of course content at least, this
type of ability would be seemingly zmnecessary. Of the forty best
predictors shown in Table 17, the Mathematics Proficiency Test appears
twelve times; the. Terman-McNamar, ten times; the Numerical Ability,
seven times; the English Proficiency Test, six times; the Spelling Test,
three times; and the Sentences and Mechanical Reasoning Tests, one time
each.

The subject areas open to the highest prediction are English,
social studies, science and mathematics, or what are commonly called the
tool subjects. Every correlation presented in these subject: areas is
significantly different from zero. This is in part due to the larger
numbers of students selecting these areas for majors. 0f the forty
coefficients presented for the remaining subject areas, i.e. industrial

arts, foreim language, home economics and business education, ten are

111

 

 

uncapmoswa
“004.0 .0.H sashes A000.0 00HHH0<_.sez AH~m.V .Hesm .aooz anHo anoaHnsm
Abwq.v thHHbm .Edz AHHm.v .G.H smegma Amwm.v .momm nmflawcm mHHHu moHEocoom msom
.G.H sesame
A~om.0 ens oaoaoeaom Ammm.0 MWcHHHoem A40m.v .00a0 .0002 0000 mesa HoHaoneoaH
A000.0 .meam .sooz A000.0 soHHHha .ssz A~m0.0 0cHHHoam nHaHo swarmsoH
AOHN.V .noom .sooz AHH4.0 .eesm .aaoa A004.V HoHHHha .ssz seem cmHoaee
A400.0 msHHHoom A400.v 0eHHHh< .eaz AH~0.V .Henm .aooz oHsHo
A004.0 .0.H sashes A004.0 eoHHHha .saz AOH0.0 .monm .aeoa seem ooHaoEossoz
A000.0 .eesm anHchm AH00.V .0.H coerce H000.0 .00h0 .eosz nHaHo
H~m0.v 00HHH0< .saz AH40.V .0.H coeaoe A~m0.v .Hesm .000: seem oocoHem
A000.V .Hoai eoHHmcm A000.0 .0.H cheese A000.0 .oenm .0002 anHo
A000.v .mesm anHHocm A000.V .oesm .heoa A000.0 .0.H cheese 0000 ooHesem HoHoem
A~40.V .mesm anHHwam A~00.0 .0.H sashes A000.0 .0000 .heoz anHo
A4H0.v .eenm anHmcm A000.0 .0.H sashes A000.0 .Hoam .aooa seem anHHmam
oem paw pea xmm woes podfiflSm

meopoflemem mouse poem

 

 

xuw mo<m 92¢ «mm< Homwmbm $04M 2H mmwmmm><
BZHOA madmw mo mmoeonmmm swam mmmme Ema mo Mm<EZDm <

NH mqmda

112
not significantly different from zero. Six of these ten occurring in
the prediction of foreign language achievement of boys.

The greatest discrepancy in the prediction of subject areas
between boys and girls lies in the foreign language area. The correla-
tions shown in this area for boys are, on a whole, negligible, while for
girls, the coefficients are very much greater. The two best predictors
of foreign language success for boys and girls, however, (the Mathematics
Proficiency and the Numerical Ability Tests) are the same for both sexes.

For purposes of comparing the results of other investigations as
reported in the validity section of the DAT manual, Table 18 is pre-
sented. The average correlations shown have no intrinsic use and in some
cases may be . Misleading in terms of judging the battery. Since these
averages are based on those sub-tests which are poor predictors as well
as good predictors, it is not expected that the average correlations for
each subject area would be high. On the contrary, the basic premise upon
which this battery was built was the idea that different courses and
subject areas would require different abilities for success. Table 18
is shown for comparative purposes. The procedure used in averaging the
correlation coefficients is described in CHAPTER IV.

Table 18 reveals that on a whole, the average correlations of the
Differential Aptitude Tests used in this study, using grade point average
criteria, are higher than those derived from the manual. The coeffi-
cients reported in the manual also reflect slightly higher correlations
for girls than for boys but apparently not as different as those found
in this study. It is also seen that the correlations reported in the DAT
manual tend to be higher in predicting achievement in the tool subjects

TABLE 18

113

A COMPARISON OF THE AVERAGE DATr VALIDITY COEFFICIENTS
ESTABLISHED IN THIS STUDY AND THOSE DERIVED FROM THE
DATA PRESENTED IN THE DAT MANUAL, BY SEX, AND
SUBJECT AREA, FOR THE DAT PREDICTORS USED

 

 

 

 

Average r found

Average r computed
from data in DAT

 

Subject Area Sex in this study_ manual**
English Boys .th .h30
Girls .565 .h50
Social Studies Boys .h05 .390
Girls .510 .h30
Science Boys .h60 .h55
Girls .525 .h65
Mathematics Boys .hOO .320
Girls .585 .390
Foreign Language Boys .170 .350
Girls .h95 .290
Industrial Arts Boys .260 .215
Home Economics Girls .335 .395
Business Education Girls .h25 .310

 

* These averages are based on the Verbal Reasoning, Numer-
ical Ability, Mechanical Reasoning, Spelling and Sentences subtests.

** These average r's are based on either the median r's
found in the DAT manual for each.DAT subtest used or on personally
computed medians in areas in which medianS'were not reported.

111-
than in the remaining areas. This is consistent with the findings of
this study.

Eire-gt prediction p_f_ the Essential High M Content Batten
criteria. As was the case in the prediction of general academic success,
the Essential High School Content Battery (EHSCB) will now be used as
an alternate criterion of success. As mentioned in an earlier chapter,
this battery was administered'to the current senior class during the
month of May in their junior year. Since the predictors were adminis-
tered at the end of their eighth grade, prediction is thus established
over a three year period of time for this criterion.

Because the EHSCB measures achievement in only four subject areas,
i.e. Mglish, mathematics, social studies, and science, it could be used
for comparative purposes and as an alternate criterion in these four
areas only. The direct zero order correlations of the test predictors
with each of the EHSCB tests are shown in Table 19.

In general, the relationships shown in Table 19 are impressively
high especially when compared to the results of similar reported inves-
tigations. As was the case in the prediction of grade point averages, the
prediction of girls achievement is consistently higher than for boys.
Another notable aspect of this table is the consistently higher pre-
diction of achievement using the EHSCB criteria than was found when
grade point averages were used as the criteria of success. It is to be
mcpocted, however, that test scores are more likely to correlate higher
with other test scores than they are with teachers' ratings of student
performance as reflected by grades. The increased prediction} of the

EHSCB criteria over the grade point average criteria in the same subject

115

 

 

000. 0H0. ~00. 000. 040. 000. 000. 000. H0 oHAHo aoHooe
0~0. 004. ~00. 004. 400. 004. 0H0. 004. ~0H ream -oepoa
0~0. 400. 00~. 400. 000. H~0. 000. ~00. ~0H thHo

H40. 000. 00~. H~0 . N04 . H04. 004. ~40. m~H 000m 8:38
400. 000. 04~. 4~0. N00. 004. ~00. 040. 000 thHo aoHosem
004. 0H0. 4~0. 004. 404. -H. 404. 0~0. H~H seem HoHoem
~H~. 0H0. 0H0. 000. 0H~. 004. 0~0. 000. 000 thH0

000. 00~. ~0~. 000. 000. 000. H00. 000. 000 0000 shHHmsm
.0000 .0000 .0.H .eaom .HHoao .noom .HHhH .noom 2 wow enoehem
.0002 .000 sesame .0002 .002 .hao> 00000

 

 

go 92-0. mwom mom Nmmaedm
.Hzmhzoo Hoomom mon Hfiezmmmm mule mo magmam mbom
BE. 94d. ngdem-g mOHoHQm—mnw HE. szmm WZOHefim-mmco

m4” Emma-H.

 

 

116
areas is shown in Table 20.
For purposes of comparison, the eight correlations between the
predictors and the appropriate criteria, by subject area and sex , have

been averaged.

TABLE 20
A COMPARISON OF THE AVERAGE CORRELATIONS BETWEEV THE

PREDICTORS AND THE EISCB TESTS AND GRADE POINT
AVERAGES BY SEX AND SUBJECT AREA

M

 

Criteria
Subject Area Sex Grade Point Average EHSCB Tests
English Boys .1455 .655
Girls .610 .710
Social Studies Boys .1450 .505
Girls .555 .610
30161109 Boys .500 0580
Girls .570 .660
Mathematics Boys .I-35 .520
Girls .605 .625

 

Table 20 reveals that in every instance, for both boys and girls,
the average correlations between the test predictors and the EHSCB
criteria are higher than between the test predictors and the grade point
average criteria. A comparison of the average correlation coefficients
between boys and girls on the EHSCB criteria confirms the tendency shown
in a similar comparison with grade point average criteria, i.e. the pre-
dictive relationships among girls is higher than among boys. This fact
would lend support to the idea that girls are to some extent at least,

more consistent in their performance on different tests and in terms of

117
their test-grade point average relationship, than are boys. Thus, the
idea that boys may be graded more inconsistently than girls does not
answer the whole question since even on an objective achievement test,
the relationships among girls are higher than for boys. This is
certainly a fact which should be considered in future prediction studies.

Although the correlations with the EHSCB criteria are higher, it
is interesting to notice that a ranking of the subject areas in order of
their predictability is similar with.both criteria. Thus, English is,
in general, the area open to the highest prediction, followed.by science,
mathematics, and social studies.

Table 21 summarizes the best three predictors in each subject
area of the EHSCB. This table shows the TermanéMcNemar Intelligence Test
to be one of the best predictors in each of the subject areas for both
boys and girls. The English.Proficiency Test seems to be the second best
all-around.predictor, appearing six times in Table 21, followed by the
Verbal Reasoning Test appearing five times, the Mathematics Proficiency
Test and.the Numerical Ability Tests each.appearing two times and the
Spelling Test appearing one time.

In comparing Table 21 with.Tab1e 17, one finds a striking differ-
ence in the type of predictors which are most effective. Fbr example,
in comparing these four subject areas only, if the predictors were
dichotomized into a "number" or quantitative group and.another group
measuring "verbal" or linguistic abilities, it would.be found that for
the prediction of grade point averages the "number" tests, i.e. the
Mathematics Proficienqy Test and the Numerical Ability Test, appear in

nine out of twentybfour cases. In the prediction of the EHSCB test

118

 

 

Aowo.v epaaapa..esz Aeme.v .o.H sesame Acme.v .moam .eamz mates

Aemm.v .o.H sesame Amae.v epaaaba .eez Amee.v .eoam .epwz seem moapwEoepmz

Aemo.v .eonm emaamem Aeme.v .mmom flanges Amme.v .o.H sesame means

Aeso.v .mwmm Hashes Ammo.v .eoem emaemem Acme.v .o.H sesame seem mommaom

Acne.v .mmmm Hmnam> Ammo.v .eonm emaamem Amae.v .o.H sesame sagas

Amem.v .memm flanges Amae.v .eoam emaHmem Aseo.v .o.H sesame whom mmaespm Hmaoom

Amee.v weaaamdm Aoaw.v .eoam emaamem Amam.v.o.H sesame maaau

Ammo.v .mmmm flanges Abbe.v .eoam emaawem Aeoe.v .o.H sesame msem emaamem
mam new pmfi xmm mane pomnnsm

whopowvmam manna pmmm

 

 

mam w,mz¢ mwom mom HMMBB<m Bzmezoo Aoomom
mon A¢Hezmmmm mme mo mmoeonmmm Bmwm mmmme Mme mo Hmdzzsm <

Hm mqm<a

119
scores, the "number" tests appear in only four of the twenty-four cases;
the remaining predictor tests being in the "verbal" group. In general
the Verbal Reasoning and.the English.Proficiency Tests, especially the
former, have taken the place of the Mathematics Proficiency and the
Numerical Ability Tests as the best three predictors.

Verbal (primarily reading) skills are to a large extent, the
factors being measured.by the typical paper and.pencil achievement test
regardless of whether the test content is English, social studies, science
etc. By its very nature, however, the Mathematics Proficiency test
requires much.less verbal ability for achievement.

A fundamental difference between the two criteria used has been
brought to light. On the one hand, the EHSCB criterion, measuring verbal
skills and.abilities to a large extent, finds as its best predictors
those instruments measuring these very abilities. 0n the other hand,
grade point averages, apparently based to a larger extent upon quanti-
tative abilities, finds as its best predictors, those tests measuring
number abilities. The TermannMcNemar Test stands as one of the best pre-
dictors for both of the criteria. Since verbal abilities are inherent
in both criteria, and since the Terman-McNemar Test is a verbal intelli-
gence test, it probably reflects a general intelligence factor permeating
all achievement.

The variance common to both criteria is shown in the correlations
presented in Table 22.

The coefficients shown in Table 22 express moderately high rela-
tionships between the EHSCB criteria and the grade point average

criteria. The fact that the correlations are higher for girls than for

120
boys supports a previous statement that girls are graded more consis-
tently with their achievement than are boys, however, the data are

somewhat conflicting.

TABLE 22

CORRELATIONS BETWEEN THE EHSCB CRITERIA AND GRADE POINT
AVERAGES IN CORRESPONDING SUBJECT AREAS, BY SEX

W

Variable Boys Girls
N r N r

 

EHSCB English.Test vs.
English Grade Point Averages 266 .613 329 .73h

EHSCB Social Studies Test vs.
Social Studies Grade Point Averages 171 .677 209 .722

EHSCB Science Test vs.
Science Grade Point Averages 173 .589 137 .678

EHSCB Mathematics Test vs.
Mathematics Grade Point Averages 107 .736 51 .7h9

 

It is seen that the highest relationship between the two criteria
lies in the subject area of mathematics. This is probably because of
this area being more factorially pure than the other subject areas.

The lowest relationship between the criteria is in the social
studies area. Considering the differences between the two criteria,
however, all the relationships shown in Table 22 are rather high ranging
from .589 to .7h9o

In order to secure higher predictions, the methods of‘multiple
correlation and regression will now be applied to the prediction of the

criteria.

121

II. MULTIPLE PREDICTION OF GRADE POINT AVERAGES AND THE
ESSENTIAL HIGH SCHOOL CONTENT BATTERY CRITERIA

Multigle prediction g_f_‘_ grade point averages _i_n_ eight high 5%
subject‘ggggg. In an attempt to achieve the highest degree of prediction
of grade point averages in each subject area investigated, and through
the use of the independent variables used in this study, the methods of
nmdtiple correlation and regression are employed. In order to achieve
this purpose, essentially four steps are involved. The first step is to
secure the inter-correlations among the predictor tests used and the cri-
teria. The inter-correlations among the predictor tests were those com-
puted from.an analysis of the total group of boys (266) and the total
group of girls (329). The rationale for using the same inter-correlations
for each of the sub-groups studied is stated in CHAPTER IV.

The second step is the solution of the normal equations leading
to the appropriate weightings for each of the independent variables.
These weightings (beta coefficients) were solved by using the abbreviated
Doolittle method referred to in CHAPTER IV. These solutions, by subject
area and sex, are found in Appendix A.

The third step involves the solutions of the regression
coefficients. From this step, the beta coefficients which are the
weightings of the independent variables in standard score form are con-
verted to "b" coefficients which are final weightings in raw score form.
This step also includes the solution of the constant "a", and also the
computation of the multiple correlation coefficient. ‘ I

The fourth step is the writing of the multiple regression

equation based on the solution of the regression coefficients.

122

These steps will now be related in successive order. For refer-
ence to steps one and two, the reader's attention may be called to Table
13, CHAPTER 5 and to Appendix A, respectively.

The solutions to the regression coefficients, by subject area and
sex, are found in Tables )43 through 50 of Appendix B. The multiple cor-
relation coefficient appears at the bottom of the third colunm in the
latter tables. The meaning of the column headings are similar to those
described in CHAPTER V.

Through a cursory inspecton of Tables h3 through 50, one is
immediately impressed with the appearance of numerous negative beta
coefficients. Under certain circumstances, negative beta weightings may
be interpreted as the variables acting as suppressor tests. This
situation arises when the variable given the negative beta weight has a
low (and often positive) correlation with the criterion and a high
correlation with the test for which it is acting as a suppressor.lIn
the previous analyses, all the predictor variables correlate to a
moderate extent with the criteria except in the instance of foreign
language achievement of boys. In this case, the three predictors given
negative beta weights are Sentences, Terman-McNemar, and the English
Proficiency Test. The tests showing the highest correlation with the
criterion are the Mathematics Proficiency and the Numerical Ability
Tests. Since none of the negatively weighted predictors show a high
relation with the mathematics tests, it is unlikely they are correlating

to arm great degree with non-valid variance in the mathematics tests.

 

lThorndike, 22. gig” p. 192.

123

Since each of the predictor variables studied is a paper and
pencil achievement test (in the broad sense of the term), it is likely
that most of the negative beta coefficients can be explained by sampling
fluctuations of the correlation coefficients. Due to the great changes
in the values of the beta coefficients through sampling fluctuations of
the correlations, it is difficult to interpret the negative beta coeffi-
cients as being significantly indicative of suppression.2 This is a
problem which requires further analysis but is beyond the scape of this
investigation.

A summary of the multiple correlations derived from Tables 143
through 50 is contained in Table 23 along with their computed standard
errors of multiple correlation. All of these correlations proved to be
significantly different from zero at the one percent level of confidence.
It will be noticed that the subject area from which the highest multiple
correlation for boys and girls was derived is foreign language. This is
interesting particulary since foreign language majors were found to be
one of the most selected and restricted subject area groups. It should
be pointed out that the size of the beta coefficients is not only a
function of the correlation with the criterion, but also a function of the
variable's inter-correlation with other variables. In the case of foreign
language achievement, although the individual predictor correlations
with the criterion were not relatively high, apparently the particular

combination of the predictors given the most weighting had relatively

 

2Edward E. Cureton, "II. Approximate Linear Restraints and Best
Predictor Weights ," Symposium: The Need and Means of Cross Validation,

Educational and Echelogical Measurement, 13:13, Spring, 1951.

TABLE 23

12h

A SUMMARY OF THE MULTIPLE CORRELATION COEFFICIENTS WITH THE STANDARD

ERRORS, USING EIGHT TEST VARIABLES AS PREDICTORS OF GRADE POINT

AVERAGES IN EACH SUBJECT AREA AND SEX

 

 

 

 

 

 

Multiple Correlation

 

Coefficients

Subject Area N Sex R S.E.
English 266 Boys .635 .037
329 Girls . 765 .023

Social Studies 171 Boys .6h0 .0h6
209 Girls .692 .037

Science 173 Boys .692 .Ohl
' 137 Girls .715 .0h3
Mathematics 107 Boys .6h2 .059
51 Girls .766 .063

Foreign Language 33 Boys .666 .121
62 Girls .771; .055

Industrial Arts 7h Boys .hh8 .098
Home Economics h9 Girls .660 .088
Business Education 187 Girls .6hh .Ohh

 

125
low inter-correlations with the other predictors. Thus, the final mul-
tiple correlation is larger than one would expect. Apparently the two
tests given the most weighting in the prediction of foreign language
grade point averages for boys are the Numerical Ability Test and the
Mathematics Proficiency Test. For girls, the two tests given the most
weighting are the Spelling and Numerical Ability Tests. The Mathematics
Proficiency Test also is rather highly weighted in the latter prediction.

It is apparent that in the case of girls, the Spelling Test shows
a considerable correlation with the criterion (.637), while for boys the
correlation (.032) is not significantly different from zero. Why the
Spelling Test should be a good predictor of language grade point average
for girls and not for boys is not completely'understood. PrObably part
Aof the explanation lies in the difference between the variabilities on the
Spelling Test for boys and girls. For boys, the standard deviation
is 21.80, while for girls it is 27.60. Obviously, the greater restriction
of range exhibited by the boys is one determining factor of the phen-
omenon; however, whether it is the sole cause of the discrepancy is not
known.

The next area open to the highest prediction for boys and girls
is mathematics, followed by English, science, social studies, home econ-
omics, business education and industrial arts. The range of correlations
seen in Table 23 is from .hh8 to .766. It might be said that, in general,
the coefficients are larger than those generally found in the literature
concerning studies of this type.

The fourth step in this multiple regression problem involves the

writing of total regression equations based on the eight predictors.

126
These equations, for each subject area and sex, are found in Appendix D.

One of the objectives of this study was to determine whether the
two predictor combinations producing the highest multiple correlation in
each subject area and for each sex, differed significantly'from.a similar
analysis of the multiple correlations derived from.a combination of all
eight predictors. Tb this end, Table 2h presents the names and the best
two predictor combinations, separated by subject area and sex. The max-
imum.(eight variable) correlations in each subject area and sex also are
presented, together with the final test of significance of difference
between the two multiple correlations. It is seen that in no case is the
difference between the two multiple correlations statistically significant.

The range of multiple correlations derived from.the various two-
predictor combinations is form .h13 to .7h0, with a median of .6h2. The
range derived from a combination of all eight predictor tests is from
.hh8 to .766, with a median of .666.

Table 2h also shows the TermaneMCNemar and Mathematics Proficiency
Tests combination to provide the highest multiple correlations in six
out of the thirteen cases.

In every case, either the Mathematics Proficiency, Numerical
Ability Tests, or both, entered into the highest two-predictor combin-
ations. In four out of the thirteen cases, a combination of the Spelling
Test and one of the "number" tests proved to be the best dual combin-
ation. One of the reasons for this is the relatively low correlation
between the spelling and.number tests as seen in Table 13.

Since it is an objective of this study to interpret the findings

in such a way as to make them usable and understandable to counselors,

127

 

 

nowpwUSom
amm. ooo. ooo. ooo. nom. ahaaaos .saz one .eoai .eoez eoa aaaao nonsense
moasocoom
new. ooo. ooo. Noe. mam. epaaaoe .saz one .o.H sashes an naaao asem
nose
omm. ooo. one. ooo. mas. meaaaoom one .oonm .eoez ea seem HosannaonH
ooo. mmo. new. ooo. one. .neoe .eooz one eeaaaoa . 2 mo nanao omeawsea
moo. awe. ooo. ems. mom. .neom .eooz one soafiaoe .saz mm seem snaoaee
mmm. moo. ooo. ooo. mme. msaaaoam one .eoni .eaes Hm naaao noaoes
ooo. omo. moo. amo. moo. msaaaoam one .eonm .eoez eoa neon -oeoez
mod. moo. mas. moo. mos. .o.H castes ens .monm .eoez ema aaaao
mos. ago. moo. moo. moo. .o.H ceases one .eoam .esez maH seem oesoaom
nee. Hmo. moo. omo. eeo. .o.H sesame one .oonm .eoez mom aanao noaoaom
ram. ooo. ooo. moo. mao. .o.H sesame one .moni .eoez H~H seem aeaoem
are. moo. mop. mmo. one. .o.H sesame one .eoni .eoez awn nanao
Nae. emo. mmo. Heo. pom. .o.H sesame ens .eoam .esea oom seem eaaamsm
eaoeaue .m.m a .m.m m seaeeeaosoo 2 saw eons
oaeeaeae noose oaoeaaas are noooaeonm 0:9 room oooooam

 

 

 

 

Nam 92¢ <mm< Homwmbm modm mom
mZOHemqmmmoo mquBADZ 03B mmmma zmmzemm MozmmmmmHm mo
moz<onHonm mme 924 aZOHB<ZHmZOo mOBoHQmmm A<HOB are amZOHemszzoo
EOBOHQmmm 039 Bmmm mme mo mmommm QMdQZdBm Qz< mZOHB<Ammmoo mquHHDZ mmh

am mqm¢e

 

 

128
the regression equations based on each of the twoepredictor combinations
of Table 2h were graphically represented in nomograph form. These nomo-
graphs, by sex and subject area, are seen as Figures 3 through 15 in
Appendix C.

We prediction 2; the Essential E252 m1 Content Batten
criteria. Since the EHSCB criteria play a secondary role to the grade
point average criteria, and since it was found that in the latter’pre-
dictions there were no significant differences in the multiple correl-
ations derived through the use of two and eight variable combinations,
the multiple prediction of the EHSCB tests was accomplished by selecting
the two predictor tests producing the highest multiple correlation. These
tests with the multiple correlations and respective standard errors for
each EHSCB test is shown in Table 25.

Table 25 shows correlations ranging from .687 to .860. The
subject area open to the highest prediction is seen to be English,
followed.by science, mathematics, and social studies.

In five out of the eight cases, the TermanéMcNemar and the Mathe-
matics Proficiency Test combination showed the highest two-predictor
multiple correlation. In the case of boys, this is explained by the
relatively low inter-correlation of these tests (.575). The Terman-
McNemar Test appears in seven out of the eight cases and shows, on the

whole, to be the best single predictor of the EHSCB tests.
III. THE SIGNIFICANCE OF DIFFERENCE BETWEEN SUBJECT AREA MAJORS

On preceding pages, the ability of the predictor tests to correlate

with the various criteria has been evaluated. In this section, an

129

 

 

Hmo. ooe. aaaaaos aeoaaosaz see .o.H sesame Hm naaao

moo. roe. eonoaoaaonm .easz one .o.H easing Noe neom aoaossoeaea
omo. ooo. eesoaeaaoai .eoez one .o.H sesame and naaao

emo. mes. Aesoaeaooni .eoez one .o.H assess nae seem ooeoaeo
omo. ome. eoaoaeaoonm .eooa one .o.H sesame mom naaao

Heo. ooo. eoeoaoaoena .eeaa one .o.H sesame Had seem noaeaom Heaoeo
ooo. ooo. .o.H sesame one meaaaoom awn naaao

moo. moo. soaoaoeeenm .eoez one eoeoaoaeenm enaaeem ooN seem essence
.m.m m arena 2 son snoooam momma

neoaoaeansoo aoaoaeonmuese seem

 

 

mquo 92¢ mwom mom Mmmea<m Hzmhzoo Aoomom mon A¢Hazmmmm mmH 20 mmmoom
GZHBOHQmmm 2H mZOHB¢szZOU mOBOHQHmmloze mme m0 mmommm mmmmzeem 924 mZOHedqmmmoo mquBHDE

mm mqm<a

130
attempt will be made to evaluate the magnitude of the differences in
perfbrmance on each of the predictor tests, comparing the means of the
students majoring in a given subject area with those not majoring in
that area. These evaluations will be made for boys and girls separately.

Table 26 presents the mean scores on the predictor tests made by
boys majoring in a given subject area and those not majoring in that
subject area. The significance of differences between each pair of
means is indicated by the "t" ratios.

An examination of Table 26 shows that only two out of the forty
t-ratios presented failed to reject the null hypothesis. The two ex-
ceptions are in comparing the Mechanical Reasoning means of foreign I
language majors and non-majors and in comparing the Spelling Test means
of the mathematics majors and.non-majors.

. It is also evident from Table 26 that, in general, the group of
boys majoring in foreign language represents the most able group of
students. Their relatively higher performance on the two achievement
test predictors, the Mathematics Proficiency and English Proficiency
Tests, is worthy of note.

The science majors seem.to represent the second most capable
group although this group is not greatly different from.the mathematics
majors in terms of their mean test performance. The social studies and
industrial arts majors, on a whole, rank fourth and fifth, reSpectively,
in their mean test performance.

It must be remembered that the rankings above are on a group
basis and are valid and meaningful in this context only. The reason for

this is that each student has majored in at least three subject areas;

131

TABLE 26

THE MEAN SCORES AND "t" RATIOS ON THE PREDICTUR
TESTS MADE BY BOYS MAJORS AND NON-MAJORS
IN EACH OF FIVE SUBJECT AREAS

 

 

 

 

 

 

 

Predictor Social Studies Science Mathematics
Test ijor' Non-Major Major Non-Major Majgr’ Non-Major_
Verb. N 171 95 173 93 107 87
ReaSe M 1501.13 18.01 l7e36 1,4911? l7e08 1,4035
t 2.7h 2.97 2.60
Nun. N 171 95 173 93 ‘ 107 87
Abil. M 1h.81 18.1h 16.98 1h.18 17.50 12.79
t 3.83 3.21 h.98
Mech. N 171 95 173 93 107 87
R683. M 330111 38093 36.69 32.96 38.67 310711
t 3.29 2.h8 h.20
Spell. N 171 95 173 93 107 87
M 26.13 2h.20 27.39 21.81 123.17 23.h6
t .69* 2.17 .09*
Sent. N’ 171 ' 95 173 93 107 87
M 18.17 23.13 22.02 16.07 22.38 17.06
t 2.98 3.62 3.02
Terman N 171 95 173 93 107 87
1.0. M 107.85 113.81 111.90 lO6.b1 112.20 106.07
t 3.50 3.08 3.22
Eng. N 171 95 173 93 107 87
Prof. M: 97.u0 107.3h 103.92 95.h3 10h.55 96.86
t 3.62 2.86 2.h6
Math. N 171 95 173 93 107 87
Prof. M 55.29 6h.11 60.07 55.hl 61.28 50.35
t 3.92 2.17 h.25

 

confidence.

* Lacks significance of difference at the five per cent level of

132
TABLE 26 (continued)

W

 

Predictor Foreign Language Industrial Arts
Test Major Non-Major Maj or Non-Major
Verb. N 33 233 7).; 192
Hoes. M 21.85 15.57 12.51 17.83
t h. 36 5.91
Num. N 33 233 7h 192
Abil. M 19.61 15.h9 13.6h 16.91
t 3.110 3.89
Mech. N 33 233 7b 192
R983. M 37015 35013 320,48 36050
t .96.“.- 2.56
Spell. N 33 233 7h 192
M 113.15 22.93 17.68 28.113
t h.10 6.95
Sent. N 33 233 7’4 192
M 28.06 18.79 12.31 22.88
t 11.10 6 .95
Terman N 33 233 71L 192
1.0. M 119 .69 108 .60 103.09 112.614
t 5. 23 5.71
Eng. N 33 233 7’4 192
Prof. M 120.118 98.18 88.024 105.93
’0 7.79 6.19
Math. N 33 233 7h 192
Prof. M 70.18 56.76 52.08 60.89
t 6.11 3.88

 

* Lacks significance of difference at the five per cent level

of confidence.

133
thus, a student majoring in a foreign language may also have elected to
major in inudstrial arts and science. In general, however, students
elect certain sequences of subject areas for majors, e.g. college bound
students generally elect a foreign language, mathematics and/or science
as high school majors.

An analysis similar to the one for boys is shown in Table 27 for
girls. The analysis and interpretaion of’Table 27 is seen to be much
the same as it was for boys; the outstanding features being:

1. Only one pair of means lack statistical significance of
difference; this involving the Spelling Test in the subject area of
mathematics.

2. In general, the differences between the means shown for girls
is of a greater magnitude than the differences shown for boys in Table
26.

3. A relative ranking of the subject area majors from high to low
shows that, in general, foreign language majors tend to be the most able
group followed by the mathematics, science, social studies, business
education and home economics majors.

h. By reference to the mean TermannMcNemar I.Q.'s for each group,
it will be noted that there is a distinct break between the foreign
language, mathematics, and science majors whose mean I.Q.'s are 119.68,
118.h1, and 112.09, respectively, and the mean I.Q.'s of the social
studies, business education, and home economics majors, whose mean I.Q.'s
are 105.09, 102.82, and 97.30, respectively. These facts probably can I
be explained in terms of the similar sequence of majors established by

students. For example, pupils majoring in foreign language also

TABLE 27

13h

THE MEAN SCORES AND ~"t" RATIOS ON THE PREDICTOR

TESTS MADE BY GIRL MAJORS AND NONAMAJORS

IN EACH OF SIX SUBJECT AREAS

 

 

Predictor Social Studies Science thhematics
Test Major NonéMajor Major Non-ngor Major Non-Major
Verb. N 209 120 137 192 71 258
Reas. 7M 1h.70 18.15 18.12 1h.h2 21.55 lh.h2
3.60 ' h.08 6.15
Num. N 209 120 137 192 71 258
Abil. M 13.28 16.87 16.09 13.52 20.21 13.0h
t h.30 3.23 8.06
Mech. N 209 120 137 192 71 258
R888. M 21069 25e58 2he91 21.83 29.112 2103?
t 3.81 2.58 .
Spell. N 209 120 137 192 71 258
t 3.26 1.1162:- h. 25
Sent. N 209 120 137 192 71 258
M 21.69 28.76 - 26.80 22.h6 31.96 22.15
t h.53 2.87 5.67
Terman N 209 120 137 192 71 258
1.0. M. 105.09 112.30 112.09 10h.60 118.h1 10h.78
t h.19 11.51; 8.07
Eng. N 209 120 137 192 71 258
Prof. M 99.95 113.72 108.98 100.55 118.11 100.19
’0 5.00 3.19 6.68
Math. N 209 120 137 192 71 258
Prof. M 50.71 60.88 58.31; 51.62 68.66 50.50
t b.87 3.3h 8.99

 

* Lacks significance of difference at the five per cent level

of confidence.

TABLE

27 (continued)

135

 

Predictor
Test

Verb.

Reas.

NUMe
Abil.

MGGhe
R6830

SpBlle

Sent.

Terman
I.Q.

PTOfe

Math.
PTOfe

c+§§=z c+2§=Z c+t§=z c+t§=2 (+£352 c+I§=Z c+§§=Z c+IESZ

Home Economics
Nbflor Noonajor

h9 280
11.12 16.81
6.62

h9 280
10.h1 15.32
5.85

N9 280
17.29 2h.13
h.75

h9 280

20.71 35.77
5.12

89 280
18.65 25.25
3.57

N9 280
97.30 109.5h
6.69

h9 280
89.20 106.66
5.10

h9 280
h3.63 56.31
h.h5

Foreign Language
Major NoneMajor

62 267

22.51 1h.h3
6.62

62 267
20.h8 13.22
7.72

62 267
29.05 21.73
5.23

62 267
h9.06 29.92
5.08

62 267
3h.06 22.00
6.h8

62 267
119.68 10h.9h
7.37

62 267
119.16 100.56
6.01

62 267
67.8h 51.30
7.62

Business Education
Major Non-Major

187 182
13.59 19.08
5.90

187 1h2
12.97 16.72
5.69

187 1&2
20.2h 26.89
5.59

187 1&2
29.08 39.39
3.66

187 1N2
20.71 28.96
5.39

187 1h2
102.82 11h.l7
6.72

187 1h2
97.98 112.07
5.3h

187 lh2
h9.92 60.35
5.11

 

136
frequently major in science and/or mathematics.
In the next chapter, the summary, conclusions and implications

for further research will be presented.

CHAPTER‘VII

3111mm, CONCLUSIONS, AND DIPLICATIONS
FOR FURTHER RESEARCH

I. SUMMARY

This study was designed to inquire into the single and combined
value of seven measures of aptitude and achievement to predict academic
success in each of eight subject areas at the high school level, i.e.
English, social studies, science, mathematics, industrial arts, foreign
language, home economics, and.business education, in addition to pre-
dicting the general academic success of high school students.

The predictive measures used consisted of two achievement tests,
the Mathematics Proficienqy, and English Proficiency tests; four special
aptitude tests of the Differential Aptitude Battery, i.e. Verbal
Reasoning, Numerical Ability, Mechanical Reasoning, and Language Usage
(consisting of'Spelling and Sentences sub-tests); and a measure of
general scholastic aptitude, the Terman-MCNemar'Tests of Mental Ability.
These measures were administered to the students in their latter months
of eighth grade.

Two criteria of academic success were used; grade point averages
and.Scores on the Essential High.School Content Battery. Grade point
averages in high school majors were computed by averaging teachers'
marks in three or more courses within a subject area. The total grade
point average, consisting of the average grades of all one-unit courses
taken in high school, was used as a criterion of general academic success.

The Essential High School Content Battery was administered during

138
the latter months of the student's eleventh grade. The four sub-tests of

this battery, i.e. English, Social Studies, Science, and Mathematics,
were used as alternate criteria of success in these subject areas. The
composite score of the EHSCB served as an alternate criterion to total
grade point average as a measure of general academic success.

The period of time elapsing between the administration of the
predictor tests and the securing of the criterion measures is seen to
be at least three years for both criteria.

The senior classes from six Cincinnati Public High Schools formed
the original experimental population. In an effort to insure uniformity
among the grade point average criteria from.achool to school, an analysis
of covariance technique was used. This technique served to adjust the
criterion means relative to the initial ability levels of the students
comprising a given school. The A.C.E. Psychological Examination was
used as the ability measure in this analysis; it was administered during
the midqyear when the students were high school juniors.

Using the same type of covariance technique for each of the
subject areas, as well as for total grade point average, three schools
were singled out which seemed to grade uniformly; The senior students
from.these three schools comprised the sample used in the evaluation of
the tests as predictors of academic success.

Only those seniors for whom complete test records were available
were used as subjects in this study. The final group conSisted of a
total of 595 students made up of 266 boys and 329 girls.

The statistical techniques used in analyzing the raw data con-

sisted of Pearson product-moment coefficients of correlation and

139
Student's t-ratio. The former was used as an index of relationship
between the predictor and criterion measures, while the latter was used
for determining the significance of difference between means of various
sub-groups.
Methods of determining the significance of difference between
correlation coefficients were also used in addition to the use of

Fisher's z-function for averaging correlations.
11. CONCLUSIONS

The conclusions of this study may center around the hypotheses as
set forth in CHAPTER I. Each will now be discussed in light of the
findings.

1. Hyppthesis. Assuming that academic achievement is due largely #1
to a generalized verbal factor, and assuming further that the Terman-
McNemar Intelligence Test is a valid measure of this verbal factor, it is
hypothesized that this test will be the best all-around.predictor of /J
academic success.

This general verbal factor, however, is likely to play a more
important role in the prediction of the Essential High School Content
Battery criteria, since the latter criteria are probably more dependent
upon verbal skills than are the characteristics upon which school marks
are based.

Conclusion. For both boys and girls, the Mathematics Proficiency
Test proved to be the best predictor of total grade point average. The
TermannMcNemar Test of Mental Ability showed the highest correlations

with the composite score of the EHSCB.

1h0

2. Hypothesis: Since twelfth grade girls probably represent a
more heterogeneous population than do twelfth grade boys, it would be
expected that the aptitude and achievement predictors would correlate
higher with both criteria for girls than they would for boys.

Conclusion. Although tests of significance of difference between
individual pairs of correlations did not, in general, prove to be statis-
tically significant, there is little doubt that the cumulative
probabilities indicate that the girls represented a more predictable
group in terms of predictor test performance, than did boys. In only
one case out of seventy-six possible comparisons did boys show a
higher correlation between a predictor and the criterion, than girls.
This fact is partially explained by the greater heterogeneity of high
school girls in comparison to boys.

3. Hypothesis: Due to the more highly verbal nature of the
Essential High School Content Battery criteria over the grade point
average criteria, it would be expected that much of the difference
between these two criteria would be in terms of the degrees to which the
verbally loaded measures differentially correlate with them.

Conclusion. On the whole, it was found that the Mathematics
Proficiency and Numerical Ability Tests were among the best predictors
of grade point averages. Although these tests also proved to be excellent
predictors of the EHSCB criteria, in many cases they were displaced by
the English Proficiency and Verbal Reasoning Tests.

The Terman-McNemar Test stood as a high predictor of both criteria
but proved to be a relatively better predictor of the EHSCB criteria.

At least in part, this is probably due to the highly verbal nature of

the TermanéMcNemar Test.

lbl

From this analysis, it appears that grade point average criteria
are more highly related to quantitative abilities, while the EHSCB cri-
teria are more highly related to linguistic abilities. The correlations
between these two criteria, however, were rather high indicating that
both criteria are measures of certain common factors.

h. Hypothesis: Boys would.be expected to perform higher on the
"number" tests, the Mechanical Reasoning and the Terman-McNemar Intel-
ligence Test, while girls probably would exceed.boys on the Spelling,
Sentences, English Proficiency and Verbal Reasoning Tests. This
hypothesis is based on the findings of numerous studies showing that on
the whole, boys exceed girls on quantitative measures while girls
generally exceed boys on linguistic measures. Even though the Terman-
MoNemar Test is verbally oriented, it is nevertheless, a general
aptitude test, and since it is expected that the remaining twelfth
grade boys represent a more homogeneous and academically select group
than do girls it, therefore, would be expected that they would perform
higher on this test.

Conclusion. Boys were found to achieve higher than girls on the
Verbal Reasoning, Numerical Ability, and the Mechanical.Reasoning Tests
of the Differential Aptitude Battery. The two groups were not found to
differ significantly, however, on the Verbal Reasoning Test. Boys also
achieved higher mean scores on the Terman-McNamar Test and the Mathe-
matics Proficiency Test; however, the means of the former test did not
prove toifitatistically significantly different.

Girls showed higher mean scores than boys on the Spelling,

Sentences, and.English Proficiency Tests. The differences between boys

1h2
and girls on the Spelling and.Sentences Tests proved to be significant
while the English.Proficiency Test means did not Show a significant
difference.

5. Hypothesis: It is hypothesized that there will be significant
differences in mean performances on the aptitude and achievement
predictors when students majoring in a subject area are compared to those
not majoring in that subject area. This hypothesis results from the
notion that , on the whole, the brighter students tend to enter the
college preparatory curriculum, i.e. mathematics, science, and foreign
language, while the less capable students tend to select other subject
areas for majors.

Conclusion. If one accepts the findings stated previously and
assumes the existence of a general intelligence factor as being the
factor which accounts for the high relationships shown in this study,
one may speculate with confidence that students tend to elect certain
combinations of majors. Thus, college preparatory students generally
elect majors in foreign language, mathematics and/or science. This
group generally constitutes the brighter pupils while the less able
students tend more toward the vocational subject areas. This would, of
course, account for the significance of differences in mean performance
of majors and non-majors in a subject area on the predictor tests.

A general ranking of mean performance on the predictor tests
shows the foreign language, mathematics and science majors to con-
stitute the highest subject area groups while the industrial arts, home
economics and business education majors constitute the lowest ability

subject area groups.

lh3
III. IMPLICATIONS FOR FURTHER RESEARCH

One of the most outstanding outcomes of this study is the rela-
tively high relationship seen between the predictor tests and the
various criteria. For the most part, studies Of this type result in
correlations between .hO and .60. Why should this particular sample
prove to be a more predictable group? Does the answer lie in the nature
of the characteristics of the sample group, or in the methods of
teaching and administering marks?

If the answer lies in the nature of certain population charac-
teristics, do these characteristics vary from population to population
or are they stable? If the answer lies in the methods of teaching and
administering marks, what methods are most conducive to high prediction?

Another question which needs further research is why quantitative
or "number" measures are better predictors of grade point averages than
linguistic measures. DO the rated outcomes of instruction, as reflected
by grades, actually depend to a greater extent upon quantitative skills
or do the latter simply reflect an aspect of "global" intelligence?
hbuld the answer to this question vary among different levels of intel-
ligence and educational achievement?

A further question arising from this study, and related to the
latter question, is whether or not it is fruitful to attempt to isolate
so called "special abilities" for purposes of predicting academic
success. In other words, is there sufficient differentiation among
abilities needed for success in the typical school setting to call for

tests Of Special abilities? It is the writer's Opinion that a molar

lhh
approach would.prove more useful for predictive purposes.

Special abilities, as such, would.probably play a subordinate
role, in the prediction of academic success, to such factors as general
attitudes toward school, subject area and teaching personnel; along with
other factors such as interest, motivation and past success.

The question may be raised as to why girls, in general, tended
to be a more predictable group, in terms of their high school
achievement, than boys. Is this purely a statistical phenomenon,
explainable in terms of the greater variance among girls at the twelfth
grade level, or is there a genuine difference in the ability to predict
boys'and girls'achievement in high school? Further research is necessary
to determine the answer to this question.

Due to the relatively high relationships between predictor and
criterion variables shown in this study, there is no doubt that the
effectiveness and reliability of'prediction should be checked. This
could be done through the use of cross-validation procedures. Shrinkage
would probably result, but to what extent it would occur would have to be
determined through further research.

It is apparent that this study has raised as many questions as

it has answered. ’

1.

2.

3.

h.

5.

7.

8.

9.

10.

12.

BIBLIOGRAPHY
A . BOOKS

Cronbach, Lee J ., Educational Psychology. New York: Harcourt, Brace
and Compamr, 19511.

Guilford, J .P. , Fundamental Statistics in chhology__ and Education.
New York: McGraw-Hill Book Company, 19112.

Pintner, Rudolf, Intelligence Testy 5 Methods and Results. New York:
Henry Holt and Company, 1931.

Super, Donald E., A rais' Vocational Fitness. New York: Harper
and Brothers, DEE

Thorndike, Robert L., Personnel Selection. New York: John Wiley and
Sons, 1113., 19,490

B . PERIODICALS

Amos, Viola, "Factors Related to High School Achievement, “ J_p_____urnal
of Educational P cholo , 311:229-36, April, 19113.

Cureton, Edward E., "II. Approximate Linear Restraints and Best
Predictor Weights," Symposium: The Need and Means of Cross Valida-
tion, Educational and Pachological Measurement, 13:11-16, Spring,
1951.

Curtis, Erta A., and Nemzek, Claude L., "The Relation of Certain
Unsettled Home Conditions to the Academic Success of High School
Pupils ," Journal 2; Social P cholo , 9:h19-35, November, 1938.

Dodes, I.A. , "Prediction of High School Success ," Hgg' h Points,
31:5-111, November, 19h9.

Douglass, H.R. , "Special Methods on High School Level: Mathematics ,"
Review 2; Educational Research, 2:7-20, 81-82, February, 1932.

Embree, Royal B., "Prediction of Senior High School Success at
Various Levels of Intelligence ," Journal of Educational Psychology,
28: 81-91, February, 1937.

Finch, F.H. , and Nemzek, C.L. , "The Relationship of the Bernreuter
Personality Inventory to Scholastic Achievement and Intelligence ,"
School and Society, 36:5914-96 , November, 1932.

1H6

13. Frandsen, Arden N. , and Sessions, Alwyn D., "Interest and School
Achievement ," Educational 33d Pachological Measurement,

13: 911-101, Spring—,l 1953. '

1h. Cough, Harrison 0., "Factors Relating to Academic Achievement of
High School Students, Journal 9__f Educational P cholo
140: 65-78, February, 19119.

15. Heimann, Robert A. , and Schenk, Quentin F., "Relation Of Social Class
and Sex Differences in High School Achievement ," School R__e____view,
62. 213-21, April, 19511.

16. Hinkelman, Emmet A., "Relation of Certain Personality Variables to
High School Achievement ," School Review, 60:532-3h, December, 1952.

17. Jordan, A.M., "Correlations of Four Intelligence Tests with Grades ,"
Journal 2;; Educational P cholo , 13:1419-29, October, 1922.

18. Kaulfers, Walter V., "Present Status of Prognosis in Foreign Language ,"
School Review, 39:585-96, June, 1931.

19. Lee, J. Murray, and Hughes, w. Hardin, "Predicting Success in Algebra
and Geometry,“ School Review, h2:188-96, March, 1931:.

20. Limp, C.E., "A Work in Commercial Prognosis ," J_q____urna1 of Educational
Research, 16: h6-56, June, 1927.

21. McDaniel, H.B., "The Use of Tests in Guidance," California Guidan___£_e
Newsletter, 3: 3, November, 19119.

22. Mitchell, Blythe 0., "The Relation of High School Achievement to
the Abilities Measured by the Holzinger-Crowder Uni-Factor Tests,"
Educational and chhological Measurement, 15:1187-90, Fall, 1955.

23. Mitchell, Claude, "Prognostic Value of Intelligence Tests ," J_3____urna1
2; Educational Research, 28. 577-81, April, 1935.

21;. Nemzek, Claude L., "The Value of Certain Factors for the Direct and
Differential Prediction of Academic Success ," Journal 2; meri-
mental Education, 7:199-202, March, 1939.

25. "The Value of the Bernreuter Personality Inventory for Direct
and Differential Prediction of Academic Success as Measured by
Teachers' Marks ," Journal 9_f Applied dP cholo , 22:576-86,
December, 1938.

26. "Value of Certain Factors for Direct and Differential Predic- 1
tion of Academic Success ," J_q_____urnal of S_q____cial chholog, 12:21-30, {
August, 1910. ,2.

k.

27.

28.

29.

30.

31.

32.

33-

31.10

35.

36.

37.

38.

39.

140.

1117

Ohlson, David, "School Marks and Intelligence," Educational Admin-
istration 5pc}. S ervision, 13:90-102, February, 1927.

Powers, S.R., "Intelligence as a Factor in the Selection of High
School Subjects," School Review, 30:1455, June, 1922.

Proctor, William, "Psychological Tests as Means of Measuring the
Probable Success of High School Pupils," Journal 2;: Educational
Research, 1:258-70, April, 1920.

Ross, 0.0., and Hooks, W.T., "How Shall We Predict High School
Achievement," Journal 9; Educational Research, 22:1814-95,
October, 1930.

Seagoe, M.V., "Prediction of Achievement in Foreign Language,"
Journal pf Applied ngcholog, 22:632-6140, December, 1938. .

Segel, David, "The Validity of a Multiple Aptitude Test at the
Secondary Level," Educational ppg Pflchological Measurement,
73695'705: Spring, 19 0

Shaw, Duane 0., "A Study of the Relationships Between the Thurstone
Primary Mental Abilities and High School Achievement," Journal
pf Educational Pacholog, 140:239-2119, April, 1919.

Super, Donald E.,"The Bernreuter Personality Inventory: A Review of
Research," Pflchological Bulletin, 39:914-125, March, 19142

Tozer, C.E., "A Statistical Prediction of High School Success for
Purposes of Educational Guidance," Journal p; Educational Research,
22: 399-1402 , November, 1930.

Thorndike, Edward L., "Interests and Abilities," Journal 2; Applied
Monolog, 28:143-52, April, 191414.

Townsend, A., "Achievement and Interest Ratings for Independent
School Boys," Educational Records Bulletin, 143:149-514, January,
1915. """“"" """"""' """'"""'

Wesman, Alexander G. , "Separation of Sex Groups in Test Reporting,"
Journal pf Educational chholog, 140: 223-221; , April, 19149.

Mliteley, Sarah 3., "Predicting Stenographic Success Through
Prognostic Tests," Th3 Balance Sheet, 18:2h2-hh, March, 1932.

Wolking, William D. , "Predicting Academic Achievement with the
Differential Aptitude and the Primary Mental Abilities Tests,"
Journal 3; Applied P cholo , 39:115-18, April, 1955.

14-10

h2.

h3.

hh.

h5.

h6.

h7.

h8.

h9-

SO.

1118

C. PAMPHLETS AND BULLETINS

Committee on Test Standards for the AERA and NCMUE, Technical
Recommendations for Achievement Tests (washington, D. 0.: National
Education Association, 19531, p.17.

Kelly, T. L., Educational Guidance (New York Teachers' College,
Contributions to Education, No. 71, New York, 19lh), p.116.

Miles, W.R., A Comparison_ of Elementagy‘and Hi 1gh School Grades
(University of Iowa, Studies in Education, Vol. 1, No. 1, Iowa
City, Iowa, 1911), p.22.

Ross, C.C., Thg'Relation Between Grade School Record‘ggg High
School Achievement-(New York Teachers' College, Contributions to
Education, No. 166, New York, 1925), p.70.

Seashore, Harold G., "Aptitude, Intelligence, and Achievement,"
Test Service Bulletin, No. 51 (New‘York: The Psychological
Corporation, 19§51, p.1.

Swineford, Francis and Holzinger, Karl J. ,A Study_ in Factor
Analysis:_ TheIReliabilit of Bi-Factors and Their Relation to
Other Measures (University_ of Chicago, Supplementary Mbnographs,
No. 33 .

Wesman, Alexander G., A Study 2;. Transfer 2; Training (New York
Teachers' College, Contributions to Education, No. 909, New
York, 19145), 13025.

D. ENCYCLOPEDIA ARTICLES

.Archer, Clifford.P., "Student Mbrtality," Engyclopedia of Educational
Research, p. 1158. New York: The Macmillan Company, 1930.

Enrich, Alvin G., and Cain, Leo F., "Prognosis in Secondary Schools,"
Engyclopedia 2§,Educational Research, pp. 8hh-859. New York: The
Macmillan Company, 19

E. TEST MANUALS

Allen, Richard D., Bixler, Harold H., et. al., "Content of the
Series," Metropolitan Achievement Tests: Intermediate and
Advanced Arithmetic Tests, p.1. New York: The World Book Company,

19E7.

51.

52.

53.

5h-

55.

lb?
American Council on Education, Manual of'Directions: American

Council on Education ngchological Examination, p.1. Princeton,
New Jersey, 19 7.

Bennett, George, K., Seashore, Harold G., and wesman, Alexander G.,
Differential Aptitude Tests, Manual-Second.Edition, p.1. New
York: The Psychological Corporation, 1932.

Harry, David P. and.Durost, Halter N., Manual'ggiDirections,
Essential High School Content Battegy, p.1. New York: The world
Book Company, 19 l.

Terman, Lewis M. and.McNemar, Quinn, Construction of the Tests: Terman-

McNemar Test of Mental Ability-Manual, p. 3. New York: The World
Book Company, 19 2.

F. DOCTORAL THESES

Aaron, Sadie, "The Predictive Value of Cumulative Test Results,"
Doctor's Thesis. Stanford University, l9h6, p.227. V

APPENDIX A

151

 

 

mmao. oooo.a
ward. oooo.a
moao.- oooo.a
Smo. 884..
Pgma. oooo.H
ooam. oooo.a
:pzo.- oooo.a
moi”; 8%. 88.." 0
8R. moma. 2%. m
oamm.m cpmm. 0000. a .4 HHH>
>wam.a oomo. mmaa. oooo.a o
pmhm. wgoo. mmgo. mamm. m
ompm.m ommm. osmm. oooo.a <.HH>
mmmm.a mpmm. Hana. mpmm. oooo.a o
mmpm. mace. mane. mama. ngam. m
ommm.m ommm. ompm. ommm. oooo.a < H>
cmap.a Emma. HHma. :mom. Nana. oooo.a o
gnaw. mwo. opmo. :mmo. mmwo. mm::. m
ommm.m ogzm. ooam. cram. ommm. oooo.a « >
mmmm.a mmaa. maoo. worm. oped. «ham. oooo.a o
amdm.a ammo. Haoo. wmmm. mmaa. mpma. ommm. m
o»om.: omm:. ozmm. cmmm. omwm. 03H». oooo.a < >H
a.“ mm o. maoa. mace. mzpo. :wmo. m:ma. - oooo.a o
m mm. m: o. mama. osoo. wowo. pmmo. mood. . Hmom. m
oamm.m ooam. omen. omam. oaog. ommm. omma. oooo.a « HHH
moaa.m mmma. mamm. mama. mmom. mama. oaoa. :owa. oooo.a o
mamm.m mamm. pama. mzma. orga. HHPH. «mad. HHma. mmmw. m
ommm.m cmam. oagp. omam. omam. ommm. oamg. omam. oooo.a «.HH
omm>.m 03mm. oamm. ommm. ohm». capo. ompm. omoa. ommm. oooo.a H
Momno .pfigo .uopm .Honm .a.H .pnom .Haomm .muom .Han< .muom
flu: .mnm 33mm. . 30m: .52 58>

 

 

momdnmzé 936m 0.3.8 H.308 whom mnavoflvonm you

mamma pnwam go mpamHOHmwmou wpmm may now soapsaom mappfiaooa

mm age

152

 

mmmo.u oooo.H

comm. oooo.a

mpoo. oooo.a

Homufi o 0800 3”

:pzo. oooo.a

:oom. oooo.a

mmao.- oooo.a
Hmmm.a ommm. oooo.a o
mamm. mmmo. down. m
cpmm.m Omar. oooo.a <.HHH>
Haom.a wmmo. Pmam. oooo.a o
:omm. mmao. axmo. mmmm. m
omm:.w owmm. 0pm». oooo.a < HH>
oaoo.m meow. rmmm. pmmm. oooo.a o
szm. meo. ammo. mwga. mmpm. m
comm.m Odom. opmp. ommm. oooo.a «VH>
mHOm.H mmoa. omoa. mama. :moa. oooo.H o
ppmm. :m:o. mmzo. Homo. mmgo. mdms. m
.owwm.m owmm. ommm. 0:0». cmum. oooo.a < >
omm:.m mmmm. mgma. maaa. Hmmm. mawm. oooo.a o
Pmmm.a m:>a. maoa. mamm. :mga. magm. momw. m
omOm.m omwm. cpmm. camp. ommm. omgm. oooo.a .4 >H
Hmam.a ::mo. mmao. gmao. ommo. mmaa. mmgo..- oooo.a o
08>. $4.0. No.8. Elmo. maze. mayo. :80. .. {1%. m
o:om.: 0pm . cams. owma. comm. ommg. ovam. 0000. < HHH
momm.m mma . «mam. oppm. momm. mamm. mmmm. mmmm. oooo.a o
mwpo.m oaom. Hmam. :ama. mmma. mgwa. mmma. mama. pumm. m
o:mo.m omwm. 0mm». ommm. ompm. ommm. o:am. adam. oooo.H «.HH
oowa.m ommm. oamw. cmaw. omoo. oamm. ooam. oa:m. ooam. oooo.H A< H

Momgo .pflno .mopm .%0hm .a.H .pqmm .aammw .mmmm .Hflp< .mmmm
gvmfir. .mam, adahmfi‘ . .mumz, .852, .nhwbl

 

 

mammnw>¢ pqflom madam Hdpoa mHnHo mnfipowdmym Mom
mumme pgmfim no mpanOHmmmoo upmm any you soapsaom mappflaoon

mm magma

153

HOHH.I OOOO.H

 

mmmm. oooo.a

mmao.- _ oooo.a

mmma. oooo.a

mmwo. oooo.a

ommm. cocoa

mmoo. oooo.a
Emma mamm. oooo.a 0
8mm . mmmo. aomm. m
oaam.m comm. oooo.a 4.Haa>
mmom.a mmmo. pmmm. oooo.a o
Foam. mmao. mmmo. «mam. m
ommm.m ommm. ommm. oooo.a < HH>
ampa.m mmmm. pmmm. pmmm. oooo.a o
ommm. amoa. mmmo. mmma. mmpm. m
Ommm.m ommm. ommm. ommm. oooo.a <_a>
ammm.a mpma. omoa. mmma. mmoa. oooo.a o
ammm. mmmo. mmmo. aomo. mmmo. mmmm. m
ommm.m ommm. ommm. omou. ompm. oooo.a < >
mmmm.m mmmm. mmma. mamm. ammm. mapm. oooo.a o
mmmm.a moma. maoa. mama. mmma. mmmm. momm. m
ommm.m oopm. ommm. oamm. ommm. ommm. oooo.a _< >H
maom.a oaao. mmao. mmmo. ommo. mmaa. mmmo..- oooo.a o
mmpm. awoo. moao. mpmo. ammo. ammo. mmmo. . mmmm. m
om>~.m oaam. ommm. ommm. ommm. ommm. owam. oooo.a < HHH
Nmmm.m PmOm. «mam. ommm. mmmm. mmmm. mmmm. mmmm. oooo.a o
mamo.m ammm. ammm. mama. mmma. mmwa. mama. mama. Pmmm. m
ommo.m oamm. camp. ommm. ommm. ommm. omam. oamm. oooo.a < Ha
omma.m ommm. oamm. emap. omom. oamm. ,oomm. oamm. ommm. oooo.a m a

Mommo .pauo .mopm .moym .m.a .pamm .aamAWI. .mwom .aan< .mamm
.npwz .mnm adahma . - .nomz .352 , s,.£Hm>

 

mmaamqm ma momanmp< pqaom momma manam muapoammgm you
upmma pmmam mo magmaoammuoo amom mm» you qoaesaom mappaaoon

Om UHQwB

15h

 

mwma . 0000.4”

mama. oooo.a

mmma.- oooo.a

maoa. oooo.a

ammo. oooo.a

Emma. oooo.a

mamo . 008.."
mamm.a mamm. oooo.a o
ommm. ammo. mmmm. m
ommm.m ommm. oooo.a .4 aaa>
ommma mmmo. mmma. oooo.a o
mmmm. momo. mmmo. mamm. m
ommm.m omam. ommm. oooo.a _< HH>
mmmm.a mmam. amma. mpmm. oooo.a o
mmmm. ammo. mmmo. mmma. nmmm. m
comm.m ommm. empm. ommm. oooo.a .4 H>
maam.a ammo. aama. mmom. pmma. oooo.a o
mmam. ammo. opmo. mmmo. mmmo. mmmm. n
open.m cmmm. comm. ommm. ommm. oooo.a .< >
momo.m mmma. maoo. mmpm. owma. «ham. oooo.a o
mmmm .a ammo. aaoo. mumm. mmaa. mpma. ommm. m
owam.m ommm. ommm. ommm. ommm. omam. oooo.a A< >H
mmmm. Emma.- mmoa. mmoo. - ammo. . mmmo. mmma. - oooo.a o
ommm. aama:. mmmo. omoo. . momo. pmmo. mooa. . amom. m
omOm.m cmma. ommm. omam. oaom. ommm. omma. oooo.a «.HHH
mmpm.m mmmm. mmmm. mmma. mmom. mama. ommm. moma. oooo.a o
moma.m mamm. pamm. mmma. Ema. aawa. mmma. aama. mmmp. m
cmma.m ommm. cam». omam. ommm. ommm. oamm. ommm. oooo.a ¢.aa
322m ommm. ommm. ommm.. ommm.. oamm. ommm. 83. ommm. oooo.a a
moose .pano .uonm .mogm .a.a .pqmm .aammm .mmom .aagm .maom
_ m.. .. .mpmz .mcm .quanms ‘ .momz gasz .mnm>

 

muaamam ca mmmmnmpm pqaom omauu whom weapoammpm you
mamas pmmam mo mpnmaoammooo spam mam mom soapsaom mappaaoon

Hm. magnum.

155

HHH>

 

mmmo.- oooo.a
ammo. oooo.a
mmmo.- oooo.a
mmma. oooo.a
mmaa. oooo.a
mmmm. oooo.a
omaa. oooo.a
mmmm.a ammm. oooo.a o
mmmm. momo. mmmm. m
85m ommm. oooo.a <
mmmm.a mmma. mmma. oooo.a o
maam. mmmo. mmmo. mmmm. m
oamm.m ommm. ommm. oooo.a <.aa>
mamo.m mmmm. amma. mumm. oooo.a o
mmom. mmaa. mmmo. mama. mmmm. m
ommm.m cmmm. cmmm. ommm. oooo.a < H>
cmm>.a . moam. aama. mmow. mmma. oooo.a o
mmmm. mmmo. ommo. mmmo. mmmo. mmmm. m
ommm.m oamm. comm. ommm. ommm. oooo.a _< >
mmmm.a mamo. maoo. mmmm. omma. «mam. oooo.a o
mmma.a mmmo. aaoo. mmmm. mmaa. mpma. ommm. m
oomm.m oamm. ommm. ommm. ommm. omam. oooo.a <_>a
mmmoa ommo: mmoa. mmoo. .. mmS. mmmo. mmma. .. cocoa o
momm. mmmo: mmmo. omoo. .. momo. mmmo. mooa. .. amom. m
comm.m omma. ommm. omam. oaom. ommm. omma. oooo.a a aaa
mmmm.m ammm. mmmm. mmma. mmom. mmmm. omom. moma. oooo.a o
ommad mamm. mamm. mmma. mmma. aama. mmma. aama. mmmm. m
oama.m cmmm. oamm. omam. ommm. ommm. oamm. ommm. oooo.a «.aa
ommm.m 0mmm. ommm. ommm. ommm. oamm. ommm. omom. ommm. oooo.a a
Mommo .paho .mogm .moym .m.a .pqmm .aammm .mmmm .aagm .mwmm
. mum: .mam adapme .nooz .asz .pnm>

 

mmamspm awaoom ma mmmwnoh< pqaom «mama whom mnapoammgm no“
mpmms pmmam mo mpqmaoammmoo spam mm» new aoaasaom cappaaoon

mm magma

156

mmoo.

 

oooo.a

ammo. oooo.a

mmoo. oooo.a

mmmo. oooo.a

mmaa. oooo.a

ommm. oooo.a

amoo.- oooo.a
mmmm.a mmmm. oooo.a o
momm. oamo. aomm. m
ommm.m ommm. oooo.a <_aaa>
mmflza mm? PTR. 259a o
mmmm. mmao. mmmo. mmma. m
ommm.m 0mmm. ommm. oooo.a H m aa>
mmma.m mmmm. mmmm. mmmm. oooo.a o
mamm. mamo. mmmo. mmma. mmma. m
ooam.m ommm. ommm. ommm. oooo.a <.a>
mmmm.a mama. omoa. mmma. mmoa. oooo.a o
ommm. ommm. mmmo. aomo. mmma. mmmm. m
oamm.m oamm. ommm. omom. ommm. oooo.a < >
mammd mmma. mmma. mamm. ammm. mam? cocoa o
mmmm.a mema. maoa. mama. mmma. mmmm. momm. m
oamm.m omOm. ommm. oamm. ommm. ommm. oooo.a .< >a
mmmm.a mmmo. mmao. mmmo. ommo. mmaa. mmmo..- oooo.a o
mmmmf mmmo. moao. 2.8. mmma. mmmo. mmmo. .. mamm. m
omm>.m ommm. ommm. ommm. ommm. ommm. omam. oooo.a < aaa
mmmm.m mmmm. mmmm. ommm. mmmm. mmmm. mmma. mmmm. oooo.a o
mmmm.a amma. ammm. mama. mmma. mmma. mmma. mama. mmmm. m
ommm.m ommm. ommm. ommm. ommm. ommm. omam. oamm. oooo.a < aa
omma.m ommm. oamm. cmam. omom. oamm. {oomm. oamm. ommm. oooo.a < a

Momma .pauo .monm .moym .m.a .pqwm .aammm .mmmm .aan< .mmmm
mum: .mqm qmsnma .w» .momz .asz .ppo>

 

 

mmamspm amaoom ma mmmwnm>¢ pnaom mmwuo manam mqapoamopm How
mpmma pmmam mo magmaoammmoo upmm map mom aoapsaom mappaaoon

mm magma

 

157

 

mooa.- oooo.a

maao. oooo.a

mmma. oooo.a

mmaa. oooo.a

mmma. oooo.a

mama. oooo.a

mooa. oooo.a
mmmma mmmm. oooo.a o
mmmm . mmma. mmmm . m
oamm.m oamm. oooo.a 4 aaa>
momm.a mama. mmma. oooo.a o
mmam. mmmo. mmmo. mmmm. m
ommm.m omOm. ommm. oooo.a 4 aa>
mmmm.a ammm. amma. mmmm. oooo.a o
ammm. momo. mmmo. mmma. mmmm. m
oaam.m oamm. ommm. ommm. oooo.a A< Hm
mmmm.a mmam. aama. mmom. mmma. oooo.a u
mmmm. mmmo. ommo. mmmo. mmmo. mmmm. m
cmmm.m oaOm. comm. ommm. ommm. oooo.a .< >
momm.a mmao.- maoo. mmmm. omma. mmam. oooo.a o
mmma.a omoo.- aaoo. mmmm. mmaa. mmma. ommm. m
ommm.m omom. ommm. ommm. ommm. omam. oooo.a <_>H
mmmma ammm. mmoa. mmoo. mmmo. mmmo. mmma. - cocoa o
momoa mmma. mmmo. omoo. momo. mmmo. mooa. . amom. m
omam.m ommm. ommm. omam. oaom. ommm. omma. oooo.a 4 aaa
mmao.m momm. mmmm. mmma. mmom. mmmm. omom. moma. oooo.a o
mmma.m mmmm. mamm. mmma. mmma. aama. mmma. aama. mmmm. m
omma.m ommm. oamm. omam. ommm. ommm. oamm. ommm. oooo.a < aa
omom.m ommm. ommm. ommm. ommm. oamm. ommm. omom. ommm. oooo.a a
moomo .pauo .mopm .mopm. .m.a .pamm .aaumw .auom .aaam .naom

mmmz, .mmua quanma, . . .mmmzl, .asz . mom

 

 

monoaom na
mamas pmwam no apnoaoammooo «pom any you noapsaom mappaaoon

.mm

manna

momduog. 950m 03.8 Enom mnapoavonm .aom

158

 

 

 

mmoo. oooo.a

mmmo 88a

mOmo. oooo.a

mmoa. oooo.a

mmao - oooo.a

mmom oooo.a

oooa. oooo.a
mmmm.a mmmm. oooo.a o
ommm. mmmo. aomm. m
ommm.m cmmm. oooo.a <.aaa>
mmmm.a mmao.- mmmm. oooo.a o
ommm. mmoo.- mmmo. mmmm. m
oamm.m comm. ommm. oooo.a m aa>
mmma.m mmmm. mmmm. mmmm. oooo.a o
ammm. mmmo. mmmo. mmma. mmmm. m
ommm.m oamm. ommm. ommm. oooo.a <_a>
ommm.a mmmo. omoa. mmma. mmoa. oooo.a o
amam. mmmo. mmmo. aomo. mmmo. mmmm. m
ommm.m ommm. ommm. omom. ommm. oooo.a 4 >
mmmm.m aamm. mmma. mamm. ammm. mamm. oooo.a o
mmom.a mmma. maoa. mamm. mmma. mmmm. momm. m
ommm.m ommm. ommm. oamm. ommm. ommm. oooo.a 4 ma
mmmm.a mmmo. mmao. mmmo. ommo. mmaa. mmmo..- oooo.a o
moam. ammo. moao. mmmo. mmmo. mmmo. mmmo. 44mm. m
ommm.m ommm. ommm. ommm. ommm. ommm. omam. oooo.a < aaa
mmam.m mmmm. mmmm. ommm. mmmm. mmmm. mmmm. mmmm. oooo.a o
momm.a amam. ammm. mama. mmma. mmma. mmma. mama. mmmm. m
oomm.m ommm. ommm. ommm. cmmm. ommm. omam. oamm. oooo.a m aa
cmma.m oamm. oamm. cmam. omom. oamm. comm. oamm. ommm. oooo.a < a
moomo .pauo .moum .moum .m.a .pnom .aaomm .uaom .aag< .uaom

: and: .mqm nuance .. .momz. .aaz .nuom
monoaom ma nowduvh« pnaom nudge nauam mdapoadmnm you

«page mmmam mo

mm «anus

manuaoammooo upon any you aoapsaom cappaaoon

159

 

mmmo.u oooo.a

mmao. oooo.a

mmmo. oooo.a

mwam. oooo.a

mmoo. 0000..”

mama. 0000...”

mmma.- oooo.a
mmam.a mmam. oooo.a o
ammm. omma. mmmm. m
ommm.m ooam. oooo.a < H:
mmma.a mmmo.- mmma. oooo.a o
ommm. omoo... mmmo. mmmm. m
oamm.m omom. ommm. oooo.a <_aa>
mmmm.a mmma. amma. mmmm. oooo.a o
mmmm. ommo. mmmo. mmma. mmmm. m
omom.m cmmm. ommm. ommm. oooo.a a Hm
mmmm.a mmma. aama. mmom. mmma. oooo.a o
aomm. mmmo. ommo. mmmo. mmmo. mmmm. m
oomm.m ommm. comm. ommm. ommm. oooo.a _< >
mamo.m amom... maoo. mmmm. omma. mmam. oooo.a o
aaOm.a mmma. .aaoo. mmmm. mmaa. mmma. ommm. m
oomm.m oaom. ommm. ommm. ommm. omam. oooo.a «.ma
mmmaa ammo. mmoa. mmoo. mmmo. mmmo. mmma. .. cocoa o
mmmm. mmmo. mmmo. cmoo. .. momo. mmmo. mooa. .. amom. m
ommm.m oamm. ommm. omam. oaom. ommm. omma. oooo.a <_aaa
mmmo.m mmom. mmmm. mmma. mmom. mmmm. omom. moma. oooo.a o
mmamm mmmm. mamm. mmma. mmma. aama. mmma. aama. mmmm. m
oama.m ommm. oamm. omam. ommm. ommm. oamm. ommm. oooo.a «.aa
oomm .m ommm . ommm. ommm. ommm. oamm. ommm. omom. ommm. ooooa a
moomo .pyyo .yoym .yoym .m.a .pyom .aaomm .ooom .aaam .ooom
. ......M _. gm: .myymf adayme . .nomz .Esz .o..a0>

 

mum: ya oomoym>m poaom omoyo whom moapoamoym you

mpmoa mmmam yo apnoaoayyooo opom amp yoy noapsaom mappaaooo

mm oaooa

I . l- I"!
._

‘ I It

 

160

HHH>

 

mmao.u oooo.a

amma. oooo.a

omma. oooo.a

momm. oooo.a

mmoo.- oooo.a

mmmo. oooo.a

mmmo.- oooo.a
mmmma mmmm. oooo.a o
mmmm. mmmo. aomm. m
ommm.m oamm. oooo.a .4
mmmm.a ammo.. mmmm. oooo.a o
mmmm. mmoo. mmmo. mmmm. m
omom.m omam. ommm. oooo.a 4 aa>
mmmm.a mmoa. mmmm. mmmm. oooo.a o
mmam. ommo. mmmo. mmma. mmmm. m
ooOm.m ommm. ommm. ommm. oooo.a 4 am
mmmm.a ammo. omoa. mmma. mmoa. oooo.a o
mmom. mmoo. mmmo. aomo. mmmo. mmmm. m
ommm.m ommm. ommm. omom. ommm. oooo.a 4 >
mmmm.m mmmm. mmma. mamm. ammm. mamm. oooo.a o
mmmmya mamm. maoa. mamm. mmma. mmmm. momm. m
oamm.m ommm. ommm. oamm. ommm. cmmm. oooo.a 4_>H
mmmm.a mmma. mmao. mmmo. ommo. mmaa. mmmo.. oooo.a o
mmmm. omoa. moao. mmmo. mmmo. mmmo. mmmo. mmmm. m
ommm.m omom. ommm. ommm. ommm. ommm. omam. oooo.a 4 aaa
mmmm.m mmmm. mmmm. ommm. mmmm. mmmm. mmmm. mmmm. oooo.a o
mmmo.m mmmm. ammm. mama. mmma. mmma. mmma. mama. mmmm. m
ommo.m ommm. ommm. ommm. ommm. ommm. omam. oamm. oooo.a .4 Ha
ooma.m ommm. oamm. omam. omom. oamm. ,oomm. oamm. ommm. oooo.a 4 a

Moomo .payo .yoym .moyy .m.a .yoom .aaomm .mmom .aag4 .moom
‘ mpmz .mym nosyoe .. .momz .asz .pyo>

 

 

man: my momoyo>4 poaom omoym mayao mqayoaooym yom

mpmma mmmam yo mpyoaoaymooo opom on» yoy ooaaoaom oapyaaoon

mm 0..”an

161

ammo.

HHH>

 

oooo.a
mmmm. oooo.a
mmma. oooo.a
mmmo. oooo.a
mama.- oooo.a
moom.- oooo.a
mmmmm. oooo.a
mmmm.a mmmm. oooo.a o
oaam. amma. mmmm. m
cmma.m oaam. ooOOea 4
mmao.a mama.- mmma. oooo.a u
mmmm. mamo.- mmmo. mmmm. m
ommm.m ommo.- ommm. oooo.a 4 aa>
mmmm.a mmmm.- amma. mmmm. oooo.a o
mmmm. mmma.- mmmo. mmma. mmmm. m
ommm.m ommo.- ommm. ommm. oooo.a 4 H>
ammm.a momm.- aama. mmom. mmma. oooo.a o
mmmm. mmoa.- ommo. mmmo. mmmo. mmmm. m
oamo.m omao. comm. ommm. ommm. oooo.a .4 >
mamm.a mmma.- maoo. mmmm. omma. mmam. oooo.a o
mmmoa mmmo... aaoo. mmmm. mmaa. mmma. ommm. m
oaom.m ommo. ommm. ommm. cmmm. omam. oooo.a 4.>a
mmmma mmma. mmoa. mmoo. .. mmmo. mmmo. mmma. .. cocoa o
mmao.a mmma. mmma. omoo. . momo. mmmo. mooa. . amom. m
ooom.m omam. ommm. omam. oaom. ommm. omma. oooo.a 4 aaa
mmmm.m momm. mmmm. mmma. mmom. mmmm. omom. moma. oooo.a o
ommm.m mmmm. mamm. mmma. mmma. aama. mmma. aama. mmmm. m
oama.m ommm. oamm. cmam. ommm. ommm. oamm. ommm. oooo.a 4.aa
ommm.m oamo: ommm. ommm. ommm. oamm. ommm. omom. cmmm. oooo.a a
Moomo .payo .yoym .yoym .aa 5.4% .aaomm .moom .aap4 .ooom
spa: .mom ooayoa . .. .momz .anz .oymm

 

 

mmmaqufl ca mmmmymzw 958 @695 whom mmapoammym no.“

mmooe mmmam yo opyoaoaymooo opom om» yom ooapsaom mappaaoon

mm mamm.a.

162

 

 

 

mmma.- oooo.a

mmmm. oooo.a

mmma.- oooo.a

mmmm. oooo.a

mmmo. oooo.a

mmom. oooo.a

mmmm.- oooo.a
ammma. ammm. oooo.a o
mmmm. mmmo. aomm. m
ommm.m ommm. oooo.a 4.aaa>
ammo.a mmma.- mmmm. oooo.a o
momm. ammo. .. mmmo. mmmm. m
om0m.m omam. ommm. oooo.a 4 aa>
mmmm.a mmma. mmmm. .mmmm. oooo.a o
mmmm. mmmo. mmmo. mmma. mmmm. m
omam.m ommm. ommm. ommm. oooo.a 4_a>
mmmm.a mmmo. omoa. mmma. mmoa. oooo.a o
mmam. amao. mmmo. aomo. mmmo. mmmm. m
ommm.m ommm. ommm. omom. ommm. oooo.a 4 m
momm.m mmmm. mmma. mamm. ammm. mamm. oooo.a o
mmmm.a maom. maoa. mamm. mmma. mmmm. momm. m
ommm.m ommm. ommm. oamm. ommm. ommm. oooo.a 4_>H
mmmm. mmma.- mmao. mmmo. ommo. mmaa. mmmo..- oooo.a 0
8mm. mmma. - moao. mmmo. mmmo. mmmo. mmmo. mmmm. m
ommm.m omam. ommm. ommm. ommm. ommm. omam. oooo.a 4 aaa
mmmm.m mmam. mmmm. ommm. mmmm. mmmm. mmmm. mmmm. oooo.a o
mmma.m oamm. ammm. mama. mmma. mmma. mmma. mama. mmmm. m
oaao.m ommm. ommm. ommm. ommm. ommm. omam. oamm. oooo.a 4.aa
0mmm.m oamm. oamm. omam. omom. oamm. comm. oamm. ommm. oooo.a 4_a
moomo .payo .moym .yoym .m.a .poom .aammm .ooom .aao4 .ooom

5d: . mam Guava . . . £002 .452 . 98> III
omdsmnda ca mommym>4.9naom undue mayao wnavoamoym you

.opooa pmmym yo

mm manda

myooyoayyooo ovum omv yoy ooayoaom oapyaaooo

163

 

mmmo... oooo.a

mmam.- oooo.a

mmmo. oooo.a

mmmm. oooo.a

mmmo. oooo.a

mmma. ooooza

mmma.: coco...”
mmmma mmmm. oooo.a o
mmmm. mmma. mmmm. m
ommm.m ommm. oooo.a .4 aaa>
mmma.a mmmo.- mmma. oooo.a o
momm. mmao.- ammo. mmmm. m
Ommm.m oomm. ommm. oooo.a 4 aa>
mmmm.a oama. amma. mmmm. oooo.a o
ammm. mamo. mmmo. mmma. mmmm. m
ommm.m omom. cmmm. ommm. oooo.a 4 H>
mmmm.a mmaa. aama. mmom. mmma. oooo.a o
mmmm. mmmo. ommo. mmmo. mmmo. mmmm. m
ommm.m omom. comm. ommm. ommm. oooo.a .4 m
ooma.m mmmm. maoo. mmmm. omma. mmam. oooo.a o
aamm .a mmma. aaoo. mmmm. mmaa. mmma. ommm. m
oamm.m ommm. ommm. ommm. ommm. omam. oooo.a 4_>a
mmom.a ommo. mmoa. mmoo. . mmmo. mmmo. mmma. - oooo.a o
mmmm. mmmo. mmmo. emoo. . momo. mmmo. mooa. .. amom. m
ommm.m omma. ommm. omam. oaom. wommm. omma. oooo.a 4_aaa
mmmmm mmfln. mmmm. mmma. mmom. mmma. 30m. :09”. 0000...” o
m moo . m wamo . mam: . mmma. 33H. H.R..” . mama . H.R.—u . mwmb. m
ommm.m ommm. oamm. omam. ommm. ommm. oamm. ommm. oooo.a 4.aa
ommm.m ommm. ommm. ommm. ommm. oamm. ommm. omom. ommm. oooo.a a

mommo .mayo .yoym .moym .m.a .poom .aaomm .moom .aag4 .moom

. .. .. . floz .myym quaymm . . moo: .asm .nyms t,

 

 

opy4 aoayomomya ma oomoyom4 poaom oooyo whom myapoamoym you
mmooa pmmam yo opnoaoayyooo opom mam yom ooamsaom oaypaaoon

om oaoma

16h

HHH>

HH>

 

mmmm.- oooo.a

mamm. oooo.a

momm.- oooo.a

mmam.- oooo.a

mmmm. oooo.a

mmam. oooo.a

maom. oooo.a
mmam. mama.- oooo.a o
mmmm. - mOmo.- aomm. m
00mo.m omom. oooo.a 4
moom.a mmmm. mmmm. oooo.a o
mmmm. mmmo. +38. mmmm. m
oomm.m ommm. ommm. oooo.a 4
ELM .N mwmm . Pmmm. mem. OOOO.H O
memm. moma. mmmo. mmma. mmmm. m
ommm.m oaam. ommm. ommm. oooo.a 4 am
mmmm.a mmmm. omoa. mmma. mmoa. oooo.a o
mmmm. mmma. mmmo. aomo. mmmo. mmmm. m
0mm; 83. ommm. omom. ommm. 88a 4 m
mmoa.m ®:®O.I mmma. mama. Hmmm. mflhm. oooo.H o
mmmm.a ommo.- maoa. mamm. mmma. mmmm. momm. m
omma.m Omam. ommm. oamm. ommm. ommm. oooo.a 4 >H
amao.a mama.- mmao. mmmo. ommo. mmaa. mmmo.. oooo.a o
mmmm. moaa.u moao. mmmo. mmmo. mmmo. mmmo. mmmm. m
ommm.m omma. ommm. ommm. ommm. ommm. omam. oooo.a 4 aaa
ommm .m ammm. mmmm . ommm. mmmm. mmmm. mmmm. mmmm . ooooa o
mamo.m ammm. ammm. mama. mmma. mmma. mmma. mama. mmmm. m
ommm.m ommm. ommm. ommm. ommm. ommm. omam. oamm. oooo.a 4 Ha
ommm.m ommm. oamm. omam. omom. oamm. comm. oamm. ommm. oooo.a 4 a
Moomo .payo .moym .moym .a.a .poom .aaomm .omom .aan4 .moom

. gym: .mom omayoa .. .nooz .EJMI. .nyo>

 

 

moaaonoom maom
mpmoe mmmam yo

.qmmagma

ma mmmdum>< pqaom mmmyo mayao mmapoammym how
mpamaoammmoo mpmm map now noapdaom mapvaaoon

165

 

maaa... 88a
mmma. oooo.a
mmmo. oooo.a
HHNH. OOOO.H
mmmo... . cocoa
mmmo. oooo a
mmao. cocoa
Mmmma mmmm. oooo.a o
mmm. mmoa. aomm. m
omma.m oamm. oooo.a 4. aaa>
ommm . a mmoa. mmmm. 88. a o
aaom . mm mo. mmmo. mmmm. m
ommm.m ommm. ommm. oooo.a 4 a:
ammma mmma. mmmm. mmmm. oooo.a o
mmmm. mmmo. mmmo. mmma. mmmm. m
ommm.m ommm. ommm. ommm. oooo.a 4 a>
mmmma mmmo... omoa. mmma. mmoa. oooo.a o
mmmm. maao... mmmo. aomo. mmmo. mmmm. m
oammm oamm. ommm. omom. ommm. oooo.a 4 >
mmmm .m mmma. mmma. mamm. ammm. mamm. oooo.a o
mmmm .a mmaa. maoa. mamm. mmma. mmmm. momm. m
oomm.m omom. ommm. oamm. ommm. ommm. oooo.a 4 B
mmmma mmmo. mmao. mmmo. ommo. mmaa. mmmo. .. cocoa o
mmmm. mmmo. moao. mmmo. mmmo. mmmo. mmmo. .. mmmm. m
ommm. 8mm. ommm. ommm. ommm. ommm. omam. 88a 4 aaa
mmam. mmmm. mmmm. ommm. mmmm. mmmm. mmmm. mmmm. 88a o
3.8. m ammm. ammm. mama. mmma. mmma. mmma. mama. mmmm. m
omom.m 8mm. cmmm. ommm. ommm. ommm. omam. oamm. cocoa 4 Ha
ommm.m oomm. oamm. omam. omom. oamm. 8mm. oamm. ommm. oooo.a 4 a
Moomo .payo .yoym .yoym ..oa .poom .aaomw .moom .aap4 .ooom
. 5oz . mmmw Eeyoa . . . moo: .mmymm . my!

 

 

noavdodmm maoaamdm ma mommym>< Pnaom omdyu mdyao wdapoamunm pom
Boom. woman yo upooaoayyooo $3 2a. yoy ooapzaom oapmaaoon

mm canoe

APPENDIXB

167

 

 

 

 

mmmm.- a o
mmmm.m .az mummm.
HEN . Ind Nmnmxmm.
mmmm.- mm.mm aoao. mmmo. mmma. omm. mamm. .yoym apoz
moao.- mo.moa aooo. mmmo. omoo. mmm. mmoo. .yoym .mqm
ommm.- mm.moa amao. mmmo. mmmm. mmm. mmmm. .m.a ooayoa
mmaa.- mm.mm mmoo. mmmo. mmmo. mmm. mmmo. ooooopoom
mo a.- mm.mm mmoo. mamo. mmmo. omm. mmma. moaaaomm
oo o. aa.mm maoo.- mmo. mmoo.- aam.. mmao.- .ooom .mooz
ammm... mm.ma mmmo. mmaa. mmma. amm. mmmm. 5234 .32
moma. mm.ma moao.- mmmo. mmmo.- mmm. aoaa.- .ooom aopyo>
oayao

amam.
ommo.m .az mummm.

mmmmaé .w . mmuomom.

mmmm.- mm.mm mmoo. oomo. mmma. omm. mamm. .yoym gym:
mama.- mm.ooa maoo. momo. mamo. mam. mamo. .moym .mom
mmmm.- mm.moa mmoo. memo. mamo. omm. mmma. .m.a omayoa
mmmo.- mm.ma maoo. mmmo. mmao. mmm. mmmo. omooopyom
ommo.- mm.mm mmoo. mmmo. mmmo.. mmm. maoa. mnaaaomm
mmmm. mm.mm moa .- mmmo. mamo.- mma. mmma.- .moom .nooz
momm.- oo.ma mmao. amoa. momo. mmm. mama. mpaaao4 .ssz
aaom.- mm.ma mmao. mmmo. ommo. mmm. mmma. .moom aopyom

mNom
z . oapoayo>
mapamz-m A: man .\9n\ maymam may mam yopoamoym

 

 

WHmHU 9,2 mwom mom mmHAUzm ZH mmom‘mmca BZHOA @053 MO

ZOEOHQBE mam. mom mEHOHEOO ZOHmmmmmxmm a ho ZOHanom

mm mam4e

 

8
IO
1

 

 

 

 

mmom.- . o
omma.m uaz mu mmm.
mmm.. w mm.mmmy...
mmmm.- am.0m mmao. mmmo. amma. mmm. mmmm. .yoym mpoz
ooao. mm.mm aooo.- mmmo. maoo.- mmm. amoo.- .yoym .mqm
mmmm.a- mo.moa mmao. mmmo. mmma. mmm. cmmm. .m.a qoayoe
mmma.- mm.am mmoo. mmmo. mmmo. amm. mmaa. moooopyom
aamo.- mo.0m mmoo. mmmo. mamo. mom. mmmo. moaaawmm
omao.- mm.am mooo. mmmo. omoo. mmm. mmoo. .moom .mooz
mmmo.- mm.ma mmoo. mmaa. mmao. mmm. ammo. mpaaao4 .ayz
mmoo.- om.ma mooo. mooa. maoo. mmm. mmoo. .moom aopyom
mayao
omwool I d
oama.m .a2 4. omm.
OOHN.NIH ‘ Nmnmmomo
mmom.- mm.mm oaao. mmmo. mmma. mmm. ammm. .moym mum:
momm.- om.mm mmoo. 0mmo. ammo. mmm. omaa. .moym .mom
mmam.a- mm.moa amao. mmmo. moma. mmm. mmmm. .m.a omayoa
mmma.- ma.ma mmoo. aomo. mmmo. amm. mmaa. mooyopoom
mmma.- ma.mm mmoo. mmmo. mmmo. amm. mmma. moaaaomm
omma. am.mm mmoo.- mmmo. mmao.¢ mma. mmmo.- .ooom .mooz
mama... am.ma mmoo. omaa. mmmo. mmm. ammo. , .3234 .52
mmmo. mm.ma mmoo.- mmmo. mmmo.- mmm. mmmo.- .owom aopyo>
mNom
J4 4 magmaydym
mapamz-m m: map .xomxx maymam may mmm yopoammym

 

 

maHmHU de mwom mom mmHQDBm 300m ZH agd BZHQH mega ho
ZOHBDHQMmm mmm. mom mBszHOHEOU ZOHmmmmomm Mme ho ZOHBDAOm

a: mam4a

169

 

 

 

 

 

mmma.- a

H+~+~ o N I 2 ml o

mmmm.m-n.w mmummmw.

aomm.- mm.mm mmao. mamo. mmmm.. mmm. mmmm. .yoyy gym:
mmmm.- mm.moa mmoo. ommo. ommo. omm. oooa. .yoym .mom
mmma.a- mo.maa moao. oamo. mmma. amm. mmom. .m.a oaayoa
mmmo. om.mm aaoo.- mmmo. moao.- mmm. mmao.- mooyopoom
maaa.- am.mm amoo. mmmo. mmmo. mmm. mmoa. moaaaomm
mmmo.- am.mm mmoo. mmmo. mamo. mmm. momo. .omom .nooz
mmmo.- mo.ma mmoo. maoa. ammo. mmm. mmmo. myaaag4 .asz
mmoo.- ma.ma mooo. oomo. mmoo. amm. mmoo. .moom awoyom

oayao

ommm.- . o

oomm.m uaz mu mmm.

cmam. -u_w mmummmm.

omom.- mo.om amao. ammo. mmmm. amm. mmmm. .yoym gym:
mmmm.- mm.moa mmoo. mmmo. mamo. mOm. mooa. .yoym .mnm
ammm.- om.aaa mmoo. mmmo. mmmo. amm. mama. .m.a omayoa
mmma.- mo.mm mmoo. mmmo. mmmo. aom. mmma. mooooyyom
mmoa.- mm.mm mmoo. mmmo. mmmo. mOm. mmaa. myaaaumm
mamm.- mm.mm mmao. ammo. mmmo. mmm. mmma. .ooom .nooz
ammo.- mm.ma maoo. mmaa. mmoo. mmm. maao. myaaao4..asz
mmma. mm.ma mmoo.- mmmo. mmmo.- mmm. mooa.- .moom ampyom

mNom
m . oapoayo>

mapmmzap m2 map \mA\\ maymam may maml yopoamwmm

 

 

mquw Q24 mwom mom HDZMHow 2H mmu¢mm>< 82Hom mn<mw ho

ZOHEoHnmmm Ema mom mEZMHOHmhmoo onmmmmomm Ema ho ZOHEDuom

mummam4y

170

 

 

 

 

 

 

mmom. u o
omm.m naz my mmm.
mamm .m-u w mmummmm.
mmmm.- mm.mm mmao. ammo. mmma. amm. mmmm. .yoym gym:
mmma. aa.maa maoo.- mmmo. aomo.- mam. mmmo.- .yoym .mym
mmmm.- am.maa omoo. mmmo. mmmo. mmm. mmmo. .m.a omayoe
mmoo. mm.am mooo.- mmmo. maoo.- mmm. mmoo.- mooooyoom
ommm.- om.mm ooao. momo. mmom. mmm. momm. myaaaomm
mmmm.- mm.mm mmao. mmmo. ommo. mom. omma. .moom .mooz
cmmm.- am.om mmmo. moma. mmma. mmm. amma. mpaaaa4 .asz
mmmo. mm.am maoo.- mmmo. aoao.- mmm. mmao.- .moom aoyyom
mayam
2%-.6
cmma.m .az mu mmm.
mmam.m-u u mmuomam.
mmmm.a- mm.am ommo. mOmo. mmam. oam. mmam. .yoym myoz
mamm. mm.moa mmoo.- mmmo. mmmo.- mom. mmma.- .moym .mym
mmmo.a- om.maa mmoo. momo. mmmo. mmm. mama. .m.a omayoe
mooa.- mm.mm mmoo. mmmo. mmmo. mmm. mmmo. moooopaom
aomm.- ma.mm mmoo. mmmo. mmmo. aom. mmam. moaaaoym
omaa.- mm.mm Omoo. mmmo. mmoo. amm. mmmo. .mwom .mooz
mmmo.- cm.ma amoo. omma. mmoo. mmm. mmao. mpaaap4 gaoz
amma. mo.ma mmoo.- mmma. aomo.- mmm. mmmo.- .moom aooyom
whom

.4 . oaowayo>

mafimz.M ya man xmxxx maymam may mam yopoammym

 

 

mquw 92¢ mwom

 

 

mom mOHB<EmmB¢E 2H mmu<mm>< BZHom mm¢mu %o

ZOHEOHQmmm mmB mom mBZHHOHmhmOo ZOHmmmmwmm mma ho ZOHBDAOm

04 mqm¢H

171

 

 

 

 

mmmm. - o
ooam.m «a: m. mmm.
mmaam-.. N mm.mmmm .
mmmm.- mm.mm mmao. oamo. mmma. mmm. ammm. .moym mpoz
mmmm. ma.maa omoo.- mmmo. mmaa... mam . mmmm... ..yoym mam
mmam.a- mm.maa oaao. ammo. maaa. mmm. mmom. .m.a ooayoe
mmmo.- mo.mm maoo. ommo. mmao. mmm. mmmo. moooomoom
mmom.- mo.mm mmao. ommo. mmmm. mmm. mmmm. mmaaaomm
mmam. mo.mm mmao.- ammo. mamo.- mam. mmma.- .moom .mooz
mmmm.- mm.om mmmo. mmaa. oomm. mmm. mmmm. myaaam4 .auz
mmmm. am.mm mmao.- mmmo. mmmo.- amm. mmma.- .mamm aomyo>
mayam
mmam.m u m
ommm.m «am mu mmm.
mmmmwu "nu mmmmmmwn
mamm.a- ma.om mmao. mmmo. mmma. aam. mmmm. .yoym mpmz
mmmm.a mm.oma maao.- mmmo. maao. mmo.- mmmm.- .moym .mmm
mmmo.m om.maa amao.- ommo. mmoo. mmo.- moom.- .m.a ymayoa
momm. mo.mm ooao.- mmmo. mmoo.- mao. mama.- mooyoyoom
mmaa.s ma.mm mmoo. mmmo. mmoo. mmo. mmmo. moaaaomm
mmam.- ma.mm mmoo. mmmo. mmmo. mam. mmma. .mmom .momz
maao.a- am.ma mamo. mmmo. mmmm. mmm. mmmm. myaaam4 easz
mmmo.- mm.am amoo. mamo. maoo. amo. mmmo.- .moom awpyom
oNom

k 4 mammaym>

maohmz.M mm map mmm‘x maymam may mam yomowwoww

 

mum mam4a

mquu 92¢ mwom mom mmu¢mm>¢ BzHom mm¢mc mu<DU2¢A 20
2OHBUHQmmm 228 mom m822HOH22moo 20Hmmmmwmm 225 20 2OHBDAOm

172

 

 

 

ammm. . o

omom.m mam mu mmm.

E... .H mmumoom.

mmma.a- mo.mm mmmo. mmmo. mama. mmm. mmmm. .moym mmmz

mmmm. mm.mm mmoo.- mmmo. momm.- omm. mmma.- .moym .mnm

mmma.a- mo.moa maao. ommo. mmmo. mom. mmma. .m.a ooayoa

mamo.- am.ma mmoo. mmmo. oamo. mom. mmmo. moooomoom

mmom.- mm.ma mmao. mmmo. ommo. mmm. mmmm. myaaaomm

amma.- mm.mm mmoo. mmmo. mmao. mma. mmmo. .mmom .momz

mmmm. mm.ma mmmo.- mmma. mmmo.- mmm. mmam.- mmaaao4_.asz

mama. am.ma mmoo.- mmma. mmao.- mmm. mmmo.u .moom awpyo>
r . manmmam>

mamnmzuv m2 map tm\\\ maymam may mam yomoamoym

 

 

mm mam4e

mwom mom memd AdeHmDQ2H 2H mmu¢mm>< B2Hom 22420 20
2OHBOHQmmm 228 mom mE2mHUH22HOU 20Hmmmmumm 228 20 onabqom

173

 

 

 

moHSo2oom 2202 2H mmgd E2Hom mac .20 205“.ng mmm. morn.
2OHB<AEOO EH9; 9,2 mezmHOHEOo 20memmomm E .20 ZOHBDAOW.

mm mamas

 

omam.- n m

oamm.m "a: mu omm.

omoo.m... w mmummmm.

momm. mm.mm mmoo.- mmmo. mmmo.- mom. mama.- .yoym mum:

mmma.a- om.mm mmao. m0mo. mmma. mmm. maom. .moym .mom

mmmm.m- om.mm ommo. mmmo. cmam. aam. mmam. .m.a qmayoa

ammm.- mm.ma mmao. mmmo. mmmo. mam. mmmm. mooaomyom

moam. am.om omao.- mmmo. mmmo.- mam. mmam.- myaaaomm

mmmm. mm.ma mmao.- momo. mmmo.- mma. momm.- .moom .mooz

mmmm.- am.oa mOmo. mmaa. mmam. mmm. mamm. mpaaam4 .amz

mmmm. ma.aa mmmo.- moma. mmmo.- mmm. mmmm.- .ooom awmyom

ma m ma wwxw\ ma ma ma ma oapmayom
mm z-~ 2 p rrl y .m y m yopoamoym

mamam mom

 

III}

17h

 

 

mamw. u m

ommm.m .az mu mmm.

mmmmanu N mmuomam.

mmmm.- mm.mm mmao. mmmo. mmama amm. mmmm. .yoym mymz

mmmo.- mm.mm mooo. mmmo. mmoo. mmm. mmao. .yoym .mom

mamm.- mm.moa amoo. mmmo. mamo. mmm. mmmo. .m.a qmayoe

ommo. am.om mmoo.- mmmo. mmmo. amm. mmmo.- mooqopyom

mmmo.- mo.mm mmoo. ammo. mmmo. mom. aama. mnaaaomm

momo.- . mm.om omoo. mmmo. mmao. omm. mmmo. .mwom .mooz

mmmm.- mm.ma momo. mmoa. mmmo. mmm. mmma. mmaaam4 .emz

mama. mm.ma moao.- mmmo. mmmo.- omm. maaa.- .mwom aonyom
r oamaayg

mafimz.m m2 man \\..\ ma ymam may mam yoyoamoym

 

 

om mam4e

mJEHo mom 20He<obmm mmmambm 2H magma? E8 @9420 .20
2OHHOHQHE 222. 202 mEEMHUHEOO 20Hmmmmomm mmm. 20 onBRHom

APPENDIX C

m‘i- i
I f”

*__ 130
t:
{Ci—125
C:
:-
'L-leo
E—lls

llill'TFlllllWIWIIIIIHIIIHIIIH
\0
O

.Tqrm-McNm Test .of.Menm.Ah111ty ..(I-Q. Score) -

CD (I)
O \n

4
\J'l

 

-—"150

—1h-0

 

   
   
   
 
  

135

130

 

 
     

'P-'3.0

 

§°
0
4:
*5
'8'
n. 2 O
120 o'_—- '
E?
m
115
z. 33
5 a
rdF——'llO
3 E
E~ 105 o_ 1 5
$1 $2 ‘
'_"”‘100
h 95
—"" 90
*“'l.0
Figure 1

l ' 'mmW-Ripfilpllof4cFieEJcY1quPj (Ra; §E°rl_e)_l 1.4 L1 1 L l 1 L 14 1.; l 1 1 1

_ A A

gl‘\|!.lll

HIIIHH
..q
\n

[Flinn]
.1
UI

LL

gm
0

IIrTrALIHTIHHPrIIgIIIl'HHIrmll g
8 \n ‘8 % \n ‘0" 3:

llll

|-’
U1

 

A meogruph for the Prediction of Boy's Total Grade Point Average
and Total Score on the EHSCB from the Terman-McNemar Test
of Menta1.Ab111ty and the Mathematics Proficinncy Tests

176

177

 

 

 

 

a m s o s m ...» m m... .o. m m m m .1...
_______m__~________m__m_m___a__: m,: _ a__ ~___ __ ma___m_m____agm_._ m .21—
?yoom Wauoa. oayoyrfiamo oym 333394:
e
a
r
O
M
o ...; o ..., o
3 2 2 l l
1 ...... IL: m 1
, 445.8 oooyu amao.a ofioaooym
..
1..
..
5 0 5 0 5 O c) 0 :J 5
m. a .... ...... ...... m .... n m. m 9
_ _ _
_ m _ _ m _ m * _ 2 _
Aoyoom oyooqfimv oyoom ma ammo.a momma ofioaooym
m m m m m m m ..m. a... m m m o

-—- 11.5
::'lh0

 

:_ _227:;—ILZZTZLZZ_:C—:: _EmbC;_:L::r:;

 

mayoom ..a am 33.34 35: yo away. gastgyyoa

Figure 2
Test of Mental Ability and the Mathematics Proficiency Test

A Nomograph for the Prediction of Girl's Total Grade Point
Average and the Total Score on the EHSCB from the Terman-MoNemar

Terman-McNemar Test of Mental Ability (I.Q. Score)

~—l’+5

 

HIIIHH lllilllll

135

:-l25

 

5
o

 

115

llll llll ll!

 

105

 

§

 

 

 

 

ITI'H Illl Ilil HFFIIH llll HH
8

4
\J'I

 

A Nomograph for Predicting Boys English Grade Point
Averages from the Terman-McNemar Test of
Mental Ability and the Mathematics Proficiency Tests

Predicting English Grad Point Average

+——3.o

V81?
W

"1.0

 

Figure 3

Test (Raw Score)

IIIW‘HHWHI‘ITHlIIIIIHIIINI[IIHIIHIIIiHTIHIIlIHT'HHITIIW
\fl

Mathematics Proficiency

CD
U1

(D
0

‘1
U1

‘1
O

0\
\n

O\
0

U1
0

4:-
\n

5

w
U1

L»
O

J,

8

H
\n

 

178

‘I'erman—McNemar Test of Mental Ability (I.Q. Score)

 

179

 

 

 

 

-—1h5 _ 3.5 r85
~——1ho : 80
[I
—135 : 75
I 130 ”3" E—70
E E
:‘125 :‘55
5 3E
:- 120 g) £9360
[-115 é “:55
_ ..p vy—
: M—
M 3350
L— '8 o—
: y 8:.
105 a a 1&5
: .ca ""2'0 .3:
.... on ‘8..—
"—100 3 tiff-1+0
I ea.
2 rd «3:
.:_ 95 .3 §;—35
:2 3 ,5:
._ 0 __
—- 90 5: b 1-5 .53—30
E L:
C :
_—_— 85 :‘25
p— L...
E— 80 {—20
I —--1.0 T:
— 75 —-15
Figure 1|-

A Nomograph for Predicting Girls English Grade Point
Averages from the Terman-McNemar Test of
Mental Ability and the Mathematics Proficiency Test

Terman-Manor Test of Mental Ability (LQ. Score)

HT! HITI‘HII‘IflllllflIHHIHHIHHIIIIIIIIHIHIIIIIIIIIIII

a

-—1h5

lllll

H
5

 

 

75

 

E
4
E
8

  

Predicted Social Studies

 

‘— 3.0

g 2.0

- 1.5

—_ 1.0

 

Figure 5

(D
\n

(D
O

4
U'I

.1
O

0\
U1

 

A

  

A Nomograph for Predicting Boys Social Studies Grade Point

Averages from the Temn-McNmr Test of
Mental Ability and the Mathematics Proficiency Test

HIHIIIFTITIIHTUHI

Mathematics Proficiency Test

 

180

181

 

 

 

 

 

 

5 O 5 O 5 [w 5 5 O 5 0 5 5
8 8 7. 6 h. 3 3 2 J l
.r_._ a; _____a_ _.m___ ara__ ______.___ __: _: E I; _____.__ ._.___ _ _.__
3.30m #65 .....m .H. hokoaoa Roam moans—0562
8
e
a
r.
e
m
Aw cu am no nw cu Am cu
h. 3 3 2 2 l l
_ _ L h _ - m _
omoyo>4 pyaom oooyo o 95m aoaoom oopoaooym
emmngmsm... meymye
_______n_ ____ ________ma_amm_______mmmr mm_________m__________________
.20 8m may .3334 away o2 yo Soy. yosoeyozéoayoa

 

Figure 6

A Nomograph for Predicting Girls Social Studies Grade Point

Averages from the Terman-McNemar Test of

Mental Ability and the Mathematics Proficiency Test

Terman-McNemar Test of Mental Ability (I.Q. Score)

AFT?!

]:H1VIIW7UT|HIWTHI

!!F{

 

H
[0
U"!

...I

ll]! lI'Tfl"FIV‘IT

I T
\\\§;\

P‘ F’
8 a E
Predicted Science Grade Point A erage

['0
O

\0
U1

\0
O

(I)
U1

(1)
O

.4
U1

2 5

 

1.0

 

Figure 7

re)

I
Raw

PIIWIHI[HTHIIH]HTTIWTI1HYiHTHIIHIIHII)H[TH| HIII‘WI

Mathematics Proficiency Test (

A Nomograph for Predicting Boys Science Grade Point

Averages frcm the Termsn-MCNemar Test of

Mental Ability and the Mathematics Proficiency Test

 

 

 

 

 

 

(I)
\J‘l

(I)
O

.4
\n

.4
O

O\
\J'I

O\
O

\J'I
\n

\r.
O

4:—
\n

5

U)
\J'I

Lu
0

N
U1

8

...;
U1

182

 

 

 

 

 

 

 

f ”" 3-5
{——1b.5
t:
t—lho
F—
I:
17135
t: F— 3.0
1713—130
8..
“3 C
.. 125
0.5—-
5: 1:
h— “I
fl9:120 §_ 2.5 3
r4:: fi‘—fl—fl—flfl_“§32£§59..—a—e.ufi
B<~—- 5
<11‘:-‘-"ll5/£V 3:
r+__ -p
E— ’3' ‘2?
ar—-llO 04 E—"
z I o h
“a: g 20
stiles 0— ' '3
a 5 8 3;"
“ “ 100 8 E
g ._ ”3
st: 3 3‘3
__ +L
’5 a 95 g a
__ H
.5
EE: c:-- 1.5 =5
24 90 A: §~
::
--u ‘ C35
I
E- 8o
- ‘~——l.0
Figure 8

 

HI

,. 70
L— 65
:60
E
:7-55
l. 50
l.
p__
L ’45
Lao
:‘35
:30
‘:-25
t..—
F.
‘L20
E
._15

 

A Nomograph for Predicting Girls Science Grade Point

Averages from the Terman-McNamar Test of

Mental Ability and the Mathematics Proficiency Test

183

8’

.4
U1

.4
O

0\
U1

0\
O

\11
\D

 

U1
0

 

Spelling Test (Raw Score)

TTlllTli TITTTI Ifll IWHFI illl Il1ll11l11Ir1llllll‘llllll IITI

4:-
U1

 

5

 

U)
\J'I

w
0

l1)
U1

r—-
p—
r
P'—

—10

 

 

-—4yo
— 2L5

0

to

G3

‘4

:‘i

4:

.p

£-—-2x>

0

"'3.

U

a

-3

E

5-— 1.5

E!

'd

3

U

13

0

&:
~——-1uo

 

Figure 9

Mathematics Proficiency Test (Raw Score)

|IIII|IIII IIIIilTITIIIII Tll1 Tlllllll1[1TT1lllTlll11T

60

U1
\11

\J'l
O

4:-
\n

 

’6'

 

U.)
U1

U)
0

'0
U1

 

DO
0

1...:
\n

p
O

 

A.Ncmograph for Predicting Boys Mathematics Grade Point

Averages tram the Spelling Test and the
Mathematics Proficiency Test

18h

185

 

 

 

am «I, m. a.» w mu m, M, m 5 m 6 m b w.
___:__:___-E__::___:_::r:__::__EL::___:_:___::__EL
thoom swmv honvaoaMOHA moapdavnvmz
5 o. 5 o .0 < o
3 3 2 1 .
_ _ L H _ l
03.85 0.38 «8.6 augments: Aconcagua
m 4.0 m. a.» a w m, M, .0. fi m. 5 m m m.
__ ___________ ______—____ ________ ____‘ _____.P__ ____> ___ __ __ w._ _____

 

 

 

 

 

Aouoom sumv Pena waHHHomm

Figure 10

Averages from the Spelling Test and the
Mathematics Proficiency Test

A NOmograph for Predicting Girls Mathematics Grade Point

 

Mechanical Reasonin Test Raw Score)

 

E-ao
F-
:25
:
'_'_—20
E
E—l5
:2
—:L_ o
E- 5
2. 0

 

186

F 3-5
_ 3O

.0 ’-
3 ‘_25

‘1 2.5

A Nomograph for Predicting Boys Foreign Language

 

 

1
4;

Numerical Ability est (Raw Score)
‘ I

 

 

Figure 11

Grade Point Averages from the Mechanical Reasoning

Test and the Numerical Ability Test

r:
E
L—
L:

50

$83

8‘

\n
\11

 

 

 

 

 

 

 

 

’TgE
8:45
CD.—
a:
32 to
Mr—
*3:
c3__ 35
,3":
8 30
d2:
01-
(5..
'a’L—25
O—
H—
g_
.3__ 20
>3:
_. 15
+_ 10
__ 5
z. o

  
 

 

 

Figure 12

A Nomograph for Predicting Girls Foreign Language
Grade Point Averages from the Mechanical Reasoning
Test and the Numerical Ability Test

*'—- h.0
to
d
3...
0
2
g— 3.5
H
0
° ‘2 a
O
E .3
ON
3 4,
3.0 m
3 :2
fl 5’
g 3
so g
'd
:3 ‘3
° 2
gL 2.5 g
:13 z
—— 2.0

'F— 20

 

187

188

 

 

 

 

 

 

 

 

 

 

 

O 5 O 5 O 5 O 5 O 5 O a) 0 5 O
8 7 7 (My 6 5 5 .4 .4 3 3 2 2 1 l
.2: :_L:.:..EL.C__r:_::b:.rFZTZLIZ__:;_:__:Z 2:
3.33 5 name hoqofiofloam moflpwemfimz

S

.3

ma

um

18

um

ahorn...

de

mam.

0 Ru 0. se

7 , t B new

8383‘ p58 03.6 3.2 SE» 3 uopfiewna m m. n

s as

.1 is

F 3.

Ha

rm

.mt

hm

Po

mm

mm

NG

A
.w ,6. N «w m w m, M; m fl w so , w m
ELEL:_,;C:_::_::_::_E;::_EL_E_::_._ LE;

Avhoom mev PmmB JUCwHHQO

Mathematics Proficiency Test

[..n
4:
W

H
4:-
O

....
U.)
U1

...:
w
O

53

...
8

E

g...
[...I
O

H
O
\11

100

i— 3.0

 

Terman-McNemar Test of Mental Ability (I.Q. Score)

[TIHPIHIHH [WWW] HIHIIITIHIHWIIIHIHHIHHIHHIHHIHIII

 

95

85

80

75

Predicted Home Economics Grade Point Average

L— 2.0

 

Figure 1h

Av

Numeric

A Nomograph for Predicting Girls Home Economics

Grade Point Averages from the Terman-McNemar Test
of Mental Ability and the Numerical Ability Tests

 

HITITIIIHT[TIIIIIIIIIllll

Ability Test Lnaw Score)

 

'[IHII

.3:—
O

on
\J'l

U.)
C

25

8

[.1
\11

...I
O

189

190

m t. m m s o

m.) m U
r_______r_____t_r_______a___pa ___ul___u

1‘

Aouoom ~55 pace .0: ca. 33.3852

 

 

nw
2

2.5
1.5

F'— 3.0
F—

 

omeuoea. 930m 83.5 soavaod e353?! oopoacoum

 

 

 

 

kw % B mw hm w my 0 .% m «w W 3, mm B
__:L::_::_::___:_:it: ___L::_:: :: _:L_:__::_

Aonoom ~35 amen. Honofloaonm 333.05%

Figure 15

A Nomograph for Predicting Girls Business Education
Grade Point Averages from the Mathematics Proficiency

and Numerical Ability Tests

APENDDC D

192

 

wxmqao.+~xmooo.+oxaaoo.+mxmqoo.-axamoo.+mxomoo.+Nxsomo.+flx:oao.u mama
mxmooo.-axmmfio.+oxoqmo.+mxmmao.+£xomao.umxomao.-mxmomo.+axmmmo.u «Hex
mxmmmo.+~xmooo.-wmeHo.+mxoooo.+gxqefio.+mxmmoo.+quomo.+axamoo.- uamx

wxqqao.+axowoo..wxoaflo.+mxmaoo.+axmmfio.+quqao.-mxwmgo.+axmmao.u uaux
wxqmfio.+~xmfifio.uoxaafio.-mxooao.-qxwmoo.+mxomoo.+mxofimo.+axamoo. uamx

wxmmao.+~xqaoo..oxo:oo.+mxmooo.Lgxooao.+mquao.+mxmmmo.+HN©Hoo.u naux
axoomo.+exmooo..wxmmoo.+mxmqoo.+gxmmoo.+mxomoo.+mxamoo.+axmmoo.n uamx

mxaqao.+~xmmoo.+oxqoao.+mxaaoo..gxamoo.+mxmmoo.+mxmmoo.+axqooo. uaox
wxamao.+sxmmoo.+oxmooo.+mxm~oo.+axamoo.+mxmmao.+mxmfioo.+fix~moo.u uHmN

mxmwao.+~xfiooo.-oxamao.+mxoooo.+qxsmoo.+mxoooo.+Nxmgoo.+axmooo. uaox
mNOflHo.+~xsmoo.+omeHo.+mxmooo.+qxsaoo.+mxmmoo.-mxmmoo.+fixomoo.u ume

wxfloao.+~xaooo.+oxaafio.+mxo:oo.+axmgoo.+mxmaoo.-Nxmmmo.+HxNOHo.u aflox
mxsmoo.+~xmaoo.+wxmsoo.+mxmaoo.+zxmmoo.+mxmoao..NN4:HO.+memHo. uamx

wHHHU
wHMHU

whom

wasfiw
whom

wanflc
whom

wanflo
whom

wahflo
whom

wahfic
whom

coapmodom wwocflwsm

woflsozoom wEom

wand HwflppwdosH

owmsmcwm :wfionom

mafipmsmgpmz

wonwflom

mmficspw Hsfloow

nmflfiwcm

 

wCOflpwsmm

 

 

Kmm

mesa pomwnsw

 

 

 

 

mmmmd Homwmbm BmUHm 2H mmw<mm>¢ 92Hom mm<mo
mquo mz<.mwom OZHBonmmm mom mZOHemmom ZOHmmmmo

Hm mmmda

mm

x‘ ; ,
a “ a 4". " "‘ .. ‘5: 3 ’
. ' . .. . '
-.1 .1. DA is: ' 'l "

./

 

               

 
 
 

HIGH 9 T T

16 N S A E UNIV. LIBRARIES
IV HHII! IllIIIWIN||1|||1INIWIHIIWHIIHI
9 1 02889536

312 3