I
I

7 1 - 1 1 ,7 9 3
BRANDT, R ic h a r d P a u l , 1 9 4 2 THE RELATIONSHIP OF SELECTED PREADMISSION DATA
TO GRADUATION, MEASURES OF GRADUATE PERFORMANCE,
AND DEPARTMENT PROFILES OF COLLEGE OF EDUCATION
MASTER'S STUDENTS AT MICHIGAN STATE UNIVERSITY.
M ic h ig a n S t a t e U n i v e r s i t y , P h . D . , 1 9 7 0
E d u c a t io n , h i g h e r

University Microfilms, A XEROXCom pany, Ann Arbor, Michigan

THE RELATIONSHIP OF SELECTED PREADMISSION DATA TO
GRADUATION, MEASURES OF GRADUATE PERFORMANCE. AND
DEPARTMENT PROFILES OF COLLEGE OF EDUCATION
MASTER' S STUDENTS AT MICHIGAN STATE UNIVERSITY

By
R ichard P. B randt

A THESIS

Submitted to
M ichigan State U niversity
in p a rtia l fulfillm ent of the re q u ire m e n ts
for the d eg re e of

DOCTOR OF PHILOSOPHY

College of Education

1970

ABSTRACT
THE RELATIONSHIP OF SELECTED PREADMISSION DATA TO
GRADUATION, MEASURES OF GRADUATE PERFORMANCE, AND
DEPARTMENT PROFILES OF COLLEGE OF EDUCATION
MASTER' S STUDENTS AT MICHIGAN STATE UNIVERSITY
By
R ichard P. B rand t

Statem ent of the P ro blem
While th e re is co n sid erab le d ifferen ce of opinion a s to what
im portance should be attached to the M a s t e r 's d eg re e, it is evident
that la rg e nu m bers of people have earn ed and w ill continue to ea rn
the M a s t e r 's d eg re e.

This is p a rtic u la rly tru e of education, w here

on e-half of a ll M a s t e r 's d e g re e s a r e aw arded.
E a rly educational sy ste m s only provided education for the
v e ry se le c tiv e .

However, c u rre n t th eory holds th at ev ery m an is

entitled to the m axim um am ount of education he can profitably u se.
O ur sy stem of h ig h er education h as responded to th is p r e s s u r e by
estab lish in g Ju n io r C olleges, b ranch cam puses, and off-cam pus
cam pu ses.

T his has allowed m o re people to com plete un dergraduate

work and s t a r t gradu ate work.

In addition, th e re is a national trend

R ich ard P . B ra n d t

tow ard continuing into g rad uate work im m ed iately following the
u n d erg rad u ate work.

These two fa c to rs have brought about a la rg e

in c re a s e in demand for grad u ate education.
This in creasin g dem and, in tu rn , h as brought about p r e s ­
s u re fo r selectiv e ad m issio n s from the a d m in is tra to rs and faculty of
g rad u ate schools who find th e ir available re s o u r c e s depleted.
yet, th e re is no stan dard m ethod fo r selectio n .

As

V ery little is known

about the ac cu ra cy o r the long range effects of selection.

O rganization of the Study
The p rim a ry focus of the p re s e n t study, then, was an
investigation of the relatio n sh ip of se lec te d p rea d m issio n data to
p red icted su c c e ss in grad u ate school.

S uccess w as m ea su red a s

m em b ersh ip in one of five c a te g o rie s rela tin g to graduation o r no
graduation, d uration of the M a s t e r 's p ro g ra m , m e a s u re s of p e r ­
siste n c e in the M a s te r 's p ro g ram , and m e a s u re s of acad em ic
p erfo rm an ce in the M a s t e r 's p ro g ram .

In addition, the rela tio n sh ip s

of the p rea d m issio n data to d ep artm en t p ro file s was studied.
In o rd e r to in vestigate th ese co n c ern s, a sam ple of 358
M a s t e r 's d eg re e students belonging to one of the five r e s e a rc h
ca te g o rie s as of w inter, spring, su m m er, o r fall 1969 was selected .
A s e r ie s of four m ultiple d isc rim in a n t functions w as calculated to

R ic h a rd P. B ran d t

te s t the relatio n sh ip of the selected p rea d m issio n data to predicted
categ o ry m em b ersh ip .
The group of students who had su ccessfu lly com pleted the
M a s te r1 s d eg re e was then selected from the to tal sam ple.

A

m u ltiple r e g re s s io n an a ly sis w as then calculated to explore the
re la tio n sh ip between the p re a d m issio n data and the duration of the
Master* s p ro g ram .

In addition, two canonical c o rre la tio n s w ere

computed to re la te the p rea d m issio n data to a m e a su re of p e rs is te n c e
m ade up of the total te r m s attended, the to tal te rm s m isse d , the
duration, and the average co u rse load and a m e a su re of academ ic
p erfo rm an c e made up of the to tal g rad e point av erag e, the total
c re d its , and the total c re d its d e fe rre d .
F inally, factor an a ly se s with the oblique solution for a ll
fa c to rs with eigenvalues g r e a te r than 1. 00 w ere run fo r the to tal
g raduated group and for each m a jo r d ep a rtm e n t w ithin the to tal group.

M ajor F indings of the Study
The re s u lts of th ese an a ly se s supp o rt the following co n clu ­
sions b ased on te s ts of significance a t the . 05 level w here applicable:

1.

The se le c te d p read m issio n data is significantly re la te d to
m em b ersh ip in the graduated, g raduated to d o cto ra l pro g ram ,
not en ro lled fo r 5 y e a rs , academ ic action, o r do not rea d m it
categ o ry .

R ic h a rd P . B ra n d t

2.

Selected p rea d m issio n data is significantly d iffe ren t fo r the
g raduated and the academ ic action g ro u p s.

3.

Selected p rea d m issio n data is significantly d iffe ren t for the
g raduated and the not en ro lled for 5 y e a rs group.

4.

Selected p rea d m issio n data is significantly d iffe ren t for the
g raduated and the do not re a d m it groups.

5.

P re a d m issio n data for graduated groups is
to the d uration of the M a s t e r 's p ro g ram .

6.

P re a d m iss io n data for the graduated group is significantly rela te d
to an index of p e rs is te n c e m ade up of to tal te rm s attended, to tal
te rm s m isse d , duration, and av erag e c o u rse load.

7.

P re a d m issio n data fo r the graduated group is significantly rela te d
to an index of acad em ic p erfo rm an ce m ade up of total g ra d e point
a v erag e, to ta l c re d its , and total c re d its d e fe rre d .

8.

A fa c to r a n a ly sis of p rea d m issio n data for the graduated group
y ield s an in te rp re ta b le s e t of fa c to rs .

9.

The s e t of fa c to rs for each m a jo r d ep artm en t is a lm o st id entical
to the s e t of fa c to rs fo r the to tal group and fo r each o th er
dep artm ent.

significantly related

ACKNOW LEDGM ENTS

The w r ite r w ishes to e x p re ss h is ap p reciatio n and
acknowledge a debt of g ratitu d e:
To D r. Van C. Johnson for h is ability to guide without
undue r e s tric tio n and for h is encouragem ent and c ritic is m .
To D r. N orm an T. B ell, D r. W alter W. Scott, and
D r. Eldon N onnam aker for th e ir w illingness to se rv e on the
guidance com m ittee.
To D r. T. C. Cobb fo r providing the opportunity to work
in the College of Education, which in tu rn provided the w r ite r with
the n e c e s s a ry background for th is study.
To John F. D ra p e r fo r h is s ta tis tic a l a s s is ta n c e .
To my wife, P a tric ia , for h e r encouragem ent and supp ort
and a lso fo r coding the data.
To M arsh a L. Webb for typing the rough d ra ft of the text
and the tab les.

TABLE OF CONTENTS

Page
LIST OF T A B L E S ..................................................................................................

v

CHAPTER
I.

RATIONALE OF THE S T U D Y .............................

1

S tatem ent of the P r o b l e m ............................
1
6
P u rp o se of the S t u d y .......................................
T heory ..................................................................................
O rganization of the S t u d y ................................10
II.

8

REVIEW OF THE L IT E R A T U R E .......................... 15
Im p o rtan ce of the M a s t e r 's D egree
. . . .
15
P o litic a l o r Social T h eo ries
Governing Selective A dm issio ns . . . .
16
Changing P r e s s u r e s in A dm ission
S y s t e m s .............................................................. 19
The E ffects of Selective A dm issions . . . .
22
P re d ic tio n Studies R elated to
Selective A d m i s s i o n s ............................... 28
..........................................28
M e a su re s of P e rfo rm a n c e
P re d ic tiv e M e a s u r e s ..........................................31
S u m m a r y ................................................................... 34

III.

GENERAL DESIGN AND METHODOLOGY . . . .
The S a m p l e ..............................................................44
The N ature and Source of D a t a .....................47
A naly sis P r o c e d u r e s ..........................................49
S u m m a r y ...................................................................53

iii

44

CHAPTER
IV.

Page
FIN D IN G S........................................................................................
H ypotheses C oncerning Group
M e m b e r s h ip ..................................................................
H ypothesis 1
H ypothesis 2
H ypothesis 3
H ypothesis 4
Sum m ary of Group M em bership
H y p o t h e s e s ..................................................................
H ypotheses C oncerning M e asu res of
P e rfo rm a n c e fo r G r a d u a t e s ..............................
H ypothesis 5
H ypothesis 6
H ypothesis 7
Sum m ary of G raduate P erfo rm a n c e
H y p o t h e s e s ..................................................................
H ypotheses C oncerning D ep artm en tal
C h a r a c t e r i s t i c s ........................................................
H ypothesis 8
S u m m a r y .................................................................................

V.

SUMMARY AND C O N C L U S IO N S .....................................

55

55

66
67

79
79
90
92

Im plication s fo r P re d ictio n of Group
M e m b e r s h ip ..................................................................
96
Im plication s fo r M e a su re s of G raduate
P e r f o r m a n c e .............................................................
97
Im plications for G raduate P re a d m issio n
P r o f i l e s ..............................................................................100
R ecom m endations fo r F u tu re R e se a rc h . . . 102
B IB L IO G R A P H Y ............................................................................................................ 105

iv

LIST OF TABLES

TOTAL SAMPLE DIVIDED BY CATEGORY

. .

46

GRADUATES DIVIDED BY DEPARTMENT

. .

47

VARIABLE ABBREVIATIONS AND
...................................................................
DESCRIPTIONS

56

SIGNIFICANCE OF THE DISCRIMINANT
FUNCTIONS FOR ALL G R O U P S ...............................

57

DISCRIMINANT FUNCTION COEFFICIENTS
FOR PREADMISSION DATA FOR ALL
..................................................................................
GROUPS

58

SIGNIFICANCE OF THE DISCRIMINANT
FUNCTION FOR GRADUATE AND 5 YEAR
NOT ENROLLED G R O U P S ..............................................

60

DISCRIMINANT FUNCTION COEFFICIENTS
FOR PREADMISSION DATA FOR
GRADUATE AND 5 YEAR NOT ENROLLED
GROUPS
..................................................................................

61

SIGNIFICANCE OF THE DISCRIMINANT
FUNCTION FOR GRADUATE AND ACADEMIC
ACTION G R O U P S ...................................................................

62

DISCRIMINANT FUNCTION COEFFICIENTS
FOR PREADMISSION DATA FOR GRADUATE
AND ACADEMIC ACTION GROUPS
.....................

63

v

SIGNIFICANCE OF THE DISCRIMINANT
FUNCTION FOR GRADUATE AND
DISMISSED G R O U P S ........................................................

64

DISCRIMINANT FUNCTION COEFFICIENTS
FOR PREADMISSION DATA FOR
GRADUAtA a ND DISMISSED GROUPS
. . . .

65

REGRESSION COEFFICIENTS, BETA WEIGHTS,
STANDARD ERRORS, AND STEPWISE R FOR
PREADMISSION DATA FOR GRADUATES WITH
DURATION AS DEPENDENT VARIABLE
. . .

68

CORRELATION MATRIX OF PREADMISSION
DATA FOR G R A D U A T E S ..............................................

70

CORRELATION MATRIX OF MEASURES OF
PERSISTENCE GOR G R A D U A T E S ..........................

71

CORRELATION MATRIX OF PREADMISSION
DATA AND MEASURES OF PERSISTENCE
FOR G R A D U A T E S .............................................................

72

X2 TEST OF SIGNIFICANCE OF CANONICAL
CORRELATIONS OF MEASURES OF
PERSISTENCE FOR G R A D U A T E S ..........................

73

CANONICAL COEFFICIENTS FOR MEASURES
OF PERSISTENCE AND PREADMISSION
D A T A ............................................................................................

74

CORRELATION MATRIX OF MEASURES OF
ACADEMIC PERFORMANCE FOR
GRADUATES .............................................................................

75

CORRELATION MATRIX OF PREADMISSION
DATA AND MEASURES OF ACADEMIC
PERFORMANCE FOR GRADUATES
....................

76

vi

TABLE

Page

4. 18

X2 TEST OF SIGNIFICANCE OF CANONICAL
CORRELATIONS OF MEASURES OF
ACADEMIC PERFORMANCE FOR
G R A D U A TES.............................................................................................77

4 .1 9

CANONICAL COEFFICIENTS FOR
MEASURES OF ACADEMIC PERFORMANCE
AND PREADMISSION D A T A .........................................................78

4 .2 0

HIGHEST FACTOR LOADINGS FOR
PREADMISSION DATA FOR GRADUATES
FROM ALL DEPARTMENTS USING THE
PROMAX ROTATIONAL A N A L Y SIS ..........................................81

4.21

HIGHEST FACTOR LOADINGS FOR
PREADMISSION DATA FOR GRADUATES
ADMITTED NONDEGREE USING THE
.....................................83
PROMAX ROTATIONAL ANALYSIS

4. 22

HIGHEST FACTOR LOADINGS FOR
PREADMISSION DATA FOR GRADUATES
ADMITTED IN ELEMENTARY EDUCATION
USING THE PROMAX ROTATIONAL
A N A L Y S IS ..................................................................................................85

4.23

HIGHEST FACTOR LOADINGS FOR
PREADMISSION DATA FOR GRADUATES
ADMITTED IN SECONDARY EDUCATION
USING THE PROMAX ROTATIONAL
A N A L Y S IS ..................................................................................................86

4. 24

HIGHEST FACTOR LOADINGS FOR
PREADMISSION DATA FOR GRADUATES
ADMITTED IN ADMINISTRATION USING
THE PROMAX ROTATIONAL A N A L Y S I S .......................... 88

4.25

HIGHEST FACTOR LOADINGS FOR
PREADMISSION DATA FOR GRADUATES
ADMITTED IN COUNSELING AND
GUIDANCE USING THE PROMAX
ROTATIONAL A N A L Y S IS ..............................................................89

v ii

CHAPTER I

RATIONALE OF THE STUDY

Statem ent of the P ro b lem
The problem of controlling grad u ate a d m issio n s and
en ro llm e n ts is causing in c re a sin g concern among the a d m in is tra to rs
of grad u ate schools.

The g rad u ate a d m issio n s o ffic er a t the U n i­

v e rs ity of Michigan re p o rte d th at grad u ate application s in c re a se d
20 p erc en t in the academ ic y e a r 1966 -157 o v er the academ ic y e a r
1965-66. * The Dean of the College of Education a t M ichigan State
U niversity recen tly estab lish ed a faculty com m ittee to work on the
problem of ad m issio n s and en ro llm e n ts.
problem a t M ichigan State.

E n ro llm en t h as becom e a

The G raduate Student A ffa irs Office

re p o rts an annual in c re a s e in en ro llm e n t of approxim ately 12 p e rc e n t
in its annual re p o rt fo r the acad em ic y e a r 1968-69.
Spaeth notes th at the in c re a s e in p r e s s u r e i s p a rtly due to
the fact th a t a deg ree from a good g rad u ate school is the b e s t way
to in su re a place in a p ro fessio n .

2

M ore im p o rta n t fo r C olleges of

Education is the tren d of in c re a se d education re q u ire m e n ts c u rre n tly

1

2

being dem anded fo r ce rtifica tio n by the sta te le g is la tu re s and te a c h e r
3

o rg an izatio n s.

The p r e s s u re fo r ad m issio n to g rad u ate study is also

being inten sified by the d ec rea sin g tim e lag betw een the end of the
u nderg rad u ate p ro g ram and the s t a r t of the g rad u a te p ro g ram .

4

D avis rep o rted th at 32. 6 p e rc e n t of the college s e n io rs he sam pled
in 1961 planned to go to g rad u ate school the following acad em ic y e a r.
5
An additional 44 p e rc e n t planned to attend a t a l a te r date.
Grigg
re p o rte d that a study of college s e n io rs in 1956 showed th at 83 p e rc e n t
intended to go on.

A two y e a r follow -up on this sam e group showed
g
th at 48 p e rc e n t had actually en ro lled a s of June 1958.
The in c re a se d dem ands for g rad u ate education a r e being felt
esp ecially in C olleges of Education.

In 1940, o n e -th ird of a ll

M a s t e r 's d e g re e s w ere aw arded in education.

By 1960, on e-h alf

of a ll M aster* s d e g re e s w ere aw arded in education.

7

Grigg re p o rts

th at twenty thousand M a s t e r 's d e g re e s w ere aw arded in education in
g
1950. He e s tim a te s that fifty thousand w ill be aw arded in 1975.
The production of M a s t e r 's d e g re e s by the colleges of education in
M ichigan is even m o re strik in g .

The C itizen s C om m ittee on H igher

Education in M ichigan rep o rted th at 2, 553 M a s t e r 's d e g re e s w ere
aw arded in education in 1963.

The second hig h est num ber of d e g re e s

w ere aw arded in E nglish, which had 616.

g

As g rad u ate schools re a c h the lim it of th e ir re s o u rc e s ,
faculty and a d m in is tra to rs concerned with a d m issio n s policy seek

3

ways of lim iting e n ro llm e n ts.

B erelso n notes that g rad u ate schools

have trad itio n a lly depended on a high ra te of a ttritio n to keep e n r o ll­
m ents in line with re s o u rc e s .
a ttritio n of 40 p erc en t.

T his r e s u lts in a n o rm a l ra te of

Some g rad u ate schools, how ever, a r e turning

to ca refu l sc ree n in g of applicants to lim it en ro llm en t.

T his has been

done for som e tim e in the p ro fessio n al schools, according to B e r e l ­
son, and has re su lte d in a n o rm al a ttritio n ra te of about 10 p e rc e n t. ^
R ecently, g rad u ate schools have a ls o used the techniques of lim ited
ad m issio n s and estab lish m en t of off-cam pus c e n te rs . ** The College
of Education a t M ichigan State h as had lim ited en ro llm e n t in specific
M aster* s p ro g ra m s and a nondegree p ro g ram which functioned often
as an open door to g rad u ate work.

The open door policy in the non-

d eg ree p ro g ram has re c en tly been elim inated by faculty action.

In

addition, ad m issio n re q u ire m e n ts in c e rta in d ep a rtm e n ts a r e going
up.

T his follows the p attern suggested by K urland.

K urland points

out that the n o rm al p a tte rn is to s e t c e rta in lo g ical ad m issio n r e q u i r e ­
m en ts such a s m inim um grad e point o r m inim um t e s t s c o re s .

As

p r e s s u r e fo r ad m issio n in c r e a s e s , the n o rm a l tendency is to r a is e
the m inim um re q u ire m e n ts.

E ventually, the re q u ire m e n ts a r e s e t

so high th at they no longer b e a r any rela tio n sh ip to su c c e ss in a
grad uate p ro g ra m . ^
A s ad m issio n s re q u ire m e n ts go up, the p r e s s u r e becom es
g r e a te r to s e le c t the m o st acad em ically com petitive ap p lican ts.

The

4

selectio n problem in graduate a d m issio n s, then, re la te s quite closely
to the problem of p erso n n el selectio n .

In p erso n n el selectio n the

im p o rtan t question i s w hether the p erso n being co n sid ered w ill be
able to p erfo rm the job o r not.

13

In g rad uate a d m issio n s the

im p o rtan t question is w hether o r not the p erso n w ill g rad u ate.
problem of effective selectio n h as been puzzling r e s e a r c h e r s .

The
Irvine

and F e d le r, in dealing with the problem of d ifferen tiatin g between
g ra d u a te s and nongraduates, found th at the p red ic tio n s based on an
a n a ly sis of av ailab le data w ere sub ject to la rg e e r r o r s with re s p e c t
to grad uatio n o r no graduation.
problem .

14

O ther a u th o rs have a ls o noted th is

Mayhew sug g ests th at r e s e a r c h e r s should be content with

sortin g ap p lican ts into upper, m iddle, and low er c a te g o rie s.

15

Bogue su g g ests a m odel for estab lish in g cutting s c o re s on ad m issio n
re q u ire m e n ts which elim in ates only the bottom stu den ts for whom
failu re is p ra c tic a lly guaranteed.

16

In addition to the problem of selectio n of su c c e ssfu l s t u ­
dents, the problem of efficient u se of student and u n iv e rsity
re s o u rc e s i s involved in a good a d m issio n s p ro c e s s .

B ru b ach er,

H enderson, and M eder advance the concept th at ad m issio n s r e q u i r e ­
m ents m u st g u aran tee efficien t u se of the student* s talen t, the b e s t
use of the school* s unique q u a litie s, and the b e s t fulfillm ent of
s o c ie ty 's needs.

17

It is p a rtic u la rly im p o rtan t that ad m issio n s

5

p ro g ra m s m e e t th ese re q u ire m e n ts in light of the d isp ro p o rtio n a te
am ount of higher education re s o u r c e s needed to o p era te a g rad u ate
p ro g ram .

One study a t the U n iv ersity of M ichigan noted th at the co st

ra tio for fresh m an -so p h o m o re, ju n io r-s e n io r, and g rad u ate p ro g ra m s
was 1:3:8.

18

M cConnell em p h asizes th at r e s e a r c h e r s should d ire c t

th e ir attention to the problem of efficient u se of U niv ersity re s o u rc e s
and not to b e tte r selectio n p ro c e d u re s.

He contends th at "planning

fo r the future should be based on w hat se em s to be an inescapable
assum ption, nam ely, th at in the long run A m erican H igher Education
as a whole w ill not becom e m o re se lec tiv e. "

19

S ev eral questions in

addition to the question of g radu ation need to be answ ered in o rd e r
to g u ara n tee the effective u se of student and U n iv ersity re s o u rc e s .
It would be n e c e s sa ry to know the a v e rag e duration of a M aster* s
p ro g ram in v ario u s d e p a rtm e n ts, the av e rag e co u rse load of the
student, and the a v e rag e num ber of te rm s m isse d by the student in
o rd e r to g u aran tee m axim um u se of the school* s re s o u rc e s .

It

would be n e c e s s a ry to know the d e p a rtm e n ta l c h a r a c te r is tic s of each
m a jo r in o rd e r to in su re the b e s t u se of the student* s talen t.

F inally,

the p a rtic u la r co u rse offerings and d eg re e p ro g ra m s of the U n iv e r­
sity m u st be aligned with the needs of the so ciety serv ed by the
U niv ersity.

6

P u rp o se of the Study
The purpose of th is study is to inv estigate the rela tio n sh ip
of selected p rea d m issio n data with s e v e ra l m e a s u re s of g rad u ate
su c c e ss.

The f i r s t question of in te r e s t is the pred icted m em b ersh ip

in one of five groups.

The f i r s t of th ese is the group th at g raduated

within the p re s c rib e d tim e lim its but have not continued to the P h.D .
p ro gram a t M ichigan State.

The second group is com prised of s t u ­

dents who g raduated within the p re s c rib e d tim e lim its and continued
in the P h .D . p ro g ram .

The th ird group is m ade up of students who

have had academ ic action taken ag a in st them as a r e s u lt of poor p e r ­
form ance in the M a s t e r 's p ro g ram .

The fourth group is m ade up of

students who have been asked to w ithdraw from the p ro g ram for
v ario u s re a so n s.

The la s t group is made up of stud ents who w ere

accepted and enrolled for v a rio u s num bers of te rm s , but have not
re -e n ro lle d for a t le a s t five y e a r s and a re , th e re fo re , no longer
eligible to use th e ir earn ed c re d its fo r a M aster* s d eg ree.

The

purpose of exploring the question of group m em b ersh ip is to exam ine
the b a sic question of graduation o r no graduation.
A second purpose of th is study is to re la te the two graduated
groups to indices of duration, p e rs is te n c e , and academ ic p erfo rm an ce.
These re la tio n sh ip s w ill provide m e a s u re s of re s o u rc e use.
s e v e ra l p re d ic to r v a ria b le s w ill be involved, the m u ltiv a ria te

Since

7

techniques of m ultiple re g re s s io n , d isc rim in a te a n a ly sis, and
canonical c o rre la tio n will be used.
An additional purpose of this study is to exam ine the d e p a r t­
m ent p ro file s of the g raduate group.

T his inform ation is needed in

o rd e r to s e le c t the d ep a rtm e n t in which a p a rtic u la r student would
be m o st likely to succeed.
The following hypotheses w ere form ulated to exam ine the
above questions:

1.

Selected p rea d m issio n data is significantly re la te d to
m em b ersh ip in the gradu ated, g raduated to P h .D . p ro g ram ,
not en ro lled fo r five y e a rs , academ ic action, o r do not
re a d m it category.

2.

Selected p rea d m issio n data is significantly d iffe re n t fo r the
graduated and the not enrolled for five y e a rs groups.

3.

Selected p rea d m issio n data is significantly d iffe ren t for the
graduated and the academ ic action groups.

4.

Selected p rea d m issio n data is significantly d ifferen t for the
grad uated and the do not re a d m it g ro u p s.

5.

P re a d m iss io n data fo r the g raduated groups is significantly
re la te d to the duration of the Master* s p ro g ram .

6.

P re a d m issio n data fo r the graduated group is significantly
re la te d to an index of p e rs is te n c e for the Master* s p ro g ram .

7.

P re a d m iss io n data for the graduated group is significantly
re la te d to to ta l grade point, to ta l c re d its , and to ta l c re d its
d e fe rre d considered to gether.

8.

A fa c to r a n a ly sis of p rea d m issio n data fo r die graduated
group w ill yield an in te rp re ta b le s e t of fa c to rs which w ill
be d iffe ren t fo r each dep artm en t.

8

Theory
T h e o rie s re la te d to the g rad u ate ad m issio n s p ro c e ss a re
g en erally concerned with explaining p a tte rn s of a d m in istra tiv e behavior
o r the p red ictio n of acad em ic su c c e ss.

Bogue sug g ests a th e o re tic a l

m odel which com bines the a d m in istra tiv e p a tte rn s with the p red ictiv e
sy stem to yield a decision c r ite r ia fo r ad m issio n s.

20

B ottenburg

and C h rista l th eo rized th at a d m in istra tiv e p a tte rn s among a group
of r a t e r s could be grouped o r c lu ste re d into a sm a ll num ber of p a tte rn s which would explain r a t e r policy.

21

W h erry s used a facto r

an a ly sis technique to identify the p rin cip al fa c to rs of a rater* s
decision; r a t e r s with s im ila r p ro file s w ere then grouped in a s im ila r
m ann er to B ottenburg and Christal* s w ork.

22

Roscoe theo rized that

som e unknown, and th e re fo re unidentifiable, v a ria b le s w ere included
in a rater* s a s s e s s m e n t of a student.

He p re se n te d each r a t e r with

a lis t of graduated studen ts and asked each r a t e r to s e le c t the top
ten g rad u a tes from the lis t.

He fe lt th at th is technique would help

to include any n onm easurable values in the sy stem .

23

The stu dies

which attem p t to explain a d m in istra tiv e behavior do not g en e rally
try to p re d ic t acad em ic s u c c e ss .

The in ten t is to identify the values

underlying adm issio n d ecisio n s.
T h eo ries re la tin g p rea d m issio n data and o th er fa c to rs to
m e a su re s of academ ic s u c c e ss a r e num erou s.

The g rad e point is

9

the m o st u n iv ersa lly accepted c rite rio n of su c c e ss in the academ ic
setting.

M ost r e s e a r c h e r s use the grad e point a t som e point in the

p ro g ram , th e re fo re , as a m e a su re of s u c c e ss .

M ehrabian c o r r e ­

lated ad v iso r ra tin g s with g rad e point a v e rag e on the assum ption
that faculty ratin g s of g rad u ate studen ts w ill be co n sisten t w hether
they a r e in the form of g ra d e s o r ra tin g s.

24

Owens re la te d p e e r

ratin g s to g rad e point a v e rag e to te s t the assum p tio n th at students
a r e able to s o r t th em selv es into ab ility grou ps.

25

C areton, Payne,

Newman, Borg, H e rb e rt, and S trie k e r a re among s e v e ra l au th o rs
who th eo rize th at g ra d e point a v e rag e can b e s t be p red icted by using
som e com bination of fa c to rs .

26

The p a rtic u la r com bination of

fa c to rs seem s to v a ry with each author.

M e a su re s rela tin g to b io ­

grap h ical data, acad em ic p erfo rm an c e, and te s t data a r e usually
included.
F e d le r notes, how ever, that the ab ility to p re d ic t g rad e
point a v e rag e m ay be as n e c e s s a ry a s putting a r a z o r edge on a hoe.
Bogue, Irv in e, and Mayhew concur with F e d le r in theorizing that the
m o st re le v a n t question in p rediction stu d ies m ight be the p red ictio n
of graduation o r no graduation.

28

Irvine notes, how ever, that d e a l­

ing with the dichotom ous question of graduation o r no graduation is
sub ject to la rg e s ta tis tic a l e r r o r s .

29

Mayhew th e o riz e s, th e re fo re ,

that i t se em s to be m o st p ro fitab le to co n sid er so rtin g app licants

10

into s e v e ra l broad c a te g o rie s.

30

Bogue su g g ests fu rth e r that

s ta tis tic a l m ethods should be em ployed to define a th e o re tic a l
ad m issio n s m odel which not only allow s fo r selectin g applicants
into s e v e ra l c a te g o rie s, but a lso m in im izes the pred ictio n o r sorting
erro rs.

31

O rganization of the Study
Selected p read m issio n data for five grad u ate student popu­
latio ns w ill be run in a d isc rim in a te a n a ly sis to d e te rm in e the r e l a ­
tionship between group m em b ersh ip in a p a rtic u la r population and
the p re a d m issio n data.

P re a d m issio n data fo r the sam ple of students

who receiv ed the M a s t e r 's d eg re e w ill also be re la te d to in dices of
p e rs is te n c e , duration , and acad em ic p erfo rm an ce to d eterm in e the
relatio n sh ip of the p re a d m issio n data to m e a s u re s of re s o u rc e use.
In addition, the g raduated group w ill be exam ined by d ep artm en t to
d eterm in e if significant d ep a rtm e n ta l c h a ra c te r is tic s a r e p re s e n t in
the p rea d m issio n data.
The relev an ce of the problem of selectio n in g rad u ate a d m is ­
sions and a ratio n ale for a study of selectio n p ro ce d u res has been
p resen ted in C hapter I.

The lite ra tu r e re la te d to selection and p r e ­

diction in g rad u ate a d m issio n s w ill be review ed in C hapter II.

The

specifications of the sam ple, n atu re of the p rea d m issio n data and
the c r ite r ia , and a d e sc rip tio n of the m ethodology for analyzing the

11

data w ill be p resen ted in C hapter III.

The r e s u lts of the an a ly sis

w ill be p resen ted in C hapter IV, and the conclusions and a sum m ary
w ill be p resen ted in C hap ter V.

FOOTNOTES

Byron L. G roesbeck, "G raduate A dm issions - -Are We
P re p a re d for the Y e a rs A head?" College and U niv ersity, XLII,
No. 4 (Sum m er 1967), 504-510.
2

Joe L. Spaeth, "A llocation of College G rad u ates to G ra d u ­
ate and P ro fe ssio n a l Schools, " Sociology of Education, XLI, No. 4
(F all 1968), 342-349.
3

H igher Education in a Decade of D ecision (W ashington,
D .C . : N ational Education A ssociation , E ducational P o lic ie s C om ­
m issio n , 1959), p. 3.
4

G roesbeck, op. c i t . , pp. 504-510.

5

Ja m e s A. D avis, G re a t A sp ira tio n s (Chicago: Aldine
Publishing Company, 1964), p. 43.
0
C h a rle s M, G rigg, G raduate Education (New York: The
C enter for Applied R e se a rc h in Education, Inc. , 1965), p. 88.
7

B ern ard B erelso n , G raduate Education in the United S tates
(New York: M cG raw -H ill Book Company, 1960), p! 182.
8

Gr i gg , op. c it. , p. 57.

9

R ep o rt of the C itizens C om m ittee on H igher Education in
M ichigan, The C itizens C om m ittee on H igher Education, H arold T.
Smith, Executive D ire c to r, 709 South W estnedge Avenue, K alam azoo,
M ichigan, 1965, p. 3.
10

B erelso n , op. c i t . , p. 168.

11T. R. McConnell, A G en eral P a tte rn fo r A m erican P ublic
H igher Education (New York! M cG raw -H ill Book Company, 1962),
p. 9.

12

13
12

N orm an D. K urland, "M ore P eople Than P la c e s , o r How
to E n su re Equality of O pportunity Among E quals When T h e re A re
M ore E quals Than O p p o rtu n itie s ,11 College and U n iv e rs ity , XLIII,
No. 2 (W inter 1968), 221-224.
13

J . P. Guilford and W illiam B. M icheal, The P re d ic tio n
of C ateg o ries from M e asu rem en ts (B everly H ills, C alifo rn ia: Sheridan
Supply Company, 1949), p. 2.
14

Donald W. Irv in e , "M ultiple P re d ic tio n of C ollege G ra d u a ­
tion from P re a d m issio n D ata, " Jo u rn a l of E x p erim e n ta l E ducation,
XXXV, No. 1 (F a ll 1966), 84-89; and D aniel P. F e d le r, ''In trig u in g
P ro b le m s of D esign in P re d ic tin g College S uccess, " E ducational and
Psychological M easu rem en t, XXV, No. 1 (Spring 1965), 29-37.
15

Lew is B. Mayhew, "N ontest P re d ic to r s o f A cadem ic
A chievem ent, " E ducational and P sychological M e a s u re m e n t, XXIX,
No. 1 (Spring 1969), 39-46.
16

E.
G. Bogue, "A pplication of a M inimum L o ss D ecisio
S trategy in the Selection of Cutoff P o in ts in College and U n iv ersity
A dm issions, " College and U n iv ersity , XLIII, No. 2 (W inter 1968),
131-141.
17

John S. B ru b ach er, B a ses fo r P olicy in H igher E ducation
(New York: H a rp e r and B ro th e rs, P u b lis h e rs , 1960), p. 49; and
A lb ert E. M eder, "A dm issions and the C o lle g e ," C ollege A d m issio n s
(P rinceton: College E n tran ce E xam ination B oard, 19631" p. 41.
18

G rigg, op. c i t . , p. 35.

19
20

21

M cConnell, op. c it. , p. 9.
Bogue, op. cit. , pp. 131-141.

Raymond E. C h rista l, "JAN: A T echnique fo r A nalyzing
Group Judgm ent, " The Jo u rn a l of E x p erim e n ta l E d u catio n , XXXVI,
No. 4 (Sum m er 1968), 24-27; and R o b ert A. B ottenburg and Raymond A.
C h rista l, "Grouping C r ite r ia - - A Method Which R etain s M aximum
P re d ic tiv e E fficiency, " Jo u rn a l of E x p erim e n ta l E ducation, XXXVI,
No. 4 (Sum m er 1968), 28-34.

14
22

R o b ert J . W herrys and Ja m e s C. N aylor, "C om parison
of Two A pproaches - -JAN and P R O F --F o r C apturing R a te r S tra te g ie s, "
Educational and P sychological M easurem ent, XXIX, No. 2 (Sum m er
1966), 267-286.
23

John T. Roscoe and Sam uel R. Houston, "The P re d ic tiv e
V alidity of GRE Scores fo r a D octoral P ro g ra m in Education, " Educa tional and P sychological M easurem ent, XXIX, No. 2 (Sum m er 1969),
507 -509.
24

A lb e rt M ehrabian, "U ndergraduate A bility F a c to rs in
R elationship to G raduate P e rfo rm a n c e , " E ducational and P s y c h o ­
logical M e asu rem en t, XXIX, No. 2 (Sum m er 1969), 409-419.
25

Thom as R. Owens, "P re d ic tin g A cadem ic S uccess in
Master* s D egree P ro g ra m s in Education, " Jo u rn a l of Educational
R e se a rc h , LX, No. 3 (November 1966), 124-126.
26

Edw ard E. C areton and L ouise W. C areton, "P red ictio n
of Success in G raduate Study of Psychology a t U niversity of T ennessee, "
A m erican P sy ch o lo g ist. IV, No. 8 (August 1949), 361-362; David A.
Payne and Cynthia E, T uttle, "T he P re d ic tiv e R elationship of the
M iller A nalogies T e st to O bjective and Subjective C r ite r ia of S uccess
in a G raduate School of Education, " E ducational and P sychological
M easu rem en t, XXVI, No. 2 (Sum m er 1966), 427 -430; R ichard I.
Newman, "GRE S cores as P re d ic to rs of GPA for Psychology G ra d u ­
ate Students, " Educational and P sychological M e asu rem en t, XXVII,
No. 4 (W inter 1968), 433-436; W alter E B org, "GRE Aptitude
S cores a s P re d ic to rs of GPA for G raduate Students in Education, "
Educational and P sychological M e asu rem en t, XXIII, No. 2 (Sum m er
1963), 379-382; David J H e rb e rt, r,A P re d ic tiv e Study of Q uality
P oint A verages in G raduate Education C o u rse s, " Jo u rn a l of E d u ca­
tional R e sea rch , LX, No. 5 (Jan u ary 1967), 218-220; G eorge S trie k e r
and J . T. H uber, "The GRE and U ndergraduate G rad es a s P re d ic to r s
of Success in G raduate School, " The Jo u rn a l of E ducational R e sea rch ,
LX, No. 10 (July-A ugust 1967), 466-468.
27
28

F e d le r, op. cit. . pp. 29-37.
Bogue, op. c it. , pp. 131-141.

29
Irv in e , op. c i t . , pp. 84-89.
30
31

Mayhew, op. c it. . pp. 39-46.
Bogue, op. cit. , pp. 131-141.

C H A P T E R II

REVIEW OF THE LITERATURE

Im portance of the
Master* s D egree
T h ere is co n sid erab le d ifferen ce of opinion in the lite r a tu r e
a s to the place of the M a s t e r 's d eg re e; B e re lso n notes th a t sc h o la rs
debated w hether it was a te rm in a l d eg re e o r a steppingstone to a
d o cto rate a s e a rly a s 1902. * In 1910, Calvin Thom as of Colombia
d escrib ed it as "a slig htly c u ltu ra l d e g re e , p a rtly a re s e a rc h d eg ree,
but everyw here a teaching d e g re e , m ainly for the second ary schools. "
B erelson quotes J . P. E ld e r, then g rad u a te dean of H arv ard , a s
saying, "The M aster* s d eg re e is , a t p re s e n t, a b it like a s t r e e t 3

w a lk e r--a ll things to a ll m en (and a t d iffe re n t p ric e s ). "

W hatever

the place of the d eg re e, it is evident th a t i t is an ex tre m e ly popular
d eg ree.

In the academ ic y e a r 1962-63 a lm o st 88, 000 people earn ed

the M a s t e r 's d eg re e.

In 1953-54 it w as the h ig h est d eg re e held by

half of a ll college and u n iv e rsity faculty m e m b e rs .

It w as the highest

d egree held by 60 p e rc e n t of a ll new faculty m e m b e rs in 1962 -63.

15

4

16

B erelso n re p o rts that a s of 1960, alm o st 600 in stitutions
w ere aw arding the Master* s. Over a q u a rte r of these w ere lib e ral
5
a r t s o r te a c h e rs colleges.
Snell r e p o rts that 30 p erc en t of the
lib e ra l a r t s colleges, 63 p erc en t of the te a c h e rs colleges, and a ll
g
of the m a jo r u n iv e rsitie s aw ard the M a s te r 's d egree.
The highest
proportion of M a s te r 's d eg re es a re aw arded in education.

In 1940,

o n e-th ird of a ll M a s t e r 's d eg re es w ere aw arded in education.
n early one-half a re aw arded in education.

Now,

In 1950, this amounted

to 20 thousand d e g re e s in education; 50 thousand d e g re e s a r e p ro jected for education by 1975.

7

A larg e p a r t of this output is accounted

for by a few la rg e u n iv e rsitie s.

In 1961-62, for instance, the top

ten d egree-producing in stitu tion s accounted for approxim ately 15.5
g
p ercen t of the to tal output of Master* s d e g re e s.
It is in te restin g to
note that two of the top ten degree-producing in stitu tio ns in this study
w ere the U niversity of M ichigan and Michigan State U niversity.

P o litic al o r Social T h eo ries
Governing Selective A dm issions
The availability of education has always been related to
c u rre n t p olitical and social p rac tic e.

H igher education fo r the few

had its origins in the G reco-R om an cu ltu re.
for the upper c la ss.

L iberal education was

A risto tle taught th at the mind ru le s the body

and it was a sim ple extension of th is that led to the theory th at the

17

upper c la ss ru le s the low er c la s s .

g

In P la to 1s R epublic, the u ltim ate

control of the sta te w as given to a sm a ll body "whose in te lle c tu a l
p o ten tialities have been winnowed from the chaff of the populace by
a s e v e re and exacting d ia le c tic .

The educational th e o rie s of the

re n a issa n c e called fo r education of the e lite , a lso .

In the e a rly

h isto ry of United S tates education, T hom as J e ffe rs o n called fo r p r o ­
g re s s iv e sc re e n in g on the b a s is of in te lle c tu a l ab ility . **
C u rre n t thought se e m s to be supporting the id ea of providing
m axim um education fo r a ll who w ant it.

T. R. M cConnell s ta te s :

"In a d em ocracy i t is ju s t a s g r e a t a c rim e to p rev e n t a m an, stro n g
in mind o r c h a ra c te r, o r body, from accom plishing w hat n a tu re gave
him power to do, a s to p rev e n t a weak m an from e x e rtin g h is pow er
to the fu lle st extent in com petition o r cooperation w ith h is fellow s. "

12

The E ducational P o lic ie s C om m ission of the NEA co n c u rred in th e ir
statem en t:

"While governing b o a rd s of fa c u ltie s of educational

in stitu tio n s can do much to reg u la te college atten d an ce, e ith e r by
r e s tric tio n o r expansion, the A m eric an people th e m se lv e s u ltim ately
decide who w ill go to college. "

13

H enderson s ta te s th at the problem

of who should be educated should be d ete rm in ed by the needs and
d e s ire s of the individual, the r e s o u rc e s , re s p o n s ib ilitie s , and
p u rp oses of each in stitu tio n , and the ideology, m anpow er, and
citizenship needs of the nation.

14

B ru b a c h e r quoted H orace Mann

18

as saying, "E ducation is the g r e a t e q u a liz e r of the conditions of m en
--th e balance wheel of the so c ia l m ach in ery . "

15

While c u r re n t thought se em s to su p p o rt m axim um opportunity
for everyone, th e re is a ls o the r e s tr a in t of lim ited re s o u r c e s .

Many

a u th o rs recogn ize th is and su g g e st se le c tiv e a d m issio n s p ro g ra m s
which seek to a t tr a c t the b e s t qualified stu den ts to g rad u a te p ro g ra m s.
The NEA E ducational P o lic ie s C om m ission called for heavy em ph asis
in a ttra c tin g the gifted student, and se le c tiv e re c ru itm e n t as opposed
to ad m issio n .

16

B ru b a c h e r n otes th a t the talented student is often

the forgotten m an.

He say s ju s tic e dem ands th at a student with

s u p e rio r ab ility should have s u p e rio r opportunity.

17

Snell in his

recom m endations fo r im proving the M a s t e r 's d eg re e ad vocates
m inim um ad m issio n sta n d a rd s and p ro g ra m s no long er than a calen d ar
y e a r in length so th at m o re people can go through them .

18

H a rte and Thompson, in th e ir su rv ey of legal c a se s
involving ad m issio n re q u ire m e n ts , r e p o r t that the c o u rts have c o n ­
siste n tly upheld the rig h t of the u n iv e rsity to s e le c t students.

The

case of B u rel v s. D avision in 1894 is a landm ark c a se which e s t a b ­
lished education a s a p riv ile g e and not a righ t.

The case of L en er

vs. the State B oard of E ducation of New York upheld the rig h t of
Brooklyn College to in s is t on a m inim um en tra n ce sc o re .

19

19

Changing P r e s s u r e s in
A dm ission System s
The e a r lie s t a d m issio n s sy stem belonged to the G reek s and
was op erationally sim p le.
fam ilies of nobility.
blood and sex.

The G re ek s only educated m ales from

So, the only ad m issio n re q u ire m e n ts w ere

20

E a rly re q u ire m e n ts in A m erican u n iv e rs itie s w ere also
quite sim p le.

B room e re p o rts th at the e a r lie s t re q u ire m e n ts on

re c o rd w ere published by H arv ard in 1692.

The H a rv ard r e q u i r e ­

m ents read a s follows:
When any sc h o la r is able to re a d Tully o r such like
c la s s ic a l Latin au th o r ex tem p o re, and m ak e and speak tru e
L atin in v e rs e and p ro se , neo (ut aiunt) m o rte , and decline
p erfectly the p arad ig n s of nounes and v e rb e s in ye G reeke
tongue, then m ay hee bee adm itted into ye college, n or sh a ll
any claim e a d m issio n before such qualifications.'**
The exam ination o v er the above re q u ire m e n ts w as usually
given by the p re sid e n t of the school.

H illg a rth re p o rts that by 1872,

the ad m issio n d ecisio n had shifted to the faculty.

The p ro ced u re in

1872 was to apply to the acad em ic council, statin g p re s e n t q u a lific a ­
tions and the y e a r of expected g radu ation .

M ost g rad u ate students

w ere from the u n d erg rad u ate p ro g ram a t H a rv ard .

Thus, the faculty

had se v e ra l y e a r s in which to decide who should be given the o p p o rtunity to go on.

22

20

The ad m issio n s problem in grad u ate school h as evolved from
a sim ple beginning into a s e rio u s problem fo r ad m in istra tio n .

Snell

se e s the in cre asin g p re s s u r e a s being caused by five h is to ric a l
fac to rs.

F ir s t, by the period of 1870-1890 th e re w ere m any teaching

positions in colleges for those with a M a s t e r 's d eg re e.

Second, the

rapid expansion of high schools and colleges p re p a re d m any m o re
students for grad uate work and eventually cre ate d a demand fo r m ore
te a c h e rs with the M aster* s d eg re e in the high schools and colleges.
T hird, the electiv e sy stem , which was widely adopted in the colleges
in the 1890* s, p re p a re d students for the sp e cia lize d work in g rad u ate
school.

Fourth, A m erican women, dem anding equal rig h ts, w ere

f ir s t adm itted to g rad u ate school ju s t p rio r to the civil w ar.

In the

period 1870-1958 women earned 34 p e rc e n t of a ll M aster* s d e g re e s.
Fifth, the developm ent of su m m er schools enabled public school
te a c h e rs to continue th e ir stu d ies.

Sum m er school en ro llm en t

nationwide jum ped from 132,500 in 1919 to 943,000 in 1949.

23

G roesbeck feels th at the m o re re c e n t in c re a s e in a d m issio n s
p re s s u re is due to the in c re a sin g num ber of people receiv in g the
B. A. d eg ree, in cre asin g num bers of th ese people who want g rad u ate
work, and the d e c re a sin g tim e lage between the end of the B. A.
program and the s t a r t of the M aster* s d eg re e p ro g ram .

24

Davis docum ented Groesbeck* s contention th at g r e a te r
num bers of people receiving a B. A, d eg re e want g rad u ate education.

21

He found in h is survey of graduating s e n io rs in 1961 that 32. 6 p e rc e n t
w ere expecting to go to g rad u ate school in the following y e a r.
additional 44 p e rc e n t expected to go a t a la te r date.

25

An

Davis found,

fu rth er, th at the u n derg raduate g rad e point c o rr e la te s w ell with
graduate plans.

He found th at 54 p e rc e n t of the students graduating

in the top fifth of th e ir c la s s planned to go on in the next academ ic
y ea r.

By com parison, only 35 p e rc e n t of the students graduating

in the top half and 22 p e rc e n t of the students graduating in the bottom
2q
half planned to go on.
G rigg re p o rts that a National Science F ounda­
tion two y e a r follow -up study found th at 83 p e rc e n t of the graduating
se n io rs in 1956 e x p re sse d a d e s ire to go to g rad u a te school.

A

survey of the sam e sam ple two y e a r s la te r showed th at 48 p e rc e n t had
actually enrolled in g rad uate school a s of June 1958.

27

G rigg in te rp re te d th is data to m ean that m any students w ere
postponing th e ir g rad u ate work and would be applying to g rad uate
school a t a la te r date.

The data on graduating se n io rs in education

compiled by D avis su p p o rts th is opinion.

He found that m o re e d u c a ­

tion students said they w ere going to g rad u ate school a t a la te r date
than those planning on attending im m ediately.
nev er planned to go to g rad u ate school.

V ery few said they

28

T. Ft. M cConnell notes an o th er so u rc e of p r e s s u r e for
graduate ad m issio n s.

College and u n iv e rsity a d m in is tra to rs often

22

e stab lish off-cam pus c e n te rs and branch cam puses in o rd e r to sa tisfy
the demand for s e rv ic e s that cannot be m et a t the m ain cam pus.
E stab lish m en t of th ese c e n te rs, according to M cConnell, not only
sa tisfie d th ese dem ands but g e n e ra te s additional dem ands from
students in the com m unity who a r e encourated to attend g rad u ate
school because it is now m o re a c c e ss ib le .

29

H uganir, in an a r tic le dealing with p a rt-tim e studen ts,
em phasizes the fact that in d u stry is often a ttra c te d to an a r e a
p a rtia lly on the b a sis of the educational fa c ilitie s that w ill be a v a il­
able to th e ir em ployees.

Tuition refund plans offered a s a fringe

benefit thus becom e an im p o rtan t fa c to r in the num ber of p a rt-tim e
students attending an institu tion.

30

The E ffects of Selective A dm issions
As en ro llm en t p r e s s u r e in c r e a s e s , U n iv ersity a d m in is tr a ­
to rs often r e s o r t to se le c tiv e a d m issio n s to balance the dem and for
se rv ic e s a g a in st the available supply.

Many opinions have been

offered a s to who should do the selectin g , how the selectio n should
be done, and what the scope of the selectio n should be.
H illg arth , a s w as m entioned e a r lie r , re p o rte d th at the
selection d ecision w as given to the faculty in the e a rly y e a rs of
graduate education.

G rigg re p o rts th at the grad u ate school g en erally

d e te rm in e s broad p o licies such a s re sid e n c e re q u ire m e n ts, foreign

23

language re q u ire m e n ts and m inim um c re d it re q u ire m e n ts.
i t is the d ep a rtm e n t th at actually s e le c ts the stu d en ts.

31

However,
B e re lso n

concurs with the statem en t, "The graduate school a d m its stu den ts, to
be s u re , but the d ep a rtm e n ts re a lly s e le c t them (what one u n iv e rsity
p re sid e n t d e s c rib e s a s *som ething am ounting a lm o st to a guild
system with r e s tric tio n of en try a t the d e p a rtm e n ta l le v e l' ). "

32

Thus, the fact th at selectio n is done a t the d e p a rtm e n ta l lev el is w ell
estab lish ed .

T h ere i s som e d ifference of opinion a s to w hether

selectio n should be done a t the d ep a rtm e n ta l level, how ever.

M ay-

hew and Jenson both feel th at selection should be done a t th is level.
They feel th at the objectives and p o licies of each d ep artm en t a re
sufficiently d iffe ren t a s to m ake se le c tio n s outside the d ep a rtm e n t
ineffective.

33

The work of B ottenburg and C h ris ta l a ls o tends to

support selectio n a t the d e p a rtm e n ta l level.

T h eir stu d ies of rating

c h a ra c te ris tic s showed that any grouping of ra tin g s tended to reduce
p red ictiv e efficiency.

They did, how ever, m ake a stron g c a se for

combining ra tin g s in the m o st efficien t m an n er p o ssib le so as to keep
prediction lo se s within to le ra b le lim its .

34

H ills, in h is su rv e y of

prediction re s e a rc h , noted, how ever, that m ultiple R
v ery little differen ce between s im ila r d ep a rtm e n ts.

2

r e s u lts show

The only s i g ­

nificant d iffe ren c es o ccu r between w idely d iffe ren t d ep a rtm e n ts such
as physics and education.

35

Irvine re p o rts , also , th at the U n iv ersity

24

of G eorgia u se s a m u ltip le p red ictio n equation to p re d ic t th e ir
a p p lic a n t's likelihood of su c c e ss a t the un d erg rad u ate level.
sam e equation is applied fo r the whole u n iv ersity .

The

However, a h igh er

sc o re is so m etim es n e c e s s a ry to be adm itted to som e d e p a rtm e n ts.
M ethods of se lec tio n v a ry widely.

36

One type of selectio n

which is becom ing m o re recognized is the p re se le c tio n which tak es
place when the student decid es to apply to a given in stitu tio n .

G rop per

and F itzp a tric k r e p o rt th a t the a c tu a l decision to go to g rad u ate school
usually o cc u rs in the th ird o r fourth y e a r of college.

37

B e re lso n

found th at only 35 p e rc e n t of the students applying fo r the D octorate
degree had m ade the decision by the end of college.

T his is c o n ­

tra s te d with the field of m edicine w here a lm o st half of the m ed ical
students m ade the decisio n before college.

38

Davis noted th at a

definite geographic selectio n is m ade by the student.

He found th at

70 p e rc e n t of the stu den ts in h is sam ple went to grad uate school
within a four hour d riv e of th e ir hom e.

39

B e re lso n re p o rts , also ,

th at m o st stud ents apply to m o re than one school.

His su rv ey of

grad u ate applications fo r the acad em ic y e a r 1958-59 showed that
74.000 students applied to the schools in h is sam p le.

Of th ese,

51.000 w ere accepted and 39,000 new students actu ally showed up.
O r a s B e re lso n s ta te s :
showed up som ew here. "

"About o n e-h alf of the app lications anyw here,
40

Since m o st stud ents apply to m o re than

25

one school, th is lead s to the conclusion th a t everyone g e ts in
som ew here.
Selection p ro c e d u re s a t the individual in stitu tio n s v a ry
from open door p o licies to app licatio n of com plex s ta tis tic a l p r e d ic ­
tion form ulas with asso c ia ted cutting s c o re s .

The b asic re q u ire m e n t

a t any in stitu tio n is a Bachelor* s deg ree with a good acad em ic
rec o rd .

The v ario u s d e p a rtm e n ts add re q u ire m e n ts to th is .

One

of the m o re popular re q u ire m e n ts is the G raduate R e co rd s E x am in a­
tion or the M ille r A nalogies test.

How ever, B e re lso n se e m s to

su m m arize the conclusions of m any a u th o rs with re g a rd to te s ts in
h is statem ent:
Im proving the se lec tio n d ev ices is a t b e st doubtful. At one
tim e, high hopes w ere held fo r the G raduate R ecord E x am in a­
tion, but now the consensus is , I think, th at i t i s useful only for
the candidate from the unknown college, and then only m arg in ally
so.
K urland se e s the n o rm al se le c tiv e a d m issio n s p ro c e s s a s
sta rtin g out a s a logical sy stem and p ro g re ssin g to an ir re le v a n t
system .

He fee ls th at m o st schools o r d e p a rtm e n ts s t a r t with

reaso nable re q u ire m e n ts which a r e re la te d to acad em ic su c c e ss .
Then, a s the num ber of ap p lican ts in c r e a s e s , the re a c tio n of the
d ep artm en t is to keep the sam e m e a s u re s and only r a is e the m inim um
sc o re s.

In th is way, according to K urland, re q u ire m e n ts a r e ra is e d

up until they no longer b e a r any re la tio n sh ip to acad em ic su c c e ss .

42

26

M eder a lso fe e ls th at ad m issio n s sy ste m s g en erally s t a r t
out with reaso n ab le re q u ire m e n ts and becom e ir re le v a n t in tim e.
He w rite s:
A re q u ire m e n t o rig in ally im posed b ecause of its relev an ce
to college cu rricu lu m , having lo s t i ts relev an ce, has been c o n ­
tinued, p erh ap s in a m odified form , p a rtly a s a m a tte r of v ested
rig h t enjoyed by som e college d ep artm en t, p a rtly for re a so n s
of trad itio n , p a rtly for re a so n s of em otion, a ll ju stified by a p p r o ­
p ria te r a t i o n a l i z a ti o n s .^
C h rista l, in h is work quantifying the values underlying ra tin g s, noted
that one of the p ro b lem s in se lec tiv e a d m issio n s is th at we a r e not
capable of weighing the data provided by s e v e ra l v a ria b le s without
the help of a com puter.

44

Yet, v ery few ad m issio n s sy ste m s a r e

based on a c o m p u ter-g e n erate d com posite of the a p p lic a n t's re c o rd .
The scope of the selectio n sy stem is a topic th at is receiving
some attention esp ec ially by advocates of com puter m anaged a d m is ­
sion sy ste m s.

The tra d itio n a l ap proach is fo r the ad m issio n s

com m ittee to sc re e n the ap p lican ts and tu rn the accepted students
over to the faculty fo r the p re s c rip tio n of a p ro g ram .

Walton s u g ­

g e s ts, how ever, that with the ab ility of the com puter to p ro c e s s and
su m m arize la rg e am ounts of data about an applicant, ed u cato rs
should s t a r t to co n sid er m erg in g the ad m issio n and acad em ic advising
functions into a continuous p ro c e s s .

45

E ric R odgers a g re e s with

Walton in h is sta te m e n t th at the ad m issio n s p ro c e s s i s not th at

27

im p o rtan t as long a s i t does not g u aran tee a d eg re e.
is an advocate of a continuous selection p ro c e ss .

Thus, R odgers

46

The p rim a ry goal of a se le c tiv e a d m issio n s p ro g ram is to
control the num ber of stud ents adm itted to g rad u ate p ro g ra m s.

How­

ev er, se lec tiv e ad m issio n s can produce som e im p o rtan t side effects.
One side effect th at h as been noted by H ills is that the student popu­
lation becom es m o re uniform when selectiv e ad m issio n s p ro c e d u re s
a re im posed.
to go up.

The population av erag e of the se lec tio n m e a s u re s tends

A ccording to H ills, the in c re a se d population a v e ra g e affe cts

grading p ra c tic e s in a school.
grade tends to go up.

47

The m inim um stan d ard fo r a given

D uration of the g rad u ate p ro g ram and

a ttritio n r a te s a lso seem to be affected.

G oodrich re p o rts that, in

an open door a d m issio n s situ atio n, b o rd erlin e stud ents a r e often
adm itted and often succeed sim ply b ecau se they a r e p e rs is te n t.
is , they stay in school u ntil they w ear down th e ir p ro fe s s o rs .

That

48

B erelso n re p o rts th at we can g et an idea of the effects of selectio n
on the a ttritio n ra te by com paring the g rad u ate schools with the
m edical schools.
fully.

The m ed ical schools s c re e n app lican ts v e ry c a re -

T his r e s u lts in ap proxim ately 10 p e rc e n t a ttritio n .

The

grad u ate schools, on the o th e r hand, have a m uch e a s ie r screen in g
p ro c e ss, and an a ttritio n ra te of about 40 p erc en t.

49

28

P red ictio n Studies R elated
to Selective A dm iasions
Many au th o rs have gone beyond th e o re tic a l co n sid eratio n s
about se lec tiv e ad m issio n s and have attem pted to re la te p re a d m issio n s
data to actu al p erfo rm an ce in g rad u a te school.

T h e ir stu d ies can be

categorized u nder the broad heading of p red ictio n stu d ies.

The

au th o rs of th ese stud ies use a wide v a rie ty of m e a s u re s of both p e r ­
form ance and p red ictiv e data.

T h e re fo re , i t se e m s usefu l to co n sid er

th e ir m e a s u re s se p a ra te ly .
The rele v an t stu d ies w ill be su m m arize d by m e a s u re s of
p erform an ce f irs t.

The stu d ies w ill then be su m m arize d by p r e ­

dictive m e a s u re s .

R esu lts w ill be su m m arize d with the sectio n on

p red ictiv e m e a s u re s .

M easu res of P erfo rm a n ce
In pred iction stu d ies re la tin g to g rad u a te education the
m e a su re of p erfo rm an ce is alw ays designed to define su c c e s s in
g raduate study.

The m o st common m e a su re of p erfo rm an c e is the

g rad e point av erag e.

Depending on the p urpose of the study, grad e

point can be m ea su red a t v a rio u s points in the p ro g ra m o r calculated
on only a portion of the to tal work com pleted.

Newman c o rre la te d

the g rad e point av erag e fo r 66 psychology stu den ts who had 9 or
m o re c re d it h o u rs with the G raduate R eco rd s E xam ination.

50

P latz

29

used the f i r s t s e m e s te r g rad e point a v e ra g e a s a c r ite r ia to c o rre la te
with u n d erg rad u ate grad e point a v e ra g e and M ille r Analogy T est
s c o re s .

51

Jen so n used the g rad e point av erag e of the c o u rse s taken

in the stu d en t1 s m a jo r a s the c r ite r ia for a m ultiple c o rre la tio n with
un derg rad u ate g ra d e point av e rag e and M iller Analogy T est s c o re s .

52

Borg used g rad e point a v e rag e a fte r 15 c re d it h o u rs in a c o rre la tio n a l
study of the G raduate R eco rd s E xam ination.

53

C areton, Eckhoff,

H e rb e rt, Owens, Payne, and S trie k e r a ll used o v e ra ll g rad e point
av erag e a t graduation in c o rre la tio n a l stu d ies with v ario u s p re d ic to r
v a ria b le s.

54

Payne brought out the point in his study th at the g rad e

point av erag e should not be re g a rd e d a s ra tio data.

T h erefo re , the

m edian g rad e point should be used in place of the m o re commonly
used m ean g rad e point.

T his could be quite a sig n ifican t point since

a ll stu d ies concerned with g ra d e s use the m ean g rad e point.

How­

e v e r, Payne found that the m ean g rad e point and the m edian g rad e
point c o rre la te d . 98.

Thus, he concluded th at i t i s acceptable to

u se the sim p le r m ean g rad e point.

55

A nother m e a su re of p erfo rm an ce is the a s s e s s m e n t o r the
rating of the ad m issio n s com m ittee.

T his type of a m e a su re is

sim ply a sta te m e n t of c u rre n t ad m issio n s policy, a s the p ro sp ectiv e
student h as not enrolled fo r any c o u rs e s a t the tim e of the m e a su re .
M ehrabian used a m ultiple re g re s s io n equation of s e v e ra l p rea d m issio n

30

v a ria b le s to p re d ic t the ratin g of an interv iew com m ittee.

56

Houston

used a re g re s s io n technique developed by B ottenburg and C h ris ta l to
group the r a t e r s into s im ila r policy g ro u p s.

57

W h erry s used a p r o ­

file of fa c to rs a n a ly sis to acco m p lish a grouping of r a t e r s in his
study.

58
A few au th o rs have used faculty ra tin g s of grad u ate students

as a p erfo rm an c e m e a s u re .

This type of a m e a su re is a sta te m e n t

of academ ic policy, a s the student has now taken c o u rse s and engaged
in o th er a c tiv itie s around the school which a r e c h a r a c te r is tic of
graduate stu d en ts.

R oscoe p re se n te d g rad u a te a d v is o rs with a lis t

of re c e n t g ra d u a te s and asked the a d v is o rs to pick the top ten
g rad u a tes.

59

P la tz c o rre la te d p re a d m issio n data with faculty ratin g s

of f ir s t s e m e s te r d o c to ra l stu den ts.

60

C areton, in his study of the

pred ictiv e efficiency of the M ille r A nalogies T est, c o rre la te d te s t
s c o re s with faculty ra tin g s of f i r s t s e m e s te r d o cto ra l students.

61

Some a tte m p ts have been m ade to deal with the graduation no graduation question.

W right re la te d s e v e ra l b io graphical and

academ ic v a ria b le s to g raduation in a ten y e a r longitudinal study.

62

Roscoe and Houston studied a sam ple of d o cto ra l students who
graduated from th e ir p ro g ram v e rs u s a sam p le who w ere d ism isse d
from the p ro g ra m .

63

Irv in e used a m ultiple re g re s s io n technique

to p re d ic t graduation fo r u n d erg ra d u ates on the b a s is of s e v e ra l
academ ic and t e s t m e a s u re s .

31

O ther stu d ies of p erfo rm an ce m e a s u re s include the w ork of
S trie k e r and Huber, who studied the rela tio n sh ip of te s t s c o r e s to
the num ber of m onths between the end of co u rse work and the o ra l
exam inations for d o cto ral students, and the study of Spaeth, who
related s e v e ra l v a ria b le s to the quality of grad u ate school attended.

65

M ehrabian studied the re la tio n of th re e m e a s u r e s of p e r ­
form ance to p re a d m issio n data by combining the m e a s u re s into one
o v era ll index found by adding the Z -sc o re of the th re e m e a s u re s .

66

P re d ictiv e M easu res
P re d ic tiv e m e a s u re s re la tin g to g rad u ate education a r e
g en erally e ith e r m e a s u re s of p rio r p erfo rm an ce o r b iograp hical
data.

B iographical data includes such ite m s as age, sex, m a r ita l

sta tu s, num ber of dependents, o r o th er data which d e s c rib e s the
p erso n al a ttrib u te s of the applicant.

M e a su re s of p r io r p erfo rm an c e

include g rad e point a v e ra g e s for p r io r w ork, te s t s c o r e s , ra tin g s,
o r other v a ria b le s which d e sc rib e the app lican t re la tiv e to som e
stan dard of p erfo rm an ce.
G rade point a v e ra g e s from p r io r work a r e included in a lm o st
a ll p rediction stu d ies in g rad u ate education.

T h e re fo re , i t se em s

useful to su m m a riz e th ese stu d ies by the o th e r v a ria b le s studied.
T e st s c o re s from v ario u s types of stan d ard ized te s ts a r e
often used as p red ictiv e m e a s u re s .

The G raduate R eco rd s

32

Exam ination and the M ille r A nalogies T e st a r e two of the m o re
popular te s ts used.

T hese te s ts a r e u sually used in com bination

with o th er m e a s u re s in a m u ltip le c o rre la tio n equation.

The hig h est

c o rre la tio n between the M ille r A nalogies T e st and g rad e point w as
found by C areton in h is study of 38 g rad u a te students in psychology
in 1949.

He found th at the M ille r A nalogies T e s t c o rre la te d . 68 with

f ir s t s e m e s te r g ra d e point av e rag e .

67

T his study has been rep o rted

consistently as the hig h est c o rre la tio n found in the rev iew s of the
lite ra tu re of o th er au th o rs who a r e studying the p re d ic tiv e power of
standard ized te s ts .
considerably low er.

M ost c o rre la tio n s involving te s t s c o re s a r e
Payne found a c o rre la tio n of .26 between the

M iller A nalogies T e st and g rad e point a v e ra g e for 219 stud ents who
com pleted a M aster* s d e g re e in education.
R

2

68

Jen son found a m ultip le

of . 489 using the u n d erg rad u ate g ra d e point a v e ra g e and the

M iller A nalogies T e st to p re d ic t g rad u ate g rad e point av erag e.
Jenson also used s c o re s from the Iowa M athem atical Aptitude T e st
in his study, but was unable to e s ta b lis h a significant rela tio n sh ip
between th ese s c o re s and the g rad u ate g ra d e point.

69

Eckhoff, in

h e r study of Master* s d e g re e stu d en ts in education, found a m ultip le
R

2

of , 51 using u n d erg rad u ate g ra d e point av e rag e , the M ille r

Analogies T est, and the G raduate R eco rd s E xam ination.

70

M ehrabian

and P la tz used the M ille r A nalogies T e st s c o r e s to p re d ic t faculty

33

ratin g s.

They both found the MAT to be significantly re la te d to the

faculty ra tin g s.

71

Studies involving the G raduate R eco rd s E xam ination seem to
produce c o rre la tio n s of ap prox im ately the sam e m agnitude a s the
M iller A nalogies T est.

Borg, in h is study of grad u ate students in

education a t Utah State, found th at the GRE v e rb a l s c o re c o rre la te d
. 36 with g rad u ate g ra d e point av e rag e .
c o rre la te d . 37 with GPA.

The GRE quantitative sc o re
72

N e ith e r of th ese was significant.

Newman

found th at the GRE v e rb a l and the quantitative s c o re s c o rre la te d . 21
and . 18 re sp e c tiv e ly with g rad u ate GPA fo r his sam ple of 66 grad u ate
students in psychology.

73

R oscoe, S trie k e r, and Houston each used

GRE s c o re s in m ultiple c o rre la tio n s stu d ies.

T h eir re s u lts con cu rred

with the re s u lts of Jenson and Eckhoff in finding that the un derg rad u ate
grade point accounted for the m a jo rity of the v aria n ce in the m ultiple
c o rre la tio n equation.
H e rb e rt, in h is study of education g rad u ate students a t
Duquesne, used s c o re s from the N ational T e a c h e rs Exam ination in
a m ultiple c o rre la tio n study to p re d ic t g rad uate g rad e point.

How­

ev er, he found th at the only significant c o rre la tio n (.57) w as between
the u n d ergradu ate g rad e point and the g rad u ate g ra d e point.

75

Owens used the u n d erg rad u ate g ra d e point, a s c o re on the
W atson G a se r C ritic a l Thinking T e st, graduation from a p riv ate o r

34

public un derg rad u ate school, and w hether the su b ject was an Ohio
State g rad u ate o r not to p re d ic t g rad u ate g rad e point a v e rag e for
education stud ents.

He re p o rte d a m ultiple R

significant a t the . 01 level.

2

o r .524, which was

76

R atings of the quality of the un d erg rad u ate school o r d e p a r t­
m ent have also been used a s p red ictiv e m e a s u re s .

M ehrabian,

Owens, and Spaeth used ra tin g s of the u n d ergraduate school
developed from a selectiv ity index published in a docum ent which
surveyed u n dergrad uate schools.

77

H e rb e rt used se v e ra l m e a s u re s including y e a rs of teaching
experience, u n derg raduate m ajo r, u n d ergradu ate student teaching
grade and y e a r graduated with B. A. in a m ultiple c o rre la tio n study.

78

Age, sex, and m a rita l statu s a r e the m o st common b io ­
graphical data used in p red ictiv e studies.

Spaeth used father* s

education and fam ily incom e, in addition to age and sex.

79

W right,

in his 10 y e a r longitudinal study, found age to be an in v e rse in d ic ato r
of su c c e ss.

80

M ehrabian found sex to be an insignficant in d icato r

in h is study of graduate stud ents a t UCLA.

81

Summary
While th e re is consid erable d ifferen ce in the lite ra tu re as
to what im portan ce should be attached to the M aster* s d eg re e, i t is

35

evident that la rg e n um b ers of people have earned and w ill continue to
e a rn the M a s t e r 's d eg ree.

This is p a rtic u la rly tru e of education,

w here o n e-h alf of a ll M aster* s d e g re e s a r e aw arded.
E a rly educational sy ste m s only provided education for the
children of nobility.
v ery selectiv e.

Even the e a rly system in the United S tates was

However, c u rre n t th eo ry holds that ev e ry man is

entitled to the m axim um am ount of education he can profitably u se.
Our sy stem of h ig h er education has responded to th is p r e s s u r e by
establishing Ju n io r C olleges, branch cam puses, and off-cam pus
grad uate c e n te rs in addition to expanding the fa c ilitie s a t the m ain
cam puses.

This h as allowed m o re people to com plete underg raduate

work and s t a r t gradu ate work.

In addition, th e re is a national trend

toward continuing into g rad u ate w ork im m ed iately following the u n d e r ­
graduate work.

These two fa c to rs have brought about a la rg e in c re a s e

in demand fo r g rad uate education.
This in c re a sin g dem and, in tu rn , has brought about p r e s ­
su re for se lec tiv e ad m issio n s from the a d m in is tra to rs and faculty
of graduate schools who find th e ir available re s o u rc e s depleted.
yet, th e re is no stan d ard method fo r selectio n .

As

V ery little is known

about the a c c u ra c y o r the long range effects of selection.
P re d ictio n stu d ies a r e often used to a tte m p t to in c re a s e the
accu racy and a s s e s s the effects of se le c tiv e a d m issio n s p ro g ra m s.

36

The o v erall g rad u a te g rad e point is by fa r the m o st popular c rite rio n
in predictio n stu d ies.

How ever, faculty ra tin g s, p e e r ra tin g s, d u r a ­

tion, and g raduation o r no graduation have been used a s c r ite r ia .
The m o st popular p red ictiv e v a ria b le s have been u n d erg rad u ate g rad e
point av e rag e s and stan d ard ized te s t s c o r e s .

H ow ever, sex, m a rita l

sta tu s, age, quality of the u n d erg rad u ate school, u n d erg rad u ate m a jo r,
and o th er v a ria b le s have also been used.

FOOTNOTES

1 B ern ard B erelso n , G raduate E ducation in the United
States (New York: M cG raw -H ill Book Company, I960), p. 185.
2Ibid.
3Ibid.
4

John L. Snell, "The M a s t e r 's D egree, " in G raduate E d u ­
cation Today, ed. by E v e re tt W a lte rs (W ashington, D .C .: A m erican
Council on Education, 1965), p. 74.
5
B erelso n , op. c it. , p. 186.
Snell, op. c it. , p. 80.
7

C h a rle s M. G rigg, G raduate Education (New York: The
C enter fo r Applied R e se a rc h in Education, Inc. , 1965), p. 57.
O

Snell, op. c it. , p. 80.
9

J oh n S. B r u b a c h e r , B a s e s f o r P o l i c y in H ig h e r E d u c a ­
tio n (N ew Y ork : M c G r a w - H i l l B o o k C o m p a n y , 1965)7 p 3^
10Ibid. , p. 4.
n ibid. , p. 5.
12

T. R. McConnell, A G en eral P a tte rn for A m erican P ublic
H igher Education (New York: M cG raw -H ill Book Company, 1962),
p. 15.
13

H igher Education in a Decade of D ecision (W ashington,
D .C . : N ational E ducation A ssociatio n, E ducational P o lic ie s C o m ­
m issio n , 1959), p. 21.

37

38
14

Algo D. H enderson, P o lic ie s and P r a c tic e s in H igher
Education (New York: H a rp e r and B ro th e rs, P u b lis h e rs , 1960),
p. 49.
15
10

B ru b ach er, op. cit. , p. 8 .
H igher Education in a Decade of D ecision, op. c i t ., p. 33.

17

B ru b ach er, op. c i t . , p. 13.

18

Snell, op. c i t . , p. 1 0 0 .

19

Reid F. H arle and John T. Thompson, "L eg a l and P o li t i ­
cal Im plications of P o lic ie s Governing A d m issio ns to P ublicly Sup­
ported In stitu tio n s, " College and U n iv ersity , XLHI, No. 3 (Spring
1968), 274-285.
20

C atherine R. Rich and Thom as G a rre tt, eds. , Philosophy
and P ro b le m s of College A dm issions (W ashington, D. C. : C atholic
U niversity of A m erica P r e s s , 1963), p. 4.
21

Edwin A. B room e, A H isto ric a l and C ritic a l D iscussio n
of College A dm ission R eq u irem en ts (New York: Colom bia U n iv ersity
P r e s s , 1903), p. IT!
22

J . W. H illgarth, "G raduate A dm issions - -Havoc into
C h a o s," College and U niv ersity , XLII, No. 4 (Sum m er 1967), 498504.
23

Snell, op. c it. , p. 76.

24

Byron L G roesbeck, "G raduate A d m ission s - -A re We
P re p a re d fo r the Y e ars A h ead ?" C ollege and U n iversity, XLII,
No. 4 (Slimm er 1967), 504-510.
25

Ja m e s A. D avis, G re a t A sp ira tio n s (Chicago: Aldine
Publishing Company, 1964), p. 43.
2 6Ibid. , p. 81.
27
28

G rigg, op. c i t . , p. 8 8 .
D a v i s , o p . c i t . , p. 4 9 .

39
29
McConnell, op. c i t . , p. 10.
30

G eorge H. H uganir, "The A p p raisal and Retention of the
P a r t Tim e Student, " P ro c ee d in g s of the Fifth Annual M eeting of the
Council of G raduate Schools in die U. S. (W ashington, D. C. : Council
of G raduate Schools, 1965).
31
32

G rigg, op. c i t . , p. 27.
B erelso n , op. cit. , p. 121.

33

Ralph E. Jenso n, "P re d ic tin g Scholastic A chievem ent of
F i r s t Y ear G raduate Students, " E ducational and P sychological
M easu rem ent, XIII, No. 2 (Sum m er 1953), 322-329; and Lewis B.
Mayhew, "N ontest P re d ic to rs of A cadem ic A chievem ent, " Educa tional and P sy cholog ical M easu rem ent, XXV, No. 1 (Spring 1965),
39-46.
34

R o b ert A. B ottenburg and Raymond E. C h rista l, "G ro u p ­
ing C r ite r ia - - A Method Which R etains M aximum P re d ic tiv e E fficiency, "
Jo u rn a l of E x p erim en tal Education, XXXVI, No. 4 (Sum m er 1968),
28-34.
35

John R. H ills, M arilyn B. Gladney, and Josep h A. Klock,
"Nine C ritic a l Q uestions About Selective College A dm issions, " The
P erso n n e l and Guidance Jo u rn a l, XLV, No. 7 (M arch 1967), 640647.
36

Donald W. Irv in e, "M ultiple P re d ic tio n of College G ra d u a ­
tion from P re a d m issio n D ata, " Jo u rn a l of E x p erim en tal Education,
XXXV, No. 1 (F a ll 1966), 84-89.
37

G eorge L. G ropper and R o b ert F itz p a tric k , Who G oes to
G raduate School? (P ittsb u rg h , P a. : A m erican In stitu te for R e se a rc h ,
1959), p. 13.
38
39
40

B erelso n, op. c i t . , p. 144.
D avis, op. c i t . , p. 81.
B e r e l s o n , o p . c i t . , p. 144.

4 1 Ibid, , p. 145.

40
42

N orm an D. K urland, "M ore People Than P la c e s o r How
to E n su re Equal O pportunity Among E quals When T h ere A re M ore
E quals Than O pportun ities, MCollege and U n iv e rsity , XLIII, No. 2
(W inter 1968), 221-224.
43

A lb e rt E. M eder, "A dm issions and the College, " College
A dm issions (P rin ceton : College E n tran ce Exam ination Board] 1963),
p. 42.
44

Raymond E. C h rista l, "JAN: A Technique for Analyzing
Group Judgm ent, " The Jo u rn a l of E x p erim en tal Education, XXXVI,
No. 4 (Sum m er 1968), 24-27.
45W esley W. W alton, "P o te n tia litie s of the C om puter fo r
M easurem en t and P re d ictio n with R esp ect to College A dm issions
P ro c e s s e s , " E ducational and P sychological M e a su re m e n t, XXV,
No. 1 (Spring 1965), 47-53.
46

E r ic R odg ers, "A dm ission of the D om estic G raduate
Student, " P ro c ee d in g s of the Fifth Annual Meeting of the Council of
G raduate Schools in the U .S . ^Washington, D .C .: Council of G radu ate Schools, 1965), pp. 17-19.
47

H ills, Gladney, and Klock, op. c it. , pp. 640-647.

48

Max G oodrich, "The A p p ra isal and R etention of G raduate
Students, " P ro c ee d in g s of the Fifth Annual Meeting of the Council of
G raduate Schools in the U .S . (W ashington, D .C . : Council of G ra d u ­
ate Schools, 1965), pp. 33-37.
49

B erelso n , op. c it. , p. 168.

50

R ichard I. Newman, "GRE S co res a s P re d ic to rs for
Psychology G raduate Students, " E ducational and P sychological
M easurem ent, XXVIII, No. 2 (Sum m er 1968), 433-436.
51

A rth u r P latz and C h a rle s M cClintoch, "U ndergraduate
G rades and the M ille r A nalogies T e s t a s P re d ic to rs of G raduate
S u c c e ss," The A m erican P sy ch o lo g ist, XIV, No. 5 (May 1959),
285-289.
52

Ralph E. Jenson, "P re d ic tin g S cholastic A chievem ent of
the F ir s t Y ear G raduate Student, " E ducational and P sychological
M easurem ent, XIII, No. 2 (Sum m er 1953), 322-329.

41
53

W alter R. B org, "GRE A ptitude S co res a s a P re d ic to r of
GPA fo r G raduate Students in Education, ME ducational and P s y c h o ­
logical M easu rem en t, XXIII, No. 2 (Sum m er 1963), 379-382.
54

Edw ard E. C areto n and L ouise C areton, '’P re d ic tio n of
Success in G raduate Study of Psychology a t U n iv ersity of T en n essee, "
A m erican P sy ch o lo g ist, IV, No. 8 (August 1949), 361-362; C onstance M.
Eckhoff, "P re d ic tin g G raduate S uccess a t Winona State C ollege, "
E ducational and P sychological M easurem ent, XXVI, No. 2 (Sum m er
1966), 483-485; David J . H e rb e rt, "A P re d ic tiv e Study of Q uality
P o in t A v e rag e s in G raduate Education C o u rse s, " The Jo u rn a l of E d u ­
cational R e se a rc h , LX, No. 5 (Jan u ary 1967), 218-220; Thom as R.
Owens, "P re d ic tin g A cadem ic S uccess in M aster* s D egree P ro g ra m s
in E ducation, " The Jo u rn a l of E ducational R e se a rc h , LX, No. 10
(July-A ugust 1967), 124-126; David A. Payne and Cynthia E. T uttle,
"The P re d ic tiv e R elationship of the M ille r A nalogies T e s t to the
O bjective-S ubjective C r ite r ia of S uccess in a G radu ate School of E d u ­
cation, " E ducational and P sy cholo gical M easu rem en t, XXVIII, No. 2
(Sum m er 1968), 427-430; G eorge S trie k e r and J . T. H uber, "The
GRE and U ndergrad uate G ra d es a s P re d ic to r s of S uccess in G raduate
S chool," The J o u rn a l of E ducational R e s e a rc h , LX, No. 10 (Ju ly August 1967), 466-468.
55

Payne and T uttle, op. c i t . , pp. 427-430.

56

A lb e rt M ehrabian, "U ndergraduate A bility F a c to rs in
R elation to G raduate P e rfo rm a n c e , " E ducational and P sycho lo gical
M e a su re m e n t, XXIX, No. 2 (Sum m er 1969), 409-419.
57

Sam uel R. Houston, "G enerating a P ro je c te d C rite rio n
of G raduate School S uccess Via N orm ative Judgm ent A n alysis, " The
Jo u rn a l of E x p erim en tal E ducation, XXXVII, No. 2 (W inter 1968),
64-69.
58

R o b ert J . W h e rry s and J a m e s C. N aylor, "C om parison
of Two A pproaches - -JAN and P R O F --fo r C apturing R a te r S tra te g ie s ,"
E ducational and P sycho logical M easu rem en t, XXVI, No. 2 (Sum m er
1966), 267-268.
59

John T. R oscoe and Sam uel R. Houston, "The P re d ic tiv e
V alidity of GRE S cores fo r a D o cto ral P ro g ra m in E ducation, "
E ducational and P sycho logical M easu rem en t, XXIX, No. 2 (Sum m er
1969), 507 -509.

42

an
61

P la tz and M cClintoch, op. c it. , pp. 285-289.
C areton and C areton, op. c it. , pp. 361-362.

62

C h a rle s R. W right, "S uccess o r F a ilu re in E arning
G raduate D eg rees, " Sociology of E ducation, XXXVIII, No. 1 (F a ll
1964), 73-97.
63
64

R oscoe and Houston, op. c i t . , pp. 507 -509.
Irv in e, op. c it. , pp. 84-89.

65

S trie k e r and H uber, op. c i t ., pp. 466-468; Jo e L. Spaeth,
"The A llocation of College G raduates to G raduate and P ro fe ssio n a l
Schools, " Sociology of Education, XLI, No. 4 (F a ll 1968), 342-349.
66

M ehrabian, op. c it. , pp. 409-419.

6V
C areton and C areton, op. c it. , pp. 361-362.
68

Payne and T uttle, op. c it. , pp. 427 -430.

69
70

Jen so n , op. c it. , pp. 322-329.
Eckhoff, op. c i t . , pp. 483-485.

71

M ehrabian, op. c it. , pp. 409-419; P la tz and M cClintoch,
op. c it. , pp. 285-289.
72
73

Borg, op. c it. , pp. 379-382.
Newman, op.

c i t . , pp. 433 -436.

74

R oscoe and Houston, op. c it. , pp. 507 -509; S trie k e r and
Huber, op. c it. , pp. 466-468; Houston, op. c it. , pp. 64-69.
75
76
77
78

H e rb e rt, op. c i t . , pp. 218-220.
Owens, op. c it. , pp. 124-126.
M ehrabian, op. c it. , pp. 409-419.
H e r b e r t , op. c i t . , pp. 2 1 8 -2 2 0 .

7 9 S p aeth .

op. c i t . ■ PP- 3 4 2 -3 4 9 .

80W r ig h t. OP- cit_- ■ p p ' 7 3 "7 9 ’
8 1 M eh ra b ia n .

QP- c i t . . PP- 4 0 9 -4 1 9 .

C H A P T E R HI

GENERAL DESIGN AND METHODOLOGY

The design of the study is p re se n te d in th re e g e n e ra l
sectio n s:

( 1) the sam p le, (2 ) n atu re and so u rc e s of data, and

(3) a n a ly sis p ro c e d u re s.

The Sample
The population from which the sam p le was se lec te d c o n ­
siste d of stud ents adm itted into the g rad u ate p ro g ram a t the
M a s t e r 's level who fit in one of five c a te g o rie s during the w inter,
spring, su m m er, and fall te rm s of 1969.

The five c a te g o rie s w ere:

(1) students who graduated with a M aster* s d eg re e, (2) stud ents who
graduated with a M aster* s d eg re e, and went im m ediately to a
d o cto ral p ro g ram a t M ichigan State, (3) stud ents who w ere su b ject
to academ ic action, (4) stu dents who w ere req u e sted to leave the
p ro g ram , (5) students who had not en ro lled for five y e a rs .
The sam ple of students who g rad u ated with a M aster* s
d egree w as selected by g ath erin g the n am es of g ra d u a te s from the
final d eg re e c e rtific a tio n lis ts published by the re g istra r* s office
44

45

for w in ter, sp rin g , su m m er, and fall te rm s 1969.

T hese sam e

lis ts w ere then com pared to the file of c u r r e n t d o cto ra l stu den ts to
s e le c t the sam ple of stud en ts who continued into the d o c to ra l p ro g ram
in the C ollege of Education a t M ichigan State U n iversity .
The sam ple of stu d en ts su b ject to academ ic actio n was
selected from the lis ts of students rem oved from re g u la r sta tu s to
p ro v isio n al sta tu s o r from p ro v isio n al sta tu s to nondegree sta tu s.
T hese lis ts a r e published by the G raduate Student A ffa irs Office in
the College of Education.
this office.

A cadem ic action is u n d er the co n tro l of

A student is su b ject to academ ic action when h is to tal

grade point a v e rag e fa lls below a 3. 0 fo r re g u la r and p ro v isio n al
students o r below 2 .0 for nondegree students.

Specific actio n taken

is also governed by the num ber of c re d its a studen t h as accum ulated.
The sam ple of students who w ere asked to leave the p ro g ram
consisted of the total population of stud ents d ism isse d from the
graduate p ro g ram a s of fa ll te rm 1969.

Students a r e g e n e ra lly d i s ­

m issed a s a r e s u lt of not m eeting the p ro v isio n s of p rio r academ ic
action.

However, som e students a r e in th is category b ecause of

cheating o r poor re la tio n s with the U n iv ersity.
The sam ple of stu d en ts who have not been en ro lled fo r five
y e a rs w as selected by se a rc h in g the g rad u a te student re c o rd file s
on the College of Education IBM 1130 com puter.

A FORTRAN

46

p ro g ram which p rin ted a lis t of a ll M aster* s students who had not
en ro lled fo r a t le a s t 20 te rm s p rio r to fall 1969 was used fo r th is
purpose.

Since a ll c re d its used fo r a M aster* s d eg re e m u st be taken

within five y e a rs p r io r to the te rm of graduation, this sam ple r e p r e ­
sen ts those stud ents who w ere in good standing a t the tim e of la s t
en ro llm en t, but a r e not elig ib le to count any of th e ir work for a
M aster* s d eg re e a t th is tim e.
The to ta l sam ple num bered 358.

The num ber in each c a te ­

gory is su m m arize d in Table 3 .1 .

TABLE 3. 1
TOTAL SAMPLE DIVIDED BY CATEGORY
Num ber
in C ategory

C ategory

1.

Students graduating with a M aster* s d eg re e

199

2.

Students graduating with a M aster* s d eg re e and
continuing to Ph. D.

16

3.

Students subjected to acad em ic action

50

4.

Students asked to w ithdraw

19

5.

Students not en ro lled fo r 5 y e a rs

90

The sam ple of stud en ts who grad u ated with a M aster* s d e g re e w as
exam ined by d ep a rtm e n t fo r one hypo th esis.

The num ber of g r a d u ­

a te s in each d ep a rtm e n t is su m m arize d in Table 3 .2 .

47

TABLE 3.2
GRADUATES DIVIDED BY DEPARTMENT

D epartm ent

Num ber
in D epartm ent

Nondegree

36

E lem en tary Education

53

Secondary Education

44

H ealth, P h y sical Education, and R ecreatio n

6

A dm in istratio n and H igher Education

24

Counseling, Guidance, P e rso n n e l S e rv ic e s, and
E ducational Psychology

36

The N ature and Source of Data
The following v a ria b le s w e re rec o rd ed o r computed from
the o rig in al application fo r grad u ate study for each su b ject in the
study:
1.

Age a t tim e of application to grad u ate p ro g ram .

2.

M arital sta tu s.

3.

Sex.

4.

Born in M ichigan o r out of sta te .

5.

Legal resid e n ce in M ichigan o r out of sta te .

6.

Y e ars a t legal resid e n c e .

48

7.

N um ber of colleges attended for B .A .

8.

N um ber of y e a rs from s t a r t to finish of B .A ,

9.

G raduated with B .A . from M ichigan State, o th er Michigan
u n iv e rsity , o r out of sta te college.

10.

N um ber of y e a rs from end of B .A . to M. A. application.

11.

U ndergraduate grade point av erag e.

12.

G raduate g rad e point av erag e.

13.

G raduate c re d its .

14.

A dm ission sta tu s (nondegree, provisional,

15.

M ajor d ep artm en t a t tim e of ad m issio n s.

o r re g u la r).

In addition, the following v a ria b le s w ere reco rd ed o r c o m ­
puted from the grad u ate tr a n s c r ip t for each su b ject in the graduated
sam ple:
1.

T otal te rm s attended for M a s t e r 's deg ree.

2.

T otal te rm s m isse d since f i r s t te rm enrolled.

3.

A verage co u rse load.

4.

T otal g rad e point av erag e.

5.

T otal g rad u ate c re d its.

6.

Total c re d its d e fe rre d .

7.

T otal d uration of the M aster* s p rogram in te rm s .
The purpose of th is study is to explore die relatio n sh ip of

p read m issio n v a ria b le s to v ario u s m e a s u re s of p erfo rm an ce.

49

T h erefo re , the independent v a ria b le s w ere se le c te d b ecau se they did
not involve any m e a s u re s of p erfo rm an ce a f te r ad m issio n .

The

dependent v a ria b le s w ere se lec te d because they w ere m e a s u re s of
p erfo rm an c e a fte r the com pletion of a p ro g ra m .

A nalysis P ro c e d u re s
D isc rim in a n t a n a ly sis, m ultiple r e g r e s s io n a n a ly sis,
canonical c o rre la tio n , and fa c to r a n a ly sis w ere the b a sic p ro c e d u re s
used in the s ta tis tic a l tre a tm e n t of the data.
E xtensive screen in g of the data w as n e c e s s a ry p r io r to the
ac tu al a n a ly sis p ro c e d u re s.

T his w as n e c e s s a ry b ecause the sam ple

included students who w ere adm itted a s soon a s one y e a r p r io r to
graduation and a s long a s seven y e a r s p r io r to the tim e of the study.
R ecord keeping p ra c tic e s have not rem ain ed co n siste n t during th is
tim e span.

This re su lte d in m any su b jects with incom plete data.

The m a jo rity of th ese people w ere elim inated in the data coding
operation.

A fter the data w ere coded and tr a n s f e r r e d to IBM c a rd s,

an IBM 1130 p ro g ram w as w ritten to fu rth e r sc re e n the data.

A ll

v a ria b le s w ere checked fo r rea so n a b le range and im p o ssib le co n d i­
tions.

The e r r o r s w ere flagged and c o rre c te d .

four re c o rd s w ere found to contain m issin g data.

A fter th is p ro ce d u re,
The m issin g data

fields in these re c o rd s w ere assig n ed the m ean of the sam ple for that
p a rtic u la r v a ria b le .

In addition, the sam e IBM 1130 p ro g ram

50

converted the y es o r no a n sw e rs fo r nondegree, p ro v isio n al, o r
re g u la r ad m issio n sta tu s into an index of 1 fo r nondegree, 2 for
p ro v isio n al, and 3 fo r re g u la r.

A y es o r no resp o n se for p r io r

g rad u ate ex p erien ce was also c re a te d and the a v e rag e co u rse c o m ­
puted from the total c re d its and the total te rm s attended v a ria b le s.
The c o rre la tio n of se v e ra l p re d ic to r v a ria b le s with som e
c rite rio n of s u c c e ss i s a common p ro ce d u re.

However, Tiedem an

points out th at when the c rite rio n is m em b ersh ip in a group, and the
m ultiple c o rre la tio n o r m ultiple re g re s s io n technique is used, the
in v e stig a to r is forced to co n sid er the grou ps in p a irs . 1 Thus, when
it is d e s ire d to p re d ic t group m e m b e rsh ip in one of s e v e ra l g ro u p s,
the m ultiple re g re s s io n m odel i s not a p p ro p ria te .

The m odel s u g ­

gested by Tiedem an is m ultiple d isc rim in a n t a n a ly sis.

A s a re s u lt,

m ultiple d isc rim in a n t a n a ly sis w as used to te s t the p o ssib ility of
predicting group m em b ersh ip in one of the five sam ple g ro u p s.
Since the independent v a ria b le s ap p eared to be m o stly u n c o rre la te d ,
all of the p rea d m issio n v a ria b le s w ere used in th is a n a ly sis.

The

an aly sis w as run on the CDC 3600 com puter using the m u ltiv a ria te
s ta tis tic a l package which w as p ro g ram m ed by J e re m y Finn.

2

In o rd e r to ex plore the d iffe ren c es in the independent v ari ables between the g raduated g rou ps and the o th er g roup s in the
sam ple individually, th re e m ultiple d isc rim in a n t functions w ere

51

calculated.

The m u ltiv a ria te s ta tis tic a l package w as also used for
3

these calcu latio n s.

The s e r ie s of four m u ltiple d isc rim in a n t

a n a ly se s com pleted the testin g of the hypotheses concerning group
m em b ersh ip .
The g rad uated group w as se lec te d fo r the rem aining s t a t i s t i ­
cal tre a tm e n ts.

This was done in o r d e r to exp lo re the rela tio n sh ip

of the p re a d m issio n data with s e v e ra l m e a s u re s of p erfo rm an ce in
the M a s t e r 's d e g re e p ro g ra m .
The IBM 1130 Step-W ise M ultiple R e g re ssio n P ro g ra m was
used to form a p redictio n equation re la tin g the independent v a ria b le s
with the duratio n of the M a s t e r 's p ro g ra m .

4

The duration was

m ea su red in te r m s and w as co n sid ered to be the to ta l of the num ber
of te rm s en ro lled and the num ber of te rm s m isse d from the f ir s t
te rm en ro lled to the te rm of graduation.
D uration a s a m e a su re of p erfo rm an c e is actually a com bination of total te rm s attended, total te r m s m isse d , and av erag e
course load.

Cooley and Lohnes su g g est that the technique of

canonical c o rre la tio n should be used when i t i s n e c e s s a ry to re la te
5
two s e ts of m e a su re m e n ts m ade on the sam e su b je cts.
This te c h ­
nique produces the m axim um c o rre la tio n betw een lin e a r functions of
two se ts of v a ria b le s .

It is , in effect, the c o rre la tio n between two

m ultiple re g re s s io n equations.

In o r d e r to te s t the rela tio n sh ip

52

between d u ratio n , te rm s attended, te r m s m isse d , a v e ra g e co u rse
load, and the independent v a ria b le s , then, a canonical c o rre la tio n
was run using the m u ltiv a ria te s ta tis tic a l package on the CDC 3600
com puter. 6
The to ta l g rad u ate g rad e point, total g rad u a te c re d its , and
total c re d its d e fe rre d w ere a ls o c o rre la te d with the independent
v a ria b le s to provide a m e a s u re of acad em ic p erfo rm an c e.

The

canonical c o rre la tio n technique w as a lso used fo r th is a n a ly sis.
F inally, a fac to r a n a ly sis of a ll of the independent v a ria b le s
was run for the graduated sam ple.

The purpose was twofold.

F ir s t,

it w as co n sid ered d e s ira b le to sim plify the p red ic tiv e m e a s u re s .
T hat is , a few com posite p red ic tiv e m e a s u re s would be p re fe ra b le
to m any seem ingly disconnected m e a s u re s when evaluation of an
ap p lican t1 s potential is d e s ire d .

Second, i t w as d e s ire d to explore

the p ossib le d ifferen ce in en tra n ce c h a r a c te r is tic s among the g r a d u ­
a te candidates by d ep artm en t.

In th is way, the p o ssib ility of p r e ­

dicting the m o st a p p ro p ria te d e p a rtm e n t for an ap p lican t could be
explored.
The fa c to r a n a ly sis p ro g ra m in the IBM 1130 S ta tistic a l
System w as used for this purpose.

7

The p ro g ram w as run for the

e n tire gradu ated sam ple and fo r each d ep artm en t re p re se n te d in the
sam ple.

53

The p ro g ram p erfo rm ed the p rin cip al ax is, orthoginal o r
v arim ax , and the oblique o r prom ax ro tatio n s.

F a c to rs w ere c o m ­

puted whose c h a ra c te ris tic v e c to rs had asso c ia ted c h a r a c te r is tic
ro o ts g r e a te r than o r equal to one.

The r e s u lts of the oblique, or

prom ax, solution w ere used in the a n a ly sis.

This solution was

selected because the re s tr ic tio n of independence between fa c to rs is
relax ed .

T his re s u lts in fa c to rs which a r e e a s ie r to in te r p r e t than

the o ther solutions.

How ever, the fa c to rs a r e slightly i n te r c o r r e -

lated in the oblique solution.

A sam ple of M a s t e r 's d e g re e students belonging to one of
five c a te g o rie s a s of w in ter, sp rin g , su m m er, o r fall 1969 was
drawn.

The rela tio n sh ip of se le c te d p rea d m issio n data with p red icted

category m em b ersh ip was explored

In addition, the rela tio n sh ip of

the p rea d m issio n data fo r the sam ple of graduated students with
se v e ra l m e a s u re s of p erfo rm an c e w as explored.

F inally the c h a r ­

a c te r is tic s of the p rea d m issio n data for the graduated sam ple w ere
exam ined by m a jo r d ep artm en t.

FOOTNOTES

*David V. Tiedem an, "The M ultiple D isc rim in an t Function
--A Symposium, " H arv ard E ducational Review, XXI, No. 2 (Spring
1951), 167-181.
2

J e re m y D. Finn, Multi v aria n ce : F o rtra n P ro g ra m fo r
U nivariate and Multivariage~~Analysis of V arian ce and C ovariance
(Buffalo, N. Y. : The State U n iv ersity of New York a t Buffalo, 1967).
3Ibid.
4

M ason P . R osenthal, 1130 Step-W ise M ultiple R e g re ssio n
P ro g ra m (San F ra n c isc o , C alifornia: IBM C orporation, 1966).
5
W illiam W. Cooley and P aul R. Lohnes, M ultiv ariate P ro cedures for the B ehavioral Sciences (New York: John Wiley and
Sons, 1962), p. 39.
6 Finn, op. cit.
7

1130 S ta tistic a l SyBtem (White P la in s, New York: IBM
C o rp o rati on] 1967).

54

C H A P T E R IV

FINDINGS

The hypotheses w ere teste d and the data analyzed by
computing m ultiple d isc rim in a n t a n a ly sis, m ultiple re g re s s io n
an a ly sis, canonical c o rre la tio n s , and fac to r a n a ly sis.

The . 05

level of significance w as p re s e le c te d a s the c r ite rio n fo r reje c tin g
the null hypothesis w here applicable.

Table 4. 1 p re s e n ts the v a r i ­

able d e sc rip tio n s and the corresponding ab b rev iatio n s th a t w ill be
used in the tab les in th is chapter.

H ypotheses
Group M em bership
H ypothesis 1. --H y p o th e sis 1 s ta te s th at the p rea d m issio n
data is significantly re la te d to m em b ersh ip in the graduated,
graduated to d o cto ra l p ro g ra m , not enro lled fo r five y e a rs , academ ic
action, o r d ism isse d group.

To te s t th is hypothesis, a m ultiple

d isc rim in a n t a n a ly sis was computed using the p re a d m issio n data a s
the independent v a ria b le s and m em b ersh ip in one of the five groups
as the dependent v a ria b le .

M ultiple d isc rim in a n t a n a ly sis w as used

55

56

TA BLE 4. 1
VARIABLE ABBREVIATIONS AND DESCRIPTIONS
D escrip tio n

A bbreviation
M-S

M a rrie d o r single

SEX

Sex

RES

In sta te o r out of sta te re s id e n t

YRS-RS

Y e ars a t resid e n ce

AGE

Age a t tim e of application to g rad u ate school

BORN

Born in s ta te o r out of sta te

GRAD

G raduated from M. S. U. , o th er M ichigan u n iv er sity , o r out of sta te u n iv e rsity

NO-COL

N um ber of colleges attended fo r B .A .

BA-MA

N um ber of y e a rs from end of B .A . to M. A. appli cation

BA-BA

N um ber of y e a r s from s t a r t of B .A . to end of B .A .

U-GPA

U ndergraduate g rad e point av erag e

G -E X P

G raduate e x p e rie n c e --y e s o r no

ADM

A dm ission sta tu s

MAJOR

D ep artm en t assig n ed a t the tim e of ad m issio n

T -G P A

T otal g rad e point a v e rag e fo*. ]V :a s t e r 1 s work

T - CRED

T otal c re d its

C R -D E F

T otal c re d its d e fe rre d in g rad u ate work

T -A T T

T otal te rm s attended

T-M ISS

T otal te rm s m isse d from ad m issio n to graduation

DURAT

Total d u ratio n of M a s te r 1 s work in te rm s

AVG- LD

A verage c o u rse load in M a s te r 1 s work

57

because it n o rm ally lead s to d ra m a tic reduction in the space
dim ensionality of the independent v a ria b le w ithout su b stan tia l lo ss
of inform ation when the dependent v a ria b le is a categ ory.

The

r e s u lt of th is a n a ly sis is a s e t of d isc rim in a n t functions which
tra n sfo rm a s e t of m e a s u re s fo r an individual into a single s c o re
which lo ca te s th at in d iv id u a l's p osition in one of the five g roup s.
The m axim um num ber of functions th a t can be com puted i s one le s s
than the num ber of groups.

The significance of the d isc rim in a n t

functions fo r H ypothesis 1 a r e p resen ted in Table 4 .2 .

TABLE 4. 2
SIGNIFICANCE OF THE DISCRIMINANT FUNCTIONS
FOR A LL GROUPS

Function

NDF

x2

P e rc e n t of
V ariation

I

56

177.40

54.47

II

39

83. 93

32. 33

X2 > 23. 6 8 , P < .05

As Table 4 . 2 in d ic ates, two of the p o ssib le four functions
w ere significant a t le s s than the . 05 level, thus supporting
Hypothesis 1.

The f i r s t function is the m o st pow erful, accounting

for 54. 47 p e rc e n t of the total v a ria n c e .

The second function accounts

58

for 32. 33 p e rc e n t of the v a ria n c e .
o r independent.

The two functions a r e orthogonal,

Thus, the com plete evaluation of an in d iv id u a l's

p rea d m issio n data is reduced to the co n sid eratio n of two independent
fa c to rs .

The d isc rim in a n t function coefficients fo r the o rig in al

v a ria b le s a r e p resen ted in Table 4. 3.

TABLE 4. 3
DISCRIMINANT FUNCTION COEFFICIENTS
FOR PREADMISSION DATA FOR ALL GROUPS
Raw C oefficient

Standardized

V ariable
I

II

I

II

M -S

. 431

1. 389

. 208

. 671

SEX

. 302

- . 624

. 148

-. 307

RES

687

. 155

-. 259

. 058

YRS-RS

015

- .001

-. 158

-. 016

AGE

. 033

. 079

. 270

. 641

BORN

. 346

. 158

. 174

. 079

GRAD

. 012

. 008

. 010

. 007

NO- COL

. 130

. 343

. 116

. 306

BA-MA

015

- .063

-. 094

-. 385

BA-BA

. 002

- . 064

. 014

302

U -G PA

. 423

-1.472

. 178

-. 619

G -E X P

. 038

. 057

. 018

. 028

ADM

948

. 499

-. 711

. 374

MAJOR

261

- . 117

-. 432

-. 195

59

The sta n d ard coefficients have been c o rre c te d for
d iffe re n c e s in the stan d ard deviations of the o rig in a l v a ria b le s .

As

Table 4. 3 in d ic a te s, function I is loaded p r im a rily on m a jo r, age,
resid e n ce , and m a r ita l sta tu s .

The loadings fo r function I range

from .010 fo r college of g rad u atio n to -.711 fo r ad m issio n sta tu s.
Function II is com posed p rim a rily of m a r ita l sta tu s, age, u n d e r­
g raduate g rad e point a v e ra g e , y e a r s from end of B .A . to s t a r t of
M. A. , ad m issio n sta tu s , n um ber of co lleg es attended fo r B .A . ,
sex, and num ber of y e a rs from s t a r t to end of B .A .

The loadings

for function II range from -. 016 fo r y e a r s a t re sid e n c e to .671 for
m a rita l sta tu s.

H ypothesis 2 . --H y p o th e sis 2 s ta te s th a t the p re a d m issio n
data is significantly d iffe re n t fo r the g rad u ated and the not enro lled
for 5 y e a rs groups.

A m u ltiple d is c rim in a n t a n a ly sis w as p erfo rm ed

to te s t th is hypo th esis, using the p re a d m issio n data fo r the g rad uated
and the academ ic action g ro u p s.

Since only two g ro u p s w ere

involved, only one d isc rim in a n t function w as calcu lated .

The s i g ­

nificance of the d isc rim in a n t function fo r H ypothesis 2 is p re se n te d
in Table 4. 4
As Table 4 .4 in d ic a te s, the d isc rim in a n t function is s i g ­
nificant a t le s s than the . 05 level, thus supporting H ypothesis 2.

60

T A B L E 4. 4
SIGNIFICANCE OF THE DISCRIMINANT FUNCTION
FOR GRADUATE AND 5 YEAR NOT ENROLLED GROUPS

Function

NDF

X2

P e rc e n t of
V ariation

I

14

6 6 . 46

26. 79

X2 > 2 3 .6 8 . P < . 05

The function acco u n ts fo r 26. 79 p e rc e n t of the v a ria n c e in the in d e ­
pendent v a ria b le s .

The d is c rim in a n t function coefficients for the

o rig in al v a ria b le s a r e p re s e n te d in Table 4. 5
The function i s loaded high on age, u n d erg rad u ate g rad e
point a v e ra g e , a d m issio n s ta tu s , and m a jo r a t the tim e of adm issio n.
The stan d ard ized co efficien ts ran g e from -. 023 fo r m a r ita l sta tu s to
-.752 for a d m issio n s ta tu s .

H ypothesis 3. - -H y p o th esis 3 s ta te s th at the p rea d m issio n
data is significantly d iffe re n t fo r the graduated and the academ ic
action g ro u p s.

The g rad u ated and the acad em ic action groups w ere

selected out of the to ta l sam p le and a m ultiple d isc rim in a n t function
was calculated to t e s t the above h y p o th esis.

Again, since only two

groups w ere involved, one d is c rim in a n t function w as calculated.
The significance of the function i s p re se n te d in Table 4. 6 .

61

TABLE 4. 5
D IS C R IM IN A N T F U N C T IO N C O E F F I C I E N T S
F O R P R E A D M IS S IO N D A T A F O R G R A D U A T E A N D
5 Y EA R NOT E N R O L LE D GRO U PS

V ariable

Raw
C oefficient

M -S

- .048

023

SEX

. 311

. 153

RES

- .635

236

YRS- RS

- .011

-. 113

AGE

. 053

. 413

BORN

. 086

. 043

GRAD

. 232

. 190

NO-COL

- . 033

-. 028

BA-M A

- .037

-.230

BA -BA

- .013

063

U -G PA

1 . 026

. 445

G -E X P

- . 105

-. 050

ADM

- . 967

752

MAJOR

- .251

-.431

Standa rdi z ed

62

TA B LE 4. 6
SIGNIFICANCE OF THE DISCRIMINANT FUNCTION
FOR GRADUATE AND ACADEMIC ACTION GROUPS

Function

NDF

X2

P e rc e n t of
V ariation

I

14

23. 92

10.48

X2 > 2 3.68, P < .05

The function is significant a t l e s s than the . 05 level.
Table 4. 6 also in d ic ates th a t the v a ria tio n accounted fo r by the
function is 10.48 p erc en t.

The d isc rim in a n t coefficients for the

original v a ria b le s a r e p resen ted in Table 4 .7 .
The v a ria b le s with the hig h est coefficients for th is function
a re age, y e a rs from end of B .A . to s t a r t of M. A. , m a r ita l sta tu s,
sex, graduation code, y e a r s from s t a r t to finish of B .A . , ad m issio n
statu s, and m a jo r a t ad m issio n .

The coefficients range from -.019

for num ber of colleges attended for B .A . to -1. 188 fo r age.
sign of the coefficient is a function of sc a le d irec tio n .

The

F o r in stan ce,

the la rg e negative coefficient fo r age in d ic ates th at o ld er students
a re m ore likely to have acad em ic actio n taken a g a in st them .

H ypothesis 4 . --H y p o th esis 4 s ta te s th at the p rea d m issio n
data for the graduated group is significantly d iffe ren t from the

63

TA B L E 4. 7
D IS C R IM IN A N T F U N C T IO N C O E F F I C I E N T S
F O R P R E A D M IS S IO N D A T A F O R G R A D U A T E A N D
A C A D E M IC A C T IO N G R O U P S

V ariab le

Raw
C oefficient

M -S

-1. 278

- . 628

SEX

. 948

. 473

RES

- .297

- . 124

.400

. 052

- . 157

-1.188

BORN

. 521

. 263

GRAD

- . 505

- .415

NO -COL

- . 021

- .019

BA-M A

. 128

. 756

BA -B A

. 095

.412

U -G PA

.477

. 196

G -E X P

. 128

. 061

- . 521

- .414

. 234

.406

YRS-RS
AGE

ADM
MAJOR

Standardized

64

p read m issio n data fo r the d ism isse d g ro u p .

A m ultiple d isc rim in a n t

function w as calculated for the graduated and the d ism isse d groups
to te s t this hypothesis.

The resu ltin g function w as sig nifican t a t

le s s than the . 05 level and accounted fo r 38. 98 p e rc e n t of the to tal
v arian ce in the independent v a ria b le s.

T hese re s u lts a r e p resen ted

in Table 4 .8 .

TABLE 4. 8
SIGNIFICANCE OF THE DISCRIMINANT FUNCTION
FOR GRADUATE AND DISMISSED GROUPS

Function

NDF

x2

P e rc e n t of
V ariation

I

14

6 8 . 79

38 . 98

X2 > 23. 6 8 , P < . 05

The coefficients of the d isc rim in a n t function, which a r e
rep o rted in Table 4. 9, range from . 030 fo r the n um ber of y e a rs
from the s t a r t of the B. A. to the end of the B. A. to . 506 fo r age.
M arital sta tu s, age, born in s ta te o r out, ad m issio n sta tu s , and
m ajo r a ll had standardized coefficients h ig h er than .359.

65

T A B L E 4. 9
D IS C R IM IN A N T F U N C T IO N C O E F F I C I E N T S
F O R P R E A D M IS S IO N D A T A F O R G R A D U A T E A N D
D IS M IS S E D G R O U P S

V ariab le

Raw
C oefficient

Standardized

M -S

1. 039

. 504

SEX

- . 251

126

RES

- .487

-. 198

YRS-RS

- . 008

-. 090

AGE

. 070

. 506

BORN

. 707

. 359

GRAD

- . 122

-. 101

. 299

. 288

BA-M A

- .049

-. 272

BA -BA

. 030

-. 126

U -G P A

- .498

-. 201

G -E X P

. 076

. 036

ADM

- .517

-.401

MAJOR

- . 282

462

N O -CO L

66

Sum m ary of Group
M em bership Hypotheses
The p o ssib ility of p redicting group m em b ersh ip using
p rea d m issio n data was te ste d with H ypotheses 1 through 4.

H ypothe­

sis 1 could be considered the m a jo r hypothesis of the s e t a s it
tested the d iffe ren c es in p re a d m issio n data fo r s e v e ra l g ro u p s.

The

m ultiple d isc rim in a n t a n a ly sis for H ypothesis 1 yielded two orthogonal
o r independent functions.

Thus, the p re a d m issio n data tends to p r e ­

dict group m em b ersh ip by two independent m e a s u re s .
H ypotheses 2, 3, and 4 teste d the d iffe re n c e s in p r e a d m is ­
sion data betw een the graduated group and one of the o th er groups in
the sam ple.

Since H ypothesis 1 d em o n strated th a t the p re a d m issio n

data d is c rim in a te s by two independent m e a s u re s , and since the
m axim um num ber of d isc rim in a n t function fo r two groups i s one,
Hypotheses 2, 3, and 4 w ere r e s tr ic te d in the am ount of v a ria n c e
that could be accounted for in the p rea d m issio n data.
Function 1 of H ypothesis 1 was the m o st pow erful, a c c o u n t­
ing fo r 54.47 p erc en t of the v a ria n c e .

The o th e r functions ranged

from a low of 10.48 p e rc e n t fo r H ypothesis 3 to 38. 98 p e rc e n t for
Hypothesis 4.

67

Hypotheses Concerning M easu res
of P e rfo rm a n c e fo r G radu ates
H ypothesis 5. - -H y pothesis 5 s ta te s th at p rea d m issio n data
for the graduated groups is significantly re la te d to the duration of
the M a s t e r 's p ro g ra m .

The graduated group and the graduated group

who continued to the d o cto ra l p ro g ram w ere selected from the to ta l
sam ple to te s t th is hypothesis.

A ll of the p re a d m issio n v a ria b le s

w ere used a s independent v a ria b le s .

The duration of the M aster* s

p ro g ram , which w as m e a su re d a s the num ber of te rm s from the f i r s t
te rm of g rad u ate en ro llm en t to the te rm of g raduation, w as used a s
the dependent v a ria b le .

M ultiple re g re s s io n w as the technique chosen

to te s t this hypothesis because i t pro vides an a n a ly sis of the re la tio n s
among a single c rite rio n m e a su re and s e v e ra l p re d ic to r m e a s u re s .
In addition, the a n a ly sis re s u lts in an equation for predicting the
unknown c rite rio n sc o re of a new sub ject from this known s e t of
p red ic to r s c o re s .

The re g re s s io n and beta coefficients for H ypothe­

sis 5 a r e rep o rted in Table 4. 10.
The m axim um R

2

The m axim um R attained was . 499.

w as .2497, which in d ic ates th at 24. 97 p erc en t of

the v arian ce in the independent v a ria b le s w as accounted fo r by the
re g re ss io n equation.

The value of F w as 4. 374, which i s sign ificant

beyond the .05 level, thus supporting H ypothesis 5.

Exam ination of

the beta w eights shows th at age, re sid e n c e , and the num ber of y e a rs
from the end of the B. A. to the M. A. application w ere the m o st

68

T A B L E 4 . 10
REGRESSION COEFFICIENTS, BETA WEIGHTS, STANDARD ERRORS,
AND STEPWISE R FOR PREADMISSION DATA FOR
GRADUATES WITH DURATION AS DEPENDENT VARIABLE

V ariable

R eg ressio n
C oefficients

Standard
E rro r

Beta
C oefficients

Standard
E rro r
of Beta

Stepwise
R

M -S

- . 229

1 . 022

016

. 079

. 499

SEX

. 623

. 964

. 046

. 071

. 498

RES

-5.188

1. 359

314

. 082

. 407

YRS-RS

- . 015

. 049

023

. 075

. 499

. 362

. 182

. 389

. 195

. 482

BORN

-1.171

1. 067

-. 088

. 080

. 487

GRAD

. 091

. 657

. 011

. 081

. 499

NO- COL

- . 139

. 566

018

. 072

. 499

BA-MA

- .268

. 184

-. 236

. 162

. 490

BA-BA

- . 191

. 207

119

. 129

. 495

U-G PA

- .822

1. 372

051

. 085

. 497

G -E X P

-1. 897

1. 014

133

. 071

. 445

ADM

- . 087

. 747

010

. 090

. 499

MAJOR

- .471

. 274

123

. 071

. 463

AGE

The m ultiple R = .499, R2 = .2497, F * 4 .3 7 , P < . 05
Standard e r r o r of the estim a te is 6 . 077; the constant te rm is 18. 457.

69

significant fa c to rs in the final equation.

The p redictio n equation,

form ed from the re g re s s io n coefficients, has the following form :

P re d ic te d D uration = - .229 M -S + . 623 SEX - 5. 188 RES
- . 015 YRS-RS + . 362 AGE - 1. 171 BORN
+ . 091 GRAD - .139 NO-COL - .268 BA-M A
- . 191 BA-BA

- . 822 U -GPA

- 1. 1897 G - EXP

- .087 ADM - .471 MAJOR + 18.457

H ypothesis 6 . --H y p o th esis 6 s ta te s that the p re a d m issio n
data fo r

the graduated groups i s significantly re la te d to an index of

p e rs is te n c e

fo r the M aster* s p ro g ra m .

The index of p e rs is te n c e

was m ade up of the to ta l te rm s attended from the f i r s t te rm of
attendance to the te rm of g rad uatio n, the to ta l te rm s m isse d during
this sam e in te rv a l, the du ration, and the a v e rag e co u rse load.

A

canonical c o rre la tio n was calculated to re la te the p re a d m issio n data
with the v a ria b le s which m ade up the index of p e rs is te n c e .

Canonical

c o rre la tio n is a sp e cia l case of m u ltip le re g re s s io n w here the num ber
of c r ite r ia is g r e a te r than one.

Thus, canonical c o rre la tio n i s the

m axim um c o rre la tio n between lin e a r functions of the two s e ts of
v a ria b le s.

The num ber of p o ssib le canonical c o rre la tio n s is equal

to the num ber of v a ria b le s in the s m a lle r s e t su b ject to the r e s t r i c ­
tion that each canonical v a ria te be orthogonal to a ll o th er canonical
v a ria te s on its side of the equation.

The canonical c o rre la tio n m odel

TABLE 4. 11
CORRELATION MATRIX O F PREADMISSION DATA FOR GRADUATES
J
0
U
I
o

w

£

m

»

X

3

w

W

CM OT

CD

.150

.224

i
w

w
o

2
«
0
to PQ

00

£

c

b
O

CD

Si

o $
PQ

^

'

- P

CM

a,
x
w
i
o

tf

0
•-5

e*5

0

1 M-S
2 SEX

. 168

3 RES

.171

4 YRS-RS

.310

5 AGE

.330

6 BORN

.435

7 GRAD

.466

.477

8 NO-COL
9 BA-MA

.279

10 BA-BA
11 U-GPA

.529

.402
.164

. 165

13 ADM

. 166

R > .138, P < .05

-. 139

.238

12 G-EXP
14 MAJOR

.749

-. 178

.199

-. 302

-.291
.258 -.256

.218

568
241

.308

71

p roduces th re e useful s e ts of c o rre la tio n s in addition to the
canonical c o rre la tio n .

T hese a r e the c o rre la tio n s betw een the

independent v a ria b le s, the c o rre la tio n s between the dependent
v a ria b le s, and the c o rre la tio n s betw een the independent and the
dependent v a ria b le s .

The c o rre la tio n s between the independent

v a ria b le s a re p re se n te d in T able 4. 11.

The c o rre la tio n s which had

a value of R g r e a te r than . 138 w ere sig n ifican t a t the . 05 level.
The highest c o rre la tio n s w ere between age and the n um ber of y e a rs
from end of B. A. to application to M a s t e r 's p ro g ram (. 749), age
and the num ber of y e a rs from s t a r t of B. A. to the end of B. A.
(. 529), and u n d erg rad u ate g rad e point a v e rag e and ad m issio n sta tu s
(. 568).
The c o rre la tio n s between the m e a s u re s of p e rs is te n c e a r e
p resen ted in Table 4 . 12.

TABLE 4. 12
CORRELATION MATRIX OF MEASURES OF PERSISTENCE
FOR GRADUATES
1
T -A T T

2
T-M ISS

1

T -A T T

2

T - MISS

. 317

3

DURAT

. 705

. 882

4

AVG- LD

-. 759

353

R > .138, P < .05

3
DURAT

-. 613

4
AVG-LD

TABLE 4.13
CORRELATION MATRIX OF PREADMISSION DATA AND
MEASURES OF PERSISTENCE FOR GRADUATES

J
0
u
i

CO
tf

co
I

co
eo

W

2

I

w

OS

0

PS

I*

to PQ

0

o
00 2

<
n

0)

I.j
o <
pq

C
Oh
0
rH '
- P

&
X
w
N '
1-H 0

os
3
eo 0

f-H <J

o

2 S
-3
to

193

-.396

.218 -.259 -.175

2 T-MISS

-. 309

. 160 -. 144

. 150

-.217

-.205

3 DURAT

-.407

.218 -.232 -. 169

.147

-.245

-. 191

.227

. 153

1 T-ATT

4 AVG-LD

-. 160

. 197

R > . 138, P < .05

.515

-.189

.266

.216

73

This table in d ic ates th a t d uration and a v e rag e co u rse load a r e quite
highly c o rre la te d with the o th er m e a s u re s of p e rs is te n c e .
The c o rre la tio n s between the independent and the dependent
v a ria b le s a r e p re se n te d in Table 4. 13.

T his table in d ic ates that

re sid e n c e , age, born in s ta te o r out, and g rad u ate ex p erien ce a ll
c o rre la te significantly with a ll of the m e a s u re s of p e rs is te n c e .
The significance of the canonical c o rre la tio n fo r H ypothe­
s is 6 is p resen ted in Table 4. 14.

The h ig h est canonical R w as .613,

which is sig n ifican t a t the . 05 level.

H ypothesis 6 w as th e re fo re

supported.

TABLE 4. 14
X2 TEST OF SIGNIFICANCE OF CANONICAL CORRELATIONS
OF MEASURES OF PERSISTENCE FOR GRADUATES
Num ber
of Roots
Removed

Canoni cal
R

x2

NDF

0

. 613

116.272

56

=

. 05

1

. 26 4

27.411

39

>

. 05

2

.218

13.692

24

>

. 05

3

. 153

4. 487

11

>

. 05

P

The stan d ard ized canonical coefficients, which co rresp o n d
to the beta w eights in a m ultiple r e g re s s io n equation, a r e p resen ted

74

in Table 4. 15.

The m e a s u re s of p e rs is te n c e w ere loaded high est

on te rm s m isse d and av erag e co u rse load.

The loadings ranged

from . 168 fo r te rm s attended to . 845 for av erag e co u rse load.

The

v a ria b le s with the h ig h est coefficients for the p rea d m issio n data w ere
resid e n ce , age, and num ber of y e a rs from the end of the B. A. to the
M .A . application.

TABLE 4. 15
CANONICAL COEFFICIENTS FOR MEASURES OF
PERSISTENCE AND PREADMISSION DATA

V ariable

Standardized
C oefficient

V ariable

Standardized
C oefficient

M -S

-0.130

U -G PA

-0.115

SEX

0. 167

G - EXP

-0. 218

RES

0. 716

ADM

YRS-RS

0. 090

MAJOR

-0. 194

AGE

0. 620

0. 061

BORN

-0.116

GRAD

0. 023

T -A T T

0 . 168

NO-COL

0. 135

T - MISS

0. 605

BA-MA

-0.428

DURAT

-0.383

BA-BA

-0.274

AVG- LD

-0.845

75

H ypothesis 7. --H y p o th e sis 7 s ta te s th at the p rea d m issio n
data fo r the graduated groups is significantly re la te d to an index of
academ ic p erfo rm an c e.

The index of acad em ic p erfo rm an ce was

m ade up of the total g rad e point a v e ra g e fo r the M aster* s p ro g ra m ,
and the to ta l c re d its d e fe rre d .

A canonical c o rre la tio n was again

calculated to re la te die p re a d m issio n data with the m e a s u re s of
academ ic p erfo rm an c e.

The c o rre la tio n s of the independent v a r i ­

ables a r e the sam e a s in H ypothesis 6 and a r e p re se n te d in Table 4. 11.
The c o rre la tio n s of the m e a s u re s of academ ic p erfo rm an ce a r e p r e ­
sented in Table 4. 16.

This table in d icates th at the m e a s u re s of

academ ic p erfo rm an ce a r e not c o rre la te d significantly.

TABLE 4. 16
CORRELATION MATRIX OF MEASURES OF
ACADEMIC PERFORMANCE FOR GRADUATES

T -G P A
1 T -G P A
2 T -C R E D
3 C R -D E F

R > .138, P < . 05

-. 145

T - CRED

C R -D E F

TABLE 4.17
CORRELATION MATRIX OF PREADMISSION DATA AND
MEASURES OF ACADEMIC PERFORMANCE FOR GRADUATES

CO

® pq
.351

1 T-GPA
2 T-CRED

. 191

.145

3 CR-DEF

R > . 138, P < . 05

.281

. 174

. 162

77

The c o rre la tio n s betw een the independent and dependent
v a ria b le s a r e p resen ted in Table 4. 17.

As indicated in th is table,

the independent and dependent v a ria b le s a r e m ostly not significantly
c o rre la te d .
The significance of the canonical c o rre la tio n fo r H ypothe­
s is 7 is p resen ted in Table 4. 18.
attained was .426.

The m axim um canonical R

Two of the canonical ro o ts w ere significant

beyond the .05 level, thus supporting H ypothesis 7.
TABLE 4. 18
X2 TEST OF SIGNIFICANCE OF CANONICAL CORRELATIONS OF
MEASURES OF ACADEMIC PERFORMANCE FOR GRADUATES
Num ber
of Roots
Removed

Canonical R

X2

NDF

P

1

. 426

86.902

42

= . 05

2

. 409

48. 961

26

= . 05

3

. 270

14.313

12

> . 05

The stan d ard ized canonical coefficienta a r e p re se n te d in
Table 4. 19.

The to ta l g rad e point a v e ra g e and c re d its d e fe rre d had

the highest coefficients on the dependent v a ria b le side of the e q u a ­
tion.

The coefficients for the p re a d m issio n data ranged from . 056

78

for graduation code to -1. 036 for age.

The independent v a ria b le s

with the hig h est coefficients w ere age, the num ber of y e a rs from the
end of B. A. p ro g ram to M a s t e r 's application, the n u m ber of y e a rs
from the s t a r t of the B. A. to the end of the B. A. , und erg rad u ate
grad e point av e rag e , and ad m issio n sta tu s.

TABLE 4. 19
CANONICAL COEFFICIENTS FOR MEASURES OF
ACADEMIC PERFORMANCE AND PREADMISSION DATA
Standardized
C oefficient

V ariable

Standardized
C oefficient

V ariab le

M -S

-0.361

B A -B A

0.435

SEX

0. 132

U -G P A

0. 521

RES

0. 074

G -E X P

-0.065

YRS-RS

0. 130

ADM

AGE

-1. 036

BORN

-0.110

GRAD

0. 056

NO- COL
BA-MA

0. 489

MAJOR

-0. 200

T -G P A

0. 897

-0.104

T -C R E D

-0.133

0. 726

C R -D E F

-0.309

79

Summary of Graduate
P e rfo rm a n c e Hypotheses
The relatio n sh ip of the p rea d m issio n data for the graduated
groups to th re e m e a s u r e s of p erfo rm an ce was explored with
Hypotheses 5, 6, and 7.
te s t Hypothesis 5.

The R

A r e g r e s s io n equation was calculated to
2

for this equation was .2497, which m eans

the equation accounted for 24. 97 p e rc e n t of the v a ria n c e in the in d e ­
pendent v a ria b le s with relation to duration of the M a s t e r 's p ro g ram .
Canonical c o rre la tio n was used to t e s t the relation sh ip
between the p re a d m issio n data and m e a s u r e s of p e r s is te n c e and
academ ic p erfo rm an c e for Hypotheses 6 and 7 resp ec tiv e ly .
canonical R for Hypothesis 6 was . 613.

The

The canonical R for Hypothe­

sis 7 was .426.

Hypotheses Concerning
D epartm ental C h a ra c te r is tic s
Hypothesis 8 . --H y p o th e sis 8 s ta te s th at a fa c to r an a ly sis of
p read m issio n data for the graduated group will yield an in te rp re ta b le
se t of fa c to rs which w ill be d ifferen t fo r each departm ent.
hypothesis was tested in two p a r ts .

This

F i r s t, a facto r an a ly sis with a

rotation c rite rio n of a ll fa c to rs with eigenvalues g r e a t e r than 1. 00
was computed using the p rea d m issio n data for the e n tire graduated
sam ple.

This was done to t e s t the sta te m e n t that the facto r an aly sis

80

of the p rea d m issio n data will yield an in te rp re ta b le s e t of fac to rs.
Next, the graduated group was divided into groups according to th e ir
m ajo r at the tim e of adm ission.

A facto r an a ly sis using a rotation

crite rio n of a ll fa c to rs with eigenvalues g r e a t e r than 1. 00 was c o m ­
puted for each m a jo r group.

This was done to t e s t the sta te m e n t that

the fa c to rs will be d ifferen t for each departm ent.
The purpose of the fac to r a n a ly se s was to group tog eth er
under one category a s many v a ria b le s a s w e re m ea su rin g the sam e
factor and a ssig n a common d e s c rip tiv e nam e to the group.

The

principal ax is, orthoginal o r v arim ax , and the oblique o r prom ax
solutions w e re calculated.

The orthoginal solution is commonly used

because it m ain tain s independence between fa c to rs .

However, the

oblique solution was selected for this a n a ly sis because i t re la x e s the
req u irem e n t of independence and allows the v a r ia b le s to group into
m ore identifiable c a te g o rie s.

However, the fa c to rs can be somewhat

in te rc o rre la te d in the oblique solution.
The r e s u lts of the prom ax rotation fo r a ll g ra d u a te s a r e
presen ted in Table 4. 20.

The highest fac to r loadings fo r the p r e ­

adm ission data for the 5 fa c to rs with eigenvalues g r e a t e r than 1. 00
a r e rep o rted .
was 69.251.

The p erc en t of v aria n ce accounted fo r by the 5 fa c to rs
The com m unalities ranged from .423 for g raduate

experience to .916 for age.

TABLE 4 . 2 0
HIGHEST FACTOR LOADINGS FOR PREADMISSION DATA FOR GRADUATES
FROM A L L D EPA RTM EN TS USING THE PROMAX ROTATIONAL ANALYSIS

1

2

3

4

5

Eigenvalues

2.343

2.140

1.828

1.437

1.252

Cumulative
Proportion
of Variance

18.028

34.490

48.557

59.615

69.251

Communalities

Variable
M-S
SEX
RES
YRS-RS
AGE
BORN
GRAD
NO-COL
BA-MA
BA-BA
U-GPA
G-EXP
ADM

.640
-.762
.887
.863
-.810
-.773
-.777
.906
-. 827
.862
-.569
.796

.581
.576
.699
.763
. 916
.683
.662
.621
.822
.753
.746
.423
.751

82

The f i r s t facto r for the total graduated sam ple could be
c h a ra c te riz e d a s a post B. A. factor.

It was composed of age, BA-

MA, and previous graduate experience.

The second fac to r, c o m ­

posed of resid e n ce , born in o r out of state, and graduation code,
could be considered a resid en ce facto r.
undergraduate academ ic facto r.

The third fac to r was the

It was composed of undergraduate

grade point av e rag e , and ad m issio n statu s.

These two v a ria b le s

load to g eth er because ad m issio n sta tu s i s d eterm in ed by the strength
of the undergrad uate grade point av e rag e .

The fourth facto r was

composed of m a r i ta l sta tu s and y e a r s a t resid e n ce .
considered a s a dom esticity fac to r.

This could be

Number of colleges attended for

the B. A. and the num ber of y e a r s between s t a r t and finish of the
B.A. formed the la s t factor which could be labeled the u n d e rg ra d u ­
ate p e rs is te n c e facto r.
The r e s u lts of the prom ax rotation for the graduated groups
adm itted nondegree a r e p resen ted in Table 4 .2 1 .
cent of variance attained was 64. 38 p ercen t.
eigenvalues g r e a t e r than 1. 00.

The highest p e r ­

Four fa c to rs had

The com m unalities ranged from . 35

for adm ission statu s to . 846 for resid e n ce .

F a c to rs 1 and 2 w ere

the sam e a s fa c to rs 1 and 2 for the total group.
a c te riz e d by post B.A . and re s id e n c e fa c to rs.

They w ere c h a r ­
The th ird facto r

was composed of m a r ita l sta tu s, y e a r s of resid en ce, graduated in

TABLE 4. 21
HIGHEST FACTOR LOADINGS FOR PREADMISSION DATA FOR GRADUATES
ADM ITTED NONDEGREE USING THE PROMAX ROTATIONAL ANALYSIS

1

2

3

4

Eigenvalues

3.181

1.810

1.511

1.222

Cumulative
Proportion
of Variance

26.510

41.595

54.191

64.382

Communalities

Variables
M-S
SEX
RES
YRS-RS
AGE
BORN
GRAD
NO-COL
BA-MA
BA-BA
U-GPA
G-EXP
ADM

.598
.664
-.927
-.742
-. 745
-. 702
. 601
-. 634
-.767
-.691
.588
.882

.645
.584
.846
.600
.789
.749
.656
.521
.709
.614
.377
.629
.351

84

state or out, and num ber of colleges attended for the B.A .
v a ria b le s seem to m e a s u re m obility.

These

The fourth facto r se em s to

be a m e a s u re of the underg rad u ate ex p erien ce.

It i s composed of

undergraduate g rad e point av erag e, BA - BA, and sex.
The r e s u lts of the prom ax ro tatio n for the e le m e n ta ry e d u ­
cation group a r e p resen ted in Table 4 .2 2 .

Five fa c to rs had e ig e n ­

values g r e a t e r than 1. 00 and accounted for 71. 01 p e rc e n t of the total
varian ce.
The five fa c to rs for the e le m e n ta ry education group a r e
alm o st identical to the five fa c to rs fo r the total group.

F a c to r 1,

composed of age, BA-M A, and BA-BA could be labeled a s a p e r ­
sisten ce facto r.

F a c to r 2 was the re s id e n c e o r m obility factor.

F acto r 3 was the u n derg rad u ate p erfo rm an c e factor.

F a c to r 4, the

dom esticity facto r, was composed of m a r ita l sta tu s and y e a r s a t
residence.

F a c to r 5 had only one v a ria b le and rem ain ed the g r a d u ­

ate experience factor.
The r e s u lts of the prom ax ro tatio n al an a ly sis for the
secondary education group a r e p resen ted in Table 4 .2 3 .

The m a x i ­

mum varian ce attained was 69. 96 p e rc e n t a f te r four fa c to rs w ere
rotated.

C om m unalities ranged from . 473 fo r m a r i t a l sta tu s to . 937

for age.
M arital sta tu s, sex, resid e n ce , and y e a r s a t resid e n ce
combined to form the f i r s t facto r.

Num ber of colleges attended for

TABLE 4. 22
HIGHEST FACTOR LOADINGS FOR PREADMISSION DATA FOR GRADUATES
ADMITTED IN ELEM ENTARY EDUCATION USING THE PROMAX ROTATIONAL ANALYSIS

1

2

3

4

5

Eigenvalues

2.807

1.922

1.880

1.569

1.052

Cumulative
Proportion
of Variance

21.596

36.386

50.848

62.918

71.012

Communalities

Variable
M-S
SEX
RES
YRS-RS
AGE
BORN
GRAD
NO-COL
BA-MA
BA-BA
U-GPA
G-EXP
ADM

-.626
-. 674
-.926
. 959
-.756
-.828
.503
.662
.863
-.900
-.824
-. 898

.660
. 338
.607
.804
.952
.591
.665
.568
.833
.858
.790
.727
.832

TABLE 4. 23
HIGHEST FACTOR LOADINGS FOR PREADMISSION DATA FOR GRADUATES
ADM ITTED IN SECONDARY EDUCATION USING THE PROMAX ROTATIONAL ANALYSIS

1

2

3

4

Eigenvalues

2.994

2.658

1.950

1.491

Cumulative
Proportion
of Variance

23.035

43.489

58.493

69.964

Communalities

Variables
-.450
-.684
-.527
-.924

.489
-.880
.779
.901
-.736
-.969
•

CO
00
oo

1

M-S
SEX
RES
YRS-RS
AGE
BORN
GRAD
NO-COL
BA-MA
BA-BA
U-GPA
G-EXP
ADM

-.606
.618
.677

.473
.516
.785
.747
.937
.692
.816
.545
.890
.811
.501
.740
.637

87

B.A. , B A-BA , and u n d erg rad u ate g rad e point a v e rag e formed
factor 2, which was the underg rad u ate p erfo rm an c e factor.
F a c to r 3 was the post B .A . facto r and was identical to the post B.A .
factor for the total group.

F a c to r 4 is a re s id e n c e factor, being

composed of resid e n ce , born in o r out of sta te , graduated in o r out
of state, and ad m issio n sta tu s.
The r e s u lts of the prom ax an aly sis for the ad m in istra tio n
group a r e p re se n te d in Table 4 .2 4 .

The cumulative p ro p o rtio n of

varian ce attained a f te r five fa c to rs w ere rotated was 80. 375 p ercen t.
The com m unalities ranged from . 574 for grad u ate exp erien ce to . 915
for BA-MA.
F acto r 5 was loaded on age and thus re m a in s an age facto r.
F acto r 1 was loaded on age and num ber of colleges attended for B.A.
F acto r 2, the underg rad u ate factor, was loaded on sex, u n d e r g r a d u ­
ate grade point av erag e, and ad m issio n sta tu s.

F a c to r 3 w as loaded

on residen ce, born in o r out of state, graduated in o r out of sta te ,
and g raduate experience.

I t could be labeled a re s id e n c e factor.

F acto r 4, the d om esticity factor, was loaded on m a r ita l sta tu s, sex,
and y e a r s a t resid en ce.
The r e s u lts of the prom ax an aly sis for the counseling and
guidance group a r e p resen ted in Table 4. 25.
values g r e a t e r than 1. 00.

Six fa c to rs had e ig e n ­

The cumulative p erc en t of v aria n ce was

TABLE 4 . 2 4
HIGHEST FACTOR LOADINGS FOR PREADMISSION DATA FOR GRADUATES
ADM ITTED IN ADMINISTRATION USING THE PROMAX ROTATIONAL ANALYSIS

1

2

3

4

5

Eigenvalues

2.950

2.709

2.142

1.510

1.136

Cumulative
Proportion
of Variance

22.696

43.537

60.015

71.635

80.375

Communalities

Variable
M-S
SEX
RES
YRS-RS
AGE
BORN
GRAD
NO-COL
BA-MA
BA-BA
U-GPA
G-EXP
ADM

.933
.560

-.633
-.823

.717
-.481

-.673
-.830
-.520

-.917

-.817
-.511
-.837

.852
.809
.779
.708
.883
.801
.816
.753
.915
.888
.859
.574
.787

TABLE 4. 25
HIGHEST FACTOR LOADINGS FOR PREADMISSION DATA FOR GRADUATES
ADM ITTED IN COUNSELING AND GUIDANCE USING THE PROMAX ROTATIONAL ANALYSIS

1

2

3

4

5

6

Eigenvalues

2.337

2.215

1.973

1.591

1.286

1.035

Cumulative
Proportion
of Variance

17.981

35.022

50.205

62.451

72.345

80.311

Communalities

Variable
M-S
SEX
RES
YRS-RS
AGE
BORN
GRAD
NO-COL
BA-MA
BA-BA
U-GPA
G-EXP
ADM

.985
.820
-.911
.887
.854
.804
-.887
.945
-.818
.985
-. 968

.722
.833
.649
.891
.949
.692
.758
.821
.910
.813
.745
.846
.804

90

80.311 percent.

The com m unalities ranged from . 692 for born in

state o r out to . 949 for age.
F a c to rs 5 and 6 loaded on sex and grad u ate experience
resp ectiv ely and, th e re fo re , retained th e ir original v a ria b le nam es.
F acto r 1 loaded on resid e n ce , born in sta te o r out, and graduated in
state o r out.

Thus, factor 1 could be labeled a re sid e n c e fa c to r.

F acto r 2 loaded on age and BA-BA.

F a c to r 3 loaded on u n d ergrad u ­

ate grade point a v e rag e and could be considered an u ndergraduate
perform ance facto r.

F a c to r 4 could be considered to be a m e a su r e

of mobility as it loaded on y e a r s a t resid e n ce and num ber o f c o lle g e s
attended for B .A .
In su m m ary , the facto r an a ly sis of the total graduated group
yielded an in te rp re ta b le s e t of fac to rs, thus supporting p a r t A o f the
hypothesis; however, the factor a n a ly se s of the m a jo r group y ield ed
a s e t of fa c to rs which w e re a lm o st id en tical to the total group and to
each o ther.

T h erefo re , p a r t B of the hypothesis was not supported.

Summary
The p o ssib ilitie s of predicting group m em b ersh ip from
p read m ission data was explored by testing four hypotheses through
calculating a m ultiple d isc rim in a n t a n a ly sis for each hypothesis.
All hypotheses w e re supported a t the . 05 level of significance.

The

91

f i r s t hypothesis, dealing with a ll five groups, was the m o st powerful
as two orthogonal d isc rim in a n t functions re s u lte d from the c a lc u la ­
tions .
The graduated groups w ere then selected from the total
sam ple.

The relatio n sh ip of the p rea d m issio n data to th re e m e a s u r e s

of p erfo rm an c e was tested.

A m ultiple r e g r e s s io n a n a ly sis to p r e ­

dict duration of the Master* s p ro g ra m was significant a t the . 05
p ercen t level.

Canonical c o rre la tio n s between the p re a d m issio n data

and m e a s u r e s of p e rs is te n c e and academ ic p erfo rm an c e w ere a ls o
supported a t the . 05 level.
Finally, a facto r an a ly sis with oblique rotation was computed
for the total graduated group.

The graduated group was then broken

down by m a jo r a t the tim e of ad m issio n and facto r a n a ly se s w e re run
for each group.

These an aly ses re su lte d in a s e t of in te rp re ta b le

fa c to rs for the total group and for each subgroup.

The fa c to rs fo r

the subgroups w e re n ea rly identical with the fa c to rs of the to tal
group and with the fa c to rs for each o th er subgroup.

CHAPTER V

SUMMARY AND CONCLUSIONS

The purpose of this study was to in v estig ate the relatio n sh ip
of selected p re a d m issio n data to s e v e r a l m e a s u r e s of graduate s u c ­
cess.

The f i r s t question of i n t e r e s t was predicted m em b e rsh ip in

the group of students who eventually graduated, graduated and c o n ­
tinued on to the d o ctoral p ro g ra m , had ac ad em ic action taken
ag a in st them, w ere asked to leave the p ro g ra m , o r did not e n ro ll
for 5 y e a r s .

In addition, it was d e s ire d to te s t the stren g th of the

differen ces between the two graduated c a te g o rie s and the th re e c a t e ­
g o rie s of students who did not grad u ate.

T hese te s ts w ere m ade in

o r d e r to explore the relatio n sh ip of the p re a d m issio n data to the
question of graduation o r no graduation.
A second question of i n te r e s t was the relatio n sh ip of the
p rea d m issio n data for the graduated sam ple to s e v e r a l m e a s u r e s of
p erfo rm an ce in the M a s t e r ' s p ro g ra m .

The m e a s u r e s of perfo rm an ce

w ere duration of the M a s t e r 's p ro g ra m , an index of p e rs is te n c e in
the M a s t e r 's p ro g ram , and an index of academ ic p erfo rm an ce in the

92

93

M a s t e r ' s p ro g ram .

These m e a s u r e s w ere explored in o r d e r to

provide a m e a s u re of student use of college of education r e s o u r c e s .
The final purpose of this study was to exam ine the d e p a r t ­
m ental p ro file s of the p re a d m issio n data for the g raduated sam ple.
This was done in o r d e r to examine the d iffe ren c es that m ight ex ist
in the entran ce qualifications of su c ce ssfu l g ra d u a te s of the v ario u s
d ep a rtm e n ts in the college of education.
These p a r tic u la r questions w ere examined in re sp o n se to
s e v e r a l p ro b lem s r a is e d in the review of the l it e r a t u r e concerning
graduate work and g rad u ate ad m issio n s.
F ir s t, the l it e r a t u r e seem ed to indicate that th e re is a
national trend toward a g r e a t e r percen tag e of people seeking g radu ate
education.

In addition, th e re is a trend toward continuing to g rad u ate

work d ire c tly a fte r the u n d erg rad u ate p ro g ra m .

T hese fa c to rs , in

combination, have in c re a s e d the p r e s s u r e for g rad u ate ad m issio n s
o fficers.
Secondly, the lite r a tu r e seem ed to indicate that g raduate
schools a r e doing a poor job of selectio n when com pared to the p r o ­
fessional schools.

This is evidenced by a high a ttritio n r a te in the

graduate schools.
Finally, the stud ies of selectio n p ro c e d u re s in g raduate
school have been quite unresponsive to the basic question of

94

graduation o r no graduation.

In addition, they have been quite

stereotyped in th e ir use of g rad e point a v e ra g e s o r faculty ratin g s
a s c r i t e r i a for su c c e ss.
In o r d e r to investigate these concerns, a sam ple of 358
M a s t e r ' s d eg ree students belonging to one of the five r e s e a r c h c a t e ­
g o rie s a s of w inter, spring, s u m m e r, o r fall 1969 was selected .

A

s e r i e s of four m ultiple d is c rim in a n t functions was calculated to te s t
the relatio n sh ip of the se lec te d p rea d m issio n data to pred icted
category m em b ersh ip .
The group of students who had su ccessfu lly completed the
M a s t e r 's d eg re e was then se lec te d from the total sam ple.

A

m ultiple r e g r e s s io n an a ly sis was then calculated to explore the
relatio n sh ip between the p re a d m is sio n data and the duration of the
M a s t e r 's p ro g ram .

In addition, two canonical c o rre la tio n s w ere

computed to r e la te the p re a d m issio n data to a m e a s u r e of p e rs is te n c e
m ade up of the total te r m s attended, the total t e r m s m isse d , the
duration, and the av erag e co u rse load and a m e a s u r e of acad em ic
p erfo rm an c e m ade up of the to ta l g ra d e point av erag e, the total
c re d its, and the total c r e d its d e fe rre d .
Finally, facto r a n a ly se s with the oblique solution for a ll
fa c to rs with eigenvalues g r e a t e r than 1. 00 w ere run for the total
graduated group and for each m a jo r d ep a rtm e n t within the total

95

group.

The r e s u lts of these an a ly se s support the following conclusions

based on te s ts of significance a t the . 05 level w here applicable:

1.

The selected p rea d m issio n data is significantly rela te d to
m em b ersh ip in the graduated, graduated to do cto ral
p ro g ram , not enrolled for 5 y e a r s , academ ic action, or
do not re a d m it category.

2.

Selected p rea d m issio n data is significantly different for the
graduated and the acad em ic action groups.

3.

Selected p rea d m issio n data is significantly d ifferen t for the
graduated and the not enrolled for 5 y e a r s group.

4.

Selected p rea d m issio n data is significantly different for the
graduated and the do not re a d m it groups.

5.

P re a d m iss io n data for graduated groups i s significantly
re la te d to the duration of the Master* s p ro g ram .

6.

P re a d m iss io n data for the graduated group is significantly
rela te d to an index of p e rs is te n c e made up of total te r m s
attended, total t e r m s m isse d , duration, and a v e rag e course
load.

7.

P re a d m iss io n data for the graduated group is significantly
re la te d to an index of academ ic p erfo rm an ce m ade up of
total grad e point av erag e, total c re d its, and total c re d its
d e fe rre d .

8.

A fac to r a n a ly sis of p re a d m is sio n data for the graduated
group y ield s an in te rp re ta b le s e t of fa c to rs.

9.

The s e t of fa c to rs fo r each m a jo r d ep artm en t is a lm o st
id en tical to the s e t of fa c to rs fo r the total group and for
each o th er d ep artm en t.

96
Im plications for P red ictio n
of Group M em bership
Several au th o rs have suggested that it would be d e sira b le
to operate grad u ate ad m issio n s on the b asis of a prediction formula
o r on the b a s is of established cutting s c o r e s .

Bogue suggests a

sy stem for establishing a lo ss ratio for every possible sc o re in a
s e t of a d m issio n s data.

The s c o r e s with the low est num ber of p r e ­

diction e r r o r s would be the s c o r e s with the lowest loss ratio .

These

s c o re s would be selected a s the cutting s c o r e s . * Irvine su ggests
that a m ultiple prediction equation should be employed that d i s tinguishes between potential g rad u a tes and potential dropouts.

2

Mayhew th e o riz e s that the b e st we can do with our p re s e n t m e a s u r e s
3

is to s o r t out the v e ry low est students.

The data in this study s u g ­

g e s ts , however, that these concepts m ay be inadequate for the
M a s t e r 's p ro g ram in the College of Education at Michigan State
U niversity.

The te s t of Hypothesis 1 indicates that the p rea d m issio n

data is related to group m em b ersh ip along a t l e a s t two dim ensions.
M a rita l statu s (.208), resid e n ce (-.2 5 9 ), age (.270), and ad m ission
status (-.7 1 1 ) w ere the m o st im p o rta n t v a ria b le s in function I which
accounted for 54 p erc en t of the v aria n ce .

M a rita l sta tu s (. 671), sex

(-. 307), age (. 641), num ber of colleges attended for the B. A. (. 306),
num ber of y e a r s from the end of the B. A. to the M. A. application
(-. 385), num ber of y e a r s from the s t a r t to the finish of the B. A.

97

(

302), u n derg rad u ate g ra d e point a v e rag e (

619), and adm issio n

sta tu s (. 374) w e re the m o s t im p o rtan t v a ria b le s in function U which
accounted for 32 p e rc e n t of the v arian ce.

Thus, function II, the

le a s t powerful of the two functions, is composed of the trad itio n a l
m e a s u r e s of u n d erg rad u ate perfo rm an ce and m e a s u r e s of resid en ce.
Function I, the m o st powerful, is composed of m e a s u r e s of resid en ce
and ad m issio n statu s.

T h erefo re, the concept of a m ultiple p r e d i c ­

tion equation involving m e a s u r e s of p a s t p erfo rm an ce is inadequate.
Since th ere a r e two independent functions, one m ultiple prediction
equation would be inadequate to d e s c rib e the data.
powerful function is composed of resid en ce fa c to rs.

Also, the m o st
These fa c to rs

a r e not usually considered in m ultiple prediction equations based on
p a st perfo rm an ce.
The coefficients for adm ission statu s and m a jo r a t the time
of adm ission both have in te re stin g im plications for the nondegree
p ro g ram .
sion (

A dm ission sta tu s (-.7 1 1 ) and m a jo r a t the tim e of a d m i s ­

432) have high negative loadings which indicate that nondegree

adm ission sta tu s d is c rim in a te s r a t h e r highly ag ain st graduation.

Im plications for M e asu res
of G raduate P e rfo rm a n c e
The m e a s u r e s of p erfo rm an c e for the graduated group which
w ere studied w ere duration, p e rs is te n c e , and academ ic p erfo rm an ce.

98

The m ultiple R for the r e g r e s s io n equation fo r duration was . 499.
The m ultiple R

2

was .2497.

Thus, while the m ultiple was s ig n ifi­

cantly g r e a t e r than zero, the total v aria n c e accounted for was only
25 p erc en t.

With a standard e r r o r of the e stim a te of 6. 077, the

duration could be predicted plus o r m inus 6 te r m s .

The highest

Beta weights w ere a sso c ia te d with resid en ce (-. 314), age (. 389), and
num ber of y e a r s between the end of the B.A . and the M. A. a p p lic a ­
tion (-.2 3 6 ).

Thus, it is possible th at o ld er Michigan re s id e n ts

who delay the s t a r t of th e ir g rad u ate work tend to take longer in
getting th e ir Master* s deg ree.
The co rre la tio n m a trix of the p rea d m issio n data which was
p a r t of the canonical c o rre la tio n calculations revealed a few high
relatio n sh ip s between the p re d ic to r v a ria b le s.

Residence was c o r ­

rela te d with born in o r out of state (. 435) and graduated in o r out of
state (. 466), indicating that sta te re s id e n ts tend to be natives and tend
to rec eiv e th e ir B.A. in Michigan.

Age was c o rre la te d with num ber

of y e a r s from the end of the B.A. to the M.A. application (. 749) and
with the num ber of y e a r s from the s t a r t to the finish of the B .A .
(.529), indicating that o ld er students take longer to com plete the
B.A . and delay the s t a r t of th e ir grad uate work.

Num ber of colleges

attended for the B .A . i s c o rre la te d with the num ber of y e a r s fo r the
s t a r t to the finish of the B. A. (. 402), indicating quite logically that

99

the m o re colleges attended for the B. A. , the longer it takes to get
the B.A.

Finally, ad m issio n sta tu s was c o rre la te d with u n d e r ­

grad u ate grade point av erag e (. 568), indicating that ad m issio n sta tu s
is determ in ed by the strength of the underg rad u ate g rad e point
averag e.
The c o rre la tio n s of the m e a s u r e s of p e rs is te n c e showed
that m o re te rm s attended m eant longer duration (. 705) and a s m a lle r
cou rse load (-.7 5 9 ).

The canonical coefficients indicated that te r m s

m issed (. 605) and av erag e cou rse load (
im p o rtan t m e a s u r e s of p e rs is te n c e .

845) w ere the m o st

Residence (.716), age (.620),

and num ber of y e a r s from the end of the B.A. to the M. A. a p p lic a ­
tion (-.4 2 9 ) w ere the b est p re d ic to rs of p e rs is te n c e .

Thus, o lder

Michigan re s id e n ts who delay the s t a r t of th e ir g raduate work not
only take longer, but a r e le s s p e rs is te n t.
The c o rre la tio n m a tr ix of the academ ic p erfo rm an ce
m e a s u r e s showed th at these m e a s u r e s w ere not c o rre la te d .

How­

ev e r, the canonical coefficients indicated that total g rad e point
av erage (. 897) was by far the m o s t im p o rtan t m e a s u r e .

Age (-1. 036),

num ber of y e a r s from the end of the B .A . to the M. A. application
(. 726), num ber of y e a r s from the beginning to the end of the B.A.
(. 435), and u n d erg rad u ate g rad e point a v e rag e (.521) w ere the b e st
p re d ic to rs of acad em ic p erfo rm an ce.

Thus, the younger student

100
who takes longer to complete the B.A. and delays the s t a r t of
graduate work tends to have strong academ ic p erfo rm an ce.
Age, resid e n ce , and delay of the s t a r t of grad u ate study
seem to figure in the m e a s u r e s of p e rs is te n c e and academ ic p e r ­
form ance.

The relatio n sh ip of resid e n ce to graduate study is s u p ­

ported in Davis' study.

Davis rep o rted that m o st g raduate students

attend school within a four hour d rive of th e ir home.

4

W right found

age to be an in v e rse factor in his 10 y e a r longitudinal study.

5

Howg
ever, Spaeth did not find age to be a significant facto r in his study.

H e rb e rt, also, did not find a relatio n sh ip between the y e a r of B.A.
graduation and M a s t e r 's grad e point.

7

Im plications for Graduate
P re a d m iss io n P ro file s
The factor an aly ses of the total g rad uate group and the
d ep artm en tal groups yielded in te rp re ta b le facto rs which w ere a lm o st
identical.

The analy ses resu lted in 4, 5, and 6 fa c to rs which would

support the finding of Hypothesis 1 that p rea d m issio n data is made
up of m o re than one s e t of independent m e a s u r e s .

The fact th at the

se ts of fa c to rs a r e n early identical is supported by H ills' r e s e a r c h .
Hills found that m ultiple re g r e s s io n prediction equations w e re only
Q

different for widely different fields such as physica and education.

101
While a ll of the s e ts of fa c to rs w ere a lm o st identical, the
se t for the d ep a rtm e n t of secondary education is the e a s ie s t to
in te r p r e t.

These fac to rs show the p re s e n c e of a definite d o m e s ­

ticity index m ade up of re sid e n c e (-.5 2 7 ), y e a r s a t resid e n ce (-. 924),
sex (-.6 84 ), and m a r ita l statu s (-.4 5 0 ).

A B.A. p erfo rm an ce factor

also ex ists which co n sists of num ber of colleges attended for the
B.A, (-.7 3 6 ), num ber of y e a r s to get the B.A. (-.8 8 8 ), and u n d e r ­
g raduate grade point av erag e (-. 606).

T here is also a post B.A.

factor consisting of age ( -. 880), num ber of y e a r s from the end of the
B.A

t o M .A . application (-. 969), and graduate experience (. 618).

Finally, th ere is a resid en ce factor m ade up of resid en ce (.489),
born in o r out of state (. 901), and ad m issio n status (. 677).
The im plications for the College of Education a r e twofold.
F ir s t, the c u r r e n t method of evaluating a M a s t e r 1s application is
to send it to a d ep a rtm e n t re p re se n ta tiv e who in turn weighs the
inform ation available on the application blank and m akes the decision
to ac ce p t or r e je c t the applicant.

The sim ila rity of the factor s e ts

between d ep a rtm e n ts se em s to suggest that a good a lte rn a tiv e might
be college-wide ad m issio n s.

A com m ittee m ight be form ed, with

re p re s e n ta tiv e s from each d ep artm en t, to estab lish college-wide
ad m issio n s sta n d ard s.

T hese sta n d a rd s could then be applied in

weekly application evaluation m eetings.

102
Second, the fact that the p re a d m issio n data group th em selv es
into easily identifiable fa c to rs would seem to suggest that any college
ad m issio n stan d ard should be based on a system of four o r five
independent m e a s u r e s of an applicant* s potential.

Each of these

m e a s u r e s could be considered, in turn, in some s o r t of weighting
schem e.

Recommendations for
F uture R e sea rch
The exp lorato ry n atu re of this study su g g ests se v e ra l a r e a s
which might yield profitable r e s u lts in future r e s e a r c h .

F ir s t, the

College of Education could p ro fit by the extension of the m ethods of
group prediction used in this study to the a r e a of faculty decisions
in ad m issio n s.

That is , the c r i t e r i a used by v ario u s faculty m e m ­

b e rs resp o n sib le for ad m ission s screen in g in accepting o r rejecting
an applicant m ight be considered.
Another approach m ight be to apply the group prediction
methods to o ther groups.

F o r instance, the group of students who

change th e ir sta tu s o r th e ir m a jo r during th e ir Master* s pro g ram
might be examined.
The a r e a of m e a s u r e s of gradu ate p erfo rm an c e also needs
some new p ersp e c tiv e s.

The relatio n sh ip of p e rs is te n c e to a ttritio n

needs to be explored in o r d e r to form a b e tte r definition of a graduate

103

dropout.
needed.

B e tte r m e a s u r e s of academ ic p erfo rm ance a r e also
The total grad e point av erag e o ffers too narro w a view of

the total grad uate experience.
Finally, the v a ria b le s relating to m obility and resid e n ce
exhibited a consistently high relatio n sh ip with a ll of the c r i t e r i a in
this study.

Residence, y e a r s at resid en ce, born in o r out of state,

graduated in or out of state, age, m a r ita l sta u s, and num ber of c o l­
leges attended for the B.A. all a r e m e a s u r e s of m obility o r resid en ce.
These fa c to rs and o ther fac to rs relating to resid en ce mobility should
be studied in relatio n sh ip to graduation and m e a s u r e s of g raduate
p erfo rm an ce.
It is hoped th at fu rth e r r e s e a r c h in these a r e a s will be of
benefit to the C o l le g e of Education personnel concerned with the
quality of the adm ission p ro g ram .

FO O TN O TES

E, G. Bogue, "Application of a Minimum Loss Decision
Strategy in the Selection of Cutoff P oints in College and U niv ersity
A dm issions, " College and U niversity, XLIII, No. 2 (Winter 1968),
131-141.
2

Donald W. Irvine, "Multiple P re d ic tio n of College Gradua
tion from P re a d m is s io n Data, " Jo u rn a l of E x p erim en tal Education,
XXXV, No. 1 (F a ll 1966), 84-89.
3

Lewis B. Mayhew, "Nontest P r e d i c t o r s of A cadem ic
Achievement, " E ducational and Psychological M easurem ent,
XXIX, No. 1 (Spring 1969), 39-46.
4

J a m e s A. Davis, G re a t A sp ira tio n s (Chicago: Aldine
Publishing Company, 1964), p. 43.
5

Jo e L. Spaeth, "Allocation of College G rad u ates to G r a d u ­
ate and P ro fe ss io n a l Schools, " Sociology of Education, XLI, No. 4
(F all 1968), 342-349.

g
David J. H e rb e rt, "A P re d ic tiv e Study of Quality Point
A v erag es in G raduate Education C o u rse s, " J o u rn a l of Educational
R e s e a r c h , LX, No. 5 (Janu ary 1967), 218-220.
7

C h a rle s R. W right, "S uccess o r F a ilu re in Earning
Graduate D egree, " Sociology of Education, XXXVIII, No. 1 (F all
1964), 73-97.
8

John R. Hills, M arilyn B. Gladney, and Jo sep h A. Klock,
"Nine C r itic a l Q uestions About Selective College A dm issions, "
The P e rso n n e l and Guidance Jo u rn a l, XLIX, No. 7 (M arch 1967).

104

B IB L IO G R A P H Y

B erelson, B ern ard . Graduate Education in the United S ta te s.
York: M cGraw-Hill Book Company, 1960.

New

Bogue, E. G. "Application of a Minimum Loss Decision Strategy
in the Selection of Cutoff Points in College and U niversity
A dm issions. " College and U niversity, XLIII, No. 2
(Winter 1968), 131-141.
Borg, W alter E. "GRE Aptitude S cores a s P r e d ic to r s of GPA for
G raduate Students in Education. " Educational and Psycho logical M easurem ent, XXIII, No. 2 (Summer 1963), 379382.
Bottenburg, R o bert A. , and C h rista l, Raymond A. "Grouping
C r i t e r i a - - A Method Which R etains Maximum P re d ic tiv e
Efficiency. " Jo u rn a l of E x p erim en tal Education, XXXVI,
No. 4 (Summer 1968), 28-34.
Broom e, Edwin A. A H isto rica l and C ritic a l D iscussion of College
A dm ission R e q u ire m e n ts. New York: Colombia U niversity
P r e s s , 1903.
B ru bach er, John S. B ases for Policy in Higher E ducation.
York: M cG raw -H ill Book Company, 1965.

New

C areton, Edward E. , and Careton, Louise W. " P re d ic tio n of S u c­
c e s s in G raduate Study of Psychology a t U n iversity of
T en n essee. " A m erican P sychologist, IV, No. 8 (August
1949), 361-362.
C h rista l, Raymond E. "JAN: A Technique for Analyzing Group
Judgm ent. " The J o u rn a l of E x p erim en tal Education, XXXVI,
No. 4 (Sum m er 1968), 24-27.

105

106

R eport of the C itizens C om m ittee on Higher Education in M ichigan.
The Citizens Com m ittee on Higher Education, Harold T.
Smith, Executive D ire c to r, 709 South Westnedge Avenue,
Kalamazoo, Michigan, 1965,
Cooley, William W. , and Lohnes, P aul R. M ultivariate P ro c e d u r e s
for the Behavioral Sciences. New York: John Wiley and
Sons, 1962.
Davis, J a m e s A. G re a t A s p ira tio n s .
Company, 1964.

Chicago: Aldine Publishing

Eckhoff, Constance M. " P re d ic tin g G raduate Success at Winona
State College. " Educational and P sychological M e a s u r e ­
m ent, XXVI, No. 2 (Sum m er 1966), 483-485.
F ed ler, Daniel P. "Intriguing P ro b le m s of Design in P red ictin g
College S u c c e s s ." Educational and Psychological M e a s u r e ­
m ent, XXV, No. 1 (Spring 1965), 29-37.
Finn, J e r e m y D, M ultivariance: F o r tr a n P ro g ra m for U nivariate
and M ultivariate A nalysis of V arian ce and C o v a ria n ce.
Buffalo, N. Y. : The State U n iv ersity of New York a t Buffalo,
1967.
Goodrich, Max. "The A p p ra isa l and Retention of G raduate Students. "
P ro c ee d in g s of the Fifth Annual Meeting of the Council of
G raduate Schools in the U. S. Washington, D. C. : Council
of G raduate Schools, 1965.
Grigg, C h a rle s M. G raduate E ducation. New York: The C enter
for Applied R e s e a rc h in Education, Inc. , 1965.
G roesbeck, Byron L. "G raduate A dm issions - - A r e We P re p a r e d
for the Y e a rs A head?" College and U n iv e rs ity , XLII, No. 4
(Sum m er 1967), 504-510.
G ropper, George L. , and F itzp a tric k , R obert. Who Goes to G r a d u ­
ate School? P ittsb u rg h , Pa. : A m eric an Institute for
R e se a rc h , 1959.
Guilford, J . P. , and Micheal, W illiam B. The P re d ic tio n of C a te g o r ie s from M e a s u r e m e n ts . B everly H ills, C alifornia:
Sheridan Supply Company, 1949.

107

H arle, Reid F. , and Thompson, John T, "L egal and P o litical
Im plications of P o lic ie s Governing A dm issions to Publicly
Supported Institutions. " College and U niversity, XLIII,
No. 3 (Spring 1968), 274-285.
Henderson, Algo D. P o lic ie s and P r a c t i c e s in Higher Education
New York: H a rp e r and B ro th e rs, P u b lish e rs, 1960.
H e rb ert, David J. "A P re d ic tiv e Study of Quality Point A v erag es
in G raduate Education C o u rses. 11 Jo u rn a l of Educational
R e s e a r c h , LX, No. 5 (Jan u ary 1967), 218-220.
Higher Education in a Decade of D e cisio n . Washington, D, C. :
National Education A ssociation, Educational P o lic ie s
C om m ission, 1959.
H illgarth, J. W. "G raduate A d m is s io n s - - Havoc into Chaos. "
College and U niversity, XLII, No. 4 (Sum m er 1967), 498504.
Hills, John R. ; Gladney, M arilyn B; and Klock, Jo sep h A. "Nine
C ritic a l Questions About Selective College A dm issions. "
The P e rso n n e l and Guidance Jo u rn a l, XLV, No. 7 (March
1967), 640-647.
Houston, Samuel R. "G enerating a P ro je c te d C rite rio n of G raduate
School Success Via Norm ative Judgm ent A nalysis. " The
Jo u rn a l of E x p erim en tal Education, XXXVII, No. 2 (Winter
1968), 64 69.
Huganir, G eorge
Student.
Council
Council

H. "The A p p ra isa l and Retention of the P a r t Time
" P ro c ee d in g s of the Fifth Annual Meeting of the
of G raduate Schools in the U. S. Washington, D .C . :
of G raduate Schools, 1965.

Irvine, Donald W. "Multiple P re d ic tio n of College G raduation from
P re a d m is s io n Data. " J o u rn a l of E x p erim en tal E ducation.
XXXV, No. 1 (F a ll 1966), 84-89.
Jenson, Ralph E. "P re d ic tin g Scholastic Achievem ent of the F i r s t
Y ear G raduate Student." Educational and Psychological
M e asu rem en t, XIII, No. 2 (Sum m er 1953), 322-329!

108

Kurland, Norman D. "M ore People Than P la c e s o r How to E n su re
Equal Opportunity Among Equals When T here A re More
Equals Than Opportunities. " College and University, XLIII,
No. 2 (Winter 1968), 221-224.
Mayhew, Lewis B. "N ontest P r e d ic to r s of Academic Achievement. "
Educational and Psychological M easurem ent, XXIX, No. 1
(Spring 1969), 39-46.
McConnell, T. R. A G eneral P a tte r n for A m erican Public Higher
E ducation. New York: McGraw-Hill Book Company, 1962.
M eder, A lb e rt E. "A dm issions and the College. " College Admis sions. P rinceton: College E ntrance Examination Board,
1963.
M ehrabian, A lbert. "U ndergraduate Ability F a c to rs in Relation to
G raduate P e rfo rm a n c e , " Educational and Psychological
M e a su re m e n t, XXIX, No. 2 (Summer 1969), 409-419.
Newman, Richard I. "GRE S cores as P r e d ic to r s of GPA for
Psychology G raduate Students." Educational and P s y c h o ­
logical M e a su re m e n t, XXVII, No. 4 (Winter 1968), 433-436.
Owens, Thomas R. "P red ictin g Academ ic Success in Master* s
D eg ree P ro g r a m s in Education. " Jo u rn a l of Educational
R e s e a r c h , LX, No. 3 (November 1966), 124-126.
Payne, David A. , and Tuttle, Cynthia E. "The P re d ic tiv e R e la tio n ­
ship of the M iller Analogies T e st to the Objective Subjective C r ite r ia of Success in a G raduate School of E d u ­
cation. " Educational and Psychological M easu rem en t,
XXVIII, No. 2 (Sum m er 1968), 427-430.
P latz, A rth u r, and McClintoch, C h a rle s. "U ndergraduate G rad es
and the M ille r Analogies T e s t as P r e d ic to r s of G raduate
S u c c e s s ." The A m eric an P sychologist, XIV, No. 5 (May
1959), 285-289.
Rich, C atherine R. , and G a r r e tt, Thomas, eds. Philosophy and
P ro b le m s of College A d m iss io n s. Washington, D . C . :
Catholic U niversity of A m e ric a P r e s s , 1963.

109

Rodgers, E ric . "A dm ission of the D om estic Graduate Student."
P ro ceed in g s of the Fifth Annual Meeting of the Council of
G raduate Schools in the U . S . Washington, D . C . : Council
of G raduate Schools, 1965.
Roscoe, John T. , and Houston, Samuel R. "The P re d ic tiv e Validity
of GRE S cores for a D octoral P ro g r a m in Education. "
Educational and Psychological M e a su re m e n t, XXIX, No. 2
(Sum m er 1969), 507 -509.
Rosenthal, Mason P. 1130 Step-Wise Multiple R e g re ssio n P r o g r a m .
San F ra n c isc o , California: IBM Corporation, 1966.
Snell, John L. "The M a s t e r 's D egree. " G raduate Education Today.
Edited by E v e re tt W a lte rs. Washington, D. C. : A m erican
Council on Education, 1965.
Spaeth, Joe L. "Allocation of College G raduates to G raduate and P r o ­
fessio n al Schools. " Sociology of Education, XLI, No. 4
(F all 1968), 342 -349.
1130 S tatistical System.
1967.

White P la in s, New York:

IBM C orporation,

S triek er, George, and Huber, J. T. "The GRE and U ndergraduate
G ra d e s a s P r e d i c t o r s of Success in G raduate School. "
The Jo u rn a l of Educational R e se a rc h , LX, No. 10 (Ju lyAugust 1967), 466-468.
Tiedeman, David V. "The Multiple D isc rim in a n t F u n c tio n --A
Symposium. " H a rv ard Educational Review, XXI, No. 2
(Spring 1951), 167-181.
Walton, Wesley W. "P o te n tia litie s of the C om puter for M e asu rem en t
and P re d ictio n with R e sp ect to College A dm issions P r o c e s s e s . "
Educational and Psychological M e a su re m e n t, XXV, No. 1
(Spring 1965), 47-53.
W h erry s, R obert J . , and Naylor, J a m e s C. "C om parison of Two
A pproaches - - J AN and P R O F - - F o r Capturing R a te r
S tr a te g ie s ." Educational and P sychological M easurem ent,
XXIX, No. 2 (Sum m er 1966), 267-286.
“
Wright, C h a rle s R. "Success o r F a ilu re in E arning G raduate D e g ree s. "
Sociology of Education, XXXVIII, No. 1 (F a ll 1964), 73-97.