USING UNIVERSITY BRAND ASSET VALUE TO FORECAST ENROLMENT

fancyfantasicΤεχνίτη Νοημοσύνη και Ρομποτική

7 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

91 εμφανίσεις



USING UNIVERSIT
Y BRAND ASSET VALUE
TO FORECAST
ENROLMENT

John B. Green, Jr., PhD

The Ralph W. Ketner School of Business, Catawba College, United States

jbgreen@catawba.edu

Abstract

This work examines the early, pr
omising results of bench
-
marking a University's brand asset value
and

forecasting subsequent enrollment. It was accomplished by sampling the brand perceptions of
230 newly accepted students at a Southeastern US University using an adaptation of the seminal

Brand Asset Valuator® (BAV) framework (Young and Rubicam, 20
13
.) The brand's
Differentiation,
Relevance, Esteem
and

Knowledge

along with student
Psychographic Profiles

were used to: (1)
Describe the

University
X”
strategic
brand
position; (2) Develop a d
escriptive bench
-
marking linear
regression function

to track University branding progress
;
(
3)
D
evelop
multiple
enrollment forecasting
model
s

to discriminate probable matriculants from non
-
matriculants

based on the same original BAV
data
; (4)
Develop “user

friendly”

marketing
information
resource allocation scheme
s in
IBM
®
/SPSS
®

Modeler™
; and (5) Pursue future research in retention prediction based on these, and similar
methods, with the aim of
testing

an “early warning risk level classification model
.


U
sing

Modeler’s


Auto
Numeric and Auto
Classification NODE
s
, the
six and
nine
most accurate
mathematical
model forms were estimated and ranked

based

on
20

Numeric and Classification
specific variables
.

Multiple Regression and Logistic R
egression
/Discrimina
nt Analysis

were
the most
a
ppropriate
choice
s

explaining 55.42%
of
the variation in
University X’s brand asset value

and
predicting
93.5%
of the
matriculation

decision
s

accurately
.


Provocative results indicated: (1) The University's brand
value
was
led

b
y
Esteem

and
Knowledge
,
BAV’s
3rd and 4th most important
“Pillars
,


while the more important
Differentiation

and
Relevance

constructs followed well behind; (2) T
he regression function explaining

b
rand latent construct variation
proved

satisfactory

given th
e heteros
cedasti
c
c
haracteristics of the

benchmark
ing sample
; and (3) The
Discriminant Analysis
/Logistic Regression

combination

showed promise, especially if
the
ir
primary
limitation
s
, imbalance of sample matriculants vs. non
-
matricula
n
ts,
is
corrected

in
future research
.




This line of research holds out hope
,
help
and collaboration opportunities
for those institutions unable
to participate in Y&R's 50,000+ brand syndication program
along with
the
Harvard, Cambridge, NYU
and Stanford
s of the world
.

More
over, these
preliminary
results are
noteworthy for
Enrollment
Officers.
The author

reasons
that,
a
fter running
customized
intake models, and discovering the
probability of each applicant's matriculation,
administrator
could
easily calculate t
he expected va
lue of
each applicant to the admissions process.

This
would
suggest
the appropriate levels of m
arketing
resources to be expended
per applicant
.


NB
Dr

Green welcomes new
collaborators in
his
ongoing work
, especially

new
cross
-
cultural model
comparison
s

wi
th Spanish, French and Italian Universities.

Keywords: University branding,
R
esource allocation,
E
nrollment forecasting, Young & Rubicam,

BAV,

E
xpected value,
Linear
r
e
gression,
L
ogistic
r
egression, Discriminant
a
nalysis
, Collaboration



INTRODUCTION



Br
and awareness

is
“the consumer’s ability to
identify the
brand
under different conditions, as reflected
by their brand recognition or recall performance” (
Kotler and Keller 20
06
.
)
As the very
foundation of
a
brand
’s

actual
v
alue

it’s interesting to note th
at

while
o
ver

50,000
institutions participat
e

in Young and
Rubicam’s
(Y&R)
Br
and
Asset Valuator (BAV) Program (Lebar 2011
,
)
very
few are
institutions of
higher

learning
(Table 1
; C
hapleo 2011
.
)

This work contends that Colleges and Universities
lacking

Ha
rvard
-
C
ambridge
-
like”

marketing
budgets
are at a severe
competitive
disadvantage

without such
system
s

to help the
ir brand c
ompete
.


Table 1: Sampling of Young and Rubicam's University Brand Rankings

Client

Countries

US News Rank

ARWU Rank

Cambridge Univ
ersity

PRC

2**

4**

Columbia University

USA

11

6

Cornell University

USA

15

10

Harvard University

PRC

1

1

Massey University

New Zealand

299

402
-
501

National Univ
ersity
Singapore

Singapore

30

101
-
151

New York University

USA

52

32

Peking University

PRC

52

201
-
302

Stanford University

PRC

16

2

Uuniversity of Auckland

New Zealand

61

201
-
302

Tsing Hua University

PRC

223

201
-
302

University of Canterbury

New Zealand

188

402
-
501

University of Otago

New Zealand

125

201
-
302

Victoria University

New Zealand

229

402
-
501

T
his research

d
emonstrate
s

one
simple
solution

to
help
fill the gap in less endowed institutions’
branding “tool boxes”;
a

user friendly,
open source
turnkey

process
for

leveling the
branding
playing
field.


This entails emulating the Y&R me
thod of brand valuation
,

with the aid of
IBM
®
/SPSS
®

Modeler™

analytics
,
to narrow the gap
be
tween the “haves and have
-
nots”.
Unfortunately, t
he
alternative is
to
s
peculate
on
the characteristics and value of
one’s
“University
B
rand”
us
ing

d
escriptive stat
istics, graphs, intuition
and/
or personal bias

in lieu of
modern
“analytics
.

Regrettably,
t
his
can
l
ead

t
o sponsor/customer brand confusion (Kocyigit, Orhan a
nd Christian M. Ringle, 2011.)

2. CONCEPTUAL FRAMEWORK


Determining brand value is not an exact

science. None
-
the
-
less, prominent schools of thought assess
brand worth as it relates to their disciplines; commonly economics/finance vs. marketing. The
financially focused see brand asset value as an economic asset and are likely to prefer frameworks

advanced by BrandEconomics™ and/or EVA™ (Gerzema et al. 2009; Mizik and Jacobson 2005.)
Marketing oriented managers
value

a brand’s awareness, strength, vitality and/or communication
ability (Scovotti and Spiller 2006) and are more likely to
prefer

Inter
brand™, BrandZ™ or the
practicing brand manager’s

first choice, Y&R’s
Brand Asset Valuator™

(
Mills

2005.) This research
agrees with practitioners
favor
ing

the BAV conceptual framework (Fig.1)
,

by far the largest, most
profitable syndication with
over 50,0
00 brand
s participating. Obviously, developing a free model
already popular with brand managers would
narrow
Lilien’s “Academic
-
Practitioner divide” (2011.)


3. METHODOLOGY

Specifying variables required assigning eight gnostic (product market proprietary
) statements
measuring the importance of each pillar to the University X brand’s primary target market. Using


Saunder’s 1987 recommendation, still the best treatise on model/variable specification, Dr Green over
-
specified variables from theory and practic
e.



FIGURE 1: BAV CONCEPTUAL FRAMEWORK



The
n,
both sets
were subjected to
statistical elimination
.
E
ight ways each University X pillar was
Different

from its competition,
Relevant

to its markets,
Esteemed

by
its
primary target market and well
Known

by

this
market
were measured on a 5 point scale from Strongly Agree (5) to Strongly Disagree
(1) (
see
Fig.

2 and
APPENDIX A.)





Similarly

scaling was used to
specif
y

a 7
0 variable Psychographic Profile
from Critical to my decision
(5) through Not Impor
t
ant

(1)
(see APPENDIX B.)


3.
1

Sam
pl
ing Plan and Data Acquisition


Two sampling frames were used: S
0
, students who were accepted but declined enrollment (n=489)
and S
1

students who were accepted but
who did enroll (n=822.) The
1311 records

were reduced to
437
after eliminating incorrect/missing email addresses, duplicates, graduate students and “out of
office” returns.
All were contacted and asked the strength of their agreement with the
102 variable
statements
.

Those failing to respond received a second
follow
-
up
email about 3 weeks later to
increase response rate.


Th
e

S
0

response rate
was
9.2% and
the
S
1

response rate
was

26.6% yielding an overall response rate
of 19.9%. The low response rate by those who had just recently rejected admission to Univers
ity X
was not unexpected. None
-
the
-
less, nonresponse error was addressed according to Lambert and
Harrington’s recommendation (1990 p
.
8) that “In every survey involving mail questionnaires there
should be a provision for at least one follow
-
up questionna
ire so that any bias in the answers of
original respondents can be partially corrected . . .”
Thus, after

waiting another 3 weeks, a telephone
sample measuring surrogate long term nonresponse error was conducted. In an independent t
-
test,
variable mean r
esponses for twenty
-
one late non
-
responders from S
0

and forty
-
eight late non
-
responders from S
1

(n=69) proved to be the statistical equivalent of the original combined sample.
Thus, it was decided that the late observations would be added to the original
sample


3.2 Defining University Brand Position


Pillar
b
rand
value was calculated
as: (1) T
he
number
of
statistically significant
variables

per
P
illar

e.g.
2 significant
Differentiation

variables out of 8
would equal
a
Differentiation

strength of
25%;
or

(2)
The
s
um of the P
illar

statistically significant

v
a
riables


correlation
e.g.

∑D
r
R
r
E
r
K
r

where D, R, E and K are
BAV Pillars and
r

was
all statistically significant statement

correlation
s
.

Both methods yielded the
same
positioning
result

so
the first, simpler method was
chose
n
.


Two

measures

were employed to
benchmark
the
latent c
onstruct
University Brand Position
: (1)
Graphical d
ifference between
Ideal

P
illar
position
and the
A
ctual

Pillar position
when
juxtaposed
on
Y&R’s Brand Asset Consulting
Brand
s
cape PowerGrid
;

and (2) A proper
mathematical
metric
’s

change over time.






3
.3

Strategic Branding Gap
®


Fig. 3a illustrates ideal brand

S
trength

(score of 100% for brand
Differentiation
and brand
Relevance

on Y axis) and ideal brand
stature

(score of 100% for brand
Esteem
and
bra
n
d
Knowledge

on Y axis.)
Brand/category leadership i
s obtained in Quadrant 3 (Q3.) This results only when each pillar
approaches 8 significant variables out of 8 possibilities at p<=.05. Much less than that translates into
a significant branding gap (see Fig. 3b), unrealized potential (falling short in Q1

or Q2) or a decline of
a past brand leadership into a ubiquitous commodity status (Q4.)
In deference to Ansoff’s
Strategic

Gap

(1965),
this work defines a new term

-

Strategic Branding Gap
®

(SBG
)
,
when one’s actual
position fall
s

significantly short of
t
he
ideal

position
.

SBG
,

ultimately
, also
defines the
measure of
one’s brand positioning goal

to be attained
.







3.
4.

Regression Benchmark
and Classification Models


Also c
ritical to benchmarking
one’s
brand
position change over
time is

measuring
the
va
riation
explained by the
mathematica
l form measuring that
change
.
Here
, based on
IBM
®
/SPSS
®

Modeler
’s


AutoNumeric and AutoClassification NODE’s ranking of 6 Numeric and 9 Classification
methods,
the author
chose
M
ultiple
L
inear
R
egression in the form: Y

= a + b
1
X
1

+ b
2
X
2

+ b
i
x
i

+
e

where
Y is the dependent variable, a subjective assessment of the overall effects of

32 independent
BAV
variables

(BAV32) defining
Difference
,
Relevance
,
Esteem

and
Knowledge

and
a 70 variable
psychographic profile (
AIO
70) def
ining one’s
A
ctivity, Interest and
O
pinions
. The dependent variable,
Y, was scaled as an 11 point variable from 0 to 10, where 0 meant the
complete absence of
difference, relevance, esteem
and

knowledge

and 10 meant a complete compliment of
Difference
,
Re
levance
,
Esteem

and
Knowledge
.


3.5
Model

Selection


M
ultiple regression was
IBM
®
/SPSS
®

Modeler’s™

second choice but was chosen over the first choice,
CHAID, due to an insignificant difference in overall
model
correlation (.868 vs
.

.873,) the greater
fam
iliarity of regression to practitioners

and
its robust reputation.
Though
Logistic Regression

and

Bayesian Networks
ranked ahead of
Discriminant Analysis

(Fig 4)
, a
ll three ranked from 96.52% to
100% accuracy, over
-
fitting notwithstanding
.

Using re
asoni
ng similar to the Auto
N
umeric

model
decision, Discriminant Analysis was chosen over
Bayesian
Networks and Logistic Regression due to
the insignificant difference in their accuracies and the
wide
familiarity of
the technique among

practitioners.

Logistic R
egression was
, ultimately,
chosen to support
the
Discriminant Analysis

procedure due to comparable accuracy and its
immunity

to many Discriminant Analysis assumptions.
When each
procedure

was run, the variables recommend were combined yielding
a “Common20
.

These

included
d4 d8 r2 r8 e2 e7 e8 k1 k5 k8 a24 a28 a32 a38 a40 a41 a48 a58 a60 a62

and
are
embolden

and
enlarged

in APPENDIX A and B
.




Fig.

4: IBM®/SPSS® Modeler’s™ AutoClassification NODE Model Selection



4.
Results

Both
the
full
data set

and the

“C20”

were internally consistent with Chronbach Alpha
s

of .961
and .842
respectively.
The “C20”
were

also
: (1)
I
n harmony with Saunder

s
recommendations

for variable
specification; (2)

F
riendlier


with “
once collected
model construction
data


use
d

in th
ree different
models
;

(3)

P
arsimon
io
u
s; and (4)

Practitioner/
M
odel
R
elativ
e


(
Lilien 1
975
.
)

Thus,
wider
use by
administrators
is
to be
expected
.



4.1 Goal 1
Strategic Branding Gap


B
rand leadership
exists
in
Q3.

Fig
.
3b

demonstrates a

significant
fa
ll
over time
into
a
Q4
commodity
with

predictable
, and actual,
increase
s

in
pric
e
competition.

While
the SBG position found
Difference

relatively sound at

62.5%
, Relevance at 12.5%
dragged
the SBG
downward
. University X still had the
good will of many
, b
ut had become
painfully irrelevant

to current target market interests.
With a
poor
product mix (missing,
old fashion or otherwise w
ounded majors)
the former
brand leader
could not
compensate for
the irrelevance score.


4.2 Goal 2:
Multiple Linear Regre
ssion

Benchmark


Goal 2 was achieved using t
he
“C
2
0


variables
. The

initial benchmark r
egression
was reasonable
(see
Table 2
.
)


Multicollinearity was not
an issue due to a KMO score

of .855

fo
r
the
Orthogonal
Factor Analysis
.

The function
s
statistically

significant

F
actor
s, in

rank order, and the
function’s
R
2

of
.561
suggest a most reasonable low point
for future
branding
improvement
.


4.
3

Goal
3a
:
Discriminant Analysis


Goal #3
using the
d
iscriminant
analysis
form D
p

= λ
0

+ λ
1
X
1

+ λ
2
X
2

+ λ
i
X
i

+
e

separated the
matriculants from the nonmatriculants
reasonably well
(See Fig. 5.)




Fig.

5: Discriminant Analysis

20 Variable Prediction




T
he
classification

accuracy
was
93.5% and cross
-
classification
(Leave one out method) accuracy was
91.7%
, possibl
e over fitting notwithstanding
.

And while the

Wilks’ Lambda
of .511 a
nd a Canonical
Correlation of .699 account
ed

for 48.2% of variation, the exercise was considered a 1
st

time success.

However, these results give pause with the function’s Eigenvalue rou
nded to
exactly
1
,
a severe
imbalance in the nonmatriculants (n=40) vs
.

matriculants (n=190) causing lack of
homoscedasticity

and some correlation in the
C20 variables
.

But,
discriminant analysis models developed from small
samples have been found robust,

even with covarying multivariate and
heterogeneous

matrix
problems (Frank et al, 1965; Klecka 1980.)
With

c
lassification rates
being the ultimate judge of validity
(Chakrapani 2004) and these found to be
much higher than
apriori
expectation
,
one can conf
idently try
again with more data

probably improving all validating statistics and bringing assumptions closer to
rigorous

guidelines
.



None
-
the
-
less,
as
a supporting
precaution
,

given

Logistic Regression
’s

me
re robust
reputation,

the


C20 variable p
rocedu
re

demonstrated similar classification rates
in 5 t
hrough

7 iterations
(89.1
%

to
96.3%
accuracy
.
)

Moreover, a
superior aposteriori nonmatriculant classification accuracy

demands
further investigation
.

Thus,
both techniques
look
especially

promising
,
espec
ially
if
one assumes
the
severe imbalance of matriculants is remedied.



5.
Limitations

and Future research



The most obvious limitation of this work is the sample size of
only 40
“no
nmatriculants. These
respondents were, obviously,

less motivated t
o participate

than matriculants
.


S
uggest
ing
reason
s

for
matrix homogeneity issues
,
judgment suggest that
future attempts will be display more
homoscedasticity
.
S
olutions should get more
robust as
new data
is acquired at each
intake.



This work is
not
presumptuous
enough
to suggest that one linear regression model and
a
discriminant
analysis model
are exhaustive in measuring brand equity/awareness

and also predicting future
enrollment
.
But there are many o
ther procedures, such as
this works
temptin
g
Logistic Regression

classification
comparison, that
may be as
good as or

even better

than Discriminant Analysis
. Further
research using nonlinear techniques is
strongly
recommended.


Finally, s
uccessfully
reproducing these results in similarly profile
d US institutions,

along with real
-
time
external validation of this and other efforts, is needed. The author welcomes inquiries into greater
details
concerning
this work and eagerly supports external validations

and
global cultural explorat
ory
work
, espec
ially
with
Spanish, French and Italian institutions
both
large and small
.


6.
CONCLUSION


This research has derived an easily understood
, user friendly BAV
system which
could
: (1)
S
tandardize
brand name

assessment; (
2
)
S
uggest
graphical and empirical
(
re)positioning
branding
strategies and metric;
(
3
)
A
ccurately predict

ultimate enrollment
as well as
aid in marketing to
indecisive students
; and (4) Allow
expected value theory to be part of an overall model to include
optimizing enrollment marketing expe
nditures. Bot
h

large and small institutions
may well
benefit from
the

free
,

open source methods

demonstrated here
.

To
wit:

“While analytics are not perfect, we prefer
them to the shoddy alternatives of bias, prejudice, self
-
justification and unaided int
uition” (Davenport et
al. 2010, p. 17.)
This process’s first iteration is credited with contributing to the 103% increase an
enrollment last year; by far, the largest growth in memory. The end product of these methods, while
only Excel simulation today,
may someday be the
“apps”

for the
tech
-
de’jour
.

APPENDIX

A: BAV 32 Variables

Please choose your level of agreement with the 32 statements below:

5=Strongly agree; 4=Agree; 3=Neither agree/nor disagree; 2=Disagree;1=Strongly Disagree


5 4 3 2

1


d1 I believe
University X

is unique among
Universities

I have considered


d2 Compared to other
University
s
,

University X

really cares about developing each student’s
individual talents completely

5 4 3 2 1

d3 The primary difference b
etween
University X

and other small
Universities,
is
its

emphasis
on academic excellence

5 4 3 2 1

d4 Compared to other
Universities
,
University X

really cares about developing
her students’ academic potential

5 4 3 2 1

d5 The
primary difference between
University X

and other small
Universities

is
its

emphasis on
athletic excellence

5 4 3 2 1

d6 Compared to other
Universities
,
University X

really cares about the personal and
professional success of each of her stud
ents

5 4 3 2 1

d7
University X

is different because she offers a balance in academics, athletics and social
activities

5 4 3 2 1

d8 The primary difference between
University X

and other
Universities
is her
ability to provide la
rge financial aid packages

5 4 3 2 1

r1
University X

has the campus culture and social environment I look for in a
University

5 4 3 2 1

r2
University X

has a network of distinguished alumni, something that is very
important fo
llowing one’s graduation

5 4 3 2 1

r3
University X

has the major(s) most appropriate to my career/professional interests

5 4 3 2 1

r4
University X

has the courses that most appeal to my personal interests

5 4 3 2 1

r5
University X

has the majors I require for success in graduate school

5 4 3 2 1



r6
University X

has the job/career placement services I need following graduation

5 4 3 2 1

r7
University X

has a distinguished faculty who
really care about each student’s success

5 4 3 2 1

r8
University X

has the small, friendly environment I look for in a

school
University


5 4 3 2 1

e1
University X

is an excellent Liberal Arts
University

5 4 3 2 1

e2
University X

is highly respected throughout the Southeast

5 4 3 2 1

e3 A
University X

degree from is extremely valuable; it “paves the way” to a very bright future

5 4 3 2 1

e4
University X

is held in high esteem by my
family

5 4 3 2 1

e5
University X

is held in high esteem by my friends

5 4 3 2 1

e6
University X
’s academic reputation is superior to most small Liberal Arts
Universitie
s

5 4 3 2 1

e7
University X

has the best and

brightest faculty in the Southeast

5 4 3 2 1

e8
University X

gets the best and brightest students in the Southeast

5 4 3 2 1

k1 I am familiar with
University X

advertising, promotion, events and/or news
articles

5 4 3

2 1

k2 I know many of the majors offered by
University X

5 4 3 2 1

k3 I follow a
University X

athletic team or know someone who played a sport for
University X

5 4 3 2 1

k4 I am familiar with the
University X

guarantee: “
Our Purpose: Your Promise”

5 4 3 2 1

k5 I have attended a musical, theatrical and/or other public performance at
University X

5 4 3 2 1

k6 My parents know a lot about
University X

5 4 3 2 1

k7 My friends know a l
ot about
University X

5 4 3 2 1

k8 My teachers/professors know a lot about
University X

5 4 3 2 1


APPENDIX

B: AIO 70 (Psychographic Variables)

Please rate how important the following ACTIVITIES, INTERESTS AND OPINIONS were in
your final UNIVERSITY
DECISION: (5) Critical; (4) Very Important; (3) Important; (2) Somewhat Important
; (1) Not important.


1
University

reputation

2 Great faculty

3 Great sports

4 Liberal political environment/viewpoint

5 Distinguished speakers

6 Dist
inguished alumni

7 Number of majors

8 Quality of intended major

9 Great theatre/performing arts events

10 Great parties

11 Great music

12 Religious affiliation/environment

13 Attending an event (e.g. theatrical, sporting,
homecoming etc..) with a faculty m
ember

14 Inexpensive tuition

15 Scholarships

16 Loans

17 Plenty of on campus work
-
study jobs

18 International travel opportunities

19 Graduation job opportunities

20 Graduate school opportunities

21
University

close to girlfriend/boyfriend

22 Close to home

23 Attending an event (e.g. theatre, sports) with
University
's current student(s)

24
University

offers spring break (for
course credit) in cool countries

25
University

degree completion programs e.g. 4
year
University

has program with 2 year college
to ac
cept all their/your transfer credits

26 # of other
University
s that already accepted
39
University

guide books

40 High S
chool/counsellor recommendation

41 Large
University

42 Friendly admissions staff

43 Friendly stud
ent body

44 Experiencing an actual class in session

45 One
-
on
-
one discussions with faculty from my major

46 Student guided campus tours

47 Seeing a
University

advertisement in a magazine

48 Seeing an advertisement for my
University

in a newspaper

49 Seeing

an advertisement for my
University

on a bill
-
board

50 Hearing an advertisement for my
University

on the
radio

51 Interacting with the
University
's web page

52 Going to a
University
's sporting event

53 Going to
University
's theatrical/musical performance

5
4 Meeting a faculty member in their office

55 Hearing from a faculty member by telephone

56 Hearing from the
University

President by telephone

57 Listening to a
University
's PodCast

58 Attending a
University
's a local event in my
hometown sponsored by the
University
/Alumni

59 Going to a
University
/Career fair

60 Having a
University

representative speak at
my High School

61 Guaranteed ability to live in campus dorms all 4
years

62 Feeling the campus would be a safe place to
live



you

27 Scholarship offers already received

28 Cool city

29 Laid
-
back rules

30 Housing quality

31 Food quality

32 Fraternity/sorority possibilities

33 Tutoring services

34
Other learning disability services

35 Parents' alma mater

36 Parents' wishes

37 Friends attending

38 Friends reco
mmendation

63 Meeting a graduate of the
University

while shopping,
working, going to a concert etc.

64 Feeling I could succeed at that
University

65 Meeting a
University
's current student or alumnus(a)
while on FaceBook/other social WWW site

66 Having LIVE instant messaging CHAT with
University

representative while on
University
's WWW page

67 Being near OTHER
University
s in the area

68 Reading about the
University
's high ranking in US
News and World Report

69 Having a family member work at the
University

70
University

will accept all/most of cred
it earned

elsewhere


REFERENCES


[
1
]

Ansoff, H. Igor. (1965)
.

Corporate Strategy: an Analytic Approach to Business Policy for Growth
and Expansion. Harmondsworth: Penguin.

[
2
]

Chakrapani, Charles (2004)
.

Statistics in Market Research, London: Arn
old Publishers.

[
3
]
Chapleo, Chris (2011)
.

Exploring Rationales for Branding a University: Should We be Seeking to
Measure Branding in UK Universities?,


Journal of Brand Management

18 (April),

pp
.

411
-
422.

[
4
]
Davenport, Thomas H., Geanne G. Harris, an
d Robert Morison (2010)
.
Analytics at Work.
Cambridge MA: Harvard Business School Publishing Corporation p
.
17.

[
5
]
Frank, R.E., W.F. Brassy, and D.G. Morrison (1965)
.

“Bias in Multiple Discriminant Analysis,”
Journal of Marketing Research
.

2(8)
pp.
250
-
258.

[
6
]
Gerzema, John, Ed Lebar and Anne Rivers (2009)
.


Measuring the Contributions of Brand to
Shareholder Value
.


The Journ
al of Applied Corporate Finance,

21(4)
.

[
7
]
Klecka, William R. (1980)
.

Discriminant Analysis
.

Newberry Park, CA: Sage Publica
tions.

[
8
]
Kocyigit, Orhan and Christian M. Ringle (2011)
.
The Impact of Brand Confusion on Sustainable
Brand Satisfaction and Private Label Proneness: A Subtle Decay of Brand Equity
.

Journal of Brand
Management 27 (May)
.

[
9
]
Kotler, Phillip and Kevin
Lane Keller (2012), Marketing Management. Upper Saddle River, NJ:
Prentice Hall.

p. 140.


[
1
0
]


Lambert, Douglas M. and Thomas C. Harrington (1990)
.

“Measuring Non
-
response Bias in
Customer Service Mail Surveys
.

Journal of Business Logistics.

11(2), 8.

[
1
1
]
Lebar, Edward (2011)
.

Brand Asset Consulting: Young and Rubicam Brands, accessed January
4, 2011, http://brandassetconsulting.com.

[
12
]
Lilien, Gary L. (2011)
.

“Bridging the Academic

Practitioner Divide in Marketing Decision Models
.

Journal of Ma
rketing
.

75(4)
pp
.

196
-
210
.

[
13
] Lilien, Gary L. (1975). “Model relativism: A Situational approach to model building.”
Interfaces
,
5(3)
pp.

11
-
18.

[
1
4
]
Mills, Roger (2005)
.

Manager Update.

Henley
-
on
-
Thames: Spring 16(3) 5
.

[
1
5
]
Mizik, Natalie and Robert Jacobson (2008)
.

The Financial Value Impact of Perceptual Brand
Attributes
.

Journal of Marketing Research
.

45(1), 15.

[
1
6
]
Saunders, John (1987)
.

“The Specification of Aggregate Market Models
.


European Journal of
Marketing
.

21(
2),
pp.
5
-
47.

[
1
7
]
Scovotti, Carol and Lisa D. Spiller (2006)
.
Revisiting the Conceptual Definition of Direct
Marketing: Perspectives from Practitioners and Scholars
.

Marketing Management Journal
.

16(2): 4
.


[
1
8
] Young and
R
ubicam Consulting. 20
13
http://bavconsulting.com/
.