. _ . ‘. _ y
m V .... A , ..
mxm mmfl ,5 g :
Rmu nmn_ ,._.:A, ,@
ADm PEN , . ..
V0 .1. U ., V.
T. 0V0
SWN ,m.
., mom en“. 7
PmR. eTMl
0 hs. _
. “Hm tuflun
mmrnu. mmm
WARN flR
DAN fimbl
AS.” hM
GAN T...
_ m 0.
. S C.
U
o . .

 
  
  

.326. _. .r

 

 

This is to certify that the
thesis entitled

USING ADOPTION PROCESS VARIABLES AS A PREDICTOR
OF PRODUCT CONTINUANCE OR DISCONTINUANCE

presented by

Patrick Michael Dunne

has been accepted towards fulfillment
of the requirements for

Marketing

 

 

Ph.D

degree in

 

0-7639

 

   

   
 

dial .3! :27
HUM: & SUNS'
353K blNDtRY NC“
anamv emoeas

SPIINEMIT. “ICING”
“ ‘ 4M

  

U

5W
‘A‘S

M.
cow

ABSTRACT

USING ADOPTION PROCESS VARIABLES
AS A PREDICTOR OF PRODUCT
CONTINUANCE OR DISCONTINUANCE

BY

Patrick Michael Dunne

Over $300 billion will be spent on new product in—
novation in the 1973 fiscal year. Yet, over 70 per cent of
this cost will go to products that will not be successful
in the market place. Nowhere is this problem of new product
introduction more prevalent than in the retail food industry.
Thus, the stated purpose of this study was to provide the
retail food manager a means of predicting which of the
products distributed through his retail food outlet should
be eliminated from his product line at a point in time far
earlier than the usual analysis of thirteen weeks sales
data. Instead of reviewing the initial three months sales
results, it was hypothesized that knowledge of the $2331
of consumer adOption process variables, as well as knowledge

of the rate 9g growth of these variables, could be used to

I' i '
35:12: management 5 c

 

:e‘-‘§rec"1ct at a PM!“1
:23. the point 1"
asks after product i
as an arbitrary deCl
s:;:;ficant reductior
tieccntinuation deci
Six variable:
.Ei'eis of activity f
595115: seven new pr
iiiilarket during t

.
~J.f‘:‘h
I

#4.
' «Cs ‘-
LO detErmln
| ‘ apPrOac
lice
sL 1
' lnea
1' d
18c
"35?. r
\‘ pe

Patrick Michael Dunne
predict management's decision to continue or discontinue a
new product at a point in time earlier than in present use
today. The point in time chosen for this study was five
weeks after product introduction. While admittedly, this
was an arbitrary decision, it was chosen so as to make a
significant reduction in the amount of time needed to make
the continuation decision.

Six variables of the adOption process were selected
for testing in this research. The research measured the
levels of activity for these six predictor variables
against seven new products introduced in the Des Moines,
Iowa market during the summer of 1972. The seven products
studied were of a similar nature in both terms of level of
newness and in terms of consumer product classification.
The study presented its own analysis of this definitional
classification as well as what items were to be considered
important in the adOption process. A telephone survey was
conducted of customers for a selected chain store in Des
Moines to determine the level of activity for the products.

The approach to analyzing the data was twofold.
First, linear discriminate analysis was used to test if the
weekly percentages for the second through fifth week after

introduction of the six predictor variables chosen were

21m discriminate '.
22:51:; to its cri:
21:12 at the end of t
32:11: hypothe s is w a
ext-tests. Also,

5:295 on the mean I
53.5615: ntinued gr'

Lie same time oerig

.
n
i

‘.
J; .,

tile‘ _
i=5.“

m 3

Standard

'932::

. Of ,

he “leap

Patrick Michael Dunne
able to discriminate between the products which management,
according to its criteria, decided to continue or discon-
tinue at the end of thirteen weeks. Second, the rate of
growth hypothesis was tested by means of eighteen independ—
ent t-tests. Also, two—way analyses of variance were per—
formed on the mean rates of growth between the continued
and discontinued groups by the six predictor variables for
the same time period as the earlier tests.

The study was able to differentiate between the con—
tinued and discontinued products by using the weekly per—
centage levels of the six variables. In the fifth week the
discriminate function was found to be significant at the
.01 level and to correctly assign all seven products to
their prOper grouping. The thesis, also, analyzed the data
from the second, third and fourth week. This was to see if
an earlier time period could produce significant results,
likewise. While the second week's function was found to
be significant at the .02 level and correctly predicts the
outcome of all seven products, the observed difference in
the means of the two product groups was only nine times the
week's standard error. The fact that the observed differ—
ence of the means for the fifth week was seventy times as

great supported the notion that the fifth week is able to

gritie more conc his 1

~ 1
a. p'fi:flnsfl*
.... .uv‘hu» v. o

The function

(u

2: :. '79 teen inflcer.
35:, :‘rey were sti'.
}IZ§'LC‘. groupings c:
“aria‘tles were four.
:: the continuation
333932 weak inf:
.3512“. type.

The analys

Presented 9.11

gas in ?_ ~

A
...

'4
’~..\

He‘serthe 1‘

Patrick Michael Dunne
provide more conclusive evidence as to the continuation of
the product.

The function for the other two weeks was believed
to have been influenced by extraneous variables. Neverthe—
less, they were still able to predict five of the seven
product groupings correctly. Three of the six predictor
variables were found to exercise a great deal of influence
on the continuation of a new product. They were weak
interest, weak information seeking, and knowledge of the
product type.

The analysis of the rates of growth as a predictive
tool presented evidence, which tends to support the hypothe-
sis that the rates of growth cannot be used to predict the
continuation or discontinuation of the product.

Nevertheless, it was shown that for at least one
product grouping at one time and in one geographic location,
a linear discriminate function of six adOption variables
could be used to predict product continuation or discontinu-
ation. Prior to this time the adOption variables have been
used as a post—Operative tool to eXplain what has happened.
Now these variables have been shown to be useful as a pre-

dictive tool as well.

csrx:

f‘“
J.

USING ADOPTION PROCESS VARIABLES
AS A PREDICTOR OF PRODUCT
CONTINUANCE OR DISCONTINUANCE

BY

Patrick Michael Dunne

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Marketing and Transportation Administration

1972

COpyright by
PATRICK MICHAEL DUNNE
1972

To Joe, Ray

and their families

ii

like completj
f‘;:..::;on of one iridl
sent the author fro-r
::;':';:ec‘. to that cor
be especially noted
Dr. Donald 1
35;artnent of Marke‘
Echigan State Uniw
littee and academic
5“" 5‘4PPort through
3:“ Tal‘lor's advice
SE‘I‘.’e as a mOdel f’:
Dr. Richard

.
r3252”

thcltion Admi

w
I. ‘
,31FA

d.‘.

ACKNOWLEDGMENTS

The completion of a doctoral program is never a
function of one individual. While space limitations pre-
vent the author from thanking all those whose efforts con—
tributed to that completion, the following peOple should
be especially noted for their assistance:

Dr. Donald A. Taylor, Professor and Chairman,
Department of Marketing and Transportation Administration,
Michigan State University, as chairman of the research com—
mittee and academic advisor, has provided close guidance
and support throughout the author's graduate program.

Dr. Taylor's advice and philOSOphy of life will always
serve as a model for the author.

Dr. Richard J. Lewis, Professor of Marketing and
Transportation Administration, Michigan State University,
has always been a helpful counselor to the author. As both
a member of the research committee and a neighbon,Dr. Lewis
was always available with advice to solve the problem at
hand.

Dr. Leo G. Erickson, Professor of Marketing and

iii

   
  
  

frazsror-tation Admin;
as a center of the re
it: fine inception of

:esm appreCiated.

D-

Dr. Eddie V.

'01

.‘I§ -‘
~10. '

Marketing,

['3

Q"
s “"65

........., Assistant 2

any, were instrure:

......

.;:‘-.i;, and MIS. C

:v ,
‘I

C ‘9. ‘h x
"’ “9 triankeé

.3558";
{9- In an:

5%,:
‘i.")‘r‘.’ .‘ ‘
‘ '0 W‘l

‘ 4 Ci) 'T

‘A

tile *
A flna‘L
5 f5:

Transportation Administration, Michigan State University,
as a member of the research committee served tirelessly
from the inception of the research. His assistance is
deeply appreciated.

Dr. Eddie V. Easley, Professor and Chairman, Depart-
ment of Marketing, Drake University, and Dr. Vasanth B.
Solomon, Assistant Professor of Statistics, Drake Univer-
sity, were instrumental in the deve10pment of the research
design. Dr. Solomon's assistance with the statistical
methodology is greatly appreciated.

Miss Mary Hershberger, who did all the rough draft
typing, and Mrs. Jo McKenzie, who typed the final copy,
are to be thanked for putting up with the author's constant
pressure. In addition, sincere appreciation goes to the
company, which must remain anonymous, that permitted the
author to use its store for this study.

A final word of thanks must go to the members of
the faculty of the Graduate School of Business Administra—
tion at Michigan State University for their assistance in

the develOpment of the author's professional career.

iv

TABLE

ACKNOWLEDGMENTS. . . . .
LIST OF TABLES . . . . .

LIST OF FIGURES. . . . .

Chapter
I INTRODUCTION. .

Background. .

OF CONTENTS

Statement of the Problem.
Implications for Marketing Problems
Conceptual Framework.

Methodology .

Limitations of the Study.
Some Possible Contributions
of the Study .

II LITERATURE REVIEW

The Adoption Process.
Ryan and Gross Hybrid Corn Study.
Five-Stage Ad0ption Process Model
"Hierarchy of Effects" Model.
"AIDA Model".

Critique of Adoption Process Models

Innovation-Decision Model

Marketing's Explanations of Consumer

Decision-Making Process.
Summary of Ad0ption Models.

Industrial Goods and the AdOption Models
The Diffusion Process

The Diffusion Process in Marketing.
Adopter Categories.
The Innovation-Decision Period.

V

Page
iii

viii

ll
14
17
19

20

25

26
27
29
3O
31
32
35

36
43
46
46
48
49
54

Iagter
Diffusi
Opinior
Indust:
Di fu
Innovati
Rogers
Other
Produc
Backgt
Tradit
Theort
Reta:
Summary

Chapter Page
Diffusion Effect. . . . . . . . . . . . . 57

Opinion Leadership. . . . . . . . . . . . 59
Industrial Goods and the
Diffusion Process. . . . . . . . . . . . 60
Innovation Characteristics. . . . . . . . . 61
Rogers' Characteristics . . . . . . . . . 61
Other Studies . . . . . . . . . . . . . . 63
Product Elimination Studies . . . . . . . . 64
Background. . . . . . . . . . . . . . . . 64
Traditional Approaches. . . . . . . . . . 66
Theoretical Approaches. . . . . . . . . . 67
Retail Grocery Practices. . . . . . . . . 74
Summary . . . . . . . . . . . . . . . . . . 76
III RESEARCH DESIGN . . . . . . . . . . . . . . . 84
Product Classification. . . . . . . . . . . 84
Hypotheses. . . . . . . . . . . . . . . . . 94
Sample Design . . . . . . . . . . . . . . . 95

The Sampling Frame. . . . . . . . . . . . . 95
Data Collection . . . . . . . . . . . . . . 98
Analysis of the Data. . . . . . . . . . . . 102

IV PRESENTATION OF FINDINGS. . . . . . . . . . . 113

The Management Decision . . . . . . . . . . 113
Weekly Percentages of the Six
Variables as Prediction Tools. . . . . . . 115
Second Week Findings . . . . . . . . . . 121
Second Week Summary. . . . . . . . . . . 124
Third Week Findings. . . . . . . . . . . 124
Third Week Summary . . . . . . . . . . . 128
Fourth Week Findings . . . . . . . . . . 128
Fourth Week Summary. . . . . . . . . . . 132
Fifth Week Findings. . . . . . . . . . . 132
Fifth Week Summary . . . . . . . . . . . 134
Summary of Weekly Percentage Tests . . . 135
Rate of Growth of the Six Variables
as Prediction Tools. . . . . . . . . . . . 136

V SUMMARY’AND CONCLUSIONS . . . . . . . . . . . 145
Objectives of the Study . . . . . . . . . . 145

Empirical Findings. . . . . . . . . . . . . 148
Implications of the Research. . . . . . . . 150

vi

A Census Tra:
Des Moines

Trading Ar

E Telephone
viewing Pi
Recording

- ' s

-" ‘flnyapk

..~..4.. wMy
J c

Page
Appendix
A Census Tracts of Polk County, Iowa;
Des Moines, Iowa; Selected Store's
Trading Area. . . . . . . . . . . . . . . . . 157
B Telephone Questionnaire, Telephone Inter-
viewing Procedures, Telephone Data

Recording Procedures. . . . . . . . . . . . . 162

Bibliography . . . . . . . . . . . . . . . . . . . . . 167

vii

'7

Hyootnetice
Six Predic

Mean Value
Discrimina

Values: Se

Hypothetic
Six Predi:

ANTCA Tat
HYpotheti

Second we
prEdiCto:

Third WEE
Prefiicto;

FDDrth w
Pr9d1CtQ

Fifth We
prEdith

Mean Va

Table

3-1

LIST OF TABLES

Hypothetical Second Week Levels of the
Six Predictor Variables . . . . . . . .

Mean Values of Predictor Variables,
Discriminate Weights, and Importance
Values; Second Week . . . . . . . . . .

Hypothetical Rate of Growth for the
Six Predictor Variables, Third-

Second Week . . . . . . . . . . . . . .

ANOVA Table for Rates of Growth of the
Hypothetical Third-Second Week. . . . .

Second Week Levels of the Six
Predictor Variables . . . . . . . . . .

Third Week Levels of the Six
Predictor Variables . . . . . . . . . .

Fourth Week Levels of the Six
Predictor Variables . . . . . . . . . .

Fifth Week Levels of the Six
Predictor Variables . . . . . . . . . .

Mean Values of Predictor Variables,
Discriminate Weights, and Importance
Values: Second Week . . . . . . . . . .
Second Week Results of Assignment Tests
Mean Values of Predictor Variables,
Discriminate Weights, and Importance
Values: Third Week. . . . . . . . . . .
Third Week Results of Assignment Tests.
Mean Values of Predictor Variables,
Discriminate Weights, and Importance

Values; Fourth Week . . . . . . . . . .

Fourth Week Results of Assignment Tests

viii

Page

105

107

109

111

116

117

118

119

123

123

126

126

130

130

\-
M
.P

.-..9

J -
I
0—4
F—0

Meat
Disc
wait

.-;2 Fifi

.
‘. I Q'\W
‘ .4

n

17:.
J...

ls.

-- N

. LJ 7-
.u
Va

l',

3-»

.’ s
~gl‘.
‘ .I 9-
a
5
C
§
'i
7‘1
s
I.
sea
J

Table

4-11

4—19

Page
Mean Values of Predictor Variables,

Discriminate Weights, and Importance
Values; Fifth Week. . . . . . . . . . . . . . 133

Fifth Week Results of Assignment Tests. . . . 133

Summary of the Relative Importance
Percentages . . . . . . . . . . . . . . . . . 136

Rate of Growth for the Six Predictor
Variables; Third-Second Week. . . . . . . . . 137

Rate of Growth for the Six Predictor
Variables; Fourth-Third Week. . . . . . . . . 138

Rate of Growth for the Six Predictor
Variables; Fifth-Fourth Week. . . . . . . . . 139

t-test Scores for the Rate of Growth
for the Six Predictor Variables . . . . . . . 140

ANOVA Table for Rates of Growth of the
Third-Second Week . . . . . . . . . . . . . . 142

ANOVA Table for Rates of Growth of the
Fourth-Third Week . . . . . . . . . . . . . . 142

ANOVA Table for Rates of Growth of the
Fifth-Fourth Week . . . . . . . . . . . . . . 142

ix

Figure

1-1

2‘6

2‘7

2‘8
2‘s

‘1

LIST OF FIGURES

Planned Contribution of New Products
to Sales Growth, 1963-1967. . . . . . .

Illustration of a Discriminate Function

Relationship of the Three AdOption
Process Models Hierarchy. . . . . . . .

Relationship of AdOption Process Stages
of the Four Campbell Models . . . . . .

The Nicosia conceptualization of the
Purchase Decision Process . . . . . . .

The Andreasen Conceptualization of the
Purchase Decision Process . . . . . . .

The Howard-Sheth "Theory of Buyer
Behavior" . . . . . . . . . . . . . . .

Summary AdOption Model. . . . . . . . .

Adopter Categorization on the Basis
of Innovativeness . . . . . . . . . . .

The Berenson Abandonment Model. . . . .
The Kotler Abandonment Model. . . . . .

New Product Dimension Model . . . . . .

Page

16

32

35

38

4o

41

45

50

69

71

88

Each y

zillign ll new"

CHAPTER I
INTRODUCTION

Background

Each year American business firms introduce several

million "new" products to the market place. The introduc-

t:iJDn of these so-called "new" products is an attempt by the

firm to obtain a differential advantage by means of product

differentiation over their competition. However, some

t171<3ught should be given to the question of what is a "new"
product.

Wasson has noted that a product can be classified

as "new" in at least thirteen possible ways. Each of these

w . . . . .
aYs Will have some effect, either positive or negative, on

t
he introduction strategy chosen by the firm. Robertson

hag defined "new" innovations in three ways, depending upon
11:3 effects on established patterns of consumer behavior.
3112-‘-ze11 and Nourse have defined new products as the products
abQ looked upon by the processor and distributor. Yet none
Qs these definitions have become universally accepted so

1:
11§t the term new product means the same to all readers.

1

  
 
  
 
   

Nevertheles'
a'zariety 0f “new 0
anon for this beta
zit-.Lderson. Alder
afim seeks power
:3 resolve negotia'
:zthe firm‘s favo
mint difterentf
jasiict physicall
repute it from at
aSteptable met‘no
though conmunic
eryironnent; (3‘
53598 of satisf

31"
- ‘Ce and terms

2
American firms continue to introduce

Nevertheless,
A possible explan-

a variety of "new offerings" each year.
ation for this behavior can be found in an interpretation

of Alderson. Alderson has been interpreted as saying that
a firm seeks power through product differentiation in order

to resolve negotiations between the firm and the consumer
in the firm's favor. One of the most common methods of
product differentiation is by means of differentiating the

product physically from all competitive offerings so as to
Other

remove it from any margin of perfect substitution.
acceptable methods include (1) psychological differentiation

(2) differentiation in the purchase

through communicat ion;
(3) differentiation in after—purchase assur-

enVironment:
ances of satisfaction in use; and (4) differentiation in

price and terms of sale.5
It is through these power—seeking, differentiation

a - . . . -
cthItleS, that the firm seeks to reduce risks and create

a . .
Preference among purchasing units that is suffIClently

s
t1"brig to withstand the efforts of competing firms. The
P .
rg fits a firm earns are in part a payment for the risk

5.
r1"'leed in the firm's efforts to achieve this power, and

i
11 part to guide a firm to other than low risk alternatives.

8
M [[935 quotes a Chicago industrial designer as saying

t
hQ se risks are causing companies to face "situations they

 

:e'.'er dreamed of be

Today‘s :15-
;rsi'ict innovation
mansion. Each de-

frtr products not ;

J
*1

m
(3
()

ount for a
“sass National Pr 3-”

:Y‘ “ta

.u.“ 5

Whats 5

resulted from prod;

 

55:11.18? e

that the

 

 

never dreamed of before."6

Today's American economy has reached a stage where

product innovation has become a major factor of economic

expansion. Each decade brings a higher percentage of sales

from products not in production in the previous decade.
This statement is most pronounced for those industries

which account for a growing prOportion of the United States'

Gross National Product. (See Figure 1-1) . A study by

Printers' Ink has shown that 43 per cent of 1957 gross sales

resulted from products not in production in 1947 and has
estimated that the figure has increased to 56 per cent for

1966 using 1956 as a base year and 62 per cent for products

Offered during the decade ending in 1971.8 Of the companies

with the highest growth rate for the past several years. over

50 per cent of their sales have come from products introduced

during the previous decade while only 10 per cent of the

s . .
ales of low growth firms came from this source. Of speCial

i .
n1:3rest is the fact that the majority of high-growth com—

p - . . . . . . . .
anles achieved divers1fication by acquiring other companies

1 I
r1 addition to internal developments. The situation is

f
u): hher verified when one considers the results of a 1965

s
tufly of United States firms. Of 742 firms studied in the

S i
3": months starting January 1, 1965, 522 firms launched some

1 o 2 . 1
36 new products in that period.

 

 

 

_ . '1 -'|| I ll a '1. I ‘ . - n . ‘
0:0... _ o. 5.... 9.5.: co: 0 OCC 9 <
:;..o ._...< $62. 93.3 k

ads-27.: .3 or:
QNEFZOO QWZZle

 

 

\

.Roonmoma ..,..C.SOCU mu-_<m Ou. MAODQOEL \SmZ to 20:3.

 

 

 

 

 

 

.m .a .82: £3» .65
.couHHEmm can coHH<INoom

 

muoscoum 3oz mo usoEommsmz ..ocH “condom

 

j

N E: 3232 1 573568 38» .23: seesaw

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

F

 

_ _ H ages: sage.
803%...nt WIN“ is: e: :2:
2269a .32 DE 5%:
“sebum” TM 71.3% 2.8.... $23.22
~561an «3e a :3 £35
FAN}.-- \ , SEE
h 2% . mafia:
ml H - 3:52:
W l .s! :1 .§ 23:53 .
1 1:! \N\\\\\N\\\ Seas; 33.58:
— p i\\.\\.\.\\VV\\ . we: a was: .83
7 _ i J..\\\.W\\s\\\\\\thsx\\\\\ as
80.2 88.2 83 83 80..” o
32.8 .o 22.55 $5.52. $43

Romaimmma ._...5..o~._o mmfi<m O... 953.com; >52 ”.0 29.595200 9.2qu

I see:

\

 

V

1968 fiscal year “

cent of this cost

The C051 C

a

1.". the market pla:

 

2::s eliminated ir
5;;31: of ten dense}
in; on projects t‘:
:ercial usefulnes
However ,
Tore unsuccessful
the"

a
,-
Hub

MS 0f the tart

any e
xaggerated

231:39
W On what

“Ch a failure

91:1 l

.I Ce
nt of
the
kite st
kuan b
teak
NEVErth

5

The cost of this new product innovation for the

1968 fiscal year amounted to over $250 billion, or over 25
11

per cent of the Gross National Product. Yet, over 70 per

cent of this cost went to products that were not successful

in the market place, in fact over two-thirds went to prod-

‘ucts eliminated in the develOpment stage. Thus, about

(eight of ten develOpment engineers may be said to be work-

irng on projects that will not be justified in terms of com—

mercial usefulness. (Basic research is not included here.)

However, in spite of the increasing efforts to re-
rnc>xre unsuccessful products in the develOpment stages before

‘tIleay reach the market place, many new products fail when

'tlieajr are finally introduced. (It should be noted that esti-

‘naltzeas of the rate of new product failures are almost invari-

ably exaggerated. The actual rate of failure depends. of

(:CD‘IJEfise, on what products are included in the base against
which a failure rate is computed, as well as on the criteria
e»

“RE>:l.oyed to identify failures. Weiss has found that over

8 c
C) I;>er cent of new products are not "new" but "simply modi-

13

f.
J“C=E3Ltions" of existing products. ) Booz—Allen and Hamilton

5ft:
:E>‘=>rt that in a survey of all industry groups only 62.5
1: cent of the products presented to the consumer will do

ugh o u
see than break-even over their first three years of sales.

Nevertheless, compare the above figure to an

1,!

analysis by the P:

 

I
eluded that 19 of -

. 15
all: a 1972 EU-
saips. food snack:

11‘. the 1970' s rou

1“” . .
.". .ert'c “N ‘r-v .'.
gvlm‘l fine ‘15.

 

sipernarket produ<

Pals 17

A more co
new products can
on the food indus
PIDiucts they exe

if:e
it test marke1

42 3
.er cent were

6
analysis by the American Management Association which con-
cluded that 19 of 20 new products could be expected to

fail,15 a 1972 Business Week prediction that of the 120,000

 

soaps, food snacks and other supermarket products introduced
in the 1970's roughly 10,000 will bomb out16 or a report of
Advertising Age which predicted that 80 per cent of new
supermarket products will fail as they will not meet sales
goals.17
A more conservative estimate of the failure rate of
new products can be found in a study by Buzzell and Nourse
on the food industry. Of the 127 distinctly new food
products they examined, 39 per cent were discontinued either
after test marketing or after regular introduction. Also,
42 per cent were classified by their sponsors as either ex-
tremely unsuccessful or moderately unsuccessful. Another
criterion of product performance is the length of time
required for the contribution to profit earned from a prod-
uct to offset its develOpment and introduction costs. By
this criterion, 44 per cent of the products failed to break
even after two years of regular distribution.18
In summary it can be stated that advancing tech-
nology and increasing research and development give no

assurance that new products will have a high probability of

success. It is not uncommon to find studies which reveal

 

the new product 1
cent.

Rescurce
development Pres‘
notability is 8<
framed with pro<
threaten to limi1
consumer spending
firm: can take c<
an to one. The
linited the amour
Winter. Yet, i:
55:th the cons:
"ill also suffer

34336 needs .

seen that the fir
not least keep

develenient

7
the new product failure rate to be between 50 and 90 per
cent.

Resource allocation for new product research and
development presents a problem for the firm when the failure
probability is so high. While industry is constantly con-
fronted with product obsolescence as new develOpments
‘threaten to limit the market life of existing goods, and
cnonsumer spending patterns undergo constant shifts, few

fiirms can take comfort even in a success—failure average of
izvvo to one. The profit squeeze of today's economy has
limited the amount of financial set-backs a firm can en-
counter. Yet, if a firm doesn't seek out new products to
Satisfy the consumer's ever-changing wants and desires, it

‘A?j_ll also suffer financial disaster as competitors fulfill

these needs .

Statement of the Problem
In view of the above research findings, it can be
Seen that the firm faces a high risk situation if it fails
't:‘=> at least keep pace with its competitors in new product
<3tevelopment. Similarly, the firm experiences a high finan-
<==iiwal risk every time it undertakes to introduce a new prod—
‘k1‘==t in the market place. If one subscribes to the theory

that the ultimate objective of a firm is that of survival,

 

{ten the firm is f.

. ‘ . ‘.
:ntrozuctlon. Th-

 

 

 

zage presents only.
2:: duce the highl

fin. could re-exa:

 

I .I
. 1
f”
-l

the first sta
final stage of tes
seek out newer res
able marketing dec
No longer

guards predictin

SthSS. Today t

:15 r .

\

flfiv,

y” 3‘-
I

.1‘9
1-! S tage .

or the 1

532:5
35 Or failU]

-«

”559 are:

l.

8
then the firm is forced into a condition of new product
introduction. This means of seeking a differential advan-
tage presents only two alternatives for the firm in order
to reduce the high financial risks. First of all, the
firm could re-examine its internal develOpmental process
from the first stage of determining firm objectives to the
final stage of test marketing. Secondly, the firm could
seek out newer research techniques in order to make profit—
able marketing decisions during the introduction process.
No longer can the firm take the leisurely approach
towards predicting product success during the introduction
process. Today there is a dollar premium on time which is
greater in the first month of a new product's life than at
any other stage. Crawford specifically listed six important
reasons for the firm to rapidly determine a new product's
success or failure probabilities immediately after launch.
These are:
l. The attention span of consumers in the market
place is short.
2. Changes can be made rapidly thanks to the
speed of mass communication.
3. The size of the initial investment grows with
each new product to the point where the launch

of important new items is backed by dollar

9

budgets which strain the resources of Oper—

ating units.
4. In any creative, market oriented company,

Opportunity cost decisions abound.
5. TOp management isn't known for patience.
6. Finally, launch of a new product signals the

start of many ongoing problems throughout the
corporation.19

This study will concern itself with the introduction
of new food products because nowhere is the problem of new
product introduction more prevalent than in the retail food
industry. The ex-Secretary of Agriculture, Orville Freeman,
stated, "Each year about 5,000 new food products are offered
to stores that already carry 8,000 different items."20
Business Week in a 1972 survey of new product marketing
problems found that the three year payout is some eighteen
months too long. During the last ten years, as new brands
introduction more than doubled in the frozen-food and dry—
grocery business, average product "1ife eXpectancies" fell
from 36 months to 12 months.21
Thus, one can see that today's supermarket is faced

with the problem of selecting those products which offer

the potential of success and eliminating those with limited

possibilities. This thesis will attempt to provide a

;;;ie‘;ine for the
I
:cbe retained. ".

sales analysis, in

:13: are examined

we're information.

.333: of a three

cs '
AW

1“»
.L..‘-§“"fl

”Messful prs
..:.a<e his decisi

331‘“: ‘

Bing tool whi

 

 

cor»: -
...itution to th
115:, h .5
.uutor as ‘W'nl
In View oc
3f 0.

10
guideline for the store manager in his selection of products
to be retained. The current supermarket method includes
sales analysis, in which the sales of the products in ques—
tion are examined over a three month introduction period be-
fore a final decision is reached. However, in view of the
above information, the manager can no longer afford the
luxury of a three month trial program. His low Operating
margin will not permit him to use valuable shelf Space on a
non-successful product. Today's manager needs to be able
to make his decision at a much earlier point in time. Any
marketing tool which could be used at this stage would be a
contribution to the individual manager and likewise to the
distributor as well as producer.

In view of these above reasons the stated purpose
of this research undertaking will be to provide a means of
predicting which food products should be eliminated from
the firm's product line at a point in time far earlier than
the usual analysis of thirteen weeks sales data. Instead
of reviewing initial three months sales results, it is
hoPed that the prOposed method will enable the retail food
manager to make a profitable prediction on the basis of
consumer behavior patterns during the initial 5 weeks of

introduction.

ll

Implications for MarketingProblems

 

A great deal of research has been conducted by mar-
keting scholars and professionals concerning the way new
products are conceived and progress through different
stages of the introduction process. These may be classified
into two analytical frameworks. New sales information can
be thought of in terms of diffusion processes or adOption
processes.

Diffusion process is the name given to the process
by which "new ideas are communicated to the members of the

social system."23

To model a diffusion process, one works
with a few macr0parameters that will locate a curve that
describes the path of the innovation over time. Some of
the major diffusion models which have been develOped by
those who have worked on the first purchase forecasting
problem include concave models, S-curve models, epidemio~
logical models and reliability engineering models.24
The adOption process, however, focuses on the

"mental process through which an individual passes from
first knowledge about an innovation to a decision to adoPt

"25 Adop-

or reject and to a confirmation of that decision.
tion itself is the act of buying the product in the case
of nonrepurchasable products, or the decision to use the

product regularly, in the case of repurchasable products.

12
These models require a more behavioral and detailed render-
ing of the individual's process of moving toward the trial
and use of a new product. Here some common models are the
DEMON model, urban's model and Alba's model.26
Other works in this area include Lavidge and

27

Steiner's examination of the use of advertising for pre-

28

dicting effectiveness, and Bader, as well as Britt and

Lucas,29 who examined the use of point of purchase displays
for existing products. Yet none of these models have pre-
sented the retail store manager with a useful tool for pre-
dicting product success at an early stage in the product
introduction phase. All have used sales as the measurable
variable for predicting success. The closest anyone has
come to predicting success without waiting for complete
sales returns has been Crawford. Crawford used a trajectory
projection to predict sales results. However, these trajec-
tory curves, which were based on consumers' awareness and
knowledge of the new product, percentage of repeat sales
and dealers' promotional activities, were obtained over a
long time period and at great expense, something most re-
tailers would be unable to afford.30

In an attempt to provide the retail store manager

with a useful tool for predicting product performance at an

early stage in the introduction process, this thesis

grapcses a model ;
t3 failure. The :
2:.1‘91, which will
tern “new product

:ew product inforr

”newness" of the 3

 

each of the dimen:

 

 

neaness as percei‘
1:.gc'r.annel: the 1

zit-w
J ~uers u The In

are level of newt“:
czaracteristics .
“c? «S ' '

. . t.a351ficatic
;::duc‘

~5 in comp;

:ETel
‘5 0f newnes

7163: D

l3
pr0poses a model for predicting which products are doomed
to failure. The model is a two phase product classification
model, which will be used to define what is meant by the
term "new product" and to provide a means for analyzing
new product information. The first phase classifies the
"newness" of the product in a three dimensional matrix with
each of the dimensions reflecting the different levels of
newness as perceived by the different members of the market-
ing channel; the producers, the middlemen and the ultimate
consumers. The model is further structured by combining
the level of newness with an analysis of the product's
characteristics. Miracle's31 revision of Aspinwall's prod-
uct classification is used to relate the behavior of new
products in comparison with other new products of similar
levels of newness and product classification.

By combining these product characteristics with the
new product classifications and then measuring the levels
of activity of the stages of the adOption process, it is
felt that a contribution can be made for determining which
new products should be eliminated from a firm's total
product line at a point earlier in time than is presently
the case. It is to be noted that while this model does not
Imake use of the repeat purchase patterns of the consumers

111 an attempt to predict product retention, it does serve as

 

areans for predi'
anagenent. Thus

elizination from

ices not necessar.

 

izes, however, (3:31
rinate those prod
far continuation.
:r. the fact that

LIA ' - .
.e a Signif ica:

14

a means for predicting which products will be retained by
management. Thus, the fact that the model does not predict
elimination from the line for the product in question, it
does not necessarily predict success for that product. It
does, however, guide management in their attempts to eli—
minate those products which will have a very low probability
for continuation. The low probability for success is based
on the fact that the eliminated products will never experi-

ence a significant level of first time purchasers.

Conceptual Framework

 

The conceptual framework for this study involves the
selection of various phases of the adOption process which
can be used to differentiate between those new products
which management will either retain or drop at a later date.
It is hypothesized that by measuring the level of activity
for the initial six phases of the adoption process presented
in this thesis, a significant reduction in the time needed
to evaluate the retention of a new product can be achieved.
The general retailing practice is use today is for retail
food management to analyze sales data for the initial thir—
teen weeks and then make a judgment as to continuation or
discontinuation on the basis of analysis of sales trends,

ROI or profit.

This thee;
:he six predictor

sessing a common 1

 

 

it. an attempt to F
tween the two groi
can serve as
instions to their
if product groupir.

«as of their thi

mm this analysis

 

will enable

,
inc ‘
a

‘7‘
sure product

.. reasurements ta

8 Keri
~.ES
(X1, X2, . .
on a

bum n .
The resul
Cen‘

15

This thesis will measure the level of activity for
the six predictor variables from a group of products pos-
sessing a common level of newness and product classification
in an attempt to see if a significant difference exists be—
tween the two groups, the retained and drOpped products,
which can serve as a tool for assigning future product intro-
ductions to their correct grouping. The initial selection
of product grouping will be made by the management on the
basis of their thirteen week analysis of sales results.
From this analysis, a discriminate function will be derived
which will enable the management to classify both present
and future products to their prOper grouping on the bases
of measurements taken before significant sales results are
available.

An illustration of the preposed is as follows:
Given that we were able to measure the levels of six vari—
ables (xl, x2....x6) during the fifth week fater introduc-
tion and mark them on a graph with x's representing con-
tinued products and 0's representing discontinued products,
as classified by management at the end of thirteen weeks as
shown in Figure l-2.

The resultant ellipses in Figure 1-2 enclose some
Specified pr0portion of the points, say 98 per cent in each

Product group. A straight line is drawn through the two

 

ILLUSTRAT

____
- ______
mt

,

.
‘5.“

 

 

 

16

FIGURE 1-2

ILLUSTRATION 0F.A DISCRIMINATE FUNCTION

 

 

 

points where the ellipses intersect and then projected to
a new axis Z. The axis Z condenses the information about
group separability into a set of points on a single axis.
The axis Z is the discriminate function in this illustra-
tion and it can be used to predict correctly the product
grouping for present as well as future products on the
basis of measurements of the activity occurring in the
early phases of the adOption process in the initial five
weeks of product introduction.

The information shown in Figure 1-2 can also be
expressed in terms of an equation were x1, x2,...,x6 repre-
sent the six predictor variables. In such a case, a numeri-

cal value can be determined for each of the products under

:uesrion and the
I

:5 which group me
."al'ie.
Thus, the
:easurenents of t

it be analyzed a:

seeks with regard

 

tension can be c:
ezgnt weeks later
lgpothesized that

25:] 3
‘

IOdUCt Signi

17
question and the products assigned to a group on the basis
of which group mean value is closest to the new product's
value.

Thus, the stated objective of this study is that
measurements of the initial stages of the adOption process
can be analyzed and a decision can be made at end of five
weeks with regard to the continuation or rejection of the
product from the line if thirteen weeks sales results were
available. For purposes of verification, the fifth week
decision can be compared with the actual management decision
eight weeks later to determine the assignment error. It is

hypothesized that such an analysis of the adoption process

can product significant results.

Methodology

The data to be used for this study will be obtained
by means of a phone survey of supermarket shOppers in the
Des Moines, Iowa market. One supermarket chain whose retail
sales accounted for 26 per cent of Des Moines' 1971 retail
grocery sales c00perated with this study. The choice of
Des Moines as well as this chain was fortuitous. Any
analysis of the results of this study with the total United
States population can only be made insofar as the consumers

of the above chain in Des Moines, Iowa are representative of

18
the United States market.

The proposed research design involved the selection
of seven products from a list of products the store manage-
ment was planning to introduce within the next three months.
This was the maximum number of products available for the
study since basic criteria for the product selection were
that all the products were defined as having the same level
of newness and that each fit into a similar classification
in Miracle's groupings. After selecting the products a
phone survey of the store's customers was conducted to
determine the amount of customer awareness, knowledge,
interest (strong and weak) and information seeking activi-
ties (strong and weak) with regard to the new products at
the end of the second, third, fourth and fifth week after
introduction.

Linear discriminate analysis was used to test if
the weekly measurements of activity of the six predictor
characteristics chosen were able to differentiate between
the continued and discontinued products. The rate of
growth hypotheses were tested by means of eighteen inde-
pendent t-tests. These eighteen tests, each with the
standard assumptions of normality, equality of variance
and independence, were used to analyze the data to deter-

mine if there was a significant difference in the means of

 

:he continued an:

:redictor variab‘.

 

titrd week period
tie-way analyses
:ate of growth be

.- Coe six predic

:lie t~tests,

The the $1

the data were col

data COUld ha‘le b

are Compared to t1

$97.72 hax'ye been 0}

 

 

3'u'~ ‘
”like to the 1

lect‘.

SUIn

t was admittex

“hie, WhiCh may he

3‘ L'

39 n99ded t0

19
the continued and discontinued products for each of the six
predictor variables for the third—second week period, fourth-
third week period and the fifth-fourth week period. Also,
two-way analyses of variance were performed on the mean
rate of growth between the continued and discontinued groups
by the six predictor variables for the same time periods as

the t-tests.

Limitations of the Study

 

The thesis has several limitations. First of all,
the data were collected during one time period. If the
data could have been collected over different time periods,
and compared to these different periods, different results
might have been obtained. However, because of the resources
available to the researcher, the idea of a longitudinal col-
lection process was not considered. The five week cut-off
point was admittedly an arbitrary decision. This point in
time, which may need readjusting in future studies, was
chosen so as to make a significant reduction in the amount
of time needed to make the continuation decision.

Due to time and expense constraints the researcher
was able to study the consumer purchasing behavior for one
particular product grouping at one retail location. This

selection of product grouping and location also provided no

 

26335 IO:

Q

- e
“Yo." 1"
"Nun“. -

- .
eve.

..:..'lg pc

i y D ‘
1N1 6‘ ‘
«he: _‘S‘

“PA ‘_
:‘VduCt (

uufl'n. “v
M: ‘ {Jo
F .

5:?fir;

20
means for measuring the advertising effectiveness of the
products due to the chosen retail store's established adver-
tising policies. It can be noted that the absence of store
advertising might not be a limitation, since the advertising
of the manufacturer could be considered to be a part of the
product offered for sale. Nevertheless, these limitations
might produce some variance in the thesis results. Every
effort, however, was made to have customer selection by day
and hour representative of the total store p0pulation and
all findings were considered to be valid only for the par—
ticular store studied. Another possible limitation is that
this study assumed that there was a common homogeneity be-
tween the households selected for interviewing for each of
the four different weeks. It is further noted that a final
limitation of the thesis was the fact that only a limited
number of new products of the same grouping were available

for study during the time of this study.

Some Possible Contributions of the Study
The major contribution of this study was that by the
use of a product classification systems, which considered
both the level of newness and the product's consumer and
market characteristics, a useful tool was develOped which

could enable one particular store to predict the

 

.
arty?! ‘u‘fl
‘ I'...c

.5}...

i F
122131 5.

J

.2 ‘Q
.e res“
-M'- s
y‘vClC. l

‘ »
:‘Rn. ‘ n
nth.‘Un

~

21
continuation or discontinuation of a product before mean-
ingful sales results were available.
The thesis further provided a basis for examining
the results of combining three marketing tools: new
product dimensions, product-market classifications and

adOption studies into a useful research technique.

9—4
.

Cl)

\‘1

|
‘k.

"New P
Week,

Cheste

Journa
.“

1‘ -
'. 10.2135

ENC“:

Ԥ-Q.

(Janus

Robert
Inner;
\
arva:
P- 2

Thoma

1m

Pren:

10.

11.

12.

13.

14.

15.

22

FOOTNOTES

"New Products: The Push is on Marketing," Business
Week, March 4, 1972, p. 72.

 

Chester R. Wasson, "What is "New" about a New Product?"
Journal of Marketing, XXV (July, 1960), 52-56.

Thomas S. Rdbertson, "The Process of Innovation and the
Diffusion of Innovation," Journal of Marketing, XXXI
(January, 1967), 14-19.

 

Rdbert D. Buzzell and Rdbert C. M. Nourse, Product
Innovation in Food Processing: 1954-1964 (Boston:
Harvard university, Division of Research, 1967),
pp. 22-31.

Thomas A. Staudt and Donald A. Taylor, A Managerial
Introduction to Marketing (Englewood Cliffs, N.J.:
Prentice-Hall, 1970), pp. 12-13.

 

Business week, March 4, 1972, p. 72.

Booz-Allen and Hamilton, Inc., Management of New
Products (New Ybrk, 1968), pp. 5-6.

 

"New Product Revolution," Printers' Ink, October 31,
1958, 21-27.

 

Boos-Allen and Hamilton, pp. 14-15.

Press Release, May 1965, Sales and Marketing Executives
International, New YOrk.

Daniel D. Roman, Research and Development Management:
The Economics and Administration of Technology (New
York: Appleton Century-Crofts, 1968), p. 35.

 

Booz-Allen and Hamilton, pp. 10—11.

E. B. Weiss, "That Malarky About 80 Percent of New
Products Failing," Advertising A99. XXXVI (August 2,
1965), 101.

 

Booz-Allen and Hamilton, pp. 10-11.

Philip Marvin, "Why New Products Fail," in Thomas Berg

l6.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

23

and Abe Shuchman, Product Strategy and Management
(New YOrk: Holt, Rinehard and Winston, 1963), p. 351.

Business week, March 4, 1972, p. 73.

Theodore Angelus, "Why Do Most New Products Fail,"
Advertisipg Age, XL (March 24, 1969), 85.

Buzzell and Nourse, pp. 169-170.
C. Merle Crawford, "The Trajectory Theory of Goal

Setting for New Products," Journal of Marketing
Research, III (May, 1966), 117.

 

 

Orville Freeman, This Week, JUne 26, 1966, 2.

 

Business Week, March 4, 1972, p. 7.

 

It should be noted that two models (the Demon and the
Ayer) have already been prOposed for predicting product
performance before the product introduction stage
occurs. See Henry J. Claycamp and Lucien E. Liddy,
"Prediction of New Product Performance: An Analytical
Approach," Journal of Marketing Research, VI (November,
1969), 414-420 and D. B. Learner, ”Profit Maximization
through New-Product Marketing, Planning and Control,"
in Frank M. Bass, Charles W. King and Edgar A.
Pessemier, editors, Applications of the Sciences in
Marketing Management (New Ybrk: John Wiley and Sons,
Inc., 1968), pp. 151-167.

 

Everett M. Rogers with F. Floyd Shoemaker, Communica-
tion of Innovations (New YOrk: The Free Press, 1971),
p. 99.

 

Philip Kotler, Marketing Decision Making (New YOrk:
Holt, Rinehart and Winston, Inc., 1971), pp. 524—536.

 

Rogers with Shoemaker, p. 7.
Robert J. Lavidge and Gary A. Steiner, "A Model for
Predictive Measurements of Advertising Effectiveness,"

Journal of Marketing, XXV (Octdber, 1961), 59-62.

Samuel Bader, "The Sign: Its Power Ever Rises,"

Printe
i3 Darrel

F‘fect

cranI
l‘J.
1966)

‘ Sordo
.n '
Stra:

18-24

 

 

29.

30.

31.

24
Printers' Ink, May 29, 1964, 135.

Darrell Lucas and Stewart Britt, Measuring Advertising
Effectiveness (New Ybrk: McGraw-Hill, 1963).

 

Crawford, JOurnal of Marketing Research, III (May,
1966).

Gordon Miracle, "Product Characteristics and Marketing
Strategy," Journal of Marketing, XXIX (January, 1965),
18-24.

 

 

a oeoret
"in”: .-
H“ " U: s.

f“,

CHAPTER II
LITERATURE REVIEW

The primary objective of Chapter II is to establish
a theoretical background for the research by reviewing the
relevant literature. Chapter II is divided into four sec-
tions: AdOption Process, Diffusion Process, Innovation
Characteristics, and Product Elimination Studies.

No review of the literature in this area could ever

be made without use of Everett M. Rogers and his BibliOg-

 

raphy on the Diffusion of Innovation.1 As of July, 1968,
when the last bibliography was compiled, Rogers had
gathered, analyzed and summarized 1,084 diffusion and adOp-
tion studies. Some sixty-four of these studies were classi-
fied as marketing studies. It should be noted that only

six years earlier Rogers rejected Katz and LeVin's classi-
fication of marketing as a research tradition because as

he stated, "there are relatively few research studies avail-

able in this field (marketing)."2

Also, the work of Thomas
S. Rdbertson who has been most active in relating the

studies in the area of diffusion and adOption to basic

25

26
marketing concepts was extremely useful. Thus, the most
recent texts of Rogers and Robertson3 will serve as bench

marks for this chapter.

The AdOption Process

 

This review of the literature begins by studying
the adoption process and then going to the diffusion pro-
cess. This is the reverse of the traditional case because
of the nature of the research being conducted. In this
study the stages of the adOption process are used as the
predictive variables. Also, it can be argued that the
adeption process actually takes place before the social
system completes the diffusion process.

Rogers and Shoemaker have defined the adeption
process as "the mental process through which an individual
passes from first knowledge of an innovation to a decision
to adapt or reject and to a confirmation of that decision."
This is a revision of Rogers' earlier definition of the
process as it allows for the rejection of the innovation
under question.5 The adeption process should be distin—
guished from the diffusion process by which "new ideas are
communicated to the members of the social system."6 Thus
diffusion occurs within a social system, while adoption

takes place within an individual.

‘I

‘u

at.

«\~
‘4‘

N4
\..

e

27

Several adOption models have been prOposed. The
one common bond between them is their conceptualization of
a flow Of events rather than an "instant metamorphosis".
These models prOpose a series Of "stages" to represent the
sequence of mental and/or behavioral events assumed to be
antecedents of adOption. However, there is a great deal of
differences in the number of stages and in the nature of
these stages. Thus, the reader should realize that all
these models are conceptual frameworks only and that their
value must be determined by how well they Operate in a
given situation. This section of Chapter II will attempt
to review the major models pertaining to the adOption Of
new innovations in their chronological order. It should
also be noted that the innovation as used in these adoption
studies will refer to any product or idea which is perceived
as new by the individual in question. This may be either
a product for which a new use has been develOped, which has
undergone minor changes or is a completely new Offering

with no direct substitute.

Ryan and Gross Hybrid Corn Study
The landmark study of the adOption process is the
1943 study of the introduction of hYbrid corn in Iowa. The

innovation of hybrid corn was the result of years Of

 

.!
I

Am1

 

lIlCGSSJ

Pica

.:Eeri

28
intensive research by agriculture scientists. The hybrid
vigor of the seed did not continue in the second generation,
so farmers had to purchase hYbrid seed each year, whereas
previously they had selected their own Open-pollinated seed.

The major advantage of the innovation was a 20% increase in

yield. While the new corn was first made available in 1928, TI
it wasn't completely adOpted by Iowa farmers until 1941.7 E‘

Ryan and Gross found that the first use of the rag
hybrid seed followed a bell—shaped (but not exactly a

 

normal) distribution when plotted over time. They also
noted that the process contained at least three stages:
awareness, or first hearing of the new idea; trial, or
first use; and adOption, or complete use of the new seed.
The average time for this process was nine years. A final
finding of their study was that the typical farmer first
heard of the new seed from a salesman, but neighbors were
the most influential source in leading to adOption.
Rogers has pointed out several criticisms Of this
pioneering effort. Two of the criticisms are:
1. Ryan and Gross made no analysis of Opinion
leadership in the process.
2. They defined "acceptance" as first use of
the seed (the trial stage), and largely

ignored the adOption (complete use of the

 

long bee:
area. I

"N994. ‘*
BVu‘;‘~ L

5; ~ .
..r.ec tn

29
new product) dimension in the data they

had gathered.8

Five—Stage Ad0ption Process Model

Yet, in spite of these shortcomings, this study has
long been regarded as the forerunner Of all others in this
area. In 1955, when the North Central Rural Sociology sub—
committee for the Study of Diffusion of Farm Practices de—
fined the stages Of the adOption process, they relied
heavily upon this work and that Of Wilkening.9 The five
stage model adapted by the committee was:

1. Awareness: The individual learns Of the
existence Of the new idea but lacks infor—
mation about it.

2. Interest: The individual develOps interest
in the innovation and seeks additional infor-
mation about it.

3. Evaluation: The individual makes mental appli-
cation Of the new idea to his present and
anticipated future situation and decides
whether or not to try it.

4. Trial: The individual actuall applies the
new idea on a small scale in order to deter-

mine its utility in his own situation.

5. Adoption: The individual uses the new idea
on a full scale.10

However, this conceptualization has been subject
to criticism in recent years. First of all, the process

seldom ends with adOption as the individual is exposed to

 

51553.13"1c
latest t!
”marthtvv

;t as th

renamed '

S ' .
ecs in .
:s possr:

In 9

2e furtnc

. ‘, .
"’4‘." 1'
Jiu‘vld‘h

eiF

.ects
Steiner
We ha

and Cona‘

30

dissonance in evaluating his behavior.11 McCarthy in his
latest text has listed a sixth stage to the above model.
McCarthy calls this final stage confirmation, and defines
it as the process of seeking reinforcement. He, also, has
renamed the fifth stage, decision.12

Another criticism is that the above model always
ends in adoption decisions. RObertson points Out that it
is possible for the individual to reject the innovation.
He further points out that no provision is made for the
individual to skip stages, nor are any feedback 100ps pro-

vided.13

Thus, this model fails to account for non-adOp-
tions, impulse buying behavior and feedback of past eXperi-

ences which can affect present behavior.

"Hierarchy of Effects" Model

Marketing researchers have prOposed a hierarchy—Of-
effects model as a result of a 1961 study by Lavidge and
Steiner.l4 This model consists Of the six steps in the
purchase process, which they found to be related to the
three basic psychological states: cognitive, affective,
and conative. Their six stages were:

1. Awareness: The individual is aware of the
product's existence.

2. Knowledge: The individual knows "what the
product has to offer".

“It.

31

3. Liking: The individual has favorable attitudes
toward the product.

4. Preference: The individual's favorable atti—
tudes have "developed to the point
of preference over all other possi-
bilities".

5. Conviction: Preference is coupled "with a
desire to buy" and confidence "that
the purchase would be wise".

6. Purchase: "Attitude" is translated into
actual buying behavior.

Lavidge and Steiner have stated that since time and
difficulty involved in each stage depends upon both product
and consumer characteristics, the stages are not necessarily
equidistant. However, a buyer may sometimes go through
several stages simultaneously (an impulse purchase).
RObertson has pointed out this model, also, made no effort
to use feedback behavior.15 Nevertheless, this model was
the first to explicitly rely on an information—attitude—

behavior theory Of communication effect.

:AIDA Model"

 

Another marketing effort to define the process is
the "action-oriented framewor " called AIDA. This model
which was first prOposed in personal selling texts consists
of four fundamental tasks which have been recognized for
many years: (1) to get attention, (2) to hold interest,

C3) to arouse desire, (4) to Obtain actigp. The

29.3th

 

. .
If", -
s... '.
\ 1

‘PL, ‘

M...“

.
!-:.
hn‘v‘.
‘

a.
he.

"a:

 

 

-
\
\

A I

p);

32

relationship of these models is shown in Figure 2-1.

FIGURE 2-1

RELATIONSHIP OF THE THREE ADOPTION
PROCESS MODELS HIERARCHY

 

 

Hierarchy
Of Effect AIDA McCarthy
Purchase Confirmation
Behavioral Level Decision
Conviction Trial
Desire
Preference Evaluation
Attitude Level
Liking
Interest Interest
Knowledge
Information
Awareness Attention Attention

Critique of Adoption Process Models

 

Robertson in his text, Innovative Behavior and Com—

 

munication summarizes several critiques Of these models.

 

He repeats Mason's16 challenge that five discrete stages
(Mason was referring to the basic rural sociology model,
but this critique will apply to all three models) are nec—
essary to account for adoption. Mason found several forms
of the process occurring and no single process amounted for

all forms of behavior. He did find that only two stages

33

are actually necessary: awareness and adoption (or rejec—
tion as supported by later studies of Rogersl7), with
awareness always occurring first. Robertson, also, notes
that Mason found that evaluation apparently occurred before
interest and that adOption was not the terminal stage, but
was followed by interest and information seeking.18 This
finding appears to be in agreement with those who hold to
the theory Of cognitive dissonance, mentioned earlier in
this chapter, in that some purchasers seek out information
to evaluate and reconfirm their previous purchase behavior.

RObertson, therefore, has concluded that a consist-
ent adOption process conceptualization seems unlikely across
any range of consumer goods. For inexpensive, low-risk
products, deliberateness in purchase may not be as necessary
as for expensive high-risk products.19 Important differ—
ences also seem to exist in the process for continuous as
opposed to discontinuous innovations.20

RObertson continues his critique by reviewing Camp-
bell's two contentions that first of all the process may
not be rational and secondly, that it might not start at
awareness. Campbell has sought to show that the previous
models, which only pertained to rational buying situations,
failed to consider a situation where the perception of a

problem could be the first stage. As has been noted,

ll

I‘e"er

Ink-ll

beginni

“in-

example

freow

‘AVAA

.
0
.33 Fr;
In!
‘ \

CF“:

Q..

'4-

In

34
whether consumers are relatively active or passive at the
beginning stages probably depends on the situation, for
example, whether the products involved are expensive, in-
frequently purchased items or inexpensive, frequently pur-
chased ones.

Campbell advocates four adOption processes. These
four processes, which follow, are outgrowths of his two
criticisms of previous models.

1. Rational/problem solving: Here the consumer
becomes aware of a problem, looks for a solu-
tion, and carefully evaluates any product
which potentially solves his problems.

2. Rational/innovation: Here the consumer becomes
aware Of the product before the prOblem and
rationally judges it.

3. Nonrational/problem solving: Here the problem
is perceived in advance of product awareness,
but “in seeking a solution, the consumer
impulsively accepts the innovation without
careful consideration or evaluation".

4. Nonrational/innovation: Here the individual
sees something new and impulsively adOpts the
item without deliberation as to its utility.

The relationship Of these four models by Campbell

is shown in Figure 2-2.

.
::"".n 1
{Wt-fig"; A

a ' Q
v'fl" OM
‘. 4" -bnu
“

"2"; ‘
5‘ J~-€IT.

interes
.ik'ifeneS
Mimi

ReZectio
Rejec

+4

| Y‘
v‘ ‘.

.
Mora:

y
H.023
~‘v
H-,‘
O“‘:
5.. ,
.‘ ‘
.
N‘ ‘
4‘ L
the;
“V‘s

ll

35

FIGURE 2-2

RELATIONSHIP OF ADOPTION PROCESS STAGES
OF THE FOUR CAMPBELL MODELS:

 

 

 

Nonrational/
Rational/ Rational Problem Nonrational/
PrOblem Solving Innovation Solving_ Innovation
PrOblem or Awareness Problem or Awareness
interest interest
Awareness Interest
Evaluation Evaluation Adoption Re- AdOption Re-
jection jection
Rejection Trial Trial
Rejection AdOp- 2
tion AdOption Resolution Resolution

Innovation-Decision Model

 

Rogers in an attempt to respond to the above men-

tioned critiques prOposed a revision of the earlier adOption

process which he titled the "Innovation-Decision Process".

This model is the result of his effort to put some concep-

tual order as to the number Of stages in the process in view

of the many conflicting theoretic approaches to the process.

This proposed model consists Of four functions or stages:

1. Knowledge: The individual becomes exposed
to the innovation's existence and
gains some understanding of how
it functions.

2. Persuasions: The individual forms a favorable
or unfavorable attitude toward
the innovation.

36
3. Decision: The individual engages in activi-
ties which lead to a choice to
adOpt or reject the innovation.

4. Confirmation: The individual seeks reinforce-

ment for the innovation-decision
he has made, but he may reverse
his previous decision if exposed
to conflicting messages about the
innovation.

Rogers states that this model is designed to account
for the major criticisms raised about earlier models, to
profit from recent research, and to be consistent with the
learning process, theories of attitudes change and general
ideas about decision making. He, also, notes that the
knowledge stage is influenced by the: a) social system
norms, b) communication integration and, c) the tolerance
of deviancy. The persuasion stage is influenced by the in—
novation perceived characteristics of: a) relative advan—
tage, b) complexity, c) compatibility, d) trialability and,

22 (These characteristics will be dis-

e) Observability.
cussed in the third section of this chapter.) Rogers cites
a 1960 study by himself and Beal as the empirical evidence

for the validity Of this process.23

Marketing's Explanations of Consumer
Qpcision-MakingrProcess
The past decade has seen the introduction of three

major consumer behavior models in marketing literature as

~h

~vo- ~-
3 a.c..

Howard a

n q‘

r
“.u‘v)! ad
5..

...9 COT;

Edge 3

 

relatic
basic 1
itid c

v.“
‘* him a

i4). Cr
a

I ' p.

is’ihf\

KN“

37
an attempt to explain the dynamics of consumer decision-

making with regard to new product introduction: Andreasen,24

25 and Nicosia.26

Howard and Sheth
The basic assumptions of the Nicosia model (Figure
2-3) are that the firm is introducting a new brand and that
the consumer is unfamiliar with it. Nicosia used the tech-
nique of computer flow charting to designate elements and
relationships, and it should be noted that there are four
basic fields in Figure 2-3. It is explicitly assumed that
field one includes the output of an advertising message
from a business firm and that the consumer recipient was
not previously familiar with the product. As the message
(subfield one) reaches the consumer, it serves as an input
into subfield two, consumer attitude. As this message is
received and acted upon, the output hOpefully is formation
of an attitude toward the product, which then serves as the
input for field two. Field two represents a search for and
an evaluation of the advertised product and other available
alternatives as well. The Output from this field may or
may not be a motivation to buy the advertised brand. If
such a motivation emerges, it serves as the input for field
three-the transformation of motivation into purchasing
action. Finally, field four is storage or use Of the pur—

chased item, and the output is feedback of sales results to

38
FIGURE 2-3

THE NICOSIA CONCEPTUALIZATION OF THE
PURCHASE DECISION PROCESS

m We!" M A Summary Flow Chart
Roldan: Fro. “madam to tho Cam's Attitude

 

 

 

  
  
   

Sum.“ 01. Subfiold Iwo
Cam's
"""' MN" up, W. .5 Amtm pp) mm

 

(“fiddly
podiipaiflonl) Field Two: Such fat,
7 and Evoluaflon of,
50"" Moons-End (.) human to)
(Pro-action Field)

 

 

 

 

 

 

 

 

Evaluation

 

rem "M"
o! Purchac

 

 

 

 

 

 

 

 

"chafing
Octavia

Source: Francesco M. Nicosia, Consumer Decision Processes
(EngleWOOd Cliffs. N.J.: Prentice-Hall, 1966), p. 156.
the business firm and retention of the consequences Of the
purchase in the buyer's memory.
Nicosia noted that a major advantage of his model
13 that his is amenable to simulation techniques for analyz-
lng the effects of the various variables. Also, the model
Indicates the occurrence of feedback and successfully inte-
gra . . . o 27
tes communication input and response output variables.
Robertson points out a criticism Of this model.

Thfi model consists Of "long listings Of variables that

19111: possibly enter into a consumer behavior model with

(L

:z'rative l
3&- 7

SI .. .ea.

”VD" M~
“‘uv~-.‘q

73583 in
.m‘ ‘

:11 11‘. d
” a J...
3.1;“
A.

3 source

SURE): W‘" 4

0“
ation t5
‘ I
:elef' f

39

little, if any, explicit treatment of how they are inter-
related".28

The Andreasen model (Figure 2-4), also, assumes in-
novative behavior since the product is either a new product,
or at least new to the individual. Andreasen advises the
marketing manager to make note of the type of newness in—
volved in the product. This model begins with the individ-
ual in a state of unawareness, therefore, having develOped
no attitude toward the given innovation. One Of a number
of sources communicates a form Of information to the con-
sumer which he filters out (selective process). The infor—
mation then affects his attitude, defined in terms Of
belief, feeling and disposition components. At this point
he may do one of three things: (1) select the product,
(2) search for more information, or (3) take no action. A
"select’ decision will be mediated by certain constraints,
and other purchase decisions (store, quantity, etc.) will

29 The model

have to be made before ownership is final.
contains all the advantages as well as the disadvantages Of
the previous model including an even more incomplete speci—
fication of variable interrelationships.

Unlike the preceding models, the Howard and Sheth

model (Figure 2-5) focuses on repeat purchase behavior and

has four major components: stimulus variables, response

I
kn.

 

.acmmaoo one anamouom .uuoom

 

.mom .a .xmoma

 

"mwOGwHHH .30w>soawv How>mzom Hmaamsoo ca

mo>flvommmuom .:Omuuonom .m mmaona was amenHMmmmx .m oaoumm ca oovsflnmom
.:Hoooz coamfloon d uuofl>msom uoEoumso can moosuauu<= .sommouos< .m Guam "mousom

.1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
l
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
l
I
I
I
I
I
I
I

.1
I
I
I
I
I
I
I

J

 

 

 

 

 

_ mmvas. ZOE—Duo «$20.58 Curr—bi—

 

        

  

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

        

 

 

 

 

 

 

 

 

 

 

 

 

 

_
_
_
_
_
""" J l"'|'l"'|'|"'|'|'ll"’"l ' ‘l""|' “mason“
—1 w q!"ll""ll|.l """"" "'|)I“"U" lllll rho-“ancho‘c— "
_
_ _ _ a . _
_ _ _ . 38:3 _
_ _ _ £93.“ :33 3:25.80 3:029”. Ki
_ _ _ 35>» 886 209.809.. .
— — — w """""" ‘ J H —
_ _ _.L- E . , _
3050a .
0 u n o. .39. .3325 14 b:3o=§<
4 _ _ qII|l_ c r .o . Snoo>u< _ coca
‘ . .
. r IIIIIIII I_ 538 _A K 20:15 .E “ Mwwwwzw
. L 3818
. 3c. 3 93° . . .8353
. «co soon _II_|I.1| . . m .362an .9.— 32:2.
. . .coocoaoOE .
Suuhfia oz 628 _ c2552...
.
_ «or 883m
. 5:23.: 3 XS. rh
_ 040: 230i
_
n 2050 “cauficufi “NH.”
.0 ooaua>
_ a . . . . 8852
.3an 29.026... 98803600 .2502 2.2.00
w. IIIIIII .1 Eons-o .82.: 85:33 3288.. 3.333.. I I I I
333.5 .330 839i 2.30... 23E 89.3 I
083:. .3352
£2.28

mmmvomm ZOHmHUmQ Mméugm NEH. .mO ZOHBfiNHAd—HHQHUZOU ZNmfiflMQ—fim ”Han.

film gUHh

 

 

/

in

ntl I I..II|

.. N—h.vH.>S~rhnm

 

N~n.~>hiw FN

 

puflc \hoAvanur—t...

a. ..I .3; v-1~u v - ..~

~ npruiuhm IAWHNH ANZPJN ~

hvnh Nuns

41

.mmm .m .Amoma .Ommoanov m sapwoooum oucouowcoo Housflz homa
.=How>mzwm woman no huoonez .fiuosm .z nmflommb one oumBom .< GAOO

mhuukuu 02¢

Wu4m4u¢d> mnOZMUOXN no u“ lumbar: IIII
mkuumuu xuqmomwu IIIIII
20:.«510uz. no in C Ffimhao.k3mz_

 

 

        

20:23.:

zofizwxmmnis
mmoathhd I

20:70.5:

 

  

 

zolkuflumnbdm +4
_
.

r IIIII l
o

  

    

_

muim

 

..0.h....._h.ou..:

 

"monsom

 

    

  

o... :SEmMn a

 

Jthmqumm

   

 

.__....__J

 

U. mHUulwuhuz

 

 

  

  

20....«2c0m2.

 

 

h2m220m§du Jdauom

 

35:3 .0
>..._:._o....__.,.>... .v
wnmzuxahuurhflo .n
unia .~

C3430 ..

 

U_J0m\..>w

 

 

 

 

«.8 223% _

 

 

 

 

Gogdx um

 

  

  

 

 

 

 

wmflxumna

  

”53:30

 

 

 

 

 

 

 

 

muzéum .n
ytfmdjfia .o
mww2u>Cuzfim5 .n
uoia .N

T1193”. ._

 

u>:4unh.u....w_m

 

 

wordzm

 

 

 

 

 

 

     

 

 

 

all.

=MOH>¢mmm MHMDm ho Mmommfiz mfimmmIQM¢ZOm HEB

‘IL
. nut...” (ammo — mmuhnao _
a .

 

 

 

 

 

 

I

   

$835; _ onwxuuu _

to 835.22. .

 

> .— 34/03:

 

mIN mgUHh

 

 

mhjaz.

42
variables, hypothetical constructs, and exogenous variables.
The inputs to the buyer's internal state are stimuli from
the environment, either commercial or social. These com-
mercial stimuli are the marketing activities of the various
firms. They may come either via the physical brands them—
selves Or some linguistic or pictorial representations of
the attributes of the brands. The social stimulus input I"-
refers to the information that the buyer's social environ- I»
ment provides regarding a purchase decision, for example, I
word of mouth communication. The hypothetical constructs
are enclosed within the large rectangular box, which repre—
sents the consumer's internal state. Howard and Sheth have
noted that the two hypothetical constructs, learning and
perceptual, serve the role of endogeneous variables. The
learning constructs include motives, evoked set, decision
mediators, predispositions, inhibitors and satisfaction:
whereas, the perceptual constructs include sensitivity to
information, perceptual bias and search for information.
The exogenous variables, shown at the tOp of the diagram,
influence the hypothetical constructs and provide for
adjustment for interpersonal differences. The response
variables are ordered to create a hierarchy similar to the

variety Of hierarchies of the AIDA and Lavidge and Steiner

models. However, Howard and Sheth have incorporated

 

D 5‘
.3 ”E

43
30
several feedback effects.
While admitting that it is difficult to do justice
to these models in so short a description, it is important

that the reader notes the significant step forward that

these models have provided the marketing manager.

Summary of Adoption Models

After an analysis of the different adOption models
presented thus far, one is left with the feeling that the
concept of an adOption model is logical. If a behavioral
act is to occur, it must have antecedents. Therefore, such
a model, by merely forcing attention on these antecedents,
provides a service for the marketing manager. The concept
of an adoption model is of invaluable use in increasing our
understanding of consumers behavior.

However, regardless of which model the market re-
searcher chooses to follow, certain observations should be
remembered.

First of all, the process need not always conform
to a single form. Granted, while most behavior is of the
rational/decision-making form, the market researcher should
keep in mind that the nonrationalpsychosocial form as well

as the nonrational/impulse form are possible eXplanations

of human behavior.31

Q i
‘ fluwr

-. JJH¢n
I
(.5.-
anJ o

and EC

t1”. “‘5‘“

An

44

Second, there is no maximum number of stages al-
though the minimum number appears to be two.

And thirdly, no specified sequence of stages must
occur. Allowance must be made for consumers to skip stages
and for the occurrence of feedback.

For the above reasons the author prOposes another
model of the adoption process. This model is derived from
the 1971 innovation-decision process of Rogers and makes
alterations for the above observations. This model is
shown in Figure 2-6.

The model is a logical extension of the previous
discussion and has four advantages.

1. The model may be conceptualized as either an
innovation-oriented process, that is starting
with awareness, or a problem—oriented process,
starting with problem perception.

2. Feedback effects are taken into consideration.

3. The model enables the reader to trace the dif—
ferent sequences of behavior which might occur
as the consumer may skip some stages in the
adoption process.

4. The model's final stage, confirmation, allows
either a continuation of the action taken in

the decision stage, or the rejection of that

jam Ado: ZOHPH.hhadAH< s3 m/szam

MU [IN Tu Vfi ulumuvi. v."

45

 

Amie sowumahawcoo

 

\
T!,|.W

 

 

coamwomn

.k\\
‘A// mafia

wGUHSOm GOfiumUHfiSEBOU

 

 

 

 

 

 

 

soauoofiom
pmssfiusou

l

 

 

 

. sowuuov< umumq

 

4

/LF

 

 

 

AWLWMuQOp< pmsafiusoWTMHllllllr

 

m NUGNDGfiufioumHQ i

 

nW\\\\\\\l

 

mmlllllllg cofiuumhmm

sowuaop<

 

 

 

 

AHDOZ ZOHBmond MMdZZDm

mlN NMDUHm

 

 

owvmasosu

 

 

soauaoouom amanonm

 

 

F‘ngvd-

yn‘v -
I

‘i'c ‘n
~.u- ‘1‘

~.' .
inch

. Q

46
decision.32

This model will be discussed in greater detail in

Chapter III.

Industrial Goods and the Adoption Models

Ozanne and Churchill have examined the process in
the industrial goods market and found the industrial pro—
cess to be the same as the traditional one except that the
trial stage might be eliminated with indivisible innova-
tions.33 They did state that where limited scale prototypes
could be used, the trial stage would be continued. Contrary
to predictions, however, they found "personal sources (in
particular personal selling) were more important at early
stages, while impersonal sources (especially the price quo—
tation and tooling proposal) were paramount at the evalua—
tion stage. The available evidence also suggests that as
the final decision approaches, the need for informational
inputs increases. At the evaluation stage the industrial
decision makers employ a larger number and a greater

0 I I u 34
variety of information sources than at the earlier stages."

The Diffusion Process
Rogers has defined diffusion as the process by
which innovations spread to the members of a social system.

(He had previously defined it as the process by which one

47
follows the spread of a new idea from its source of inven-

35 Thus,

tion or creation to its ultimate users or adOpters.
as he sees it, there are four crucial elements in the dif—
fusion of new ideas: the innovation which is communicated
through certain channels over time among the members of a
social system. Here the term innovation is the idea, prac-
tice or object perceived as new by an individual. Rogers
states that it is the perceived or subjective newness of
the idea, and not the "objective" newness, for the individ-
ual that is important. Communication is the process by
which messages are transmitted from a source to a receiver.
The channel is the means by which these messages move from
source to receiver.

Time is the important element in the process. Katz
has been quoted as saying "time is the key to diffusion
research". Time is usually measured in three dimensions:

1. The length of time in innovation-decision pro—

cess, that is the length of time during which
an individual passes from first knowledge of
the innovation to the adOption or rejection
of that product.

2. The innovativeness of the individual, that is,

the relative earliness-lateness with which an

individual adOpts an innovation when compared

 

48
with other members of his social system.
3. The innovation's rate of adoption in a social
system, usually measured as the number of
members of the system that adOpt the innova-
tion in a given time period.
A social system is defined as a collectivity of 1
units which are functionally differentiated and engaged in
joint problem solving with respect to a common goal. The
members or units of a social system may be individuals,

informed groups, complex organizations or subsystems.36

The Diffusion Process in Marketing

 

The diffusion process in marketing can be conceptu-
alized as: the adoption of new products and services over
time by consumers within social systems as encouraged by
marketing activities. Adoption refers to the use of a new
innovation. New products can be any product perceived as
providing additional utility by the consumer. The time
dimension will distinguish early adOpters from late
adOpters. Consumers will refer to the consumer adopting
unit, be it individual, family, organization or political
unit. Social systems constitute the boundaries within
which diffusion occurs. This may range from family to

friendship groups to the entire market place. Marketing

. _ ‘—

“III?!“ 5 [in

|
H. \
'I

5‘ ”-
a. L

:w
1. )1

A;

49
activities will refer to those activities undertaken by the
firm in order to gain consumer patronage. These most gen-
erally include the mixing of the various marketing variables
of product, price, promotions and place in forming an Opti-
mum marketing mix.

These aspects of the diffusion process are interde-
pendent. For example: the attributes of the new product
will affect the rate of adOption over time, the types of
consumer who will adOpt, the kinds of social systems within
which diffusion will take place, and the marketing efforts
needed to achieve diffusion. Similarly, the marketing
manager must realize that successful new product diffusion
is critically dependent upon the communication of relevant
product information and matching the self images of social
system members with the perceived product images. That
marketing activities can guide and control the rate of
adoption is witnessed by Zaltman, in which the various
marketing strategies are explained in terms of past studies

from the area of behavioral sciences.37

Adopter Categories

 

Not all individuals in a social system adopt an in-
novation at the same time. Rather, individuals adopt in an

ordered time sequence and they may be classified into

 

x /_

*9
eb

CE.

m

C»

50
adOpter categories on the basis of when they first begin
using a new idea. Figure 2-7 shows the traditional five
categories of classifying adOpters. The figure, also,
shows the approximate percentage of individuals in these
five categories (innovators, early adopters, early majority,
late majority, and laggards). Rogers notes the above clas-
sification is not symmetrical and it fails to account for
incomplete adOption. However, neither of these criticisms

has distracted from the model.38

FIGURE 2-7

ADOPTER CATEGORIZATION ON THE
BASIS OF INNOVATIVENESS

  
    

 
        

 

N h----------b---

:
. : :
muovnons i :
EARLY . EARLY um: :
.Aooprms: MAJORITY MAJORITY . LAGGARDS
2.5%: 13.5% g 34% 34% : 15% A
I—zsd I-sd I+sd

The innovativeness dimension, as measured by the time at which an individual
adopts an innovation or innovations, is continuous. However, this variable may be
partitioned into five adopter categories by laying off standard deviations from
the average time of adoption.

Source: Everett M. Rogers with F. Floyd Shoemaker,
Communication of Innovations (New Ybrk: The
Free Press, 1971), p. 182.

 

 

::e f

i,"F\“'r

fiend} I

~4

A e

Stanton,

51

in reviewing the work of Rogers, has drawn

the following observations concerning these five categories.

Innovators:

Early adOpters:

Early majority:

Late majority:

Laggards:

They are the first to adOpt and have been
referred to as venturesome. They tend to
be young and, at the same time, high in
social and economic status. They are
cosmOpolites, with many contacts outside
their own social groups and community.
This group tends to rely on impersonal
and scientific information sources or
other innovators rather than personal
salesmen.

 

This group is likely to be relatively high
in social status, probably being Opinion

leaders. They may be younger, more mobile

and more creative than later adopters.
Their social relationships are confined to
local groups and they have the greatest
contact of all the groups with salesmen.

This category consists of those with above
average social status. They usually will
not consider an innovation until early
adOpters have tried it. A long period may
elapse between trial and adoption. This

deliberate group has considerable contact

with mass media and salesmen and early
adOpters.

PeOple in this group tend to be below
average in social status and income. They
are less likely to follow Opinion leaders
and early adopters. Some social pressure
might have to be applied to this group in
order for the product to be tried. They
are a skeptical group. This group makes
little use of mass media and of salesmen.
They tend to be oriented more to other
late adOpters than to outside sources

of information.

 

This group has the lowest social status
and income, and tends to be tradition—

 

bound. Their main source of information

is other laggards.39

 

112' 316
CEIELI

s‘abcg‘
' 5‘:

29.131?

52

From a content analysis of over 3,000 research

findings relating various independent variables to innova-

tiveness, Rogers has made thirty-two generalizations con—

cerning innovativeness on the basis of (l) socio-economic

status, (2) personality variables and (3) communication

behavior.40

These findings are reprinted here in summary

form along with the number of supporting and non-supporting

studies.

Socio-economic Characteristics

1.

Earlier adopters are no different from later
adopters in age. (228: 44 younger, 108 no
relationship, 76 older)

Earlier adOpters have more years of education
than do later adOpters. (275: 203—72)41

Earlier adOpters are more likely to be literate
than are later adOpters. (38: 24-14)

Earlier adOpters have higher social status than
later adopters. (402: 275-127)

Earlier adOpters have a greater degree of upward
social mobility than do later adOpters. (5:
5-0)

Earlier adOpters have larger size units (farms,
etc.) than do late adOpters. (227: 152-75)

Earlier adopters are more likely to have a com-
mercial (rather than a subsistence) orientation
than are later adopters. (28: 20-8)

Earlier adopters have a more favorable attitude
toward credit than later adOpters. (25: 19-6)

Earlier adopters have more specialized opera—
tions than later adOpters. (15: 9-6)

w—

 

53
Personality Variables

10. Earlier adOpters have greater empathy than
later adopters. (14: 9-5)

11. Earlier adOpters are less dogmatic than later
adOpters. (36: 17-19)

12. Earlier adopters have a greater ability to
deal with abstractions than do later adopters.
(8: 5-3)

13. Earlier adopters have greater rationality than
later adopters. (14: 11-3)

14. Earlier adOpters have greater intelligence than
later adOpters. (5: 5-0)

15. Earlier adopters have a more favorable attitude
toward change than later adOpters. (57: 43-14)

16. Earlier adopters have a more favorable attitude
toward risk than later adopters. (37: 27—10)

17. Earlier adopters have a more favorable attitude
toward education than later adopters. (31:
25-6)

18. Earlier adopters have a more favorable attitude
toward science than later adopters. (27: 20-7)

19. Earlier adOpters are less fatalistic than later
adopters. (17: 14-3)

20. Earlier adOpters have higher levels of achieve-
ment motivation than later adOpters. (23:
14-9)

21. Earlier adOpters have higher aSpirations (for
education, occupations, etc.) than later
adOpters. (39: 29-10)

Communication Behavior

22. Earlier adOpters have more social participation
than later adOpters. (149: 109-40)

VAL
“1‘

l‘.

54

23. Earlier adOpters are more highly integrated
with the social system than later adOpters.
(6: 6-0)

24. Earlier adOpters are more cosmopolite than
later adOpters. (174: 132-42)

25. Earlier adopters have more change agent con-
tact than later adOpters. (156: 135-21)

26. Earlier adOpters have greater exposure to mass
media communication channels than later
adOpters. (116: 80-36)

27. Earlier adopters have greater exposure to inter—
personal communication channels than later
adopters. (60: 46-14)

28. Earlier adopters seek information about inova-
tions more than later adOpters. (14: 12-2)

29. Earlier adOpters have greater knowledge of
innovations than later adOpters. (80: 61-19)

30. Earlier adOpters have a higher degree of
Opinion leadership than later adOpters.
(55: 42-13)

31. Earlier adopters are more likely to belong to
social systems with modern rather than tradi-
tional norms than are later adOpters. (46:
32-14)

32. Earlier adOpters are more likely to belong to
well integrated systems than are later adOpters.
(15: 8-7)42
The Innovation-Decision Period
The innovation-decision period is the length of
time required to pass through the innovation—decision pro—

43

cess. The length is usually measured from first knowledge

until the decision to adOpt (or reject), although in a

55
strict sense it should perhaps be measured to the time of
confirmation. This last step is often impractical or im-
possible because the confirmation functions may continue
over an indefinite period of time. Rogers has listed ten
generalizations concerning variables affecting this period
and the length of time involved. By way of providing a
summary for the reader of the research already performed
in this area the Rogers list follows along with the number
of supporting and nonsupporting studies.

1. Earlier knowers of an innovation have more
education than later knowers. (24: 17—7)

2. Earlier knowers of an innovation have higher
social status than later knowers. (28: 18-10)

3. Earlier knowers of an innovation have greater
eXposure to mass media channels of communica—
tion than later knowers. (29: 18-11)

4. Earlier knowers of an innovation have greater
exposure to interpersonal channels of communi-
cation than later knowers. (18: 16-2)

5. Earlier knowers of an innovation have greater
change agent contact than later knowers.
(26: 23-3)

6. Earlier knowers of an innovation have more
social participation than later knowers.
(13: 11-2)

7. Earlier knowers of an innovation are more
cosmopolite than later knowers. (5: 5-0)

8. Later adopters are more likely to discontinue
innovations than are earlier adOpters.
(6: 6-0)

::~‘l
A‘s».

 

i926

~f‘, n
st .

56
9. The rate of awareness-knowledge for an inno-
vation is more rapid than its rate of adOption.

(2: 2-0)

10. Earlier adOpters have a shorter innovation-

decision period than later adopters. (6: 5-1)44

ROgers' earlier text, also, went into the differ-
ences between personal and impersonal communications as
factors of increasing awareness. At that time he made the
generalization that "impersonal information sources are
most important at the awareness stage, and personal sources
are most important at the evaluation stage in adoption

" 45 . .
process . A second generalization from that text was
that "cosm0polite information sources are most important at
the awareness stage, and localite information sources are

. . n 46 .
most important at the evaluation stage . Later marketing
studies have supported these generalizations. Arndt dis—
closed that "product-related word of mouth was found to
flow from early to late adOpters and noncadOpters. More
than two-thirds of the comments were received by respondents

47 Two other interesting findings

who had not bought yet".
of his were that "compared with the non—exposed individuals,
those receiving favorable word of mouth pressure were more
likely to buy the product, while those exposed to unfavor-

48

able word of mouth were less likely to buy", and ”respond—

ents low in generalized self-confidence seemed to react to

Ha.-

word 01

me

Sumzer

:eerin

(I)

57
word of mouth in an ego defensive manner. They also tended
to be less likely to be exposed to word of mouth."49
Summers stated that his research suggests that the volun-
teering of unsolicited product information is generally
more common in interpersonal channels than information

seeking behavior.50

Diffusion Effect

 

The diffusion effect51 is the cumulatively increas-
ing degree of influence upon an individual to adopt or re—
ject an innovation, resulting from the increasing rate of
knowledge and adOption or rejection of the innovation in

52 The diffusion effect is often listed

the social system.
as the reason for the increasing rate of growth of the in-
novation in the diffusion process. It is thought that if
every consumer considered adopting on an individual basis,
without social influence, then the probability of adOption
would be the same for everybody regardless of time period.
However, if consumer influence is introducted than a "snow—
balling" effect will occur, since other's previous experi—
ence with the innovation will influence the present deci—
sion. Summers' work can lead to the generalizations that

personal influences gain as the risk involved increases

(higher prices, greater complexity of product) and that

‘o
A n
U one

VIN!
. \lu
.

H ‘ y
Hg.

‘ d

‘h

5"

'l-A

58

other factors which will induce the diffusion effect include
product ownership and competence.53

Other research to support the diffusion effect in-
clude the studies used in abandonment of the hypodermic
needle model which postulated that the mass media had
direct, immediate and powerful effects on a mass audience.
Another is the introduction of the "two-step flow" model
which hypothesized that information is moved from sources
to Opinion leaders, who in turn influence their followers.
The three most famous studies here include the 1940 Erie
County Election Study, 1954 Decatur Study of Opinion Lead-
ership and the Coleman, Katz and Menzel Drug Study.54

Subsequent research by Allvine and Arndt in the
area of retail grocery sales have reconfirmed the diffusion
effect and use S-shape growth patterns in diffusion process.
Allvine's findings in a study of the acceptance of pro-
motional games by supermarkets found that the growth pat-
terns suggested both a diffusion effect (he called it
interaction) and a S-shape diffusion process. He also
found that the rate of diffusion was prOportional to the

55 Arndt found two measures

importance of the first adOpter.
of sociometric integration (number of close friends and

number of persons with whom you are likely to discuss new

food products) were significantly positively related both

59
to whether a respondent received word of mouth communication
and to acceptance of the new product. Generalized self-

confidence was also positively related.56

Opinion Leadership

 

Much has already been written in this chapter with
regard to Opinion leaders, communication flows and inter-
personal relationships affecting buyer behavior. Yet,
Mancuso's reminder to marketing managers that ”Opinion
leaders have not been fully utilized...in assisting with
new product introduction", still remains true.57

Opinion leadership,58 the degree to which an indi—
vidual is able to influence informally other individual's
attitudes or overt behavior in a desired way with relative

59

frequency, have been widely discussed in recent marketing

60 yet marketing managers still know very

publications,
little concerning its profitable use.

At the present time, there appear to be four basic
strategies with regards to use of Opinion leadership in new
product introduction. The first is to create leaders in a
manner similar to Mancuso's record shOp experiment, in
which teenage panels were used to enable "select" records
61

to reach the TOp Ten charts only in panel cities.

Another example would be to offer certain selected

F?

 

 

 

02‘s

5w.

Tu;

oug

‘L'
(.1:

‘ I

‘1

lo

(I)

60

consumers "special deals" on new products. This method is
costly and usually does not locate new Opinion leaders.
The third method would be to locate and identify Opinion
leaders. However, this becomes more difficult as market
size increases.

The most common approach toward influencing the
buyer is to focus on the characteristics of the Opinion
leaders in general and then aim a promotional campaign at

those characteristics.

Industrial Goods and the Diffusion Process

Martilla has found in research conducted in three
industrial markets that word of mouth communication within
firms is an important influence in the later stages of the
adOption process. Word of mouth communication between firms
was found to be more situational. Opinion leaders were
found to be more heavily exposed to impersonal sources of
information than other buying influentials in the firm. The
study also reported that, as in consumer marketing, indus-
trial Opinion leaders are difficult to locate and identify
using available demographic data.62 Webster, a year earlier
in interviews with industrial buyers, failed to identify a
significant amount Of word of mouth communication in indus—

trial markets and suggested a key role for manufacturers'

Db.

 

61

63
salesmen.

Innovation Characteristics

 

Rogers' Characteristics

As noted earlier, the rate of growth and the extent
of product diffusion are largely a function of the per—
ceived attributes of the innovation. Rogers has prOposed
a set of five characteristics which contribute to the ex—
planation of the different rates of adOption. While
realizing that they are not a complete list, but at least
the most important characteristics, he has found that
(1) relative advantage, (2) compatibility, (3) complexity,
(4) trialability and (5) Observability are useful in de—
scribing the rate of adOption.64

Relative advantage is the degree to which an inno-
vation is perceived as being better than the idea it super-
sedes. While this characteristic is often measured in terms
Of utility by the user. For example, Rogers notes that the
major advantage of 2,4-D weed spray was the reduction in
unpleasant labor tasks, rather than in financial gains per
se. In his review of the literature, Rogers found 29 of 43
studies agreed that relative advantage was positively re-

65

lated to rate of adOption. Thus, marketing managers have

sought means of encouraging the consumer to perceive a

62

greater value in their product than that of the competition.

Compatibility is the degree to which an innovation
is perceived as consistent with the existing values, past
experiences and needs of the consumers.66 The introduction
of self-cleaning ovens required no changes in the way a
housewife went about baking. Electronic ovens, however,
cook much more rapidly and don't "brown" food to the same
extent. Because they require a change in the way cooking
has traditionally been done, electronic ovens are likely to

67 Rogers has found 18

encounter a slower rate of adOption.
of 27 studies agreeing with the premise that the greater
the need for consumers to restructure their thinking and to
engage in new forms of behavior, the less quickly the item
is to be adOpted.68

Complexity is the degree to which an innovation is
perceived as relatively difficult to understand and use.
Nine of sixteen studies have agreed to the negative rela—
tionship between complexity and rate of adOption. An ex—
ample of this was diffusion of canasta and television among
different social classes. Television was considered to be
less complex for the lower classes.

Trialability is the degree to which an innovation

may be experimented with on a limited basis. In—store

sampling of a new food product or the introduction of trial

63
size packages account for the factor in which nine of
thirteen studies agree.

Observability is the degree to which the results of
an innovation are visible to others. Fashion trends move
rapidly through their life cycles due to their Observability.
Seven of nine studies agree to this finding.

As mentioned earlier, the important point is how
these characteristics are perceived by the consumers and
not the subjective evaluation. Thus, it can be generalized
that the rate of adOption is positively related to relative
advantage, compatibility, trialability and Observability

and negatively related to complexity.69

Other Studies

 

Other studies have pointed out other characteris-
tics. One of these studies was by Mansfield and the re—
sults of that study have been used to support the idea of
the diffusion effect. Mansfield studied the rapidity with
which twelve innovations spread through the industrial
sector. A major finding was that the prOportion of firms
already using an innovation would increase the rate of
adoption. This lends support to the notion of the band-
wagon characteristic. Mansfield's hypotheses were:

1. Profitability of an innovation relative to
others that are available will increase the

64
rate of adOption.

2. The larger the investment required, assuming
equally profitable innovations available, the
slower the rate of adOption.

3. The type of industry will affect the rate of

adOption depending on its aversion to risk,
market competitiveness, and financial health.

70
Finally, as mentioned earlier, the only other mar—
keting studies in this area are with regard to risk percep—
tion. These studies have shown that the risk perceived by
consumers in new product adOption is negatively related to
buying behavior. Studies have also shown it to be a major

. 7l
factor in buyer response to new products.

Product Elimination Studies

 

Background

 

No other area of marketing probably has as little
written on it as that pertaining to products which are to

72 Berenson,

be eliminated from the firms product line.
Grashof and Rothe73 all have commented that the literature
on product elimination is extremely sparse and vaguely
defined: no body of knowledge exists that can be referred
to for guidance for action in this idea. The few contribu—
tions have all been theoretical in nature and of somewhat

limited use to small-to—medium size retailers. The product

elimination area is further clouded as to whom should have

RE

.
‘I
to

‘ 1

f1

65

the responsibility for such action. The traditional ap-
proach has been to centralize such authority in the home
office of the chains,74 however at the present time a
change is being made toward giving the individual store
managers some authority on their product line.75 (It
should be noted that the chain co-Operating with this re-
search project was an early adOpter of the latter method.)

Rothe has studied the different factors involved in
the product elimination decision in the food industry, as
well as the drug, clothing and major and minor appliances
industries. One of his findings was that firms in the food
industry rated product elimination activities approximately
one-third as important as new product activities. In the
recognition stage of locating weak products, little atten-
tion was given to product profitability. This came at the
next stage, analysis. The major factors for food items at
the recognition stage were: minimum dollar volume, minimum
unit volume, minimum market share percentage, some compari-
son of today's market share with previous years and percent-
age of total company sales this product contributes. Prod—
uct profitability is the major factor in the analysis stage.
A final finding of Rothe was that while much of the litera—

ture dealing with this subject dwells on the formality

issue, the food industry respondents were less formal in

66

their elimination decisions.76

Traditional Approaches

 

Before reviewing the theoretical approaches to
product elimination suggested in the literature and moving
on to a discussion of actual practices in the retail grocery
industry, it will be useful to first review some of the
traditional lines of thought. According to Berenson,
product elimination decisions have traditionally been domi-
nated by four different viewpoints: the accountant's, the
economist's, the sales manager's and, perhaps to a lesser
extent, the government policy maker's.

The accountant's customary view of a product line
deletion involves a comparison of the dollar costs of reten—
tion with the dollar cost of abandonment. This approach is
concerned with quantitative measures of depreciation or
product disposal costs of a food item, current expenditures
and revenues. The primary emphasis is on quantifiable
financial items. While Berenson makes a distinction between
the accountant's and the economist's view, both are basi—
cally cost-revenue decisions.

For the economist, Berenson notes that product eli—
mination is a matter of emphasizing the future and leaving

the past for historical record. The prime considerations

67
in this method are alternative choices and marginal costs.
It involves questions of incremental profits - for example,
the possibility that the product may be in the black on an
out-of—pocket cost basis and can therefore make some con-
tribution to general overhead.

HOpefully, the sales manager's vieWpOint would be a
synthesis of both the accounting and economic traditions.
However, this expectation is usually unfulfilled. The
sales manager's approach to the problem has been largely
intuitive. It stresses the factors that may make the line
easier to sell but not necessarily more profitable — for
example, it favors carrying a full line and seeking to
build volume at the expense of over-all, long-term profit.

The decision criteria for the government policy
maker relates to public interest. The government tends to
consider continued satisfaction of the consumer as an over-
riding criterion. Hence, railroads regulated by the ICC
cannot readily abandon trackage or other services when the

line as a whole is making a profit.77

Theoretical Approaches

 

As has been noted earlier, the vast majority Of
firms do not have established procedures for pruning their

products. Such action is usually undertaken either (1) on

68
a piecemeal basis, as in instances where the product's
money-losing status is incontrovertible, conspicuous, and
embarrassing, or (2) on a crisis basis, wherein the pre-
cipitating event may be a financial setback, a persistent
decline in total sales, piling inventories or rising costs.

However, neither of these practices has been suc-
cessful in the long run.78 In an attempt to provide the
firm with a more reliable method of eliminating non-produc-
ing products from their line a number of theoretical models
have been prOposed. These models have as their basis a
thorough analysis of the basic concepts of marketing,
finance, management, psychology and accounting.

One of the earlier theoretical models prOposed was
that of Berenson. He presented a model (Figure 2-8) which
considered five major decision factors: financial security,
financial Opportunity, marketins strategy, social respon-
sibility, and the possibility of organized intervention
against product deletion. The first two criteria are
readily quantified: the first relating an evaluation of the
basic profit criteria of the firm, and the second which
provides an Opportunity to consider the profitability of
the product in terms of opportunity costs, phase of the
product's life cycle, and the amount of return in excess

of the firm's minimum goal. He suggested that a judgment-

(59

 

.mcowfluom mmmcfimsm .

<4 coauom oxmu

.qx afloOmEonudmm ma >w ma

m< coauom mxmu

.mx kHOumE«xouadm ma >w ma

N4 coauom oxmu
.Nx mamumefixouaam mH >w we
H< coauom mxmu

.Hx kaoumauxouadm we >w we

>w so ao<

=OCwq uospoum wzu mcflcnum:

.66 .m .momH
.GOmcmumm pmucou “mousom

coauom oamwooam m sues pmumHouuoo ma umnu >w mo Hm>ma OLu u
>w he paumowpafi sofiuom man u <

mousmmoa onu «0 85m ozu I
maumufiuo one mo mocmuuooefl m>HumHmu osu mo musmmoa m u

.HQEEDW

 

 

 

 

 

 

coauco>uOucH conacmwuo
wawuommmm muouomm

 

 

 

 

 

 

muaaanflmaoamou Hmauom
wcfiuummmm muouomm

 

 

 

 

 

mwmumuum wcfiuoxuma
mcwuomwmm mHOuomm

 

 

 

 

huficsuuoaao Hmfiocosfiw
wsauuowmm muouomm

 

 

I m _
>
a
uq< _ >
um< >w. u m>
.~< l
~>
"H<
munwuos mwumufiuo
and ee< use
unwaas

AWQOZ Bzmzzonz<m< ZOmmemm Mme

wlN WMDDHE

 

augusumm Hmwwwmcam
msauummmm muouumm

mfiumufiuo ucm>oamm
mnu macho

70
determined numerical scale could be established to accom—
modate the remaining three factors. The subjective weights
assigned to these factors are to reflect the degree of im-
portance attributed them by management. The score for each
category is then multiplied by the weighting factor, and
the summation of the five weighted scores becomes the over-
all rating of the product under question.7

The Kotler model (Figure 2-9) for product pruning
is, as he admitted, an expensive one in terms of executive
time, but the cost must be compared to the greater cost of
keeping a sub-optimal product in the line. The model, a
PERT approach, is made up of a creation and Operational
level. The creation level is composed of the develOpment
of a representative corporate team and the establishment of
objectives and procedures related to product pruning.

The Operational level is a six-step approach, the
first of which consists of management preparing a data
sheet for every company product. The data sheet contains
all the important information about the product during the
past three years. The data sheet is to provide information
for judging the product by the management team. Step 2
consists Of a computer program to review the data from
Step 1 for any signs of weakness. For example, the firm

might set the standard of a sales decline in any four

THE KOTLER ABANDONMENT MODEL

Cltl-IA'I'ION STAGES

OPERATIONAL STEPS

 

U

Ito-cognition

 

 

U

AirfliniS

 

lesc Out

i

 

Source: Philip Kotler,

71

FIGURE 2-9

 

Appoint a product
review (mnrnittcc

 

 

Hold mr-olings to set

Objectives a t id proccd u; cs

 

 

related to product pruning

 

.—

i

l. Controller's office
fills out product

   

 

(lain—streets

  

 

 

. Computer program

no. 1 determines
dubious products

 

 

 

3. Mmmgzr-mr‘nt tr-am

 

‘i’

 

fills out ruling [onus
for dubious products

 

 

 

Y

 

5. Managtuunt trrsun reviews
thcsc indexes and decides
on products to drop

 

 

 

 

(r

 

6. Management loam develops
policies and plans {or
phasing out “dropped”
products

 

 

 

 

. Computer program

no. 2 dr‘tr‘rmiur‘s
prodm-t rrtontion
index for molt
dubious product

 

J

"Phasing Out Weak Products",
Harvard Business Review, March—April, 1965,

 

p. 112.

 

 

72

periods during the three years as a sign of weakness. The
remaining three steps are similar to Berenson's model. A
rating form is prOposed for detailed analysis of the dele-
tion candidates. Management then assigns a numerical score
to the categories on the form. weighted factors are applied
to the scores, and the weighted scores are summed to obtain
an overall "product retention index”. Product deletion
decisions can then be made using a cut—off point in the
retention index. At this point management may make some
subjective judgments as to possible customer reactions.
Lastly, plans and policies for phasing out "drOpped" prod-
ucts are developed. For each product, management must
determine its obligations to the various parties affected
by the decision. Here it may decide to stock a reasonable
amount of replacement parts or to seek out another manu-
facturer for the product.80

Hamelman and Mazze in 1972 introduced an extension
of the Kotler model called PRESS (Product Review and Evalu-
ation Subsystem). This model is different from other
product—elimination models in that it is capable of c0ping
with a company's total product line rather than a segment
of the line thought to be weak.

The program consists of four integrated parts,

PRESS I through PRESS IV. PRESS I contains the primary

73
model and uses standard cost accounting and marketing per-
formance data, while PRESS II, III, and IV perform analyses
concerned with price changes, sales trends, and product
interaction.

The PRESS model differs from Kotler's approach in
that it reduces the amount of executive management decision
time and that it looks at the entire product line. Whereas
Kotler provided broad guidelines for his model, PRESS con—
siders product line interactions and Operational aspects of
deletion decisions. The retention index of Kotler yielded
a single number indicating the degree of product desirability
from the weighted ratings on the seven subjective scales.
PRESS offers cutoff points for deletion decisions by a
systematic review of Selection Index Numbers, which are
based on a series of performance ratios using standard Cost
accounting data.81

It is interesting to note at this point, that any of
the adOption process models mentioned earlier in this chap-
ter can, also, be used as a product elimination model. The
adOption criteria and the retention criteria are basically
the same, the difference is that one involves forecasting
and the other measurement. However, both include a subjec—

tive weighting of the processes' elements.

74

Retail Grocery Practices

 

At the present time the actual practice being under-
taken in the retail grocery industry appears to being moving
away from the main office and back to the store manager with
regard to product elimination decisions. The supermarket
operations of today while approaching the problem in a
manner similar to the theoretical concepts mentioned earlier
do it with far less sOphisticated techniques. Grashof re—
viewed the process and noted that the chains have two pro-
cedures for the identification of items that should be con—
sidered for deletion. The first method is that one old
product be drOpped for every new one added. Since all
figures indicate that the number of retail food items car—
ried in stock by the average supermarket increases by over
20% each year, one must realize that this rule is not
adhered to 100 per cent.

The second procedure is for a periodic review of
all items the chain handles. This review can be conducted
by the buyer for each product family, the head buyer who
examines all items carried by the chain or, as the trend
is now moving, by the manager of the individual store who
reviews all his own products.

As reported by Grashof and confirmed by later

interviews with three chain executives, the prime factor in

75
the identification of items for possible deletion, as well
as the most important criterion for use in the final deci-
sion is lack of consumer demand for an item as indicated by
a low rate of sales for the item.

Another important criterion is the level Of the
gross margin percentage of the item. By combining these
two factors a third criterion - gross margin dollars gener-
ated per unit of time - can be determined. While this
third method is not as important as the first two, it does
permit a comparison between dissimilar items in a product
family, as well as across product families.

At this point, two other factors should be men-
tioned. First, the chain will view the trend as well as
absolute values for the three criteria stated earlier: and,
second, the chains, in their desire to maintain variety on
the store's shelves, will hesitate to delete one—of—a—kind
items.

Thus, one can see that while the chains have never
develOped a mathematical model for eliminating products
from their stock, they have a set of criteria. Unfortu—
nately, they still appear to make these important decisions
by weighing these three factors and adding a fourth one,
the individual's making the decision, personal interest in

the product.

76
Summary

Chapter II has reviewed the relevant literature
with regard to the adOption and diffusion of innovations,
the relationship between these processes and the consumer's
perception of the innovation's characteristics and the pres-
ent policies of product elimination.

These four areas were chosen for study since this
thesis develOps a new method of predicting which products
should be eliminated from the product line on the basis of
their rate of diffusion into a social system.

The adOption process has been presented as a means
of determining what the final diffusion cycle will be. It
is hypothesized that the measurements of the level of
activity in the early stages of the process can be used to
predict the ultimate outcome of the decision stage.

It might also be noted that while this thesis is
concerned with the consumer's adOption process in an attempt
to predict product elimination it can also be used to sup-
port the notion that product elimination is the central
focus in the decision or trial stage of the retailer's

adOption process.

10.

ll.

12.

77
FOOTNOTES
Everett M. ROgers, Bibliggraphy on the Diffusion of

Ipnovations (East Lansing, Michigan: Michigan State
University, College of Communication Arts, 1968).

 

 

Everett M. Rogers, Diffusion of Innovation (New York:
The Free Press, 1962), p. 23.

 

Everett M. Rogers with F. Floyd Shoemaker, Communica-
tion of Innovations (New York: The Free Press, 1971)
and Thomas S. Robertson, Innovative Behavior and
Communication (New York: Holt, Rinehart and Winston,
Inc., 1971).

 

 

 

 

Rogers with Shoemaker, p. 99.

Rogers, Diffusion of Innovation, p. 17.

 

Rogers with Shoemaker, p. 7.

Bryce Ryan and Neal C. Gross, "The Diffusion of HYbrid
Seed Corn in Two Iowa Communities," Rural Sociology,
VIII (March, 1943), 15-24.

 

Rogers, Diffusion of Innovation, p. 35.

 

Eugene A. Wilkening, Acceptance of Improved Farm
Practices, North Carolina Agricultural Experiment
Station Bullegin #98, 1952.

 

Rogers with Shoemaker, pp. 100-101.

For a good review of cognitive dissonance theory in
marketing, the reader should see Jack B. Cohen and
Marvin E. Goldberg, "Dissonance Model in Post-Decision
Product Evaluation," Journal of Marketing Researgp,
VII (August, 1970), 315-321, for their instant coffee
experiment results, Shelby D. Hunt, "Post—transaction
Communications and Dissonance Reductions," Journal of
Marketing, XXXIV (July, 1970), 46-51 for means of dis-

 

 

Marketing, XXXIII (October, 1969), 50-55.

 

E. Jerome McCarthy, Basic Marketing: A Managerial
Approach, 4th ed. (Homewood, R. D. Irwin and Co.,
1971), p. 194.

 

13.

14.

15.

l6.

17.

18.

19.

20.

21.

22.

23.

24.

78
Robertson, Innovative Behavior, pp. 58-62.
RObert J. Lavidge and Gary A. Steiner, "A Model for
Predictive Measurements of Advertising Effectiveness,"

Journal of Marketing, XXV (October, 1961), 59-62.

RObertson, Innovative Behavior, p. 59.

 

Robert Mason, "An Ordinal Scale for Measuring the
AdOption Process" in Wilbur Schramm, editor, Studies
of Innovation and of Communication to the Public
(Stanford, California: Stanford University Institute
for communication Research, 1962).

Rogers with Shoemaker, p. 101.

RObertson, Innovative Behavior, pp. 61-62.

 

For a discussion of how perceived risk might affect
the individual's purchasing behavior the reader should
see Jagdish N. Sheth and M.‘Venketesan, "Risk-Reduc-
tion Processes in Repetitive Consumer's Behavior,"
Journal of Marketing Research, V (August, 1969), 307-

 

310, Scott M. Cunningham, "Perceived Risk as a Factor
in the Diffusion of New Product Information," in
Proceedings of 1966 Fall AMA Conference (Chicago,

 

1966), pp. 698-721, and Johan Arndt, "Perceived Risk,
Sociometric Integration and Word of Mouth in the
AdOption of New Food Products," 1966 Proceedings,

op. cit., p. 644-48.

 

Robertson, Innovative Behavior, p. 62.

 

Rex R. Campbell, "A Suggested Paradigm of the Indi-
vidual AdOption Process," Rural Sociology: XXXI
(December, 1966), 458-466.

 

Rogers with Shoemaker, pp. 101-03.
Ibid., p. 120.

Alan R. Andreasen, "Attitudes and Customer Behavior:
A Decision Model," Reprinted in Harold H. Kassayian
and Thomas S. RObertson, Perspectives in Consumer
Behavior (Glenview, Illinois: Scott, Foresman and
Company, 1968), pp. 498-510.

 

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

79

John A. Howard and Jagdish N. Sheath, "Theory of Buyer
Behavior," Proceedings of 1967 Winter AMA Conference
(Chicago, 1968), pp. 253-262.

Francesco M. Nicosia, Consumer Decision Processes
(Englewood Cliffs, New Jersey: Prentice-Hall, 1966).

 

gpgg., pp. 153-191.

Robertson, Innovative Behavior, pp. 69-73.

 

Andreasen, "Attitude and Customer Behavior," pp. 498-
510.

Howard and Sheth, 1967 Winter Proceedings. PP. 254-62.

 

For a more complete definition of the different be-
havioral types see George Katona, The Powerful Con-
sumer (New York: McGraw—Hill, 1960), p. 138.

 

It should be noted that Robertson, also, presented a
redoing of the adOption model on pages 73 to 77. His
model was a take off of the earlier models and in-
cluded eight stages. This writer feels confident in
presenting his own summary model in view of the intro-
duction of the Rogers innovation-decision model which
was published after Robertson's work and results of a
case study by Fred D. Reynolds, "Problem Solving and
Trial Use in the AdOption Process," Journal of Market-
ingResearch, VIII (February, 1971), 100-102 which
found evidence for problem—oriented behavior.

 

Rogers has defined indivisible innovations as those
innovations which may not be experimented with on a
limited basis. Rogers, Diffusion of Innovation,

p. 131.

 

Urban B. Ozanne and Gilbert A. Churchill, "AdOption
Research: Information Sources in the Industrial Pur—
chasing Decision," Proceedings of 1968 Fall AMA Con—
ferengg (Chicago, 1968), p. 359. (The reader might
also consult Frederick E. Webster, Jr., "New Product
AdOption in Industrial Markets: A Framework for
Analysis," Journal of Marketing, XXXIII (July, 1969),
35-39.

 

Rogers, Diffusion of Innovation, p. 13.

36.

37.

38.

39.

40.

41.

42.

43.
44.
45.
46.
47.
48.
49.

SO.

80
Rogers with Shoemaker, p. 18-28.
Gerald Zaltman, Marketing: Contributions from the

Behavioral Sciences (New York: Harcourt, Brace and
World, Inc., 1965).

 

Rogers with Shoemaker, p. 175-182.

William J. Stanton, Fundamentals of Marketing, 3rd ed.
(New York: McGraw-Hill, 1971), pp. 206-208.

 

Rogers with Shoemaker, p. 185.

In the remaining listings of generalizations the first
number refers to the number Of studies concerned with
the generalization. The following two refer to the
number of studies supporting and not supporting the
generalization.

ROgers with Shoemaker, Appendix A. It should be noted
that most research in this area has been concerned
with the earlier adOpters. However, Uhl, Andrews and
Poulsen (Journal of Marketing Research, VII, 51-54) in
their article "How Are Laggards Different? An Empiri-
cal Inquiry" present an example of the research being
done in this area by marketing peOple. They found
that laggards may not be as different in terms of some
socioeconomic factors from earlier adOpters. The major
differences were in income, brand loyalty and family
size.

 

Ibid., p. 128.
Rogers with Shoemaker, Appendix A.

Rogers, Diffusion of Innovation, p. 99.

 

Ibid., p. 102.

Arndt, 1966 Winter Proceedipgs, p. 646.

 

Ibid., p. 647.
Ibid., p. 648.

John Summers, "New Product Interpersonal Communica-
tions," Combined Proceedings, Spring and Fall AMA

 

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

81
Conferences, 1971. (Chicago, 1971), 433.
In his 1962 edition, Rogers used the term "interaction
effect" and defined it as "the process through which
individuals in a social system who have adOpted an
innovation influence those who have not yet adOpted".
(Rogers with Shoemaker, p. 138).
Rogers with Shoemaker, p. 161.
Summer, p. 433.
ROgers with Shoemaker, pp. 203-209.
Fred C. Allvine, "Diffusion Of a Competitive Innova—

tion," Proceedings of 1968 Fall AMA Conference
(Chicago, 1968), pp. 341-352.

 

JOhan Arndt, "Role of Product-Related Conversations in
the Diffusion of a New Product," Journal of Marketing
Research, IV (August, 1967), p. 293.

 

 

Joseph R. Mancuso, "Why Not Create Opinion Leaders for
New Product Introductions?" JOurnal of Marketing,
XXXIII (Ju1y, 1969), 20-25.

 

For a more complete look at Opinion leadership, leader
traits and leadership overlap, the reader should con-
sult RObertson pp. 175-189, Rogers, 1971, Chapter 6
and Johan Arndt, Word of Mouth Advertising: A Review
of the Literature (New York: Advertising Research

 

 

Foundation, 1967).
Rogers with Shoemaker, p. 224.

Mancuso, pp. 20-25, Lawrence G. Corey, "PeOple Who
Claim to be Opinion Leaders: identifying Their
Characteristics by Self-Report," Journal of Marketing,
XXXV (October, 1971), 48—53. David B. Montgomery and
Alvin J. Silk, "Clusters of Consumer Interests and
Opinion Leaders' Spheres of Influence," Journal of
Marketing Research, VII (August, 1971), 317-321.

 

 

 

Charles W. King and John O. Summers, "Overlap of
Opinion Leadership Across Consumer Product Cate-
gories," Journal of Marketing Research, VII (February,
1970), 43-50.

61.

62.

63.

64.

65.
66.

67.

68.
69.

70.

71.

72.

73.

74.

82
Mancusco, pp. 20-25.
John A. Martilla, "Word-of—Mouth Communication in the

Industrial AdOption Process," Journal of Marketing
Research, VIII (May, 1971), 173-178.

 

 

Frederick E. Webster, Jr., "Informal Communication in
Industrial Markets," Journal of Marketing Research,

 

Rogers with Shoemaker, pp. 22-23. It should be noted
that in Chapter 5 of his 1962 text, Rogers referred

to the last two stages as "divisibility" and "communi-
cability".

Ibidoo pp‘ 138‘144.

Ibid., p. 145.

Robert D. Buzzell, Robert E. M. Nourse, John B.
Matthews, Jr., and Theodore Levitt, Marketing: A

Contemporapy Analysis (New YOrk: McGraw-Hill, 1972),
p. 175.

 

 

Rogers with Shoemaker, Appendix A.
Ibidol pp. 154-157.
Edwin Mansfield, "Technical Change and the Rate of

Imitation," Econometrica, XXIX (OctOber, 1961), 741-
766.

 

See footnote #19.

Peter F. Drucker, "Care and Feeding of the Profitable
Product," Fortune (March, 1964), 133.

Conrad Berenson, "Pruning the Product Line," Business
Horizons (Summer, 1963), 63, John R. Grashof, "Infor-
mation Management for Supermarket Chain Product Mix
Decisions: A Simulation Experiment" (unpublished Ph.D.
dissertation, Marketing Department, Michigan State
University, 1968), pp. 74-75, and James T. Rothe, "The
Product Elimination Decision," MSU Business_Tgpics
(Autumn, 1970), 45.

 

 

 

Information Obtained in a letter from Dr. Willard R.

75.

76.

77.

78.

79.

80.

81.

82.

83

BishOp, Jr., Director of Research, Super Market
Institute, JUne 7, 1972.

Stated by executives of three major Mid-west chains
during personal interviews in April and May, 1972.

Rothe, 45-52.
Berenson, 64-66.

Philip Kotler, "Phasing Out Weak Products," Harvard
Business Review (March-April, 1965), 107-18.

 

Berenson, 63-70.
Kotler, 107—118.
Paul W. Hamelman and Edward M. Mazze, "Improving

Product Abandonment Decisions," Journal of Marketing,
XXXCI (April, 1972), 20-26.

 

Grashof, pp. 74-79.

CHAPTER I I I
RESEARCH DES I GN

Set forth in this chapter are the framework and
methodology employed in this thesis. The first section of
Chapter III will identify the method of product classifica—
tion used in this study. This will be followed by a dis-
cussion of the sample design, the data collection proce-

dures and the techniques used to analyze the data.

Product Classification

Each year businessmen are confronted with research
studies which list the percentage of product failures for
the previous year's new offerings between 40 and 90 percent.
Yet, as pointed out in Chapter I, most of these failure
estimates are exaggerated. The actual rate of failure
depends of course, on what products are included in the
base against which the failure rate is computed, as well as
the criteria employed to identify failures. Weiss noted
that 80 per cent of the new products studied in these past

Studies are not really new products.1

84

85

In an attempt to overcome the problem of mis—defin-
ing what is meant by the term new product and to provide a
means for obtaining and analyzing new product information
in a more useful manner, a two phase product classification
model was develOped for this study.

The first phase was an attempt to provide the
reader with a more definitive classification of product
newness. A three dimensional matrix was develOped. Each
of these dimensions reflected the different levels of new-
ness perceived by the different members of the marketing
channel; the producers, the middlemen or distributors and
the ultimate consumer.

The producer level was divided into three groups:
distinctly new products, product line extensions and product
improvements. The distinctly new products were substan-
tially different in form, technology or ingredients from
any of the company's previous offerings. Product line
extensions were merely new package sizes, flavors or shapes
of existing products. Product improvements included
changes in existing products, such as changes in taste,
ingredients, appearances or textures.

The distributor segment was categorized by levels
(*5 newness. In descending order, these are new product

tYIPes, new brands and new items. New product types are

86
those substantially different in form, basic ingredients,
and/or method of use in the home from any other product
previously stocked by the middleman in question. New brands
are any brands not previously carried in stock by the
middleman. New items are any products added to stock for
the first time.

The consumer view of product newness is broken into
two segments. The first segment is that of non-perception,
that is the consumer doesn't really perceive the product as
being new. Products in this category included those items
which while considered new by the manufacturer or middleman
fail to provide the consumer with any additional utility
over the previous offerings. An example of this type of
newness is the unobservable improvement of a laundry soap.
Thus, while the soap package might declare "new, improved",
the customer considers it the same as his old package of the
product.

The other segment of the consumer dimension refers
to cases in which the consumer does perceive a difference
in the product. Here the consumer perceives the product as
laew'to the firm, that is the firm produces no product which
.Possesses a positive cross-elasticity with the product under
question. The second grouping refers to the conditions when

no firm produces a product with a positive cross-elasticity.

87

By grouping these three dimensions together, it is
now possible to explain what is meant by the term "new
product" in a more definite manner. Now it is possible to
explain the term in 27 different ways, with each way taking
into account the level of newness as perceived by the mem-
bers of the channel. An example of these can be shown by
referring to Figure 3-1 where the box selected is marked
with an X. The X in this box indicates that the new prod-
uct selected was considered to be a product improvement by
the producer, a new brand by the middleman, and a product
which is not in direct competition with other brands of the
manufacturer.

However, the introduction of a 27 matrix diagram
for defining the level of newness of a product only answers
half of the question of how to define a new product in a
more definitive manner which will be useful as a tool of
market prediction? Some manner of utilizing product charac-
teristics as a means of determining the market mix must be
develOped. The traditional approach to the classification
(XE goods has been that of convenience, shopping and
S£>ecialty goods. The definitions of these goods are based
0ft consumer buying habits. Miracle, in a revision of an
eaJi‘lier work by Aspinwall notes that this is not an alto-

geiiher satisfactory solution as they focus on consumer

Producer Dimensions

 

88

Figure 3-1

New Product Dimension Model

Buyer Dimensions

Non-Perceptive

Perceptive

No firm

substitute

No substitute

 

Distinctly
new
products

 

 

 

 

Product
line
extensions

 

 

 

Product
improveme ts

 

 

 

 

 

 

New

New

New

New

New

New

New

New

New

Product types

Brand

Items

Product types

Brand

Items

Product types

Brand

Items

 

snotsuauta iounqtinsrq

89
behavior and can't answer all questions as to why a con-
sumer "shOps" for some goods and not for others. Miracle,
thus, redefined consumer and market characteristics in
order to develOp a single list of characteristics. The
list consists of
1. Unit value

2. Significance of each individual purchase to
the consumer

3. Time and effort spent purchasing by consumers
4. Rate of technological change

5. Technical complexity

6. Consumer need for service

7. Frequency of purchase

8. Rapidity of consumption

9. Extent of usage

Using these characteristics and their interdepend-

ence he projected five product groups.

9O

 

 

 

Product Characteristics of Five Groups
Product Group
characteristic I II III IV V
1 Very low Low Medium High Very high
to high
2 Very low Low Medium High Very high
3 Very low Low Medium High Very high
4 Very low Low Medium High Very high
5 Very low Low Medium High Very high
to high
6 Very low Low Medium High Very high
7 Very high Medium Low Low Very low
to high
8 Very high Medium Low Low Very low
to high
9 Very high High Medium Low to Very low
to high medium

product classifications described earlier,

By utilizing these product groupings with the new

it is felt that

a contribution for research methodology has been made by

this thesis.

A contention of this thesis is that this new

categorization will be useful in attempts to relate the

behavior of new products in comparison with other new

products of the same category.

The author's summary model (Figure 2-5) prOposed in
Chapter II was an attempt to overcome the shortcomings of
the previously mentioned adOption models. The model made

allowances for these shortcomings by:

91

l. The model accounted for both rational and non-
rational behavior as it is possible to go
directly from the knowledge stage to the
adOption-rejection stage or to make use of
the intermediate stages.

2. The model made allowances for problem solving
situations as it may begin with either the
point of problem perception or knowledge.

3. The model made allowances for all possible
consumer behavioral patterns by allowing one
to skip some stages and/or by the use of feed-
back to redo others.

However, this thesis is primarily concerned with
the activities of the consumers in their rational consider-
ation of a new product. Thus, the research design made an
attempt to measure the amount of consumer activity in the
different stages of the process before adOption or rejection.

The model for this research was conceived as begin-
ning with the knowledge stage, which commences when the
individual is exposed to the innovation's existences

(awareness) and gains some understanding of how it functions

 

(knowledge). Most past research studies have conceptual-

 

ized awareness as occurring due to random or nonpurposive

. . . . . . 3 .
actiVities of the indIVIdual. However, knowledge—seeking

92

was felt to be an initiated and not a passibe activity.
The predispositions of individual influence his behavior
toward communication messages and to the responses these
messages generate. Hassinger notes that even if an indi-
vidual is exposed to messages concerning the innovation,
there will be little effect of such exposure unless the
individual perceived the innovation as relevant to his needs
and is consistent with his existing attitudes and beliefs.4

At the persuasion stage the individual forms a
favorable or unfavorable attitude toward the innovation.
Whereas the mental activity at the knowledge stage was
mainly cognitive (or knowing), the main type of thinking
at the persuasion stage is affective (or feeling).

At this stage the individual becomes more psycho-

logically involved (interest seeking) with the innovation.

 

He actively seeks information about idea. His personality,

 

as well as his social system's norms, will affect where he
seeks out this information, what messages are perceived,
and how they are interpreted. It is at this stage that a
general perception of the innovation is develOped. Such
perceived attributes of an innovation as its relative
advantage, compatability, and complexity are eSpecially
important at this stage.

In forming a favorable or unfavorable attitude

93
toward the innovation, the individual may mentally apply
the new product to his present or future needs. As he
progresses with his mental application, he will seek rein-
forcement of his attitude toward the new product. The
individual is likely to seek convictions that his attitude
is correct from peers by means of interpersonal communica-
tion channels. Mass media messages are too general to pro—
vide the specific kind of reinforcement that the individual
needs to confirm his beliefs about the new product.

At the decision stage, the individual engages in
activities that will lead to a choice to adOpt or reject
the innovation. This decision is confirmed or rejected at
the final stage of the model, the confirmation stage.
Throughout this terminal stage, the individual seeks to
avoid a state of dissonance or to reduce it if one occurs.

Thus, this thesis is an attempt to measure the early
activities of the adOption process which occur before the
decision stage and compare these activities with products
from the same classification by use of the previously de-
fined methods. The hypotheses of this study contended that
either the absolute measurement of these early activities
or the changes in their relative growth can be used to pre-
dict which products should be eliminated from a firm's line

at a point earlier in time than in present use.

94

Hypotheses

 

This thesis will provide a model which should be
able to show through measurement of the initial phases of
the consumer's adOption which products are unlikely to be
retained by store management after a thirteen week analysis
of sales data. Thus, the model presented in this thesis
will identify those products which are prime candidates for
elimination from the firms product line. If these products
are not eliminated then the model will predict a very low
probability of success for them.

The more specific hypotheses of the study follow.
These hypotheses are listed in the null.

1) The knowledge of the 13331 of consumer adop-
tion process variables (awareness of the new product,
knowledge of its product type, weak and strong interest in
the product and weak and strong information seeking activi-
ties toward the product) within an earlier period of time
makes no difference in management's ability to identify
products, which, according to its criteria, should be con-
tinued relative to those which should be discontinued.

2) The knowledge of the £333 pf growth in consumer
adoption process variables (awareness of the new product,
knowledge of its product type, weak and strong interest in

the product and weak and strong information seeking

95
activities toward the product) within an earlier period of
time makes no difference in management's ability to identify
products which, according to its criteria, should be con—

tinued relative to those which should be discontinued.

Sample Design

 

The Sampling_Frame

 

The data to be used for this study was obtained by
means of a phone survey of supermarket shOppers in the Des
Moines, Iowa market. Several regional and national chains,
a major voluntary chain and a number of strongly competitive
local supermarkets are presently Operating in the Des Moines
market. Among the local supermarkets is the Abel Chain (a
fictious name) which was chosen for COOperation in this
thesis. Abel, which has nine stores in Des Moines, accounted
for 26% of the city's 1971 retail grocery sales. With the
COOperation of the chain's tOp management one of these nine
stores was randomly selected to participate in this study.
The selection of Des Moines was fortuitous. Any analysis
of the results of this study with the total United States
population can only be made insofar as the consumers of the
above chain in Des Moines are representative of the United
States market.

The selected store's trading area (Appendix A shows

96
the 1970 Census Tract data for Des Moines, Polk County and
the selectes store's trading area) covers approximately
twenty-eight miles and has a pOpulation of over 20,000 in
some 5,500 households. The trading area's boundaries in—
clude an interstate highway system on two sides, a major
east-west thoroughfare and the city's incorporation limits.
These boundaries are in agreement with the research findings
Of Cox and Cooke5 in their study of dimensions involved in
shOpping preference.

The selected store had a weekly sales volume of
over $100,000 and an average inventory of over 11,000 food
items. A major reason for selecting this particular chain
was the fact that it has competed successfully in this mar-
ket without the use of any means of advertising. While it
is noted that national advertising will have some effect on
consumer behavior, the retailer will consider this to be
part of the total product being offered to the consumer.
Thus, this thesis has attempted to eliminate the effects of
local sales promotion from its results.

With the COOperation of store management and repre-
sentatives of Des Moines' food wholesalers, seven products
were chosen for this study. The basic criteria for selec-
tion was that all the products come from the same cell of

the matrix of this study's prOposed new product

97

6 and that each product was from the

classification system
same product group in the Miracle classification. A further
set of criteria was placed upon the selected products. The
products could not be ones in which Des Moines was to serve
as a test market for analysis by their producer. These
products were withdrawn from consideration since the market-
ing variables could be altered by the manufacturer or dis—
tributor. Likewise, products being introduced with either

a sampling or couponing campaign were not considered. These
promotional strategies were felt to be capable of presenting
an unfair bias to the six predictor variables chosen for
analysis in the five weeks past introduction. The Abel
Chain also agreed to hold prices constant for the products
and not to vary the amount or location of shelf space during
the initial 13 weeks. The seven products selected which
were considered to be a product improvement by the producer,
a new brand by the middleman, a product which is not in
direct competition with other brands of the manufacturer

and belonging to same product grouping, according to
Miracle's model. Since the study concerned itself with new
products distributed through retail food outlets, the prod-
ucts were all members of Miracle's second group. The com-
bination of this product grouping, along with the level of

newness mentioned above, resulted in the largest number of

98

products available for study during the summer of 1972.
The seven products included a:

1. cooking oil

2. ready-to-eat meal in a can

3. floor cleanser

4. furniture conditioner

5. snack food

6. fabric softener

7. instant dessert

Data Collection

 

After the selection of the products a phone survey
of the selected store's customers was conducted to determine
the amount of customer awareness, knowledge of product type,
strong and weak interest in trying the product and strong
and weak information seeking activities with regard to the
seven products at the end of the second, third, fourth and
fifth week after introduction. A COpy of the questionnaire
used is shown in Appendix B.

In an effort to determine the sample size needed
for this study, the following assumptions were made. First,
a 95% confidence level with a maximum of a 3.0% error in
estimating prOportions in the 25%.to 30% range was selected.

These confidence limits were in agreement with previous

99
studies in the area and store management estimated that
approximately 30% of their customers become aware of a new
product during its initial month of introduction. By sub-
stituting this above information into the formula for deter—

mining sample size ('/i = /E§), a sample size of 800 was
2' n

19.2 z/(17Yt-73l
2 n °

 

determined for this experiment,

Nevertheless, while the store's manager did not
know what was the total number of regular customers for his
store, it was shown in Appendix A that the number of house—
holds in the store's trading area was 5,552. Thus, the
sample of 800 households for interviewing can be considered
as being greater than ten per cent of the population.

Since past studies by the Drake University Research
Center indicated a completion rate of 66%, an effort was
made to randomly select 1,200 households for interviewing.
In order to assure that these 1,200 households were randomly
selected, a list of 3,600 households was prepared during
the three weeks prior to the introduction of the new prod-
ucts. This prepared list was derived by selecting auto-
mObile license plate numbers from the supermarket's parking
lot in prOportion to that day's sales volume and tracing
them with the assistance of the State of Iowa's Motor
Vehicle Registration Office. The Abel Chain does not make

either hourly or intraday cash register tape readings so

100
that the license plates were not selected in accordance
with the store's hourly sales volume. To overcome this
prOblem and prevent any sampling bias a random selection of
the store's hours was performed by groupings the hours into
four groups of four hours each and randoming selecting two
groups for each of the twenty ones in which the license
plate numbers were gathered. Thus, a total of 3,600 license
plate numbers were selected and these numbers were grouped
into 1,200 groups of three names each.

The phone survey was Operationized by.randomly
selecting the second number from each group as the one to
be called first and then proceeding to the first, then the
third number of the group is no response could be gathered
from the original selection. The interviewer would start
with the second name from each group, regardless of which
household she contacted in the previous group. If the name
selected belonged to a non-household or had a unlisted num—
ber, the interviewer was to go to the next number in that
group. This grouping of names was an attempt to give all
households using the store's facility an equal chance of
being interviewed and reduce the number of non-completions
by eliminating all non-households. A total of 200 inter-
views a week were made in this manner with fifty households

being asked about one product and three groups of fifty

101
households about two products. The products were rotated
each week. The calls were, also, rotated between product
groups, so as not to produce a unfair day—product bias.

The survey Operation was conducted during the sum-
mer of 1972. Two female interviewers with prior instruction
and identifying themselves as being from the Drake Univer-
sity Business Research Center were hired for the study. A
copy of the procedures followed by these interviewers is
shown in Appendix B.

These interviewers followed the questionnaire shown
in Appendix B. This questionnaire was pretested by 25
senior level marketing students in an effort to remove all
ambiguity. The final copy of the questionnaire was again
pretested by a random selection of 25 homemakers from the
Greater Des Moines area in an effort to confirm its meaning-
fulness. Thus, it is felt to be fair and unbiased.

The data were recorded according to the procedures
shown in Appendix B.

These procedures were followed as the female inter-
viewers contacted fifty households per week for each of the
seven products. The thesis assumed that the fifty house-
holds selected per week for each product possessed a common
homogeneity in their buying behavior. The data were then

recorded in terms of percentages of households who responded

102
affirnatively for each predictor variable for that week.

These percentages were non-cumulative from week to week.

Analysis of the Data

The approach to analyzing the data was twofold.
Firmst, linear discriminate analysis was used to test if the
weeflcly percentages of the six predictor variables chosen
were able to differentiate between the continued and dis-
ccnitinued products. Second, the rate of growth hypothesis
was: tested by means of eighteen independent t—tests. Also,
th>—way analyses of variance were performed on the mean
rates of growth between the continued and discontinued
SIOUps by the six predictor variables for the same time
IPeriods as used in the t—tests thus resulting in a 2'by 6
design. These time periods were the third-second week,
fourth-third week and fifth-four week.

This research used the decision rule that if the
teStvalue of any test exceeded the critical value of .02
the hypotheses was rejected.

Linear discriminate analysis was chosen to deter-
‘mine if some function could be used to separate the two
Product groups (the continued and the discontinued) on the
basis of the level of the six predictor variables chosen

for this study. The major advantage of this particular

 

103
statistical tool is that it provides the researcher with a

function that best discriminates between the continued and

 

discontinued products. Also, since the research assumed
that the data was obtained from a multivariate normal popu-
lation, such statistical tools as analysis of variance
analysis were unusable. This is because analysis of vari-
ance is only able to use data from a untivariate sample.
Linear discriminate analysis, thus, assumes that the
dependent variable must be a dichotomy, there must be a
random sample, that the relationship between the independent
and the dependent variables is linearity. There must be a
normal distribution and there must be a homogeneity of
variance.

Other tools such as multiple regression and canonical
analysis could also have been used in certain situations
when working with multivariate data. However, it was the
intent of this research to determine if a function could be
derived Which could serve as a means of predicting which of
the two product groups a new product would ultimately belong.
Thus, multiple regression which is used to provide a con-
tinuous function, and canonical analysis which seeks only
to determine the linear combinations of the predictor vari—
ables and the two product groups that are very highly corre—

lated with each other were not used.

 

 

104

Green and Tull listed the major Objectives of dis-
criminate analysis. They are:

1. Determining which variables account most for

intergroup differences in average profile.

2. Determining whether significant differences
exist among the average "score" profiles of
two (or more) a priori defined groups, assuming
group covariation and dispersion are equal and
the distributions are multinormal.

3. Determining linear combinations of the pre-
dictor variables that be used to represent the
groups by maximizing among-group relative to
within-group separation.

4. Establishing procedures for assigning new
products whose profiles, but not group identity,
are assumed to be from one of the a priori de-
fined groups.

An example of the use of linear discriminate analy-
sis can be shown by assuming that Table 3—1 is the result
for the second week. (Since it is unknown how the test
results will end, four products will be placed into each
group.)

The linear discriminate function for this hypothet-

ical example is B = 1.00x1 + 4.69x2 - 2.54x3 + 11.24x4 +

 

105

 

 

 

 

 

 

sm~.m Nmk.m Noo.m Nm~.e Moo.e Nom.m weaned can:
rim H IJ. alum. rim Ilm
e e N m e N muoocoum
m o m o c M
Na we we Na ne Na pussflusooman
Noo.e Noe.m Nm~.e Nom.e Nm~.e Nmk.m weaned ewe:
Ilm. .llm llm im llm llml
o m N m m 0H muospoum
m m m o o m
we Nm No Nu Nw Nm poscwucoo
wcouum xmuz msouum xmoB mmpuasocx mmucuum3<
lwcfixoom aowumeuomaH umwumucw

 

Amuusnauuum cosmoupsfi mcammummoa madame nuaswcou mo ummusouuomv

mmqm<Hm<> MOHUHQmmm me mmH mo
mam>mg xmm3 9200mm A<UHHmmHomwm

Him mam<H

106
16.34x5 + 7.89x6 where x1 is awareness of the new product,

x2 is knowledge of its product type, x is having a weak

3
interest in the product, x4 is having a strong interest in
the product, x5 is undergoing weak information seeking acti-
vities concerning the product, and x6 is undergoing strong
information seeking activities concerning the new product.

One can see from Table 3-2 that weak information
seeking activities make the greatest contribution in dis—
criminating between the two groups. Its importance value,
69.45, may be interpreted as the contribution this variable
makes toward over-all product continuation. Strong infor-
mation and strong interest are the only other variables
with a relative importance of greater than 10%.

If the six predictor variables scores are placed
into the discriminate equation and a 8 value is determined
for each product, the products can be assigned to the con-
tinues or discontinued category on the basis of their being
closer to 5c = 247.38 or id = 131.10 where EC and 2d

represent the average means of the continued and discontinued

products. The results are shown below:

Predicted
Continued Discontinued Total
Continued 4 0 4
Actual Discontinued .9 §_ .3
Total 4 4 8

 

107

at»

Vii

 

 

 

OH.HmH n muospoua pwscwucoomfip

IN

 

mm.mq~ u muospoua poscfiucoo w
as 0e.HN 00.e 00.0 00.0 eoaaaeuoeeH weouam
mm mq.mo em.oa mm.m 00.x cowumauomcH xmoz
HM mo.aH em.HH oo.m mm.q umuuoucH mcouum
m mm.n am.~i mu.¢ om.o umououaH xmuz
0 mm.0s 00.0 00.4 mm.0 mweaeaoex
e mm.m oo.H om.~ mm.m mmucoum3<
uocmuuoaaH mucouu0deH unwfioz and: cmoz
o>wumamm mumsfiefiuomwa oscwusoomfin oscfiusou moapmfium>

 

Mama 0200mm “mm04<> moz<am0m=H oz< .msmonz
mseszHmomHn .mmam<He<> mosoHemmm so mm04<> zen:

Nlm mqm<8

108

These results indicate that the function makes no
assignment error. Presumably, this function could be used
to categorize new sets of data for the same time period.

Individual tests were used to determine whether or
not there was a significant difference between the means of
the two sample groups' rate of growth.8

An example of the use of t-tests can be shown by
assuming that Table 3-3 shows the rate of growth for the

third-second week. The test values for the six variables

are :

awareness 2.726 strong interest 4.404
knowledge 3.750 weak information 6.109
weak interest 2.896 strong information 6.123

Since the critical value for t is 3.143 at level of .02
with six degrees of freedom, therefore, the thesis is able
to discriminate between continued and discontinued products
by the levels of knowledge, strong interest and weak and
strong information seeking activities. This conclusion is
reached because the research is able to reject the hypoth-
eses of the equality of means in all four cases.

A two-way analysis of variance analysis was per-
formed on the mean rate Of growth to determine whether
there were significant differences due to the product pre—

dictor variables and also whether there were significant

 

109)

ll-
[W‘s

 

 

 

 

 

 

 

Amo.iv Amo.rv «N. HH. mH. om.

Ho. 0. wH. am. mm. mm. muoscoum

mo. No. Hm. Hm. mH. 00.

No. Aoo.|v mm. ma. wH. me. cmocHusoomHa

NH. OH. me. me. am. no.

«H. NH. mm. mm. Hm. me. nauseoum

do. mH. Hm. Ho. mm. mm.

mo. «H. mm. no. mm. mm. poacHuaoo
wcouum xwoz mcouum Moos owpmHaosM mmocmum3<
.wdemom soHumeuomsH umouuucH

«3
Am: i mzv x663 paooomlpanH

mmHm<Hm<> KOHUHQmmm NHm mmfi
mom mfizomu mo MH<M H<UHHmmHomwm

Min mHm<H

110

differences due to the product predictor variables and also
whether there were significant differences due to continua-
tion or discontinuation of the seven products. This par—
ticular statistical tool was chosen since the data to be
analyzed was in the form of a ratio. In this case the re-
searcher had no knowledge as to whether the data would still
fall into multivariate patterns. Therefore, by using the
means of the two product groups, the researcher was able
to reduce the problem to a univariate situation and hOpe-
fully test for the significance of the difference between
the mean ratios of the predictor variables and the product
groups.

This technique assumed the model

Yij = u + “i.+ Bj + Eij i = l=2; j = l-o

where Yij was the mean rate of growth due to the i-th level
of factor A (the continuation or discontinuation of the
product) and the j-th level of factor B (the six products
predictors). Where further a1 is continuation of the
product and a2 is the discontinuation of the product. Bl

is the level of awareness, 32 is the level of knowledge of
product type, 33 is the level of weak interest activities
toward the product, B4 is the level of strong interest acti-

Vities toward the product, is the level of weak infor—

B5

mation seeking activities toward the product and B6 is the

r";

 

111
level of strong information seeking activities toward
the product. u is an overall mean effect, ai is the
effect due to the i-th level of factor A and Bj is the

effect due to the j-th level of factor B. 's are the

6..
13
errors which are assumed to be normally independently dis-

9
tricuted with a mean 0 and variance 02. p

 

The results of the analysis of variance for the FE
hypothetical example are in agreement with the t-tests. F4—
The results Of Table 3-4 show that the means are signifi-

cantly different between the six predictor variables.
Also, the means between product groups are significantly

different. Thus, the hypotheses can be used as predictive

 

 

tools.
TABLE 3-4
ANOVA TABLE FOR RATES OF GROWTH OF THE
HYPOTHETICAL THIRD-SECOND WEEK
Source d.f SS MS F C.V.F@ 2%
Products 1 .1386 .1386 19.25 8.25
Product
Variables 5 .5105 .1021 14.18 12.13
Error 5 .0358 .0072

 

112
FOOTNOTES
E. B. Weiss, "That Malarky About 80 Percent of New

Products Failing," Advertising_Age, XXXVI (August 2,
1965), 101.

 

Gordon Miracle, "Product Characteristics and Marketing
Strategy," Journal of Marketing, XXIX (January, 1965),
18-24 0

 

Everett M. Rogers with F. Floyd Shoemaker, Communication
of Innovations (New York: The Free Press, 1971), pp.
104-105.

 

 

Edward Hassinger, "Stages in the Adoption Process,"
Rural Sociology, XXIV (March, 1959), 52-53.

 

William E. Cox, Jr. and Ernest F. Cooke, "Other Dimen-
sions Involved in Shopping Center Preference," Journal
of Marketing, XXXIV (October, 1970), 12-17.

 

A survey of 30 Des Moines homemakers was used to deter-
mine the buyer dimension. The distributor's and pro-
ducer's dimensions were obtained from interviews with
Des Moines retail and wholesale food executives.

Paul E. Green and Donald S. Tull, Research for Marketing

 

Decisions (Englewood Cliffs, N.J.: Prentice-Hall,
1970), pp. 368-377. The reader might also consult

 

For a detailed explanation of the test see John R.
Stockston, Introduction to Business and Economic
Statistics (Cincinnati: Southwestern Publishing Co.,

 

1966), pp. 251-264.

Maurice G. Kendall and Alan Stuart, The Advance Theopy
of Statistics, Volume III (New York: Hafner Publishing
Company, 1966), pp. 1-54.

 

 

 

m

 

CHAPTER IV

PRESENTATION OF FINDINGS

The objective of Chapter IV is to present the find-
ings of the study. The chapter is divided into three sec-
tions and follows the general outline of the hypotheses as
presented in Chapter I. The first section will review the
management decision with regards to the continuation or dis-
continuation of the seven products studied in this thesis.
Section two will analyze the six hypotheses concerned with
the level of the weekly percentages of the six predictor
variables. The final section is concerned with the changes

in the six variables' rate of growth as a prediction tool.

The Management Decision

 

The management review of the new products found
that three of the seven products were considered to be a
complete success at the thirteen weeks post introduction
stage, one product showed signs of success as its sales
level had doubled during the past six weeks, and three

products were deemed by management to be non-successful

113

 

114
and were discontinued. It should be noted that the store
manager was reluctant to discontinue one of the three un—
successful products, because while its sales were small
(less than two percent of products of a similar nature) the
product was felt to be of a superior nature. Another one
of the unsuccessful products was considered to be a failure
by the wholesaler as well as by the retail management. In
fact, while the retail manager had already decided in dis-
continuing the fabric softener, the store chosen for this
study experienced the highest sales level of the 256 stores
serviced by the food wholesaler.

The three products deemed to be complete successes
were the floor cleanser, the furniture conditioner and the
instant dessert. The other continued product was the snack
food. The discontinued product which management was re-
luctant to discontinue because of its superior nature was
the cooking oil. The fabric softener and "ready to eat
meal in a can" were the other products discontinued.

Management felt the reasons for the three products'
lack of success were as follows

1. The cooking oil was produced by a small re-

gional producer and had no introduction
advertising campaign. The product was more

expensive than other similar products and its

 

115
physical appearance was also different.
The supposedly superior quality could only
be determined by use.
2. Both the fabric softener and the "ready to
eat meal in a can" were expensive and not
felt to be superior to other product offerings.

Weekly Percentages of the Six Variables
as Prediction Tools

 

r21

 

 

The percentage levels of the six predictor variables
for the four weeks studied are shown in Table 4-1 to Table
4-4. The tables, also contain a mean rating of both the
continued products and discontinued products for each of
the predictive variables.

As mentioned in Chapter III the Objective of dis—
criminate analysis is to product a linear function that will
discriminate between continued products from discontinued
products.1 Weights are assigned to the variables such
that the ratio of the difference between the means of the
two groups to the standard deviation within groups is maxi-
mized. The linear discriminate function can be expressed as:

Z = wlxl+ w2x2+ . . . +

“axe
where xl . . . x6 represent the independent variables (in
this case awareness, knowledge, interest and information

seeking activities) and w1 . . . W6 represent the

116

 

 

 

 

 

 

N0. 004 00.0 3.0 3.0 00 madame can:
0 0 0 0 0 00 do 0:380
O O c 4 OH ON emu OH How:
N a O O O NN possumom OHunmm
nauseoum OwscHucouwHO
N OH m.OH m.mH NN ON mcHuwu sou:
O NH «H OH OH «N nonsmoHo coon
N O a NH OH eH coom Human
N O OH ON ON mm uOOOHuHOcoo uuauchsm
4 OH «H On an on uuummua
nauseoum cmscHusoo
wsouum sous mcouum sou: owOuHaocx mmmcuumsa
.wawamm GOHumEpowcH ummuuucH

 

AmounnHuuum emuMOHch wchmmmooa uHasmm Huesmcoo mo owmucmoummv

mmqm<Hm<> “OHUHommm me mmh ho mHm>mH MMN3 onomm

Hie mqm<H

117

 

 

 

 

00.0 00.0 00.0 00.0 00.00 00.00 000000 000:
0 0 0 0 00 00 000 0000000
0 0 0 00 00 00 000 00 000:
a a N OH ON 0N umcmuwom OHHAmm
muoapoum vmscHuaoomHO
0.0 00 00 0.00 0.00 0.00 000000 0002
0 00 00 00 00 00 00000000 00000
0 00 0 00 00 00 0000 00000
0 0 00 00 00 00 00000000000 000000000
0 00 00 00 00 00 0000000
muoscoum posaHusoo
wcouum x003 wsouum xmmB mmpmHsoGM mmmamuma<

 

 

wconom :0H0m80oMCH umuuouaH

 

AmouspHuuum OODMOHOGH wchmommoa OHOEOO uoasmsoo mo owmuaoouumv

mMHm<HM¢> MOHUHQNmm me mmh mo mHm>MH Mmm3 QMHmH

qu mHm<H

 

 

 

 

 

0 0 00.0 00.00 00.00 00 000000 000:
0| .0| 0 00 00 0.00 000 0000000
O OH O ON 0m «m use OH H002
O NH O «H On On Housewom OHuncm
nauseoum OmscHucoomHO
8
n 0 0.: 00 00 00 0.00 000000 0002
00... 00 .04 0.0. mm. 00 0000000 00000
O O NH NH Nm Nm coom xomcm
O NN NH eN Oe Oq HOGOHqucoo upsuchsm
NH ON OH ON Om Nn uuummon
muospoum coscHusoo
wcouum xmma wcouum xmuz owOOHBocM mmmduum3<
.wstomm aoHumauoOCH umuumucH

 

Amouanuuus OOHOOHOGH wchmmmOOQ deamm uneconou mo uwmuaoouumv

mMHm<Hm<> MOBUHQmmm NHm NEH mo mHm>mH MNW3 EHMDOm

mtg mHm<H

119

 

 

 

 

 

00.0 00.0 00.0 00.00 00.00 00.00 000000 000:

IIIAw iIIAMI iIIAWI «H Nm Nm HHo wconoo
0 0 00 00 00 00 000 00 000:
0 00 0 00 00 00 00000000 000000

muozpoum OmscHuaoomHO

0.0 00 00 00 00 00 000000 000:

.IM wiw MOI. dim 90. 40.0.0. 000000003 .0002
0 00 00 00 00 00 0000 00000
0 00 00 00 00 00 00000000000 000000000
0 00 00 00 00 00 0000000

muuscoum coscHucou

 

maouum x003 wcouum x003 owcuHaosx mmmcuumz<
wcquum coHumauomcH umuuousH

 

 

 

AmouspHuuum pouOOHpsH wcHOOOOOOQ mHoemw uoasmcoo mo mwmucmouwmv

mmHm<Hm<> MOHoHommm me mmh mo mHm>mH Mmmz :HhHm

«I« MHQ<H

120
discriminate coefficients or weights to be applied to the
independent or predictor variables. 5 represents the
product's point score. Based on the point score, it should
be possible to predict continued and discontinued products.

Mosteller and Wallace have prOposed a method for
determining the relative importance of the independent
variables. The value of the relative importance index
measures the contribution of each variable to the difference
in the average point scores between the gwo groups (EC — ED),
in which EC refers to average mean score of the continued
products and 8 is the discontinued products.

Given that one mean value is exactly at the average
of the continued group and another at the average of the
discontinued group, then the difference in score is a mea—
sure of the importance (Yi) of the variables, indicating
the contribution it makes to the total difference in con-
tinued versus discontinued point scores.

Yi = wixic - wixiD = wi (xic - xiD)

where wi is the discriminate weight for the variable under

consideration while iic is the mean score of the continued

products and i is the mean score of the discontinued

iD

sample.2

 

121
Second Week Findings
From the information given in Table 4-1, the linear
discriminate function for the second week was

3 = -l.0075x1 + 2.9025x2 - 4.0875x3 + .945x4 + .285x5 + 7.30x6

If x1 is made an unity value the equation becomes

2 = x1 - 2.89x + 4.06x3 - .938x - 2.83x - 7.25x

2 4 S 6 f

where x1 is awareness of the new product, x2 is knowledge r”

of its product type, x3 is having a weak interest in the

 

product, x4 is having a strong interest in the product, x5
is undergoing weak information seeking activities concern-

ing the product and x is undergoing strong information

6
seeking activities concerning the new product.

The fact that four of the six predictor variables
are negative should not be alarming. This was caused by
the among group and between group interrelationship of the
variables.3 Also, it must be noted that the relative im—

portance of x x and x is only 15% of the total.

4' 5 6

Thus, one can see from Table 4-5 that weak interest
made the greatest contribution in discriminating between
the two groups. Its importance value, 60.21, may be inter—
preted as the contribution this variable makes toward over-
all product continuation. Knowledge was the second greatest

contributor even though its discriminate weight was negative.

No other variable had a relative importance of greater than

122

10%, thus, one can conclude that knowledge and weak interest
wemzthe two most important predictors of product continu-
ation at the second week past introduction.

The significance of this discriminate function can
be evaluated by using Fisher's 1936 test. Fisher took the
variance 8 from the difference of the Observed mean values ‘

16.41 - 5.57

where Var Z = = 2.15. This produced the vari-
5(9l.0075)

ance of a single value. The variance of the difference of

 

two means, one of four and the other of three numbers, is
1% of this, namely 1.25, giving a standard error of 1.18.
The Observed difference of the means, 10.84, is nine times
as great as the standard error and we conclude that the dis-
criminate is likely to be effective.

The prObability of misclassification is easy to
estimate. The standard deviation of B is «5:15 = 1.47 and

the distance between the ED + B means is 10.84. One half

C
of this is 5.42 and 5.42/1.47 = 3.69 which is significant
at .02.4
Nevertheless, while the function has been found to
be statistically significant, it should also be able to
provide a useful tool as an assignment for predicting prod-
uct continuation on the basis of the six variables used in

this thesis. Thus, if the six predictor variables scores

are placed into the discriminate equation and a 8 value is

123

TABLE 4-5

MEAN VALUES OF PREDICTOR VARIABLES, DISCRIMINATE
WEIGHTS, AND IMPORTANCE VALUES; SECOND WEEK

 

 

 

 

 

Dis-
Continue continue Discriminate Relative
Variables Mean Mean Weight Importance Importance
Awareness 28.00 20.00 1.00 8.00 62
Knowledge 22.00 8.67 —2.89 38.52 31%
Weak Interest 19.50 4.67 4.06 60.21 48%
Strong Interest 10.50 3.33 - .94 6.74 5%
Weak Infor-
mation Seeking 10.00 1.33 - .28 2.43 2%
Strong Infor-
mation Seeking 2.00 .67 -7.25 9.64 8%
EC a 16.41 Z = 5.57
Discriminate Value - 10.99
TABLE 4-6
SECOND WEEK RESULTS OF ASSIGNMENT TESTS
Z 3
Dessert 12.88 Fabric softener 10.10
zFutniture conditioner 18.48 zMeal in can 3.59
zSnack £000 13.83 3Cooking 011 3.00
zFloor cleanser 20.41
Predicted
Continued Discontinued Total
Continued 4 0 4
Actual Discontinued Q_ .2 _3
Total 4 3 7

 

 

124
determined for each product, the products can be assigned
to the continued or discontinued category on the basis of

their being closer to EC = 16.41 or ED = 5.57. Where

BC and ED represent the average means of the continued and
discontinued product. This was done for the second week
figures with the results shown in Table 4—6.

The results shown in Table 4-6 illustrate that the
second week discriminate function made no assignment error.
Presumably, this function could be used to categorize new

sets of data on products after only two weeks of introduc-

tion to predict continuation or not.

Second Week - Summary

 

The second week discriminate function was seen to
have been both significant at the .02 level and useful as
a tool for classifying products from unknown groups to their
prOper group. Two variables were found to be more impor-
tant, weak interest and knowledge, at this point in weak

two as a means for differentiating between the two groups.

Third Week Findings

 

From the information presented in Table 4—2, the
linear discriminate function for the third week was
8 = —.0407xl - .523Ox2 + 2.7852):3 - 1.0022):4 - 1.9787x5 —

7.4785X6.

125

Table 4-7 shows the mean value of the predictor
variables, the discriminate weights and the importances
values for the third week. Once again weak interest makes
the greatest contribution in discriminating between the two
product groups. However, knowledge is no longer the second
greatest contributor as weak information seeking activities
now account for 30% of the difference between the two
groups. Strong interest, strong information seeking activi-
ties and knowledge are the next in order of importance.
Awareness is very unimportant as a predictor tool for the
third week.

Once again by using Fisher's 1936 test the signifi-
cance of the third week's discriminate function can be
established.

Var Z = 674.29 - 112.28 = 2761.72

5 (.0407)
Var (EC - ED) = Var a (1/4 + 1/3) = 1610.91

 

XD - ic = 40.14

Thus, the observed difference of the two means,
562.01, is 14 times greater than the standard error and
since the standard deviation of B is 727€IT72 = 52.5 and
one half the distance between the two means is 281.005, we

can conclude that function is significant at the .01 level

/2761.72

126

TABLE 4-7

MEAN VALUES OF PREDICTOR VARIABLES, DISCRIMINATE
WEIGHTS, AND IMPORTANCE VALUES; THIRD WEEK

 

 

 

 

 

Dis-
Continue continue Discriminate Relative
Variables Mean Mean Weight Importance Importance
Awareness 38.50 19.33 1.00 19.17 1%
Knowledge 32.50 16.67 12.85 203.42 11%
Weak Interest 18.50 8.67 -68.43 672.67 35%
Strong Interest 12.00 2.67 24.62 229.70 12%
Weak Infor-
mation Seeking 15.00 3.33 48.62 657.40 30%
Strong Infor-
mation Seeking 2.50 1.33 183.75 214.99 11%
2C = 674.29 2 - 112.28
Discriminate Value = 393.29
TABLE 4-8
THIRD WEEK RESULTS OF ASSIGNMENT TESTS
2 3
Dessert 948.14 Fabric softener 575.42
zFurniture 280.76 ZMeal in can -269.28
zSnack food 992.70 zCooking oil 33.34
zFloor cleanser 475.54
Predicted
Continued Discontinued Total
Continued 3 1 4
Actual Discontinued 1_ 2_ .3
Total 4 3 7

 

127

Table 4-8 shows that the discriminate function for
the third week while being significant at even the .01 level
was not a perfect predictor of product groups. One possible
explanation is offered here as a reason for the third week's
discriminate function failure to provide a perfect predic-
tion of product grouping. This explanation, however, has i
not been substantiated by additional research. It pertains
to the high mean value given to the discontinued fabric 111
softener. This high mean value is directly attributable to
the level 4 for the strong knowledge variable. Since the
discriminate weight for this factor is 183.75, this factor
is one of the most important factors, assuming equal mean
differences for the six variables. However, this abnormally
high mean value for this product could have been caused by
what Rogers terms the non-compatibility of this product's
use with existing norms for the use of other fabric sof—
teners.5 The fabric softener studied in this experiment
required a different type of application than the other
softeners available on the market. It is felt, that this
compatibility factor could have been significant enough for
the consumer to seek out additional information on the

product.

128

Third Week Summary

The third week discriminate function was seen to
have been significant at the .01 level but effective as a
predictive tool on only five of seven products. However,
a possible explanation was offered to account for this in-
ability to predict product grouping. If the level of
strong information seeking activities for the product under
question is reduced to the level of the other products of
the same group, the function will correctly predict all
seven products correctly. (In the future weeks, the
strong information activities of the other two products
of this group, increase above the zero level as can be ex-
pected for all new products thereby reducing the possi-
bility for future errors of this type.) The relative im-
portance of the six variables for the third week has three
distinctive groupings. Weak interest and weak information
seeking were both over the thirty per cent level, strong
interest, strong information seeking and knowledge account
for ten to twelve per cent and awareness accounted for only

one per cent.

Fourth Week Findings

 

The discriminate function for the fourth week was

2 = -.3003x1 - .4245x2 + .3960x3 + .4895x4 + .2478x5 +

 

129
.1993x6 which becomes 5 = -x1 - 1.41x2 + 1.32x3 + 1.63x4 +
.83x5 + .66x6 when x1 is made an unity value.

Five of the six predictor variables are deemed to
be relatively important as shown in Table 4-9. Weak
interest, however, was not the most important as it was for
the second and third week. Knowledge, which was second for
the second week and a member of the second grouping for the
third week, had the highest relative importance percentage.
Strong interest was just slightly less important. It is
not important that the means for both groups are negative;
this is a result of the design of the discriminate function.
This function is designed to give higher values to the
desired group (the continued group, in this case) and
lower values to the undesired group. It is, also, interest-
ing to note that a negative sign doesn't necessarily mean
that the reverse of the two variables would imply a higher
probability of success. A negative sign can, also, be
caused, as noted earlier, by the interaction among and
between the variables. With regard to the case at hand,
it seems highly unlikely that both total lack of awareness
and knowledge will indicate a greater potential for the con—
tinuation of the product in question. Nevertheless, although
it will soon be shown that the discriminate function for

this fourth week is the least significant of the four

 

130

TABLE 4-9

MEAN VALUES OF PREDICTOR VARIABLES, DISCRIMINATE
WEIGHTS, AND IMPORTANCE VALUES; FOURTH WEEK

 

 

 

 

 

 

Dis-
Continue continue Discriminate Relative
Variables Mean Mean Weight Importance Importance
Awareness 41.50 32.00 -1.00 9.50 18%
Knowledge 41.00 31.33 -1.41 13.63 26%
Weak Interest 20.00 14.67 1.32 7.04 13%
Strong Interest 13.00 5.33 1.63 12.50 24%
Weak Infor-
mation Seeking 17.50 8.00 .83 7.89 15%
Strong Infor-
mation Seeking 7.00 4.00 .66 1.98 4%
8C - -32.575 SD = -38.843
Disciminate Value = -35.709
TABLE 4—10
FOURTH WEEK RESULTS OF ASSIGNMENT TESTS
Z ' 3
Dessert -29.96 Fabric softener —46.58
zFurniture conditioner -37.40 ZMeal in can —30.24
zSnack food -31.12 ZCooking oil -39.72
zFloor cleanser -31.82
Predicted
Continued Discontinued Total
Continued 3 1 4
Actual Discontinued 1_ 2_ 3-
Total 4 3 7

 

131
tested, this is not necessarily the result of the negative
awareness and knowledge weights.
The Fisher test produced a standard error of 1.56
which was nearly 25% of the observed difference of the two
means. The function was significant only at the .20 level

which was not considered to be significant for this study.

-32.575 - (#38.843)

Var Z = 5(.3003) = 4.1744

Var (6 - SD) = Var a (1/4 + 1/3) = 4.1744 (7/12) =
C 2 435

OEC - ED = 1.56

1/2 (6.268) = 1.54
/4.17

This function, also, proved useful for predicting
only five of the seven products' correct classifications.
(Table 4-10) An explanation of this random behavior was
that the fourth week was the week prior to July 4th week
end and that this might have caused an undue amount of
interest in a "ready meal" since the housewife might have
wanted to conserve some time over that period. Six of the
ten peOple who displayed a weak interest in the "meal in
can", which is considered to be the probable cause for the
failure of the function as a prediction tool, were contacted
at a later date by the researcher. Two of these peOple

felt that the holidays might have made them more aware of

H-

 

I"

132
products of this nature. The remaining four stated that
the holidays in no way influenced their awareness of the

new product.

Fourth Week Summary

The fourth week actually produced very insignifi—
cant results and the author, after accounting for the activi-
ties of that week and possible explanations for its lack of

significance, feels justified in not deriving any results

 

from this week's activity. At most, the only valid conclu-
sion that can be drawn from an analysis of this week's
activity is that no positive tool can be derived to predict

product continuation or discontinuation.

Fifth Week Findings

 

The discriminate function for the fifth week was
2 = -5.l765X1 - 21.088x2 + 30.1535x3 + 4.0180x4 + 56.2518x5
2.00x6 which becomes 8 = 'Xl - 4.07x2 + 5.83x3 + .78x4 +
10.87x5 + '39X6 when x1 was reduced to an unity value. As
shown in Table 4—11 weak information seeking activities
followed by weak interest make the two greatest contribu-
tions to discriminating between the two groups. It should
likewise be noted that awareness and knowledge once again
had negative discriminate weights. However, as in the

previous week this was not felt to be the result of non—

133

TABLE 4-11

MEAN VALUES OF PREDICTOR VARIABLES, DISCRIMINATE
WEIGHTS, AND IMPORTANCE VALUES; FIFTH WEEK

 

 

 

Dis-

Continue continue Discriminate Relative
Variables Mean Mean Weight Importance Importance
Awareness 45.00 34.67 —1.00 10.33 5%
Knowledge 44.00 34.67 -4.07 37.97 19%
Weak Interest 22.00 14.67 5.83 42.73 22%
Strong Interest 11.00 7.33 .78 2.86 1%
Weak Infor-
mation Seeking 18.00 8.67 10.87 101.42 52%
Strong Infor—
mation Seeking 5.50 1.33 .39 1.63 1%

3C = 110.57 3D = 10.23

Disciminate Value I 60.50

 

TABLE 4-12

FIFTH WEEK RESULTS OF ASSIGNMENT TESTS

 

 

ZDessert 105.44 zFabric softener 3.86
aFurniture conditioner 114.74 zMeal in can 15.70
zSnack food 108.82 8Cooking 011 11.02
aFloor cleanser 113.26
Predicted
Continued Discontinued Total
Continued 4 0 4
Actual Discontinued Q_ 3. .3
Total 4 3 7

 

134
awareness and non-knowledge being positive predictors of
product continuation.

The two other evaluations of the five week's dis-
criminate function, also, showed it to be significant. The
Fisher test found the difference in means to be almost
seventy times as great as the standard error and that the
function was significant at the .01 level. The function's
assignment value was similarly, seven for seven. (Table
4-12)

110.57 - 10.23 100.34

var 5 = 5 (5.1765) = 25.88 = 3'88

 

Var (EC - ED) = Var a (1/4 + 1/3) 3.88 (7/12) =

2.26

L/211100-34) _ 50.17 _
/3T88 1.97

 

Fifth Week Summary

 

The fifth week's results were unsurpassed in any
category by any of the other week's function. It is felt
that this is only natural as the predictive values should
increase as time passes. The fact that the four continued
products point value ranged from 105 to 114 and the point
value of the discontinued products from 3 - 15, tends to
support the predictive efficiency of this function.

That this week's results were so significant seems

.flu‘h

1‘1:

135
even more significant, in view of the fact that the same
relative importance has been attached to the six variables
as in the first three weeks studied. Weak interest, weak
information seeking make the two greatest contributions in
the fifth week, in an average of the second, third and fifth
week and in the average of all four weeks. The reason the
fourth week was drOpped in the one group was to determine
if the results of that week would distort the combined

findings. They did not.

Summary of Weekly Percentage Tests

 

The discriminate function for three of the four
weeks was found to be significant. In two of the four
weeks the function correctly predicted the product grouping
for all seven products and in the other two weeks it cor-
rectly predicted five out of seven. Two possible explan-
ations were offered for this non-perfect prediction pattern.

An examination of the relative importances (Table
4-13) of the six variables for the four week period, shows
that three variables, weak interest,\weak information and
knowledge of product type, are the most important contrib—
utors to the discriminate function. The others were not

felt to be contributors.

 

136

TABLE 4-13

SUMMARY OF THE RELATIVE IMPORTANCE PERCENTAGES

(Parentheses indicate weekly rank)

 

 

 

2nd, 3rd

2nd 3rd 4th 5th 4 Week 5th Week
Variable Week Week Week Week Average Average
Awareness (4) 6 (6) l (3)18 (4) 5 (5) 7.5 (6) 4.0
Knowledge (2)31 (4.5)11 (1)26 (3)19 (3)21.7 (3)20.3
Weak Interest (1)48 (1)35 (5)13 (2)22 (1)29.5 (1)35.0
Strong Interest (5) 5 (3)12 (2)24 (5.5) 1 (4)10.5 (5) 6.0
Weak Infor-
mation Seeking (6) 2 (2)30 (4)15 (1)52 (2)24.8 (2)28.0
Strong Infor-
mation Seeking (3) 8 (4.5)11 (6) 4 (5.5) 1 (6) 6.0 (4) 6.7

 

Rate of Growth of the Six Variables
as Prediction Tools

The rates of growth for the three time periods of
this experiment are shown in Tables 4-14 to 4-16. As indi-
cated in Table 4-17, the t—test analysis found that only two
variables, awareness for the third-second week and knowl—
edge for the fourth—third week, were significant at the .02
level. Two other growth rates, awareness and strong
interest both for the fourth-third week, were significant
at the .10 level. The remaining fourteen variables were
not deemed significant, since the .10 level is the normal

maximum test for significance.

137

 

 

 

 

 

 

oo. oo. co. co. om. “44.18 ago mangoes
co. nc.N om. om.~ cw. Aoa.lv gwo a“ and:
oo.H oo. Ano.|v mm. om.H mo. umamumom ofiunwm
muusvoum pmsafiusoomaa
oo. mo.~ mm. mm. HH.H so. umamoau uooam
oo. mm.~ oo.H oo. oo.H ms. boom xomam
oo. oo. o4. Aoa.nv an. HN. umaowufieaou muauwaunm
on. Amm.lv «H. Ao~.1v co. mm. uuommwn
muosvoum poaafiucoo
wcouum xmo: wcouum Mums owvoaaoax mmoaouoz¢
lwcaxoom coaumsuowcH ummumuaH
TWPIIV
N m #003 vacuum I phase

3I3

mmqm<Hm¢> mOHUHnmmm.NHm Huh mom mfizcmu mo NH<m

ealc MAM<H

138

 

 

 

 

 

 

 

8. 8.4 8.4 8.4 84 on. do 96.368
oo.¢ no. mm. me. mw. aw. amo a“ How:
on. oo.~ oo.N ow. om. mm. umcmumom cannon
muoapoum posaaunoomfin
oo.a Aem.lv om. Aom.lv Amo.lv Aoa.lv “mamoao uooam
oo.~ Aoo.lv on. oo. oo. ow. coon Joann
oo.~ no.~ AcH.Iv mm. mm. 00. umaowufivsoo ouauficush
oo.H an. 00. NH. mm. mo. uuommua
muospoum ponawuaou
woouum Jams waouum 3mm: mwpoaaoau mmmaouma<
mcaxmow aowumauomaH umououaH
A m3 8
m3 1 «3 3003 phase I :uusom

mmgm<Hm<> KOHUHnmmm NHm Hma mom mhzcmo mo wa<m

mHIq mqm<H

139

=4.-

 

 

 

 

 

 

 

oo. oo.m oo.N oq. mm. mm. Hfio waaxooo
Amm.uv AoN.IV mm. Aom.uv Aoo.-v Aoo.nv :66 as Ham:
Aoo.HIV ANH.IV oo. «H. Ha. no. nonmuwom cannon
muozvoum vmsnwuaooman
oo.H mm. Aom.Iv mm. Ha. «N. umammao uooam
mm. oo.N 00. on. ma. ma. boom xomsm
Amn.nv Aao.nv oo. oo. mo. ac. 86864046666 «unawausm
Amm.lv AHm.Iv AmH.IV 50. oo. Aqo.Iv uummmon
muospoum poaawusoo
wsouum xmoz wcouum x603 omvmasocu mmmcouoa<
wcfixmom soaumauomaH ummumucH
14:8
«3 I m3 3003 nuHDOh I Sousa

wmqdem¢> MOHUHQmmm NHm NEH mom wh3omu ho mhdm

0HI¢ mqm<H

140

 

 

 

 

Nmo.¢ 6 Ho. mom.m a Na. Hmn.~ u no. mHo.N m 0H. onc.a a om.
“mum u mo mosaw> Hmowufiuu
Anamo.HIv Hmmm. oaqm.H Ammmm.Iv Howe. oqmmo. nuHSOMIsuMfim
Akamm.-v seam. mNmH.~ mamn.d mmwn.m Nmm4.~ consensuuaom
coco. Amqm~.Iv Aammc.HIV omqa. mmom. Aomum.MIv pnoommIvuana
wcouum 3663 wsouum 3mm: owpoH3OGM mmmdmumz< poauoa mafia
wdfixmmm coaumauomaH umououaH

 

mmqm<Hm<> MOHOHQmmm NHm mmB mom mB3QmU mo MH<M Nah mom mmmoom Hmmhlu

NHIQ mam<H

141

These results are in agreement with the analysis
of variance tests, as only during the fourth—third week was
there found to be a significant (.10) difference due to the
continuation or discontinuation of the seven products. No
significant differences were found due to the predictor
variables. See Tables 4—18 to 4-20.

Thus, evidence has been presented which tends to
support the null hypothesis that the rates of growth of the
six variables studied cannot be used to predict the continu-
ation or the discontinuation of the product. It should be
noted, however, that this finding is based on what is
admittedly a small sample population. This small sample
size was beyond the control of the researcher as a major
contention of the thesis was that only products of a similar
nature should be studied in comparison with each other.
Nevertheless, the fact that awareness was significant at
.01 in the first time period and at the .10 level for the
second period should not be overlooked. Another possible
explanation for these results could be found by considering
the bases upon which the rate of growth ratios were built.
For a continued product, if the innovation eXperienced a
high initial acceptance, it is more difficult to experience
a high rate of growth due to the initial higher base.

While the discontinued product might never experience a

J

142

TABLE 4-18

ANOVA TABLE FOR RATES OF GROWTH
OF THE THIRD-SECOND WEEK

 

 

 

 

Source d.f 88 MS F C.V.F@ 10% C.V.F@ 2%
Products 1 0 0 0 4.06 8.25
Predictive
Variables 5 1.27 .254 1.6 3.45 12.13
Error 5 .77 .154
11 2.04
TABLE 4-19

ANOVA TABLE FOR RATES OF GROWTH
OF THE FOURTH-THIRD WEEK

 

 

 

Source d.f SS MS F C.V.F@ 10% C.V.F@ 2%
Products 1 2.55 2.55 5. 4.06 8.25
Predictive
Variables 5 2.93 .58 1. 3.45 12.13
Error 5 2.53 .51
11 8.01
TABLE 4-20

ANOVA TABLE FOR RATES OF GROWTH
OF THE FIFTH-FOURTH WEEK

 

 

Source d.f 88 MS F C.V.F@ 10% C.V.F@ 2%
Products 1 .044 .044 .34 4.06 8.25
Predictive

Variables 5 .757 .151 1.16 3.45 12.13

Erggr 5 .651 .130

 

 

143
high rate of growth due to the initial higher base. While
the discontinued product might never experience a high
level of acceptance and thus any subsequent chance may
cause a high degree of change in the growth rate. Thus,
the comparison of the mean rates of growth could be an

invalid test.

144

FOOTNOTES

For a more complete discussion of how to work linear
discriminate prOblems, the reader should consult
Maurice G. Kendall, A Course in Multivariate Analysis.
(London: Charles Griffin and Company, Ltd., 1965),
pp. 144-170.

 

Frederick Mosteller and David S. Wallace, "Inference
in an Authorship Prdblem," Journal of the American
Statistical Association, MVIII (June, 1963), 275—309.

 

 

Paul E. Green and Donald S. Tull, Research for
Marketing Decisions. (Englewood Cliffs, N.J.:

 

 

Prentice-Hall, 1970), p. 376.
Kendall, pp. 149-150.

Everett M. Rogers with F. Floyd Shoemaker, Communication
of Innovations. (New York: The Free Press, 1971),

 

 

 

CHAPTER V

SUMMARY AND CONCLUSIONS

The objective of this chapter is to review the
research project from its inception to its conclusion. The
first section of this chapter will examine the objectives
of the study. The second portion focuses on the empirical
findings of the investigation. The final section sets
forth the major implications of the study, its limitations

and notes several areas for future research.

Objectives of the Study

Over $300 billion will be spent on new product in-
novation in the 1973 fiscal year. Yet, over 70 per cent of
this cost will go to products that will not be successful
in the market place. Nowhere is this problem of new prod—
uct introduction more prevalent than in the retail food
industry. As long ago as 1966, the ex-Secretary of Agri-
culture, Orville Freeman, stated, "Each year about 5,000
new products are offered to stores that already carry 8,000

1

different items." Business Week has estimated that of the

 

145

 

146
120,000 supermarket products introduced during the 1970's

roughly 100,000 will bomb out2 or a study by Advertising

 

Agg which predicted that 80 per cent of new supermarket
products will fail as they will not meet sales goals.3
Nevertheless, even in view of the above information, the
retail food manager of today still must rely on an analysis
of the new product's initial thirteen week's sales data be-
fore making a decision on whether to continue to stock the
product in question or discontinue it.

Thus, the stated purpose of this research project
was to be able to provide the retail food manager with a
means of predicting which of the products distributed
through his retail food outlet should be eliminated from
his product line at a point in time far earlier than the
usual analysis of thirteen weeks sales data. Instead of
reviewing the initial three months sales results, it was
hypothesized that knowledge of the lgygl of consumer adOp-
tion process variables, as well as knowledge of the ra£§.gf
growth of these variables, could be used to predict manage—
ment's decision to continue or discontinue a new product at
a point in time earlier than in present use today. The
point in time chosen for this study was five weeks after
product introduction. While admittedly this was an arbi-

trary decision, it was chosen so as to make a significant

147
reduction in the amount of time needed to make the continu-
ation decision.

The consumer adOption variables chosen for study
were awareness of the new product, knowledge of its product
type, weak and strong interest in the product and weak and
strong information seeking activities toward the product.
These variables were determined to occur before the con-
sumer makes a decision to adOpt or reject the new product.

This research study measured the levels of
activity for these six predictor variables against seven
new products introduced in the Des Moines, Iowa market
during the summer of 1972. The seven products studied were
determined to be of similar nature in both terms of con-
sumer product classification and in level of newness. In
so far as these products were of a similar nature, it was
hypothesized that they should possess similar characteris-
tics with regards to the level of activity occurring in
the six predictor variables. A telephone survey was con—
ducted of 1200 customers of a selected retail food store
in Des Moines to determine the level of activity in the
adoption process variables for each of the seven products.
The two specific null hypotheses studies were

1. The knowledge of the level of consumer adOp—

tion process variable within an earlier period

148
of time, makes no difference in management's
ability to identify products, which, according
to its criteria, should be continued relative
to those which should be discontinued.

2. The knowledge of the £333 9; growth in consumer
adOption process variables within an earlier
period of time, makes no difference in manage-
ment's ability to identify products which,
according to its criteria, should be continued

relative to those which should be discontinued.

Empirical Findings

 

The approach to analyzing the data was twofold.
First, linear discriminate analysis was used to test if the
weekly percentages for the second through fifth week after
introduction of the six predictor variables chosen were
able to discriminate between the continued and discontinued
products. Second, the rate of growth hypothesis was tested
by means of eighteen independent t—tests. Also, two-way
analyses of variance were performed on the mean rates of
growth between the continued and discontinued groups by the
six predictor variables for the same time period as the

earlier tests.

 

Linear discriminate analysis was able to discriminate

149
between the continued and discontinued products by using
the weekly percentage levels of the six variables. During
the fifth week, the discriminate function was found to be
significant at the .01 level and was able to correctly
assign all seven new products to their prOper group.

Data from the second, third and fourth weeks was
also analyzed. This was an attempt to see if an earlier
time period could produce significant results. While the
second week's function was found to be significant at the
.02 level and correctly predict the outcome of all seven
products, the observed difference in the means of the two
product groups was only nine times the week's standard
error. The fact that the fifth week's mean difference was
seventy times greater than its standard error supported the
notion that the fifth week is able to provide more conclu—
sive evidence as to the continuation of the product.

The functions for the other two weeks were believed
to have been influenced by extraneous variables. Neverthe-
less, they were still able to discriminate between the two
groups and predict five of the seven product grouping
correctly.

Three of the six adOption process variables chosen
were found to exercise a great deal of influence on predict—

ing the continuation of the new product. These three

150
variables were weak interest in the product, weak infor-
mation seeking activities toward the product and knowledge
of the product type.

Evidence was presented which tended to support the
null hypothesis that the knowledge of the rate of growth
of the six variables studied can not be used to predict
the continuation or the discontinuation of the product.

It should be noted, however, that this finding is based on
what is admittedly a small sample pOpulation. This small
sample size was beyond the control of the researcher as a
major contention of the thesis was that only products of a
similar nature should be studied in comparison with each

other.

Implications of the Research

 

It was shown that for at least one classification
or products at one point in time and in one geographic
location, a linear discriminate function of six adOption
process variables could be used to predict product continu-
ation or discontinuation before meaningful sales data was
available. Prior to this time the adoption process has
been used as a post-operative tool to explain what had
happened concerning an innovation. Now the early phases

of this process have been shown to be useful as a predictive

151

tool as well.

It might be well to consider if sales data at the
end of the fifth week could have predicted the same results.
In this particular case, sales results would not have pre-
dicted the ultimate product groupings. First, for all
products in general, retail sales data is usually very small
for the initial month after introduction. Granted the whole—
saler will experience high initial sales as retailers first
begin to stock the item. However, any further sales will
only result as retailers seek to restock their inventory.
This is accomplished only after the consumer passes through
the early stages of the adOption process and as past studies
have indicated this is not a rapid occurrence.

In this particular case, two products experienced
sales for the fifth week which were not indicative of their
ultimate classification. One continued product sold less
than two cases during the fifth week. This product did not
begin to experience sufficient sales results until the ninth
week. However, the model presented in this research was
able to correctly predict its continuance on the basis of
the high level of activity in the interest and information
seeking activities of the consumers toward the product.
Another product experienced "good" sales results during the

fifth week but its sales began to decline in the later

152
weeks. The model at five weeks correctly predict its dis-
continuance at the thirteenth week. It is interesting to
note that not only did the model correctly predict two
product's ultimate result different than their sales data
would indicate, but it predicted them in both of the product
groups.

Thus, this thesis has presented a model which in
this instance was able to correctly predict at five weeks
after introduction of a new product the ultimate outcome of
managements decision to continue or reject it after a review
of thirteen weeks sales data.

This thesis is, also, of use to the manufacturer
and middlemen, since the ultimate success of their new
products depends upon the retailer's decision to stock the
product in question. Therefore, it is possible that this
model can serve as valuable aid to these other members of
the channel, since they too would like to be able to dis-
continue an unsuccessful offering at an earlier point in
time.

The financial burden of new product introduction
falls on the manufacturer and it is within his interest to
likewise determine as early as possible how retailer's will
evaluate his offerings. The fact, test marketing is designed

to assess consumers reaction to a new product and the

153
retailer's willingness to continue to stock is inferred from
such tests.

It is conceivable, then, that a manufacturer could
use the measurements of the six predictor variables used in
this thesis as an alternative to his present use of sales
data in analyzing test market results. By establishing a
series of retail locations, the manufacturer will be able
to examine consumer reaction to the various marketing mixes
with which the new product may be introduced. The model
presented here will offer the advantages of early knowledge
of product continuance on discontinuance based on evidence
which has been found to be more conclusive than initial
sales data.

However, before drawing any conclusions from the
above findings to other product classifications or geo-
graphic locations, the reader should consider two limitations
of the study. First, this study was able to study a limited
number of new products of the same classification. This
resulted in a low number of degrees of freedom being
available for determining the significance of the statisti-
cal tests. Secondly, the study assumed that there was a
common homogeneity between the households selected for
interviewing for each of the four weeks. While there is no

present evidence to refute this assumption, the reader

154

should consider this in drawing his conclusion.

Thus, this research study may have posed more ques—

tions than it has answered - if this is correct the re-

searcher will consider the undertaking a worthwhile and

rewarding experience. Nevertheless, some of the questions

that must now be looked at include:

1.

If the resources were available to the re—
searcher, would the final different results

if more products were used in the sample size?
The increasing of the sample size would enable
the researcher to have an increased number of
degrees of freedom in making the statistical
tests.

Could a different set of variables produce a
significant improvement to the model? This
study contended that the activities of the
first two stages of the adOption process in-
cluded awareness, knowledge, interest and infor-
mation seeking. What would be the results if
it were possible to measure the activity that
the early rural sociologists referred to as
evaluation or the activity of liking or prefer—
ence as proposed by Lavidge and Steiner. If

such attitudes could be quantified, they might

155

produce different results.

3. This model studied only one particular class
of products. Before universal conclusions can
be develOped it should consider some other
product classes. An especially interesting
question would concern its adaptability to
durable goods.

4. While this model considered only one retail
location, it does appear that the model can
be used at other locations, as well as a
national predictive tool, rather than as a
tool for an individual store manager.

This present study may be of great value if the

empirical findings can be used successfully in other mar-

kets.

156

FOOTNOTES

Orville Freeman, This Week, June 26, 1966, 2.

Business Week, March 4, 1972, p. 73.

Theodore Angelus, "Why Do Most New Products Fail,"
Advertising Age, XL (March 24, 1969), 85.

APPENDIX A

Census Tracts of

Polk County, Iowa

Des Moines, Iowa

Selected Stores Trading Area

157

APPENDIX A

CENSUS TRACTS 0F POLK COUNTRY, IOWA; DES MDINES, IOWA
Selected Stores Trading Area

 

Selected Store

 

Polk Co. Des Moines Trading Area
RACE
All persons 286,101 200,587 19,170
White 272,983 188,179 19,130
Negro 11,916 11,425 40
Percent Negro 4.2 5.7 .2
AGE BY SEX
Male, all ages 136,234 93,958 9,558
Under 5 years 12,756 8,539 998
3 and 4 years 4,965 3,300 382
5 to 9 years 14,413 9,291 1,232
5 years 2,760 1,798 236
6 years 2,874 1,833 258
10 to 14 years 14,665 9,464 1,257
14 years 2,816 1,838 220
15 to 19 years 12,914 9,015 881
16 years 2,746 1,798 211
17 years 2,567 1,724 175
18 years 2,677 1,988 165
19 years 2,266 1,746 103
20 to 24 years 10,478 7,968 513
20 years 2,140 1,647 84
21 years 1,905 1,486 86
25 to 34 years 17,644 11,745 1,444
35 to 44 years 15,585 10,117 1,217
45 to 54 years 15,341 10,680 912
55 to 59 years 6,296 4,569 360
60 to 64 years 5,366 4,024 269
65 to 74 years 6,891 5,421 329
75 years and over 3,895 3,126 146
Female, all ages 149,867 108,629 96,174
Under 5 years 12,193 8,246 945
3 and 4 years 4,808 3,180 388
5 to 9 years 13,569 8,778 1,188
5 years 2,472 1,641 222
6 years 2,552 1,664 207
10 to 14 years 14,083 9,186 1,136
14 years 2,725 1,804 207
15 to 19 years 13,775 9,902 839
15 years 2,686 1,773 222
16 years 2,583 1,696 182
17 years 2,569 1,727 167
18 years 2,968 2,279 159
19 years 2,969 2,427 109

158

APPENDIX Ar-Continued

 

 

Selected Store

 

Polk Co. Des Moines Trading Area
20 to 24 years 13,231 10,105 689
20 years 2,810 2,250 118
21 years 2,790 2,171 141
25 to 34 years 18,866 12,445 1,600
35 to 44 years 16,515 10,913 1,184
45 to 54 years 16,684 12,040 913
55 to 59 years 7,246 5,572 335
60 to 64 years 6,419 5,042 256
65 to 74 years 9,855 8,171 369
75 years and over 7,431 6,229 183
RELATIONSHIP TO HEAD
OF HOUSEHOLD
All persons 286,101 200,587 19,115
In households 278,187 194,123 19,133
Head of household 93,415 68,586 5,342
Head of family 72,739 50,658 4,872
Primary individual 20,676 17,848 471
Wife of head 64,118 43,628 4,500
Other relative of head 115,666 77,443 9,051
Not related to head 5,488 4,546 169
In group quarters 7,414 6,464 42
Persons per household 2.98 2.83 3.57
TYPE OF FAMILY AND
NUMBER OF OWN CHILDREN
A11 families 72,739 50,658 4,872
With own children
under 18 years 40,874 26,938 3,179
Number of children 92,639 60,307 7,616
Husband-wife families 64,118 43,628 4,490
With own children
under 18 years 36,102 23,128 2,928
Number of children 82,168 51,841 7,013
Percent of total
under 18 years 84.3 81.0 88.2
families with other
mail head 1,405 1,072 76
With own children
under 18 years 470 335 36
Number of children 932 690 81
families with
female head 7,216 5,958 306

With own children
under 18 years 4,302 3,475 215

159

APPENDIX A—-Continued

 

Selected Store

 

Polk Co. Des Moines Trading Area
Number of children 9,539 7,776 522
Percent of total
under 18 years 9.8 12.2 6.6
Persons under 18 years 97,488 63,981 79,403
MARITAL STATUS
Male, 14 years old
and over 97,216 68,502 6,263
Single 24,439 17,796 1,384
Married 66,859 45,844 4,603
Separated 932 776 34
Widowed 2,565 2,037 108
Divorced 3,363 2,825 169
Female, 14 years old
and over 112,747 82,223 6,555
Single 26,231 19,272 1,194
Married 67,707 46,439 4,659
Separated 1,371 1,174 54
Widowed 13,492 11,187 438
Divorced 6,815 5,275 264
All housing units 98,325 72,349 5,552
Vacant - seasonal
migratory 28 12 6
A11 year-round
housing units 98,297 72,337 5,546
TENURE , RACE , AND
VACANCY STATUS
Owner occupied 65,000 45,408 4,705
COOperative and
condominium 90 85 9
White 62,823 43,380 4,688
Negro 2,017 1,906 7
Renter occupied 28,412 23,098 638
White 26,657 21,388 634
Negro 1,579 1,542 1
Vacant year-round 4,882 3,831 203
For sale only 750 514 58
Vacant less than 6 mos. 613 408 47
Median price asked $16,100 $12,100 4,500
For rent 2,616 2,171 45
Vacant less than 2 mos. 1,803 1,459 27
Median rent asked 594 589 85
Other 1,516 1,146 100

160

APPENDIX A--Continued

 

Selected Store

Race Polk Co. Des Mbines Trading Area

 

LACKING SOME OF ALL
PLUMBING FACILITIES

All units 4,418 3,319 385
Owner occupied 1,394 798 230
Negro 106 76 -
Renter occupied 2,336 2,017 81
Negro 219 203 -
Vacant year-round 688 504 74
For sale only 23 10 7
For rent 427 373 17
COMPLETE KITCHEN FACILITIES
AND ACCESS
Lacking complete kitchen
facilities 2,091 1,538 149
Access only through other
living quarters 95 85 2
ROOMS
1 room 2,294 2,195 17
2 rooms 4,287 3,868 61
3 rooms 9,296 8,001 308
4 rooms 20,778 15,851 1,357
5 rooms 28,670 20,965 2,199
6 rooms 17,000 11,624 1,169
7 rooms 8,684 5,517 1,459
8 rooms 4,797 2,716 145
9 rooms or more 2,491 1,600 70
Median 4.9 4.8 5.0
All occupied housing
units 93,415 68,506 5,243
PERSONS
1 person 17,997 15,554 433
2 persons 28,192 21,227 1,355
3 persons 15,679 11,184 959
4 persons 14,281 9,395 1,103
5 persons 9,172 5,827 760
6 persons or more 8,094 5,319 733
Median, all occupied units 2.5 2.4 3.2
Median, owner occupied units 2.9 2.7 3.3
Median, renter occupied units 2.0 1.9 3.1
Units with roomers,
boarders, or lodgers 2,185 1,875 62

161

APPENDIX A—-Continued

 

Selected Store

Race Polk Co. Trading Area

Des Moines

 

PERSONS PER ROOM

1.00 or less 87,815 64,514 4,694
1.01 to 1.50 4,676 3,313 549
1.51 or more 924 579 100
Units with all plumbing
facilities - 1.01 or more 5,304 3,820 586

VALUE

Specified owner

occupied units 58,851 42,052 3,939
Less than $5,000 1,320 933 173
$5,000 to $7,499 3,498 2,776 343
$7,500 to $9,999 6,374 5,390 455
$10,000 to $14,999 15,098 12,678 907
$15,000 to $19,999 14,120 10,729 1,228
$20,000 to $24,999 8,851 5,002 549
$25,000 to $34,999 6,465 3,014 240
$35,000 to $49,999 2,332 986 38
$50,000 or more 793 544 11
Median $16,100 $14,700 15,275
CONTRACT RENT

Specified renter

occupied units 27,690 23,017 575
Less than $30 389 292 17
$30 to $39 496 422 22
$40 to $59 2,598 2,313 70
$60 to $79 5,256 4,698 118
$80 to $99 4,973 4,500 92
$100 to $149 8,400 6,821 150
$150 to $199 3,566 2,423 30
$200 to $249 511 397 3
$250 or more 398 370 1
No cash rent 1,073 781 72
Median $98 $94 91

 

 

APPENDIX B

Telephone Questionnaire

Telephone Interviewing Procedures

Telephone Data Recording Procedures

162

Telephone Questionnaire

1. Have you heard of a new product called

 

which was recently introduced in the Des Moines area?

Yes or No

2. Do you know what type of product is?

 

(If the shopper asks for help, tell her Wonder would be

a bread and Folgers would be a coffee.)

 

If you get a no to both question 1 and 2, you should

end the interview.

3. a.

Do you have any desire to try this new product?

Yes or No If no, go to #4

If yes, have you already tried the product?

Yes or No

Do you plan to purchase this product within the

next seven days? Yes or No

Have you talked to any friends, some store personnel
or read anything about this new product? Yes or No

If yes, what was the source or to whom did you talk

to?

 

Did you seek out this information? Yes or No

Thank you for your co-Operation

 

163

Telephone Interviewing Procedures

Interviewer will identify herself as being from Drake
University Business Research Center.

Interviewer will ask to speak to the household's food
shOpper.

All duplication of households will be removed prior to
making the sample list.

A maximum of three calls will be placed to a given
household.

Telephone calls will be made only during the following

hours:
Morning 9:15 a.m. - 11:30 a.m.
Afternoon 1:30 p.m. - 3:45 p.m.
Evening 7:00 p.m. — 8:00 p.m.

No calls will be made over the weekend or July 3rd

or 4th.

All disconnected numbers will be referred to the phone
company for a new listing. If the new listing remains
within the selected market area, attempts will be made
to reach the new number. If the party has moved out
of the area, a new number will be chosen as per prior
instructions.

Consecutive calls to the same household will always

 

164
be made at different times.
If the interviewer is unable to reach a household
within three attempts, that household will be perma-
nently drOpped from the sample.
Each household initially selected for the survey will
be given a respondent number for purposes of identifi-

cation. ReSpondent numbers will range from 1 to 1200.

 

be

165

Telephone Data Recording Procedures

The data for each of the six predictor variables will

analyzed as follows:

The data will be kept in terms of percentage of house-
holds who answer the predictor question in the
affirmative.

A yes answer to question 1 will indicate awareness of
the product.

A knowledge of general product family type (e.g., Epic
being listed as a new type of coffee, a new brand of
instant coffee or simply as being a coffee will be
sufficient. To be listed as a new food product or a
new drink will not be sufficient.) must be indicated

to be considered to have a knowledge of the product
type.

An affirmative answer to 3a will indicate a weak intent
to purchase.

An affirmative answer to 3b or Be will indicate a strong
intent to purchase.

An affirmative answer to 4a will indicate weak informa—
tion seeking activities toward the product.

An affirmative answer to 4c will indicate strong infor-

mation seeking activities toward the product.

 

166
8. A total of 50 households will be contacted for each
product for each week of the test. 7
9. The outcome of elimination or continuation of the

product will be management's decision after 13 weeks.

 

BIBLIOGRAPHY

Books

Andreasen, Alan R. "Attitudes and Customer Behavior: A
Decision Model," reprinted in Harold H. Kassayian
and Thomas S. Robertson, Perspectives in Consumer
Behavior. Glenview, Illinois: Scott, Foresman
and Company, 1968.

 

Bliss, Perry. Marketing and the Behavioral Sciences.
Boston: Allyn and Bacon, 1967.

 

Buzzell, Robert D., and Nourse, Robert E. M. Product Inno—
vation in Food Processing: 1954-1964. Boston:
Howard University, Division of Research, 1967.

 

 

 

, , Matthews, Jr., John B. and Levitt,
Theodore. Marketing: A Contemporary Analysis.
New YOrk: McGraw-Hill, 1972.

 

Engel, James F., Kollat, David T. and Blackwell, Roger D.
Consumer Behavior. New Yerk: Holt, Rinehart and
Winston, 1968.

 

Green, Paul E. and Tull, Donald S. Research for Marketing
Decisions. Englewood Cliffs, New Jersey: Prentice—
Hall, 1970.

 

 

Howard, John A. Marketing Theory. Boston: Allyn and
Bacon, 1965.

 

Kassarjan, Harold H. and RObertson, Thomas S. Perspectives
in Consumer Behavior. Atlanta: Scott, Foresman
and Company, 1968.

 

Katona, George. The Powerful Consumer. New York: McGraw-
Hill, 1960.

 

Kendall, Maurice G. A Course in Multivariate Analysis.
London: Charles Griffin and Company, Ltd., 1965.

 

and Stuart, Alan. The Advance Theory of Statistics.
Volume III. New YOrk: Hafner Publishing Company,
1966.

167

168

Kotler, Philip. Marketing Decision Making. New YOrk:
Holt, Rinehart and Winston, Inc., 1971.

 

Learner, D. B. "Profit Maximization through New Product
Marketing, Planning and Control," in Frank M. Bass,
Charles W. King and Edgar A. Pessemier, editors,
Applications of the Sciences in Marketing Management.
New Yerk: John Wiley and Sons, Inc., 1968.

Lucas, Darrell and Britt, Stewart. Measuring Advertising
Effectiveness. New YOrk: McGraw-Hill, 1963.

 

Luck, David J., Wales, Hugh B. and Taylor, Donald A.
Marketing Research. Englewood Cliffs, New Jersey:
Prentice-Hall, 1970.

 

Marvin, Philip. "Why New Products Fail," in Thomas Berg
and Abe Shuchnan, Product Strategy and Management.
New YOrk: Holt, Rinehart and Winston, 1963.

 

Mason, Robert. "An Ordinal Scale for Measuring the AdOp—
tion Process," in Wilbur Schramm, editor, Studies
of Innovation and of Communications to the Public.
Stanford, California: Stanford University Institute
for Communication Research, 1962.

McCarthy, E. Jerome. Basic Marketing: A Managerial
Approach, 4th ed. Homewood, Illinois, R. D. Irwin
and Co., 1971.

 

Nicosia, Francesco M. Consumer Decision Processes. Engle-
wood Cliffs, N.J.: Prentice-Hall, 1966.

Robertson, Thomas S. Consumer Behavior. Atlanta: Scott,
Foresman and Company, 1970.

. Innovative Behavior and Communication. New York:
Holt, Rinehart and Winston, Inc., 1971.

 

Rogers, Everett M. Diffusion of Innovation. New York: The
Free Press, 1962.

. Bibliogrgphy on the Diffusion of Innovations.
East Lansing, Michigan: Michigan State University,
College of Communication Arts, 1968.

 

with Shoemaker, F. Floyd. Communication of

 

169
Innovations. New York: The Free Press, 1971.
Romans, Daniel D. Research and Development Management:
The Economics and Administration of Technology.

New York: Appleton Century-Crafts, 1968.

Smith, Dilman M. K. How to Avoid Mistakes When Introducing
New Products. New York: Vantage Press, 1964.

Stanton, William J. Fundamentals of Marketing, 3rd ed.
New Yerk: McGraw-Hill, 1971.

 

Staudt, Thomas A., and Taylor, Donald A. A Managerial
Introduction to Marketing. Englewood Cliffs, N.J.:
Prentice-Hall, 1970.

 

 

Stockton, John R. Introduction to Business and Economic
Statistics. Cincinnati: Southwestern Publishing
Co., 1966.

 

Zaltman, Gerald. Marketing: Contributions from the
Behavioral Sciences. New York: Harcourt, Brace
and World, Inc., 1965.

 

Periodicals

 

Ahl, David H. "New Product Forecasting Using Consumer
Panels," Journal of Marketing Research, VII (May,
1970), 160-167).

Angelos, Theodore. Why Do Most New Products Fail,"
Advertising Age XL (March 24, 1969), 85-90.

 

Arndt, Jehan. "Role of Product-Related Conversations in
the Diffusion of a New Product," Journal of
MarketingyResearch, IV (August, 1967), 291-295.

 

 

Bader, Samuel. "The Sign: Its Power Ever Rises,"
Printers' Ink, May 29, 1964, 134-138.

 

Berenson, Conrad. "Pruning the Product Line," Business
Horizons (Summer, 1963), 63-70.

 

 

Campbell. "A Suggested Paradigm of the Individual AdOption
Process," Rural Sociology, XXXI (December, 1966),
458-466.

 

 

170

Claycamp, Henry J. and Liddy, Lucien E. "Prediction of New
Product Performance: An Analytical Approach,"
Journal of Marketing Research, VI (November, 1969),
414-420.

 

Cohen, Jack B. and Goldberg, Marvin E. "Dissonance Model
in Post-Decision Product Evaluation," Journal of
Marketing Research, VII (August, 1970), 315-321.

 

 

Corey, Lawrence G. "PeOple Who Claim to be Opinion Leaders:
Identifying Their Characteristics by Self-Report,"
Journal of Marketing, XXXV (October, 1971), 48-53.

 

Cox, Jr., William E. and Cooke, Ernest F. "Other Dimensions
Involved in Stepping Center Preferences," Journal of
Marketing, XXXIV (OctOber, 1970), 12-17.

 

 

Crawford, C. Merle. "The Trajectory Theory of Goal Setting
for New Products," Journal of Marketing_Research,
III (May, 1966), 117-25.

Drucker, Peter F. "Care and Feeding of the Profitable
Product," Fortune (March, 1964), 133-139.

Fourt, Louis A. and Woodlock, Joseph W. "Early Prediction
of Market Success for New Grocery Products,"
Jourggl of Marketing, XXV (January, 1960), 31-38.

 

 

Freeman, Orville. This Week, JUne 26, 1966, 2.

 

Hamelman, Paul W. and Mazze, Edward M. "Improving Product
Abandonment Decisions," Journal of Marketing, XXXVI
(April, 1972), 20-26.

 

Hassinger, Edward. "Stages in the AdOption Process," Rural
Sociologx. XXIV (March, 1959), 52—53.

 

Hunt, Shelby D. "Post-transaction Communications and Dis-
sonance Reductions," Journal of Marketing, XXXIV,
(July, 1970), 46-51.

 

King, Charles W. and Summers, John O. "Overlap of Opinion
Leadership Across Consumer Product Categories,"
Journal of Marketing Research, VII (February, 1970),
43-50.

 

Kotler, Philip. "Phasing Out Weak Products," Harvard

171

Business Review (March-April, 1965), 107-118.

Lavidge, Rdbert J. and Steiner, Gary A. “A Model for Pre-
dictive Measurements of Advertising Effectiveness,"
Journal of Marketing, XXV (October, 1961), 59—62.

Mancuso, Joseph R. "Why Not Create Opinion Leaders for New
Product Introductions?" Journal of Marketing,
XXXIII (July, 1969), 20-25.

 

Mansfield, Edwin. "Technical Change and the Rate of Imita-
tion," Econometrica, XXIX (October, 1961), 741—766.

Martilla, John A. "Word-of-Mouth Communication in the
Industrial AdOption Process," Journal of Marketing
Research, VIII (May, 1971), 173-178.

 

Miracle, Gordon. "Product Characteristics and Marketing
Strategy," Journal of Marketing, XXIX (January,
1965), 18—24.

 

Montgomery, David B. and Silk, Alvin J. "Clusters of Con-
sumer Interests and Opinion Leaders' Spheres of
Influence," Journal of Marketing Research, VII
(August, 1971), 317-321.

 

Mosteller, Frederick and Wallace, David S. "Inference in
an Authorship Problem," JOurnal of the American
gratistical Association, MVIII (June, 1963), 275—
309.

 

Namias, Jean. "Intentins to Purchase Related to Consumer
characteristics," Journal of Marketing, XXV (July,
1960), 32—36.

 

"New Product Revolution," Printers' Ink, October 31, 1968,
21-27.

 

"New Products: The Push in on Marketing," Business Week,
March 4, 1972, 72—77.

 

Oskikowa, Sadaomi. "Can Cognitive Dissonance Theory Explain
Consumer Behavior?" Journal of Marketing, XXXIII
(October, 1969), 50-55.

 

Parfitt, J. H. and Collins, B. J. K. "Uses of Consumer
Panels for Brand-Share Prediction," Journal of
Marketing Research, V (May, 1968), 131—145.

 

 

172

Reynolds, Fred D. "Problem Solving and Trial Use in the
AdOption Process," Journal of Marketing Research,
VIII (February, 1971), 100-102.

Robertson, Thomas S. "The Process of Innovation and the
Diffusion of Innovation," Journal of Marketing,
XXXI (January, 1967), 14-19.

 

Rothe, James T. "The Product Elimination Decision," MSU
Business Topics (Autumn, 1970), 45-52.

 

Ryan, Bryce and Gross, Neal C. "The Diffusion of Hybrid
Seed Corn in Two Iowa Communities," Rural Sociology,
VIII (March, 1943), 15-24.

 

Sheth, Jagdish N. and Venketesen, M. "Risk-Reduction Pro-
cesses in Repetitive Consumer's Behavior,"
Journal of MarketingyResearch, V (August, 1969),
307—310.

Wesson, Chester R. "What is "New" About a New Product,"
Journal of Marketing, XXV (July, 1960), 52-56.

 

Webster, Jr., Frederick E. "New Product AdOption in Indus-
trial Markets: A Framework for Analysis," Journal
of Marketing, XXXIII (July, 1969), 35-39.

 

 

 

. ”Informal Communication in Industrial Markets,"
Journal of Marketing Research, VII (May, 1970),
186-189.

Weiss, E. B. ”That Malarky About 80 Percent of New Products
Failing," Advertising Age, XXXVI (August 2, 1965),
101-107.

 

Uhl, Kenneth; Andrus, Roman and Poolsen, Lance. "How Are
Laggards Different? An Empirical Inquiry," Journal
of Marketing Research, VII (February, 1970), 51-54.'

Proceedings of American Marketing
Association Conferences

 

 

Allvine, Fred C. "Diffusion of a Competitive Innovation,"
Proceedings of 1968 Fall Conference. Chicago, 1968,
341—352.

 

Arndt, Johan.

173

"Perceived Risk, Sociometric Integration

and Word of Mouth in the AdOption of New Food
Products." Proceedings of 1966 Fall Conference.
Chicago, 1966, 644-48.

 

Cunningham, Scott M. "Perceived Risk as a Factor in the
Diffusion of New Product Information." Proceedings
of 1966 Fall Conference. Chicago, 1966, 698-721.

 

 

Ehrenberg, A.

S. C. "Predicting the Behavior of New Brands,"

Abstracts of 1970 Fall Conference. Chicago: 1970,

 

92.

Haines, Jr.,

George H. "A Study of Why PeOple Purchase New

Products," Proceedings of 1966 Fall Conference.
Chicago, 1966, 685-697.

Howard, John A. and Sheth, Jagdish N. "Theory of Buyer
Behavior." Proceedings of 1967 Winter Conference.
Chicago, 1968, 253-262.

 

Jacoby, Jacob. "Innovation Prowess as a Function of
Personality," Abstracts of 1970 Fall Conference.
Chicago: 1970, 121.

Kemm, Thomas

 

R. "Scientific Experimentation in Product

Innovation," Proceedings of 1966 World Congress.
Chicago, 1966, 241-246.

 

King, Charles W. and Summers, John 0. "Technology, Innova—

tion

and Consumer Decision Making." Proceedings

 

of the 1967 Winter Conference. Chicago, 1968,

 

63-68.

Ozanne, Urban B. and Churchill, Gilbert A. "Adoption
Research: Information Sources in the Industrial
Purchasing Decision." Proceedings of 1968 Fall
Conference. Chicago, 1968, 353-362.

 

 

Rockwood, Peris Emmett, moderator. ”The Diffusion of
Innovation Viewed as an Information System."
Proceedings of 1967 Winter Conference. Chicago,

1968,

 

395-396.

Rosenbloom, Richard S. "Product Innovation in a Scientific
Age," Proceedings of 1966 World Congress. Chicago,

1966,

 

247-259.

174

Summers, John. "New Product Interpersonal Communications,"
Combined Proceedings Spring and Fall 1971 Confer-
ences. Chicago, 1971, 433.

Taylor, Stanford H. "Keys to Success in Product Introduc-
tion,"Ԥrpceedings of the 1966 World Congress.
Chicago: 1966, 229—237.

Wentz, Walter B. and Eyrich, Gerald I. "New Product Fore-
casting Without Historical, Survey or EXperimental
Data," Proceedings of 1968 Fall Conference.
Chicago, 1968, 215—221.

 

Other Sources

 

 

Beal, George M. and Bohlen, Joe M. The Diffusion Process,
Special Report No. 18, Agricultural Extension
Service - Iowa State College, Ames, Iowa, 1957.

 

Bell, William E. "Consumer Innovation: An Investigation
of Selected Characteristics of Innovators," Unpub-
lished D.B.A. dissertation, Michigan State univer—
sity, East Lansing, 1962.

Bohlen, Joe M. How Farm Pegple Accept New Ideas. Special
Report No. 15, Agricultural Extension Service -
Iowa State College, Ames, Iowa, 1955.

 

 

Booz-Allen and Hamilton, Inc. Managgment of New Products.
New York, 1968.

 

Cunningham, William H. "An Analysis of Market Segmentation
on the Basis of Emotional-Attitudinal Variables in
the United States Compact Car Market." unpublished
Ph.D. dissertation, Michigan State university, East
Lansing, 1970.

Grashof, John R. "Information Management for Supermarket
Chain Product Mix Decisions: A Simulation Experi—
ment." Unpublished Ph.D. dissertation, Michigan
State University, East Lansing, 1968.

Gross, Neal C. "The diffusion of a Culture Trait in Two
Iowa Tornships." Unpublished M.S. thesis, Iowa
State College, Ames, Iowa, 1942.

175

Press Release, May, 1965, Sales and Marketing Executive
International, New York.

Ryan, Bryce. Social and Ecological Patterns in the Farm
Leadership of Four Iowa Townships. Iowa Agri-
cultural Experiment Station Bulletin 306, Iowa
State College, Ames, Iowa, 1942.

Wilkening, Eugene A. Acceptance of Improved Farm Practices.
North Carolina Agricultural Experiment Station
Bulletin #98, Raleigh, N. C., 1952.