DEMAND ESTIMATION IN A PUBLIC UTILITY
AN EMPIRICAL MODEL

BY

Richard Frank Sauter

A THESIS

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

DOCTOR OF BUSINESS ADMINISTRATION

Department of Marketing and Transportation Administration

1970

ABSTRACT

DEMAND ESTIMATION IN A PUBLIC UTILITY:
AN EMPIRICAL MODEL

BY

Richard Frank Sauter

Public utilities often must serve a large number of

different types of industrial users. These customers differ

greatly in their demand requirements and purchase character—
istics. Estimating the demand requirements of these

customers is necessary if effective and equitable rates,

product offerings, and selling strategies are to be

developed.

This study provides increased understanding of the
major factors that influence demand in the industrial
market segment of a public utility. It does so by (a) pro-
\dding a conceptual framework for analyzing the variables
thatnflght affect the industrial concern's demand behavior,
0” employing empirical data to measure and test the
degree to which these factors influence the demand for a
SPGCific utility's offerings, and (c) developing statistical

Inc“3-813 that specify the relationship between the most

critical explanatory factors and selected demand criteria.

Q

-
.Dt'

 

- a
Mac ~.-

Niovya-

 

 

:‘V'As..'

~"‘¥-‘-

 

 

._ V
N.' ‘
at
a
Q.
Q. '
\ '~.. ‘
an“ '-
. “
o. ..
..
.

I)!

 

 

Richard Frank Sauter

The utility selected for analysis was the Michigan

Bell Telephone Company. A stratified sample of manufacturers,

wholesalers, retailers, realtors, and dry cleaners was

drawn from a geographic segment of Michigan Bell's market.

Questionnaires were sent to owners or managers of eadh

sampled firm. Information was obtained concerning the

structural characteristics of the firms and the respond-
ent's attitudes toward the value and usage of various

telephone communication offerings. These data were used

as the independent variables in the analysis.
The principle demand variables investigated in the
study include both intrastate and interstate toll service,

Yellow Page Advertising usage, total telephone equipment

empenditures, trunk demand, and the number of telephone

hxmtions demanded. Data concerning these variables were
Sflthered from Michigan Bell Telephone Company records and

serve as the dependent variables in the study.

Factor analysis and simple correlation analysis
““53 used to identify and extract the most important

e’<E>.'l.«:-1natory variables from the large set of initial pre-

‘fiCtors. Multiple regression analysis was then performed

 

Richard Frank Sauter

on the reduced set of predictors and each demand criterion.
The regression models served as the basic forecasting
models in the study.

Both structural and attitudinal variables were
analyzed to determine their effect on the firm's telephone
communications patterns. Structural factors best explained
the variances in individual firm demand behavior. Variables
associated with firm size were correlated highly with
equipment demand for most business segments. Structural
factors related to the geographical location of customers
and suppliers were correlated highly with toll usage--par-
ticularly for firms in the manufacturer and wholesaler
segments. In all cases, linear forecasting models best fit
the relationships between structural characteristics and
demand criteria .

Attitudinal factors did not contribute a significant
amount of additional information to the structurally based
(kmand models. Although these factors varied among business
ESegments, they did not differ substantially for respondents
Vfithin each segment. Variations in attitudinal responses,

however, did suggest that marketing or selling strategies

 

Richard Frank Sauter

should vary for different customer types.

The respondent's perceived or expected reaction to
an increase in toll rates varied both within and across
customer groupings. In general, respondents from smaller
firms expected toll usage to decrease much more in
response to a rate increase than did those in larger firms.
Also, telephone communications with suppliers were expected
to be affected more by a rate increase than were communica-
tions with customers.

Several findings also pointed to a need for more
effective communications between the utility and its
customers. A large number of respondents perceived their
understanding of the communications alternatives available
to them to be quite poor. Also, many respondents indicated
that they depended almost completely on the utility sales—
man for suggestions concerning both the types and amount
0f telephone communications equipment or services to utilize.

In summary, this research focused on the factors
influencing the business firm's demand for a number of
telephone utility offerings. However, many of the findings.

implications, and recommendations derived from the study

 

Richard Frank Sauter

are of value to other types of utilities and non-regulated
industries as well. It is hoped that one of the most
significant contributions of this research is to provide

a framework or focus for structuring and carrying out
further utility-related studies. Certainly, further
managerial and regulatory-oriented research is needed in
all utility-based industries if a better matching of
utility resources with customer needs and a more efficient

allocation of company marketing efforts is to be attained.

CDCopyright by
RICHARD FRANK SAUTER

1971

 

 

ACKNOWLEDGMENTS

Many individuals contributed to the successful
completion of this dissertation. Although it is impossible
to acknowledge all of those who provided time, energies,
and encouragement to this work, special mention should be
made of some.

To the anonymous companies that responded to requests
for information, I extend my thanks. Their assistance
provides hOpe for an expanded interraction between the
business and academic communities.

The Michigan Bell Telephone Company provided detailed
data on customer demand characteristics even though the
gathering of such data resulted in a partial disruption
of their normal business activities. I especially want to
thank Mr. Al Mathers of Michigan Bell for his many hours
Of help in searching out and tabulating much of the required
information. My appreciation is also extended to a number
of other Michigan Bell employees for their free exchange

Of ideas and helpful suggestions.

ii

 

The Institute of Public Utilities of Michigan State
University, directed by Professor Harry M. Trebing, pro-
vided funds for printing and mailing questionnaires.
Professor Trebing also served as a member of the dissertation
committee and contributed freely of his time and energies.
His advice and counsel were most helpful.

A Ford Foundation dissertation grant also provided
funds which helped insure the completion of the study.

Professor Jerome Herniter served as a member of the
dissertation committee and his constructive suggestions
during the formative stages of the research were most
valuable.

To Professor William Lazer, Who served as chairman
of both my graduate committee and the dissertation committee,
go my sincerest thanks. He was a constant source of
encouragement and assistance. Most important, however,
Professor Lazer made the thesis research a true learning
experience. For this, I am forever grateful.

Finally, to my wife Karen I owe much more than can be
acknowledged here. For the many drafts she typed, I offer
my sincere appreciation. For the encouragement, patience,

and inspiration she continually provided, I can offer only

my love. ...
111

TABLE OF CONTENTS

Page
AcmOWI‘EDGMmTSOO..0000.00000 ......... 0......0...0000 ii
LIST OF TABLES.....00.0...000..... 00000000000000 oooooViii
Chapter
I INTRODUCTION0000000000....... ..... 0.0....0.0... 1
Purpose of the Research............... ..... .... 6
The Telephone Communications Industry.......... 8
Local, Intrastate, and Interstate Exchange
Rate8000000000.00000.0.000....000000000 ..... 0 12
Background of the Problem...... ....... . ..... ... 15
Problem Statement......... ...... .... ........ ... 18
General Hypotheses00000000.000000 ..... 00.0...00 21
methOd Of Investigation0.0.0.....000...00..00.0 22
Limitations of the Study....... ..... ........... 25
Terms and Definitions.............. ...... ...... 26

contributions00.0.0.0000.0000....0.0..0......00 29

II MODELING DEMAND IN THE INDUSTRIAL MARKET
SECTOR-éA CONCEPTUAL FRAMEWORK

IntrOduCtionoo00000000000000.0000.coo... 000000 . 34
Classification Methods in Industrial Markets... 34

Classification by Product................... 36
Segmentation by Class of Trade.... ......... . 37
Classification by End-Use................... 42

Factors Influencing the Demand for

Industrial Goods............................. 44
Techniques Used to Model Demand in the

Industrial Market............................ 49

Sales Force Estimates....................... 50
Regression/Correlation Analysis............. 51
Discriminant Analysis....................... 53
Probabilistic Models........................ 55

iv

TABLE OF CONTENTS (Continued)

Chapter Page

III

IV

summary000..00000000000.0000.....00.000000...00 58

RESEARCH DESIGN AND METHODOLOGY

Questionnaire Construction............. ........ 6O
Dependent Variables................ ............ 71
Sample Selection................ ....... . ..... .. 73
Questionnaire Administration.............. ..... 87
Statistical Techniques...... ................... 89
Summary.......................... .............. 94

DETERMINATION OF THE FUNDAMENTAL
VARIABLES AFFECTING DEMAND

General Characteristics of Firms......... ...... 96

Some General Characteristics of Firm Demand.... lOl

Attitudinal Characteristics of Responding
Firms........................................ 109

Relationships Between Criterion Variables ...... 119
Simple Correlations Between Independent
and Criterion Variables.............. ..... ... 123
Manufacturer Segment Associations ....... .... 124
Wholesaler Segment.............. ............ 127
Retailer Segment............... ..... ........ 130

Realtor and Dry Cleaner Segments............ 133

Some General Comments and Comparisons

of the Correlation Matrices.................. 133
Sources of Variation in Demand Variables—-

Factor Analysis.............................. 142
Summary........................................ 152

DEMAND MODELS
Sources of Demand Variation--Regression Models. 154

Preliminary Regression Results.............. 159
Variable Addition and Deletion.............. 162
Contribution of Additional Explanatory
Variables to Demand Models................ 167
v

TABLE OF CONTENTS (Continued)

Chapter Page
Yellow Page Advertising Demand Models .......... 171
Toll Forecasting Models.............. .......... 177

Expenditures for Equipment..................... 185
Equipment Demand--Number of Lines and

Locations....................... ............ . 188
Validation Sample Results.......... ...... . ..... 192
Effect of Attitudinal Factors on the

Demand Models. ......... ....... ............... 196
Probabilistic Demand Models: Concept

and Illustration.......... ...... . ............ 210
Summary.................... ................ .... 222

VI SUMMARY AND CONCLUSIONS

Introduction............ ..... . ..... . ........... 225
General Findings................... ....... ..... 227
Implications of the Study.......... ...... ...... 231
Suggestions for Additional Research.. ...... .... 239

Concluding Comments............................ 244
BIBLImRAPHYoo'o0.0.00.0000000000o 00000000 00000000 245
APPENDIX

I Ratios for Selected Criterion and
Independent Variables. ..... .. ............. .. 254

II Kendall Coefficient of Concordance
for Business Segments on the Time and
Importance Placed on Various Telephone
Communication Activities.................... 255

III Percent of Firms in Each Segment that
Perceived Their Telephone Usage to be
Less than Equal to or Greater than Firms
of Comparable Type and Size................. 260

IV Rotated Factor Loadings for Each
Criterion Variable and Associated
Explanatory Variables--By Business
Segment..................................... 267

vi

TABLE OF CONTENTS (Continued)

Chapter

APPENDIX (Continued)

V Cover Letters, Permission Request
Forms and Questionnaires........ ..... .......

vii

Page

289

LIST OF TABLES
Table Page

3.1 Classificatory Factors Hypothesized to
Affect Demand--By Business Segment............. 63

3.2 Attitudinal Factors Hypothesized to
Affect Demand00....0..0....00.0.0.0.0... ...... . 68

3.3 Dependent variables Reflecting Firm
Demand for Telephone Offerings... ......... ..... 72

3.4 Business Segments Selected for Analysis--
by SIC and Number of Firms..................... 80

3.5 Pilot Study Results Showing Variance in
Firm Usage Characteristics for Five
Business Segments.............................. 82

3.6 Number of Establishments Included in
Manufacturer, Wholesaler, and Retailer
Samples--by Firm Size.......................... 86

3.7 Number of Establishments Included in the
Realtor and Dry Cleaner Samples--by Firm

Size00...000000..00000..00......00.0.0.0...0... 88

3.8 Questionnaire Mailing and Response
Informtion.0.0000000000000000.000....000.000.0 91

4.1 Size Distribution of Firms--Number of
Employees by Business Segment........... ..... .. 98

4.2 Size Distribution of Firms--1967 Gross
Revenue by Business Segment....................100

4.3 Selected Characteristics of Firms
Sampled--by Business Segment...................102

4.4 Mean Intrastate and Interstate Toll

Billings for February, 1968--by Business
segInent..0........00....0..0..0..0.0...........104

viii

LIST OF TABLES (Continued)

Average Equipment and Service Utilization
for February, 1968--by Business Segment........

Demand and Usage of Other Equipment and
Services for February, 1968--by Business
segnlent.00...........00........................

Comparison of Mean Time and Importance
Rankings of Local Telephone Usage--by
Business Segment....................... ....... .

Comparisons of Mean Time and Importance
Rankings of Toll Telephone Usage--by
Business Segment......................... ..... .

Perceived Telephone Usage Relative to
Firms of Similar Type and Size............ .....

Correlation Matrix of Criterion
Variables--by Business Segment...... ...... .....

Simple Correlation Matrix for Selected
Explanatory and Criterion Variables--
Manufacturer Segment.................. .........

Simple Correlation Matrix for Selected
Explanatory and Criterion Variables--
Wholesaler Segment.................... ..... ....

Simple Correlation Matrix for Selected
Explanatory and Criterion Variables--
Retailer S€gIn€nt00.0..0.000...0.0.0..0. ...... .0

Simple Correlation Matrix for Selected
Explanatory and Criterion Variables--
Realtor segment00000.0.00000.0....00.0.0.0.0...

Simple Correlation Matrix for Selected
Explanatory and Criterion variables--
Dry Cleaner Segment............................

ix

Page

105

108

112

113

115

120

125

129

131

134

135

5.2

5.3

504

LIST OF TABLES (Continued)

Number of Independent Variables
Significantly Correlated with Selected
Criterion Variables--by Business Segment....... 139

Rotated Factor Loadings for Total Toll
and Associated Explanatory Variables--
Manufacturer Segment........................... 147

Explanatory Variables Having Highest

Loadings on Each Factor in the Rotated

Factor Loading Models--by Dependent Variable

and Business Segment........................... 151

Explanation of Independent Variable Symbols
Used in the Chapter V Tables.......... ..... .... 158

Coefficients of Determination for

Regressions of Demand Criteria on the

Explanatory variables Selected from Factor
Models--by Business Segment.................... 161

Coefficient of Determination for Regressions

of Demand Criteria on a Reduced Set of

Explanatory Variables Selected from Factor
Models-~by Business Segment.................... 165

Predicting Equations for Yellow Page
Advertising Usage-~by Business Segment......... 172

Predicting Equations for Total Toll
Usage-~by Business Segment..................... 178

Predicting Equations for Intra-State Toll
Usage--by Business Segment.......... ........ ... 181

Predicting Equations for Inter-State Toll
Usage--by Business Segment..................... 182

Predicting Equations for Telephone Equipment
Expenditures--by Business Segment.............. 186

X

5.13

II.l

II.2

LIST OF TABLES (Continued)

Page
Predicting Equations for Number of
Telephone Lines--by Business Segment ........... 189
Predicting Equations for Number of
Telephone Locations-~by Business Segment ...... . 190

Comparison of Regression Results Obtained
from Analysis and Validation Samples........... 195

Number of Firms Expecting Declines in
Inter-State Toll Usage Resulting from a
20% Increase in Toll Rates. .......... . ......... 203

Expected Responses of Heavy and Light
Inter-State Toll Users to a Toll Rate
Increase--by Business Segment.................. 206

Conditional Probability Matrix Relating
Line Demand to Number of Employees.... ......... 215

Conditional Probability Matrix Relating
Line Demand to Number of Offices.... ........... 217

Conditional Probability Matrix Relating
Line Demand to Number of Offices and
Number of Employees...................... ..... . 218

Ratios for Selected Criterion and
Independent Variables-—by Business Segment..... 254

Mean Rankings of Local Telephone Usage
Time Spent on Various Communication
ACtiVj—ties000000000000000000000.....0....00.... 256

Mean Rankings of Toll Telephone Usage

Time Spent on Various Communication
ACtiVities...000000000000...00 000000 000000 00000 257

xi

Table

II.3

II.4

III.1

III.2

III.3

III.4

III.5

III.6

IV.1

IV.2

LIST OF TABLES (Continued)

Page

Mean Rankings of Importance of Local
Telephone Usage on Various Communication
ACtiVitieS...000....00000.00.....00 ......... .0 258

 

Mean Rankings of Importance of Toll
Telephone Usage on Various Communication
ACtiVj—tie500000000000.00.0 .................. 0. 259

 

Perceived Amount of Local Telephone
Usage As Compared to Firms of Similar
Type and Size-~by Segment.......... ......... .. 261

Perceived Amount of Toll Time Used As
Compared to Firms of Similar Type and
Size-~by Segment................... ........... 262

Perceived Amount of Telephone Equipment
Used as Compared to Firms of Similar
Type and Size-~by Segment........ ........ ..... 263

Perceived Amount of Telephone Time

Used to Contact Suppliers As Compared

to Firms of Similar Type and Size--by
Segment............................... ........ 264

Perceived Amount of Telephone Time Used
to Contact Customers As Compared to
Firms of Similar Type and Size--by Segment.... 265

Perceived Amount of Yellow Page Advertising
Used as Compared to Firms of Similar Type
and Size--by segmentoo....o................... 266

Explanation of Abbreviated Variable
Symbols Used in Tables IV.2 - IV.20........... 267

Rotated Factor Loadings for Total Toll

Demand and Associated Explanatory
variables--Manufacturer Segment....... ........ 270

xii

Table

IV.3

IV.4

IV05

IV.6

IV.7

IV.8

IV.9

IV.10

IV.11

Iv. 12

LIST OF TABLES (Continued)

Page
Rotated Factor Loadings for Intrastate
Toll Demand and Associated Explanatory
variables--Manufacturer Segment......... ...... 271
Rotated Factor Loadings for Interstate
Toll Demand and Associated Explanatory
variables--Manufacturer Segment.... .......... . 272
Rotated Factor Loadings for Equipment
Demand and Associated Explanatory
Variables--Manufacturer Segment..... .......... 273
Rotated Factor Loadings for Number of Lines
Demanded and Associated Explanatory
Variables--Manufacturer Segment...... ..... .... 274

Rotated Factor Loadings for Number of
Locations Demanded and Associated Explana-
tory variables--Manufacturer Segment.......... 275

Rotated Factor Loadings for Total Toll
Demand and Associated Explanatory
Variables-~Wholesa1er Segment................. 276

Rotated Factor Loadings for Intrastate
Toll Demand and Associated Explanatory
Variables--Wholesa1er Segment................. 277

Rotated Factor Loadings for Interstate
Toll Demand and Associated Explanatory
Variables-~Wholesaler Segment................. 278

Rotated Factor Loadings for Equipment
Demand and Associated Explanatory
Variables--Wholesa1er Segment................. 279

Rotated Factor Loadings for Number of

lines Demanded and Associated Explanatory
Variables-~Wholesaler Segment................. 280

xiii

Table

IV.13

IV.14

IV.15

IV.16

IV.17

IV.18

IV.19

IV.20

LIST OF TABLES (Continued)

Rotated Factor Loadings for Number of
Locations Demanded and Associated

Explanatory Variables--Wholesa1er
Segment........................... ............

Rotated Factor Loadings for Yellow Page
Advertising Demand and Associated Explana-
tory Variables--Retai1er Segment..............

Rotated Factor Loadings for Total Toll
Demand and Associated Explanatory
Variables--Retai1er Segment...................

Rotated Factor Loadings for Intrastate
Toll Demand and Associated Explanatory
Variables--Retai1er Segment........... ....... .

Rotated Factor Loadings for Interstate
Toll Demand and Associated Explanatory
variableS--Retailer segment. 0 0 0 0 0 0 0 0 . 0 0 0 . . 0 0 0 0

Rotated Factor Loadings for Equipment
Demand and Associated Explanatory
Variables--Retai1er Segment.......... ...... ...

Rotated Factor Loadings for Number of
lines Demanded and Associated Explanatory
Variables--Retailer Segment...................

Rotated Factor Loadings for Number of

Locations Demanded and Associated
Explanatory Variables-~Retai1er Segment.......

xiv

Page

281

282

283

284

285

286

287

288

CHAPTER I

INTRODUCTION

Firms concerned with the production, marketing,
and distribution of such products as gas, electricity,
and communication devices comprise the "public utility"
sector of our economy. As part of the "regulated industries"
they have long been the object of detailed local, state
and federal regulation.

The subject of marketing in the regulated industries
has received scant attention and study. Various authors
have concerned themselves with the general subject of

1 But little has been done in

regulatory economics.
examining the application of various marketing principles
to the public utilities, or in developing systematic

analysis of marketing strategies in regulated environments.

Although many of the modern marketing concepts and techniques

 

1For examples of this see: Charles F. Phillips, Jr.,
The Economics of Regulation (Homewood, Illinois: Richard D.
Irwin, Inc., 1965), pp. 656-707; Dudley F. Pegrum, Public
Regulation of Business (Homewood, Illinois: Richard D. Irwin,
Inc., 1965), pp. 637-721; Martin G. Glaeser, Public Utilities
in American Capitalism (New York: The MacMillan Company, 1957).

 

l

apply to regulated as well as the non-regulated industries,
research and testing have been performed primarily with
firms existing in the nonregulated environment.

Yet the regulated industries, particularly the public
utilities, are performing an increasingly vital role in
the workings of our economy. The importance of these
industries can be seen in their contribution to National
Income. Approximately eight percent of the National Income
originates in the regulated industries with close to
one-half of this originating in the public utilities.1

The communications sector of the public utilities,
on which this study concentrates, contributes nearly two
percent to total National Income and since 1960 has
accounted for nearly ten percent of all new plant and

equipment expenditures.2

Also, as measured in total assets,
the telephone segment of the public utilities industries
has grown at a faster rate than any other--nearly tripling

in size since 1947. Over eighty percent of the families

in the United States, as well as an even larger share of

 

1The Economics of Regulation, p. 7.

2Business Statistics, 1969 Edition, U. S. Department
of Commerce, p. 9.

the business sector, presently utilize this service.1

The fact that the regulated industries of this
country are a combination of private ownership and public
control causes significant problems and conflict between
the goals of profit maximization and public service at
the lowest possible price. It generally is felt that
regulation of public utilities seeks to provide goods and
services to the consumer at "reasonable rates." However,
if regulation in such industries is to provide the impetus
and motivation for technological innovation, then it must
be approached in terms of its effect on the total benefits
to society and not just in respect to some particular
"reasonable rate of return."

One of the prime ways in which a firm benefits
society is by Satisfying the changing wants and needs of
its customers. The relative success of a firm Operating
in a competitive environment often depends greatly on how
well it meets the needs of the consumer. The search for
a differential advantage through greater want satisfaction

is what leads to continuing technological innovations and

 

1The Economics of Regulation, pp. 8-11.

a dynamic competitive environment.1 Although most public
utilities operate under regulatory constraints and in
monopolistic types of market structures, varying amounts
and types of competition do exist. For example, firms
operating in the natural gas industry face competition
in many markets from the electric and oil utilities. Most
transportation industries compete directly with each
other for certain segments of the market. In most utilities
there is the long run threat of substitute products being
developed by other regulated or non—regulated firms. Also,
new technology is a constant threat to the monopolistic
position that many utilities enjoy today.2 Even where no
direct substitutes are available for a particular utility's
offerings, the firm still must vie for its share of the
consumer's increasing discretionary income.

Competition in all of these forms has caused the

public utilities in recent years to become more market

 

For a rigorous discussion of the concept of "competi-
tion for differential advantage" see Wroe Alderson, Marketing
Behavior and Executive Action (Homewood, Illinois: Richard
D. Irwin, Inc., 1957), pp. 101-129.

2

 

For an example of this type of threat, see B. Herber,
"The Impact of Microwave on the Telephone Business," Public
Utilities Fortnightly, LXX (August 16, 1962), pp. 214-15.

oriented. Utilities have increased their advertising
budgets, developed a variety of new and diversified
products and services, and conducted studies to determine
the effect of both demographic and socio-psychological
variables on demand.1 These are but a few indications of
the utility's effort to understand its customers better
and to compete with other durables and non-durables for
the consumer dollar. Independent researchers are performing
other studies which will provide additional knowledge of
consumer purchase behavior and buyer attitudes toward the
products and services which the public utilities offer.2
The attempts of the utilities to anticipate, stimulate,
and foster change in customer thinking are not only
occurring in the consumer or residential sector of the
market, but in the business sector as well. However, before
the requirements of either of these segments can be deter-

mined and before successful marketing strategies can be

 

1For example, see Report Issued jointly by the Opinion
Research Corporation and the Long-Range Planning Group of
Michigan Bell Telephone Company entitled Profile of the
Michigan Bell Consumer in May, 1967.

 

2For example, see William Stephenson, "Public Images
of Public Utilities," Journal of Advertising Research
(December 1963), pp. 34-39, and Thomas S. Robertson, "An
Analysis of Innovative Behavior and Its Determinants,"
(unpublished Ph.D. Dissertation, Northwestern University, 1967).

developed, the needs of individual customers in each
buying situation and the roles of the major buying

influences must be understood.

Purpose of the Research

"Effective marketing strategies, with respect to both
industrial and consumer goods and services, require
knowledge of the buying process, the buyer's behavior, and
the various influences and motivations that affect this

behavior."1

This is especially true in the industrial
market Where there are fewer customers but where the amount
of a product or service that a single buyer demands is often
large. To develop proper product offerings and marketing
strategies for these customers, the selling firm must
profile the characteristics associated with various types
of business customers and understand their respective
effects on demand.

This study seeks to provide increased understanding

of the major determinants and variables that influence demand

in the industrial market segment of a public utility. It

 

1Patrick J. Robinson, Charles W. Paris, and Yoram
Wind, Industrial Buying and Creative Marketing (Boston:
Allyn & Bacon, Inc., 1967), p. 214.

does so by (a) laying a theoretical framework for analyzing
the factors that might affect the industrial concern's
demand behavior; (b) employing empirical data to measure
the degree to which these factors influence the demand for
a specific utility's offerings; and (c) developing
statistical models that indicate the relationship between
the most critical explanatory factors and specific demand
criteria.

The utility selected for analysis in this study is
the telephone industry. Telephone utilities are in the
somewhat unique position of renting and selling their
equipment and services to all business firms in a given
geographical area. It would be very useful, therefore,
for such a utility to be able to estimate the potential
usage of its equipment and services by individual firms.

Accordingly, the main purposes of this study are:

1. To develop statistical forecasting models that

recognize the role that selected "structural"l

 

l"Structural" will be used in this study to character-
ize specific variables associated with the firm much as
"demographic" is used to represent certain variables related
to households. Since these "structural" variables are
traditionally used as the basis for classification (i.e.,
number of employees, gross sales, etc.), these variables
will also be referred to as "Classificatory" factors. For
a comprehensive listing of the basic structural or classi-
ficatory variables used in this study, see Chapter III.

variables associated with a given industrial
concern play in determining the demand for
various types of telephone communications
equipment and service.

2. To determine the degree to which the decision
maker's personal attitudes toward the value
of communication influence the accuracy of
these demand models.

3. To implement and test the models.

4. To evaluate the managerial implications of
the models.

5. To investigate the regulatory implications of
the models.

6. To illustrate how the models can be used
to assist management in strategy and planning

decisions.

The Telephone Communications Industry

A comprehensive analysis of all aspects of the
telephone industry and its various regulatory constraints
is not critical to the development of this study. However,
a brief review of some of the unique characteristics of

the telephone market is necessary. This section of the

paper examines briefly the changing technology and market
structure in certain areas of the communications industry.
Also some of the factors that affect rate-making policy
in the local exchange and toll telephone markets will be
described.

The telephone industry in the United States is
privately owned but heavily regulated by the Federal
Communications Commission and various state agencies.
American Telephone and Telegraph is the dominant firm in
the industry; however, a few moderate sized firms such as
International Telephone and General Telephone and a large
number of small independent firms exist to serve primarily
rural areas.

The telephone utility, like most other utilities,
is characterized by a large investment in plant and
equipment. To satisfy the potential demand that exists
for communications services at any one time, a high fixed
investment in lines, switching apparatus, and associated
equipment is needed. Despite the fact that this equipment
is fully utilized only during peak demand periods, rates
must be set to recover the costs associated with full capacity

requirements. This means that a portion of the rates are

10

paid by potential as well as actual users.

There is little direct competition today among
communication industries for much of the residential and
business telephone services. This is particularly true
in the residential sector where, for many product and
service offerings, the local telephone utility enjoys a
near monopoly position. In recent years, however, several
factors have led to increased competition in the communica-
tions industry. Direct competition among communication
utilities has developed for such offerings as private lines
and data transmission services. Technological advances
have also introduced competitive elements into the larger
scale areas of supplying communication needs to both big
business and government.1 Competition was also intensified
following the 1959 Federal Communications Commission's

2 which allowed firms to develop their own microwave

decision,
systems for the transmission of both voice and nonvoice

messages. This decision led several large firms and the

federal government to adOpt or increase their usage of private

 

1Charles F. Phillips, Jr., pp, cit., p. 692.

2In the Matter of Allocation of Frequencies in the
Bands Above 89Mc., Docket No. 11866, Temporary Decision,
August 6, 1959 (issued on July 30, 1959).

 

ll

microwave systems,1 and in turn caused AT&T to offer new
communications offerings to customers with extensive
communication needs.2
An additional result of the microwave decision was
the raising of some critical questions concerning the
role of regulatory policy in the communications industry.
It has been suggested that there are certain inequalities
in existing regulatory practices since ". . . one side
of the competitive market--the private systems and their
suppliers--face little or no regulation. However, the
other side of the competitive market, the established
common carriers such as Bell and Western Union, are
closely regulated as public utilities.”3
Directly related to the research in this study is
the suggestion that a new regulatory philOSOphy be developed.

This new philosophy is based on a recognition of the

monopolistic character of the residential market and the

 

lHerber, "The Impact of Microwave . . .,” pp, cit., p. 220.

 

2"Offering Similar to Telpak Would Have Been Made,"
Telecommunications Report (July 16, 1962).

3Bernard P. Herber, "Telephone Industry Reaction to
Microwave Competition,” Public Utilities Fortnightly, LXX
(October 25, 1962), pp. 634-35.

 

12

more competitive situation in the business-government
market.l Before any new regulatory or rate-making policies
can be developed, more detailed analysis of the demand
elasticities of the various communications markets is
needed. Also required is an evaluation of the possible
effects that a dynamic technology and an increased competi-
tive environment will have on the structure of the
different communication markets. Only by accurately
identifying and measuring the effects of these forces can

a flexible pricing policy be developed. This regulatory
policy must not unfairly discriminate between the residential
and commercial market segments or among the many sub—markets

within these segments.

Local, Intra-State, and Inter—State Exchangp Rates
Little is known about the exact characteristics or
elasticities of telephone demand. Based on a general
understanding of customer demand elasticities, however,
some discriminatory rates have been established. For
example, most local service demand and some residential
and commercial long distance demand is considered highly

inelastic. Some other types of demand, such as long distance

 

lIbid., p. 636.

13

social calls and certain types of business toll communica—
tions are more elastic. While the rate structures for
these telephone services are based primarily on the costs
associated with each offering, rates are often also based
on the perceived price sensitivities of the different
demand segments. A familiar example of this rate-making
strategy is the different rates charged day and night
users of toll services. In this case, the higher day rate
applies to the more inelastic business calls while the
lower evening rate applies to the more elastic family or
social calls. Several other examples of discriminatory
rate-making policies are given in the following paragraphs
in Which the structure of both local exchange service and
toll service is examined in greater detail.1

The fundamental unit used to establish local rates
is called an exchange. The exchange typically includes
all inhabitants of a city or town and certain outlying
or suburban areas. The geographical limits of the exchange
are not fixed but may vary as local population configurations

dictate. Price differentiation among exchange units is

 

1For a more detailed examination of the telephone
industry rate structure, see Chalres F. Phillips, Jr.,
220 Cite, pp. 362-370.

14

based primarily on the size of the exchange or the number
of telephones that can be called within an exchange
without a toll charge being added. In addition to differ-
ences in cost, price discriminations within an exchange
are based on the class of user (residential vs. commercial)
and the number of subscribers using a single line. In
this way, flat rates, based primarily on cost but modified
according to differing elasticities of demand, are charged
to broad classes or categories of users. Some cities

also offer what are termed message rates to subscribers.
In these cases a flat rate is charged a subscriber as

long as he does not exceed a specified number of message
units. Any calls over the specified number of units are
then billed at some additional uniform price.

In addition to the various local exchange rates,
users are also charged for all installations of equipment
and for specialized or extra vertical types of equipment.
For example, those buyers who are willing to pay the added
price may adopt equipment such as private-branch exchanges,
extension telephones, and colored or special design
telephone equipment.

Toll rates for both intrastate and interstate service

are also differentiated for both different types of uses

15

and users. As in the local exchange case, price discrimina-

tion is based both on cost and user demand elasticities.

These rates reflect such factors as "(1) distance; (2) whether

completion is requested to a specified person or to any-

one who answers; (3) time of day; and (4) length of con-

versation (holding time)."1
In general, local and statewide exchange and toll

rates are set to provide rates that are similar under

like conditions and provide the supplying utility with a

reasonable rate of return. This process might discriminate

against some users by granting equal rates to both high

and low cost areas. However, it leads to a system whereby

users in different exchanges pay equal amounts for the same

relative amount of service.

Background of the Problem

Traditionally, most firms selling to the industrial
market sectors have relied on government statistics as a
basis for generating market demand estimates. For example,
statistics concerning the number, type, and size of
business firms located in any given city are readily avail-

able in various government publications. These statistics

 

lIbid., p. 367.

16

provide the firm with rough macro estimates of potential
demand in given regions or specific industrial groupings.1
This information does not, however, help a firm develop
estimates of individual firm demand behavior. This demand
problem has not received the consideration or emphasis
devoted to the more macro oriented estimation techniques.
There are several basic reasons for this:

1. Many firms do not feel the need of estimating
demand on other than a region or industry basis.

2. Individual firms, like individual consumers,
are extremely varied in their wants and needs.
This heterogeneity causes problems in forecasting
individual firm demand.

3. Individual biases and value systems of
purchasing agents or corporate managers often
interfere with the rational selection of goods
and services based solely on firm needs or

requirements.

 

1For example see: E. J. Hermer, "On Mapping Industrial
Markets," Dunn's Review, Vol. LXI (May, 1953), p. 39, and
Code Koning, "Effective Techniques in Industrial Marketing
Research," Journal of Marketing, Vol. XXVIII (April, 1964),
pp. 57—61.

 

l7

4. It is more difficult to forecast sales for a
particular firm than it is for an entire
industry. This stems from the fact that most
of the data needed for forecasting are available
on an industry-wide rather than an individual-
firm basis.

If demand and the determinants of demand in the
industrial sector are to be understood and accurately
quantitifed, analysis must take place on as small a
customer unit or segment as possible. For if such marketing
strategies as personal selling, new product development,
and advertising are to be effective, they must be based
on detailed knowledge of the customer's needs and require-
ments.

To anticipate and foster change in the requirements
of its business customers, a public utility should have
insight concerning:

1. The needs of the individual customer in each

buying situation.

2. The roles of the major buying influences.

3. The organizational procedures in the procurement

process.

18

4. The environment in which the customer Operates.l

Only by classifying and identifying as many buying influences
and determinants as possible can the firm adequately

(l) predict future requirements, (2) allocate marketing

and production expenses and facilities Optimally, and

(3) develop products and strategies consistent with market

requirements.

Problem Statement

 

A public utility as contrasted with a firm operating
in a more competitive industry, has a unique problem when
dealing with the commercial segment of its market. When-
ever a new firm enters the utility's geographical market
area, or an existing customer modifies its market position
through an acquisition, product-line expansion, extension
of sales to a new area, or by building a new plant, the
public utility has the problem of estimating the probable
effect of this change on existing equipment and/or service
demand. In other words, the utility must be ready and able
to meet and service all new demand that might develop.
Although a similar problem exists for some non-regulated

firms, it is especially critical for a utility that has a

 

1RObinson, pp, cit., p. 214.

l9

monopoly or near-monOpoly position in a given geographical
area. For in such an instance, it is known that the new
or enlarged firms have to purchase any needed products

and services from the local utility.

There are typically three questions to be answered
prior to modeling or predicting the demand for a given
product or service:

1. Does the firm exhibit a need for the basic

product or service under study?

2. In what quantity or amount is the firm expected

to demand the product?

3. What brand will the firm buy or from what

company will the firm purchase the product?

Since the public utility affords an environment
where such variables as competitive factors, channel con-
siderations and brand choice problems are negligible, the
last of the above three problem areas will not be considered.
The difference between a firm's actual demand for a specific
product offering and its needs or potential demand for that
product is of primary importance.

The public utility, therefore, would like to predict,

as accurately as possible, the types and amount of service

20

and/Tn: equipment that new and existing companies might

require.

1.

There are four basic reasons for this:

To insure that any additional needs required

in the near future can be satisfied from
existing equipment.

To provide an accurate prediction of future
plant or equipment needs if existing facilities
are not sufficient.

To provide its salesmen with any information
about the buying firm which might aid in the
allocation, selling and installation of the
"right equipment and/or services at the right
rates."

To aid and guide management in the development
of new products and services and in the establish-

ment of equitable rate and price structures.

If forecasts and strategies compatible with the

above four areas are to be designed, two basic problems

Immt.be solved. First, the types of information needed

tomake accurate decisions in these four areas must be

(kmermined and monitored. Second, some method must then

be developedWhere‘by this information can be used to

21

generate forecasts of the buying firm's expected usage

and purchase behavior.

General Hypotheses

This research can be considered an exploratory or

pilot study in the general area of modeling demand in the

industrial sector of a public utility's market. The five

hypotheses used to guide the research are:

l.

 

The influence that certain structural variables
associated with given types and sizes of business
firms exert on the demand for telephone communi-
cations equipment, services and usage can be
identified and measured.

Top management in a firm will have varying
attitudes and biases toward the value of
telephone communications. The effect that

these personal values and attitudes have on
individual firm demand for telephone communica-
tions equipment, services and usage can be
identified and measured.

The effect of structural variables on firm demand
for telephone communications offerings will vary

from one sector of the industrial market to another.

22

4. The predictive power of the demand model using
structural factors as independent variables will
be significantly greater than that expected
solely by chance.

5. The predictive power of the model will be
increased significantly if attitudinal factors

are added to the array of structural independent

variables.

Method of Investigation1

To make the modeling, hypothesis testing, and
associated experimentation as realistic and accurate as
possible, a regulated industry and firm that would c00perate
in furnishing accurate data concerning itself and its
commercial customers was chosen. The telephone industry
and, in particular, Michigan Bell Telephone Company, adequately
fulfilled this need. Michigan Bell offered customers a
large variety of both hardware and service offerings. Also,
its market included nearly all types and sizes of business
emtablishments. Finally, with approximately 51%.of Michigan

Be11's total revenue derived from the business sector,2 there

 

 

. 1For a more complete statement of the methodology used
1n this study see Chapter III.

2 . .
1967 Michigan Bell Telephone Company records.

23

was attzlearly stated need for the types of information
outlined.and sought in this research.

To make the study operational and to facilitate data
collection, the scope of the study was limited in three
ways. First, the research was limited to firms in a
single geographical area. This eliminated the problem of
gathering data from a large number of widely dispersed
Michigan Bell business offices. It also minimized the
effects of factors that varied according to a firm's
particular location in the state.

Second, the firms included in the study were selected
from five segments of the business sector. These five
sectors were:

1. Dry cleaning establishments

2. Real estate establishments

3. Retail firms

4. Wholesale firms

5. Manufacturing firms
These segments included firms ranging from very homogeneous
QTOUPings (based on the structural variables used in the
anaIYSis) to extremely heterogeneous categories. The

forecasting models were expected to be more accurate for

24

the homogeneous segments than for those segments that
included firms with widely varying structural Character—
istics.

Third, those few firms large enough to merit a

permanent full time telephone sales representative were

dropped from the study. These firms typically have extremely

complex communication needs, and the development of a
generalizable model for firms of this nature would be

nearly impossible.

A non-random stratified sample was selected from

several different sources, including the Dunn and Bradstreet

Reference Book, the Yellow Pages of the Greater Lansing

telephone directory, the Directory of Michigan Manufacturers

and the Polk City Directory for Lansing.

A mail questionnaire was constructed for each of the

five business segments previously listed. Data was sought

concerning both the structural and attitudinal character-

istics of the selected firms. These characteristics con-

stitute the major independent variables used in the pre~
dictive demand models.
The primary demand variables investigated in this

study include both intrastate and interstate toll service,

25

Yelemn Page Advertising usage, total equipment demand,
trunk demand, and the number of telephone locations
demanded.l Data concerning the types of equipment,
service, and usage patterns of each firm included in the
sample were gathered from Michigan Bell records and serve

as the dependent variables.

Limitations of the Study

Since this study is empirical in nature, there are
certain limitations associated with the collection and
analysis of the data. These general limitations are as
follows:

1. While an attempt was made to include all factors
associated with a given business segment that
might affect the demand for communications
offerings, it is very likely that some have
been omitted. The exclusion of certain variables
from the analysis may influence the predictive
accuracy of the demand models. It is hoped,

however, that this effect will not be significant.

For a more detailed discussion of the explanatory
and demand variables used in this study, see Chapter III.

26

2. The usual limitations associated with the use
of a mail questionnaire survey apply here.
These are discussed in more detail in the
analysis chapter.

3. Since ordinal measurements were used in the
attitudinal section of the questionnaire, the
conclusions that can be drawn from this data
are limited and subjective in nature.

4. The respondents studied represent certain types
of firms in a confined geographical area. There-
fore, any findings can only be related directly
to a limited segment of the total market.

While extrapolating the results to firms in
other segments seems fairly logical in this
instance, the statistical accuracy of such

generalizations must be verified by further study.

lErms and Definitions1

Certain terms associated with the products, services

aDdoperations of the telephone utility require Operational

 

 

1Definitions of those terms denoted by an asterisk (*)
have been taken from a pamphlet entitled Glossary of Tele-
EflQflsfig published by the Michigan Bell Telephone Company.

27

defdxrttions. Following is a list of various terms used in

this study, along with their corresponding definitions:
Telephone Communications Equipment - Telephone
communications equipment refers to the hardware that
a customer rents from Michigan Bell. The specific
types of equipment referred to in this study are:

*Call Director - The call director is a single,
compact telephone unit which provides the
capacity of up to 30 telephone lines.

*Key Set - A key set is another name for a push
button telephone wherein the buttons are used
for intercom, holding, signaling and/or pick-up
of additional telephone lines.

*PBX (Private Branch Exchange) - A switching system,
often on a telephone customer's premises, which
serves that customer's telephones over a common
group of lines from the central office. The
PBX can receive incoming calls, place outgoing
calls as well as inter-connect office extensions.
Many sizes of switchboards are available.

Eglephone Communications Services - Telephone communica-
tions services refer to the offerings-—other than

hardware--which Michigan Bell offers to its customers.

28

The specific type of services referred to in this
study are:

*Private Line — A point to point arrangement
which has no connection with the exchange or
toll system of the telephone company.

*Trunk Line - A channel from a telephone company
central office to a customer's private branch
exchange used to provide service to the PBX
central equipment.

*WATS (Wide Area Telephone Service) - A special
"access" line to a nationwide network over
which a customer can make as many calls as he
likes (within a selected wide area which may
be nationwide) for a fixed monthly charge.

*Communications Consultant - A communications
consultant is an individual who evaluates the
data transmission needs of a business customer
and recommends and specifies the types of equip-
ment and services the firm would use to best
satisfy these requirements.

*Exchange - An exchange is a defined area, served

by one or more central offices, within which the

 

29

telephone company furnishes telephone service
at the exchange rates, and under regulations
applicable in that area as prescribed in the
company's Filed Tariffs.

*Local Service Area — A local service area is
the area, consisting of one or more exchanges,
within which calls may be made without a toll
charge, in accordance with the class of service
offerings.

*Long Distance Call - Any call beyond the local
or multi-message unit calling area.

*Station - An installed telephone or teletype-
writer on a customer's premises which, in the
exchange network, has been given a telephone
number and is the instrument through which

service is furnished to the customer.

Contributions

It has been stated that one of the directions which
marketing theory should take is the development of better
analytical tools of analysis.1 That this is necessary is

illustrated by a statement taken from a recent Marketing

 

1E. T. Grether, "A Theoretical Approach to the Analysis
of Marketing," Theory in Marketing, Reavis Cox and

30

Science Institute publication:
The development of an effective and efficient
approach to the matching of buyer requirements
and seller capacities is dependent on the
marketer's understanding, and consequently his
ability to classify or structure the decision making
process underlying industrial buying behavior. The
basic objective is to try to develop and explain an
operationally-useful classification system, focusing
directly on the nature of individual customers, or
potential customer's requirements and the resulting
buying situations.1
The development of a classification and forecasting scheme
for a single utility should lead to a better understanding
of the factors that influence the purchase of telephone
communications products and services. By structuring models
that relate various Classificatory and attitudinal factors
to purchase behavior, the public utility should be able
to develop new and better marketing strategies. Also,
the models and associated experimentation focus directly
on the nature of different customer requirements by examining
the factors that influence demand across various market

segments. This information should enable the utility to

be more effective when selling, advertising, pricing, and

 

Wroe Alderson, Ed. (Homewood, Illinois: Richard D. Irwin,
1950). P. 114.

1Robinson, pp, cit., p. XII.

 

 

31

developing products for specific types of industrial
buyers.

The research results, although limited in scope,
should help determine the feasibility of using this approach
in the development of either long-range micro or macro—
models of industry demand. The ability to extrapolate
from a study similar to this to a larger scale model will
depend upon both the importance of certain variables in
the demand models and the general accessibility of data
pertaining to these variables. If this data is readily
accessible, and reasonable estimates can be made about the
movement of these factors through time, the long range
planning function of a firm should be improved substantially.

The results of this study will also lend empirical
evidence to researchers developing similar models in other
industries. Because of an increased number of influencing
factors, constructing empirical forecasting models for
firms operating in a competitive environment would be more
complex than those developed in this study. The potential
benefits of such models to industrial firms, however, should
stimulate research efforts in this direction.

Finally, the research should indicate how salesmen

might best be assigned to specific customers. With a variety

32

of product offerings, a conglomerate grouping of firms,

and a range of customer sizes, some system must be designed

whereby salesmen can be allocated to customers according
to the varying needs and requirements of the purchasing
firm. This study provides the supplying firm with

information on the individual business customer and his

perceived needs. Therefore, valuable insight should be

gained as to how the selling function and variations in

product offerings might be adjusted to best match buyer

requirements.

Overview

In Chapter II, a discussion is presented concerning
tflle previous research conducted in the areas of modeling

irniustrial markets and forecasting demand in the business

Seetor. Chapter III contains a detailed review of the

reSearch methodology including sample selection, question-

naiilte construction and administration, and the statistical

techlhiques used for analysis. Chapter IV considers the

total array of independent and dependent variables used
Mn t1'Zleanalysis and delineates those explanatory factors
w .

‘h1‘211- appear to be the best predictors of demand. In

C
hapter V, the basic forecasting models are developed, and

33

a discussion of the models' application to the problem of
salesman allocation is presented. Finally, Chapter VI
contains a review of the significant findings and a summary

of the conclusions and further research which the findings

suggest.

CHAPTER II

MODELING DEMAND IN THE INDUSTRIAL MARKET SECTOR
A CONCEPTUAL FRAMEWORK

Introduction

Much has been written pertinent to the general topic

of marketing in the industrial sector. Most of the

writings, however, focus on concepts related to industrial

firms operating in a competitive environment. This

chapter will review those writings and methodologies which
luave a significant bearing on the specific problem of

iiDrecasting demand for commercial users of telephone

Lrtility offerings.

Classification Methods in Industrial Markets

 

Before a firm that sells products or services in the

1ndllstrial market can develop realistic and accurate

marketing strategies, management in that firm must segment

It‘s Eictual and potential customers. The characteristics

u o O I I I a

Seed: in categorizing customers is of major importance.

Of . . .
tea1'1, however, these characteristics are not eaSily

i
SCHLEited due to the large number of complex variables

34

35

associated with different types of business customers.
The importance of developing and utilizing Classificatory
schemes, however, is indicated by the entymologist Sokal
Who stated:

Classification is one of the fundamental concerns

of science. Facts and objects must be arranged

in an orderly fashion before their unifying
principles can be discovered and used as a basis

for prediction. Many phenomena occur in such variety
and profusion that unless some system is created
among them, they would be unlikely to provide any
useful information.1

Developing proper classification schemes to aid in the
decision making process has long been a difficult problem
facing the management of both industrial and consumer
kxased firms. As Frank and Green state:

Marketing managers and researchers often comment

on their difficulty in developing useful ways

of classifying customers for formulating marketing
policy. The source of the difficulty frequently

stems from the abundance of alternative classifica-
tion methods rather than from a lack of possibilities.2

It: i.s important that management choose those classification

S(”denies which best meet the requirements of the specific
\

1R. R. Sokal, "Numerical Taxonomy,“ Scientific

 

 

AEEEELégpgp, 215 (December, 1966), p. 67.

in 2Ronald E. Frank and Paul E. Green, "Numerical Taxonomy
.Iketing'Analysis: A Review Article," Journal of

Marke

‘**—~..;§ing_Research, Vol. V (February, 1968), p. 83.

36

problem under consideration.

Traditionally, firms selling or buying industrial

goods have relied on several basic classification schemes.

The three most popular bases for classification are:

 

"market segments or classes of trade reached; products or

services marketed; and end-use applications. Classification

according to classes of trade characteristics helps in

. according

analyzing the company or division buyers .
to products or services marketed, and according to end-use

application in developing technical specifications and

Preparing promotional materials."1

(Zlassification by Product

The most basic classification system considers the

tIHPes of products and services sold and bought by

ifhfllistrial firms. The five groups into which industrial

goods are usually divided are (l) installations or major

equipment; (2) accessory equipment; (3) operating supplies:

(4) :fabricating materials and components; (5) Primary

matuelrials.2 The majority of products and services offered

LL~___‘_¥
1Patrick J. Robinson, pp gl., pp, cit., p. XI.

 

the 2For a more detailed description of these goods and
13laying motives associated with each, see Melvin T. Copeland,

 

5‘

I)‘

37

by public utilities to business customers are either

operating supplies or accessory equipment. More specifi-

cally, the goods and services supplied to business firms
by the telephone utilities can be classified as accessory
equipment.

An analysis of the industrial buyers of accessory
equipment must include a cross—section of many types of
business establishments. This is especially true for
telephone equipment and services since nearly all industrial
firms purchase these products. The major problem in
analyzing these buyers is to determine the basic underlying

factors that cause variations in individual firm purchase

and usage behavior .

Segmentation by Class of Trade

To make accurate and timely marketing decisions,

management must also predict the collective behavior of

market participants.1 In the industrial market, these

\

 

iggiéflfigiples of Merchandising (Chicago: A. W. Shaw Company,
S 24) , pp. 130-154, 190-215; and Ralph 8. Alexander, James
° cIross, and Ross M. Cunningham, Industrial Marketing

(Ek3nnfawood, Illinois: Richard D. Irwin, Inc., 1961), pp. 6-13.

1Paul E. Green and Donald S. Tull, Research for

MEg-ZhiéipingpDecisions (Englewood Cliffs, N.J.: Prentice-Hall,

I
“c- . 1966), p. 120.

38

market participants are different types of business firms.
Before the collective behavior of these firms can be
modeled or predicted, some system must be found to group
these customers into homogeneous segments.

The most frequent starting point in classifying
business firms or institutions is the Standard Industrial
Classification system. The federal government designed
the SIC code to categorize all types of business establish-
ments according to either the product produced or the
operation performed. This classification system is most
useful when it "can be viewed in the light of the buying
motives of industrial purchasing agents. . . If the problem
Of all industrial plants could be categorized into those
that can or cannot be solved through the use of certain

EIrOducts, it could be determined what plants are in a
CEHTtain market and the exact problems each one faces."l
Classification by SIC assumes that the problems faced
b3' tihe industrial concern originate primarily with the
product manufactured and/or the manufacturing process used.

For Some industrial goods, this is a valid premise. For
\_.

Ind! lFrancis E- Hummel, "Pinpointing Prospects for
us trial Sales," Journal of Marketing (July, 1960), p. 27.

39

others it is not. For example, when considering the
accessory equipment marketed by a telephone utility, the
traditional product or type of firm-oriented classification
schemes may be too general. Therefore, this approach is
often partially or entirely inadequate. A number of other
factors associated with the buying firm, his customers,
and suppliers may dominate the variables associated with
the product produced or the operation performed. Thus,
businesses classified under the same SIC code might manu-
facture or sell the same basic kinds of products and have
similar production problems and techniques. The degree
of dissimilarity in other phases of their operations,
however, could rule out the usefulness of the SIC system
as the basic means of forecasting product or service
requirements and demand.

Although customer behavior is multidimensional in
thert it generally is influenced by a large number of
Variables, "researchers often sidestep its complexity by
picIking some unidimensional attribute assumed to be an
incii-Cator of the more complex phenomena to be understood."1

Becnalase of its general availability, simple employment data
_LLL_____,

1Ronald E. Frank and Paul E. Green, pp. cit., pp. 83-84.

 

40

is used by a great many sellers of industrial products to
supplement the SIC system and provide additional information
about the size of a particular market or customer. This

one factor falls far short of being a good indicator of
demand for most products. For this reason, many firms

also supplement SIC data with value added, production, or

wages and salary statistics.1

In addition, variations in
the purchasing habits and marketing strategies of different
firms, as well as varying end-use characteristics of certain
products, have forced some industrial good manufacturers
to develop even more detailed classification systems.

The behavioral factors influencing the purchasing
behavior of individual firms have also been used to
categorize buyers. This has led to classification schemes
Inased on a variety of buying situations and purchase

ITIOtives,2 ”since for any given buyers purchasing the same

EIrCXiuct, individual purchasing patterns and the buying

PrOCess itself have been found to differ distinctly with

Subsequent purchases over time and in dealing with different
\_

1Robert J. Piersol, "Accuracy of Estimating Markets

J .ndustrial Products by Size of Consuming Industries,"
‘EfliEEEEEJ of Marketing Research, Vol. V (May, 1968), p. 147.

2

for: JI

Patrick J. Robinson, pp, cit., p. 22.

 

41

suppliers."1 The basis for these latter classification
schemes usually rests on the assumption that there are

2 The

several suppliers competing for the buyer's loyalty.
majority of the variables used in these classification
models, however, are not relevant for the public utility.
In the utility case, the problem is not one of classifying
customers on the basis of their likelihood of purchase
from a particular seller, but, rather to determine whether
they will purchase a given product at all. One method of
answering this question is to determine either directly

or indirectly each purchaser's end-use characteristics

for different products. Estimates are then made of the

buyer's specific needs and demand based on these varying

end-use situations.

lIbid., pp. 22-38.

2For example, a type of classification scheme often
used to categorize the industrial customers is one which
attempts to classify buyers according to the different
fa<3tnbrs that influence their decision to purchase a given
pxn3<3l1ct from one supplier rather than a competitor. This
t339€3 of classification system often forms the basis for

prediction models comparable to the consumer brand-switching

42

Classification by End-Use

The ability to expand sales to both existing and

potential customers is severely limited when specific

end-use information is not available. For this reason,

many firms marketing capital goods perform research to

discriminate among the end-use characteristics of their

products for different customers. The strength of the

end-use method of classification rests in the fact that

it "considers each customer as a heterogeneous unit by

focusing on his numerous requirements. This is essentially

the 'consumer viewpoint' which is said to be the heart of
the 'marketing concept'; the ultimate user's mode of
consumption is implicit in this scheme."1 This method of
analysis allows the salesman to determine the product, or
LIrOduct variation, which a particular firm should purchase

based on the specific characteristics associated with its

In general, formal classification schemes have not

been developed that consider the specific purposes for

“fililel industrial products are purchased. Instead, the motives

\

th 1Charles L. Hinkle, Unpublished paper presented to

Ju: American Marketing Association International Congress,
63 -17—l9, 1968 at Philadelphia, Pennsylvania.

 

43

associated with the purchase of a product, and the end-use

to which it is put are inferred or implied from the basic

type or class of firm purchasing the product. It is often

difficult, however, to locate and identify firms with

similar product applications simply on the basis of class

of trade. For in this case, the classification scheme
may not recognize that a single customer uses a product in
more than one way or that customers of the same type or

size may use fundamentally different products.
Classifying a public utility's commercial customers

by end-use should lead to a better match between customer

needs and the products and services provided. This is

particularly true for the telephone industry in which the

tYpes and amount of equipment and service purchased is

<3ependent upon a number of end-use factors. For example,

the degree to which products are used to communicate with

SPQQPliers, customers, branch facilities, or other depart—

Imerrts varies among customers. Also, firms differ in the

l"‘11.:><2>.:I:'tance they place on not having a customer receive a

k"45i)' signal.1 Finally, the use of telephone equipment for

\

Of_ . 1For example, an airline ticket agency or central taxi

ha f1} ce would be expected to be much more concerned with

ma"J-rlg'an open line available at all times than the typical
nufiacturer. This is true since the probability of a lost

 

44

computer hook-ups or data transmission varies substantially
among customers.

The relative importance of these factors can vary
both among similar firms and across industry classes. Thus,
to match accurately product and service offerings with
customer needs, the telephone salesman must determine the
variable influence which these factors have on different
industry classes. Because of the large number of customers
served by a public utility and the wide variance in
customer characteristics, this task is made extremely
difficult. Therefore, the traditional classification schemes
discussed previously may be of little value when used alone.
What is needed is a more generalized technique which is
capable of isolating the most important purchase indicators

-from.the total array of possible influencing variables.

liaszgprs Influencing the Demand for Industrial Goods

Before a predictive model can be developed to relate
'13631? characteristics to product demand, the researcher
Shol11d have some conception of those factors that are most

llkely to influence the purchase decision. Most research
~L_______»

s
tile resulting from a busy signal would be much higher in
e first two cases than in the latter.

 

45

has dealt with the consumer buying process; however, the

literature does contain a number of constructs relevant

to the problem of modeling industrial demand. The purpose

of the following paragraphs is to relate certain empirical

findings and theoretical concepts dealing with industrial

purchasing processes to the problem of modeling industrial

demand for utility products and services.

Traditionally, the industrial purchase process has

been viewed as a rational decision making procedure in

which the variables of quality, price, delivery time,

quantity, and service are most influential. The Industrial

Marketing Committee Review Board has stated that "Rational

buying motives appear to predominate in the industrial

field (as against emotional motives in the consumer field),

13nt their influence declines with the increase in product
Similarity."l Modeling buying behavior in the industrial
IDarket has long been dominated by the concepts and principles
stated in classical economic theory. "Yet there is "an
increasing recognition that industrial buyers are influenced

by emotional 'noneconomical' as well as 'rational—economical'
\

1Industrial Marketing Committee Review Board, "Funda-

m . . .
er113511 Difference Between Industrial and Consumer Marketing,"

J
.saalzraai_of Marketing, xxx (October, 1954), p. 153.

 

46

consideration.”l

Much of the recent industrial marketing literature
has focused on the relative importance of non-economic
factors in the industrial purchasing decision.2 Generally,
this research has dealt with the purchasing agent and the
various psychological and behavioral determinants associated
with his decision to buy from one supplier rather than
another. By gathering empirical evidence on industrial
buyers' purchasing habits, researchers have attempted to
quantify and model the relationships between certain
psychological and behavioral variables and their resultant
effect on buying decisions.3

This study is concerned primarily with a firm

Operating in a non-competitive environment. Brand choice

~¥

1Patrick J. Robinson, pp al., pp. cit., p. 153.

 

2For examples of this see F. R. Shoaf, "Here's Proof--
Tube Industrial Buyer's Human," Industrial Marketing, XLIV
(be, 1959), pp. 126-128: H. Lazo, "Emotional Aspects of
IEndustrial Buying," in R. S. Hancock, ed. Dynamic Marketing
329; a Changing WOrld: American Marketing Association Pro-
Siggdings of the 43rd National Conference: 1960, pp. 260-261;
Eind D. J. Duncan, "Some Basic Determinants of Behavior in
IDHdustrial Purchasing," Pacific Purchasor, XLVII (May, 1965),
pp. 17-18.

 

 

3For a discussion of some of the psychological and
ibehavioral variables that influence the industrial buyer's
Phrchase decision see Patrick J. Robinson, pp_pl,, pp, pip.,
PP. 155-571.

47

and snapplier preference considerations are not factors

in tine purchase decision. The only types of variables to

cxxusider in the monopoly situation are those influencing

the basic "buy-no-buy" decision.l There may be some

differences in the psychological and behavioral character-

istics of various purchasers or users. These differences

may cause variations in the amount of a product or service

demanded by structurally similar firms. However, for

firms in a monopoly or near-monopoly position, non-behavioral

factors should be accurate indicators of customer demand

patterns.

 

lThe "buy-no-buy" decision as used here infers that
the customer has a choice between two or more alternatives.
This means that the customer is aware of the offering and
that the product or service is actually available to him.

If information on new offerings is not disseminated to
potential customers or if certain new products are only
available to buyers in specific industries or geographical
locations, there exists an unserved market. These customers
do not actually face the "buy-no-buy" alternative and the
Predictive model is trivial.

Although the availability problem does not exist in
this study, the amount of information which various firms
haVe concerning telephone products and services may well
haVe an impact on certain demand offerings. In fact, as
later discussions will show, many firms included in this
analysis were either totally unaware or had only a slight
familiarity with certain telephone analysis. For a more
detailed discussion of this problem and the specific
Pr0ducts and services involved, see Chapter III and IV.

48

Building an industrial market forecasting model
witli structural variables requires a thorough listing of
ald.<§haracteristics that might influence a buyer's
demand behavior. Careful analysis of these factors is
necessary prior to including them in the model since:

The analysis of a market defined in this way can be

no more valid than the set of characteristics

(market factors) on which it is based. If the

market factors are really not the primary causes

of demand or if some primary factors are omitted,

the resulting market potential will be imprecise.

On the other hand, a complete list of market

factors would be impossible. Hence, it is necessary

to make judgments regarding those which are to be

included and those which are to be excluded.
In some cases, these "judgments" are difficult to make.
This is true when the buyer and seller perceive different
factors as having the greatest influence on demand. It is
also true when the seller has a large number of different
types and sizes of customers whose basic needs vary both
Within and across the traditional business segments. Even
though certain of these segments possess seemingly

hOmOgeneous characteristics, they may actually differ

greatly in their use of the product.

1Mark E. Stern, Marketing Planning: A Systems Approach
(NEW York: McGraw-Hill Book Company, 1966), p. 22.

49
Tecturiques Used to Model Demand in the
Industrial Market
Forecasts of both expected and potential demand
usually'are developed through one of the following three
procedures:

1. The selling firm relates various economic
indicators to the sale of a product and,
ignoring individual firm demand, develops
macro estimates of total potential demand.

2. The selling firm segments the total market by
geography or class of firm and based on these
classifications:

a. "identifies specific end—use situations
for the product,

b. identifies the major market segments in
which these end-use situations will
exist, and

c. estimates the potential market for the
product within each segment. (The
total market potential is simply the
sum of these.)"

3. The selling firm analyzes customers on an

individual basis and determines the factors

associated with each company that affect its

1National Industrial Conference Board, pp. cit., p. 70.

 

50

product usage. These findings are then
extrapolated to either homogeneous segments
or total markets to obtain values of total
market potential.

Which of these three methods is used depends upon
the accessibility of information and the purpose of the
forecast. If the goal is to develop strategies on a
national or regional basis, then one of the first two
methods or some modification of them is most appropriate.
However, if the purpose is to define and develop policies
on an individual customer basis, as in this research, then
some version of the third method is most beneficial.
Several techniques that are used to analyze and forecast
the demand behavior of individual firms are reviewed in

the following paragraphs.

Sales Force Estimates

Probably the most common method of estimating demand
for a particular firm is through salesmen's projections.
If a salesman is familiar with a particular firm, he usually
can provide reliable demand information about the firm.
This information should include data on both potential

demand and that portion of the total which his company can

51

expect to obtain (expected demand). The biases and
disadvantages inherent in this method are well known,
particularly where estimates are used as a basis for
establishing sales quotas.l Despite these disadvantages,
the salesman is in close contact with the market, and his
knowledge of customers can at least serve as a starting
point for more formalized estimating techniques. They can
also provide readily accessible checks on figures derived

by some other estimation method.

Regression/Correlation Analysis

Another technique used to estimate demand on the
firm level is regression analysis. Here certain quanti-
fiable factors associated with the firm are used to generate
statistical estimates of demand. These factors and their
relative influence are determined through sampling a number
of customers similar to those for which demand estimates
are to be made. By determining the quantitative inter-
relationships Of these factors and their resulting impact
on demand, regression equations are derived and used to

obtain forecasts of the demand variable. In essence, however,

 

1National Industrial Conference Board, Forecasting
Sales (New York: NICB, 1964), p. 21.

 

52

the value of demand arrived at by such a technique is an
average, or conditional mean value. As such, they may not
be reliable estimates of potential demand.

When describing how math models might be used in the
public utilities, Robert Williams suggests a technique
similar to this for establishing market potentials for
business customers:

A telephone company makes a study of the communica-
tions equipment and service usage of trucking firms
in the trucking-firm panel and determines the
average equipment and service for each of several
sizes of firms. (Perhaps the number of trucks
operated is the measure of size.) The average
communication development of the top 10 per cent in
each size group is also determined. The company's
salesmen then start 'caterpillaring,' bringing the
below average firms in all districts (pointed out

by the Model) at least up to the average, and others
up to the top ten percenters. The salesman does not
have to tell the customer that the telephone company
has decided his firm's communications needs; he merely
shows him the standards the trucking industry itself
has set.

Of course, more than one independent variable should be
used in this type of analysis, provided the accuracy of
the demand forecast is improved sufficiently. Several

techniques are available to assist the researcher in

 

1Robert H. Williams, "A Simple Math Model for Public
Utilities," Public Utilities Fortnightl , Vol. LXIX, No. 11
(May 24, 1962).

53

determining which influencing factors contribute most
to "explaining" demand fluctuations.l Thus, factors that
contribute little or no information to the forecast can

be eliminated from the model.

Discriminant Analysis

Regression analysis is a technique used by firms
to estimate the amount of a given product a buyer will
purchase, given certain characteristics associated with
the buying firm. However, if the dependent—variable is
not continuous but multichotomous, as is the case when
trying to predict whether a customer will purchase machine
A, B or C or when attempting to distinguish one group of
customers from another, multiple regression analysis cannot
be used.2 In this type of situation, N-way multiple
discriminant analysis is used frequently.

Two ways that discriminant analysis can serve the

marketer are:

 

1For an excellent discussion of the methods used to
select the "best" regression equation, see Norman R. Draper
and Harry Smith, Jr., Applied Regression Analypis (New York:
John Wiley & Sons, Inc., 1966), pp. 163-195.

Regression analysis can be used in the dichotomous
situation where only two groups exist.

54

1. as a basis for building models aimed at
predicting customer behavior.

2. as a determiner of the relative importance of
various customer characteristics as they con-
tribute to the prediction.1

A search of the literature revealed no direct appli-

cation of discriminant analysis to predicting customer
behavior in the industrial market. The literature, however,
does indicate many uses of this technique in the consumer
area. For example, a number of articles have been published
in recent years using dfis technique to predict and discrimi—
nate between the audiences of different FM stations,2 the
purchasers of major brands of automobiles,3 the households

most likely to adopt new products,4 and the individuals most

 

1Ronald E. Frank, William F. Massy, and Donald G.
Morrison, "Bias in Multiple Discriminant Analysis," Journal
of Marketing Research, Vol. II (August, 1965), p. 250.

2William F. Massy, "Discriminant Analysis of Audience
Characteristics," Journal of Advertising Research, Vol. V,
No. 1 (March, 1965), pp. 39-48.

3Franklin B. Evans, "Psychological and Objective Factors
in the Prediction of Brand Choice: Ford Versus Chevrolet,"
Journal of Business, XXXII, No. 4 (October, 1959), pp. 340-369.

4Ronald E. Frank, William F. Massy, and Donald G.
Morrison, "The Determinants of Innovative Behavior with
Respect to a Branded Frequently Purchased Food Product,"
Proceedings of the Winter Meetingyof the American Marketing
Association, New York, December, 1964, pp.312-323.

55

likelqr‘to use thrift deposits in savings and loan establish-
ments or commercial banks.1 Similar applications should
be possible for discriminating among the users of different

types of telephone communication products and services.

Probabilistic Models

It is helpful for the selling firm to have estimates
of a buyer's potential as well as actual demand. It would
be of much greater strategic value, however, if information
were also available on the likelihood of each customer
increasing his demand by some specified amount. If values
were derived to indicate the probabilities of different
buyers demanding additional products or services, customers
could then be given priority according to their demand
potential. In this way, selling expenses and effort could
be allocated in a more optimal manner.

The probabilistic approach provides a useful con—
ceptual framework for considering the expected purchase
behavior of customers--whether they be consumers or business

firms. In the consumer area, the use of probabilistic

 

1Henry Claycamp, "Characteristics of Thrift Deposit
Owners," Journal of Marketing Research, II, No. 2 (May,
1965). PP. 163-170.

56

models to reflect buyer behavior has been increasing
rapidly in recent years. This is indicated by the number
of articles published in various professional journals
idealing with such subjects as brand switching,l Bayesian
analysis,2 decision-tree analysis,3 and learning process.4
Probabilistic models should be of equal value in the
industrial market, yet the number of articles concerning
applications in this area remain quite scant.

Many of the demand models discussed in the literature
involve the use of conditional probability. Where the
amount of a particular product an individual or firm
purchases is dependent upon or affected by some other

factors, demand is viewed as ”conditional" on the occurrence

 

1Alfred A. Huehn, ”Consumer Brand Choice - A Learning
Process?" in Quantitative Techniqpes in Marketing Analysis,
Ronald E. Frank pp 31., eds. (Homewood, Illinois: Richard
D. Irwin, Inc., 1962), pp. 390-403.

2For a discussion of Bayesian analysis as it applies

to marketing problems see Ronald E. Frank and Paul E. Green,
Qpantitative Methods in Marketing_(Englewood Cliffs:
Prentice-Hall, Inc., 1967), pp. 6-30.

3John F. Magee, "Decision Trees for Decision Making,"
Harvard Business Review, XLII (July-August, 1964), pp. 126-38.

4Emir H. Shuford, "Some Bayesian Learning Processes,"
Human Judgments and Optimality, Maynard W. Shelly and Glenn

L. Bryan, eds. (New York: John Wiley & Sons, Inc., 1964),
pp. 127-152.

57

of these other events. In these cases, conditional
prObability models can be develOped. The notation p(E/F)
is used to indicate the probability of the event E given
the event F exists or has occurred.

This concept can be used to develop probabilistic
models of industrial demand for telephone communications
offerings. Using the notation given above, E would repre—
sent the probability of a firm demanding X units of a
certain product while F represents the value of some
characteristic associated with the firm, such as a specific
number of employees. This analysis can be extended to
include more than one conditional variable, in which case
multistage probability models can be developed.1

If a utility has information about particular business
customers and their respective buying characteristics, the
necessary parameters can be derived and probabilistic
estimates of demand generated. The data can be used to
guide management in its decision making procedures by

providing information concerning the likelihood of certain

 

1For a more detailed discussion of conditional
probability models see G. Hadley, Introduction to Probability
and Statistical Decision Theory (Cambridge: Holden-Day, Inc.,
1967. PP. 282-296.

58

industrial customers increasing their purchases of specific
products and services. Also, these static models of
individual firm buyer behavior can be used as building
blocks in constructing dynamic aggregative models of

industrial purchasing behavior.

Summary

This chapter has described some of the concepts and
methodologies that are useful for modeling individual firm
demand in the industrial market. The material concentrated
primarily on three areas - (l) the classification methods
available to strategically segment industrial buyers,
(2) the types of behavioral and non-behavioral variables
that must be considered in modeling firm demand, and
(3) the basic techniques that have been used to structure
the interrelationships between buyer characteristics and
purchasing behavior. Most of the literature referred to
in this chapter dealt with industrial firms operating in
a competitive environment. However, the general approaches
and fundamental concepts developed provide a framework for

structuring an empirical demand model of the business sector

of a public utility.

CHAPTER III

RESEARCH DESIGN AND METHODOLOGY

Questionnaires were constructed for each of the
five business segments listed in Chapter I. The purpose
of the questionnaire was to obtain insight and data on as
large a number of the factors hypothesized to influence
demand as possible, so that accurate models relating
individual firm characteristics to demand could be developed.
The structuring of several questionnaires was necessary
because each of the five segments is somewhat unique in
operating characteristics and basic communication needs.

A mail survey was used to obtain data on the independent
variables associated with each firm. Because each
respondent completed only one questionnaire and the data
were collected during a relatively short time period, the
research was cross-sectional in nature. The inherent
limitations associated with cross—sectional analysis and

mail surveys are well known.1 However, possible alternative

 

1For a discussion of the limitations of cross-sectional
analysis and mail surveys, see Mildred Parten, Surveys, Polls,
and Samples (New York: Harper & Brothers, 1950), pp. 391-402.

59

 

60

procedures did not prove feasible. For example, con-
sideration was given to using a longitudinal study to
examine the variation of certain independent variables
through time and their effects on firm demand. Time and
money constraints prevented such an effort. Similar
constraints prevented the use of in-depth personal inter—
views as a method of obtaining the data, although this
procedure may have provided more insight on demand variations
than did the mail survey.

Data regarding the dependent variables indicating
the different types of demand exhibited by each firm were
obtained from Michigan Bell Telephone records. Since this
information was considered confidential by Michigan Bell,
written permission from a major officer in each firm was
required before the desired data were released. This require-
ment undoubtedly created some hesitancy on the part of
certain individual respondents and lowered the final

response rate.

Questionnaire Construction

 

Two types of information were required from each
respondent. First, data were sought on those factors

characterizing firms in the five business segments that

61

‘were hypothesized to affect telephone usage behavior.
Second, information regarding the respondent's attitudes
toward telephone communication was needed. The question-
naire was divided into two parts, analagous to these two
categories.

The Objective of the first part of each questionnaire
was to obtain detailed information on certain specific
aspects of the responding firm and its activities. This
portion of the questionnaire was structured differently
for each of the five business segments, since each class
of business had unique characteristics. Interviews with
Michigan Bell personnel and a review of industrial survey
material were used to generate a list of variables relevant
to each segment. A general lack of knowledge of the
specific factors that affect customer demand behavior
for telephone offerings in the industrial market, however,
necessitated an exploratory type of study using a variety
of possible influencing variables. Statistical techniques
were then used to extract those variables which were most
important in explaining variations in the independent

factors from the larger set.1

 

1For a more detailed discussion of the statistical
procedures used, see pp. 142-152.

 

62

Certain independent variables, such as gross revenue
and number of employees, were common to all segments.
Other characteristics, such as the number of sales clerks
or number of production employees, were appropriate to
only one or two of the business categories. Table 3.1 lists
the variables included in the first part of the question—
naire for each business segment and indicates those
variables common to more than one segment. As Table 3.1
indicates, the types of telephone communications equipment
and services demanded and used were hypothesized to be
dependent not only on the characteristics of the buying
firm but on the location and composition of its customers
and suppliers as well. The measurements of most of the
variables listed in Table 3.1 conform to a ratio scale.
These factors will be referred to hereafter in the study as
structural or Classificatory variables.

In the original survey instrument, data were sought
concerning the present status of each Classificatory
variable as well as the expected status or value of these
factors five years hence. Pretesting the questionnaire,
however, Showed that respondents were either unwilling or

unable to estimate future factor values. This forced the

63

TABLE 3.1

CLASSIFICATORY FACTORS HYPOTHESIZED TO AFFECT

DEMAND-BY BUSINESS SEGMENT

 

 

Business
Variable Description Segment*
1. Number of years in operation in
present facility M,W,R,r,D

15.

16.
17.
l8.
19.
20.

210

22.

23.
24.

Number of buildings used in local
business Operation
Number of square feet of floor space

Number of separate offices
Gross Sales

Total number of employees
salaried employees

Number
Number
Number
Number

Number
Number
Number
Number
Number

Average number of orders taken/month

of
of
of
of

of
of
of
of
of

administrative employees
engineering employees
production employees

clerical employees
sales personnel
other personnel

buyers
brokers

Percent orders taken by phone
Percent orders taken by mail
Number of salesmen in field

Average number hours/months used

for data transmission

Number of additional plants or
facilities

Location of additional plants
or facilities
Location of main offices

SIC classification

Q
Q
Q

Q

32.3.2
2 2'; S
wwwsu
Hnbo

Q

Q
Q

U U

Q
Q
Q

U

Q
Q
Q

W.? W
U U H

Q

Q

Q

zzszz
sagas:

Q

Q
Q
Q
Q
U

2.? 2
w'? N
U U H

Q
Q

H w‘; 3 3

M,W
M,W
M,W
M,W

64

TABLE 3.1 (Continued)

 

 

Business

Variable Description Seqment*

Percent of revenue derived from:
Manufacturing activities
Wholesaling activities

25. MWR
MWR

Retailing activities M,W,R

MWR

MWR

26.
27.
28.
29.

Other activities
Number of suppliers

30.
31.

Location of suppliers M,W,R
Percent of supplies purchased by
location M
32. Number of customers M,
33. M

M

34.

Location of customers
Percent of revenue by location

350

Percent revenue by type of customer r,D
36. r D

Type of services offered

 

*M = Manufacturer, W = Wholesaler, R = Retailer,
r = Realtor, D = Dry Cleaner. Thus, the notion M,W,R
in this column following a variable would mean that
data concerning this variable was sought only from firms
in the manufacturer, wholesaler, and retailer segments.

65

deletion of this aspect of the survey.

The objective of the second part of each questionnaire
was to ascertain the importance of various types of
telephone services to the buyer, as perceived by the owner
or some other high ranking official of the purchasing
firm. This portion of the questionnaire was identical
for each of the five business segments, and permitted a
comparison of responses both within and across each of
the five categories.

The questions in Part II of the questionnaire were
divided into three groups. The purpose of the first
seven questions was to discern how important the manage-
ment of each firm perceived local, intra-state, and inter-
state telephone communications to be for their firm's
operations. Information was sought concerning the per-
ceived importance of each of these offerings for several
business activities. Since a measure of relative importance
was desired, the respondents were asked to rank the contri-
bution of telephone services to several activities using
the following criteria:

(a) the estimated telephone time spent on each

activity during an average month.

66

(b) the expected importance of telephone communica-
tions on various activities in the future.

(c) the percentage increase (decrease) in toll
usage for each activity that would be expected
if a decrease (increase) in rates occurred.

Information on these three criteria should provide
an accurate reading of the telephone usage habits of
individual firms. This information should also lead to
a better understanding of the communication activities
in each segment for which various types of telephone
offerings are perceived to be most vital.

The objective of question number eight in Part II
was similar to that of 1-7. However, here the respondent
was asked to compare his firm's usage of telephone
communications in certain areas with what he felt or knew
to be true of other firms of a similar type and size. By
comparing these responses to actual usage values,
generalizations could be made concerning the buyer's aware-
ness of competitor's communication behavior. It has been
inferred that a major selling point regarding telephone
services might be a reference to competition's heavier

usage patterns.1 This information, combined with an

 

1Robert H. Williams, pp, cit., pp. 730-732.

67

understanding of how important telephone communications
are to each user, should help the seller develop effective
personal selling strategies.

The remainder of the questions in Part II were con-
structed to determine the customer's awareness of the
telephone communication alternatives available to his firm.
Although this information was not sought on the pre—test
questionnaire, specific comments by pre-test respondents
indicated a basic unfamiliarity with certain types of
telephone equipment and services. This lead to the
inclusion of a general question dealing with customer
awareness in the final draft. Table 3.2 summarizes the
major independent variables included in Part II of each
questionnaire. These measures, primarily ordinal in nature,
relate basically to attitudes and/or opinions of manage—
ment and shall be referred to as attitudinal or behavioral
factors in this study.

Several questions were dropped from the survey
instrument following the pre-test. The most important
factors that forced the deletion of these questions were:

(a) a general feeling that the questionnaire was

too detailed and time consuming.

68

TABLE 3.2

ATTITUDINAL FACTORS HYPOTHESIZED TO AFFECT DEMAND

 

 

l - 6 Relative amount of time (perceived by user)
spent communicating by phone in the local
calling area:

with customers
with suppliers
with other departments
with salesmen
with branch facilities
for data transmission

mmbWNl-d

7 - 11 Relative amount of time (perceived by user) spent
communicating via long-distance phone:

with customers
with suppliers
with salesmen
with branch facilities
for data transmission

1
l

l—‘OKOCDQ

12 - 17 Relative importance of telephone usage as per-
ceived by the user in present and future
communication:

12 with customers
13 with suppliers
14 with salesmen
15 with other departments
16 with branch facilities
17 for data transmission

18 - 22 Expected percentage decrease in in-state
telephone communications:

18 with customers
19 with suppliers
20 with salesmen

69

TABLE 3.2 (Continued)

 

 

21 with branch facilities
22 for data transmission

that would result from a 20% increase in toll
rates.

23 - 27 Expected percentage decrease in out-state
telephone communication:

23 with customers
24 with suppliers
25 with salesmen
26 with branch facilities
27 for data transmission

that would result from a 20% increase in toll
rates.

28 - 35 Relative amount of:

28 local telephone service

29 toll telephone service

30 telephone equipment

31 yellow page advertising

32 data transmission

33 telephone usage to contact suppliers
34 telephone usage to contact customers
35 telephone utilization by salesmen

perceived as used by respondent's firm as com-
pared to other firms of similar type and size.

36 Perceived understanding of telephone offerings
available.
37 Methodology used when evaluating or adopting a

new service or piece of telephone equipment.

 

70

(b) both an inability and an unwillingness on
the part of the respondents to provide detailed
sales information--either by product line or
geographical area.

(c) an inability on the part of the respondent

to answer questions regarding specific types

of telephone equipment or services. For
example, it was originally planned that several
questions related to specific services such as
WATS (Wide Area Telephone Service) would be
included in the questionnaire. The pre-test
indicated that non—users were generally unaware
of the nature of the WATS service and that
respondents in the firms presently utilizing
WATS were not aware of the exact nature of its
use.

The questionnaires remained lengthy even after a
number of deletions; however, completeness of the returned
questionnaires and the total response-rate justified the
intensiveness of the survey instrument. Also, the amount
of effort required to obtain a comprehensive listing of

firms in each of the five segments and to obtain a sufficient

71

response rate from a mailed questionnaire did not, in

the researcher's opinion, justify a superficial treatment.

Dependent Variables

 

Table 3.3 lists the dependent variables used to
reflect individual firm demand for various kinds of equip-
ment, services, and usage. Figure 3.1 summarizes the
basic taxonomy of all variables examined.

Limitations were placed on both the amount and types
of information that could be gathered relevant to these
dependent factors. First, to ensure privacy, toll usage
data had to be gathered and analyzed on a summary basis
(total monthly in-state and out-state billings) rather
than on an individual call basis. Thus, no information
could be collected on geographic usage patterns other than
that reflected by summarized in- and out-state data
or inferred from information derived from the questionnaire.
Second, no information could be secured on the amount or
types of local telephone usage except that which could be
inferred from questionnaire data. This was due to the
fact that Michigan Bell does not regularly tabulate local
communication data. Third, ready access was only available

on two months toll billings. This meant that seasonal

DEPENDENT

72

TABLE 3.3

VARIABLES REFLECTING FIRM DEMAND
FOR TELEPHONE OFFERINGS

 

 

10.

11.

12.

l3.

14.

Average
Monthly
Monthly
Monthly
Monthly
Number
Utiliza
PBX
PBX
Number
Number
Number
Number

Total n

monthly toll billings.
Intra—state toll billings.
Inter-state toll billings.
billing for equipment usage.
billing for yellow page advertising.
of lines or trunks.
tion of a PBX (switchboard).
station capacity.
line capacity.
of call directors.
of Key-sets.
of extensions.
of stations.

umber of telephone locations.

 

73

variations in long-distance usage could not be incorporated
in the demand model. The demand for equipment and most
services, however, is not a function of the season or time
of year. Therefore, this limitation was not critical to

the development of demand models for these offerings.

Sample Selection

 

The original survey design called for a sampling
of all types of business firms throughout the state of
Michigan. Preliminary study indicated several problems
inherent in this approach.
A. Geographic Limitations
1. A portion of the State of Michigan is not
serviced by Michigan Bell Telephone Co.,
but by a number of other telephone utilities.
Since it was necessary to gather telephone
usage data from the particular utility
servicing the selected firms, the geo-
graphical area was first limited to that
serviced solely by Michigan Bell.
2. Early discussions with Michigan Bell personnel
indicated that both product demand and

product offerings vary according to geographic

 

 

74

 

.Oum

mEHHm
muflm paw momma umawfifim ou pwummfioo

mm mmcflnmmmo mDOHHm> mo momma pm>flmoumm

OHQMHHm>m

mw>flumcuwuam mo mcfipcmumumpcs ©o>flm0Hom

ceaumoflsoeeoo mcosmmamu mo

 

moanmflum> Hmuofl>mnom

.m

.N

OOGMUHOQEA mnouom paw “Cowman Um>fimoumm .H
L

 

 

 

ILA.

 

 

 

11

pwuflaauo

mova>umm mcosmmamu mo pasoEm paw momma
unmemwoom Ozonmmawu mo momm: HHOE
nowadays acme

Imagow mconmmamu mo wmmhu paw unsoE¢

 

 

.Oum

mumfiammom mo :oHumooq
muwwammom mo umnfisz

mHmEoumsO mo cOwumooq
mHmEoumdo mo woman can uwQEsz
Oahu an mommoaafim mo HOQEOZ

mwum Oflnmmumomm an mmHMm mo coaponwnumfla

mmamm mmouw
meQMAHm> SHOOMOAMHmmmHO

 

.h
.m
.m

HNMd‘

 

 

 

moanmflum> pampcmmma

 

mmaflmflum> ucmpcmmuch

Ammo: QZ¢SWQ OB mBDmZH UHm¢m

H.m MMDUHm

 

 

75

location. Since this variance added com-
plexity to the research design while not
being of particular relevance to the major
hypotheses under study, the decision was
made to confine the research to a small
geographic area.

3. Further discussions with Michigan Bell
personnel showed that any data required on
firm usage and demand was only available from
local business office files. To facilitate
the collection of this data, the Greater
Lansing Area was chosen as the region from
which the sample would be drawn. Collecting
the data was also extremely time consuming
and caused a partial disruption of business
office activities. The choice of the Greater
Lansing geographic area minimized both the
amount of time needed to gather the informa-
tion and the amount of disturbance of usual
business office activities.

B. Business Activity Limitations
1. It was hypothesized that the factors affecting

demand for telephone communication products

76

and services would vary across different
classes of business. This forced the use

of different questionnaires for each separate
class of business. Inclusion of all types

of business customers in the study would

have necessitated the structuring and printing
of a large number of different questionnaires.
This was impossible because of the time and
financial constraints associated with the
research design.

Usage data on certain classes of businesses
could not be released by Michigan Bell even
with management or owner consent. This was
particularly true for such institutions as
hotels, motels, hospitals and certain other
service organizations. For these businesses,
telephone usage by individuals not employed
by the firm itself made all usage information
confidential.

Because firms such as service stations, shoe
repair shops, barber shops, and many eating

establishments are very small, they exhibit

77

demand characteristics similar to resi—
dential customers. Firms in these business
classes may differ from residential users

in local phone usage, but they exhibit little
variation in their demand for equipment and
toll usage. This makes the problem of
forecasting toll and equipment demand for
these types of establishments trivial.

Some establishments of the size and scope

of the Oldsmobile Division of General Motors
and Michigan State University were located
in the survey area. These firms are so
complex in their communication needs and
behavior, however, that they are not typical
of the general class of firms investigated
in this study. The values that would be
obtained from firms of this nature for most
of the Classificatory variables listed in
Table 3.1 would be far beyond the range of
values obtained for the other firms in the
study. Thus, any analysis of firms of this

size would have to be done on an individual

78

basis and would be an exception to the

general models being developed.

One of the objectives of the study was to determine

the effects of certain independent variables on demand

across different types of businesses as well as within

certain industry segments. Because of this objective and

due to the limitations discussed above, five different

classes of commercial customers from the Greater Lansing

area were selected for analysis. The three main criteria

used for selecting these five segments were:

10

There should be at least one category which is
homogeneous with respect to the Classificatory
variables used to define member firms. There
should also be one category whose member firms
are heterogeneous with respect to these classi-
ficatory factors.

There should be a wide variance in the demand for
telephone communications products and services
between the large and small business establish-
ments in each category.

There should be a sufficient number of firms in
each category to assure an adequate sample response

for analysis.

79

Using the preceding criteria and following a study
of the frequency and size of various types of firms in
the Greater Lansing area, five business segments were
selected for analysis. Table 3.4 lists these segments and
indicates the types of firms (by SIC) included in each
segment as well as the approximate number of establishments
of each type located in the Greater Lansing area. Categories
(1) and (2) are service institutions and firms in these
groups are quite homogeneous with respect to most of the
Classificatory variables used in the analysis. Categories
(3), (4) and (5) are broad groupings and as such are com-
prised of firms reflecting extreme heterogeneity in their
Classificatory characteristics.

To determine the variance in demand and usage behavior
for each of the selected categories, a pilot study was
run where two or three small and two or three large firms
were chosen from each classification. Where possible,
thanet worth rating given in the Dunn and Bradstreet
Reference Book was used as a proxy measure of firm size.
When this was not possible, as in the case of the dry
cleaning and realtor firms, individual judgment based on

experience and knowledge of the area was used to select

80

TABLE 3.4

BUSINESS SEGMENTS SELECTED FOR ANALYSIS-~BY SIC
AND NUMBER OF FIRMSl

 

 

 

 

Approximate
SIC Number of Firms
Classes in Each Basic
Basic Classification Included Classification
1. Real Estate Establishments 65,66 1752
2. Dry Cleaning Establishments 721 703
3. Retailers 52,53,55
(except 554)
56,57,59
(except 591) 4804
4. Wholesalers 501-509 4445
5. Manufacturers 19—39 2356
l

The above categories are not mutually exclusive.
For example, a firm engaged in both retailing and whole—
saling would be included in both categories.

2Michigan Bell Telephone Directory for Lansing Area.

31bid.

4Manufacturer segments: U.S. Bureau of the Census,
Census of Business, 1963, Vol. II, Retail Trade-Area
Statistics, Part 2, U.S. Government Printing Office,
washington, D.C., 1966. Figure represents Lansing Metro-
politan Area only.

5

 

U.S. Bureau of the Census, Census of Business, 1963,
Vol. 5, Wholesale Trade-Area Statistics, U.S. Government
Printing Office, washington, D.C., 1966. Figure repre-
sents Lansing Metropolitan Area only.

6U.S. Bureau of the Census, Census of Manufacturers,
1963, Vol. III, Area Statistics, U.S. Government Printing
Office, washington, D.C., 1966. Figure represents Ingham
County.

 

 

81

the pilot firms. Usage and demand data was then obtained
for each of the selected firms from Michigan Bell records.
Table 3.5 illustrates selected demand and usage character-
istics for these firms and shows that a substantial
variance does exist in demand characteristics among firms
in each of the five categories.

Next, a comprehensive listing was developed of all
firms in the Greater Lansing Area classified in one or
more of the five segments. Compiling a comprehensive
list of these firms created a difficult problem, since
no one source provided a complete record of businesses
classified in this manner. The State of Michigan main-
tains a file on all businesses located within the State
for tax purposes, but this list is arranged alphabetically
and not by class or location of business. Michigan Bell
also maintains a listing of all of its commercial customers,
and codes each firm according to a specific classification
system. But no feasible way could be found to isolate only
those firms in particular classes that were located in the
Greater Lansing area. Various trade publications and the
Chamber of Commerce provided listings which were incomplete

and thus of limited value except for cross-reference purposes.

82

 

 

 

02 a H 0H.0H mH.m 00.0H m0.mH mom m
02 v m 0m.MN mm.mH 00.0 m0.m0 Hom Q
02 HN 0 05.m>m mh.m0HH 00.nm mm.m¢m 00m 0
02 6H m 00.00 m0.HmH mh.h mh.HmH 00m m
mm» 0H 6 0m.¢HN mm.OMH 0N.0H mN.HmH mom d
mHmHmmmHosz
oz 0 N 0m.H mm.H 00.0 0h.vN mm m
02 a N 00.0 00.0 00.m mm.0N m0 0
02 NH m mm.0m m0.mN 00.00 0m.0¢H m0 O
02 n N mh.H mn.¢H 00.00H m¢.Nw mm m
oz 0H d mw.m 00.NN 00.0aH 0m.mm m0 . d
mnouHmmm
oz N H 00.0 00. 0m.VH mm.NH HNS m
02 H H 00.0 00.0 0m.HN 00.HH HNS Q
02 m H mm.N 00.0 00.HmH 0N.mH HNh 0
oz 0 N 00.NH 0m.m 0n.h mm.vm HNb m
02 m v 00.0 mH.N m>.0HH 00.mm HNb d
mumcmmHO
6: mac.“ minute «we ,ch H60 .H60 Huamue 6H1, “665666
1mm» ImooH mumumlumucH wumumlmuucH mchHu mGHHHHm wpou OHm mmmchsm
xmm Ononm mmcHH mcHHHHm mocmumHQ lum>p¢ OOH>Hom
IOHOB mCOH >HOHOOHHQ HmOOH

 

MHMQ1MmMmD

 

 

mBZmzwmm mmMZHmDm m>Hm mom mUHBmHmMBuammmU
m0<mD ZmHm ZH WUZ<HM¢§ UZHZOmm mBHDmmm MDDBm BOHHm

m.m

Mdmdfi

83

 

 

 

 

.mmmH .Numounmm mo HucoE onp now no .mmmH .humonnmm mo mm CO>H0 mump HHNN
02 m N 00.m mN.m mN.h m0.0N mNm m
02 m H mv.v 00.n 00.0w mN.0H mmm 0
wow mnN ma 0m.00m mh.0N0 00.00m mN.mmmN Hmm 0
wow NH v 00.0 0n.mv 0N.Hm mH.N¢H Hem m
mm» hoH mH 00.H0 mm.0H 0¢.¢0 mN.hm0 Hmm 4
mHmHHmumm
02 m N 0n.mm 0m.hN mm.0m mh.vm mNN m
02 N N 00.0 00.v 00.nn om.mN mMN Q
mow mm m 0H.mwm 00.0Hm 00.0H 00.th 0mm 0
wow m0 0H 0>.Hmm me.omm 00.mv mh.nH0 vmm m
mm% mm 0H mm.hmMH mm.mv¢ 0m.0 mm.HHm mmm a
muwusuommocmz
2 20: 9:5: E HE HE .139 . Ema. 2; 223m
1mm» ImooH mo wummeHOHCH mumumlmuucH mCHmHu mcHHHHm OOOO OHm mmmaHmsm
xmm mcosm mmcHH msHHHHm mocmumHQ Inw>pd OOH>Hmm
IOHOB mQOH NHOOOOHHQ HMOOH

 

mama OUMmD

 

 

AUOSCHUCOUV m.m mqméfi

84

The Dunn and Bradstreet Reference Book1 lists firms
by city and classifies each establishment by the SIC
method. This reference was used as the major source for
developing a list of manufacturers, wholesalers, and
retailers. The manufacturer listing was supplemented by

The Directory of Michigan Manufacturers,2

and the Lansing
Chamber of Commerce data was used to supplement the
retailer list. However, no source could be located that
listed wholesalers by geographical area other than the
Dunn and Bradstreet reference.

The Dunn and Bradstreet Reference Book does not,
however, contain a listing of realtors and dry cleaners.
The yellow pages of the telephone directory, which includes
thorough listings of firms in these two categories, was
used exclusively as a source to define the population of
realtors and dry cleaners in the Greater Lansing area.

Since firm size was hypothesized to be the major

factor affecting firm demand for telephone communications

equipment and service, it was necessary to stratify the

 

1Dunn and Bradstreet Reference Book, 1968, New York:
Dunn and Bradstreet, Inc.

2The Directory of Michigan Manufacturers, 1968,
Detroit: Michigan Manufacturers and Financial Record,
Manufacturer Publishing Company.

85

firms in each segment according to a "size" variable.
There are two basic problems when stratifying firms by
size: first, to specify a relevant size variable (or
variables) for each category, and second, to obtain data
on these variables for each firm in the population. No
source of information or data on the usual size variables
such as gross sales, or number of employees, could be
located. The net worth code given in Dunn and Bradstreet,
therefore, was used as a gross estimate of the relative
measure of firm size for the manufacturer, Wholesaler, and
retailer segments.

The sample for the manufacturer, retailer, and
wholesaler segments was then chosen according to the
following criteria:

1. Since there were relatively few firms large

enough to be classified as either "A," "B," or
"C," all firms receiving these Dunn and Bradstreet
ratings were included in the sample.

2. A random sample was taken of those firms either

receiving a rating below "C” or no rating.
Table 3.6 indicates the number of firms in each rating

classification that were included in the manufacturer,

86

TABLE 3.6

NUMBER OF ESTABLISHMENTS INCLUDED IN
MANUFACTURER, WHOLESALER, AND RETAILER SAMPLES
BY FIRM SIZE*

 

 

D & B Net Worth Rating
Below E or Total

 

Business No Rating Sample
Segment A B C D E Given Size
Manufacturer 20 12 18 9 13 34 106
Wholesaler 15 18 26 16 13 20 108
Retailer 5 15 25 8 4 8 65

 

*The proxy variable used to indicate firm size is
the Dunn and Bradstreet net worth code. The net worth
assigned to each letter is as follows: A: $500,000;

B: $200,000-$500,000; C: $75,000-$200,000; D: $35,000-
$75,000; E: $20,000-$35,000.

wholesaler and retailer samples.

Stratification of realtors and dry cleaners by size
was even more difficult. The only adequate listing of
these establishments is in the Yellow Pages. A number of
firms do, however, purchase blocks of advertising space in
the Yellow Pages in addition to simply having their names
listed. These firms are usually assumed to be among the

larger realtors and dry cleaner establishments. Therefore,

\

in, .

 

H .__.. -_... ...—......_47“ +1- _ ..l
._*_H.—.._'_ ._._. ...—v ...—”M ”H

...-

......”

.._.,1--._-__._.

87

the total population of businesses in these two segments
that purchased separate advertising space were included
in the sample. A random sample was then taken of those
remaining firms whose names were listed under the general
realtor and dry cleaner headings in the yellow pages.
Table 3.7 indicates the number of establishments of each
type that were included in the realtor and dry cleaner
samples.

Upon completion of the sample for each of the five
business segments, addresses of the selected firms and,
When available, the names of owners or tOp officers in
each firm were obtained from the Polk City Directory for
Lansing. In those few cases where the Polk Directory did
not list this information, the Lansing Telephone Directory

was used to secure firm addresses.

Qpestipnnaire Administration

Five different questionnaires were prepared, one for
each of the selected business segments. (See previous
section dealing with "Questionnaire Construction"). Each
questionnaire was coded and accompanied by a permission
request form and cover letter stating the purpose of the

research and assuring the respondents of the confidential

88

TABLE 3.7

NUMBER OF ESTABLISHMENTS INCLUDED IN THE
REALTOR AND DRY CLEANER SAMPLES
BY FIRM SIZE*

 

 

Amount of Space Used
in Yellow Page Advertising
(proportion of paper used)

 

 

Business Segment 1/2 1/4 1/8-1/16 1/16
Realtors 4 8 16 12
Dry Cleaners 1 4 9 ll

 

*The proxy variable used to indicate firm size
for these segments was the amount of advertising space
purchased in the yellow pages of the telephone directory.

nature of their replies.

A total of 336 questionnaires were mailed within a
one—week period. After approximately three weeks, dupli-
cate questionnaires, permission request forms, and cover
letters were mailed under a second cover letter to those
individuals not responding to the original mailing.
Following a second three week period, personal phone calls
were made to those subjects who had still not responded.

A third set of materials were then mailed to those

subjects who stated a willingness to complete the questionnaire

89

in the telephone conversation. Table 3.8 lists the
number of questionnaires sent to each segment during each
of the three periods and the response rate following each

mailing.

Statistical Techniqges

 

The problem of segmenting markets and forecasting
demand are, in many cases, complementary. This is true
since many times a segmentation strategy is based upon the
relative amounts of a product or service demanded by
certain types or classes of customers.1 Even when market
segmentation is based upon some factor or factors other
than usage rates, the strategy may still be used as an
aid to forecasting demand. Thus, whenever a firm attempts
to forecast the demand for a particular product, market
segmentation plays a fundamental role in the development
of demand estimates. In fact, except for those industries

in which economic indicators or past sales are used to

 

1For examples of this type of segmentation, see
"Some Practical Applications of 'Heavy Half' Theory,"
Proceedings, of the 10th Annual Conference of the Adverti-
sing_Research Foundation (October, 1964) and Dik Warren
Twedt, "How Important to Marketing Strategy is the 'Heavy
User'?" Journal of Marketing (January, 1961), pp. 71-72.

 

90

forecast demand, market segmentation is a basic pre-
requisite for accurate demand estimation.
Based on this rationale, the first step in developing
a predictive model for telephone products and services was
to identify various segments of the commercial market
exhibiting different usage or purchase rates. Because
a large number of independent factors were examined,
search procedures were required to identify the most
important characteristics and to establish whether or not
significant relationships existed among specific variables.
Cross-classification analysis was one of the methods
used to describe the degree of association between specific
sets of variables. A principal advantage of this technique
is that it allows the researcher to "make relatively few
assumptions, prior to the analysis, as to the nature of
the relationships being studied."1 For example, the
assumption that a linear relationship exists between
variables is not necessary when using cross—classification
analysis. It is required when utilizing many of the other

multivariate techniques.

 

1Ronald E. Frank and Paul E. Green, pp, cit., p. 55.

 

91

mouHMCGOHummso mHnmmo mHao mopsHqu««*
oocusuou mouHmccoHummso HHm mmpoHocH*«
mausumn OOHmmO umom mom pmpmonpd¥

 

0mm

 

 

 

 

 

 

 

 

$N.0m n www.uoumm omcommmm anmmD Run ".mmm "mumm mmcommmm Hmuoe
00H mmN mm HS am mm 00 mHN we wHH 0mm mHmuoe
HHI. NH]. m1. H1. m1. Hm. mm. mmw. m1. N11. NM]. wumcmmHo sun
mN mm m m 0H m m mN HH mH 0v muouHmmm
Nv mm MH 0H NH NH 0H hm hH mN m0 mHmHHmumm
N0 Hm HN wN mN NH HN 00 aN 0m m0H meHMmmHosz
Hm mm SH HN vN mH NH HS HN 0m HOH wHOHopommsnmz
uwz Hmuoe umz Hmuoe pmHHmz umz Hmuoe pmHHmz ...umzflsmmpoe :pmHHmz pqmsmmm
Omcuspmm pwcuoumm pwauoumm pwcusumm mmwchom
mHMHOB Huompcoo OGOHQOHOD mCHHHmZ pcooom deHHmS umHHm

mcHSOHHouv

 

mcHHamz Hagan

r

ZOHB¢EmOmZH

mmZOmmmm QZ< UZHHHmE mmH¢ZZOHBmmDO

m.m mHm¢H

92

Another technique used to summarize or reduce the
large number of independent factors to a lesser number
was factor analysis. In addition to reducing the
redundancy in the original set of measurements, factor
analysis may be used to:1

1. provide a useful way to segment a population
into distinct categories based on specific
characteristics.

2. assist the researcher in extracting certain
variables from a large original set that may
then be used as independent factors in further
analysis.

3. create a smaller subset of variables that can
replace the original set of factors with little
or no loss of information.

Analysis of variance techniques were also used to
isolate particular factors that appeared to influence the
behavior of the dependent variables. Since conventional
analysis of variance is unsuitable in those cases where

the dependent variable of interest is either an attribute

or one which ”does not or cannot be made to conform to the

 

lIbid., p. 76.

93

shape of the normal distribution within each classification

"1 the use of

cell established by the factor variables,
these techniques was somewhat limited.
Following the application of various search techniques
to determine Whether or not relationships actually existed
among variables, multiple regression and multiple dis-
criminant analysis were used to specify and describe the
nature of these relationships. These techniques provided
the basic models for the predictive equations. Multiple
regression was the primary technique used but the categori-
cal nature of certain dependent variables required the
use of multiple discriminant analysis.
A fundamental problem should be stated at this
point. Studies in the consumer area have shown that the
use of socio-economic factors in multiple regression or
multiple discriminant analysis models has resulted in a
low proportion of explained variance when attempting to

predict individual family purchase patterns or usage rates.

However, as Bass pp g1. state, "The inability of

 

1William S. Peters and George W. Summers, Statistical

Analysis for Business Decisions (Englewood Cliffs, New
Jersey: Prentice-Hall, Inc., 1968), p. 298.

 

 

94

socioeconomic variables to explain a substantial part

of the variance of usage rates of persons does not
necessarily imply that there are not substantial differ-
ences in the mean usage rates for different socio-economic

1 This may also be true when looking at

market segments."
individual firm demand for telephone products and services.
Therefore, cross—classification analysis was used to
establish conditional probability models of each segment's
demand for different quantities of telephone offerings.

The development of these models was based on the measure-
ment of certain key independent factors. Thus, in addition
to focusing on individual firm demand, the differences in

mean usage rates for different classes of business firms

were also examined.

Summary

Since knowledge of the basic variables that affect
industrial demand behavior for telephone offerings is
limited, an exploratory type of investigation--where a

large number of independent factors were hypothesized to

 

1Frank M. Bass, Douglas J. Tigert and Ronald T.
Lonsdale, "Market Segmentation: Group Versus Individual
Behavior," Journal of Marketinngesearch (August, 1968),
pp. 269-270.

95

influence demand-—was necessary. To obtain information
on these factors, a mail questionnaire was developed.
The questionnaire was administered to selected firms in
five categories: manufacturers, wholesalers, retailers,
realtors and dry cleaners. Telephone communication data
on individual firm purchase and usage behavior were
collected from Michigan Bell records.

Because of the large number of independent or pre-
dictive factors, statistical search procedures such as
cross-classification analysis, factor analysis and analysis
of variance were used to determine if significant relation—
ships actually existed between and among variables. Multiple
regression, multiple discriminant and conditional probability
models were then constructed to specify more precisely
the relationship among variables and establish demand

equations.

CHAPTER IV
DETERMINATION OF THE FUNDAMENTAL VARIABLES
AFFECTING DEMAND

The purpose of this chapter is to examine the total
group of independent variables and isolate those factors
which contribute most to the explanation of firm demand
variances. These variables will then form the basis for
developing the predictive demand equations presented in

Chapter V.

General Characteristics of Firms

To better understand the nature of the firms studied
and the scope and significance of the demand equations
developed in Chapter V, some of the more important
characteristics of the firms surveyed in each of the five
business segments are presented in Tables 4.1 - 4.3 and
discussed in the following paragraphs.

In Chapter III, a number of proxy variables indicating
firm size were used to stratify the samples for each

business class. The selection of the proxy variables was

96

97

based primarily on their availability in secondary source
material. Tables 4.1 - 4.3 show data obtained from the
questionnaires on size factors that are more traditionally
used to indicate firm size. These tables indicate that the
proxy variables initially used to stratify the samples

did result in a satisfactory distribution of firms over

the more traditional size variables.

Table 4.1 gives employment figures for firms in each
business segment. It also indicates the mean number of
employees used by firms in each category as well as the
standard deviation of employment and the coefficient of
variations for each business class. It should be noted
that the realtor and dry cleaner classes are relatively
more homogeneous with respect to this variable than the
manufacturer, wholesaler, and retailer segments. This is
indicated by a coefficient of variation of 90 for the
realtor segment, 75 for the dry cleaner segment, 143 for
the manufacturer segment, 202 for the wholesaler segment,
and 270 for the retailer segment. Also, based on employ-
ment data, firms in the manufacturer segment had the
largest mean size (77.3 employees), and this segment had

by far the largest percentage of firms employing over 40

98

coHumHMm> mo ucmHOHmmmOO*

 

 

 

 

00H v.00 0.N¢ 0N 0H 0m NHH NOH H4609
mm >.m 0.n 0 0 H 0 0 umcmmHO Sum
00 N.0 0.0 0 0 N 0N NN HouHmmm
05N 0.mmH 0.00 a m 0 mN Nv umHHmumm
NON 0.0m H.0N m m 0H 00 N0 HmeMmmeoEs
mVH 0.0HH m.nh 0H 0 0 ON he Hmusuommocmz
*> m .m 00H 00H|Hv ovIHN 0N10 mEHHm acmeomm
mOHumHumum wumEEom mmmNonEm mo umnEoz .oz mmmchom

 

 

 

Bzmzomm mmMZHmDm

Mm

mmmWOHmzm m0 mmmzbz l mSMHm m0 ZOHBDmHMBmHQ WNHm

H.v MHde

ilili
lli

 

 

99

persons (47%).

Table 4.2 shows the distribution of firms in each
business segment according to gross revenue earned in 1967.
As was the case for total number of employees, the
coefficient of variation for the gross revenue factor
indicates that the realtor and dry cleaner segments are
more homogeneous than the manufacturer, wholesaler, or
retailer segments. The gross sales summary statistics
do vary somewhat from those of Table 4.1, however. For
example, based on the gross sales Size variable, firms in
the wholesaler segment have a larger mean size than manu-
facturing respondents.

Table 4.3 presents summarized data on a number of
other Classificatory variables that are used to categorize
firms in the five business segments. Although frequency
tables are not given, these summary statistics should help
the reader understand the nature of the specific businesses
selected for study.1 Of particular interest are the
variables related to geographical sales patterns and the

geographical location of suppliers. These variables are

 

1A random sampling of non-respondents was conducted
through brief telephone interviews to determine the
characteristics of firms not participating.

100

COHHMHHm> mo ucmHOHmmmoua

 

 

 

 

00H NmSv 0¢mN Nm 0N 0N 0S NOH H0909
Hm 0H Nm 0 0 0 0 0 “mammHO Sum
mHH 00m SON 00 H m 0H NN MOSHmmm
0HN 0S0¢ @00H 0 0 0H SH Ne umHHmumm
mNm 0moNH SHSm vN 0 0 0H N0 umemeoaz
mMN N0¢S SOHm 0H 0 0 0H Sv nousuommocmz
*> m .m 00NH 00NHIH00 000lH0¢ 000:0 mEHHm unmemom
mOHumemum mumEEsm ~w.000W0 wocm>mm mmouw .oz mmmchsm

 

 

Bzmzwmm mmmZHmDm Mm
MDZm>mm mmomw SmmH I mzmHm ho

N.v MHm¢H

ZOHBDmHMBmHQ WNHm

101

hypothesized to correlate strongly with a number of demand
factors in the manufacturer, wholesaler, and retailer

segments.

Some General Characteristics of Firm Demand

The firms studied in each segment varied greatly
in their telephone communication usage and demand behavior.
This was particularly true for manufacturing, retailing,
and wholesaling establishments. Before the relationships
between independent variables and demand are developed,
some general findings will be presented concerning the
average usage and demand characteristics of each business
segment.

Telephone usage in this study is separated into
three components--monthly intrastate, monthly interstate,
and total monthly toll billings. As mentioned previously,
a more detailed breakdown was not possible since this
would have required a monitoring of all calls made by a
firm during a specified period of time. Table 4.4 pre-
sents the average dollar amount of intrastate and interstate
toll usage for business firms in each segment for the
month of February, 1968. This table shows that wholesalers

had the highest mean intrastate toll usage while

102

.umew mHCmHH meCB mucwemmm mmonu ou mHQmOHHQO HOC mH mHHMHHm> «0802*

 

0.0 O.H
g. m.N
H.0 0.0H

II II 0.0m 0.HN N.¢N m.HN

II II 0.0m ¢.SH 0.0m 0.0H

S.0 H.0 II II 0.0 v.v
II II mVHN mvm N.MHH N.H0
II II H.0H 0.NH 0.0H 0.0

0.? 0.0 0.0 0.0 N.¢ 0.0
II II N.0m 0.NN H.0N 0.0H

0.HH 0.0 m.HN 0.0N N.¢H N.SH

S.NH H.S
H.HS 0.¢S
0.0H 0.NH
0.0 0.?
0.Sm H.0m

0CHmCMH mpHmuoo

CmmHHOHE CH mHmHHmmom S
mwum 0CHHHMO

HMOOH CH meHHmmom x
0CHmCMH mOprso

CmmHSOHE CH mmHMm X
mmum 0CHHHmo

HMOOH CH mmHmm X

pHmHm CH CmemHMm .oz

mumHHmmsm .oz
mmmSOHmEm OmHHMHmm .oz
mmOHmmo .oz

Hm.ooo0 momma
HOOHM mo ummm .00
SuHHHomm CH mummh .oz

 

HmCmmHO HMQ

m .m m .m m .m
HouHmmm HoHHmumm umemmHOCZ

Hmnsuommssz

 

pCmEWMm mmmCHmom

OHumHkuomumnu

 

 

(Illi
ll

Bzmzwmm mmmZHmDm Mm I QMHmE<m mSmHm m0 mUHBmHmmBU¢M¢EO QWBUmHmm

0.? mqméfi

103

manufacturers had the highest mean interstate toll usage.

Ideally, toll data for all firms should have been
collected for each month of the year to account for differ—
ences caused by seasonal influences. The difficulty in
gathering the data and the amount of time necessary to
complete such an effort made this task impossible.
Theoretically, this means that any predictive model for
toll usage developed is only valid for the month in which
the data were gathered.1 This restraint does not apply
to the predictive models developed for equipment and
services since the demand for these offerings is not normally
seasonal.

The demand for equipment and services can be repre-
sented by four summarized variables:

a. The total monthly dollar value of equipment

billings
b. The number of lines or trunks used

c. The number of telephone locations a firm maintains

 

lHowever, usage data on all firms collected for a
second month and a sampling of usage data for a third month
showed close similarity with the first month's data. In
those few cases where a radical difference did exist in
the monthly toll figures, an average of the intrastate and
interstate usage figures was used.

104

TABLE 4.4

MEAN INTRASTATE AND INTERSTATE TOLL BILLINGS
FOR FEBRUARY, 1968 - BY BUSINESS SEGMENT

 

‘1‘

Amount of Toll Billings (5)

 

 

Business ‘_Intrastate ‘_Interstate

Segment x s x 3

Manufacturer $111.02 $147.05 $153.15 $267.26
Wholesaler 165.30 767.23 106.37 223.33
Retailer 61.36 98.89 47.81 116.30
Realtor 27.45 47.01 15.91 17.42
Dry Cleaner 2.11 1.54 7.89 8.72

 

d. The amount of Yellow Page Advertising a firm
utilizes

Table 4.5 summarizes the amount of each of these offerings
used by firms in each business segment. In analyzing
Table 4.5, it is readily apparent that the mean amount of
money expended on Yellow Page Advertising is greatest in
the retailer segment. An average of $85.19/month is spent
by each retailer. This result is expected since the
retailer and realtor (who ranks second in expenditures for

Yellow Page Advertising) deal directly with the consumer

105

 

 

 

 

 

00.H 0S.H HS.0 00.H 0H.0H 00.SH 00.NN 00.NN mumCmmHO Sun
00.0H H0.0 N0.N 00.0 0H.0SH 0H.SOH 00.00 N0.00 muouHmmm
0S.N0 00.0H 00.S 00.0 0H.000 N0.HON N0.SS 0H.00 mHmHHmumm
00.00 00.0H 0N.0 NS.0 00.000 00.HOH S0.00 0H.00 mumHmmeOSS
pm.OH HN.HH Ho.m He.m mm.mmHm 0S.H6Hm mo.6mm mp.me mumusuommscmz
m .M m .M m 1 .m m .1 .m A quemmm
meHmuCHmz .mOOH Omwb va,mCHHHH0 «WV mCHHHHm mmmCHmom

mCOHQmeB .oz

mmfiflq MO .HOQEDZ

quEmstm .oz

0CHmHuum>p¢ .oz

 

pmpCmEmQ mmCHumwmo mCosmmHmB

 

 

 

Etta

 

ezmzomm Hambm Sm .. momH .wmsgmmmm mom
one¢NHqup moH>mmm 02¢.ezmzmHDom moamm>¢

0.0 mHde

 

 

106

and would benefit most by maximum exposure in this medium.
Figures in Table 4.5 Show that firms in the retailer
group also have a higher average expenditure for equipment
and a larger mean number of lines and telephone locations
than firms in the other segments. More interesting and
informative comparisons between segments might be made,
however, by looking at several ratios. Table 1 in
Appendix I presents five ratios relating the dependent
variables "number of lines" (NT) and "number of locations"
(NL) to each other and to the independent variables
"number of employees” (NE) and "number of salaried
employees" (NS). One would hypothesize that the greater
the value a firm places on a caller being able to reach
his number on the first call without receiving a busy
signal, the lower the ratios NL/NT, NE/NL, NE/NT, Ns/NT,
and NS/NT should be. In other words, these ratios should
be lowest for those segments or firms where there is a
high probability that a caller will try a competitor rather
than continue to call a busy number. This is demonstrated
most clearly in the realtor segment, where the ratios of
NL/NT, NE/NL and NE/NT are substantially below those

obtained for the manufacturer, wholesaler and retailer

107

segments. It is also interesting to note that the manu-
facturer segment has a Significantly lower number of lines
and locations/employee than either the wholesaler or
retailer segment. All three segments, however, have
approximately one phone for every salaried employee.

Data were also obtained on various other types of
equipment and services used by individual firms. This
information is summarized in Table 4.6. As indicated in
this table, the demand for these product and service
offerings was not great enough to allow a regression or
demand equation to be developed. In other words, the
number of firms sampled in any one segment that used such
offerings as WATS was so small that it prevented any
general hypotheses or conclusions from being drawn. How—
ever, some general findings and implications related to
these additional dependent variables are discussed in the
following paragraphs.

A total of nine responding firms were found to
subscribe to some type of WATS service. This included one
manufacturer, and eight Wholesalers. No retailers responding
to the questionnaire used Wide Area Telephone Service. If

these findings are representative of all similar type and

108

TABLE 4.6

DEMAND AND USAGE OF OTHER EQUIPMENT AND SERVICES
FOR FEBRUARY, 1968 - BY BUSINESS SEGMENT

 

Type of Equipment or Service

 

Business Firms Firms Firms Using
Segment Using WATS Using PBX CALL DIRECTIONS
Manufacturer 1 12 5
Wholesaler 8 9 7
Retailer O 6 6
Realtor 0 2 5

Dry Cleaner 0 0 0

 

size firms, it is difficult to hypothesize why a higher
proportion of wholesalers use WATS than do manufacturers.
In fact, no differentiating features were found between
those firms using WATS and all other wholesalers and
manufacturers.

Data were also gathered on the number of firms
using PBX'S (switchboards) and the specific types of
switchboards used. (line and station capacity) Twelve
manufacturers, nine wholesalers, six retailers, and two

realtors sampled were found to have some type of switchboard

109

equipment. In percentage terms, this means approximately
26% of the manufacturers, 15% of the wholesalers, 14% of
the retailers, and 10% of the realtors sampled use PBX's.
As would be expected, the firms in each segment that use
switchboards are substantially larger (based on number of
employees or gross sales) than non-users. For example,
manufacturing firms using switchboards have an average
of 160 employees while non-switchboard manufacturers have
an average of 62 employees. As was the case with WATS
usage, however, there were not enough firms using switch-
boards in any one segment to accurately develOp equations
for predictive purposes.1
One other type of equipment usage examined was the
demand for Call Directors. Here, as with WATS usage, only

a very small subset of those sampled used call director

equipment, and no meaningful relationships could be derived.

Attitudinal Characteristics of Responding Firms
The importance that management places on various
aspects of their telephone communications activities was

hypothesized to influence both the amount of equipment and

 

lA discriminant analysis was run to differentiate

between users and non-users of WATS and switchboard equip-
ment, but no significant relationships could be found other
than those related to the size variables mentioned above.

110

usage demanded. The survey included a number of questions
which required respondents to rank various activities

for which they used telephone communications. Rankings
were based on the amount of time the customer spent on
each activity and the importance which he felt the tele-
phone played in each communication activity. Response
means and statistics related to these results are pre-
sented in Appendix II.

Several findings of interest are evident in the
tables presented in Appendix II. First, the majority of
firms in all segments ranked the activity "communicating
with customers" as being the most time consuming.

However, firms dealing directly with the consumer (i.e.,
Retailers, Realtors, and Dry Cleaners) ranked themselves
as spending a larger portion of their telephone usage time
communicating with customers than did firms in the manu—
facturer and wholesaler segments.

Second, all segments other than realtors rated
communicating with suppliers as the most time consuming
toll telephone communications activity. Realtors as a group
ranked communicating with salesmen or brokers as using up

the most toll usage time.

111

A respondent's ranking of the time spent communica-
ting with various groups may differ from a ranking which

specifies the importance of telephone communications to

 

each of these areas. In other words, even though a
customer states that the majority of his telephone time

is spent communicating with customers, he also might rate
telephone usage as more important to some other communica-
tion activity.

The tables in Appendix II show that for the majority
of cases, the relative importance which firms attached to
particular telephone communications activities was strongly
related to the perceived amount of time spent on those
activities. There were, however, several exceptions that
are illustrated in the summarized data presented in
Tables 4.7 and 4.8. These tables illustrate the differ-
ence between the mean "importance" and mean "time" rankings
for each segment and each activity. They show that in a
number of cases there were significant differences between
these two rankings. For example, although both manufacturers
and wholesalers ranked "communicating with suppliers" as
the most time consuming toll telephone communications

activity, firms in both segments ranked "communicating with

1112

.SuH>Huom mCOHumoHCCEEou oz» mH ucwuuomEH once no UCHEouCOU oEHu whoa on» .osHo> 0C» HosoH onaac

.CoHumUHuo ms» an 00. I
a CO owmmn UCMOHMHCUHn uo: 0&3 0CHmeu wocmuuomEH cme 0C0 mEHu some wazuon ounmummuHG u.n.Cc

maHun Human»

 

 

 

 

 

 

.m.C 0.0 0.0 00. 0.0 0.0 .m.C 0.0 0.0 .m.C 0.0 0.0 .m.C 0.0 0.0 mmHUHHHUMh
gunman
" * * OmOC MON 00N .mOC ~0¢ ”CM 0W0: NOM w.” mo. w.” v.” cmenoflflm
u- . . mo. H.0 S.m .m.c o.m m.~ mo. H.0 v.m .m.: m.m ~.m .uuomo Horne
.m.c e.~ H.~ .m.c o.m o.m ...: ¢.~ S.~ ...: m.~ m.~ ....c ~.~ e.~ numHHaaom
.m.: o.H H.H .m.c m.H ~.H mo. H.H 0.H mo. m.H H.N mo. m.H ..m.~ ouoeouuoo
”nqu
0CHquHC5EEoo
.0H0 .mEH mEHa .0H0 .mEH «EH9 .0Hm .QEH «EH8 .mmw .mEH ImmHB .UHWII .deH oEHB
cam: Com: cam: cum: cam: cum: Cmoz cam: Cam: awe:
HocmmHO awn muouHmmm umHHmumC umHmmeOCS Housuommscmz

 

quemmm >0 I‘mmcchmm ooqmuHOQEH pCmImEHB Cmmz

Ezmzwmm mmmZHmDm Mm I flwdmb mzomQMHmB HdUQH ho
mUZHXZ€m NUZ¢HmOQZH D2¢ QZHB Z¢fi£ m0 ZOmHM¢flSOU

S.0 mamfifi

.NuH>Huua

mcoHumoHCsEEoo mnu mH quuuomEH once no 0CHEsoCOu oEHu once on» .HOQECC on» uoon 039cc

.CoHuouHuo an» no 00.

0CHns anon!»

 

.113

 

 

 

 

 

 

 

m :0 woman uanHchmHm no: mm3 0Cchmu muCMuHOQEH Cmmfi pCm mEHu anus Cmm3uwn oOCOHmMMHQ ..n.:«
II c r H0. 0.N H.0 .m.C N.0 0.0 .m.C 0.0 H.0 .m.C 0.N 0.N uoHuHHHunm
COCon
In . . .m.c H.H v.H .m.: m.m o.m .m.: m.m ~.m .m.: 0.m o.n :meuonm
Ho. m.m ~.H .m.: m.v m.v .m.: e.H m.H .m.: v.~ o.~ ..u.: m.~ o.~ uuoHHamsm
00. H.0 0.H 00. 0.H 0.N .m.C 0.N N.N 00. 0.H 0.N 00. 0.H st0.N numEoumso
”CuHS
mCHumoHCCEEOU
.0H0 .mEH mEHB .on .QEH \mflHB .on .QEH 05H . Hm rmfiH MEHB .mflw, mmmH‘ oEHHI NJH>Huu¢
cum: me2 com: saw: cum: cam: now: Cmmz Cam: saw:
umcmmHO 0H0 mu0uHmm0 HoHHmuwm umHmmmHon3 nousuommscmz

 

usmEmmm >0 I moCHmem mummuuanH UCM wEHB com: I

BZHZUMm mmMZHmDm Mm I mwdmb mzommmqmfi HHOB ho
wOZH&Z¢m mUZ¢HmOQZH 92¢ HZHB 20m: ho ZOmHmdmZOU

0.0 mHm¢H

114

customers" as the most important toll activity.

One other aspect of the attitudinal portion of the
study should be mentioned at this point. This deals with
the portion of the questionnaire that asked respondents
to compare their demand for equipment and usage with what
they perceived to be demanded by other firms of similar
type and size. Table 4.9 summarizes this information by
indicating the per cent of firms that perceived their
demand for various offerings to be below, the same as, or
above that of similar size and type establishments. A
finer breakdown of this data is presented in Appendix III,
Where the data is tabulated by business segments.1 Several
general findings related to the information presented in
Appendix III are reviewed in the following paragraphs.

First, the large majority of firms perceive their
local telephone usage to be average. Very few firms in
any segment perceive their usage of the telephone for local
communication to be "below average.” However, with respect

to toll usage, approximately the same number of firms

 

1Although data were collected on a five point scale

(far below average, below average, average, above average,
far above average), only three summary classifications are
used in the tables.

115

TABLE 4.9

PERCEIVED TELEPHONE USAGE RELATIVE TO FIRMS
OF A SIMILAR TYPE AND SIZE

 

 

 

Perceived Usage
% Firms % Firms % Firms
Perceiving Rerceiving Perceiving

 

Number Usage Usage Usage
of (Similar ==Similar >Similar
Activity Firms Firms Firms Firms
Local Usage 183 6% 64% 30%
Toll Usage 183 13 47 40
Total Equipment Used 181 13 57 30
Time Used in Supplier
Communication 181 ll 58 31
Time Used in Customer
Communication 182 9 53 38
Yellow Page
Advertising Used 182 26 53 21

 

perceive their usage to be both above and equal to that of
comparable type and size firms. The fact that toll usage
is a variable cost whereas local usage is typically a
minor fixed cost may help to explain these differences.
All firms can use the telephone for local usage as much as

they desire without additional cost. Therefore, it is

116

probably true that most firms of similar type and size
would perceive less variance in local usage than in toll
usage.

Second, the data presented in Table 4.9 suggest
that the majority of firms perceive their use of Yellow
Page Advertising to be equal to that of comparable firms.
This consistency of response probably derives from two
sources. First, the number of alternatives a firm faces
here is less than would be the case in the other demand
categories. Only a limited number of size options are
available to a firm for Yellow Page Advertising space.
Second, and more important, most firms are probably quite
aware of the types and sizes of Yellow Page ads their
competitors carry. In many cases, they will try to maintain
a similar amount of advertising space.

The data presented in Appendix III, however, indicate
some differences in these perceptions across business seg-
ments. The manufacturer and Wholesaler segments both have
a larger percentage of respondents that perceive their
use of Yellow Page Advertising to be below average than
either of the other three segments. This might be explained

by our earlier reasoning that manufacturers and wholesalers

117

are not as concerned and do not utilize the Yellow Page
services as heavily as do firms that deal directly with
the consumer. If this is true, then it is also likely
that firms in the retailer, realtor, and dry cleaner
segments will try to maintain ads of similar or larger
size than competitors. As Appendix III shows, these firms
do perceive their usage to be either equal to or above
that of similar type and size firms.

The response patterns associated with the perceived
demand and use of equipment are very similar across segments.
The majority of firms perceive their use of telephone
equipment to be "average" with a large percentage of the
remaining firms placing themselves in the "above average"
category.

The final two activities listed in Table 4.9 are the
amount of time perceived to be used in communicating with
both customers and suppliers relative to usage by similar
type and size firms. These findings show the large
majority of firms in all segments in the "average" or
"above average" categories. The realtor segment, in fact,
had a majority of firms (61%) view themselves as above

average in their use of the phone to contact customers. No

118

firms in this segment perceived their use to be below
average.

The fact that such a large percentage of firms View
their demand for various telephone offerings and their
usage of these offerings for various activities as either
average or above average has certain implications for the
supplying utility. Assuming that the best potential
customers for a product will be those that perceive their
usage to be below average, the large majority of firms
included in this study can be termed "poor prospective
buyers." This may not be the case, however, if the firm
which perceives its demand or usage to be average or above
average is actually "below average” and can be convinced
of this fact. This implies that a major function of the
salesmen in these situations is to relate perceived demand
to actual demand and adjust his sales approach accordingly.

The relevance of the relationship between perceived
demand and actual demand as well as the potential importance
of the other attitudinal data on the demand variables
requires further research. Such research would hopefully
lead to a better understanding of the perceived demand and

actual demand relationships and a system whereby priorities

 

119

could be assigned to customers based upon their potential

demand for additional telephone offerings.

Relationships Between Criterion Variables

 

Prior to examining in detail the interrelationships
among and importance of the explanatory variables, it
should be helpful to review some of the interdependencies
existing in the dependent or demand variables. A simple
correlation matrix in which the seven major demand variables
are included for each business segment is presented in
Table 4.10. Several basic points can be summarized from
the information presented in this table.

First, the amount of Yellow Page Advertising that
manufacturing and wholesaling firms use is not significantly
correlated with any other demand characteristic. This
is not true for the retailer, realtor, and dry cleaner
segments, however, where significant correlations exist
between the advertising criterion and a number of the
other dependent variables. As was hypothesized earlier,
this is probably due to the fact that Yellow Page Advertising
is a more effective tool for firms dealing directly with
the consumer. These firms, therefore, tend to show a much

greater variance in demand for this offering than do

O
1

685: 358....

.oosluo«uuooo so«ua~ouuoo unuosdnlll oulalvnno on Inland-s- can! usuaoauuaosuo

 

 

 

 

04 o.~ o.” o." a." o.« no. no. so. no. no. em. no. 00. no. on. c c ah. an. 00. c a 00. ha. 50. «a. no. on. so. on. to. an. no. so. un.u ocean-cod ocosn

stack we wanna:

o.~ o.“ o.” o.~ o.w no. on. ca. 00. no. a o «h. n». no. 0 o no. on. no. 06.: as. no. ca. Oh. an. so. no. no. «9.. noted an hunts:

o.” o.~ o.~ o.H o.n o c an. as. co. 0 c on. on. an. no. no. as. on. so. an. «n. no. we. «0.: o.» unuunqavu

c c o!— o..~ OJ 0 c or. on. 00. c 0 v0. 00. on. o c «v. 8... an... coco—Jana: 30k

. ou-uouuucu

o; o; 03— o; o; c c «a. no. no. 0 a no. On. nu... ccnucmaaan «an...

ouaunnuucu

o.u o.u o.u o.« o.“ so.» an. an. co. cu.n oo-ocanuao

"dob n38

.3 04 o." 94 94 ... 05-1334

1 un‘llounlIUMHIJuddliuudlfiJudi! 02-?!»

2.. 3. an. .3 an. an. a: couuvuauu
3333335

 

Exam wazugn in I wag; swan-nu ho Numb! Buhgo

Ou.¢ Hans.

121

wholesalers and manufacturers. For example, although a
large manufacturer will utilize a greater amount of tele-
phone equipment and more toll time than a small manu-
facturer, the larger firm will probably not find it
advantageous to carry more Yellow Page space than the
smaller establishment. On the other hand, the use of
Yellow Page space in the retailer, realtor, and dry cleaner
segments very likely would increase along with the demand
for toll usage and telephone equipment as the firm's size
increased. The importance of the size variable on the
demand for various telephone offerings will be discussed
more extensively in the latter part of this chapter and in
Chapter V.

Second, the total amount of toll time used by firms
in each segment is correlated significantly with the amount
of equipment a firm uses. Toll usage is correlated most
highly with the number of telephone locations that a firm
maintains. This is especially true for the wholesaler
segment where a correlation coefficient of .97 exists
between total toll billings and the number of telephone
locations maintained. These findings are similar to those

arrived at in Michigan Bell's study of the consumer

122

1 where a direct relationship was found between the

market
demand for equipment and the amount of toll time a resi—
dential customer used.

Third, since telephone lines and telephone locations
constitute the major equipment demanded by firms in this
study, the total amount of equipment dollars spent is
strongly correlated with these two offerings. If only the
ten largest firms (using number of employees as the size
indicator) in the wholesaler, manufacturer, and retailer
segments are considered, the correlations between equipment
dollars spent and the number of telephone lines and tele-
phone locations drops significantly. This reflects a
greater variance in demand by large firms for the additional
types of vertical equipment offerings that were mentioned
previously in this chapter.

Finally, in all but the wholesaler segment the number
of telephone locations a firm maintains is highly corre—
lated with the number of telephone lines it uses. The poor
correlation in the wholesaler segment is also reflected

by the large variance in the number of telephone locations

 

l
p. 42.

Profile of the Michigan Bell Consumer, op, cit.,

 

123

maintained by wholesaling firms (see Table 4.5). Although
firms in the wholesaler segment did not vary substantially
in their demand for telephone lines, they exhibited a

high degree of variance in the number of telephone locations
maintained. The factors causing the variation in these

and other demand variables will be the major topic of
discussion in the remainder of this chapter and in

Chapter V.

Simple Correlations Between Independent and Criterion
Variables

 

 

Tables 4.11 - 4.15 present product-moment R's for
the Classificatory and attitudinal variables that are
significantly correlated with specific criterion demand
variables.1 These correlation matrices are presented by
business segment. Table 4.16 summarizes this information
by presenting the total number of independent variables
that were hypothesized to affect specific demand criteria
and the number that were found to have significant associa-

tions. This table indicates that, with the exception of

 

1The tables are abbreviated to only indicate
relationships that were significant at the .01, .05, and
.10 levels because of the large number of independent
variables that were originally hypothesized to affect demand.

124

the Yellow Page Advertising criterion variable in the
manufacturer segment, between 11 and 22 explanatory factors
are correlated significantly with all dependent variables

in the manufacturer, wholesaler, and retailer segments.

Manufacturer Segment Associations
Table 4.11 indicates that only three of the 44
variables in the manufacturer segment that were hypothesized
to affect Yellow Page Advertising usage are correlated
significantly with the criterion variable. Of these three,
only the variable "% sales in the Lansing Area" is significant
at the .05 level. This means that of the total array of
explanatory variables tested, the manufacturer's Yellow
Page Advertising usage is influenced most by the per-
centage of total sales occurring in the local calling area.
For the interstate toll criterion, the explanatory
variable that has the highest coefficient of correlation
is "Gross sales outside of Michigan." Since the interstate
toll component of the total toll variable is greater than
the intrastate toll component, the explanatory variable
related to out-of-state sales also carries a high coefficient
of correlation with the total toll criterion. Several

"size" variables, particularly the "number of offices" and

125

TABLE 4.11

SIMPLE CORRELATION MATRIX FOR SELECTED EXPLANATORY AND
CRITERION VARIABLES - MANUFACTURER SEGMENT

 

Criterion Variables

 

 

Explanatory Advertising Total Intrastate Interstate Equipment Telephone Telephone
Variables § T011 5 Tollp§ Toll § § Lines Locations
Size variables
1. No. Buildings n.s. .30+ .51* n.s. .33+ .41* .38*
2. Facility Sq. Footage n.s. .32+ .41* .22' .43* .47* .52*
3. No. Offices n.s. .64' .46* .67‘ .67* .70. .60*
4. 1967 Gross Sales n.s. .28+ .24+ .26+ n.s. .24+ n.s.
S. No. Employees n.s. .19° .24+ n.s. .32+ .33+ .38'
6. No. Salaried

Employees -.21° .68* .68* .59* .70* .70* .80*
7. No. Clerical

Employees n.s. .35* .44* .26+ .30+ .40* .42*
8. No. orders/month n.s. .27+ .52* n.s. .19' .22' .21°
9. Total Outlets n.s. .l9° n.s. .21' n.s. n.s. n.s.
10. No. Suppliers n.s. .31+ .32+ .27+ .43* .45' .50*

Geographic Variable;
ll. % Sales in Lansing

Area .28+ -.51* -.48* -.47* -.40* -.43* -.47*
12. % Suppliers in

Lansing Area n.s. -.30+ -.28+ -.28+ n.s. n.s. -.23'
13. No. Outlets in

Area I'A" n.s. .28+ .20' .29+ n.s. n.s. n.s.
14. N5. Outlets in

Area '8” n.s. .31+ .21' .32+ n.s. n.s. n.s.
15. No. Suppliers Out-

side Lansing n.s. .37* .35* .34* .44. .43* .S2*
16. Re. Suppliers out-

side Michigan n.s. .47* .38* .46. .36* .38* .51'
17. Gross Sale Out-

side Lansing n.s. .30+ .25+ .29+ n.s. .25+ n.s.
18. Gross Sale Out-

side Michigan -.20' .80' .57* .B3* .47' .44* .48*
19. X Sales in Mich.

Outside Lansing n.s. .19“ .4S* n.s. n.s. .30+ n.s.
20. Distance tro-

Home Office n.s. n.s. n.s. .34* .25+ .29+ .27+
21. Ne. Suppliers in

Michigan n.s. n.s. .22' n.s. .39* .38* .40*

Afititudinal Variables
. Perceived Time

Spent on Local

Communication

‘with Customers n.s. .37. .31+ .35* .56. .48. .54*
23- Perceived Under-

standings of

Offerings n.s. -.26+ n.s. -.29+ —.26+ -.21° -.30+
\

'Level of significance is equal to or less than .10.
+Level of significance is equal to or less than .05.
*Level of significance is equal to or less than .01.
has.The coefficient of correlation for the given variables is not significant at the .10 level.

 

126

the "number of salaried employees" a firm had, also are
correlated quite strongly with all toll variables. It is
interesting to note that the salaried employee variable
has a much higher correlation with the total toll and
interstate toll variables in this segment than the
independent variable ”number of employees." As will be
seen later, however, the Opposite relationship exists for
the wholesaler and retailer segments. There are two
related factors that might cause this variation. First,
manufacturers in most cases employed a larger number of
nonsalaried personnel than did the retailers or Wholesalers.
Second, the non-salaried personnel employed by the manu—
facturer appear to have a lesser influence on the demand
for telephone usage and equipment than do the non—salaried
personnel employed by firms in the retailer and wholesaler
segments.

Finally, the demand for equipment in the manufacturer
segment is correlated most highly with the "number of
offices" and the "number of salaried employees" a firm
maintains. The variable related to the perceived amount
of time spent communicating with customers in the local

calling area is also highly correlated with the equipment

127

criteria variables. However, this attitudinal factor
is not correlated significantly with any of the explanatory

1 This implies that a manufacturer's

"size" variables.
demand for telephone equipment is sensitive to his per-

ceived local communications behavior as well as to the

absolute size of the firm.

Wholesaler Segment

There are a larger number of explanatory factors
significantly correlated with the advertising dollars
criterion variable in this segment than in the manufacturer
segment. But only two factors, the "number of years the
firm has been in business" and the "number of orders/month"
are significant at the .05 level. It is interesting to
note that this is the only case in which a variable
concerning the length of time a firm has been in business
is correlated significantly with a demand variable.

All other criterion variables are highly correlated
with a number of "size" factors and with the geographic

variable "gross sales outside of Lansing.“ The explanatory

 

lAlthough collinearity is not a problem in this case,

many of the independent variables included in Tables 4.11 -
4.15 will be dr0pped from consideration later in the
analysis since they are strongly collinear with other variables.

128

factor "gross sales outside of Michigan" was correlated

more highly with the criterion variables in the manufacturer
segment than was the factor ”gross sales outside of Lansing."
The in-state component of sales, however, has more affect

on the criterion factors in the wholesaler segment. The
difference in the effects of these two geographic variables
is greater than might be expected by looking only at the
average percentage of total sales occurring in outstate

1 As will be discussed

Michigan for these two segments.

later, however, a number of other factors weigh heavily on

the demand behavior of wholesalers and manufacturers.
Another factor that affects the criterion variables

differently in the manufacturer and wholesaler segments

is the perceived importance of the telephone in dealing

with customers and suppliers. The perceived amount of

time spent communicating with customers is correlated

significantly with all criterion variables except Yellow

 

1As indicated in Table 4.3, the manufacturer segment

had an average of 29% of their total sales in Michigan out-
side of Lansing, while the Wholesaler percentage was 42%.
These percentages reflect the fact that firms in this study
are near the center of the State of Michigan. A location
nearer the border of the state might have caused a much
greater increase in out-of-state sales, particularly for
wholesale establishments.

SIMPLE CORREIATICM MTRIX FCR SELECTED EXPIANATORY AND

129

TABLE 4.12

CRITERIW VARIABLES - WHOLESALER SEGMENT

 

Criterion Variables

 

 

Explanatory Advertising Total Intrastate Interstate Equipment Telephone Telephone
Variables fi, T011 S T011 5 T011 5 5 Lines Locations
Size Variables
1. No. Buildings .19“ n.s. n.s. n.s. n.s. n.s. n.s.
2. Facility Sq. Rxxage n.s. .26+ .25+ .25+ .39* .41* .42*
3. No. Offices -.lB“ .43* .36* .SS' .56* .61* .49*
4. 1967 Gross Sales n.s. .95“ .94. .75“ .94* .90* .9S*
5. lo. Employees n.s. .89“ .88' .68' .95“ .89* .96*
6. Ho. Salaried
Employees n.s. .67. .62* .66* .73* .74* .75.
7. No. Clerical
Employees n.s. .84“ .82* .69. .88* .86* .90*
8. No. Orders/Month .25+ n.s. n.s. n.s. n.s. .22+ n.s.
9. No. Outlets in
Michigan -.18 n.s. .36“ n.s. n.s. n.s. n.s.
10. Total No. Outlets n.s. n.s. n.s. .38' n.s. n.s. n.s.
11. No. Suppliers n.s. n.s. n.s. n.s. n.s. .21° n.s.
12. 80. Years in
Operation .25+ n.s. n.s. n.s. n.s. n.s. n.s.
Geogggphic variables
13. x Sales in
Lansing Area .21“ -.28+ -.21“ -.46* -.25+ -.24+ -.24+
14. No. Outlets in
Area “A: -.19“ n.s. n.s. .l7“ n.s. n.s. n.s.
15. No. Suppliers Out-
side Lansing n.s. n.s. n.s. n.s. n.s. .21“ n.s.
16. lo. Suppliers Out-
side Hichigan n.s. n.s. n.s. n.s. n.s. .21“ n.s.
17. Cross Sale Out-
side of Lansing n.s. .96“ .94“ .76* .94* .88“ .95“
18. Gross Sale Out—
side of Michigan n.s. .17“ n.s. .33* n.s. .18“ n.s.
l9. ,6 Sales in Mich.
Outside Lansing n.s. .26+ .24+ .26+ .23+ .21“ .22+
20. Distance from
Home Office -.10“ n.s. n.s. n.s. n.s. n.s. n.s.
Attitudinal Variable;
21. Perceived Time
Spent on Local
Comunication
with Customers .17“ n.s. n.s. -.l7“ n.s. n.s. n.s.
22. Perceived under-
standing of
Offerings .23+ n.s. n.s. -.21“ n.s. n.s. n.s.
23. Perceived Import-
ance of Telephone
in Communicating
with Suppliers n.s. .23+ .23+ .18“ .26+ .26+ .27+
24. Perceived tine
Spent on Local
Communication .
with Suppliers n.s. -.l9“ .22+ n.s. -.21“ -.18“ -.20“

 

“Level of significance is equal to or less than .10.
+Level of significance is equal to or less than .05.
*Level of significance is equal to or less than .01.
n.s.The coefficient of correlation for the given variables is not significant at the .10 level.

130

Page Advertising in the manufacturer segment. Yet the
perceived amount of time spent, and the importance of
communicating with suppliers are not correlated highly with
the demand criteria. The opposite situation is true in
the wholesaler segment in which the factors dealing with
supplier communications behavior are more highly corre-
lated with the criterion variables than are those factors
associated with customer communication. This implies that
the customer aspect of a manufacturer's communication
behavior has greater impact on the demand for selected
telephone offerings than does the supplier component. In
the wholesaler segment the effects of these two components

are reversed.

Retailer Segment

Table 4.13 indicates that only the retailer segment
has a large number of explanatory variables correlated
with the advertising criterion at the .05 and .01 levels
of significance. The majority of these explanatory factors
can be categorized as "size" variables. The correlation
coefficients indicate a positive relationship between the
size of retail institutions and the demand for Yellow Page

Advertising space.

SIMPLE CORRELATION MATRIX FOR SELECTED EXPLANATORY AND

131

TABLE 4.13

CRITERION VARIABLES

- RETAILER SEGMENT

 

Criterion Variables

 

Explanatory Advertising Total Intrastate Interstate Equipment Telephone Telephone
Variables $ T011 5 T011 5 T011 5 5 Lines Locations
Size Variables
1. No. Buildings .22“ n.s. n.s. n.s. .28+ .27+ .23“
2. Facility Sq.Pootage .66“ .60“ .51“ .60“ .91“ .89“ .90“
. No. Offices .62“ .58“ .65“ .44“ .76“ .74“ .69“
4. 1967 Gross Sales .31+ .20“ .25“ n.s. .42“ .35+ .34+
5. No. Employees .64“ .75“ .63“ .75“ .97“ .94“ .97“
6. Salaried Employees .62“ .48“ .50“ .40“ .74“ .68“ .67“
7. Sales Clerks .61“ .76“ .63“ .76“ .94“ .90“ .94“
8. No. Suppliers .42“ .53“ .40“ .57“ .82“ .85“ .88“
Geoggaphic Variables
9. % Sales in
Lansing Area n.s. -.31+ -.45“ n.s. n.s. n.s. n.s.
10. No. Suppliers Out-
side Lansing .42“ .54“ .40“ .57“ .82“ .86“ .88“
11. No. Suppliers Out-
side Michigan .44“ .56“ .43“ .60“ n.s. n.s. .90“
12. % Sales in Mich.
Outside Lansing n.s. .27+ .47“ n.s. n.s. n.s. n.s.
13. No. Suppliers in
Michigan n.s. n.s. n.s. n.s. .33+ .41“ .40“
14. x Suppliers in
Lansing .21“ n.s. n.s. n.s. n.s. n.s. n.s.
15. lo. Outlets in
Michigan Out- .
side Lansing n.s. n.s. .23“ n.s. n.s. n.s. n.s.
Attitudinal Variable!
16. Perceived Time
Spent on local
Communication
with Customers n.s. n.s. n.s. n.s. .36“ .37“ .35+
17. Perceived under-
standing of
Offerings n.s. n.s. n.s. .22“ n.s. n.s. n.s.
18. Perceived Import-
ance of local
Telephone Communi-
cations with
Suppliers .30+ n.s. n.s. n.s. .30+ .28+ .31+
19. Perceived Import-
ance of Telephone
in Attracting New
Customers -.22“ n.s. n.s. n.s. n.s. n.s. n.s.

 

“Level of significance is equal to or less than .10.
+Level of significance is equal to or less than .05.
“Level of significance is equal to or less than .01.
n.s.The coefficient of correlation for the given variables is not significant at the .10 level.

132

The remainder of the dependent or criterion variables
are also correlated most significantly with "size" variables.
This is most obvious for the dependent variable "interstate
toll dollars" where seven of the ten explanatory variables
significantly correlated with the criterion can be
categorized as size factors. Also, the low mean monthly
dollar amount spent on interstate calls by firms in the
retailer segment ($47.00) as compared to the amount spent
on interstate calls by firms in the manufacturer and whole-
saler segments (see Table 4.5) suggests that fewer factors
cause a variance in interstate toll usage for the retailer
than is the case for the manufacturer or wholesaler. For
example, since retailers have less geographical variation
in sales than do manufacturers or wholesalers, any variables
associated with geographic sales patterns have little, if
any, effect on the retailers' toll usage.

Finally, a significant positive correlation does
exist between the amount of time a retailer perceives him—
self to use in communicating with customers and the amount
of telephone equipment he rents. In other words, the
higher the retailer's ranking of the amount of time spent

communicating with customers, the greater his demand for

133

telephone equipment.

Realtor and Dry Cleaner Segments

Simple correlation matrices for the realtor and dry
cleaner segments are presented in Tables 4.14 and 4.15. For
these segments, a maximum of three explanatory variables
are correlated significantly with any one criterion variable.
For the dry cleaner segment, no independent variable
was correlated significantly with either the amount of
Yellow Page Advertising used or the amount of toll usage
demanded. The relatively small sample sizes obtained for
these two segments and the relative homogeneity of firms
in the realtor and dry cleaner categories undoubtedly
contribute to the lack of significantly correlated factors.

Some General Comments and Comparisons on the
Correlation Matrices

 

Since the majority of the explanatory factors are
associated in some way with the size of the firms, these
factors are, in general, positively correlated with the
demand variables. There are, however, several interesting
exceptions. For the manufacturer, wholesaler, and retailer
segments, the independent variable "per cent of sales in

the Greater Lansing Area” is negatively correlated with the

4
2.
l

.Ho>oa on. on» us uceuwmacman no: ms3 moannwus> cu>fim on» How codusaeuuoo mo ucoaoammeou oga.u.s
.Ho. sen» mama so on assoc ea secondmesmam mo Ho>eAe
.mo. can» mesa no ou Hence an mossoamwcmwn mo Ho>eA+

.oa. can» used no an anode

.......

ma mocmuwmwsmfim mo He>eqo

 

 

.ma. «so. .hn. «mu. .su. .Hm. .u.: nomsoadem .02 .n
«Nm. eom. eem. +mv. 9mm. +Nv. .m.: modem nacho hme .N
«mm. tom. «mm. «mm. +~¢. ems. +me. neuwmuo .02 .H
unseumooq mummm w. .w,HHoe_. whamme w mace w mummmmmmw
«cosmeama econamaoe usoemwsum ousumumucH ounuumuusH Hence mewmfiuue>ud muoueomamxm

 

moafimaum> cofiumuwuo

Bzgwmm NOEASNM I mamfimg ZOHMHBHQU

vH.¢ Mame?

. Quad ngflgmxm Qavmqmw mom gnu—”mag BHBGHHMMOU ”Hm—SH»...

 

135

 

 

 

 

 

.Hm>ma 0H. can as “snowmacmfie poo um3 meanswus> cu>am on» How coausaouuoo mo ucuaudmmeoo s:a.a.:
.oa. can» noueeum no on Hoses ma ooseoamacmae mo He>ema
.mo. can» Heueeum no on Hence we unconfinecmae mo Ho>eq+
.oa. can» Houmeum no 0» Hedge ow unconfinecmwm MO H0>0Ao
cue. .n.s ace. .n.:. .n.: .n.: .n.: soaumuomd
. ca muse» .02 .m
+mm. «on. «an. .n.: .n.: .n.: .n.: moohoamfim .02 .N
*mo. id“. *0“. Cu.“ Ono: 0".“ Cu.: amuammo .02 .H
ecoaumooq emcee w, ..... m.HHbP. .. .w_HHos w Haoa m noannaue>
encompass esocawaea assemwsum ousuuumusH eununmuucH Hones uaaefluue>u< huouessamxfl

 

 

 

moan—mane, .Bfimfiuo
g

 

“tit

Bzmzwflm mmzmmflv Mun I mmfldem¢5 ZOHMHBHKU
924 ”mom—.4258 Dmfiumfimm mom NHMBNS ZOHBSQMMOU EgHm

mH.¢ mflmdfi

136

1 This

amount of toll usage - both inter and intrastate.
verifies the hypothesis that as the per cent of total
sales occurring in the local calling area increases, toll
usage declines. Surprisingly, this same inverse relation-
ship exists between the variable "per cent of sales in

the Greater Lansing Area" and the criterion variables
"equipment dollars,” "number of lines," and "number of
telephone lOcation," but only for the manufacturer and
wholesaler segments. Thus, a manufacturer or wholesaler
selling primarily in the local calling area not only has
lower toll billings than a firm selling primarily out of
state, but also demands a lesser amount of telephone
equipment. As was mentioned, this was not the case for
the retailer segment. Here the demand for equipment was
correlated positively with the per cent of sales occurring

in the local calling area.2

 

1The variable "per cent of sales in the Greater Lansing
Area" has no meaning for the realtor and dry cleaner segments
since nearly all firms sampled in these two segments had no
sales outside of this area. Also, although no significant
correlation is indicated between "per cent of sales in the
Greater Lansing Area" and the interstate component of toll
billings in the retailer segment, the correlation was posi—
tive and was significant at the .15 level.

2The correlations between "per cent of sales in the
Greater Lansing area" and the demand for equipment in the
retailer segment were not significant at the .10 level and are
not shown in Table 4.13. However, the correlation coefficients
were positive and were significant at the .20 level.

137

The findings presented thus far in the chapter
suggest several basic conclusions regarding the effect of
geographic sales patterns on the demand for telephone
offerings in the retailer, manufacturer and wholesaler
segments.l

First, toll billings for a manufacturer or whole—
saler are a positive linear function of the amount of sales
taking place outside the local calling area. Since a very
small portion of a retailer's total sales are made outside
of his local calling area, geographic sales variation has
an insignificant effect on the retailer's toll billings.

Second, the amount of equipment, particularly the
number of telephone lines and the number of telephone
locations a manufacturer or wholesaler utilizes, is also
a positive, linear function of the amount of sales occurring
outside the local calling area. However, as was the case
with toll usage, the retailer's demand for equipment does

not appear to be affected by geographical sales variation

and, in fact, seems to be related only weakly to sales in

 

1The tentative conclusions derived here only apply
to firms of the type and size studied in this paper. How—
ever, these findings provide a basis for develping a
broader research design and testing more general hypotheses.

138

general.

A linear, positive relationship also exists between
the number of suppliers a manufacturer and retailer have
outside the local calling area and the firm's demand for
various telephone offerings. This same relationship does
not exist for firms in the wholesaler segment, however,
as indicated by Table 4.12. In fact, for firms in the
wholesaler segment, there is little or no correlation
between either the absolute number of suppliers, or the
location of suppliers, and most criterion factors. It is
difficult to draw any cause-effect implications from these
correlations. For example, it is hard to understand
why supplier factors should influence demand in the retailer
and manufacturer segment but have little effect on demand
in the wholesaler segment. A possible explanation for
this difference might be that, although wholesalers view
the use of the telephone in contacting suppliers as quite
important, the majority of communication with suppliers
is performed either through face-to-face contacts with
salesmen or through written correspondence.

One other correlation result that should be mentioned

here exists between the independent variable "understanding"

139

.m>onm Ho Hm>wH OH. wfiu pm mmHQMHHm> ooHHmuHHU map

runs @33

Imnuoo mHucmoHMHGmHm on on Odoom who? QUH£3 mmHQMHHm> ucmOGmmmOcH HHm mmOSHOGHe

 

m NH m OH NH mm HH ww NH mw
N NH m OH NH mm OH ww OH Ow
m NH m OH NH mm HH ww OH mw
O NH m OH OH mm mH ww mH mw
O NH m OH NH mm NH ww ON mw
O NH m OH HH mm NH ww HN mw
O NH H OH NH mm OH ww m mw

mGOHumooq
mcoflmmHmB

mmch woogmmHmB

w pomsmHsvm

w Haoe mumnmuwucH
w ance mumnmmunsH
m HHOB Hmuoe

w monHpuw>O4

 

emem Ompmmu a.mHm Umummp a.qu Omumwu e.me Umummu a.mHm Ompmmu

 

 

 

 

on on mmHQm on ou mmHnm on on mmHQm on on mmHQm on ou mmHQm
bosom IHHm> pcoom IHHm> pcoom IHHm> pcoom IHum> bosom IHHm>

.02 .02 .02 .oz .02 .oz .02 .oz .02 .oz
HmommHO mum HouHmom HwHHmumm HmHmmmHosz nonouommocmz

 

unmammm mmOCHmom

Bzmzwmm mmMZHmDm Mm l mmHdem¢> ZOHmmBHMU QMBUMHmm EBH3
Omedqmmmov NHHZdUHmHZOHm MMHm¢HMd§ BZMQZMWMQZH m0 mmmzbz

0H.w mqmwfi

mHanum>
COHumuHHU

140

and the different criterion variables. "Understanding"

is used here to denote the customers' perceived understanding
of the kinds of equipment and services offered by Michigan
Bell. The customer responded on a scale from one to five,
"1" meaning he felt his understanding of telephone offerings
was excellent and "5” meaning the customer felt he had a
very poor understanding of theixflephone products and
servflxxsavailable. No measure was made of the respondents
actual understanding of the alternatives available but

only his perceived understanding or knowledge. Nevertheless,
it is interesting to note that a negative correlation

exists between the variable "understanding" and nearly all
criterion variables for all segments. In other words, the
demand for and usage of telephone offerings varies directly
with managements' perceived understanding or knowledge of
telephone offerings available.1 If managements' actual
understanding coincides with their perceived understanding,

then it might be inferred that a firm's demand for various

 

1To determine whether some of the variation in the
variable "understanding" was perhaps due to some other
independent variable, all intercorrelations were examined.
No significant correlations between “understanding" and any
other independent factors were found at the .05 level.
Certain "size" variables, however, were found to be sig-
nificantly associated with the perceived understanding
factor at the .10 level.

141

offerings might increase if it were made more aware of the
total array of telephone offerings available to them.

The correlation matrices illustrated and discussed
in the preceding pages indicate that many of the criterion
variables under study are correlated significantly with a
large number of explanatory factors. However, as Farley

1 even

pointed out in his study dealing with brand loyalty,
excellent marginal relationships can often cause problems
for multivariate analysis. This is true because the
independent variables used in the analysis are often
highly correlated with one another. Thus, significant
multiple correlation coefficients should be produced if
all of the independent variables shown to be significant
in Tables 4.12 - 4.16 are used in a regression analysis.
Few of the regression coefficients, however, may be found
to be significant.

Actual regressions carried out for the manufacturer,
wholesaler, and retailer segments using all of the independent

variables found a maximum of five independent variable

coefficients significant at the .05 level for any one

 

lJohn U. Farley, "Why Does 'Brand Loyalty' Vary Over

Products?," Journal of MarketinggResearch (November, 1964),
p. 11.

 

142

equation. The signs of these coefficients also failed

to show any reasonable pattern, and multicollinearity was
suspected. Factor analysis was used to identify those
variables which contributed most to explaining the variance
in the criterion variables. Factor analysis was used

because the relationship between the total set of significant
variables can be observed simultaneously without encountering
the problems of multicollinearity that are present in the
data. This meant that a relatively "clean" subset of
independent variables could be identified and the pre-
dictive power and hypothesis testing ability of multiple
regression more accurately utilized.1

Sources of Variation in Demand Variables -
Factor Analysis

 

 

A number of factoring procedures have been used as
preliminary data—reduction techniques in multiple regression
problems. One method often used by researchers is to
summarize a large number of independent factors into a
few "factors” such that only a small amount of information

included in the large set of original variables is lost.

 

1William F. Massy, "Applying Factor Analysis to a
Specific Marketing Problem." Toward Scientific Marketing,
1963, American Marketing Association, pp. 295-296.

143

Only the independent variables are used to construct the
factor scores. The factor scores themselves are then
estimated using regression equations, and these estimated
factor scores are used to predict values of the criteria

or demand variables.l Although this technique may provide
the user with optimal results, the use of the factor scores
in the regressions may cause problems of interpretability.
This problem may be particularly acute when it is necessary
or beneficial for the user to interpret directly the effects
of various explanatory variables on the criterion. In

these instances, it may be beneficial to simply isolate

the key explanatory factors, use these in the predictive
equations, and sacrifice the additional information supplied
by the remaining independent variables.

An alternative which satisfies the criterion just
mentioned is to use factor analysis to isolate the actual
independent variables which account for the major share of
information included in the original larger set of data.

This approach was used by Twedt in his study of the variables

 

1For an example of this methodoloty, see: Farley,
pp, cit., pp. 11-12.

144

affecting advertising readership.l

Although some pre—
dictive accuracy is sacrificed, this technique does allow
the user to more easily interpret the effects of key
explanatory variables.

Based on this latter methodology, factor analysis
was performed on each criterion variable and the explanatory
variables which were correlated with each criterion factor
at the .20 level of significance.2 The principal components
extraction routine was used in the analysis with an eigen
value of one used as the limiting criteria. This meant
that only factors including variables which added five per
cent or more to the cumulative proportion of variance
were included in the principal components model. The
factor loadings were then rotated using the varimax criterion.

The varimax method has the greatest simplicity since the

factor loadings are made to approach either zero or

 

1Dik warren Twedt, "A Multiple Factor Analysis of
Advertising Readership," Journal of Applied Psychology,
Vol. XXXIII, No. 3 (June, 1952), pp. 207-215. Also for an
example and comparison of the use of both of the previously
mentioned Factor Analysis techniques on the same set of data.
see Farley, QB. gig.

2Factor analysis was not carried out for the independ-
ent variables correlated with the advertising criterion in
the manufacturer segment or for any variables in the realtor
and dry cleaner segments. There were too few explanatory
factors correlated significantly with the dependent variables
in these instances.

145

one.1 Variables exhibiting the largest absolute loadings
and that appeared to be both factorially pure and offer
the best chance for prediction purposes were chosen from
the rotated principal components for each criterion
variable.2
Tables showing the factor loadings of each independent
variable for all relevant criterion variables are given
in Appendix IV for each business segment. For explanatory
purposes, Table IV-I, which indicates the independent
variable loadings on the total toll criterion variable for
the manufacturer segment, is reproduced as Table 4.17.
The underlined values in the table represent those variables
which had their highest loadings on each of the five factors.
The squared factor loadings give the percentage of variance
for each variable that may be predicted by a given factor.

Since the dependent variable is listed first in each table,

the loadings given in the first row in each table when

 

1For a complete discussion of this methodolOgy, see
Harry Harman, Modern Factor Analysis (Chicago: University
of Chicago Press, 1960), pp. 301-308.

2It should be pointed out at this time that in addition

to multicollinearity checks, analysis was also made of a
large number of dependent-independent variable relationships
to determine whether any nonlinearities existed. Since no
nonlinearities were observed, the data were used in con-
structing only linear regression models.

146

squared indicate the percentage of variance in the
criterion variable that can be explained by each factor.
For example, in Table 4.17 the total toll criterion variable
has a loading of -.798 on Factor 1; therefore (-.798)2 = .64,
or 64% of the variance in manufacturer toll usage can be
explained or predicted from Factor 1. As is also indicated
by the last figure in Row 1 of Table 4.17, .858 or
approximately 86% of the variance in manufacturer toll
usage can be explained by all five factors. It should also
be noted, however, that the first two factors alone account
for approximately 74% of the variance in the dependent
variable. A review of the tables in Appendix IV shows
that in nearly all situations one or two factors account
for the majority of variance in the criterion variable.
Since the factor extraction was stopped at the fifth
factor, and because of the diverse nature and large number
of explanatory variables, not all of the loadings show any
pattern. For example, Factors 1 and 2 in Table 4.17 are
loaded heavily on a heterogeneous grouping of variables.
However, Factor 3 loads heavily on variables associated
with the firms relationship with its suppliers; Factor 4

is loaded most heavily on variables associated with the

147

TABLE 4.17

ROTATED FACTOR LOADINGS FOR TOTAL TOLL AND
ASSOCIATED EXPLANATORY VARIABLES - MANUFACTURER SEGMENT

 

 

 

 

Loading, Commu-
Variable Factor 1 Factor2 Factor3 Factor4 Factor 5 nality

Total toll -.798 .303 .262 .230 -.084 .858
Grossale 2* -.6l; -.012 .073 .361 -.021
Numoffices —.613' -.049 .157 -.074 -.266

Underst .666 -.007 .306 .245 .252
PctsaleL .626 .048 -.345 .200 .194
Numsalem -.46; .330 .460 -.157 -.364
Tloccust -.ggg .100 .281 —.209 -.377
Numordmo -.l72 .626_ .019 -.016 .009
Numbuild .006 .697 .499 -.194 -.018
Disthome .097 .4;6_ .010 .203 -.207
Numsuppl -.l46 -.056 fgll .056 -.179
Numsupp -.067 -.064 .666 -.039 -.148
Numsupp2 -.219 -.029 .22;_ .179 -.173

Sqfeet -.112 .256 .616

PctsuppL .387 -.079 -.62;_ -.l72 .157
NumoutB -.127 .043 .093 .222_ .045
NumoutA —.086 .049 .133 .22;_ .029

Pctsupm -.158 -.355 .228 -.616_ -.083
Grossale -.200 -.025 .051 .135 -.ggg
Grossalel -.227 -.034 .033 .151 -.905

Numempl -.092 -.010 .323 -.137 -.870
Numclrem -.190 .202 .253 -.082 -.fi
PctsaleM -.058 .349 .235 -.243 -.499

 

*For an explanation of abbreviated

Appendix IV.

variables,

see

148

location and number of branch facilities; the fifth factor
is most heavily loaded on the amount of sales of the firm
and the number of employees of various types employed

by the firm.

The factor loading matrices for the remaining
criterion variables in the manufacturer segment show the
same general pattern as those in Table 4.17 (see Tables IV.2 -
IV.6 in Appendix IV). The number of independent variables
used in each case differ, and the factors themselves vary
somewhat in their contribution to explaining the variance
in the criterion variables. However, the explanatory
variables do group themselves in nearly the same manner
for each variable.

Tables IV.7 - IV.12 in Appendix IV indicate that
for the wholesaler criterion variables, extraction was
stopped at either the second or third factors. A possible
explanation for the low number of factors in this case
is that fewer explanatory variables were correlated signi-
ficantly with the criterion variables in the wholesaler
segment than was the case in the manufacturer segment.

For both the two and three factor cases, the factor that

accounts for the greatest amount of variance is the factor

149

heavily weighted on firm size variables. (i.e., number
of employees, gross sales, and physical size of the
facility.) In the three factor cases, factor two is
loaded heavily on variables associated with the perceived
importance of suppliers to the firm's communication
behavior, While factor three loads heavily on geographic
sales variables. When extraction was stopped at only two
factors, the variables associated with supplier importance
were included in the factor emphasizing size variables,
while the second factor is weighted heavily on geographic
sales variables.

Finally, Tables IV.13 - IV.19 in Appendix IV present
the rotated factor loading matrices for the various
explanatory and criterion variables in the retailer segment.
A factor analysis was run on the criterion variable ”Yellow

Page Advertising dollars" and the associated explanatory

 

factors for the retailer segment only because of the large
number of explanatory variables that were correlated signi-
ficantly with the criterion in this segment.

As was the case for the manufacturer segment, extrac—
tion was stopped in each model at the fifth factor,

although the first two factors in each case account for

150

the largest proportion of variance in the dependent
variable. However, unlike the manufacturer and wholesaler
factor models, few of the factor loadings in this segment
show any logical relationships or pattern.

For the manufacturer, wholesaler, and retailer
segments, the rotated factor models were used to select
the initial variables used in the regression models. For
this purpose, the variables with the largest absolute
loadings on each factor and which seemed to be factorially
purest were found in the rotated factor loading models.
The independent variables selected in this manner for the
manufacturer, wholesaler and retailer segments are shown
in Table 4.18. As indicated in this table, the explanatory
variables that were chosen from the factor models differ
substantially for each criterion variable across the three
segments. Total yearly gross sales and the per cent of
sales occurring in the local calling area are the two
explanatory variables occurring most frequently. However,
they are only found in two business segments and thus
cannot be assumed to be good predictors across the total
population of firms.

There are, however, certain variables that tend to

be used consistently within certain segments for all

151.

.>H XHusemm< eon .ueHnnHue> nousH>ouAns mo cowumcsmee so Houses
.ueHnmaus> ucmucemounH pousHouuoo aHuceunchHn mo wanes: HHsBe on»
no monsoon nauseous some» :« :OHuouHHo manHuue>ue on» new use weakens no: ns3 anaHscs Hosanna:
uceemem HGHHmuom u m “newsman HonmOHogk I 3 “unmfimom Hensuummsnns I as

 

mmsnooHa
mmsuEH
Canoes!
musossz
eusOEsz

 

SmHmnuom
QOHemuom
NHmmsnEsz
HHmmsmEsz
Hmmomesz

 

X34><

>4¥§N
><¥=K

caduOEsz
NeHmmmouu
eHmmmouO
exueHossz
EmuHOEsz

 

X

K

N

N N
N N N

EstmEsz
HmEeEsz
summon
MMOESZ
seepHHsnfisz

 

m 3 E

m ><#:N
X

3 Z

M ><¥:N

.3

K N
N X
.l m 3. z

m .3 s m .3 z m «:3, sum

 

chHumooq
econmmHoa

mmcHA
oCOSQOHmB

S
w uCOEQHnum W HHOB

eumuuuoucH

. alaaoe w.awoe_anuoe w aceuauum>ea
ousuumuucH

 

«nucwfimom mnmCHnsm an anmHum> unaccomoa

 

eHAMHum>
unsucmmmusH

 

BZMZUQW mmeHmDm 92¢ mammHm§> azmnzmmmn Hm
mamQOZ OZHQéOH MOBU¢W Qmfidfiom WEB 2H mOBU¢fi EU¢H
ZO UZHEQH memUHm 02H>§ mmqmfimg MmOdesmxm

mH.w mam¢fi

152

criterion variables. For example, the number of buildings
and number of offices a firm maintains are found to carry
high weights across all criterion variables in the retailer
segment. Yet, these variables load highest on only two
factors for criterion variables in the wholesaler and
manufacturer segments.

In general, the summary group of explanatory variables
derived from the factor analysis looks quite reasonable.
In nearly all cases, the explanatory variable that carried
the highest simple correlation with each criterion variable
is included. Also, the method used to select the summary
group of explanatory variables appears to have minimized
the degree of collinearity or correlation among independent
variables. Whether this procedure has also led to the
best choice of predictive variables will be determined in
Chapter V. The results of the regressions using the
variables derived from the factor analysis will be compared

with those using other selection criteria.

Summary

A detailed description of many of the character-
istics of responding firms was presented. Special emphasis

was given to the way in which the structural or Classificatory

153

variables and specific attitudinal responses differed
across the five business segments.

Simple correlations between statistically significant
explanatory and criterion variables were then derived. In
many cases, a large number of explanatory factors were
significantly correlated with the dependent variables.

For these situations, factor analysis was used to identify
the independent variables that best summarized the
information included in the larger set of collinear variables.

Based upon the results generated from the factor
analysis, variables were chosen that had high factor
loadings and which seemed to be factorially pure. By
this procedure a group of explanatory variables that
appeared to be minimally collinear were developed for each
criterion variable. These explanatory variables will
serve as initial inputs to the regression models that are

developed in the following chapter.

CHAPTER V

DEMAND MODELS

In Chapter IV, a large number of factors were
determined to be highly correlated with a firm's demand
for telephone offerings. The purpose of this chapter is
twofold: first, to examine in detail the variation in
factor effects across firms of similar types and different
business segments; second, to develop predictive models
from these factors that can be used to forecast the demand
behavior of firms for specific telephone communication

offerings.

Sources of Demand Variation - Regression Models

One of the basic problems inherent in the use of
regression models is selecting what might be called the
"best” regression equation from all possible alternatives.
This difficulty arises for two primary reasons. First,
there seems to be no universally used criterion or
standard for rating one regression equation better than

another. Second, there are a number of judgmental or

154

155

qualitative decisions that often have to be made regarding
the inclusion or deletion of certain explanatory variables
in regression models. In some cases, the criterion used
to select the ”best" set of explanatory variables may
conflict with the criterion used to select the best
regression equation. For example, in referring to the
selection of variables to be included in a regression
equation, Draper and Smith state:
1. To make the equation useful for predictive

purposes we should want our model to include

as many x's (independent variables) as possible

so that reliable fitted values can be determined.

2. Because of the costs involved in obtaining

information on a large number of x's and

subsequently monitoring them, we should like

the equation to include as few x's as possible.
One regression equation may satisfy a specific prediction
criterion better than any other; however, certain qualita—
tive or quantitative constraints on the independent variables
may lead to the selection of a model which is less than
Optimum in satisfying the predictability criterion.

There is no one criterion available for selecting

the proper regression model: yet, there are a number of

 

1Norman Draper and Harry Smith, Applied Regression
Analysis (New York: John Wiley and Sons, Inc., 1966), p. 163.

 

156

procedures available to assist the researcher in selecting
the "best" regression equation.1 One of the most commonly
used techniques is the stepwise regression method in which
variables are added or "stepped" into the regression until
some "satisfactory" model is achieved. Often the co-efficient
of determination (R2), which relates the prOportion of
variation in the dependent variable that is associated
with variation in the independent or explanatory variables,
is used as the primary judgment criteria. In other cases,
predictor or independent variables are either added or
deleted from the equation and a partial F criterion used
to test their individual contributions to the model. Based
on some preselected significance criterion, only variables
that provide a significant contribution are then retained
in the final model.

Another statistic sometimes used by researchers to
compare predicted with actual results is that developed by

Theil.2 This statistic, termed Theil's "U,"3 has been used

lIbid.. pp. 163-177.

2H. Theil, Economic Forecasts and Poligy (Amsterdam:
North-Holland Publishing Company, 1958), pp. 31-42.

 

 

 

 

3
l
J; (Pi - Ai)2 Where Pi = prediction and
U =ef=;__. 1 Ai = the corresponding
\l-r: Pi2 4"; Ai2 actual outcomes

for each subject.

157

to evaluate the forecasting performance of regression
models. As shown in footnote 3 (page 156), this inequality
coefficient takes on values between 0 and l. A 0 value
indicates perfect forecasting and values approaching 1
indicate poor forecasting performance. This coefficient
was calculated for a number of the forecasting models
presented later in this chapter. The predicting equation
selected based on this criteria was no different than when
the R2 statistic was used to choose a best regression model.

The problem that sometimes results when using these
and other statistics as the sole evaluating criterion is
that high correlations do not necessarily indicate good
predictor equations. Therefore, although statistical
criteria provide valuable information, a complete reliance
on them as the only judgment mechanism may lead to the
selection of regression models that do not provide optimum
forecasting performance.

Attempts have been made to determine the criterion
which can be used to select optimal prediction equations;1

yet "it appears that there is no single general criterion,

 

1For example, see "The Predictive Accuracy of Empirical
Demand Analysis," pp, cit., pp. 550-575.

158

TABLE 5.1

EXPLANATION OF INDEPENDENT VARIABLE SYMBOLS
USED IN THE CHAPTER V TABLES

 

 

 

§ymbol Variable Description

XSf Square footage of customer facility

XO Number of offices

XC Number of clerks employed by Retailer

X501 Number of suppliers outside of Lansing x 100
X%sl Percentage of sales in Lansing area

X%sm Percentage of sales in Michigan outside Lansing
XS Number of suppliers

Xsom Number of suppliers outside of Michigan x 100
Xce Number of clerical employees

Xse Number of salaried employees

Xgom 1967 Gross sales outside of Michigan x 100
Xom Number of orders taken per month

Xne Total number of employees

Xgol 1967 Gross sales outside of Lansing x 100

ng Total 1967 Gross sales

xoom Number of outlets outside of Michigan

X Number of years in operation

 

159

the rule being rather that ad hoc tests of predictive
performance grow out of the logic of the particular

"1 In other words, any decision about a

investigation.
best regression model should depend on the primary objectives
of the research. Where prediction is the main purpose of

the model, the best way to arrive at an optimum regression
equation is to use forecasting performance itself as the
primary indicator of regression quality. Sensible judgment
is still needed in the selection of independent variables
(i.e., where two independent variables are highly corre-
lated, the choice of one may rest primarily on ease of
measurement). However, a "critical examination of the

"2 should accompany

model through examination of residuals
any selection procedure. Only in this way can various

discrepancies in the model be uncovered and a "best"

regression model developed.

Preliminary Regression Results
Initially, the independent variables obtained from

the factor analysis models in Chapter IV3 were used as the

 

lPalda, pp, cit., p. 287.
2Draper and Smith, pp, cit., p. 172.

 

3For a list of these variables, see Chapter IV, p. 151.

160

explanatory variables in the regression equations. The
R2 values obtained by correlating these variables with
the demand criteria are shown in Table 5.2. The high
values obtained for the coefficient of determination, R2,
for many of the demand equations indicate that a large
proportion of the variation in the dependent variables is
associated with the variation in the explanatory variables.
This is especially true for the demand equations in the
wholesaler and retailer segments Where over half of the
regression equations have R2 values greater than .90.
Three major areas should be considered, however,

2 values are synonymous with

prior to assuming that high R
good predictor equations. In particular, several factors
should be investigated before it is assumed that the
equations presented in Table 5.2 are the best possible
regression models. First, the effect of deleting one or
more independent variables from the regression models on
certain predictive criterion should be examined. Second,
the possibility that variables not included in the original
explanatory variable set may improve thegxedictability and

accuracy of the equations should be tested. Finally, the

assumption that the relationships between predictor and

161

.HmpoE Conmemmu wflu CH poms mOHQmHum> uCopCmmmOCH mo HmQECC u 2
.mmem mwmflu CHmHouomm
OCmEmO any ACHB pouMHmHHoo mHuCMOHMHCmHm mmHQmHHm> “COOCmmmUCH mo HmQECC 30H 03p
mo mmsmomn mquEmmm MmHmmmHoC3 OCm HousuommsCme mfiu CH CoHkuHHO OCHmHuum>pm may
How so muCoEmmm HowwHO New Ho HOpHmmn map CH Owns COC mm3 mHmmHMCd HouommN
.ma.v magma .>H umummno
mom .mCOHmmmHmmH omega CH poms mmHQmHum> quOCQOOCH mnu mo umHH m mom

 

 

 

 

 

 

 

 

 

H
m mm. M mm. m mm. mCoHpmooq mConmonB mo HmQECz
m wm. N wm. w NO. mOCHH mConmmHoe mo HmQEsz
m mm. m mm. m an. mmcHHHHm ucmsmnsom
m as. m mm. m on. momma Heoe mumumumucH
m as. m mm. m mo. mmmmo Haoe mnmnmmuncH
m we. m mm. m mm. momma Haoe Hmuoe
m wm. II II II II mHsquCmmxm OCHmHuHm>O¢
z m z m z m mHQmHHm>
N N m N
HmHHmumm HmHmmeo£3 Honouommosz OCmEmQ
quEmwm mmmCHmsm HQ CoHHMCHEHmme mo quHonmmoo
Bzmzomm mmmszDm Hm I qumaoz moeomm 20mm Omaomqmm HmmAmCHmm> MmOB¢Z¢qum mme

ZO NHmMBHMU QZNSHQ m0 mZOHmmmmOmm mom ZOHBNZHSMMBWD m0 mBZmHOHmmmOU

N.m WHmdfi

162

criteria factors are best described by simple linear

models should also be examined.

Variable Addition and Deletion

Assuming that the independent variables selected
from the factor analysis models were the best set of
predictor variables available, the first step taken to
improve the regression results was to examine the effect
of reducing the number of predictor variables used. In
other words, the R2 values presented in Table 5.2 were
obtained from regressions using all of the variables
selected from the factor models. Better prediction models
might result, however, if fewer explanatory variables are

l A number of criteria were

used in the regression models.
established to guide the deletion decision. First, I
decided to obtain regressions on all combinations of the

explanatory variables derived from the factor models.2 Then,

starting with one independent variable, an additional

 

l"Better" is used here to mean models that sacrifice
a certain amount of statistical precision for the sake of
using fewer, more easily measureable explanatory variables.

2Since there were never more than five factors in
the factor analysis models, and thus, no more than five
independent variables, the "all combinations" approach
was feasible at this point of the analysis.

163

variable was included in the model only when it met the
following criteria:
1. Its introduction to the model (a) increased the
value of the coefficient of multiple determination
(R2) by at least .05 and/or (b) decreased the
value of the standard error of the estimate (Syc)
by at least 10%. Although there is a relation—
ship between Syc2 and R2, a substantial decrease
in one does not necessarily imply a large change
in the other. Therefore, both are included as
judgment criteria. Also, an analysis of the
residual mean square (Sycz) as the number of
independent variables in the regression is
increased helps to determine a best cutoff point
for the number of variables included in a regression.
However, this procedure does not necessarily
guarantee that the right set of explanatory
variables has been chosen.1
2. If two independent variables were highly correlated
and contributed an equal amount to explaining

the variance in the dependent variable (as

 

1For a more complete discussion of this point, see
Draper and Smith, pp. cit., pp. 165-167.

164

indicated by the increase in R2 resulting from
separately adding each variable to the model),
the independent variable that was most easily

measureable was included.

Based on the above criteria, regressions including
all combinations of the factorially selected explanatory
variables were run and best regression models chosen.
Table 5.3 lists the variables included in these regression
models and the amount of variance in the demand variables
that is explained by the reduced set of explanatory
factors.1

Comparing the results presented in Table 5.3 with
those of Table 5.2, one sees that the regression equations
of Table 5.3 generally have fewer explanatory variables
than the basic models presented in Table 5.2. In few
instances did a substantial decrease in the correlation

coefficient, or a substantial increase in the standard error

of the estimate, result when one independent variable was

 

_1Note that equations are not presented for the adver-
tising criterion in the manufacturer or wholesaler segment,
or for any dependent variables in the realtor or dry cleaner
segments. Factor analysis models were not used to generate
the initial selection of independent variables in these cases.
Regression models are developed for these dependent variables,
however, and will be discussed later in this chapter.

165

.mHOQENm mHQmHHm> mmwflu mo CoHuMCMmem Cm How H.m mHQme mmmm

.Cmmoso mm3 meanHm> NHQCMCmmem

no new poospmu mHCu 30: mo COHumHHommp m Com

.~.m magma .m muocuoom mmmm
.> Hmummso mo wLmOH .mm mom

 

 

 

 

 

H

U Ewnxu . mm mm
x wm. x x wm. N we. mCoHumooq mConmmHoe mo HmQEsz
ox.ox mm. mmx Hm. sex.mx om. macho muonmmame mo umnasz
ox ox mmx mm. mex max mm. mmx.ox Hm. mmcHHHHm ucmsmhswm
mox mm. HmXX.mmx mo. EomX.on mm. mmmwb HHOB manpmumpCH

o . o . on .80
smax ox so. amax max mm. H x x mm. momma Haas mumummnucH

HmR .o . mm . Eo .Eom . m

0X X mo N 00 NM N mm 0 mmD HHOB HMHOB
mex.mCx.ox Hm. II II II II ousuHOCmmxm OCHmHuHm>O¢
muouoHpmum Nm muouoHOwum Nm muouOHpmum Nm JIWHQMHHC>
HmHHmuom umHmmmHo£3 Honouommosz OCMEOQ

 

quEmwm mmmCHmCm

 

Bzmzwmm mWMZHmDm Nm

NmHmQOZ mOBU<m 20mm QmBUMHmm HmMHdem¢> MmOBmZ¢meM m0 8mm
QMUDQmm ¢ 20 flHmMBHMU QZNSMQ m0 mZOHmmmmwmm mom ZOHBNZHSMMEMQ ho BZMHUHhmmOU

m.m MHmNH

166

deleted from the original regression models. In fact,
in some instances, as many as four explanatory variables
were deleted from regressions that originally included
five independent variables, with little noticeable effect
on the R2 or Syc statistics. For example, when four of
the five explanatory variables chosen from the factor
models to explain the variation in manufacturer demand
for telephone locations were dropped from the regression
equation, the R2 statistic fell from .65 to only .64. The
standard error of the estimate increased by only 2%.
Thus, even though many of the factor models developed in
Chapter IV contained four, five, or six factors, only the
variables that loaded highest on the two or three top
loading factors contributed significantly to the regression
models. Once the important variables had been included
in a demand model, both the amount of variance explained
(R2) and the residual mean square (Sycz) for the model
tended to stabilize.

The regression coefficients for each of the demand
equations summarized in Table 5.3 were also checked for
significance. In all cases the coefficients are statisti-

cally significant at the .05 level. In fact, many of the

167

coefficients are significant at the .01 and .001 levels.
The coefficients also carry the expected signs, indicating
that multicollinearity is not in direct evidence. This
finding was further verified by a check of the Durban
Watson statistic for each equation. This test indicated
no serial correlation in any model at the .05 level of
significance.
Contribution of Additional Explanatory Variables
to the Demand Models

Using the predicting equations presented in Table 5.3
as a base, all other independent variables that were
correlated significantly with the criteria variables were
individually "stepped" into these basic models. The
effect that the addition of each of these variables had

on the predicting models was then determined by using

1 2

the sequential F-test criterion and the R and standard
error criterion discussed on page156. This procedure led
to the inclusion of additional predictor variables in the
basic models of Table 5.3. In some instances, the

remaining factor analysis based variables were replaced

 

1For a discussion of the uses of this technique in
regression analysis, see Draper and Smith, pp, cit.,
pp. 71-72.

168

with a new set of predictor variables. The resulting
final regression models based on these criteria are pre—
sented in Tables 5.4 — 5.8. Although much of the remainder
of this chapter will concentrate on reviewing the details
of these models for each business segment, several general
points will be made first concerning the differences among
the various demand models developed thus far.

First, there are only three cases in which the
final predictive models are the same as those presented
in Table 5.3; and there is only one case (the total toll
model for the wholesaler segment) where none of the
explanatory variables derived from the factor models were
used in the final regression equations. This indicates
that the factor analysis technique was a useful methodology
for arriving at a basic set of explanatory variables.
As mentioned previously, however, only the variables chosen
from the two or three highest loading factors provide a
significant amount of information to the demand models.

Second, the variables used to forecast the same
demand factors vary substantially across the business
segments. There were some instances in which a single

independent variable had a significant effect on a demand

169

criterion across the three major segments (i.e., the
independent variable "number of offices" is included in
the "line" and "location" demand models for each of the
three major segments). However, the differences that
exist in the demand equations indicate that the business
segmentation strategy used in this study is an effective
requisite for developing accurate demand models. In fact,
it could be argued that a finer breakdown of the business
segments than used here would result in even more precise
predicting models.

Finally, the relatively few number of explanatory
variables used in all of the demand models indicates that
there may be only a few factors influencing the demand
for many types of telephone offerings. The regression
technique used here does not prove a cause-effect relation-
ship. But the evidence supplied by the demand models
suggests that a small number of structural variables
explain a large share of the variance in individual
business firm demand for many of the criteria factors.

It is feasible that adding other variables to the
models might increase their predictive accuracy. To check

this possibility, regressions of each demand criterion were

170

run on p;;_explanatory variables that were correlated
significantly with the criterion factors.1 This analysis
indicated that the inclusion of a large number of additional
explanatory factors did not increase significantly the
correlation coefficients or significantly decrease the
standard error of the estimate for any demand model. Even
when the value of R was increased, the reduction in the
residual sum of squares was not great enough to improve
the precision of the demand models. For an increase in
the value of R resulting from the addition of more explana-
tory variables does not lead to better predicting models
when the reduction in the residual sum of squares is less
than the original residual mean square.2 An examination
of the regression outputs showed this to be the case for
the demand models that included a large number of predictor
variables.

Including all of the independent variables for which
data was obtained in the regression models did not signi-

ficantly affect the predictive performance of the models.

 

1For a list of the independent variables that are
correlated significantly with each criterion at the .05,
.01 and .001 levels, see Tables 4.11 - 4.15.

2See Draper and Smith, pp. cit., pp. 117—118.

171

The possibility does remain that a number of relevant
explanatory factors were not included in the analysis.

For cases such as the Yellow Page Advertising models,
where predictive performance is quite poor, it is possible
that information on additional explanatory criteria would
improve the demand models substantially. The types of
information that might be added to the demand models to
improve their forecasting performance will be discussed
later in this chapter when the specific demand models are
examined, and in Chapter VI under "Suggestions for Further

Research."

Yellow Page Advertising_Demand Models

When compared to the other demand models, the linear
regressions for the Yellow Page Advertising criterion show
very poor prediction results for all but the retailer
segment. In fact, even when all variables that are
correlated significantly with the advertising criterion
are used in the regressions, the multiple correlation
coefficients and standard errors are not changed significantly
from those appearing in Table 5.4.

The Yellow Page Advertising regression model for the

retailer segment shows the best predictors to be the number

1I72

 

 

TABLE 5.4
PREDICTING EQUATIONS FOR YELW PAGE ADVERTISING USAGE
BY BUSINESS SEGMENT]:
Business “_
d d
Spgpent Stan ar /Normalized Equations R n
1 (c) (a) (C)
Retailer 23 ' - 42.6 + 9.9 x 10"4 x8f + 4.8xO .71 42
2.
(a) (a)
Manufacturer YA - 11.4 + .2X%.1 .28 47
(a) (a) (a)
Wholesaler YA = 13.0 + .007xom + “2xxsl .32 62
YAN . .2 Ken '1’ '2x%1
(a) (a)
Realtor YA - 38.5 + 4.1xo .42 20
YAN - e4x°
Dry Cleaner YA 3 n.s.
YAN - 11.8.

 

 

aSignificant at the .05 level.

bSignificant at the .01 level.

cSignificant at the .001 level.

1°Month1y dollar amount of Yellow Page Advertising
subscribed to.

2

'Normalized Yellow Page Advertising predicting equation.

n.s. - No significant relationships were found between Y5 and any

combination of independent variables.

173

of offices the retailing establishment has and the square
footage of the retail facility. The normalized equation
indicates that the square footage variable was more important
than the number of offices variable in determining Yellow
Page Advertising demand. It should be mentioned here,
however, that because of the high correlation between a
number of size variables and the square footage factor,

a number of size variables could have been substituted

for the square footage variable with little effect on the
prediction results.

An examination of residuals in the retailer model
indicates that some nonlinear model might better fit the
data. To test this possibility, a quadratic model of
the form YA = a + bX + cX2 + e was fitted to the Yellow
Page demand data. A number of different size variables
were used as independent variables in this equation. But
a partial F-test showed that the data fit the linear
models as well, if not better, than it fit the quadratic
models. A number of other data transformations were
attempted, but none improved the predictive accuracy of

the basic linear models.1

 

1A variety of data transformations were also tried

for some of the other demand models, but in no case were
the results significantly better than the linear models pre—
sented in Tables 5.4 - 5.10.

174

For the Yellow Page models in the manufacturer,
wholesaler, realtor, and dry cleaner segments, the results
vary from a correlation coefficient of .46 (significant
at the .05 level) in the realtor segment to no significant
relationships in the dry cleaner segment. One interesting
relationship appearing in the manufacturer and wholesaler
models is that which exists between the independent variable
"percentage of sales occurring in the local calling area"
and Yellow Page usage. The positive nature of this
relationship indicates that the higher the percentage of
total sales occurring in the local calling area, the
greater the demand for Yellow Page Advertising space. One
wouki expect to find an even stronger relationship between
Yellow Page usage and the "absolute" amount of sales in
the local calling area. The fact that this relationship
is quite weak implies that, regardless of a manufacturer's
or wholesaler's size, the percentage of business obtained
from the local area and not the absolute amount of local
sales is the major influencing factor. Since the nature
of the retailers, realtors, and dry cleaners' business is
such that nearly all sales are made in the local areas,

the percentage of sales variable has little effect on

175

Yellow Page demand in these segments.

Several possible reasons can be given for the poor
results obtained for all but the retailer segment. First,
the large amount of Yellow Page demand variance not
explained in the manufacturer, wholesaler, realtor, and
dry cleaner segments may indicate that a number of critical
explanatory variables are not included in the models. One
factor that might be expected to influence demand is the
type of product(s) each business firm buys and sells. Some
information pertaining to this variable is available from
the firm's SIC classification, but there are not enough
firms in any one SIC category to draw any relevant statis-
tical inferences on the impact of this variable.

Another factor that may influence the demand for
Yellow Page space is the firm's perception of the degree
of competition it faces--particularly in the local calling
area. It is logical to hypothesize that the greater the
perceived competition, the more Yellow Page space a firm
would feel it necessary to utilize. Unfortunately,
although infbrmation was obtained on a firm's perceived
demand for Yellow Page space relative to competitors, no

data were collected on the perceived competition variable.

176

A third factor that likely influences the amount of
Yellow Page space used by various firms, is the number
of and time of contacts by Yellow Page salesmen. It
might be hypothesized that the amount of Yellow Page
Advertising subscribed to by a firm varies directly with
the number of recent contacts a Yellow Page salesman has
made on each firm, and the date of the most recent contact.
In other words, a firm that has been contacted in each
of the past five years concerning a possible increase in
Yellow Page advertising is likely to have a higher usage
value than a similar firm that has been contacted a lesser
number of times. Unfortunately, accurate data were not
available regarding this possible influencing factor.
Finally, an attempt was made to determine the
importance that owners or company executives place on
Yellow Page Advertising services. Any inference as to a
cause-effect relationship between the importance criteria

and Yellow Page usage, however, is extremely difficult to make.1

 

1In situations where an owner of a firm rated Yellow
Page Advertising extremely important, and the firm used a
large amount of advertising space, one would like to infer
that the importance factor caused the heavy demand. This
may not be the case, however, since a high importance
ranking may simply be a way of rationalizing a heavy usage
of Yellow Page Advertising.

177

Toll Forecasting Models

The models used to forecast toll usage are in general
better than the Yellow Page demand models. As Table 5.5
shows, demand equations for total toll in each of the
manufacturer, wholesaler, and retailer segments have R
values of .86 or higher, using a maximum of three explanatory
variables. Somewhat higher R values and somewhat lower
standard errors were obtained by including more predictive
factors, but the slight increases in predictive accuracy
that resulted did not justify their inclusion in the
models.

A comparison of the normalized total toll models
across the manufacturer-wholesaler-retailer segments shows
that different variables explain total toll usage for
each segment. The largest amount of variance in total
toll demand for the retailer group is explained by the
size variable "number of clerks." The geographic dispersion
of suppliers and sales also contributed significantly to
explaining the variance in retailer total toll: however,
the size factor was the most important explanatory variable.

In the manufacturer case the most important explana-

tory variable was "gross sales outside of Michigan." The

PREDICTING EQUATIONS FOR TOTAL TOLL USAGE

1f78

1381.8 5.5

BY BUS INESS SEGMENT

 

Business

 

Standard/ .
Segppnt Normalized Egpations R n
1 (C) (b) (c)
2.
Ym 1.57 Xe " .9 X801 - a4 Xxsl
(C) (C) (c)
Manufacturer YTT -2.1 + 12.2 x.e + .000001 xgom .90 47
YT“! '5 xse + '6 xgom
(C) (C) (C) (C)
Wholesaler YTT 91.8 + 5.2 xne + .0000007 x901 - 21.5 xse .98 62
Ym .3 k. + 1.0 X901 - e4 Xse
(C) (c)
Realtor YTT -13.3 + 6.7x“. .90 20
YTTII ’gxne
Dry Cleaner YTT n.s.
Ym n...

 

 

aSignificant at the .05 level.

bSignificant at the .01 level.

cSignificant at the .001 level.

l’Dollar value of total tell usage for February, 1968.

2Normalized total toll predicting equation.

n.s. I No significant relationships were found between YTT and any combination of

independent variables.

179

size variable "number of salaried employees" was the

only other factor that contributed any significant amount

of information to the model. It is somewhat surprising

in the manufacturer case that neither the number or location
of suppliers had a strong impact on manufacturer toll

usage. This would seem to bear out an earlier hypothesis
that customer telephone communication is more critical

to the manufacturer than is supplier communication, and

thus has more influence on toll usage.

For the Wholesaler segment, total toll usage is
most affected by the variable "gross sales outside of
Lansing." The "number of employees" and ”number of
salaried employees" factors are of secondary importance.
As in the case with manufacturers, no supplier factors
are present in the wholesaler total toll model. A
possible explanation is that, although supplier communica-
tion is important to the wholesaler, the majority of
supplier (manufacturer) contacts are either initiated by
the manufacturer or communicated in some other fashion.
This latter reasoning is borne out by the relatively
heavy weighting of the customer or sales factor on manu—

facturer toll usage. Unfortunately, no data were collected

180

on the questions of who originates the manufacturer-
wholesaler—retailer communications process most frequently,
or what amounts of various types of communication processes
are used by firms in each segment.

For the realtor segment, only one variable is
included in the total toll model. Other size factors were
significantly correlated with the total toll criterion,
but they were correlated highly with the employee factor.
Their addition to the model explained little of the
remaining variance. In the dry cleaner case, no explanatory
variables were correlated significantly with total toll
usage.

As might be expected, the models of inter- and
intrastate toll usage are similar to the total toll models
in the types of explanatory variables used. There are,
however, several basic differences between the models
shown in Table 5.5 and those of Tables 5.6 and 5.7. First,
the size variable "number of clerks" is still the most
important explanatory variable in the retailer segment
for the intrastate toll usage case. Yet, the "number of
clerical employees" (typists, bookkeepers, etc.) is the

only variable used in the interstate toll model. It is

PREDICTING EQUATIONS FOR INTRASTATE TOLL USAGE

181

TABLE 5.6

BY BUSINESS SEGMENT

 

Business

Standard/

 

 

 

§pgppnt Normalizedfiggpptions R n
1 (c) (c) (C) (C)
Retailer YI ’ I 7.1 + 3.2 xc + 3.1 x%‘m - .00045 x301 .84 42
2.
YIN - 1.5 Xe + .5 Xx," '“ 1.0 801
(C) (C) (C)
Manufacturer Y; I 12.6 + 6.3 )(.e + .002 x6“ .79 47
(C) (C) (C) (C)
Wholesaler Y! I 53.7 + 5.3 Xhe + .0000006 xxsm - 23.3 x.e .99 62
YIN - 0‘ X“. + 1.0 Xx.m - .5 Xse
(C) (C)
Realtor YI I -20.4 + 5.6 X“. .94 20
Ym " '9 ’91::
Dry Cleaner Y: I n.s.
YIN I n.s.
.Significant at the .05 level.
bSignificant.at the .01 level.
cSignificant at the .001 level.

l‘Dollar value of Intrastate toll usage for February, 1968.

2“Normalized Intrastate toll predicting equation.

n.s. I No significant relationships were found between YI and any combination of

independent variables.

182

TABLE 5.7

PREDICTING EQUATIONS FOR INTERSTATE TOLL USAGE
BY BUSINESS SEGMENT

Business

 

 

lgpgppnt Standard/gorm.alized Eqpptions R n
1. (C)
Retailer YO I 18.2 + 4.7 Xce .76 42
YONl' I .76 Xce
(c) (b) (c)
Manufacturer YO I -l4.4 + .0000008 xgom + 6.6 x3e .89 47
YON ' '7 xgom + '3 xse
(C) (b) (C) (C)
Wholesaler Yo I 29.9 + .00001Xg. + .0000006Xgom + 16.1)!com .84 62
You I “7xgs + .quom + .3X00m
(b) (a)
Realtor YO I 7.3 + 1.1x“. .53 20
'You I .ane
Dry Cleaner Yo I n.s.

Yw - he's

 

.Significant at the .05 level.
bSignificant at the .01 level.
cSignificant at the .001 level.
1°Dollar value of Interstate toll usage for February, 1968.
2“Normalized Interstate toll predicting equation.

n.s. I No significant relationships were found between Y0 and any combination
of independent variables.

183

not clear why this factor is correlated so highly with
interstate usage or why none of the geographical variables
contribute significantly to the explanation of interstate
toll usage variance.

For the manufacturer segment, there is only one
difference between the variables used in any of the toll
usage models. The explanatory factor "number of orders
taken per month" was substituted for the geographic
variable "gross sales outside of Michigan" in the intrastate
toll model. For the wholesaler model, the explanatory
factors varied considerably among toll models. Gross
sales outside of Lansing was the most important predictor
of total toll usage. The percentage of total sales made
to out-state Michigan customers and total gross sales are
the most critical explanatory factors in the intra- and
interstate toll models respectively. Also, employee
factors are more important in the total toll and intrastate
toll models than they are in the interstate toll case.

In general, each of the toll models includes at
least one "size" factor and, except for the retailer
interstate toll model, at least one "geographic" factor.

The importance of these factors, however, varied greatly

184

among models. It should also be stated that data con-
cerning the geographic sales and supplier variables were
collected on a more rigorous basis than the simple local,
instate, and out-of-state factors incorporated in the
models might imply. As well as a "local" and "in—state”
calling area, concentric circles of three different radii
were drawn on a U.S. map to represent more distant
suppliers or customers. The respondent was then asked to
state the percentage of suppliers and sales that came from
each of these mutually exclusive geographic areas. In all
cases, however, a simple local-calling area, in-state, and
out-state breakdown proved to be optimum in terms of
explaining variation in toll usage.

Finally, despite the fact that high correlation
coefficients were obtained for most of the toll usage
models, there are a number of factors that, if included
in the regressions, might lead to better prediction
results. First, as was discussed in the Yellow Page
Advertising section, a finer breakdown of business firms
according to the nature of products sold might lead to
more accurate forecasting equations. Second, a detailed

examination of the member of the channel of distribution

185

that originates a telephone communication under various
buy and sell situations should help to explain variations
in toll usage. Finally, more detailed data on the size,
importance, and location of both suppliers and customers
would likely prove valuable, especially where a business
user sells or buys a large percentage of its goods from

a small number of firms. Because of the nature of this
investigation, much of this information was impossible to
obtain via a mail questionnaire. It seems likely that an
in-depth personal interview with a number of individuals
in each firm would have to be used to obtain information

on these additional factors.

Expenditures for Equipment

 

Table 5.8 illustrates the regression models that are
the best predictors of the amount spent on telephone
equipment by firms in each of the five business segments.
Both the retailer and wholesaler regressions yield correla-
tion coefficients of .99, while the manufacturer, realtor,
and dry cleaner predicting equations are less precise.

As expected, variations in equipment demand are best
explained by size variables in all five segments. No

geographic factors are included in any of the models. The

PREDICTING EQUATIONS FOR TELEPHONE EQUIPMENT EXPENDITURES

1136

TABLE 5.8

BY BUSINESS SEGMENT

 

Business

 

Segment Standard/Norgglized Equations R 1;;
1 (C) (C) (C) (C)
Retailer YE ' -28.1 + .0036 x8, + 21.4 x0 + 5.7 xc .99 42
2.
(b) (C) (C)
Manufacturer YE 20.5 + 13.3 x0 + 6.0 x.. .75 47
(C) (c) (c) (C) (C)
Wholesaler YE 21.4 + 10.2 x0 + .00002ng + 3.5xne - 6.9)(." .99 62
Ym .1 x0 + ...,ng + e5&.e - 03x83
(C) (C)
YEN .95Xn.
(C) (a)

 

aSignificant at

.05 level.

”Significant at the .01 level.

cSignificant at the .001 level.

1-Monthly dollar amount of equipment billings.

2'Mormalised equipment billing predicting equation.

187

variable "number of offices" is common to all three major
segments but, as indicated by the normalized equations,

is not the most important variable in any model. The
normalized equations show that the number of clerks is the
most important explanatory variable in the retailer segment;
the number of salaried employees is the most important

in the manufacturer segment; and gross sales is most vital
for the wholesaler group. Because of the high correlation
between the predictor variables in the wholesaler model,

the coefficient for Xse carries a negative sign. Since

the elimination of this variable significantly decreases

the correlation coefficient, however, it is included in

the model. Because of high inter-correlations among pre-
dictors, the realtor and dry cleaner equipment expenditure
models contain, respectively, one and two explanatory
factors. In fact, this high correlation between predictors
in the dry cleaner segment causes the gross sales coefficient
to carry a negative sign. Since a significant decrease

in R results when only one of these variables is used, both

factors are included in the model.

188

Equipment Demand - Number of Lines and Locations

As was the case in the equipment expenditure models,
Tables 5.9 and 5.10 show high correlation coefficients
for both the number of lines and number of locations
regressions in each of the five business segments. The
number of offices a firm has is included in each of the
models used to forecast line demand for firms in the
manufacturer, wholesaler, retailer, and realtor segments.
But only in the manufacturer and realtor cases is this
factor most important. For the retailer case, the number
of clerks is again the most important predictor. Gross
sales is the most critical independent variable in the
wholesaler model and number of employees is most vital in
the dry cleaner model.

It is surprising to note that only in the realtor
case is the number of offices variable included in any
of the telephone location regression models. In the other
segments, variables related to the number of different
types of employees are the most important predictors of
this criterion. The number of clerks is most important
in the retailer group, number of salaried employees in

the manufacturer case, and the total number of employees

1E39

TABLE 5.9

PREDICTING EQUATIONS FOR NUMBER OF TELEPHONE LINES

BY BUSINESS SEGMENT

 

 

 

Business
Segment Standard/germalized Equations R n
1 (b) (C) (b) (C)
Retailer YT ‘ - 1.56 + .4 x0 + .l xc + .000001 KB .96 42
YTN2' - .3 x0 + .6 xc + .2 x8
(b) (C) (b) (C)
Manufacturer YT - 1.12 + .OO9X. + .3 x0 + .0002 xom .80 47
.3x. + .7 x0 + .3 xom
(C) (C) (b) (C)
Wholesaler YT - 2.0 + .0000001 x9. + .02 xne + .1x0 .93 62
YT“ - .5 x98 + .4 xne + .2xo
(C) (c)
Realtor YT - 1.4 + .5x0 .91 20
“m - .9xo
(a) (a) (b) (b)
Dry Cleaner YT I 1.4 + '032XY - .00003Xg. + .l4xn. .97 9
aSignificant at the .05 level.
bSignificant at the .01 level.
cslgnltlcant at the .001 level.

1.

Number of telephone lines or trunks used by a firm.

2oNorInalized telephone line predicting equation.

1530

TABLE 5.10

PREDICTING EQUATIONS FOR NUMBER OF TELEPHONE LOCATIONS
BY BUSINESS SEGMENT

 

 

Business
Segment Standard/Normalized quations R n
(C) (b) (c)
Retailer YL I 3.8 + 1.1 Xc - .006 x3 .99 42
(C) (a) (e)
Manufacturer YL I 2.2 + .5 xse + .0004 x801 .81 47
(C) (C) (c)
Wholesaler YL I -.S + .000001 X9. + .3 Xh. .99 62
(C) (C)
Realtor YL I -l.l + 2.2xo .90 20
Y“ - s9x0
(a) (a) (b) (a)
Dry Cleaner YL I 1.9 + .064)!y - .oooo4xg, + .17)!"e .91 9
Y” - e SXY - e 6x98 +' . legxrie

 

8Significant at the .05 level.
bSignificant at the .01 level.
cSignificant at the .001 level.
1mumber of telephone locations maintained by a firm.

2dfiormalized telephone locations predicting equation.

191

in the wholesaler and dry cleaner segments.

In the retailer case it is interesting to examine
the relationship between the demand for lines and locations
and the predictor variable number of clerks. Holding all
other factors constant, the retailer demand for lines
increases in approximately a 1:7 ratio with the number of
clerks, While the ratio between telephone locations and
clerks is approximately l:2. In other words, for every
additional seven clerks employed, one new line and
approximately 3.5 new telephone locations are added. One
should recognize that these values are not completely
accurate since many of the factors that are interrelated
with the clerk variable are not included in the line and
location demand models.

As was the case with the criterion factors discussed
earlier, the prediction equations very likely could be
improved by including data on a number of additional
explanatory variables. The inclusion of such factors as
the frequency of telephone representatives sales calls;
the time of the most recent salesman contact; management
attitudes towards the cost and value of telephone hardware;

a finer breakdown of the classes of business firms, and

192

the importance of the telephone in selling and buying
activities, could significantly improve the equipment
prediction models. As it was explained earlier, the nature
of the study prohibited the collection of detailed data
concerning most of these additional influencing factors.
Several general findings regarding the influence of certain
attitudinal factors that were measured are presented later

in this chapter.

Validation Sample Results

To avoid the problems that arise when the same
sample is used to both construct and verify a model, the
initial sample is sometimes divided into analysis and
validation sets.1 The model is derived based on the
analysis set and tested or validated on the second set of
respondents. Since there were a relatively small number
of firms that submitted completed questionnaires, it was
not possible to use a portion of these respondents as a
validating sample. However, in order to validate specific

aspects of the prediction models, ten non-respondents from

 

1For a more complete discussion on the rationale of
using this approach, see William F. Massy, e£_§l,, Purchasing
Behavior and Personal Attributes (Philadelphia: University
of Pennsylvania Press, 1968), pp. 9-10.

 

 

193

the manufacturer, wholesaler, and retailer segments were
selected at random. Telephone interviews were conducted
with individuals from these firms and data obtained on
selected dependent and independent variables. Data were
obtained on the demand variables total toll billings,
total equipment billings, and the number of lines utilized
for the month of July, 1968.1 These variables were chosen
because information on these factors could be obtained
more easily by a telephone interview than data on the more
specific demand factors. Information also was obtained
from the respondents concerning only those explanatory
variables that were correlated significantly with the
three criterion factors. It was not possible in a tele—
phone interview to collect data on all of the structural
and attitudinal factors, but data were Obtained on the
explanatory variables used in the total toll, equipment
expenditures, and number of lines models presented in
Tables 5.5, 5.8 and 5.9. A number of individuals in certain
firms had to be contacted in order to obtain accurate data
on both the independent and criterion variables.

The regression models developed from the analysis

sample were then used to predict the three criterion factors

1Because the month that this data was collected was not
the same as that used in the analysis sample, seasonal
factors were expected to affect the predictive accuracy of

 

194

for each of the thirty firms included in the validation
sample. Product moment correlations between predicted
and actual demand for both the analysis and validation
samples are presented in Table 5.11. For the sake of
sample comparisons, significance values for each corre—
lation coefficient and mean values for all demand variables
are also presented.

An analysis of the correlations presented in Table
5.11 indicates that, in general, correlations between pre-
dicted and actual values for the validation sample are
somewhat lower than those previously obtained for the
analysis sample. Also, a number of the multiple correlation
coefficients are no longer significant because of the
decrease in R values coupled with the major decrease in
sample size. It is optimistic to note, however, that in
those cases in which highly significant correlation coeffi-
cients were obtained for the analysis samples, the pre-
dictive ability of the regression models on the validating
data remains quite high. This is particularly true for
the number of lines criterion across all three segments.

As expected, the greatest variance in the validation

and analysis sample results can be seen in the toll usage

 

the total toll models. However, these seasonal factors
should have little influence on the demand for equipment.

.m.m

.m.m

.m.m moanme CH Umume mmHQMHHm> Houoflpmum can

mnouomm pcmaop cmw3umn mmHnmcoHumamu How mucmfloflmmwoo coflemflmuuoo mamfluHDE u mee

.mmHQMHHm> :oHHmuHMU How momma mamfimme

 

195

 

 

 

 

 

 

mO. mm. m~.m OH HOO. OO. mO.m me mmcHH mo Henssz
mO. em. He.Om~ OH HOO. OO. mm.Hm~ Ne Hancoe\wv musuHecmaxm namemHsvm
.m.: me. em.em OH HOO. Om. AH.OOH me Aspcoe\wv HHoe Hmuoe
Hmafimumm
OH. HO. mO.m OH HOO. mm. me.m NO mmcHH mo amassz
mO. Os. OO.meH OH HOO. OO. OO.HOH NO Haeqoe\wv muspHoammxm unmemHsOm
.m.c Om. eO.mvm OH HOO. mm. 50.He~ NO Heucoe\mv HHoe Hmuoe
Hmammmaosz
.m.c OO. ev.m OH HOO. OO. He.m we mmcHH mo Henssz
.m.: em. em.emH OH HOO. me. ee.HOH he Anucos\wv mHOUHmammxm ucwsmHsvm
.m.c he. OH.ONNO OH HOO. OO. 5H.vmmw he Haucos\wv HHoa Hmuoe
Honouommscmz

.mHm m x : .mHm eem ex C cofluwufluu panama
moflumflvmum madfidm moflumpflam> mUHumHumum mHaEMWImeNHmc< \ucmfimmm mmmcwmom

 

 

mmqmzdm ZOHBfiQHQ<> Q24 mHmMA¢Z¢.Somm QMZHfiHmO mBQDmmm ZOHmmmmwmm m0 ZOmHM¢NEOU

HH.m aflmdfi

196

predictions. Even though there is no significant differ-
ence between the mean value of toll usage for the two
samples, the correlation between predicted and actual toll
values for the validation sample is much lower than is the
case in the analysis sample. This indicates that some
factor not included in the model has a seasonal influence
on toll demand. Whether this factor is related to past

or anticipated sales or some other variable cannot be
determined from the limited data obtained in this study.
It may be that the factor causing a variation in toll
usage over time could be included in a general model by
using a dummy variable to represent a seasonal influence.
It is more likely, however, that models have to be developed

that are based on both a finer breakdown of commerical

users and specific months or periods of the year.

Effect of Attitudinal Factors on the Demand Models

The regression models discussed previously in this
chapter are conspicuous by the absence of any behavioral
or attitudinal factors. Certain attitudinal variables
are found to be correlated significantly with several
criterion factors, but these variables contribute little

additional information to the demand models that include

197

purely structural explanatory factors. In other words,
basic communication needs and demand are best explained
by the traditional Classificatory variables related to
firm type and size, number of various types of employees,
products handled, and geographic sales and supplier
patterns. There are several possible explanations for
this.

1 for

1. Significant variances were shown to exist
several attitudinal variables across the five
business segments; however, these variances were
not great enough within segments to contribute
significant additional information to the
models developed using the structural variables.

2. Where variances in attitudinal scores or rankings
were significant among firms in the same segment,
the variances are highly correlated with certain
Classificatory variables and thus add little
information to the structural models.

3. Certain attitudes may not be relevant to the demand

criteria. In other words, although attitudes were

hypothesized to affect demand behavior, some of

 

1See Chapter IV, pp. 109-119.

models,

198

the attitudes measured in this study probably

are not relevant to telephone usage and/or

purchase behavior.

The attitude measurement techniques used in this
study may not have been exact enough to dis-
criminate among various types of users of tele—
phone offerings. Also, some of the respondents

may not have been directly responsible for tele-
phone related buy decisions. Their attitudes,
therefore, would have little, if any, impact on

the demand behavior of the firm. In other words,
significant attitudinal variances among respondents
might be determined. But if the individuals
holding these varying attitudes have no effect on
the demand behavior of the firm, little association

between attitudes and demand can be expected.

While the addition of attitudinal variables to the pre—
dicting equations adds little information to most of the
several of the attitudinal factors show interesting
variations and relationships with other explanatory and
dependent variables. Some of these relationships were

discussed in Chapter IV.1 Several others will be mentioned

 

1See pp. 109-19 and p. 140 in Chapter IV.

199

in the following paragraphs.

One attitudinal variation that has not been mentioned
previously is the relative importance of management's
attitude toward telephone communications as related to its
use as a service, cost saving, or means of obtaining
new customers. No significant correlation between this
attitude and demand could be determined. Yet, it is
interesting to note that the large majority of retailers
(80%) and dry cleaners (100%) viewed telephone communica-
tions primarily as a service. Realtors were divided in
their attitudes. 47% placed greatest emphasis on telephone
communications as a service and 53% saw it as an important
tool in attracting new customers. Wholesalers (70%)
and manufacturers (78%) placed a substantial amount of
importance on telephone communications as a cost saving
mechanism. Very few firms (approximately 5% in each seg-
ment) stressed its importance as a new customer generator.
These categories are not mutually exclusive. Nevertheless,
the emphasis placed on each activity by firms in the
various segments provides some insight concerning manage-
ment's attitudes toward the uses and value of telephone

communications.

200

If these perceptions are viewed across all segments
and related to firm size, it is interesting to note that
48 or the 55 largest1 firms in the wholesaler and manu-
facturer segments and 7 of the 12 largest firms in the
retailer segment viewed telephone communications as most
important to their firms because of its cost-saving
qualities. Since this implies heavy users of telephone
equipment and services have adapted their usage patterns
primarily on the basis of cost saving motives, a wise
selling strategy may be to emphasize some other aspects
or benefits of telephone communications to these users.
This is especially true if this motive has been heavily
emphasized in past sales approaches and its demand—creating
ability has been exhausted for the frequently contacted
large customer. On the other hand, the retailer potential
demand may be better realized by emphasizing the added
cost savings that result from increased telephone equipment
and/or service utilization. This strategy, however, would
be logical only for large firms that are burdened with heavy

amounts of written communication requirements. A heavy

1Total number of employees is used in this case as

the size variable.

201

emphasis on cost savings in smaller firms could lead to
a decrease in telephone equipment and service usage,
rather than the desired increase.

Another attitudinal factor included in the survey
concerns customers‘ forecasted response to a hypothetical
increase in toll rates. Several studies have indicated
it is difficult to predict accurately purchase or usage

1 However,

behavior solely on the basis of intentions to buy.
it was hoped that information on this "intentions” variable
might provide some insight regarding the sensitivity of
toll usage to a potential rate increase. Accordingly,
respondents were asked to estimate their firm's expected
demand response to a 20% increase in both intra- and
interstate toll rates. The 20% figure was chosen following
pilot sample results that indicated this figure represented

a change that was realistic, yet large enough to cause sig-

nificant variation in customer response.2

 

1For example, see George Katona, The Powerful Consumer
(New York: McGraw-Hill Book Company, 1960), Chapter VIII;
and James Tobin, "On the Predictive Value of Consumer
Intentions and Attitudes," Review of Economics and Statistics,
February, 1959, pp. 1-11.

2Initially, it was planned to also include questions
regarding expected responses to a 20% decrease in toll
rates, but pre-test results showed no variation in responses.
All pre-test respondents indicated their toll usage behavior

202

Analysis of responses indicates that in very few
cases did respondents expect the decrease in either intra-
or interstate toll usage time to be as large as the
corresponding increase in toll rates. In other words,
the responses in general indicate a rather strong
inelasticity of demand for toll usage. There are, however,
several interesting variations in user intentions across
business segments, different size firms, and varying
communication uses.

Table 5.12 indicates that respondents in the manu-
facturer and wholesaler segments expected a much greater
decrease in interstate toll usage to result from an increase
in toll rates then was true for firms in the other three

1

segments. A more detailed examination of responses

indicates this difference results primarily from the expected

 

would not be affected at all by a 20% decrease in toll rates.
Possible explanations for this somewhat surprising inelastic
demand response are given in this section.

1A chi-square test was used to test the hypotheses
that the rate chance had the same affect on the manufacturer-
wholesaler groups combined as on the retailer, realtor, and
dry cleaner groups combined. This hypothesis was rejected
at the .001 level. No significant differences were observed,
however, between the manufacturer and wholesaler responses
or between the retailer, realtor, and dry cleaner responses.

203

TABLE 5.12

NUMBER OF FIRMS EXPECTING DECLINES IN INTERSTATE
TOLL USAGE RESULTING FROM A 20% INCREASE IN TOLL RATES
BY BUSINESS SEGMENT

r
‘

 

 

Business Expected Decrease
Segment 0-10% 10-20% 20-30% 30-40% I>40%
Manufacturer 25 14 5 3 0
(.53) (.30) (.11) (.06) (.00)
Wholesaler 41 12 7 2 0
(.66) (.19) (.11) (.03) (.00)
Retailer 37 5 0 0 0
(.88) (.12) (.00) (.00) (.00)
Realtor 21 2 0 0 0
(.91) (.09) (.00) (.00) (.00)
Dry Cleaner 9 0 0 0 0
1.00 (.00) (.00) (.00) .00
TOTAL 133 33 12 5 0
(.73) (.18) (.07) (.03)

 

*Figures in parentheses indicate the proportion of
respondents in each segment that expect the designated
decrease in toll usage.

204

impact of the rate increase on toll communications with
suppliers. Although very few firms in any segment expected
a decrease in communications with customers to result from
a rate increase, many firms in the manufacturer and whole-
saler segments did foresee a substantial decrease in
supplier toll communications. This implies that customer
telephone communication is relatively insensitive to rate
changes while supplier telephone communication might be
severely affected by a rate increase--particularly for
those firms in which telephone communications is heavily
relied upon for supplier contacts.

Although not included in Table 5.12, these same
relationships exist for the intrastate toll communication
case. The only area where the results seem to differ
between the intra- and interstate cases is where the
amount of instate toll usage for a firm differs substantially
from the out-state usage. If a firm's toll calls are
primarily out-of-state, the proposed rate increase is
expected to have a much greater impact on interstate usage.
This is logical, since absolute cost increases resulting
from the rate increase would be much more significant in

the high usage area than in the low usage toll component.

205

If the firms are segmented into heavy and light
users1 of toll service, a somewhat different pattern appears.
The relationship between heavy and light interstate toll
users and expected responses to a rate increase is shown in
Table 5.13. This table shows that heavy users of toll
service generally are less likely to expect decreases in
either intra- or interstate toll usage to result from a
rate increase than light users. These findings are par-
ticularly true for the manufacturer, wholesaler, and
retailer segments where the variance between heavy and
light toll users is greatest. A possible explanation for
this finding is that the heavy user is also usually the
larger firm. Telephone toll costs are viewed as relatively
insignificant When compared to other expenses incurred by
the firm. Conversely, the light users, generally the
smaller sized firms, probably View telephone toll costs as
a much larger expenditure relative to the total expenditures

of the firm. This would suggest that the small firm is

much more sensitive to changes in toll rates. Unfortunately,

 

1If a firm was in the top 25% of the toll users in a
business segment (based on a month‘s total toll billing) it
was considered a heavy user; likewise, a firm in the bottom
25%.was classified as a light user.

206

TABLE 5.13
EXPECTED RESPONSE OF HEAVY AND LIGHT

INTERSTATE TOLL USERS TO A TOLL RATE INCREASE
BY BUSINESS SEGMENT

 

 

 

 

Business Averagel Expected Decrease in Usage
Segment Heavy Users Light Users
Manufacturer 6.7% 11.1%
Wholesaler 7.0 13.0
Retailer 6.0 10.5
Realtor 5.02 6.7

Dry Cleaner 5.0 5.0

 

lThe arithimetic mean is used here to denote
the average decrease. The mean was calculated using
the mid-points of the ranges shown in Table 5.12.
Thus, 5% was used for all firms expected a 0-10%
decrease; 15% for firms expecting a 10-20% decrease,
etc.

2The fact that the percentage values for heavy
and light users are nearly equal in the realtor
segment and identical in the dry cleaner segment
can be explained by the fact that 21 out of 23
respondents in the realtor group and all nine of the
respondents in the dry cleaner group expected their
usage to decline by 0-10% (see Table 5.12).

207

no data were collected concerning the cost factors for
each firm; and, without this information, the actual
relationships between toll billings as a prOportion of a
firm's variable costs and the perceived responses to
increased rates can only be hypothesized.

One other point concerning the apparent insensitivity
of the heavy toll user to increases in toll rates should
be mentioned. The heavy user, while not perceiving a
substantial decrease in usage to result from an increase
in rates, may also consider alternatives such as WATS to
be much more feasible and attractive when toll rates
increase. Thus, the heavy user's sensitivity to rate
increases might also be measured by the ability and tendency
to switch to alternatives such as WATS. Unfortunately,
such alternatives were not included in the possible responses.
Therefore, no accurate inferences can be made concerning
the cross-elasticities of demand that might exist for
these types of services.

In summary, most respondents in all segments expected
toll usage to be affected only slightly by a 20% increase
in toll rates. Respondents in the wholesaler and manu-

facturer segments did indicate, however, that telephone

208

communications with suppliers would be most seriously
affected by a rate increase. Also smaller firms or
lighter users of toll services seem to perceive more
serious cut-backs in toll usage than do the larger firms
or heavy users.

If we assume that the reactions given to the
hypothetical rate increase are indicative of what actually
would happen if toll rates were increased, certain
managerial and regulatory implications are suggested. First,
it is difficult to estimate precisely how much total revenue
would increase or decrease as a result of a change in toll
rates. But the findings presented here indicate that for
some segments of the commercial market an increase in
rates would lead to a substantial increase in revenue.

For those segments or businesses where toll communication

is extremely vital to a firm's operations, an effective
strategy might be to raise toll rates substantially. This
strategy of higher rates in inelastic segments is viewed

by some utility personnel as simply pricing services highest
where the "value of service" is greatest. For example,

a representative of Consumers Power Company stated that

"In designing a rate schedule, we price those services

209

closer to marginal costs that have a low value of service,
and we price those services further away from marginal
costs that have a high value of service.1 Thus, from a
managerial point of view, a more profitable segmentation
strategy based on the value of service concept could be
designed whereby rates are increased to specific inelastic
segments.

The social implications of this type of strategy
might lead to a different regulatory policy, however,
especially where inelastic segments are least able to
bear the burden of higher rates. For example, this study
primarily dealt with small and medium sized firms and not
with the large corporations. The majority of firms responding
may not have had the option of adOpting such alternatives
as WATS because of their limited volume of toll calls.
Where the large firm may Opt for WATS rather than accept
an increase in toll rates, the medium and small size firms
would have to simply live with increased costs. In these
cases, regulatory policies might be designed to either

protect certain inelastic segments from high rates or allow

 

1William J. Jefferson, "Comment," in Performance Under
Regulation, Ed. by Harry M. Trebing (East Lansing, Michigan:
Institute of Public Utilities, Division of Research, 1968),
p. 103.

 

 

210

toll rates to be increased in certain commercial segments.
The added revenue could then be used to subsidize lower
toll rates in more price sensitive commercial or residential
sectors. The possibility that competition would find
these "high rate" segments profitable future targets,
however, makes the long run implications of this type of
policy somewhat questionable.

In summary, the attitudinal factors discussed in
this section and in Chapter IV contribute little to the
prediction models. They do provide some interesting,
and, perhaps helpful, insights concerning possible variations
in marketing approaches to the different commercial market
segments. In addition to the strategy types of inferences,
a number of regulatory or policy implications were
suggested by these attitudinal responses. These and other
regulatory and managerial implications are discussed

again in greater detail in Chapter VI.

Probabilistic Demand Models: Concept and Illustration

 

From a strategy viewpoint, it would be advantageous
for a supplier to be able to assign some likelihood to a
firm's increasing its demand for a product and/or service.

For example, a salesman would benefit if he could determine

211

the probability of a customer increasing his demand for

a specific offering solely on the basis of some knowledge
of certain descriptive characteristics of the customer.
This ability would help the salesman allocate his time
among customers in an optimal fashion. Where the customer
is a multiple purchaser,l this knowledge would also suggest
those products and/or services that have the highest
potential for increased sales or usage.

To illustrate the applicability of this type of
analysis, the criterion variable "number of lines" was
chosen for analysis. Several reasons dictated the use of
this demand criterion. First, this variable is discreet
in nature, and the variance in this demand factor is,
generally, quite low. Second, in most segments one or
two independent variables explain over 90% of the variance
in this criterion variable. Since the sample size is quite
small, using more than two explanatory variables in the
conditional probability matrices would make many cell
sizes extremely small. Finally, this criterion variable

is the only dependent factor that is correlated strongly

 

l .

Multiple purchaser as used here means a customer who
purchases more than one type of product or service from
the same supplier.

212

with the same two explanatory variables across all
segments. This means that all responding firms can be
used as a base for developing the probabilities rather than
arriving at different probabilities for each segment. It
is likely that a separate analysis of each segment would
improve the accuracy and validity of the probability
estimates; however, the purpose of this analysis is simply
to illustrate the applicability of using this approach
for discriminating between customers of high and low demand
potential.

The two factors chosen as indicator variables for
the number of lines criterion are the number of offices
and number of employees associated with each firm. Frequency
distributions including these three variables are derived
by combining respondents from all five segments. These
distributions are then used to calculate conditional

probabilities of the form:

P = p(L/E,O) where P = probability of a firm with E
number of employees and 0
number of offices demanding
L number of lines.

These explanatory variables may not be the best pre-

dictors of the criterion for any one segment, but they are

213

the two factors which are correlated most highly with
the number of lines criterion over the combined segments.
The probability matrices relating the predictor variables,
both individually and combined, to the criterion are
presented in Tables 5.14, 5.15, and 5.16. Table 5.16
shows the probability of "trunk" or line demand for a firm
conditional on both the number of employees and number of
offices the firm maintains.

Because of the relatively small sample size and
large variance in predictor variable values, ranges are
established for both the employee ED and office (0) variables.
These ranges were determined such that the number of values
occurring in each vector of the frequency table were
approximately equal and yet large enough to calculate a
probability value for each cell. Ideally, a finer break-
down of all variables would be preferred, particularly in
the number of employees case in which only three categories
were used. Any expansion in categories, however, would
have created a number of vectors in which few if any
entries exist. This would make the calculation of
probabilities totally unrealistic, if not impossible. The

number of data points, therefore, constrains or limits

214

the size of the conditional probability matrix.

Another consideration in structuring a probability
matrix to be used for predictive purposes is that the
terms in the matrix should be as close to "zero" or "one”
as possible. This means that, given a specific number of
employees and offices, the probability of a firm demanding
some specified number of lines is either one or zero.
Varying the number, or size, of the intervals used for
the predictor variables might better satisfy the zero-one
criteria, but the amount of information provided in the
model would also decrease. In other words, even though
increasing the width and decreasing the number of intervals
used for the predictor variables might better satisfy the
zero-one criteria, the usefulness of such a strategy for
management decisions would be reduced greatly. This
problem is clearly shown in the data presented in Table 5.14.

Because the ranges established on the employee
variable are large, the managerial value of the information
provided in the probability matirx in Table 5.14 is

limited.1 For example, in the 0—10 employee column, there

1Due to the size distribution of firms, it is not
possible to expand the number of employee breakdowns. As
mentioned earlier, increasing the number of employee categories
would place an extremely small number of firms in some size
categories and make the associated probability values totally
unreliable.

 

215

TABLE 5.14

CONDITIONAL PROBABILITY MATRIX RELATING LINE DEMAND
TO NUMBER OF EMPLOYEES

 

 

 

 

 

(P(L/E))
Number of Number of Lines l—jtfi
Emplgyees 1 2 3 4 )4
0-10 .42 .31 .18 .06 .03
11-40 .03 .18 .51 .17 .11
>40 .00 .05 .29 .31 .35

 

is nearly an equal probability that a firm.will use either
one, two, or greater than two lines. There are two
possible reasons for this. The demand for lines is either
extremely sensitive to changes in the number of employees
for these smaller size firms or sensitive to changes in
some other variable. It is obvious, however, that only
knowledge of the fact that a firm has from 0-10 employees
is of little administrative use. Either a finer breakdown
of the employee factor or some information on additional
variables is needed.

This same basic problem exists on the opposite end

of the employee spectrum. For the greater than forty

216

employee category, an almost equal probability exists that
a firm will demand either three, four, or more than four
lines. As in the previous case, this could be the result
of a greater demand sensitivity to changes in the employee
factor than is shown in the matrix or due to the influence
of some other variables not included in the probability
model.

The information provided in Table 5.14 is not of
great managerial use because of the wide ranges establish
on the employee predictor; however, the data in this table
do provide some insight concerning the highest potential
demand customers. Based only on the employee variable,
the utility salesman should contact those firms that have
greater than forty employees but use only one or two
lines. It seems likely that these firms are "under-lined"
and would be the best prospects for additional line demand.

Table 5.15 shows the demand for lines conditional on
the number of offices a firm maintains. This matrix pro-
vides more useful information than that presented in
Table 5.14 since each office category has but one high
probability value associated with it. For example, for

those firms having one office, the probability is .62 that

217

TABLE 5.15

CONDITIONAL PROBABILITY MATRIX RELATING LINE DEMAND
TO NUMBER OF OFFICES

 

 

 

 

(P(L/o))

Number of Number of Lines

Offices 1 2 3 4 >4
1 .62 .30 .08 .00 .00
2 .15 .48 .22 .13 .02
3 .06 .30 .46 .16 .02
4,5,6 .00 .12 .24 .44 .20
)6 .00 .04 .14 .22 .60

 

the firm will have one line and .92 that the firm will
have not more than two lines. Likewise, for those firms
having more than six offices, the probability is .82 that
four or more lines will be utilized.

Table 5.16 presents the probabilities for line demand
conditional on both the number of offices and number of
employee variables. Because the matrix entries in Table
5.16 are closer to zero or one, the probability values
included in the table are much more useful than those

included in Tables 5.14 or 5.15. When demand is conditional

CONDITIONAL PROBABILITY MATRIX RELATING LINE

218

TABLE 5.

16

DEMAND TO NUMBER OF OFFICES AND NUMBER OF EMPLOYEES
(P(L/E.O))

 

No. of No. of

_——‘
n

Number of Lines

 

 

Empees Offices 1 2 3 4 >4
0-10 1 .75 .20 .05 .00 .00
2 .18 .64 .08 .10 .00

3 .04 .25 .58 .09 .04

4,5,6 * * * * *

)6 * * * * *

10-40 1 .41 .46 .13 .00 .00
2 .07 .15 .65 .10 .03

3 .02 .08 .62 .18 .10
4,5,6 .00 .04 .32 .43 .21

>6 .00 .02 .13 .17 .67
>40 1 ** ** ** *1: *1:
2 .00 .04 .09 .32 .55

3 .00 .07 .20 .27 .46
4,5,6 .00 .00 .05 .20 .71

>6 .00 .00 .05 .20 .75

 

*No firms with 0-10 employees had greater than

three offices.

**No firms with greater than 40 employees had

only one office.

219

on these two independent factors, a higher probability
exists that a firm will use a specific number of lines than
was the case when only one of the predictive factors was
used. One could hypothesize that, given a larger sample
and an additional explanatory factor, even more exacting
estimates of demand would be achieved.

The information provided in Table 5.16 could be of
significant assistance in the structuring of a priority
system for assigning salesmen to prospective users of
additional lines. If one assumes that the number of offices
and number of employee variables are the only factors
affecting a firm's demand for lines, a system could be
structured whereby firms are ranked according to the
expected return resulting from a salesman contact. For
example, a simple expected value calculation could be
derived for each customer based upon the information
included in Table 5.16 and some knowledge of the costs and
revenues associated with installing additional line capacity.
If the expected value of a salesman's contacting a firm

"A” is denoted by E.V.(A), then:

220

E.V.(A)

PA(Ln)X Pun-LA)

where: LA = number of lines firm ”A" presently uses.

Ln = all line values greater than LA that
firm "A" could demand.l

PA (Ln) probability of firm ”A" demanding Lh lines.

P (Ln-LA) = profit resulting from a firm adding
(Ln-LA) lines.

Since cost and revenue information related to the installa-
tion of additional line capacity was not obtained, actual
expected value calculations are not made. However, this
method should provide a more accurate system for allo-
cating sales calls than would a method derived from the
regression models. If one were to use the difference
between actual demand the predicted demand value obtained
from the regression model to arrive at a potential demand
figure, information concerning the firm's varying probabili-
ties of demanding additional numbers of lines would be lost.
Because the conditional probability matrix includes informa-

tion on the varying likelihoods of a firm demanding a

 

1For the probability matrix presented in Table 5.16
no probabilities are given for any single number of lines
greater than four. Ideally, a larger sample size would
allow the inclusion of values much greater than this and
permit greater exactness in the derivation of expected
values.

221

specific number of additional lines, it should provide a
good system for analyzing demand and allocating sales
effort among prospective customers.

There are several limitations inherent in the use of
conditional probability matrices. First, average actual
demand values are used to derive potential demand. This
means that firms in each employee—office category that
have the highest actual demand also have zero potential
demand. This, of course, may be misleading since high
users may at times also be good prospects for additional
usage.

A second limitation is that the calculation of con-
ditional prObability values was based on frequency distri-
butions for all segments combined, rather than for each
individual segment. Since the regression models were more
accurate when developed for individual segments, it is
logical to assume that probability models developed on a
segment basis would also be more accurate. The limited
sampka sizes, however, prohibited this type of analysis.

Finally, just as the small sample size prohibited an
analysis on an individual segment basis, this factor also

limited the number of explanatory variables that could be

222

incorporated in the probability models. Independent
variables had to be used which were correlated most highly
with the criterion across all segments. Since the variables
chosen were not necessarily the best predictors of line
demand for each segment, the probability matrices derived

were likely less than optimum.

Summary

In this chapter, the varying influences that differ-
ent structural and attitudinal variables had on each demand
criterion were examined in detail. Linear multiple
regression models were developed to illustrate how these
explanatory factors interacted to affect demand. The
results of these regressions are encouraging since the
structural variables were good predictors of nearly all
criterion factors for the five business segments. High
correlation coefficients between actual and predicted values
were obtained by using a maximum of four explanatory
variables.

Several attitudinal variables were significantly
associated with a number of criterion factors. These
variables, however, did not contribute a significant amount

of additional information to the structural-based forecasting

223

models.

A number of transformations were attempted on
certain independent variables in an effort to improve the
predictive accuracy of the models. In all cases, however,
these transformations led to little, if any, improvement
in the basic linear regression models.

In general, the demand models varied greatly across
the five segments--both in terms of the predictors used
and the forecasting accuracy achieved. The basic segmenta-
tion strategy used in this study clearly indicates the
varying effects that certain independent variables have
on telephone demand in each of the business classes.

To show how data on a few, easily measured structural
variables might be used to select customers of high
potential demand, conditional probability matrices relating
the number of lines criterion to the number of employee
and number of office predictors were constructed. Although
the accuracy of these matrices was limited by the relatively
small sample sizes, these matrices provide a useful way
for analyzing demand.

The managerial and regulatory implications of some of

the findings presented in this chapter and Chapter IV, as

224

well as some suggestions for additional research, will be

the major subject of the next chapter.

CHAPTER VI

SUMMARY AND CONCLUSIONS

Introduction

 

This dissertation focuses primarily on examining
the relationships between selected business firms'
structural and attitudinal characteristics and the demand
for specific telephone communication products and services.
The primary objectives of the research were:

1. To determine empirically the effect that certain
variables associated with different types of
business firms have on the demand for various
telephone utility offerings.

2. To incorporate these demand influencing variables
in models that can be used to forecast individual
firm telephone communication requirements.

3. To determine the accuracy and usefulness of the
models and to evaluate the managerial and
regulatory implications of the research findings.

The scope of the study was limited to a stratified

sample of business firms in the manufacturer, wholesaler,
225

226

retailer, realtor and dry cleaner business segments in

the Greater Lansing area. Questionnaires were sent to
owners or managers of each sampled firm. Information was
requested concerning the structural characteristics of

the firms and the respondents' attitudes toward the value
of various telephone communication offerings. Although

the "structural" portion of the questionnaires differed
somewhat for each business segment, the attitudinal sections
were the same for all business firms. Telephone inter-
views with ten firms from the manufacturer, wholesaler,

and retailer segments that had not responded to the mail
questionnaire were used to obtain data on selected
independent and dependent variables. These data were used
to test and validate several models for each of these three
segments.

Factor analysis was used to identify the meaningful
dimensions and critical variables influencing demand behavior
in each segment. These variables were incorporated in the
regression equations that served as the basic predictor

models.

227

General Findings

 

This study shows that individual firm demand for
telephone communications offerings can be estimated using
key structural or Classificatory variables. The regression
and correlation techniques used in the analysis were helpful
for examining the relationship between firm character—
istics and demand and for forecasting customer communication
requirements. Highly significant correlation coefficients
were obtained for most of the regression models.

Linear demand models fit the data examined in this
study most precisely. Transformations on certain independent
variables were tried when an examination of residuals
indicated that a non-linear model might provide a better
fit for the data. However, these transformations did not
improve the forecasting accuracy of the demand models
significantly.

Even though cause-effect relationships cannot be
proven, an analysis of the regression models implies that
structural types of factors have the greatest influence
on individual firm demand. The principal components
analysis showed that the most important explanatory

variables grouped themselves along the basic dimensions

228

of firm size, geOgraphic sales variations, supplier
characteristics, and employee types. Structural variables
from these four areas were the primary explanatory factors
used in the demand models.

Several conclusions can be stated regarding the
correlations between structural variables and demand
criteria. First, size variables generally were correlated
more highly with demand criteria across all segments than
were any other types of independent factors. However, the
correlation between specific size factors and demand
criteria varied greatly across the five segments. For
example, the total number of employees variable was corre—
lated highly with nearly all demand criteria in the Whole-
saler segment. However, this explanatory variable was
associated weakly with the dependent variables in the
manufacturer case. In the manufacturer models, the number
of salaried employees was the best indicator of demand.

Second, although there was usually a high degree of

correlation among the size variables, certain size variables

were correlated much more highly with each demand criterion
than were others. For example, in the retailer segment

the size variable total number of employees had a much

229

higher correlation with all demand criteria than did most
other dependent factors. This indicates that some other
factors besides firm size influences the demand for Yellow
Page Advertising. These relationships between firm size
and demand also suggest that a careful examination of each
type of business is needed before the indicators and causes
of demand can be determined accurately.

As expected, the explanatory factors related to
geographic variations in customer locations, sales, and
supplier locations most strongly affect the toll usage
criterion variables. This is particularly so in the manu-
facturer and wholesaler segments where differences in
these geographic factors are most likely to occur. The
absolute number of suppliers and/or the amount of gross
sales corresponding to local, in-state, and out-of-state
regions were the geographic factors most frequently used
in the demand equations. However, there were several
cases where the percentage of total sales, and not the actual
amount of sales occurring in a given geographical area, was
most critical. The wholesaler and manufacturer Yellow Page
Advertising demand models is the most obvious case. Here,

the percentage of a firm's sales made to customers in the

230

local calling area was a major variable influencing
Yellow Page usage.

The addition of attitudinal variables to the
structurally based demand models did not improve the fore-
casting accuracy of the models significantly. Most of the
attitudinal responses, although varying significantly among
business segments, did not vary significantly within
business classes. For example, there was a substantial
difference between what respondents in the retailer and
manufacturer segments perceived as being the most important
and time consuming function of telephone communications.
However, there was little difference in these perceptions
for firms in the same segment. Since the demand models
were constructed on an individual segment basis, this
meant that attitudinal factors contributed little to
explaining the variance in the demand criteria.

The respondents' perceived or expected reaction to
an increase in toll rates, however, did vary both within
and across segments. The major influence on variations
within segments was the size of the firm. In general,
respondents from smaller firms expected toll usage to

decrease much more in response to a rate increase than did

231

those in larger firms. Since toll usage varied directly
with firm size, this also meant that light users were
more insensitive to toll increases than were heavy users.
Respondents from larger firms also perceived them-
selves as more aware of the total array of available
telephone services than did respondents of smaller firms.
Thus, demand, or usage, was related positively to the
perceived understanding factor. However, because of the
mediating size variable and other possible influencing
factors, no definite statements or conclusions concerning
the effect of providing light users with additional informa-

tion can be made.

Implications of the Study

Several points should be emphasized before any
specific implications of the research are discussed. First,
public utilities cannot in general adopt changes in their
pricing or product strategies without first obtaining
approval from the appropriate regulatory agencies. There-
fore, any implications or suggestions resulting from this
study are subject to the unique limitations of a firm

operating under regulatory constraints.l

 

1For a discussion of some of these constraints, see
Chapter I, pp. 8-15.

232

Second, because the basic objectives of the study
were managerial, the implications suggested by the data
are primarily administrative in nature. However, as a
result of the investigation several regulatory implications
are also suggested. Since more detailed information is
required before specific inferences can be drawn concerning
the regulatory issues, these problems are covered in the
section dealing with future research.

Third, the study was conducted in a limited geo-
graphical area among specific types of business firms.
As such, many of the findings and implications suggested
by the research may not apply to different types and/or
sizes of firms. Also, variations in rates or product
offerings for different utilities and geographical areas
may limit the general applicability of the research findings.

Two general types of explanatory variables were
analyzed in an effort to better understand and predict a
business firm‘s demand for telephone offerings:

1. Structural or Classificatory variables, which

although uncontrollable, are in many cases
capable of measurement and prediction over time.
2. Attitudinal variables, which are less easily

monitored but which could be influenced by

233

variations in the utility's marketing strategy
over time.
The extent to which each class of variables influences
demand determines the usefulness of the models for fore-
casting and strategy purposes.

Since demand for most of the dependent variables

was found to be most sensitive to changes in the structural

factors associated with the firm, the models are useful
for forecasting purposes. A periodic monitoring of the
structural variables can help a utility identify changes
in the demand requirements of its customers. It can also
provide information concerning the best prospects for
salesmen to contact. Probability models of the type
illustrated in Chapter V, where a limited number of these
structural factors are used as predictors, provide one
method whereby only a few explanatory variables require
monitoring.

The attitudinal factors provided little information
for the demand models. However, variations in these
factors across segments suggests that different marketing

or selling strategies should be used for different market

segments. For example, the survey results showed that most

234

manufacturers viewed telephone communications as most
important to their firms because of its cost-saving
qualities. The majority of realtors, however, saw tele-
phone communications as an important tool for use in
attracting new customers. Thus, cost saving aspects of
additional telephone offerings might be used when selling

to one business segment and the use of additional telephone
offerings for attracting new customers stressed as a sales
appeal to another segment. Utility salesmen should be cogni-
zant of these influences when analyzing the wants and needs of
different classes of business customers. Changes in key
influencing factors are likely to occur over time. Salesmen
would benefit from assessing the possible impact of such
changes on the demand for each service.

The forecasting models developed for telephone
equipment and service demand suggest that similar models
might be structured for other regulated and non-regulated
industries. In regulated industries, structural variables
may be of primary significance in explaining customer demand.
In more competitive industries additional behavioral types
of variables would probably have to be included to account
for variances in the buyer's choice of suppliers. Thus, a

competitive environment offering choices among alternative

235

suppliers and products would likely lead to more complex
forecasting models. Such models would not only contain
factors indicating total demand for a particular type of
product, but also variables to predict a customer's
allocation of demand among competing suppliers.

The fact that structural variables were used to
forecast demand successfully on a limited geographic and
customer basis has other general implications for the
development of more macro-oriented models. Forecasting
equations similar to those developed in this study could
be structured for other business firms. By combining
this information with various census and trade data on
the number and sizes of firms located in geographical
areas, state-wide estimates of both actual and potential
usage patterns would be developed. Thus, instead of
using various economic indicators to forecast changes in
demand, a build-up method based on a detailed analysis of
different classes of business customers might provide a
more accurate model for predicting short and long run
changes in demand patterns.

The variations in perceived or expected reactions

to a hypothetical toll rate increase suggest that the value

236

of toll telephone services differs substantially among
business customers. In those cases where customer demand
was relatively inelastic to a toll increase, revenues would
be increased substantially through higher rates. Likewise,
the apparent sensitivity of certain firms to rate increases
suggests that a maintenance of existing rate structures
or a possible decrease in rates would be a more appropriate
pricing strategy for still other segments.

Two factors may limit the utility's ability to
increase rates in the more price—inelastic segments of
the market. First, telephone utilities today face
increasing competition in a number of product and service
areas. This suggests that a discriminatory pricing strategy
leading to highly lucrative returns from certain customers
may also lead to increased competition in those segments.
When this is combined with the regulatory consequences
that might result from certain discriminatory pricing
techniques, the long run interests of the utility might be
served better in some cases if a policy of pricing "what
the traffic will bear" is avoided.

It does seem likely, however, that a more effective

pricing strategy could be developed based on a more careful

237

analysis of the value—of-service concept or some other
segmentation criteria. For if firms differ greatly in

their needs and uses of communications offerings, as this
study shows, they may also differ in the amount they are
willing and able to pay for these services. The differences
in these demand elasticities requires more careful analysis,
however, before discriminatory strategies that are both
Optimal and fair can be developed.

Several findings also point to a need for more
effective communications between the utility and its
customers. First, many respondents perceived their under-
standing of the communications alternatives available to
them to be quite poor. Allied with this finding was the
fact that the majority of respondents indicated they
depended almost completely on the utility salesman for
suggestions concerning both the types and amount of tele-
phone communications equipment or services to utilize. A
better satisfaction of each customer's communications needs
would result from (a) a more accurate and detailed informa-
tion flow between buyers and sellers regarding the various
product and service offerings and their uses, and (b) a

greater understanding on the part of the utility salesman

238

concerning the variables that most directly affect the
demand for each offering.

Some of the findings concerning managerial recommenda-
tions that were discussed in the preceding paragraphs also
suggest certain regulatory implications. For example, a
more discriminatory pricing or rate making strategy based
on the value of service concept or the relative sensitivity
of the user to rate increases might result in high rates
for certain segments of customers that are least able to
bear the added cost burden. In the telephone utility area,
this would be particularly true for smaller firms that
have few alternatives available when performing certain
types of communications activities. A policy of charging
these relatively inelastic segments extremely high rates
may lead to regulatory constraints that essentially control
the specific prices that can be charged certain types or
sizes of customers.

Another area where regulatory policy may have greater
impact in the future concerns the manner in which "buy"
decisions are made by users of telephone or other utility
services. This study was not structured to obtain informa-

tion on how well each firm's present array of telephone

239

communications products matches its needs. Nevertheless,

the lack of understanding of offerings and the dependence

on utility salesmen for guidance suggests that there may

be many cases where firms are either under--or over--utilizing
certain types of offerings. Also, there may be specific
needs of certain classes of business customers that could

be better satisfied by some new or modified product offering.
More detailed studies patterned closely to this one are
needed in which demand is viewed from an individual firm
rather than industry level. These studies can provide
regulatory agencies with better ways of evaluating the
efficiency with which the present utility marketing system

satisfies the varying demands of different business firms.

Suggestions for Additional Research

 

Many of the implications and recommendations discussed
previously should be viewed as tentative suggestions
requiring greater research and expanded data analysis.

Some suggestions were made in each of the last two chapters
concerning the kinds of additional data that should be
collected to expand the models and improve their forecasting
accuracy. There are, however, a number of other areas where

this study suggests further research efforts are needed.

l.

240

An expanded study that considers a greater

array of customer types and sizes as well as

an expanded list of possible explanatory variables
should be constructed. Such factors as seasonal
and geographic influences on demand and a more
detailed examination of the effect of attitudinal
variables on usage rates are examples of the types
of additional predictors that should be studied.
It is likely that the detailed nature of much

of this information would require the use of
personal in-depth interviews rather than the mail
survey technique that was used in this study.

This study made little attempt to assess the
relative importance or influence of certain
controllable variables on the different demand
criteria. It seems likely, however, that factors
such as personal selling and advertising would
help to explain some of the variance in the demand
variables among different customers. By

examining the effect of such variables on demand
in different buy situations, more efficient allo-

cations of advertising and selling efforts might

241

be made. Such research should also provide a
better understanding of the role that regulation
might take in controlling the amounts and types
of promotional activity engaged in by a public
utility.

The value of various types of telephone products
and services and the relative price elasticity
of such offerings to different classes of users
also merits further research. Some implications
concerning these issues were suggested earlier
based on the results of this study. However,
more data is needed before specific recommendations
can be made regarding both managerial and regula-
tory product and pricing policies. Related to
these issues are questions concerning the impact
of existing product and pricing strategies on
different classes of business customers. For
example, what types and sizes of firms face the
greatest number of alternatives when selecting
communication offerings? For What types and
sizes of firms and for What kinds of usage needs

are offerings such as WATS really considered as

242

alternatives to the traditional unit toll exchange
communication systems? In what usage areas do
basic needs exist that are not now being satis-
fied by present communication offerings? These
kinds of questions can only be answered accurately
through detailed research studies of each business
segment's communications requirements.

Research should be conducted to assess how
discriminatory rate structures for exchange and
toll services have affected usage patterns
historically. This analysis should be helpful for
determining how value-of—service pricing has
influenced the demand for various telephone
communication offerings over time. It should

also provide information that would assist researchers
in estimating the effects of present and future
changes in utility rate structures.

Another area that requires further research is

the relationship between what utility salesmen

see as the primary factors influencing demand

and those variables that actually do affect demand.

A salesman traditionally may use certain criteria

243

associated with each customer as a basis for
recommending the adoption of specific types of
products or services. If these criteria have
little relationship to the buyer's actual needs,
firms may either under- or over-utilize many of
the available utility offerings. The extent to
which this problem exists is related directly to
the buyer's understanding of the number of alterna-
tives available and his knowledge of the substi-
tutability of the various product and service
offerings. This study indicated that many users
are unfamiliar with the total array of offerings
available and thus rely heavily on the salesman's
advice when selecting their mix of communication
services. Therefore, it is critical that the
salesman's perceptions of the major influencing
variables agree with the factors that actually
do reflect customer requirements. Only in this
way can a firm's communication needs be Optimally
satisfied.

Research should be conducted to examine the

feasibility of extrapolating the kinds of information

244

obtained in this study to more macro-oriented
demand models. By gathering additional demand
information on other sizes and types of firms and
simulating changes in certain critical demand
influencing variables over time, reliable estimates
of both individual segment and total market
communication needs should be possible. The
inclusion of controllable variables in these
models may also make it possible to simulate the
effect of modified advertising and selling
strategies on total revenue and profits. From a
regulatory and managerial perspective, further
research in this direction should provide a basis
for evaluating the feasibility and benefits of a

variety of product and rate structure innovations.

Concluding Comments

 

This research focused on the factors influencing the
business firm's demand for a number of telephone utility
offerings. However, many of the findings, implications,
and recommendations derived from the study are of value
to other types of utilities and non-regulated industries

as well. It is hOped that one of the most significant

245

contributions of this research is to provide a framework
or focus for structuring and carrying out further utility
related studies. Certainly, further managerial and
regulatory oriented research is needed in all utility
based industries if a better matching of utility resources
with customer needs and a more efficient allocation of

company marketing efforts is to be attained.

BIBLIOGRAPHY

BIBLIOGRAPHY

Books

Alexander, Ralph 8., Cross, James S., and Cunningham, Ross
M. Industrial Marketing. Homewood, Illinois: Richard
D. Irwin, Inc., 1961.

Bonbright, James C., Principles of Public Utility Rates.
New York: Columbus University Press, 1961.

Cooley, W1W., and Iohnes, P.R. Multivariate Procedures
for Behavioral Sciences. New York: John Wiley and
Sons, Inc., 1962.

 

Copeland, Melvin T. Principles of Merchandising. Chicago:
AeWe Shaw COe I 1924e

 

Draper, Norman, and Smith, Harry. Applied Regression
Analysis. New York: John Wiley & Sons, Inc., 1966.

 

Frank, Ronald E., and Green, Paul E. Quantitative Methods
in Marketing. Englewood Cliffs, N.J.: Prentice-Hall,
Inc., 1967.

 

Glaeser, Martin G. Public Utilities in American Capitalism.
New YOrk: The MacMillan Co., 1957.

Green, Paul E. and Tull, Donald S. Research for Marketing
Decisions. Englewood Cliffs, N.J.: Prentice-Hall, Inc.,
1966.

 

Grether, E.T. "A Theoretical Approach to the Analysis
of Marketing," in Theory in Marketing. Edited by
Reavis Cox and wroe Anderson. Homewood, Illinois:
Richard D. Irwin, 1950.

 

246

247

Hadley, G. Introduction to Probability_and Statistical
Decision Theory. Cambridge: Holden-Day, Inc., 1967.

 

Harman, Harry. Modern Factor Analysis. Chicago: university
of Chicago Press, 1960.

 

Hummel, Francis E. Market and Sales Potentials. New York:
The Ronald Press Cbmpany, 1961.

Jefferson, Williams J. "Comment," in Performance Under
Regulation. Edited by Harry M. Trebing. East Lansing,
Michigan: Institute of Public Utilities, Division of
Research, 1968, 103-105.

 

Katona, George. The Powerful Consumer. New York: McGraw-
Hill Book Co., 1960.

Kuehn, Alfred A. "Consumer Brand Choice - A Learning
Process?" in Quantitative Techniques in Marketing
Analysis. Edited by Ronald E. Frank, pp 11., Homewood,
Illinois: Prentice-Hall, Inc., 1967.

 

Massy, William F. Purchasing Behavior and Personal Attri-
butes. Philadelphia: University of Pennsylvania
Press, 1968.

Naylor, Thomas H., §t_al, Computer Simulation Technique.
New York: John Wiley & Sons, Inc., 1966.

Parten, Mildred. Surveys, Polls, and Samples. New York:
Harper & Brothers, 1950.

Pegrum, Dudley F. Public Regulation of Business. Homewood,
Illinois: Richard D. Irwin, Inc., 1965.

Peters, William S. and Summers, George W. Statistical
Analysis for Business Decisions. Englewood Cliffs, N.J.:
Prentice-Hall, Inc., 1968.

Phillips, Charles F., Jr. The Economics of Regulation.
Homewood, Illinois: Richard D. Irwin, Inc., 1965.

248

Robinson, Patrick J., Faris, Charles W., and Wind, Yoram.
Industrial Buying and Creative Marketing. Boston:
Allyn & Bacon, Inc., 1967.

Shepherd, William G. and Gies, Thomas G. Utility Regulation,
New Directions in Theory and Policy. New York: Random
House, Inc., 1966.

Shuford, Emir H., ”Some Bayesian Learning Processes," in
Human Judgments and Optimality. Edited by Maynard W.
Shelly and Glenn L. Bryan. New York: John Wiley &
Sons, Inc., 1964.

Stern, Mark E. Marketing Planning;_A Systems Approach.
New York: McGraw-Hill Book Company, 1966.

Theil, H. Economic Forecasts and Policy. Amsterdam: North -
Holland Publishing Co., 1958.

Trebing, Harry M. (ed.) Performance Under Regulation, East
Iansing, Michigan: Institute of Public Utilities,
Division of Research, Michigan State University, 1968.

Articles and Periodicals

Adams, walter. "The Role of Competition in the Regulated
Industries," American Economic Review, 527, 1958.

Atchison, Clyde B. "Fair-Return-on—Value-Theory in Rate
Making Loses Force," 25I.C.C. Practitioner's Journal,
(October, 1957), ll-15.

Auerch, Harvey and Johnson, Leland L. "Behavior of the
Firm under Regulatory Constraint," American Economic
Review, LII (December, 1962), 1052-1069.

Bass, Frank M. Tigert, Douglas J., and Lonsdale, Ronald T.
"Market Segmentation: Group Versus Individual Behavior,"
Journal of Marketing Research (August, 1968), 269-270.

249

Brown, Hillyer. "Value of the Service as a Ceiling on
Public Utility Rates," California Law Review, XXXIII
(1945), 283-297.

Claycamp, Henry. "Characteristics of Thrift Deposit
Owners," Journal of Marketing Research, II, No. 2
(May, 1965), 163-170.

Duncan, D.J. "Some Basic Determinants of Behavior in
Industrial Purchasing," Pacific Purchasor, XLVII

Evans, Franklin B. "Psychological and Objective Factors
in the Prediction of Brand Choice: Ford Versus
Chevrolet," Journal of Business, XXXII, No. 4 (October,
1959), 340-369.

 

Farley, John U. "Why Does 'Brand Loyalty' Vary Over
Products?," Journal of Marketing Research (November,
1964), ll.

 

Farrar, Donald E., and Glauber, Robert R. "Multicollinearity
in Regression Analysis: The Problem Revisited," The_
Review of Economics and Statistics, XLIX (February,
1967), 92-106.

 

Frank, Ronald E. and Boyd, Harper W., Jr. "Are Private—
Brand-Prone Grocery Customers Really Different?”
Journal of Advertising Research, V (December, 1965),
27-35e

Frank, Ronald E. and Green, Paul E. ”Numerical Taxonomy
in Marketing Analysis: A Review Article," Journal of
Marketing Research, V (February, 1968), 83.

 

 

Frank, Ronald E., Massy, William F., and Morrison, Donald
G. "Bias in Multiple Discriminant Analysis," Journal
of Marketinngesearch, II (August, 1965), 250.

 

"The Determinants of Innovative Behavior with
Respect to a Branded Frequently Purchased Food Product,"
.American Marketing Association, Proceedings of the
Winter Meeting, New York, December, 1964, 312-323.

 

250

Green, Paul E., Frank, Ronald E., and Robinson, Patrick J.
"Cluster Analysis in Test Market Selection,"
Management Science, XIII, No. 8 (April, 1967), 388.

 

Green, Paul E., Carmone, Frank J., and Robinson, Patrick J.
"Structural Characteristics of the Computer Market -
An Application of Cluster and Reduced Space Analysis,"
Working Paper, May, 1967.

 

Hermer, E.J. "On Mapping Industrial Markets," Dunn's
Review, LXI (May, 1953), 39.

Hellborn, Ludwig 8., "Can the SIC Yield Accurate Marketing
Data?" Journal of Marketing, 20 (October, 1956),
157-162.

 

Hummel, Francis E. "Pinpointing Prospects for Industrial
Sales," Journal of Marketipg_(July, 1960), 27.

 

Industrial Marketing Committee Review Board. "Fundamental
Difference Between Industrial and Consumer Marketing,”
Journal of Marketipg, XIX (October, 1954), 153.

 

Irwin, Manley R. "The Communications Industry and the
Policy of Competition, Buffalo Law Review, XIV (Winter,
1964), 256-73.

 

.Joyce, J., and Charron, C. "Classifying Market Survey
Respondents," Applied Statistics, XV (November, 1966),
191-215.

 

Phoning, Code. "Effective Techniques in Industrial
Marketing Research," Journal of Marketing, XXVIII
(April, 1964), 57-61.

 

133430, H. "Emotional Aspects of Industrial Buying,"
Dynamic Marketinggfor a Changing_World. R.S. Hancock
(ed.) American Marketing Association, Proceedings of
the 1960 Conference, 260-261.

LEBOntief, wassily W., "Proposal for Better Business
Forecasting," Harvard Business Review (November-December,
1964), 166-7, 172, 176, 180, 182.

 

251

 

Iflagee, John F. "Decision Trees for Decision Making,"
Harvard Business Review, XLII (July-August, 1964),
126-138.

Dhassy, William F. "Applying Factor Analysis to a
Specific Marketing Problem," Toward Scientific Marketing.

American Marketing Association, Proceedings of the
1963 Conference, 295-296.

. "Discriminant Analysis of Audience Character-
istics," Journal of Advertising Research, V, No. 1
(March, 1965), 39—48.

 

I4c>rgan, James N., and Sonquist, John A. "Problems in
the Analysis of Survey Data and a Proposal," American
Statistical Association Journal (June, 1963), 415-434.

.Fualda, Kristian S. "The Evaluation of Regression Results,"
Toward Scientific MarketingL American Marketing
Association, Proceedings of the 1963 Conference, 279—290.

I>itersol, Robert J. "Accuracy of Estimating Markets for
Industrial Products by Size of Consuming Industries,"
Journal of Marketing Research, V (May, 1968), 147.

Ihok>inson, Patrick J., and Carmone, Frank J., Jr.,
"Programming Notes," Journal of Marketing Research, V

(February, 1968), 94-95.

Schupack, Mark B. "The Predictive Accuracy of Empirical
LDemand Analyses," The Economic Journal (September, 1962),

551-575.

Shoaf, F.R. "Here's Proof - The Industrial Buyer's Human,"
ggndustrial Marketing, XLIV (May, 1959), 126-128.

snfifitII, Clifford E. and Cowles, Harold H. "Regulatory
Zanluence on Managerial Performance: An Attitude
ESurvey," Public Utilities Fortnightly, LXXVII

(September, 1966), 15-27.

Sdkadl, R.R. "Numerical Taxonomy," Scientific American, CCXV
(December, 1966), 106-116.

252

"Public Images of Public Utilities,"

Stephenson , William.
1963).

Journal of Advertising Research (December,

£3uits, D.B. "Use of Dummy Variables in Regression
Equations," Journal of the American Statistical
Association, Vol. 52 (1957), 548-551.

"Measuring Potential," Sales Management, Survey

Tobat, E.L.
(1966), A-lS, A-l7.

of Buying Power,

Tkjbin, James. "On the Predictive Value of Consumer
Intentions and Attitudes," Review of Economics and

Statistics (February, 1959), 1-11.

'vaedt, Dik Warren. "A Multiple Factor Analysis of
Advertising Readership," Journal of Applied Psychology,

XXXVI, No. 3 (June, 1952), 207-215.
"How Important to Marketing Strategy is the
Journal of Marketing (January, 1961),

'Heavy User?"
71-72e

 

"A Simple Math Model for Public

VVifllliams, RObert H.
LIX, No. 11

Utilities," Public Utilities Fortnightly,
(May 24, 1962), 728-732.

Other Sources

 

Federal Communications Commission, Investigation of the
fDelephone Industry in the United States (House Document

.3<40, 76th Congress, lst Session), washington, D.C.:
11.8. Government Printing Office, 1939.

GlOSsary of Telephonese published by Michigan Bell Telephone
Chompany.

EfiJfl<1&3, Charles L. Unpublished paper presented to the
PUHerican Marketing Association, International Congress,

Philadelphia, Pennsylvania, June 17-19, 1968.

253

Iflational Industrial Conference Board. Forecasting Sales.
New York: NICB, 1964.

 

Ddichigan Bell Telephone Company. Profile of the Michigan
Bell Consumer. A Report Prepared by the Opinion
Research Corporation and the Long—Range Planning Group
of Michigan Bell Telephone Company, 1967.

 

 

bdichigan Bell Telephone Company Records, 1967.

Phobertson, Thomas S. "An Analysis of Innovative Behavior
and Its Determinants.” Unpublished Ph.D. dissertation,
Northwestern University, 1967.

EI-S. Department of Commerce. Business Statistics. 1969
edition.

 

APPENDIX I

254

.mpcmemmm chmmao

mum pom nonammm mflu ow mHQmoHHmmm no: we :mmmmonfim UOHHmHmm: mannaum> Ones

 

 

R R O.m m.v v.H umCmmHO >HQ

a a m.N m. m.N Houammm

v.m b. m.m h.m h.m HOHHmumm

@.N O. m.h o.~ m.m HmHmmeOQB

m.m N.H m.om m.@ o.m Hmhsuommosmz

mmcHH mcoHumooq mwcHH mCOHumooq mmCHH aquamOm mmmcflmsm

\mmmmoamEm \mwmonmEm \mmmxoamEm \mmmwoamfim \mCOHumooq

OOHHMHmm pmHumamm

 

moHumm pmuomamm

 

 

Ezmzwmm mmMZHmDm Mm

mmqdem¢> BZMQZmQMQZH 024 ZOHmmBHMU QmBUmqmm mOm mOHfifim

H.H mqmflfi

mmqmfimag 9232335 32 203955 omeomqmm mom moHeHHm

H NHQmem¢

APPENDIX I I

255

APPENDIX II
KENDALL COEFFICIENT OF CONCORDANCE
FOR BUSINESS SEGMENTS ON THE TIME

AND IMPORTANCE PLACED ON VARIOUS
TELEPHONE COMMUNICATION ACTIVITIES

The values included in these tables are rank
averages and do not indicate either an absolute degree
of importance or amount of time attached to any one

activity.

256

.>HN>HDUM umnu so pammm mm3 mEHu umoE mflu mmuMONUCN ta: mo xcmm "HBOZ

.mmmcommmm Hmcmmao mun mcfiummUXMee
.Cm>Nm noncommmn 02*

 

NO. u 3 HH0.0 "RemochHchmHm

 

 

 

 

m m.m e e.e m m.m m O.N v m.m mmHuHHHumm gunmen .m

R N O.N v v.m e O.m m v.e ememmHmm .o

e m N.m m m.N N ¢.m m N.m mesmennmmmo H0300 .o

N H.N m O.m N N.N N O.N N N.N mumHHmmsm .m

H H.H H N.H H O.H H H.N H m.N mumsoumso .m

«fluNB mCHDMONCSEEOU

xcmm cam: xcmm 2mm: xcmm cmmz xomm cmwz xcmm cmmz mufi>flp0¢
mnmcmmao mum muouammm mumaflmumm mHmHMmmHOSZ mnmusuommonmz

ucmemmm Mmmcflmmm

 

 

 

mmHBH>HBU¢ ZOHBfiUHZDSEOU mDOHm¢> ZO
BZMQm MEHB m0<mD MZOEQMQMB QdUOA m0 wGZHMZ¢m Zflmz

HIHH mgmflfi

257

.>DH>Huom mCOHDMONCOEEOO HMHSOHUHMQ um3p ou usmauomafl

umoE mm3 mCOQQmHmD mg» umflu uamm pampoommmu map mODMUHUCN :H: mo Mann 4 "meoz

.momcommmu Hmsmmau mun msapmmomeR
.cm>Nm noncommmu OZR

 

mm. H 3 A.m.cv NeemUCMONMNcmNm

m H.e N e.m m H.N m O.N mmHDHHHomm Rectum .O
H ¢.H e O.N a N.N e O.N ememmHmm .o
e m.e H O.H H O.N H O.N mumHHmmsm .m
N O.N N N.N N O.N N N.N mumeOumso .a

"SDH3 mCHumONQOEEOU

 

Mcmm cmmz
muoammao hum

xcmm Cmmz Mcmm cmmz xcmm one: Mcmm Cmmz NDN>NHO<
mHOHHmmm mHmHNmuom mHmHMmmHOSB mHOHOpommocmz

 

ucofimmm wwmcflmom

 

 

 

 

 

 

J

 

mMHBH>HBU¢ ZOHB¢UHZDEZOU mDOHm¢> ZO
Bzmmm MZHB mwdmb MZOEQWAMB RACE m0 mUZHMZ¢M dez

NIHH mam¢9

258

.mmmcommmu Hmcmmau mun mcflummomee
.Cm>Hm noncommmn OZR

 

HO. u z HmO.O "eemucmonHamHm

e O.N v O.N m v.m mmHDHHHomm socmum .m
m N.e m N.N e O.N cmemmHmm .o
m O.N m m.¢ m O.N mucwsunmmmo Hmsno .o
N e.N N N.N N N.N mumHHmmOm .m
H H.H H m.H H m.H mumsoumso .a

”CHH3 mcHHMUNCOEEOU

 

xcmm :mmz
mumcmmau mum

xcmm Cmmz
mHOHHmmm

Mama one: xcmm awe: xamm Cmmz MHN>HHU¢
mHmHNmumm mumammmaosz mHmHOHUmmocmz

 

 

HomemOm mmmcflmom

 

 

n
V

 

mmHBH>HBU¢ ZOHBdUHZDZEOU mDOHMd5 ZO

 

mwde mZOSQMQMB A4009 m0 MUZ<BmOm2H m0 mOZHMZ¢m 24m2

mIHH mqmflfi

259

.mmmcommmu Hmcmwau who mCHHmmomee
.Cm>Hm uncommon 02*

 

mm. H 3 H.N.Qv "RROUCMUHMNGmNm

m e.N N N.N m N.N m O.N mmHuHHHomm acumen .O
H H.H e N.N e O.N e e.m cmemmHmm .o
e O.v H O.H N e.N N O.N mumHHmasm .m
N m.H N O.N H O.H H m.H mneEoumOo .m

"CHH3 mCHumOHCsEEoo

 

xcmm cmmz
mumcmmao mun

Mcmm Emu: xcmm cmmz Mcmm mez xcwm saw: muH>Huod
mnouHmmm mHmHHmumm mHmHMmmHOSZ mumuouommonmz

 

unmammm mmmcflmom

 

['11
1' J lllrlnl

mmHBH>H804 ZOHBfiUHZDSEOU mDOHmm> ZO
m0¢mb mzommeMB AAOB m0 mUZ¢Hm0mSH m0 mUZHMZ¢m Zfimz

 

vIHH mammB

APPENDIX I II

260

APPENDIX III

PERCENT OF FIRMS IN EACH SEGMENT
THAT PERCEIVED THEIR TELEPHONE USAGE
TO BE LESS THAN, EQUAL TO, OR GREATER
THAN FIRMS OF COMPARABLE TYPE AND SIZE

Scale Used:

= Far Below Average
Below.Average
Average

Above Average

Far Above Average

m-bwwH
ll

261

 

om do w mmH Q¢EOB

 

 

gm mm w No Hmammeosz

OH mm m we Honouommscmz

em no m we HwHHmumm

HH mm o m Hmcmwao hum

Rom Rum Xv mm Houamwm

mug m NIH mEHHm ucmfimmm

wmmum>¢ m>OQ¢ ommnm>< mmmum>< 30Hmm mo Hmnfioz mmmcflmom
x x R

xcmm

 

 

Bzmzwmm Nm I WNHm QZd mmMB mdflHSHm m0 mEmHm OB Qmmdmzoo
m4 m0<mD mzommeMB Q400A m0 BZDOE¢ Qm>HmUmmm

HIHHH mqm<9

262

 

mg mg ma mmH AfiHOB

 

 

mg mg w No HOHMmmHosz

gv mg HH he Honouommscmz

ow om 0H mg HmHHmuwm

mN we mm O HmcmmHo sun

va New New mm HouHmmm

mlv m NIH mEHHm ucmfimmm

mmmum>< m>ond mmmum>¢ mmmum>¢ BOHOm mo HOQEOZ mmmcflmom
.x. x R

xsmm

 

 

Bzmzwmm Mm I MNHm QZ¢ mmwfi m¢QHEHm m0
mEmHm OB ammmmzoo mfl QmmD mSHB AAOB m0 BZDOZ< Qm>HmUmmm

NIHHH mqmfifi

263

 

 

 

 

 

om Om MH HmH H4809
mm mm mH mo umHmmmHo£3
mm mm MH we Honouommocmz
Om mm O He HmHHmumm
O ON NN O 09808 .30
me Rom XMH mm HouHmmm
mue m NIH manm ucmsmmm
wmmum>< m>OQ¢ mmmuw>< mmmuw>< BOHOm mo HOQEDZ mmmcflmsm
x x x
xcmm
szmzomm Mm MNHm 9,2 mm? 6323 mo mam:

OB Ommémzoo m¢ QmmD BZMEQHDOM mzommmAMB m0 BZbOS< Qm>Hm0mmm

mIHHH mqmdfl

 

264

Hm mm HH HmH HdBOB

 

 

mm mm OH NO HOHMmmHo£B

mm mm O we Honouommscmz

we Hm m Hv HmHHmumm

HH mm o m HmammHU hum

XmH Rmm Xmm mm HouHmmm

mIv m NIH mEHHm unwammm

mmmum>¢ m>ond mmmum>< mmmnm>¢ Bonm mo HmQEoz mumsHmom
x x x

xcmm

 

 

Bzmzomm %m I mNHm QZd mmMB MdflHZHm m0 mEmHm OB Qmm¢m200
m4 mmmHHmmDm BU¢HZOU OB QmmD MEHB mZOmmmHmB m0 BZDOEd Qm>HmUmmm

WIHHH mqm¢H

265

 

 

 

mm mm m NmH Hmfioe
He me OH NO HmemmHoHHB
mm mm mH we Housuommocmz
mm mm m we HmHHmumm
HH mm HH 0 HmommHU hum
XHO Rom R0 mm HouHmmm
mlv m NIH mEHHm pcmammm
mmmuo>¢ m>094 mmmuo>¢ wmmuo>¢ BOHOm mo HmQEsz mmmchom
x x x
xcmm

 

 

Bzmzwmm Nm l MNHm 02¢ mmwfi mHHHZHm m0 mZMHm OB Qmmfimzou mfl mmMZOBmDU

BU¢BZOU OB QmmD mZHB mzommmHMB m0 EZDQE< Qm>Hm0mmm

mIHHH mam¢9

266

 

 

 

HN mm 0N NmH H¢HOB
mH we mm No umHmmmHoaz
v mm mm we Honsuomwscmz
Hm hm OH Ne HmHHmuwm
mm mm HH m HmsmmHU >HQ
New va XMH mN HouHmwm
mlv m NIH mEHHm unmammm
wmmum>¢ O>OQ¢ mmmnm>< mmmum>¢ 30Hmm mo HmQEsz mmmchom
x .x. x.
xcmm

 

Bzmzwmm Wm

 

WNHm 924 mmwfi m¢HHEHm ho WZMHm

 

OB

Qmm<m200 mfi QmmD UZHmHBmm>Q¢ Wwdm BOHHMN m0 BZDOE< Qm>Hm0mmm

QIHHH mqmflfi

APPENDIX IV

267

APPENDIX IV

ROTATED FACTOR LOADINGs* FOR EACH

CRITERION VARIABLE AND ASSOCIATED EXPLANATORY

VARIABLES - BY BUSINESS SEGMENT

TABLE IV-l

EXPLANATION OF ABBREVIATED VARIABLE SYMBOLS

USED IN TABLES IV-2 - IV-20

 

 

 

Symbol Variable Description

Adv$ Monthly Yellow Page Advertising billing

Totaltoll Dollar value of total toll usage for
February, 1968

Intratoll Dollar value of intra-state toll usage for
February, 1968

Intertoll Dollar value of inter-state toll usage for
February, 1968

Equip$ Monthly dollar amount of equipment billings

Numlines Number of trunks or lines used by a firm

Numloc Number of telephone locations maintained by
a firm

Numempl Total number of employees

Numclerks Number of clerks employed by a retail
establishment

Numclrem Number of clerical employees

 

*Note: For convenience, only the highest loading value
Of each explanatory variable on a factor is given in the

tables.

268

TABLE IV-l (Continued)

 

Symbolpy Variable Description

Numsalem Total number of salaried employees

Numsuppl Number of suppliers

NumsupplI Number of suppliers located outside of Lansing

Numsupp12 Number of suppliers located outside of Michigan

Sqfeet Square footage of customer facility

Numoff Number of offices

PctsupplL Percentage of suppliers located in Lansing

Pctsuple Percentage of suppliers located in Michigan

Grossale Total 1967 gross sales

Grossalel 1967 gross sales outside of Lansing

Grossale2 1967 gross sales outside of Michigan

Numbuild Number of buildings customer utilizes

PctsaleL Percentage of sales in Lansing area

PctsaleM Percentage of sales made in Michigan

Numordmo Average number of orders received per month

NumoutA Number Of additional outlets in Lansing

NumoutB Number of additional outlets in Michigan
outside of Lansing

NumoutO Number of additional outlets outside of

Michigan

269

TABLE IV-l (Continued)

 

 

Symbol Variable Description

Impsuppl Importance of telephone communications in
contacting suppliers

Tloccost Ranking of time spent communicating with
customers in the local calling area

Tlocsuppl Ranking of time spent communicating with
suppliers in the local calling area

Disthome Distance of Lansing facility from the home
office

Underst Customers perceived awareness or available
understanding of available telephone
communications alternatives

Pctmerwh Percentage of total revenue resulting from
wholesaling activities

Pctmfgr Percentage of total revenue resulting from

manufacturing activities

 

270

TABLE IV-2

ROTATED FACTOR LOADINGS FOR TOTAL TOLL DEMAND
AND ASSOCIATED EXPLANATORY VARIABLES -
MANUFACTURER SEGMENT

 

 

Variable

Totaltoll
Grossale2
Numoff
Underst
PctsaleL

Numsalem
Tloccust
Numordmo
Numbuild
Disthome

NumsupplI
Numsuppl
Numsupp12
Sqfeet
PctsupplL

NumoutB
NumoutA
Pctsuple
Grossale
Grossalel

Numempl
Numclrem
PctsaleM

Factor

1

-.798
-.813
-.672
.606
.595

-.482
-.442

Loading

Factor Factor

2 3

.303 .262

.698
.697
.410

.917
.826
.791
.618
-.391

Factor

4

.230

.929
.921
-.510

Factor Commu-
5 nality

-.084 .858

-.922
-.905

-.870
-.860
-.499

 

271

TABLE IV-3

ROTATED FACTOR LOADINGS FOR INTRASTATE TOLL
DEMAND AND ASSOCIATED EXPLANATORY VARIABLES -

MANUFACTURER SEGMENT

 

Variable

Intratoll
Numordmo
Pctmerwh
Numbuild
Pctmfgr

Disthome
GrossalesZ
PctsaleL
Underst
Numoff

PctsupplL
Tloccust
NumsupplI
Numsuppl
Numsupp12

Sqfeet
Numsalem
Grossale
Grossalel
Numempl

Numclrem
PctsaleM
NumoutB
NumoutA
Pctsuple

Factor

.691
.762
.716
.703
-.469

.393

Loading
Factor Factor Factor Factor
2 3 4 5

.116

Commu-
nality

.493 .284 -.148 .837

.765
-.639
-.615

.594

-.470
.437
.918
.835
.780

.612

.456
-.916
-.897
-.873

-.868
-.493
.934
.926
-.510

272

TABLE IV-4

ROTATED FACTOR LOADINGS FOR INTERSTATE TOLL
DEMAND AND ASSOCIATED EXPLANATORY VARIABLES -
MANUFACTURER SEGMENT

 

 

Loading
Factor Factor Factor Factor Factor Commu—
Variable 1 2 3 4 5 nality

 

Intertoll -.828 -.307 -.229 .101 -.033 .843
Grossale2 -.807
Numoff -.69l
Underst .624
PctsaleL .561

Tloccust -.447

NumoutB -.930

NumoutA -.922
Pctsuple .515
Numsupp12 -.921

NumsupplI -.826
Numsuppl -.800
Sqfeet -.650
Numsalem -.508
PctsupplL .390

Numordmo .799
Pctmerwh .720
Numbuild .666
Pctmfgr -.554
Disthome .395

Grossale —.918
Grossalel -.902
Numempl -.850
Numclrem -.846
PctsaleM -.494

 

273

TABLE IV-5

ROTATED FACTOR LOADINGS FOR EQUIPMENT DEMAND AND
ASSOCIATED EXPLANATORY VARIABLES - MANUFACTURER SEGMENT

 

Variable

Equip$
Numoff
Grossale2
Underst
PctsaleL

Tloccust
Numsuppl
NumsupplI
Numsupp12
Sqfeet

PctsupplL
Numbuild
Pctmerwh
Numordmo
Pctmerwh

Disthome
Numsalem
NumoutB
NumoutA
Pctsuple

Grossale
GrossaleI
Numclrem
Numempl
PctsaleM

Factor
1

.720
-.737
-.714

.640

.533

-.533

Loading
Factor Factor Factor Factor
2 3 4 5

Commu-
nality
.398 .172 -.078

-.021 .712

.909
.826
.773
.636

-.347
.788
.697
.658
-.596

.393
.941
.925
.911
-.411

-.917
-.905
-.863
-.847
-.541

 

274

TABLE IV-6

ROTATED FACTOR LOADINGS FOR NUMBER OF LINES
DEMANDED AND ASSOCIATED EXPLANATORY VARIABLES -
MANUFACTURER SEGMENT

 

Loading Commu-
Variable Factor 1 Factor 2 Factor 3 Factor 4 nality

 

Numlines —.620 .417 .295 -.10 .668
Numoff -.737
Grossale2 —.712
Underst .654
PctsaleL .577

Tloccust —.505
PctsupplL .354
. Numsuppl .910
NumsupplI .829
Numsupp12 .772

Sqfeet .646
Numsalem .502
Numordmo .778
Pctmerwh .675
Numbuild .639

Pctmfgr -.622
Pctsuple -.446
Disthome .405
Numclrem -.920
Grossale -.904

GrossaleI —.858
Numempl -.857
PctsaleM -.509

 

275

TABLE IV-7

ROTATED FACTOR LOADINGS FOR NUMBER OF LOCATIONS
DEMANDED AND ASSOCIATED EXPLANATORY VARIABLES -

 

MANUFACTURER SEGMENT

 

 

 

Variable

Numloc
Numoff
Grossale2
Underst
PctsaleL

Tloccust
PctsupplL
Numsuppl
NumsupplI
Numsupp12

Sqfeet
Numordmo
Pctmerwh
Numbuild
Pctmfgr

Pctsuple
Disthome
Numsalem
NumoutB
NumoutA

Grossale
GrossaleI
Numclrem
Numempl
PctsaleM

Loading
Factor Factor Factor Factor Factor
1 2 3 4 5

Commu—
nality

.668
.749
.679
-.674
-.576

.499 .242 -.067 -.049 .761

.524
-.375
.902
.828
.755

.655
.796
.683
.623
-.622

-.443
.404
.940
.905
.899

-.918
—.905
-.848
-.843
-.536

 

276

TABLE IV-8

ROTATED FACTOR LOADINGS FOR TOTAL TOLL DEMAND AND
ASSOCIATED EXPLANATORY VARIABLES - WHOLESALER SEGMENT

 

 

 

 

Loading
Variable Factor 1 Factor 2 Factor 3 Communalityi
Totaltoll -.885 .267 .071 .860
Grossale -.968
Numclrem —.959
GrossaleI -.947
Numempl ~.9l9
Numsalem -.899
Sqfeet -.565
Numoff -.555
PctsaleL -.928
PctsaleM .921
Impsupp .829
Grossale2 .809
Tlocsupp -.222

 

277

TABLE IV-9

ROTATED FACTOR LOADINGS FOR INTRASTATE TOLL
DEMAND AND ASSOCIATED EXPLANATORY VARIABLES -
WHOLESALER SEGMENT

 

 

 

Loading
Variable Factor 1 Factor 2 Factor 3 Communality__
Intratoll -.877 -.305 -.013 .862
Grossale -.965
Numclrem -.958
GrossaleI -.947
Numempl -.924
Numsalem -.89l
Sqfeet -.555
Numoff -.545
Impsupp -.739
Grossale2 -.531
Tlocsupp .294
PctsaleM .909
PctsaleL -.907

 

278

TABLE IV-lO

ROTATED FACTOR LOADINGS FOR INTERSTATE TOLL
DEMAND AND ASSOCIATED EXPLANATORY VARIABLES -
WHOLESALER SEGMENT

 

 

 

 

 

Loading
Variable Factor 1 Factor 2 Factor 3 Communality
Intertoll -.720 —.352 -.250 .705
Grossale -.972
Numclrem -.944
GrossaleI -.932
Numsalem -.922
Numempl -.904
Sqfeet -.603
Numoff -.596
PctsaleL .937
PctsaleM -.905
NumoutO -.856
Grossale2 -.655
Tlocsuppl .258

 

279

TABLE IV-ll

ROTATED FACTOR LOADINGS FOR EQUIPMENT DEMAND

AND ASSOCIATED EXPLANATORY VARIABLES

WHOLESALER SEGMENT

 

 

 

 

Loading_
Variable Factor 1 Factor 2 Communality
Equip$ .933 -.110 .883
Numempl -.958
Numclrem .949
Grossale -.936
GrossaleI -.928
Numsalem -.885
Sqfeet -.562
Numoff -.56l
Tmpsuppl -.320
PctsaleL .945
PctsaleM -.907
Tlocsuppl .50
Grossale2 -.l63

 

280

TABLE IV-12

ROTATED FACTOR LOADINGS FOR NUMBER OF LINES
DEMANDED AND ASSOCIATED EXPLANATORY VARIABLES -
WHOLESALER SEGMENT

 

 

 

 

Loading
Variable Factor 1 Factor 2 Communality
Numlines -.903 .107 .828
Grossale -.960
Numclrem -.945
GrossaleI -.93l
Numempl -.918
Numsalem -.893
Numoff -.576
Sqfeet -.572
Impsuppl -.322
PctsaleL —.943
PctsaleM .909
Tlocsuppl -.255

Grossale2 .157

 

281

TABLE IV-l3

ROTATED FACTOR LOADINGS FOR NUMBER OF LOCATIONS
DEMANDED AND ASSOCIATED EXPLANATORY VARIABLES -
WHOLESALER SEGMENT

 

 

 

 

 

 

 

Loading
Variable Factor 1 Factor 2 Factor 3 Communality
Numloc -.941 .156 —.039 .911
Grossale -.969
Numclrem -.950
GrossaleI -.938
Numempl -.929
Numsalem —.901
Numoff -.563
Impsuppl -.340
PctsaleM .920
PctsaleL -.9l9
Tlocsuppl .229
Sqfeet .853
Grossale2 -.58O

 

282

TABLE IV-14

ROTATED FACTOR LOADINGS FOR YELLOW PAGE

ADVERTISING DEMAND AND ASSOCIATED EXPLANATORY
VARIABLES - RETAILER SEGMENT

 

 

 

 

Loading,

Factor Factor Factor Factor Factor Commu-
Variable l 2 3 4 5 nality
Adv$ -.646 -.388 .147 —.106 .025 .601
Numempl -.972
Numclerks -.952
Numclrem -.924
Numsupp12 -.903
Numsuppl -.880
Sqfeet -.857
Numoff -.831
PctsupplI -.785
Grossale -.690
Numsalem -.680
Numbuild .846
Tloccust .756
PctsaleL -.775
Pctsuple .741
Tlocsuppl .766

 

283

TABLE IV-lS

ROTATED FACTOR LOADINGS FOR TOTAL TOLL DEMAND
AND ASSOCIATED EXPLANATORY VARIABLES - RETAILER SEGMENT

 

 

 

 

Loading

Factor Factor Factor Factor Factor Commu-
Variable 1 2 3 4 5 nality
Totaltoll -.799 -.307 -.l92 .134 -.177 .819
Numclerks -.968
Numepl -.952
Numclrem -.931
Numsupp12 -.895
Numsuppl -.87l
Sqfeet -.837
PctsaleL -.857
Pctsuple -.637
Numoff -.832
PctsupplL -.806
Grossale -.693
Numslaem -.686
Numbuild .820
Tloccust .723
Tlocsuppl .834

 

TABLE IV-16

284

ROTATED FACTOR LOADINGS FOR INTRASTATE TOLL
DEMAND AND ASSOCIATED EXPLANATORY VARIABLES -

RETAILER SEGMENT

 

 

 

 

Loadingp

Factor Factor Factor Factor Factor Commu-
Variable 1 2 3 4 5 nality
Intratoll -.612 -.592 -.283 -.155 -.089 .837
Numclerks -.978
Numempl -.952
Numclrem -.928
Numsupp12 -.912
Numsuppl -.890
Sqfeet -.847
PctsaleL .877
Pctsuple -.56l
Numoff —.83l
PctsupplL -.801
Grossale -.696
Numsalem —.69l
PctsaleM .815
Numbuild .844
Tloccust .718

 

TABLE IV-17

285

ROTATED FACTOR LOADINGS FOR INTERSTATE TOLL DEMAND
AND ASSOCIATED EXPLANATORY VARIABLES - RETAILER SEGMENT

 

 

 

 

Loading

Factor Factor Factor Factor Factor Commu-
Variable 1 2 3 4 5 nality
Intertoll -.821 .242 -.108 -.090 -.071 .758
Numclrem -.965
Numclerks -.946
Numempl -.928
Numsupp12 —.896
Numsuppl -.871
Sqfeet -.83l
Numbuild -.803
Tloccust -.721
Tlocsuppl -.847
Numoff -.825
PctsupplL -.812
Grossale -.709
Numsalem -.702
PctsaleL .787
Pctsuple .736

 

286

TABLE IV-18

ROTATED FACTOR LOADINGS FOR EQUIPMENT DEMAND AND
ASSOCIATED EXPLANATORY VARIABLES - RETAILER SEGMENT

 

 

 

 

Loadingyi

Factor Factor Factor Factor Factor Commu-
Variable 1 2 3 4 5 nality
Equip$ -.947 —.244 .146 .022 .015 .979
Numclerks -.978
Numempl -.957
Numclrem -.930
Numsupplz -.912
Numsuppl -.890
Tlocsuppl -.855
Numoff -.831
PctsupplL -.831
Grossale -.689
Numsalem -.682
Numbuild .853
Tloccust .733
PctsaleL .790
Pctsuple -.732
Sqfeet .813

 

287

TABLE IV-l9

ROTATED FACTOR LOADINGS FOR NUMBER OF LINES DEMANDED
AND ASSOCIATED EXPLANATORY VARIABLES - RETAILER SEGMENT

 

 

 

 

Loading

Factor Factor Factor Factor Factor Commu-
Variable l 2 3 4 5 nalipy
Numlines -.951 -.173 .148 .016 .010 .956
Numclerks -.977
Numempl -.951
Numclrem -.922
Numsupp12 -.9l8
Numsuppl -.896
Tlocsuppl -.852
Numoff -.829
PctsupplL —.806
Grossale -.697
Numsalem —.690
Numbuild .854
Tloccust .732
Sqfeet .815
PctsaleL .796

Pctsuple -.726

 

288

TABLE IV-ZO

ROTATED FACTOR LOADINGS FOR NUMBER OF LOCATIONS
DEMANDED AND ASSOCIATED EXPLANATORY VARIABLES -

RETAILER SEGMENT

 

 

 

Loading

Factor Factor Factor Factor Factor Commu-
Variable l 2 3 4 5 nality
Numloc —.971 -.158 .133 .049 —.038 .984
Numclerks -.979
Numempl -.952
Numclrem -.923
Numsupp12 -.9l9
Numsuppl -.849
Numsalem -.826
PctsupplL -.808
Grossale -.702
Numoff -.697
Numbuild .853
Tloccust .736
PctsaleL .788
Pctsuple -.733
Sqfeet -.811

 

 

APPENDIX V

 

 

289

APPENDIX V

COVER LETTERS, PERMISSION REQUEST FORMS,
AND QUESTIONNAIRES

 

 

290

STUDY OF COMMUNICATION NEEDS IN BUSINESS

MICHIGAN STATE UNIVERSITY
Department of Marketing and Transportation Administration

Graduate School of Business Administration
315 Eppley Center East Lansing, Michigan 48823

Ineifew days you will be receiving a questionnaire concerning the
rohathat telephone communications plays in certain aspects of your
The data you supply in the questionnaire will provide the

Any informa- -f

business.
inflnmmtion needed to complete my doctoral dissertation.

tionIfluch you provide will be considered strictly confidential.

The study is being undertaken with the combined support of the Michigan

Berllelephone Company and the Institute of Public Utilities at
The results of the research should aid

Michigan State University.
telephone companies in better meeting the future needs of firms

such as yours.

Many factors influence the types of telephone equipment and services
a particular business firm requires. The questionnaire, therefore,
is quite detailed. As you well realize, however, your response is
necessary if the study is to be a success and contribute to a better
understanding of the communication problems facing today's business

firms.

Should you wish an advance summary of the results, please indicate
so by a note on your questionnaire. I would be happy to send you

this as advance information.

Your cooperation is sincerely appreciated.

Yours truly,

Richard F. Sauter
Research Director

291

STUDY OF COMMUNICATION NEEDS IN BUSINESS

MICHIGAN STATE UNIVERSITY
Department of Marketing and Transportation AdministratIon

Graduate School of Business Administration
East Lansing. Michigan 48823

3l5 Eppley Center

With the aid of a grant from the Ford Foundation, I am presently engaged in
research for my doctoral dissertation to determine how various business factors
The study

affect the demand for certain communications products and services.

will investigate such problems as what future communication needs will be in

certain segments Of the business community, how these needs can best be met,
’ I hope

and what factors will be most influential in determining these needs.
you will complete the enclosed questionnaire as it is vital to aw doctoral
research and to nw Obtaining an accurate understanding of these relationships.

Since there are a large number of factors that may influence comunication
needs, some of the questions included here will ask you to provide detailed
All of the information will

information on certain aspects of your business.
be held in the strictest confidence and the name of your firm will not appear
Data will be lumped together and no infor-

anywhere in the research findings.
nation will be analyzed on an individual firm basis.

Since the basic purpose of this study is to identity those factors which deter-
mine a firm 3 need for communication offerings, it is necessary to secure data

0 e '
from Michigan Bell concerning the types of equipment and services your fine
A release form is enclosed which I hope you will sign and
Upon receipt of your release
Let

presently utilizes.
return along with the completed questionnaire.
Michigan Bell has agreed to furnish the data listed on the release form.

me reemphasize that no data will be sought or collected concerning either
individual local _o_r long distance calls nor will your firm' 8 name be associated

in any manner with the information you release.
If you have any questions regarding any aspect of the questionnaire or study,
please feel free to call me at either 355-1619 or 355-5837.

As you know, the success of this research will depend upon your cooperation,

and your assistance will be greatly appreciated.

I have enclosed a self-addressed envelope for your convenience.

Thank you for your time and effort.

Sincerely yours ,

Richard F. Sauter
Research Director

 

 

292

STUDY OF COMMUNICATION NEEDS IN BUSINESS

MICHIGAN STATE UNIVERSITY
Department of Marketing and Transportation Administration

Graduate School of Business Administration
East Lansing, Michigan 48823

315 EppIey Center

June 25, 1968

You will remember my request for assistance with a study I am conducting

concerning the role that telephone communications plays in certain types
Preliminary tabulation of responses is providing

of business firms.
valuable information about these questions.

I hope to begin final tabulation shortly and need your assistance to
complete the data collection. I would, therefore, appreciate your reply
at your earliest convenience. If for some reason the earlier question-
naire did not reach you or was mislaid I have enclosed a duplicate.
I want to reemphasize that any information which you provide will be
held in the strictest confidence. Should you wish an advance summary
of the results, please indicate so by a note on the questionnaire and
I will be happy to send you this as advance information.

As you know the validity of the findings will depend on your coopera-

tion. Thank you for your assistance.

Very truly yours,

Richard F. Sauter
Research Director

 

 

293

Permission Request

thy] have your permission to obtain data from Michigan Bell records
concerning:

 

A. Total monthly toll billings of your firm
for the past 3 months.

B. Total local service billings of your firm
for the past 3 months.

 

C. The types of telephone equipment and serv-
ices your firm presently utilizes.

to be used in a research study I am presently. undertaking at Michigan
State University. The name of your firm will not be associated with
this data in any aspect of the research.

Thank you for your cooperation.

Richard F. Sauter
Research Director

Mr. Sauter:

You have our permission to obtain the information mentioned above
from Michigan Bell records.

Date Signature

 

Title

 

STUDY OF COMMUNICATION NEEDS IN BUSINESS

 

MANUFACTURER/WHOLESALER INFORMATION
(Strictly Confidential)
Graduate School of Business Administration
Department of Marketing and Transportation
Michigan State University

The questionnaire is divided into two parts. Part A deals with certain characteristics related to your firm, its
customers and its suppliers that may affect your need for various types of communications offerings.
Part B of the questionnaire is primarily concerned with your evaluation of the present and future role of telephone

communications in various areas of your firm‘s operations. Your responses to this portion of the questionnaire
will aid us in determining where telephone communication needs are geatest and how these needs vary from one

sector of the business community to another.
Please note that the responses given in this questionnaire will be considered strictly confidential and will not in

any way be identified with any individual or firm.
The entire questionnaire should take approximately 20—30 minutes to complete. Your cooperation will be most

sincerely appreciated.

 

 

295

 

UNITED STATES
OF AMERICA
SCALE or M

terse

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I1,
I
’- art
a.
fig? rat
I
L,
C

CITIES INCLUDED IN GREATER LANSING
IANSING EATON RAPIDS
EAST LANflI‘K‘u GRAND LEDGE
I'IASLETT
omos
BATH
DEWI'I'T

MONDALE

 

“I.” GREATER LANSII‘K‘:
“M" MICHIGAN. EXCLUDM
GREATER LANSING

“A" 0300 MILE RADIUS
FROM LANSING.
EXCLUDING MICHIGAN

“3" 300-700 MILE RADIUS

—'-'

 

 

 

 

 

296

PART A

Unless stated otherwise, the following questions refer only to the facility or
plant located at the address to which this questionnaire was mailed. Any ques-
tions pertaining to branch plants located elsewhere in the city, state or country
will be clearly identified. Thus unless stated specifically in the question, any
information given should refer only to the facility to which this questionnaire

was addressed.
How long has your firm been operating in its present facility?_____ yrs.

How many separate or adjoining buildings do you have at this location?

 

 

(excluding garages, sheds, etc.)

 

Approximately, how many square feet of floor Space does your physical

plant encompass? .........................................................

 

4- How many separate offices do you have in the buildings)? ........................

 

 

 

5. What were your gross sales during the fiscal year I967?

What has been the average number of employees working at this location

 

during the past year? ...........................................

 

 

7. How many of this total were salaried employees? ............ --

Of your total number of employees, how many on the average during the

past year have been employed in each of the following areas? ....................
Past Year

 

Administrative and Supervisory

 

Production

Engineering

Clerical

Sales

Other (please specify)
Approximately, how many different suppliers do you presently receive

. products from?
Approximately how many customers do you ship or sell products to?

 

 

 

 

IO.

1 l . Approximately how many orders are placed with your firm on an average

every month?
Of this total, what per cent of these orders would you estimate are taken by.

12.
a) mail at this location ___%
b) phone at this location %
—__%

c) other (please Specify)

13. How many salesmen do you presently have working in the field? .............

14. Does your firm presently utilize telephone lines for data transmission as
well as vocal communication? Yes __ No
If yes, approximately how many hours per month are used for strictly

1 5.
data transmission activities?

 

 

 

 

 

 

297

The following questions should be answered by referring to the enclosed map
(see page 2 indicating various labeled geographic regions). Please approximate
the values asked for in the following 3 questions to the best of your ability.

16. What per cent of your sales are made to customers:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3) within the Greater Lansing Area (“L”) .............................................. %

b) outside the Greater Lansing Area but within Michigan (“M”) ............ %

c) within region “A” .............................................................................. %

d) within region “B" .............................................................................. 7c

e) within region “C” .............................................................................. %

0 outside U.S.A. “O" ............................................................................ %
Total = 100%

I7. What per cent of your suppliers are located in each of the following areas?

a) within the Greater Lansing Area (“L”) ....................... _ %

b) outside the Greater Lansing Area but within Michigan (“M”) ............ %

c) within region “A" ..................................... %
d) within region “B” .................................................. - - ......... %
e) within region “C” ........................................ %
0 outside U.S.A. “O” ............................................................................ 70
Total = 100%

l 8. Using the enclosed map, please indicate 9“?" 9“?“
in the apprOpriate column the number “$3 within“: Region Region Region 0&5:
«on

of additional outlets of facilities your ,, ,, , . “ ” a .. ‘ ”
firm maintains in each geographical L ”Mum A B ‘C

 

 

area. ............. . ..........
19. In what city and state are your main offices located?

state

 

 

city .....

20. What are the central telephone numbers to which your fum is billed? (Please include the number of any

private lines or WATS lines to which separate billings are made.)

21.
tage of your total revenue derived from each.

% of total revenue
derived from each

activity
Retailer ................................ % Agent Middleman
Merchant Wholesaler ................................ % Manufacturer
Manufacturers Agent ................................ % Other (please specify)

22. Under what Standard Industrial Classification (SIC) code (5) is your company listed?

% of total revenue
derived from each
activity

Please indicate below the category (s) into which you would place your firm and if more than one, the percen-

 

 

Major SIC Code ..................... Minor SIC Code (s)

23 . Position of person (s) completing questionnaire.

 

 

 

1'

 

298
PART B

The following questions concern your own personal attitudes and Opinions
toward the value of telephone communications — both in general and in
respect to the value of specific communication offerings to certain aspects
of your firm ’s operations.

(Table for Questions 1 — S)

A. Communicating with customers E. Communicating with branch facilities
and/or the home office

g with salesmen

B. Communicating with suppliers
F. Utilizing telephone lines for data

C. Interdepartmental communiations transmission (other than voice)

facilities and/or the home office

data transmission (otha' than

misc.)
6. Other (pleas specify)

communication

in the field

E. Communicating with branch

CUSlOMCIS

B. Communicating with

suppliers
I". Utilizing telephone lines for

A. Communicating with
C. Interdepartmental
D. Oommunicatin

 

D. Communicating with salesmen in the C. Other (please specify)
field
I. Considering telephone usage in the local calling area only (i.e. no toll
charge), please rank the following activities from I to 7 in terms of
the amount of time your firm spends in the average month on each
activity in relation to the others. (I being the most time; 7 being the

least; 0 if no time at all) ......................................................................

 

 

. Considering long distance telephone usage only. please rank the fol-
lowing activities from I to 7 in terms of the amount of time your
firm spends in the average month on each activity in relation to the
others. (I being the most time; 7 being the least; 0 if no time at all)

I“)

 

Please rank the following activities from I to 7 in terms of where you
feel telephone communications will make its most important contri-
bution to your firm 5 years from today. (I being most important;
7 being least important; 0 if no contribution ......................................

L»

 

4- If in-state long distance telephone rates were to be increased by 20%,
by what per cent would you expect your usage in each of the fol—

lowing areas to decrease? (Write 0 if you would expect no decrease.)
% decrease

 

5, If inter-state long distance telephone rates were to be increased by
20%, by what per cent would you expect your usage in each of the
following areas to decrease? (Write 0 if you would expect no decrease)

% decrease

6. Compared to other firms of a similar type and size, where would rank your firm in terms of :

A. local telephone usage

 

 

 

 

 

 

 

 

 

far above average ......... above average ......... average ......... below average ......... far below average ........

8. long distance telephone usage
far above average ......... above average ......... average ......... below average ......... far below average ........

C. the amount of telephone equipment you presently utilize
far above average ......... above average ......... average ......... below average ......... far below average ........

D. the amount of yellow page advertising you presently subscribe to
far above average ......... above average ......... average ......... below average ......... far below average ........

E. utilizing telephone wires for data transmission
far above average ......... above average ......... average ......... below average ......... far below average ........

F. utilizing the telephone to contact suppliers
far above average ......... above average ......... average ......... below average ......... far below average ........

G. utilizing the telephone to contact customers
far above average ......... above average ......... average ......... below average ......... far below average ........

H. your salesrnens’ use of the telephone
far above average ......... above average ......... average ......... below average ......... far below average ........

 

299

Which of the following statements most accurately describes your understanding of the different types of
equipment, services, and rate structures Michigan Bell offers to business firms? (Check only one.)

I. We have an excellent understanding of
the various alternatives available.
2. We have a good understanding of these

____4. We have a poor understanding of these
alternatives.

 

alternatives. __5. We have no knowledge of these alter-
____3. We have a fair understanding of these natives.
alternatives.

Which of the following statements best describes the process you would go through prior to changing all or

part of your communication system? (Assume this to be a major change such as the adoption of WATS ora
private line.) Check only one.

 

 

 

1. We would determine our needs and specify the types of equipment and/or services that would best
satisfy these needs.

2. We would determine our needs and let Michigan Bell consultants specify the types of equipment
and/or services that would best satisfy these needs.

3. We would determine our needs, suggest the types of equipment and/or services that might best
satisfy these needs, but leave the final decision up to Michian Bell consultants.

4.

 

We would let Michigan Bell consultants determine our needs and specify the types of equipment
and/or services that would best satisfy these needs.

Please rank the following phrases from 1 to 5 in terms of how you feel they reflect your firm‘s attitude toward
the value of telephone communications. (1 being most important; 5 being least important.)

___A. A necessary service to our customers and

suppliers. ___D. A competitive tool
_______B. A cost-saving service.
_-__C. A primary aid to developing new cus- ____E. Other (please specify)

tomers

 

”r

 

 

300

STUDY OF COMMUNICATION NEEDS IN BUSINESS

 

RETAILER INFORMATION

(Strictly Confidential)
Graduate School of Business Administration
Department of Marketing and Transportation
Michigan State University

The questionnaire is divided into two parts. Part A deals with certain characteristics related to your firm. its
customers and its suppliers that may affect your need for various types of communications offerings. .

Part B of the questionnaire is primarily concerned with your evaluation of the present and future role of telephone
communications in various areas of your firm’s operations. Your responses to this portion of the questionnaire
will aid us in determining where telephone communication needs are greatest and how these needs vary from one
sector of the business community to another.

Please note that the responses given in this questionnaire will be considered strictly confidential and will not in
any way be identified with any individual or firm.

The entire questionnaire should take approximately 20-30 minutes to complete. Your cooperation will be most
sincerely appreciated.

301

 

 

UNIIED StAtES
Of AMERICA
seat: or an:

o m an a. an

h4_.L__I__J

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

g . . .4-
wt; c
cmrs manor-:0 m cmrea unsmc
MNSING EATON RAPIDS
EAST LANSING GRAND LEDGE
HASLETT IDLT
omos unacswnc
BATH MASON
newm POTTERVILLE

 

“L” GREATER LANSM

"M" MICHIGAN. EXCLUDING
GREATER LANSING

“A” 0300 MILE RADIUS
FROM MNSING.
EXCLUDING MICHIGAN

“3" 300-700 MILE RADIUS
FROM MNSING

“C" REST OF THE U.S.A.

 

 

 

302

PART A

Unless stated otherwise, the following questions refer only to the facility or
plant located at the address to which this questionnaire was mailed. Any ques-
tions pertaining to branch plants located elsewhere in the city, state or country
will be clearly identified. Thus unless stated specifically in the question, any
information given should refer only to the facility to which this questionnaire
was addressed.

1.

2

10.

ll.

. What were your gross sales during the fiscal year 1967?

How long has your firm been operating in its present facility?____.__ yrs.

How many separate or adjoining buildings do you have at this location?
(excluding garages, sheds, etc.)

 

Approximately, how many square feet of floor space does your physical
plant encompass? ...........................................

How many separate offices do you have in the building(s)?-

 

 

 

 

What has been the average number of employees working at this location
during the past year?

oooooooooooooooooooooooooooooooooooooooooooooooooooooo

 

How many of this total were salaried employees?

 

Of your total number of employees working at this location, how many
on the average during the past year have been employed in each of the
following areas? ...................................... --

 

Past Year
Administrative and Supervisory Positions
Clerical
Sales Clerks -
Buyers
Other (please specify)

Approximately, how many different suppliers do you presently receive
products from?

 

Does your firm maintain a mail order department?

Yes __ No_____

If yes, what per cent of your total gross yearly sales are derived from
mail order business? %

The following questions should be answered by referring to the enclosed map
(see page 2) indicating various labeled geographic regions. Please approximate
the values asked for in the following 3 questions to the best of your ability.

12.

What per cent of your sales are made to customers:
a) within the Greater Lansing Area (“L”)

b) outside the Greater Lansing Area but within Michigan (“M”) ...........
c) within region “A” - -

(1) within region “B”
e) within region “C” - -
f) outside U.S.A. “O"

 

 

 

 

 

Total =

 

 

 

 

 

 

 

§$$$$$B§

 

303

13. What per cent of your suppliers are located in each of the following areas?

I4.

 

 

 

 

 

 

 

 

3) within the Greater Lansing Area (“L”) ............. - %
b) outside the Greater Lansing Area but within Michigan (“M”) ............ %
c) within region “A” ................... - - - - - . ........... %
(1) within region “B” .............................................................................. %
~ e) within region “C” .............................................................................. %
0 outside U.S.A. “0" ............................................................................ %
Total = 100%

Using the enclosed map, please indicate an ‘93:: Outside

IS.

I7.

18.

19.

in the appropriate column the number . . . . .
. . . . . Area WIIInn Region Region Region U.S.A.
of additional outlets of facrlmes your “Ln Michigan “A" “3.. «c» “on

firm maintains in each geographical
ma. ......

 

 

 

 

In what city and state are your main offices located?

 

city - - -- -- -- state

 

 

 

 

What are the central telephone numbers to which your firm is billed? (Please include the number of any
private lines or WATS lines to which separate billings are made.)

 

 

 

0.0.0.00— -3v“ ---------------

Please indicate below the category (5) into which you would place your firm and if more than one, the percen-
tage of your total revenue derived from each.

 

% of total revenue % of total revenue
derived from each derived from each
activity activity
Retailer ................................ % Agent Middleman ................................ %
Merchant Wholesaler ................................ % Manufacturer - - - %
Manufacturers Agent ...... -- - % Other (please specify) ................................ %

 

Under what Standard Industrial Classification (SIC) code (5) is your company listed?
Major SIC Code ..................... Minor SIC Code (s) ...................................... .. ........................................

Position of person (s) completing questionnaire.

 

 

3 04
PART B

The following questions concern your own personal attitudes and opinions
toward the value of telephone communications - both in general and in
reSpect to the value of specific communication offerings to certain aspects
of your firm's operations.

(Table for Questions I — 5)

A. Communiating with customers E. Communicating with branch facilities
and/or the home office
8. Communicating with suppliers
F. Utilizing telephone lines for data
C. Interdepartmental communications mnsmision (other than voice)

D. Communicating with salesmen in the C. Other (please specify)
field
I. Considering telephone usage in the local calling area only (i.e. no toll
charge), please rank the following activities from I to 7 in terms of
the amount of time your firm Spends in the average month on each
activity in relation to the others. (I being the most time; 7 being the
least; 0 if no time at all) ......................................................................

2. Considering long distance telephone usage only. please rank the fol—
lowing activities from I to 7 in terms of the amount of time your
firm Spends in the average month on each activity in relation to the
others. (1 being the most time; 7 being the least; 0 if no time at all)

3 Please rank the following activities from I to 7 in terms of where you
feel telephone communications will make its most important contri-
bution to your firm 5 years from today. (I being most important;
7 being least important; 0 if no contribution ......................................

If in—state long distance telephone rates were to be increased by 20%,
by what per cent would you expect your usage in each of the fol-
lowing areas to decrease? (Write 0 if you would expect no decrease.)

%decrease

5. If inter-state long distance telephone rates were to be increased by
20%, by what per cent would you expect your usage in each of the
following areas to decrease? (Write 0 if you would expect no decrease)

%decrease

A. Communicating with
customers

B. Communicating with
suppliers

 

facilities andlor the home office
P. Utilizing telephone lines for

data transmission (otha than

communication
D. Communicating with Cleaner:

in the field

E. Communicating with branch

G.0th¢(plaaaaapacify)

C. Interdepartmental

 

 

 

 

 

 

 

 

 

 

 

 

6- Compared to other firms of a similar type and size, where would rank your firm in terms of :

A. local telephone usage

far above average ......... above average ......... average ......... below average .........

8. long distance telephone usage

far above average ......... above average ......... average ......... below average .........

C. the amount of telephone equipment you presently utilize

far above average ......... above average ......... average ......... below average .........

D. the amount of yellow page advertising you presently subscribe to

far above average ......... above average ......... average ......... below average .........

E. utilizing telephone wires for data transmission

far above average ......... above average ......... average ......... below average .........

F. utilizing the telephone to contact supplier's

far above average ......... above average ......... average ......... below average .........

G. utilizing the telephone to contact customers

far above average ......... above average ......... average ......... below average .........

H. your salesmens’ use of the telephone

far above average ......... above average. ........ average ......... below average .........

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

 

305

7. Which of the following statements most accurately describes your understanding of the different types of
equipment, services, and rate structures Michigan Bell offers to business firms? (Check only one.)

 

l.
____2.
__3.

We have an excellent understanding of ___4. We have a poor understanding of these
the various alternatives available. alternatives.

We have a good understanding of these '

alternatives. ___5. We have no knowledge of these alter-
We have a fair understanding of these natives.

alternatives.

Which of the following statements best describes the process you would go through prior to changing all or

part of your communication system? (Assume this to be a major change such as the adoption of WATS or a
private line.) Check only one.

__I.

___2.
____3.
____4.

We would determine our needs and specify the types of equipment and/or services that would best
satisfy these needs.

We would determine our needs and let Michigan Bell consultants Specify the types of equipment
and/or services that would best satisfy these needs.

We would determine our needs, suggest the types of equipment and/or services that might best
satisfy these needs, but leave the final decision up to Michian Bell consultants.

We would let Michigan Bell consultants determine our needs and specify the types of equipment
and/or services that would best satisfy these needs.

Please rank the following phrases from I to 5 in terms of how you feel they reflect your firm’s attitude toward

the value of telephone communications. (I being most important; 5 being least important.)

__.A.

_.B.
_._C.

A necessary service to our customers and

suppliers. ___D. A competitive tool
A cost-saving service.
A primary aid to developing new cus- _£. Other (please specify)

tomcrs

306

STUDY OF COMMUNICATION NEEDS IN BUSINESS

 

R EALTOR IN FORMATION

(Strictly Confidential)
Graduate School of Business Administration
Department of Marketing and Transportation
Michigan State University

The questionnaire is divided into two parts. Part A deals with certain characteristics related to your firm. its
customers and its suppliers that may affect your need for various types of communications offerings.

Part B of the questionnaire is primarily concerned with your evaluation of the present and future role of telephone
Communications in various areas of your firm’s operations. Your responses to this portion of the questionnaire
will aid us in determining where telephone communication needs are greatest and how these needs vary from one
sector of the business community to another.

Please note that the responses given in this questionnaire will be considered strictly confidential and will not in
any way be identified with any individual or firm.

The entire questionnaire should take approximately 20—30 minutes to complete. Your cooperation will be most
Sincerely appreciated.

307

PART A.

Unless stated otherwise. the following questions refer only to the facility or
plant located at the address to which this questionnaire was mailed. Any ques-
tions pertaining to branch plants located elsewhere in the city, state or country
will be clearly identified. Thus unless stated specifically in the question, any
information given should refer only to the facility to which this questionnaire
was addressed.

I.
2.

IO.

How long has your firm been operating in its present facility?__.____ yrs.

How many separate offices do you have at your present location?

 

Approximately how many separate real-estate transactions did you carry
out during the past year?

 

What was your gross revenue from all transactions during the fiscal year
1967? ................................................................

Please estimate below what per cent of your total revenue during the past
year was derived from each of the following types of activities. (If you
are not involved at all in a particular activity write a 0 in the blank.)

% residential real estate

 

 

% residential property insurance

% residential construction

 

 

% commercial property insurance

% commercial real estate

 

 

% personal insurance

‘70 commercial construction

 

 

% other (please specify)

 

% farm properties

What per cent of your total revenue during the past year was derived from
property dealings or other activities taking place in each of the following
geographic areas.

Greater Lansing Area
(Local Telephone Calling Area) ...................................................................

Within Michigan but Outside
of the Greater Lansing Area .........................................................................

Outside of Michigan .....................................................................................

What has been the average number of employees working at this location
during the past year?

 

How many salesmen or brokers on the average have you had working
for you at any one time during the past year?

 

How many secretaries on the average have you had working for your firm
at any one time during the past year?

 

Please indicate the number of additional offices (if any) your firm main.
tains in each of the geographical areas listed below.

Greater Lansing Area
(Local Telephone Calling Area) ..... - - .......................

Within Michigan but Outside
of the Greater Lansing Area - ....................

Outside of Michigan ......................

 

 

 

 

 

 

 

 

 

 

 

308

In what city and state are your main offices located?

 

city ................. state ........................

What are the central telephone numbers to which your firm is billed? (Please include the number of any
private lines or WATS lines to which separate billings are made.)

oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo

Under what Standard Industrial Classification (SIC) code (5) is your company listed?
Major SIC Code ..................... Minor SIC Code (s) ...................................... .. ........................................ E‘

Position of person (s) completing questionnaire.

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

 

 

309

PART B

The following questions concern your own personal attitudes and opinions
toward the value of telephone communications - both in general and in
respect to the value of Specific communication offerings to certain aspects
of your firm’s operations.

(Table for Questions I — 5)

A. Communicating with customers E. Communicating with branch facilities

and/or the home office

B. Communicating with suppliers

F. Utilizing telephone lines for data

C. Interdepartmental communications transmiss’on (other than voice)

D. Communicating with salesmen in the C. Other (please specify)

I.

9\

field

Considering telephone usage in the local calling area only (i.e. no toll
charge), please rank the following activities from I to 7 in terms of
the amount of time your firm spends in the average month on each
activity in relation to the others. (I being the most time; 7 being the
least; 0 if no time at all) ......................................................................

Considering long distance telephone usage only. please rank the fol-
lowing activities from I to 7 in terms of the amount of time your
firm spends in the average month on each activity in relation to the
others. (1 being the most time; 7 being the least; 0 if no time at all)

Please rank the following activities from I to 7 in terms of where you
feel telephone communications will make its most important contri-
bution to your firm 5 years from today. (1 being most important;
7 being least important; 0 if no contribution ......................................

If in-state long distance telephone rates were to be increased by 20%,
by what per cent would you expect your usage in each of the fol-
lowing areas to decrease? (Write 0 if you would expect no decrease.)

% decrease

If inter-state long distance telephone rates were to be increased by
20%, by what per cent would you expect your usage in each of the
following areas to decrease? (Write 0 if you would expect no decrease)

% decrease

customers
3. Communicating with

A. Communicating with

suppliers

 

facilities and/or the home office

F. Utilizing telephone lines for
data transmission (other than

4.22s)
(3. Other (please My)

communication

in the field
E. Communicating with branch

C. Interdepartmental
D. Communicating with saleauten

 

 

 

 

 

 

 

 

 

 

 

 

Compared to other firms of a similar type and size. where would rank your firm in terms of:

A. local telephone usage

far above average ......... above average ......... average ......... below average .........

B. long distance telephone usage

far above average ......... above average ......... average ......... below average .........

C. the amount of telephone equipment you presently utilize

far above average ......... above average ......... average ......... below average .........

D. the amount of yellow page advertising you presently subscribe to

far above average ......... above average ......... average ......... below average .........

E. utilizing telephone wires for data transmission

far above average ......... above average ......... average ......... below average .........

F. utilizing the telephone to contact suppliers

far above average ......... above average ......... average ......... below average .........

G. utilizing the telephone to contact customers

far above average ......... above average ........ average ......... below average

H. your salesmens’ use of the telephone

far above average ......... above average. ........ average ......... IBIOW avemae...... .

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

far below average ........

far below run-race

 

310

. Which of the following statements most accurately describes your understanding of the different types of
equipment, services, and rate structures Michigan Bell offers to business firms? (Check only one.)

 

l.
__2.
____3.

We have an excellent understanding of 4. We have a poor understanding of these

 

the various alternatives available. alternatives.

We have a good understanding of these

alternatives. __.5. We have no knowledge of these alter-
We have a fair understanding of these natives.

alternatives.

Which of the following statements best describes the process you would go through prior to changing all or

part of your communication system? (Assume this to be a major change such as the adoption of WATS or a
private line.) Check only one.

 

2

 

. __4.

We would determine our needs and specify the types of equipment and/or services that would best
satisfy these needs.

. We would determine our needs and let Michigan Bell consultants Specify the types of equipment

.____3.

and/or services that would best satisfy these needs.
We would determine our needs, suggest the types of equipment and/or services that might best
satisfy these needs. but leave the final decision up to Michian Bell consultants.

We would let Michigan Bell consultants determine our needs and specify the types of equipment
and/or services that would best satisfy these needs.

Please rank the following phrases from I to 5 in terms of how you feel they reflect your firm's attitude toward
the value of telephone communications. (1 being most important; 5 being least important.)

__.__A.

_____B.

...—_C.

A necessary service to our customers and

suppliers. __D. A competitive tool

A costcsaving service.

A primary aid to developing new cus- __E. Other (please specify)

tomcrs

 

311

STUDY OF COMMUNICATION NEEDS IN BUSINESS

 

DRY CLEANER INFORMATION
(Strictly Confidential)
Graduate School of Business Administration
Department of Marketing and Transportation
Michigan State University

The questionnaire is divided into two parts. Part A deals with certain characteristics related to your firm, its
customers and its suppliers that may affect your need for various types of communications offerings.

Part B of the questionnaire is primarily concerned with your evaluation of the present and future role of telephone
communications in various areas of your firm’s operations. Your responses to this portion of the questionnaire
Will aid us in determining where telephone communication needs are greatest and how these needs vary from one
sector of the business community to another.

Please note that the responses given in this questionnaire will be considered strictly confidential and will not in
any way be identified with any individual or firm.

The entire questionnaire should take approximately 20—30 minutes to complete. Your cooperation will be most
Sincerely appreciated.

312

PART A

Unless stated otherwise, the following questions refer only to the facility or
plant located at the'address to which this questionnaire was mailed. Any ques-
tions pertaining to branch plants located elsewhere in the city, state or country
will be clearly identified. Thus unless stated specifically in the question, any
information given should refer only to the facility to which this questionnaire
was addressed.

I.

A,
‘-

IO.

12.

How long has your firm been operating in its present facility?______ yrs.

How many separate or adjoining buildings do you have at this location?
(excluding garages, sheds, etc.)

 

Approximately, how many square feet of floor Space does your physical
plant encompass? .......................................................................................

How many separate offices do you have in the building(s)? ........................

What was the total revenue for the store located at the address to which
this questionnaire was sent during the fiscal year I967?

 

What was the total revenue for all of your Greater Lansing Area dry clean-
ing establishments during the fiscal year 1967?

 

What has been the average number of employees working at this location
during the past year?

 

How many of this total were salaried employees? .................................... ..

Of your total number of employees, how many on the average during the
past year have been employed in each of the following areas? ....................
Past Yar

Administrative and Supervisory

 

Clerical
Sales Clerks
Other (please specify)

 

 

 

What percent of your yearly gross revenue is presently derived from the
following types of customers? ....................................................................

% Residential
% Commercial or Industrial

% Other (please specify)

Which of the following services does your firm presently offer? (please
check)

___Iaundry __~__.drapery cleaning
dry cleaning ______winter storage
_____rug cleaning ______other (please specify)

Please indicate in the appropriate column the number of additional dry
cleaning establishments your firm maintains in each of the geographical
areas listed below:

Greater Lansing Area
(Local Telephone Calling Area) .................................................

 

 

 

 

 

 

 

313

Outside Greater Lansing
but within Michigan .....................................................................................

 

Outside of Michigan ....... I ..............................................................................

 

13. In what city and state are your main offices located?

city ......................... - - - . state ................................................................

 

14. What are the central telephone numbers to which your firm is billed? (Please include the number of any
private lines or WATS lines to which separate billings are made.)

......O.......OII.‘I...’............II.I ....I‘ .............................................................................................

15. Position of person (s) completing questionnaire.

 

...-00.00................OOIIOOOOI...l.............. ......COOOCOOOOO.......OOOOOOOOOUOCOOO00......000... ......O... T ‘ v ..................

 

314

PART B

The following questions concern your own personal attitudes and opinions
toward the value of telephone communications —- both in general and in
respect to the value of specific communication offerings to certain aspects
of your firm’s operations.

(Table for Questions 1 — 5)

A. Communicating with customers E. Communicating with branch facilities

and/or the home office

8. Communicating with suppliers

F. Utilizing telephone lines for data

C. Interdepartmental communiations transm'us‘on (other than voice)

D. Communicating with salesmen in the G. Other (please specify)

1.

field

Considering telephone usage in the local calling area only (i.e. no toll
charge), please rank the following activities from I to 7 in terms of
the amount of time your firm spends in the average month on each
activity in relation to the others. (I being the most time; 7 being the
least; 0 if no time at all) ......................................................................

Considering long distance telephone usage only. please rank the fol-
lowing activities from I to 7 in terms of the amount of time your
firm spends in the average month on each activity in relation to the
others. ( I being the most time; 7 being the least; 0 if no time at all)

Please rank the following activities from I to 7 in terms of where you
feel telephone communications will make its most important contri-
bution to your firm 5 years from today. (1 being most important;
7 being least important; 0 if no contribution ......................................

lf in-state long distance telephone rates were to be increased by 20%,
by what per cent would you expect your usage in each of the fol-
lowing areas to decrease? (Write 0 if you would expect no decrease.)

If inter-state long distance telephone rates were to be increased by
20%, by what per cent would you expect your usage in each of the
following areas to decrease? ( Write 0 if you would expect no decrease)

%decreue

customers
B. Communicating with

A. Communicating with

suppliers

gwlth salesmen

 

communication

facilities and/or the home office
F. Utilizing telephone lhes for

data transmission (otha tlna

D. Communicatin

inthe field
E. Communicating with branch

C. Interdepartmental

G.0ther(plaaaaspecify)

 

 

 

 

 

 

 

 

 

 

 

 

Compared to other firms of a similar type and size. where would rank your firm in terms of:

A. local telephone usage

far above average ......... above average ......... average ......... below average .........

B. long distance telephone usage

far above average ......... above average ......... average ......... below average .........

C. the amount of telephone equipment you presently utilize

far above average ......... above average ......... average ......... below average .........

D. the amount of yellow page advertising you presently subscribe to

far above average ......... above average ......... average ......... below average .........

E. utilizing telephone wires for data transmission

far above average ......... above average ......... average ......... below average .........

F. utilizing the telephone to contact suppliers

far above average ......... above average ......... average ......... below average .........

G. utilizing the telephone to contact customers

far above average ......... above average ......... average ......... below average.........

H. your salesmens’ use of the telephone

far above average ......... above average. ........ average ......... below average. ........

far below average... .....

far below average ........

far below average ........

far below average ........

far below average ........

far below average. .......

far below average ........

far below average .....

 

 

.315

Which of the following statements most accurately describes your understanding of the different types of
equipment. services. and rate structures Michigan Bell offers to business firms? (Check only one.)

__._I. We have an excellent understanding of __4. We have a poor understanding of these
the various alternatives available. alternatives.
____2. We have a good understanding of these
alternatives. ___5. We have no knowledge of these alter-
__3. We have a fair understanding of these natives.
alternatives.

Which of the following statements best describes the process you would go through prior to changing all or
part of your communication system? (Assume this to be a major change such as the adoption of WATS or a
private line.) Check only one.

 

1. We would determine our needs and specify the types of equipment and/or services that would best
satisfy these needs.

_... We would determine our needs and let Michigan Bell consultants Specify the types of equipment

and/or services that would best satisfy these needs.

__3. We would determine our needs, suggest the types of equipment and/or services that might best
satisfy these needs. but leave the final decision up to Michian Bell consultants.

4. We would let Michigan Bell consultants determine our needs and specify the types of equipment
and/or services that would best satisfy these needs.

 

Please rank the following phrases from I to S in terms of how you feel they reflect your firm's attitude toward
the value of telephone communications. (1 being most important; 5 being least important.)

__A. A necessary service to our customers and

suppliers. __D. A competitive tool
___B. A cost-saving service.
__C. A primary aid to developing new cus- __E. Other (please specify)

tomcrs

"a! '1‘ .ID I" ra-

 

 

   

31293 03 69 3553

lilllwillIlluminatiwillful