Forecasting advertising communication effects using media exposure distribution models

MISSING IMAGE

Material Information

Title:
Forecasting advertising communication effects using media exposure distribution models test market results in South Korea
Physical Description:
xiv, 211 leaves : ill. ; 29 cm.
Language:
English
Creator:
Park, Hyunsoo, 1966-
Publication Date:

Subjects

Subjects / Keywords:
Mass Communication thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Mass Communication -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1998.
Bibliography:
Includes bibliographical references (leaves 203-210).
Statement of Responsibility:
by Hyunsoo Park.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029246829
oclc - 39556936
System ID:
AA00017634:00001


This item is only available as the following downloads:


Full Text










FORECASTING ADVERTISING COMMUNICATION EFFECTS
USING MEDIA EXPOSURE DISTRIBUTION MODELS:
TEST MARKET RESULTS IN SOUTH KOREA














By

HYUNSOO PARK


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1998






























Copyright 1998

by

HYUNSOO PARK













ACKNOWLEDGEMENTS


First of all, heartfelt thanks goes to my adviser, Dr. Kent M. Lancaster. I really
appreciate his kind guidance in developing the research design and the normative
framework, and in writing this dissertation. Without his scholarly advice, guidance, and
great help during my whole Ph.D. study, this dissertation would not have been possible. I
also would like to thank my committee professors, Dr. Joseph Pisani, Dr. Leonard Tipton,
Dr. Marilyn Roberts, and Dr. Richard Scheaffer for excellent suggestions, encouragement,
and help.
Acknowledgments are due to several professionals and professors in Korea.
Special gratitude goes to the CEO of SangAm agency, Dr. Tong-Yeol Sung, and Ms.
Nan-Hee Choi, Vice-chief of the marketing department, for their considerable efforts in
providing valuable data. Mr. Hyun-Tae Kim, Dr. Jung-Sik Jo, and Dr. Yang-Ho Choi also
provided valuable help.
Appreciation is also extended to June-Hyung Ji and my brother-in-law, Yeon-Soo
Kang, for sending me precious Korean papers and data for this study. Special thanks go
to my brothers and friends, Yoon-Soo Park, Jeong-Soo Park, Kwang-Ho Lee, Dong-
Hyun Han, Yoon-Sung Hwang, Il-Joon Yoo, Seung-Kwan Ryu, and Ho-Cheol Kim
whose encouragement and friendship made this dissertation possible.
I am really grateful to my lovely wife, Yeon-Mi, whose self-sacrificing support and
help during my Ph.D. study have given me the inspiration, confidence, and balance
necessary to complete this endeavor.
Finally, I would like to dedicate this dissertation and my Ph.D. degree to my
parents, Jong-gug Park and Bum-Rok Lee. "Thank you for life, love, and help every
moment of my life. I love you."










i4 oi 1 4 9I RE 7-44 l-*g4 Zi-
-S-a-a- "r l a-$ rB O P 4. 01 -*)0 11013) = -7
924- 471xir @AMr 8 E 44. Il 2r< 'gol golt 1
0oi- dr + MA-l, 01 7s0l 71 Ot -E 1A % f-4-


*& s-g- i-444 44.1 444g -9- l -4q-0 11% 4TSi4
Dr. Kent M. Lancaster d9 13-- A4* } -E5'-14. 0A} 4Y91
sEf 463- j4 94 rQg4 7fi4X 0 f3R j41s-4q4. O}1-Bi V
-- W 44 2 -2 xl~4 Z Mel of.l Q- t4 Dr. Joseph
Pisani, Dr. Leonard Tipton, Dr. Marilyn Roberts, Dr. Richard
Scheaffer -1+-- 7*A4 -g qq4.


"A ^S-4. 44 n# SH4 1444 F-1 S.^*0-S. ^^ F
(2)7714Q rARl014 I 44I, 4A +A4i4 61 4 t)a 7


-44-?, A \OVl SP A elm +14 4utBl a 4 ^)), 3W


i3)401, -2 S01 University of Florida .4 EIdW-4 *4A4SW4 174
718-4 WP4 -a4.
5s1-al *-14-k4% -YO-I SIx as l4AA4 A471* 1l A tr :
94 AA 7j a oI 'i, + -efl, 0) -, J g-, V ,, g
st -^s4 41941 +4g rgo sgo] al-4.
11R94 4971 M49?! -Aa 0 -1 t i-44^1 etA
4t^}% Of4 4aD14'11, p-g44 491r 40 _1 *ia-, a-^4R 4A-=
*-4A -V 44. 'LLI, 71' ^a4 "'t J 4A 1 4 sil qL 0l 9 sA-0
4M9413"-4 94% el 4^l r t ^q o 4^P 414 42ltl 3--t
^-tJ;-017 44-1 !L 44.











TABLE OF CONTENTS


page


ACKNOWLEDGEMENT N III

LIST O F T A B LE S ........................ ..... ...... ............................ ........ ....... ix

LIST OF FIG URES......... ..... ...... .. .. ............................. .. ........ .. .. xi

ABSTRACT................................ ............ .. ............. .. ................. xii

CHAPTER I IN TRODU CTION ........................... ..... .............. .... ......... .. 1

A advertising C am paign E effects ....................................... ................................... ......... 1
Com plexity of the Problem ........................................ ....... .................. ...... ...............2
M edia Exposure Distribution M odels............................. ............3
Objectives of this Study. .................. .... ...... .. .. ... ...................... 5
R atio n ale .................................................................. .... ......... 6
O organization of the D issertation................... ..................... .............. ........................ 8

CHAPTER II. MEASURING AND FORECASTING ADVERTISING EFFECTS ......9

Primary Reason for Evaluating and Forecasting Advertising Campaigns......................... 10
Problems in Advertising Campaign Measurement................................................ 11
Advertising Effects on Sales............... .... ............... .................. .......... ................. 13
Difficulties in Measuring Advertising Effects on Sales.................................... 14
Advertising Communication Effects............ ................ ..................... ............... 16
Awareness .................................................... .... ......... ....... 17
Previous Research on Aw areness ...................................... .......... 19
Recall ..................................................................... .. ........ 20
Liking and Attitude Toward Advertisement........................... ................. 21
Preference ........................... ...... ........... .. .... ..... ............... .. 23
Purchase Intention and Actual Purchase ...................... ............................24
Achieving Desired Communication Effects........... ........ ....... ... ........... 25
Effective Exposure to A dvertising................ .............................................. ............... 28
Advertising Exposure......................... ... .... ....... .. 28
R each .......................... .. ..... ............................. ....... ......... 2 9
Frequency and Frequency Distribution.............................................. ............ 30
Effective Frequency and Reach.......................................................................... 32
Problems with Using the Concept of Effective Frequency............................... 34
How to Set Effective Frequency Levels........................................... ...... .. ....37








CHAPTER III. BACKGROUND ON THE SOUTH KOREAN ADVERTISING
INDUSTRY...................... .. .. ........ ................... 42

Introduction......................... ............ ............................ .................... 42
South Korean Economic Development ....................................... ..... .......... 42
The Growth of the Advertising Industry in South Korea.................................... 43
A advertising M edia in South K orea .................................................... ........... 45
Recent Changes in the South Korean Advertising Industry ................................. 47
Studies on Measuring and Forecasting Advertising Effects in South Korea......... 48
Media Planning and Evaluation Procedures of South Korean Advertising Practitioners ..51
Purpose of the Study ......................... ......................... .................................. 52
M eth o d s .............. ..... .......... ................................. ............ 5 2
R results ................................... ... ................... ................... ........ 53
Definition of Reach and Effective Reach................................................ 56
Com m unication Effects U sed...................... ..................... ................... 57
Message Weighting and Reasons of Not Weighting ............................... 59
N eeded Im provem ents ...................................... ....... ....... ........ 60
Discussion and Implications for Further Research............................ ........... 61

CHAPTER IV. M ETHODOLOGY ............................... ... ................... ....... 64

C ase S tu d y ............ ... .. .. ..... ...... .. .. .. ... .. .. .. .. .. ... .. .. .... .. .. .. .. 6 6
A SIA N A A irline.................... .... ... ..... ...... .. ... .... 66
Target Audience and Advertising Campaign ................ ............................ 69
C o m p e tito r ......................................................................................................... 7 0
M edia D ata ................ ........... ...... ........... ........... ..... ..... .................. 72
T V R atin g s ......................................... ................................... 7 2
M magazine R atings ....................................... ..... 73
T racking D ata ........................................ ...... ... ..... .. 74
S am pling ............................................ .. .. 74
M easurem ent ......................... ... ... ... .. ............................. 75
Media Evaluation Model Used in This Study ............................................... 76
Beta Binomial Exposure Distribution Model.................................... 76
M odel Selection................................. .. ..... .. ....... .. .. 79
M odel M modification ..... .................................................. .............. ... 82
Error Estimation for the Media Evaluation Model ............................. 83
Suggestions for Improving Campaign Efficiency and Budget Levels ................. 87
Sophisticated versus N alive Comparison ............................................................. 88
Summary of the Normative Framework................... ..................................... 89
R research H ypotheses ................................................................................... .. ............. 90
Testing Procedures ...................... ......................................... 93
Statistical Procedures ............ ...... ........... ............. .......... 97
Calibrating Forecasting Procedures .......................................................... .. .. 99








CH A PTER V RESULTS............................................... ................................ 104

Intro du ctio n ............................ .. .. ... .. .. .. .. ... .. .. .. .. .. .. .. .. .. ..... 104
T est M market R esults................. .. ............... ..... ....................... ........... 106
M edia Evaluation M odel Prediction..................................... ...................... 106
T racking Study R results ............................... ...................... ....................... 109
Testing Hypotheses............................ ....................................... ..... .. 111
Awareness of the Campaign Message ......................................... .... 111
Preference ..................... ......................... .. .. .... .. .. ... .. 111
W willingness to U se ................................ ......................... .............. 112
Calibrating Procedures ................................................ 113
Critical Analysis of the ASIANA Advertising Campaign.................... ...... ....... 120
Suggestions for Future ASIANA Campaigns.................. ...............125
Selecting Appropriate Vehicles................. .............................. ....... .. 125
Appropriate Budget for Monthly Schedule................................. 127
Sophisticated versus Naive Comparison ................. ......... .......... 132
Pooled Approach............................... ............. 132
Vehicle versus Message ................ .......... ....... .............. 133

CHAPTER VI. SUMMARY, CONCLUSIONS, AND IMPLICATIONS ................. 135

Sum m ary and C conclusions ............................................................. ....... 135
Im plications ................................... ........................ ............ .. 139
Limitations and Suggestions for Future Study .......................................................... 140

APPENDICES

A. SURVEY QUESTIONNAIRE IN BOTH ENGLISH AND KOREAN...... 143
EN G LISH V ER SIO N ................................. ..................... ........ 144
KOREAN VERSION .......................................... .......... 151

B. TWO TYPES OF MAGAZINE ADVERTISEMENTS ........................... 158

C. TV PROGRAM RATINGS, ADVERTISING MESSAGE RATINGS,
MAGAZINE RATINGS, AND COSTS FOR THE ASIANA
ADVERTISING CAMPAIGN IN BOTH ENGLISH AND
KOREAN... .............................. ..... ........ 161
ENGLISH VERSION ................................................. ................ 162
KOREAN VERSION ....... ...................................... ................. 166

D. GENERAL FORM OF THE BBMD MODEL................... ............. .. 170

E. ADPLUS RESULTS FOR NETWORK TELEVISION IN FEBRUARY.... 176

F. ADPLUS RESULTS FOR NETWORK TELEVISION IN MARCH.......... 179








G. ADPLUS RESULTS FOR NETWORK TELEVISION IN APRIL............. 182

H. ADPLUS RESULTS FOR MAGAZINE ADVERTISEMENTS................. 185

I. ADPLUS RESULTS FOR NETWORK TELEVISION AND
MAGAZINES COMBINED IN FEBRUARY ............................. 187

J. ADPLUS RESULTS FOR NETWORK TELEVISION AND
MAGAZINES COMBINED IN MARCH................. .................... 191

K. ADPLUS RESULTS FOR NETWORK TELEVISION AND
MAGAZINES COMBINED IN APRIL............................ ........... 195

L. ADPLUS RESULTS WITH FLOWCHART......................................... 198

R EFE R E N C E S ............................. ...... ... .... .. .... .. .. .. .... .. ... ... 203

BIO G RA PH ICA L SK ETCH .............................................................. ........ ....... 211














LIST OF TABLES


Table page

1. Common measures for advertising communication effects.................................. 17
2. Relationship between common communication measures and needed
exposure frequency in a normal marketing, copy, and media situation ............. 27
3. Example of frequency distribution................... ....... ....... ....... ..... ......... 32
4 Marketing, copy and media factors that affect effective frequency........................ 38
5 Suggested additional points and modified effective frequency levels
for brand X ....... ... ... ....... .............. ...................................... .. 41
6. The progress of the South Korean advertising industry................... .......... .. 45
7. South Korean advertising media, total expenses, and costs...................................... 46
8. Sung's independent and dependent variables and regression equations................... 50
9. Systems used to develop advertising media plans ..........................................54
10. Factors used in evaluating media plans.............................. ....................... 55
11. Representative definitions of effective reach in comparison with two
previous Am erican studies .................................... .......... ................. 58
12. Variables used to evaluate advertising communication effects............................ 59
13. Normative framework necessary to predict advertising communication effects ........ 65
14 D om estic and international m market shares ...............................................................68
15. Advertising insertions by media category and expenses of
"Her name is ASIANA" campaign from February 17 to May 16, 1996............. 70
16. Competitor GRPs and advertising expenses.... ........................... .......... 71
17. Sample characteristics ...................................... .......................... .. 75
18. R2, and R, estimation equations for network television............ ......................82

19. Summary of the normative framework applied to the ASIANA case study............... 89








20. Suggested points and modified effective frequency levels
for ASIANA Airline campaign effects ................ ................... ......... 91
21. Adjustment options, expected effects, and decisions in this study......................... 102
22. Consumer tracking study results and model predictions.................................... 110
23. Adjustment options and changes in predictions. .......................... ................. 116
24. Consumer tracking study results and model predictions with calibration ............... 118
25. CPM-MSG figures for the most efficient 11 television programs (vehicles) ........... 122
26. CPM-MSG figures for all magazine advertisements.............. ......................... 124
27. Relatively high CPM-MSG vehicle list in the ASIANA campaign.......................... 126
28. Relatively low CPM-MSG vehicles added for schedule modification ...................... 126
29. Suggested monthly budget levels and estimated message reach 3+........................ 131

















LIST OF FIGURES


Figure page

1. The effects of top-of-mind awareness of the advertisement............... ............ 20
2. The shape of a typical m message reach curve.................................................... ...... 30
3. Sung's model to estimate advertising effects ............... ................ ............... 50
4 A D p lu s re su lts ................................................... ................................... .......... 8 1
5. M edia evaluation m odel prediction............................................. ..................... 108
6. M edia evaluation model prediction with calibration............................................... 117
7. Estimated frequency distributions for the three-month ASIANA campaign............ 119
8. Budget allocation among television programs............................. .................... 120
9. Budget allocation among m agazines....................... ......................................... 121
10. Reach percentage changes with budget increases for February vehicles,
television and magazines combined .......................................... .............. 128
11. Reach percentage changes with budget increases for March vehicles,
television and magazines combined ................................. ................. 128
12. Reach percentage changes with budget increases for April vehicles,

television and m magazines com bined ..................................... ...................... 129
13. Estimated reach (3+) for each monthly advertising schedule .................................. 130
14. Estimated message reach (3+), naive versus sophisticated approaches.................. 133
15. Estimated frequency distributions with vehicle and message data............................ 134
16. Ideal procedures for predicting and evaluating advertising effects using media
exposure distribution models ............ ... ........ ... .. ....... 136














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy

FORECASTING ADVERTISING COMMUNICATION EFFECTS
USING MEDIA EXPOSURE DISTRIBUTION MODELS:
TEST MARKET RESULTS IN SOUTH KOREA

By

Hyunsoo Park

May, 1998

Chairman: Dr. Kent M. Lancaster
Major Department: Mass Communication

The primary objective of this study is to demonstrate how to predict advertising

communication effects using a variation of the beta binomial exposure distribution (BBD)

model. A computer based model which embodies important aspects of normative theories

of media planning is used to forecast the communication effects of a South Korean

advertising campaign.

Due to the lack of systematic study of advertising effects in South Korea, the

South Korean advertising industry needs objective and scientific methods to develop a

model that is applicable to the South Korean advertising environment. Therefore, this

study applied a media exposure distribution model that is used in the United States and

worldwide to the South Korean market to probe how accurately such models and

procedures can predict the communication effects of a South Korean advertising








campaign. A consumer tracking study is used to determine the accuracy of forecasting

advertising campaign effects.

Prior to the application of the media evaluation model and procedures, survey

research was conducted among South Korean media practitioners. The goal was to show

how the media practitioners view the current state of reach/frequency estimation methods

and which communication effects are used to predict or measure potential advertising

impact. Based on the survey results, previous studies as well as marketing, copy, and

media factors, an effective exposure frequency level is estimated to predict target audience

awareness of a South Korean advertising campaign message.

In addition to advertising message awareness, this study also examines target

audience preference and willingness to purchase for the brand used as a case study.

Minimum frequencies of consumer exposure to the advertising campaign necessary to

achieve preference and willingness to purchase also are estimated with the same normative

framework. Then, the minimum effective frequency levels are analyzed in comparison

with actual consumer tracking data to determine whether more accurate predictions would

have been possible.

The study results showed that the normative framework and media exposure

distribution model used in this study can be accurately used in predicting target audience

awareness of the campaign message. The prediction for target audience awareness of the

campaign message was close to the tracking study results. With additional calibrating

options, the predicted awareness of the campaign message could be exactly matched with

the tracking study results.








This study also suggests that there may be a hierarchy in consumer processing of

advertising messages. The needed target audience exposure to achieve preference was

significantly higher than the needed exposure for awareness. The needed message

exposure for willingness to use was also significantly higher than the target audience

exposure for awareness and preference. Though many more factors should be considered

for accurately predicting or evaluating advertising effects in terms of preference and

willingness to use, this study still provided useful results for the estimation of advertising

effects on preference and willingness to use.

The advertising campaign used as a case study also was critically analyzed to

suggest appropriate monthly budgets and vehicle selections for future campaigns. The

sophisticated approach used in this study was compared with typical naive approaches to

show how the potential gaps can be misleading.

In this study, the normative frameworks, media exposure distribution models,

evaluation methods, including many influential factors, and calibrating procedures provide

useful guidance for forecasting advertising communication effects.














CHAPTER I
INTRODUCTION




Advertising Campaign Effects


Advertising, an important promotional tools of modern marketing management,

has been used worldwide to present information to potential buyers and reinforce

consumer preference for a company's products or services (Lilien, Kotler, & Moorthy,

1992). Every year billions of dollars are invested by companies around the world to test,

produce, and implement advertising campaigns. According to Aaker, Batra, and Myers

(1992), estimated advertising expenditures will be $780 billion worldwide and $320 billion

in the United States alone by the year 2000. Because of these enormous expenditures,

advertisers want the advertising planners and managers to identify exactly what are the

results obtained from the advertising investment and to provide evidence of the return on

investment. Therefore, industry practitioners and social scientists throughout the world

continue to predict and evaluate the effects of advertising campaigns not only prior to

campaign execution but also after advertising is aired or printed.








Complexity of the Problem


The effects of advertising campaigns are evaluated according to campaign

objectives. No matter what objectives an advertising campaign has, such as an increase in

sales, market share, brand awareness, advertising message awareness, recall, preference,

or something else, there seem to be limited means of accurately measuring advertising

campaign effects. The main reason is that advertising effects typically interact with other

elements in the marketing mix, such as an advertiser's image, price, competitor

promotional activity, distribution, product quality, and other promotion activities.

Although technology is making the prediction and evaluation of advertising effects easier,

skepticism about the possibility of accurately predicting or evaluating advertising effects

has continued over the years (Lancaster & Helander, 1987).

Another problem in predicting or evaluating advertising effects is what is to be

measured? Two frequently used approaches are advertising communication effects and

sales. Since one of the main purposes of advertising is to communicate persuasive

messages, measuring communication effects, such as brand and advertising message

awareness, recall, attitude toward brand, preference, and willingness to purchase, is a

natural way to evaluate advertising campaign effects (Schultz, Martin, & Brown, 1984;

McDonald, 1995; Woodside, 1996).

Measuring actual behavior of consumers, such as sales or purchases, is another

frequently used approach to measure advertising campaign effects. Since the ultimate goal

of advertising may be the increase of sales, the amount of sales has been used to measure

advertising effects (Leone, 1983; Assumus, Farley, & Lehmann, 1984; Holstius, 1990;








Schreiber, & Appel, 1991; Lodish et al., 1995). However, since sales are the result of

various marketing efforts, it is very difficult to measure the absolute effects of an

advertising campaign in terms of sales.

Advertising communication effects can be used more efficiently as a method of

evaluating the effects of advertising campaigns. Brand awareness, advertising message

awareness, recall, preference, and willingness to purchase are the most frequently used

measures in terms of communication effects. For example, advertising campaign message

awareness can be measured with much less interaction with other marketing factors than

the measure of sales. However, the communication measures should be explained in

relation to the final objectives, such as increase in sales or market share. Many previous

studies in the United States suggest that there is a positive relationship between brand or

advertising message awareness and preference, sales, and market share (Burke &

Schoeffleer, 1980; Wilson, 1981; Zufryden, 1996; Woodside, 1996).

What is best among various communication effects is up to the planners to decide

in relation to campaign objectives. However, no matter what kind of approach is used to

measure advertising campaign effects, it is still difficult to identify actual advertising

effects while completely controlling for all other important controllable or uncontrollable

dimensions of the marketing mix, such as pricing behavior, distribution intensity, product

quality, competitor target market selection, and other promotion activities.



Media Exposure Distribution Models


Because of the enormous advertising expenditures and the complexity of the

problem, studies of advertising effects have been active in the advanced countries like the








United States in an effort to overcome the problems explained above (Leckenby & Kishi,

1982a; Lancaster & Helander, 1987; Rust & Klompmaker, 1981; Rice, 1985; DeKimpe &

Hanssens, 1995a; Hansen, 1995). Models have been developed to predict and measure

advertising effects.

One of the most frequently used models in the United States is the beta binomial

exposure distribution (BBD) model (Leckenby & Kishi, 1982), which was developed by

Metheringham in 1964. The general goal of the BBD model is to estimate the probability

and distribution of consumer exposures to advertising media schedules. The model

incorporates both beta and binomial distributions of schedule exposures. The beta

distribution is used to estimate the probability of consumer exposures to any number of

advertising vehicles or messages included in a schedule, such as one, two, three, or up to

any number of exposures to advertising vehicles or messages. These probabilities are then

used with the binomial distribution which estimates the proportion of consumer exposures

to advertising vehicles or messages, for example, once, twice, or three times (Lancaster &

Katz, 1989).

Computer-based models also have been developed based on the beta binomial

exposure distribution (BBD). One such microcomputer program is ADplus, which is used

to evaluate advertising schedules based on a variation of the beta binomial exposure

distribution model. It uses procedures that vary by media category to account for

differences in available audience and cost data (Lancaster, 1993).

Three test market studies in the United States (Lancaster & Helander, 1987;

Lancaster & Katz, 1989) demonstrate the ability of the predecessor of ADplus to predict








the effects of advertising campaigns, including communication effects such as advertising

exposure and brand awareness.



Objectives of this Study


While the study of advertising effects has been active in advanced countries, it has

not been widespread in some other countries such as South Korea. Even though the

South Korean advertising industry has in recent years grown into the world's 13th largest

market (US $6.1 billion in 1995), its research remains at an immature level. Only a few

studies have been conducted about measuring advertising effectiveness in South Korea

(Sung, 1989; Korean Gallup, 1992).

Because of the lack of systematic study of advertising effects in South Korea, the

development of an advertising effects model applicable to the advertising environment of

South Korea is relatively poor. Therefore, a model of measurable advertising effects that

meets the advertising environment of South Korea is needed. Such a model should be

developed by means of an objective and scientific method for forecasting and evaluating

advertising effects. Much more research should be conducted in South Korea concerning

the evaluation of desired communication effects.

Therefore, one of the primary objectives of this study is to demonstrate how to

predict the effects of an advertising campaign using a variation of the beta binomial media

exposure distribution model in the South Korean market. The computer-based beta

binomial exposure distribution model ADplus will be used to forecast the effects of an

advertising campaign. These predicted advertising effects will be compared with actual

consumer survey results in the South Korean market. Consequently, the computer-based








BBD model developed in the United States will be used to examine how accurately the

model can predict the effects of a South Korean advertising campaign.

Little information is available about how South Korean media practitioners view

current media evaluation terms and which communication effects are frequently used to

predict or measure potential advertising impact. Therefore, a survey was conducted prior

to the application of the media evaluation model and procedures. Based on the survey

results and previous American and Korean studies, optimal advertising exposure

frequencies will be estimated to predict desired communication effects, such as consumer

awareness of an advertising campaign message, preference, and willingness to purchase.

Then, a consumer tracking study will be used to determine how accurately the model

predicts the potential advertising impact.

This study also will examine methods used to measure advertising effects as well

as factors that may influence advertising communication effects, such as advertising

message quality, market share, and competitor promotional activities. For the brand used

in this study, these factors will be used to determine the level of effective frequency, which

is a predetermined level of consumer exposure repetition to an advertising campaign that

is needed to achieve desired communication effects. For example, the brand that has

strong competitor advertising, low market share, low brand loyalty, or low advertising

message quality may need a relatively high effective frequency level.



Rationale


Advertising planners have experienced difficulties in predicting and measuring the

effects of advertising campaigns over the years. One of the major challenges in evaluating








advertising effects is the inability of researchers to estimate the accurate functions between

advertising and consumer responses. Predicting advertising effects is certainly challenging

due to the difficulty in controlling other important dimensions of the marketing mix

(Lancaster & Helander, 1987). Therefore, this study will attempt to overcome some of

these problems. Advertising expenditures will be analyzed by their essential message and

media characteristics, instead of examining the amount in relation to only sales or

communication variables. Actual advertising media schedules and messages gathered from

the South Korean market will be analyzed from the perspective of media planners in terms

of the likely effects on the intended target consumers.

Since the reliability of the beta binomial exposure distribution model has been

proven over decades in the United States (Leckenby & Boyd, 1984; Leckenby & Kishi,

1982; Kreshel, Lancaster, & Toomey, 1985; Rust 1986; Lancaster & Helander, 1987), it

will be worthwhile to apply the BBD model to a totally different market, such as South

Korea, to examine whether the model works as it does in the American market.

The total expenses for advertising in South Korea increased more than 70-fold

during the last two decades. However, in spite of the recent rapid growth of the South

Korean advertising industry, there have been few Korean studies on measuring and

predicting advertising effects. Studies should be conducted to develop a model that is

suitable for the Korean advertising environment. This study, which applies a model used

in the United States and around the world, such as Europe and South Africa, to the South

Korean market, will help South Korean advertising planners develop scientific models.

These can then be used to evaluate and forecast advertising results in terms of

communication effects.








Organization of the Dissertation


This dissertation is organized into six chapters. The introductory chapter provides

the conceptual framework of this study, including the objectives and rationale. The

second chapter presents a literature review related to the purpose of this study. The

review includes many problems of measuring advertising effects as well as previous studies

on predicting and measuring advertising communication effects. Chapter II also presents

the concept of effective exposure to advertising campaigns from the prospective of media

planners.

Chapter III presents South Korean economic development, the growth of the

South Korean advertising industry, the advertising media, recent changes in the advertising

industry, and previous studies measuring and forecasting advertising effects. Survey

results of 61 South Korean advertising agencies are also included in Chapter II. This

survey research demonstrates how South Korean media practitioners perceive media

evaluation terms and the current state of reach and frequency methods.

Chapter IV, Methodology, explains the brand used as a case study, the framework

used to analyze this case study, data collection, research hypotheses, testing procedures,

and calibrating options. Chapter V concerns the results of hypothesis testing and the

application of a commercial model to the South Korean market. Test market results are

provided with comparison to the model predictions of advertising communication effects.

Finally, a summary, conclusions, implications, limitations and suggestions for

future research are presented in Chapter VI.















CHAPTER II
MEASURING AND FORECASTING ADVERTISING EFFECTS






Over the past decades, the research of advertising effects has evolved with great

detail and diversity of theoretical, methodological, and technical approaches. These

research efforts suggest that a possible solution to the ultimate problems of measuring or

predicting advertising campaign effects may be within reach (Dutka, 1995). In this

chapter, the reasons and problems with forecasting and measuring advertising effects, and

publications conducted to help measure or forecast advertising effects are reviewed,

focusing on recent developments. The review of literature will help develop appropriate

methods to conduct this study.

Advertising target audiences should be exposed to the advertising campaign with

optimal frequency to achieve the desired advertising effects. Since there is little agreement

on precisely how many exposures to an advertising campaign message are sufficient to

achieve the advertising campaign goal, studies have been conducted to examine this issue

(Krugmen, 1972; Craig, & Ghosh, 1993; Dekimpe, & Hanssens, 1995;).

In this chapter, in addition to the previous studies on predicting or measuring

advertising effects, the factors that may influence the number of sufficient exposures are








also reviewed. This study will attempt to integrate the possible influential factors to

predict and measure advertising effects from the perspective of media planners.



Primary Reasons for Evaluating and Forecasting Advertising Campaigns


The decision process for advertising typically begins with some form of planning

and either an implicit or explicit strategy or objective. After setting or establishing

measurable objectives and a budget for the various stages and elements in the advertising

campaign, creative strategies and media plans are prepared. Consequently, the planners

need to know whether the advertising campaigns achieved the goals established. Because

of enormous expenditures on advertising campaigns, advertisers want advertising planners

and managers to identify exactly what are the results to be obtained from the advertising

investment and to provide evidence of the return on investment. The advertising planner's

role is to be able to answer the question of whether advertising is the best way to use the

company's resources. Some form of advertising research, such as advertising evaluation,

is required to answer the question.

Most advertising research can be divided into two forms. One is used to predict

what might occur in the real-world marketplace. On the other hand, evaluation research

also can be used as a basis for future actions. In this case, its basic purpose is to measure

what occurred as a result of the advertising campaign and what was returned on the

advertising investment.

According to Schultz and Barnes (1994), there are three primary reasons for

evaluating an advertising campaign. First, advertising campaign evaluation can be used to

determine whether or not the overall advertising campaign objectives are accomplished.








Individual objectives for the various elements of the campaign also can be evaluated and

then, based on these evaluations, the planners can determine why failure may have

occurred.

Second, the return on the campaign investment can be quantified and justified by

knowing what is accomplished. Management also can use the information to explain the

relationship with other potential uses for funds and determine the cost effectiveness of the

campaign.

Third, the measurement information also can be used to change, add, or correct

the campaign plans for future campaigns. Since advertising campaigns can always be

improved, evaluations of previous campaigns and predictions of current campaigns are a

great help to improve the elements and makeup of future campaigns.



Problems in Advertising Campaign Measurement


Although it is important to measure and forecast advertising campaign effects, it

has been difficult for advertising campaign planners to accurately evaluate and predict the

effects of advertising campaigns over the years.

One of the major challenges in measuring advertising campaign effects is what

should be measured. Advertising communication effects, such as consumer awareness of

a brand, or actual behavior of consumers, such as sales or purchases, can be used to

measure and evaluate advertising campaign effects. According to the campaign objectives

established, advertising planners should decide what measures are best. However, no

matter what objectives were set for the campaign, such as sales, market share,








communication effects, or some other, there are problems inherent in the measurement

process.

Schultz and Bares (1994) summarized the problems in measuring advertising

campaign effects. First, there is a problem in differentiating between what is advertising

and what is not. In most cases, advertising is one part of the whole marketing effort.

Certainly advertising and all other marketing efforts, such as sales promotion and public

relations, do not work in isolation. For example, if a customer sees advertising sales

messages carried on the manufacturer's coupon and remembers it, but not the actual

advertising media message itself, this calls into question whether the advertising campaign

should be credited for the consumers' awareness of the advertising sales message. It is

very difficult to identify actual advertising effects while controlling for all other important

controllable or uncontrollable dimensions of the marketing mix, such as pricing behavior,

distribution intensity, product quality, competitor target market selection, and other

promotion activities. Usually advertising effects can not be separated from other

marketing efforts.

Second, since most campaigns are advertised over several months or even a year, it

is hard to identify the exact effects of one campaign. In most cases, the advertising

campaign effects are built over time and there are lagged effects of advertising campaigns.

Third, the goals of most advertising campaigns are subjective to interpretation.

For example, when the advertising objectives are communication effects, such as

awareness, knowledge, preference, or recall, the measurement of advertising effects often

can not be as precise as those for direct marketing in which consumer response is

measured more accurately.








The fourth problem is attributed to human memory. No one can remember all the

advertisements he or she has seen or been exposed to. Due to the lack of information

about how human memory and information storage works, it is certainly difficult to say

exactly what should be measured to evaluate advertising campaigns.

As specified above, skepticism at the possibility of accurately measuring and

forecasting the effects of advertising campaigns has grown over the years. However, in

spite of all these problems, the results of advertising campaigns must be measured and

predicted because of the advertisers' and advertising planners' needs and requirements for

accountability. Two frequently used approaches are sales and communication effects.



Advertising Effects on Sales


The purpose of advertising is to provide information that changes consumers'

mental and behavioral responses in a manner favored by the advertiser. However, it is

difficult to measure or predict advertising effects accurately because, typically, there is

more than one response resulting from an advertisement. Therefore, advertisers should

decide what is the best consumer response describing the effectiveness of the

advertisement (McDonald, 1995).

Generally, advertising can be measured and predicted at two levels, either

communication effects or sales effects. Sometimes, advertising effects are measured on

the behavior level, which includes not only purchase behavior but also information seeking

behavior before making a purchase. Direct advertising, which is general for industrial

marketing, makes it easy to measure advertising behavior effects. For example, customers

can receive detailed information about a certain product or service, if only they send a








coupon, which is available on a certain part of the advertisement. However, academically

and practically, advertising effects are usually evaluated in terms of communication effects

or sales effects.

Since the ultimate goal of most advertising campaigns is either to increase sales or

market share or to defend achieved positions, many researchers have studied the

advertising-sales relationship (Assumus, Farley, & Lehmann, 1984; Holstius, 1990;

Schreiber, & Appel, 1991; DeKimpe, & Hanssens, 1995a). However, in spite of all the

efforts that have been made to track advertising's effect on sales, it still seems to be an

elusive goal for researchers. Multiple effects exist in advertising and the other

components of the marketing mix. The promotional activities of competitors also easily

confound the measurement of advertising effects on sales (Holstius, 1990).



Difficulties in Measuring Advertising Effects on Sales


One of the major strengths of the evaluation of advertising effects on sales is that

advertising is evaluated in terms of its ultimate objective. However, studies show that

there are some limitations in measuring and predicting advertising effects on sales

(Schreiber, & Appel, 1991; Hoistius, 1990; Lodish et al., 1995). First, although it is true

that the ultimate long-term objective of advertising is to increase sales, it is also

incontestable that advertising frequently has other important short-term objectives, such as

an increase in awareness, knowledge, or recall. It may be true that ultimately all of these

objectives should lead to an increase in sales. However, it is still practically important to

recognize whether these intermediate advertising objectives are achieved or not. The sales

effect measurement approach can not provide the information about the intermediate








objectives of advertising. It should not be underestimated that the effectiveness of each

advertising program element can be evaluated solely in terms of its contribution to sales in

the complexity of the typical advertising efforts. In other words, the measured sales

effects only provide whether or not a particular program element meets the advertising

objectives. However, unlike the intermediate factors, the measured sales effects are not

able to explain why the results occurred and what corrective action is needed.

Second, sales are influenced by so many different factors that it is difficult to

isolate the sales effects provided by one particular element of an advertising program from

those provided by other factors. It is also difficult to measure sales response because of

the lag time between the advertising and the resulting sales. Lodish et al. (1995) indicated

that increased advertising weights increased sales of established brands in only 33 percent

of cases.

Therefore, desired advertising effects are mostly stated in terms of communication

effects in both South Korean and American advertising industries. According to Kreshel,

Lancaster, and Toomey's study (1985), most (79 %) of American advertising practitioners

use communication effects when evaluating advertising media plans. The survey

conducted by the author (1996) also revealed that the majority (90.3%) of the South

Korean advertising practitioners use advertising communication effects to measure or

predict the advertising impact. Only 4.9 percent of media practitioners answered that they

do not use communication effects to predict or measure advertising campaign results.








Advertising Communication Effects


Advertising effects are more often measured on cognitive and attitudinal levels

than sales so that they may be better understood. This cognitive and attitudinal method, of

course, provides the advertiser with more detailed information about the performance of

its communication activities (Holstius, 1990). Alternative measures such as awareness of

a brand or advertising messages delivered, attitude toward the brand advertised,

preference, or purchase intention can be used to evaluate the effects of advertising

campaigns.

Sometimes the purpose of advertising is to create awareness and favorable

attitudes that will ultimately change predispositions to companies and their products.

Most techniques designed to measure this kind of effectiveness are based on theories

about the ways in which consumers process information. These techniques help

advertising planners to determine whether the advertising communication was received

and what the impact was on consumers. For example, if an advertisement successfully

increased peoples' awareness of a product or service, then it would be evaluated to be

effective. In this case, the change in consumer awareness of a product or service is used

to measure advertising effectiveness, and it is presumed that the increase in consumer

awareness ultimately increases sales of the product or service (Cohen, 1988).

The common measures for the measurement of advertising communication effects

are shown in Table 1. Frequently used measures for advertising communication effects

and previous studies are reviewed in the next section.








Table 1. Common measures for advertising communication effects


Stages of Information Common Measures
Process


Cognition Awareness/recognition
Comprehension
Recall

Affect Attitudes toward product
Liking
Preference
Conviction

Conation Intention to try product
Intention to purchase
Actual product purchase





Awareness


Providing either awareness of the brand being advertised or the advertising

message itself is the first level of advertising's communication effects and the first

objective of an advertising campaign. Especially when the product is new or unknown,

the usual objective of the advertising campaign is to increase the consumers' awareness of

the brand being advertised. Therefore, awareness of either the brand or the sales message

is the simplest and the most widely used measure of advertising communication effects

(Woodside, 1996). In a typical measurement of advertising communication effects,

awareness is obtained both for the brand and its advertising message. For an existing and

well-known brand, the usual advertising goal is to develop awareness within the target








market about a specific benefit of the brand or the consumer problem that the brand can

solve. At this stage, the advertiser does not merely attempt to determine whether the

consumers are aware of the brand. If consumers are aware of an advertising message that

communicates specific benefits of the brand, advertising planners may determine that the

advertising achieved its basic objective.

Advertising communication effects also can be measured in terms of advertising

message awareness, instead of brand awareness. For an established brand that is well

known prior to the beginning of an advertising campaign, the awareness of the new

advertising campaign message can be measured at any time after the campaign begins.

The measurement of awareness is generally accomplished on both an unaided and

aided basis. In dealing with the relative utility of measurements between unaided and

aided awareness, Donius (1986) provides a helpful perspective that focuses on the

function of market fragmentation.

If a brand is in a fractionated, highly competitive market, or if it is new, aided
awareness may be the most useful, sensitive measure for that brand. If, on the
other hand, the brand is a household word, dominates its category and/or is in a
highly advertised category with few competitors, spontaneous awareness is
probably going to prove more sensitive and useful.

According to research conducted by Lowe (1984) in the United Kingdom, the

interrelationships among three measures of awareness, such as top-of-mind awareness,

spontaneous awareness, and aided awareness, were examined with over 16 waves of a

tracking study for two leading oil brands. The results suggest that the three measurements

of awareness are, to a considerable extent, measuring the same thing because the three

measurements tend to move together and at much the same rates. The correlation

coefficients of three awareness measurements were over 0.95.








Previous Research on Awareness


To use consumer awareness as a measurement tool of advertising communication

effects, evidence should be presented as to whether consumer awareness of brand or

advertising campaign messages is related to consumers' preference and purchase of the

brand. In a recent study of advertising for box office performance of new film releases

(Zufryden, 1996), the most significant predictors of consumer intention to see a film is

awareness. This study sequentially establishes a connection between advertising

expenditure and consumer awareness, intention to watch the movie and expected ticket

sales at the box office.

According to Gallup research (Mehta & Purvis, 1994) conducted to learn

consumers' top-of-mind awareness levels of competing brand advertising, top-of-mind

awareness is also related to preference. The empirical results showed a strong and

positive relationship between unaided top-of-mind brand awareness and brand preference

as Woodside (1996) stated.

Determining the size of a brand's share of consumers who identify that brand's
advertising in the top-of-mind awareness position may be one useful indicator of
effectiveness of the brand's advertising relative to the advertising of competing
brands.

In Woodside's study, the tests of association indicated that top-of-mind awareness

of an advertising message is related to "brand-mind-position" for three brands tested and

also related to preference for four of the brands investigated. These findings suggest that

a positive relationship may occur between top-of-mind awareness of advertising message

and consumer brand preferences. This study also suggests that advertising exposure leads







to top-of-mind awareness that may affect brand purchase behavior through brand

preference. These schemes are depicted in Figure 1.



1. -1 .
F I- r -



I




i r-






Figure 1. The effects of top-of-mind awareness of the advertisement

Source: adaptedfrom Woodside (1996).


Recall

Advertising planners can use recall to determine whether or not consumers are

exposed to the advertising campaign. Those who are exposed to an advertising campaign

might repeat or play back certain portions of ideas they have seen or heard.

Although recall of sales messages may have some effect on the consumers' future

purchase behavior, the relationship between recall and future consumer purchase seems to

be somewhat tenuous. According to a recent study by Lodish et al. (1995), the








examination of 389 real-world split cable television advertising experiments shows that

recall is not strongly related to sales.

Similar to awareness tests, two types of recall, such as aided recall and unaided

recall, are used in evaluating advertising campaigns. However, since the respondents are

asked only about the product category and are expected to spontaneously remember the

advertising, the unaided recall is believed to be the more useful measure.

When the marketing or advertising manager's objective is to determine the extent

to which consumers have learned or remembered advertising content, "day-after-recall

(DAR)" has been used as a measure of advertising effectiveness (Higie & Sewall, 1991).

DAR is perceived as an important factor to increase a brand's probability of becoming a

member of a consumer's evoked set (Stewart, 1986 and 1989). Therefore, in spite of the

arguments against the use of DAR as a measurement tool of advertising effectiveness,

DAR is frequently employed by advertising managers (Walker & Von Gonten, 1989).



Liking and Attitude Toward Advertisement


Liking assumes that consumers are aware of the brand and have some knowledge

of the brand either from the advertising or from actual product use. Usually, it is difficult

to separate advertising effects on consumer liking from experience, which makes

measurement more difficult. In the liking measure, it is assumed that there is an

advertising effect on the consumer's mental condition or some kind of attitude change.

The consumers move from the awareness or knowledge stage and form a positive

opinion about the product or service. However, liking does not necessarily mean that the

consumer will purchase the product. It simply means that there exist positive feelings or








impressions about the product or service. For example, liking can be measured by having

a consumer name several acceptable brands of products in a category. The assumption is

that if the products are acceptable, they are liked. Although liking is an important step to

evaluate the results of an advertising campaign, the liking measure still does not assume an

actual purchase action. Consumers may like many products but only purchase one or two

(Schultz & Barnes, 1994).

On the other hand, an individual's evaluation of an advertisement, as measured by

his/her attitude toward the advertisement, is thought to be an important mediator of the

advertising effects on brand attitude. Recently, there has been considerable research

conducted to understand the antecedents of the relationship between attitude toward the

advertisement and brand attitudes (Homer, 1990).

Research shows that attitude toward an advertisement is affected by a brand or

non-brand processing set, such as advertising exposure level (Burke & Edell, 1986; Cox &

Cox, 1988), message involvement (Park & Young, 1986; Muehling & Laczniak, 1988),

advertising message quality/content (Hastak & Olson, 1989), and effective responses

generated during advertising exposure (Burke & Edell, 1989; Machleit & Wilson, 1988).

Attitude toward an advertisement has also been observed to be related to the recall

of an advertisement (Pieters & Klerk-Warmerdam, 1996), purchase intention (Mitchel &

Olson, 1981), and attitude toward the act of buying the brand (Mitchel, 1993).

Biehal et al. (1992) suggested that attitude toward the advertisement also had a

direct and positive effect on brand choice. This may be an important issue for advertisers,

who should be concerned that attitude toward advertisements affect not only attitude

toward the brand and purchase intention but also brand choice.








Preference


Preference is defined as one brand being selected over the others among a number

of brands within a certain product category. In advertising campaign evaluation, it is

assumed that the advertised brand is preferred and likely purchased by consumers among

available alternatives, because advertising messages create a level of acceptance for a

specific brand.

Therefore, brand preference is another frequently used measure of advertising

effectiveness and has been of interest to marketers. If the major goal of advertising is to

create product preference, whether or not any actual sales results, the advertising

campaign is usually considered successful if the advertising only creates warm feelings

about the brand. Once a brand is preferred to other brands, it is usually believed that the

advertising or sales messages can do more to trigger the purchase decision. However, the

consumers may or may not purchase the product because there are still many market

factors, such as price, availability, and sales promotion that may influence consumer

buying decisions (Clark, Brock, & Stewart, 1993).

A commercial's ability to affect brand preference usually is measured as the

difference between before and after exposure to the advertisement (Fletcher & Bowers,

1988). However, only a few studies have examined the reliability of preference. In one

study, Stewart and Furse (1986) estimated fairly good r-squared (0.86) for the use of

preference, using test-retest methods.

In another study, Higie and Sewall (1991) suggested that although measuring

preference is more closely related to actual purchase, preference is less reliable than day-








after-recall or other communication effects. They also indicated that it is difficult to

measure significant differences in preference based on a single exposure to an

advertisement.



Purchase Intention and Actual Purchase


If an advertising message is successfully communicated with consumers and other

marketing variables are favorable, the final step of consumer reactions is purchase

intention or actual purchase of the brand. Like other measures explained earlier, it is

assumed that the target audience is exposed and reacts to the advertising campaign

messages. The consumers' awareness of the brand advertised may help develop a liking,

move through preference, and drive to the final step, such as purchasing the brand.

Possible measurement of advertising effects in terms of purchase intention is to identify

advertising as the force that creates a consideration of future brand switching or purchase

(Dutka, 1995).

According to Poiesz and Henry (1994), studies for advertising communication

effects are limited to the investigation of psychological dependent variables, such as

awareness, recall, or attitude toward a brand that are only assumed to precede choice

effects. Although the current availability of product scanner data and other point-of-

purchase systems can provide behavioral data on individual households, purchase

behaviors may be the most difficult concepts to measure because of the many variables

involved, like advertising effects on sales (Poiesz & Henry, 1994).

Nedungadi, Mitchell, and Berger (1993) also argued that although advertising

may have a variety of effects on consumers, an advertising campaign that increases








awareness or improves attitudes is only partially successful if the campaign does not

ultimately influence brand choice. Most evaluation questions on this level deal with what

was purchased or what will be purchased in the future. If change, such as brand

switching, did occur, or if there is an indication of future brand switching, research can be

attempted to relate the behavior change to the advertising exposure.



Achieving Desired Communication Effects


The common measures and previous studies of advertising communication effects

have been explained in this chapter. The desired communication effects can be consumer

awareness of a brand or advertising message, recall, attitude change, preference, or

purchase intention. No matter what communication goals an advertising campaign has,

target consumer's cognitive and attitudinal process is initiated by advertising exposure.

Therefore, one of the fundamental concerns of advertising planning and practice is to

develop and evaluate advertising media schedules in terms of advertising exposure.

The desired communication effects typically can be accomplished when target

audiences are exposed to the advertising campaign with optimal frequencies. For

example, if an advertising campaign goal is to achieve 50 percent awareness of the

advertising campaign message, the advertising planner should estimate how much

repetition can help accomplish the goal. Although no study was attempted to examine

how many times a consumer should be exposed to an advertising campaign to achieve

each level of the communication hierarchy described above, it is assumed that the optimal

frequency may vary with the desired communication objectives established. For example,

conative objectives such as purchase intention may need more consumer exposure than








cognitive goals such as awareness, because the target audience should first be aware of the

existence of a brand with about 3 to 5 exposures to the advertising message and then,

more advertising exposures after awareness may develop purchase intention through

preference and positive attitude toward the brand. It is assumed that consumers need

more information about the brand to reach purchase intention than awareness. The

additional exposures to advertising message will provide more information about a certain

product or service with the target audiences.

Affective and conative measures also may be influenced more by marketing, copy,

and media factors, such as product quality, advertiser image, and message creativity, than

cognitive measures. In spite of enough exposures to advertising messages, consumers

may not reach purchase intention, for instance, if a brand has poor quality and a weak

advertiser image. Table 2 shows the possible relationship between common

communication measures and needed exposures to achieve the goals in a normal

marketing, copy, and media situation. In other words, when ignoring all other influential

conditions such as marketing, copy, and media factors, optimal frequency of consumer

exposure to advertising to achieve cognitive goals may be lower than affective and

conative objectives. For example, if a brand needs consumer exposure to the

advertisement at least three times to gain 50 percent of awareness of the brand, consumers

might need to be exposed six times to achieve preference, and ten times for purchase

intention. In this case, the minimum numbers of consumer exposures to the

advertisement to achieve awareness (a+), preference (b+), and purchase intention (c+) are

3+, 6+, and 10+ in a normal marketing situation. However, until further study identifies a

scientific method to determine the needed exposure frequency for each level of consumer








information processing, it is up to advertising planners to decide how many exposures are

needed to achieve each communication goal established.


Table 2. Relationship between common communication measures and needed exposure
frequency in a normal marketing, copy, and media situation.


Common Measures Minimum Exposure
Frequency Neededfor
Average Consumer

Cognitive Awareness/Recognition
Recall a+

Affective Attitude toward brand
Liking b+
Preference (a < b)

Conative Purchase Intention c+
Actual purchase (a < b < c)


The concepts shown in Table 2 will be integrated with marketing, copy, and media

factors later in this chapter to estimate optimal frequency level for a selected case study.

Since this study is to estimate the optimal frequency of advertising exposure from the

prospective of media planners, the terms used to explain effective exposure to advertising,

such as reach, frequency, effective reach and effective frequency are reviewed to illustrate

what should be achieved to generate the desired advertising communication effects.








Effective Exposure to Advertising



Advertising Exposure


The meaning of exposure is "open eyes or listening ears" facing the advertising

vehicle or message (Sissors & Bumba, 1996). The most common standard to evaluate the

performance of an advertising media vehicle is its rating. Rating can be defined as the

percentage of people exposed to a media vehicle or message and expressed as a

proportion of a selected population base. The rating has been widely used to measure the

performance of media vehicles in terms of advertising vehicle delivery.

However, since audiences exposed to a media vehicle are not automatically

exposed to the actual advertising message, the rating can not indicate what percentage of

the target audience that is exposed to an advertising message within the vehicle.

According to Sissors and Surmanek (1986),

Vehicle exposure represents an opportunity-to-see (OTS). The number of
opportunities-to-see that a media vehicle develops does not guarantee any
exposure to advertising.

Since it is possible to be exposed to the vehicle without exposure to the advertising

message, generally, the vehicle audience is bigger than the advertising audience. In spite

of the difference between vehicle and message exposure, due to the lack of advertising

exposure data and the technical difficulties to measure accurate advertising exposure,

many media practitioners have frequently used vehicle exposure instead of advertising

exposure (Kreshel, Lancaster, & Toomey, 1985).

However, it is certain that the cognitive process explained earlier is initiated by

exposure to an advertising message, not exposure to the vehicle. Therefore, it is the








advertising exposure that should be measured and used for the estimation of media

evaluation factors such as reach, frequency, effective reach, and effective frequency.

In this study, advertising exposure represents exposure to advertising message, not

exposure to the vehicle. The estimation of media evaluation factors is explained based on

the exposure to advertising message in order to measure or predict advertising campaign

effects.



Reach


Recent developments suggest that an important element of the advertising media

decision-making process should be the evaluation of reach and frequency in terms of

advertising exposures or communication effects (Kreshel, Lancaster, & Toomey, 1985).

Reach (sometimes called "coverage") is a measurement of the proportion of the

population that is exposed to an advertising campaign at least once within a given period

of time. Reach is generally explained as a percentage, such as 20, 30, or 40 percent. For

example, if the target market of brand X is 4 million and 2 million of them are exposed to

the brand X's advertisement at least once within a given period of time, the reach would

be 50 percent. However, since reach is accumulated, the duplication should be subtracted.

Suppose only two magazine advertisements were used for brand X's advertising

campaign. If the two advertisements were read by 30 and 20 percent of target audiences,

and 10 percent of them were exposed to both advertisements, the reach of this schedule

would be 40 percent [ (30 + 20) 10 = 40) ].

Reach is usually built in a fairly consistent pattern over time. Using television as

an example, the first time an advertisement is telecast, it accumulates the largest number of








viewers. The second time it is telecast, many of the viewers are repeat viewers. For the

third time, even fewer new viewers may see the advertisement (Sissors & Bumba, 1996).

The shape of a typical reach curve is convex as shown in Figure 2.



Households
Reach %

50
40
30
20
10

Week 1 2 3 4 5



Figure 2. The shape of a typical message reach curve

Source: adaptedfrom Sissors and Bumba (1996).



Frequency and Frequency Distribution


Frequency is a companion statistic to reach and indicates the average number of

times that target audience members are exposed to an advertising message. Whereas

reach is a measure of message dispersion, indicating how widely the advertising message

may be received in a target market, frequency is a measure of repetition, indicating how

many times a target market is exposed, on average, to the advertising message (Sissors &

Bumba, 1996).








For the prediction or measurement of the desired advertising communication

effects, such as awareness, recall, preference, or purchase, these two concepts, reach and

frequency, have been widely used (McDonald, 1995). An interesting example was

provided by a research agency, Millward Brown (Brown, 1993), claiming that the increase

in advertising message awareness can be produced by every 100 TVRs (Television Rating

Points).

The average frequency is calculated by dividing the gross rating points (GRPs) by

reach. For example, if the gross rating points of twenty TV advertisements is 80 and the

percent of the target audience exposed to the advertising schedule at least once (reach) is

40 percent, the average frequency would be 2 (80 + 40 = 2).

A frequency distribution is the percentage of the target audience exposed to each

frequency level of an advertising schedule. Since this study uses advertising message

exposure, instead of vehicle exposure, frequency exposure can show the distribution

patterns of message reach at each frequency level. For example, if an evaluation of an

advertising campaign predicted the frequency distribution shown in Table 3, it is assumed

that 70 percent of the target market is exposed to the advertising message at least once

and 30 percent of the target market is exposed to the advertising message more than two

times with the plan.

To achieve desired communication effects, target audiences need to be exposed to

the advertising message a certain number of times, such as three or four times. Since the

required number of exposures may vary with each advertising campaign or plan, frequency

distributions demonstrate more pertinent information than do individual reach and








frequency when comparing alternative plans. Frequency distributions also provide

planners with a method of determining the pattern of repetition that the plan may provide.



Table 3. Example of frequency distribution


Frequency of Message Exposure Advertising Plan


1 24%
2 16%
3 9%
4-5 6%
6-8 5%
9-12 5%
+13 5%

Average Frequency 4
Total Message Reach 70
Message Reach at 3+ Level 30

Source: adaptedfrom Sissors and Bumba (1996).



Effective Frequency and Reach


In recent years, one of the most significant changes in advertising media planning

and evaluation is the development of the concepts of effective reach and frequency

(Sissors & Bumba, 1996). With these concepts, media planners can understand a number

of important factors that help decide which plan is better among alternatives or how much

repetition is needed to accomplish the expected advertising communication effects.

Effective frequency is the number of advertising exposures needed to accomplish

desired communication effects on the target audience. Since effective reach is defined as

the percentage of the target market exposed to an advertising campaign schedule at the








level of the effective frequency, determining the effective frequency level is one of the

major steps in predicting or measuring advertising campaign effects. Advertising media

planners may decide the effective frequency level needed through test marketing and study

responses to previous advertising schedules. However, sometimes, there is no data by

which to determine objectively how much repetition is necessary to achieve the desired

communication effects. In this case, the planner's estimation may have to be based on

either experience, or specialized research.

Much of what is known about the effects of frequency can be traced to

psychologically trained researchers who explored the subject in a laboratory environment.

One of the major psychologically trained researchers is Krugmen. According to his "three

hit theory" (1972), the first exposure to an advertisement is only perceived by audiences to

raise questions about "What is it?" After the second exposure, the audiences may ask

"What of it?" Then, the second exposure makes audiences react to the advertisement and

begin to compare with alternative brands. The third exposure is a reminder of the other

two exposures and may be the optimal advertising frequency.

Many other studies (Craig & Ghosh, 1993 and Sterling, 1997) have hypothesized

that audiences begin to respond more effectively to advertising from the third exposure to

tenth exposure.

According to Sterling (1997), reach and frequency analysis does not tell us enough

about whether the advertising has been sufficiently exposed to the target audience. The

use of reach (the percentage of target audience members potentially exposed to an

advertising message at least once) is misleading because it often suggests that most any

media plan successfully reaches the vast majority of its intended target audience.








Consequently, Sterling suggests that effective frequency can solve such a problem because

effective frequency is the minimum number of times that target audience members should

be exposed to the advertising campaign in order for the advertising campaign to

accomplish its goal. He said that although the number varies from campaign to campaign,

common levels of message effective frequency are between 3+ and 6+. Threshold effects

tend to occur below three exposures. It means that there will be no or little advertising

effects below the threshold level. An S-shaped response function indicates that there is a

threshold effect.

On the other hand, Naples (1979) suggested that the optimal exposure frequency

should be at least 3+ within a purchase cycle and a "convex-shaped" response curve is

expected after three exposures. He also indicated that advertising copy, market share, and

total media expenses are significant factors to decide exposure frequency levels.

According to him, too many exposures can occur negative responses to the brand.

Sissors and Bumba (1996) also argued that many researchers of advertising

response curves have not observed the S-shape, which is the only shape that can explain

the threshold effect. In fact, "convex-shaped" curves have been frequently found. The

advertising effective frequency level may also vary in response to many other marketing,

copy, and media factors, such as consumer involvement with the advertising message or

product, extent of brand loyalty, product life cycle, purchase cycle, creative execution,

extent of clutter, and other specific market or brand conditions (Ostrow, 1982; Sissors &

Bumba, 1996).








Problems with Using the Concept of Effective Frequency


Although many studies of the concept of effective frequency (Elliott, 1985;

Kreshel, Lancaster, & Toomey, 1985; McDonald, 1995; Leckenby, & Kim, 1994) showed

that effective frequency is widely accepted among media practitioners in the United States,

there have been a number of questions about the use of the effective frequency concept.

The first problem is the differentiation within product categories. Schumann et al.

(1990) indicated that when subjects were exposed to the advertisements only under

conditions of high rather than low involvement, repeated advertising exposure led to more

positive attitudes toward the advertised brand. Recently, a growing number of media

planners suggest that there should be differences among various product category

frequency levels. Although there is some speculation that, for instance, low-involvement

products may need more frequency than high-involvement products, there is still no

practical measurement to differentiate the frequency levels needed for different kinds of

products (Sissors & Bumba, 1996).

Another problem with the concept of effective frequency is whether a "threshold

effect" exists. As Krugmen suggested, many media planners in the US still believe that

there is a threshold effect and that advertising begins to be effective with the third

impression. However, according to Simon and Arndt (1980), most results of studies

concerning the threshold effect showed that the advertising response curves are mostly

convex, not S-shaped. Furthermore, many direct marketing practitioners objected to the

threshold, arguing that there is no answer about the tremendous consumer responses to

their first impressions (Sissors & Bumba, 1996).








The relationship between advertising frequency and its wear-out is another

problem. Studies show that repeated exposure to an advertisement after a certain number

of initial exposures might actually produce negative feelings toward the brand (Calder and

Sternthal 1980 ; Pechmann, & Stewart 1988 ). On the other hand, according to Russell &

Verrill (1986), each additional frequency increases advertising communication effects,

such as recall, recognition, and purchase up to 20 exposures. Therefore, the main problem

that media planners have is when, and under what circumstances, the frequency effects

wear-out. Until further research provides the answer, planners seem to be responsible for

the judgement based on their subjective evaluation.

McDonald's study (1982) also recommended that product-purchase cycle should

be considered to decide the effective frequency level. However, since each customer in a

product category has a totally different purchase cycle, it may not be useful to help

planners decide an effective frequency level. For example, if a product has a purchase

cycle of two months, each customer may purchase the product on different days during

the 60 days, such as the first day of the first month, or the last day of the second month

(Sissors & Bumba, 1996).

There are more problems to be considered. The quality of advertising message

also may affect the effective frequency level because uninteresting copy may require much

more frequency. As explained earlier, the difference between vehicle and message

exposure also may confuse the decision of an effective frequency level. According to

Lancaster et al. (1986), due to the lack of information or technical difficulties, more than

half of advertising practitioners in the United States use vehicle exposure, in spite of their

understanding of the difference between vehicle and message exposure. However, since








the difference between vehicle and message is apparent (Joyce, 1984; Sissors &

Surmanek, 1986; Lancaster & Katz, 1988; Lee, 1996), for the media planners to use the

effective frequency concept, they should focus on how much message frequency is

necessary to achieve the desired advertising exposure. It also might be necessary for

media planners to understand the ratio of message to vehicle exposure.



How to Set Effective Frequency Levels


One of the difficult issues facing any advertising planner is that of deciding how

much effective frequency is enough for an advertising campaign. Since effective frequency

levels vary with marketing, copy, and media factors, those factors should be considered

when choosing an effective frequency level for each advertising campaign.

One suggestion about how to set a frequency level was specified by Ostrow in

1982. According to Ostrow,

The right level of frequency for an advertising campaign is the point at which
effective communication takes place. For example, getting consumers to
understand the advertising message; helping consumers become more positive
about a product or influencing the purchase decisions directly.

He itemized a number of conditions that should be considered based on marketing,

media, and creative strategy factors. For example, new brands need higher frequency

levels than established brands, because new brands need to be learned by the target

market. Planners can use Ostrow's items (see Table 4) by adding or subtracting points

from a base such as 3+ or 4+. The base levels can be chosen according to the

communication goals established. For example, the base may be 3+ for consumer

awareness of a brand while the base may be increased to 6+ to achieve consumer








preference. In Table 4, Sissors and Bumba (1996) suggested specific points with

Ostrow's categorized items. However, these items and points are only suggestions.


Table 4. Marketing, copy, and media factors that affect effective frequency


Marketing factors that affect effective frequency.

Established brands (2)-.1 +.1 +.2 New brands

High market share -.2 -.1 +.1 1 I Low market share

Dominant brand -.2 -. 1 +2 Smaller, less well-known brands
in market

High brand loyalty -.2 -.1 +.1 Low brand loyalty

Long purchase cycle -.2 -.1 +.1 +.2 Short purchase cycle

Products used daily -.2 -.1 +.1 72) Products used occasionally

+.1 ) Needed to beat competition


Copy factors that affect effective frequency


Simple copy

Copy more unique than
competition

Continuing campaign

Product sell copy

Single kind of message

To avoid wearout:
New messages

Large ad units


-.2 +.1 +.2

-.2 +.1 +.2


-.2 -.1 +.1

-.2 -.1 +.1 X

-.l +.1 +.2


-.1 +.1 +.2

0-.l +.1 +.2


Complex copy

Copy less unique than competition


New copy campaign

Image type copy

More different kinds of messages


Old messages

Small ad units








Media factors that affect effective frequency.

Lower ad clutter -.2 U1 +.1 +.2 High ad clutter

Compatible editorial -.2 -.1 +.1 +.2 Non-compatible environment
environment

Attentiveness high -.2 -.1 +.1 +.2 Attentiveness low

Continuous advertising -. +.1 +.2 Pulsed or flighted advertising

Few media used -.2 ( +.1 +.2 Many media used

Opportunities for media )-. +1 +.2 Fewer opportunities
repetition

Source: adaptedfrom Sissors & Bumba (1996) p251-52.
Circled points are applied to the example of brand X described below, while the
remainingfactors are assumed to be neutral.


For example, the effective frequency for a certain brand X can be estimated with the

factors in Table 4. If brand X is an established brand which has a strong market leading

competitor with low brand loyalty and occasionally used, the sum of the scales with

marketing factors will be (+ 0.7). For the copy factors, if brand X has a new advertising

campaign with simple image type copy that is more unique than the competition, single

kind of new message, and large advertising units, the sum of scales applied with copy

factors will be (- 0.4). Finally, media factors can be applied. If the advertising campaign

for brand X is continuous and also has low advertising clutter, high opportunities for media

repetition, and a few advertising media, the sum of scales applied to the example with

media factors will be (- 0.6). When the remaining factors are assumed to be neutral, the

sum of scales applied to brand X with three factors will be (- 0.3), [ (+ 0.7) + (- 0.4) +

(- 0.6) = (- 0.3) ].








Therefore, if the advertising campaign goal is to gain 50 percent consumer

awareness of brand X and the media planner decided the base is 3+, the estimated effective

frequency would be 2.7 or 3+. In this case, the base 3+ indicates that at least three

exposures are needed to accomplish the campaign goal when ignoring marketing, copy, and

media factors.

However, different items or scales may be applied for the estimation of effective

frequency levels as the situation dictates. If the brand X company has a relatively good

image (-. 1) and high advertising message quality (-. 1) but more active competitor

promotional performances (+.2) and relatively poor product or service quality (+.2), the

estimated sum of points will be (-. 1). Table 5 shows the additional points which are

suggested by the author and applied to brand X. It is assumed that the effective frequency

base for advertising message awareness is 3+, then the minimum bases for preference (b+)

and purchase intention (c+) can be estimated as follows:

b > 3 or at least 4+

c > 4 or at least 5+

In addition to Ostrow's items and Sissors and Bumba's point scales, the items suggested

by the author can be applied to the bases of awareness (3+), preference (b+), and purchase

intention (c+). Therefore, the modified effective frequency levels for awareness,

preference, and purchase intention estimation are shown in Table 5.








Table 5. Suggested additional points and modified effective frequency levels for brand X

SuggestedAdditional Points


Good advertiser image -.1 Advertising message quality (high) -.1
Product or service quality +.2 More competitor promotional activities +.2
(relatively poor)



Sum of Points from the Ostrow's items = (- 0.3)
Sum of Points from the additional items = (+ 0.2)

Total Sum of Points = (- 0.1)


Communication Goal Effective Modified Effective
Frequency Frequency Level
Base

Awareness of Brand X 3+ 3.0 0.1 = 2.9 or 3+

Preference b+ 4.0 0.1 = 3.9 or
(b > 3 or at least at least 4+
4+)

Purchase Intention c+ 5.0 0.1 = 4.9 or
(c > 4 or at least at least 5+
5+)



Since one of the main goals of this study is to apply a frequently used American

media evaluation model to the South Korean market, background information on the

South Korean advertising industry will be reviewed in the next chapter. Survey

research conducted by the author and presented in Chapter III also shows how South

Korean media practitioners perceive the current state of reach and frequency

estimation.














CHAPTER III
BACKGROUND ON THE SOUTH KOREAN ADVERTISING INDUSTRY






Introduction


In this study, a frequently used American media evaluation model will be applied to

a South Korean case study. Therefore, it is necessary to understand South Korean

economic development, the growth and recent changes in the advertising industry,

advertising media, and the present state of South Korean advertising. Also in this chapter,

South Korean media practitioners' perceptions of media evaluation terms and concepts,

such as effective frequency and the current state of reach/frequency methods are reviewed.



South Korean Economic Development


The Republic of Korea has become the object of growing international interest

because of its rapid rise to world industrial prominence. The South Korean economic

growth has been driven by the export market, which rose from 20.6 percent of GNP in

1972 to 45 percent by 1988. According to Choi (1994):

The composition of the South Korean exports has changed dramatically from
agricultural and simple labor-intensive goods to increasingly more sophisticated
value-added products, including steel, petrochemicals, automobiles, consumer
electronics, computers and large shipping vessels.








With the dramatic growth of exports, the South Korean per capital GNP has been

increased from $1,012 in 1977 to $4,994 in 1989 (AD DATA, 1995). The total

population of South Korea was 36 million in 1977 and 42 million in 1989. South Korea's

world trade success has not been simply the result of entrepreneurial talent and the

contribution of high quality, but of low-cost labor as well. South Korean economic

development has been planned, directed, and controlled by aggressive government

intervention and financed by government-controlled banks (Koo, 1993).



The Growth of the Advertising Industry in South Korea


In Korea, the first modem advertisement was seen in the "Han-Sung weekly

newspaper" in 1886; it advertised a German trading house which sold products imported

from Germany (Shin, 1986). However, successive political, economic, and social

disorders deterred the development of a capitalistic economic system. Accordingly,

advertising and industrial activity did not fully develop in relation to other specialized

fields and was generally held in lower esteem than other business. The stages of Korean

economic development and the circumstances of consumer life were so different from

those of advanced countries that consumer awareness of advertising was relatively low

(Yoon, 1980).

After the Korean War ended in the 1950s, the Korean advertising agency business

had to start over because the few Japanese agencies operating in Korea during its

occupation (1910-1945) left Korea after World War II (Shin, 1989). The introduction of

television broadcasting, which was initiated by the American corporation RCA, greatly








facilitated the growth of advertising in South Korea (Yoon, 1994). South Korean TV

broadcasting has quantitatively and qualitatively broadened since the introduction.

Along with the influence of the United States on South Korean TV broadcasting,

foreign professionals also initiated the establishment of advertising agencies in the 1960s.

A media representative company, Impact, started its operation as an advertising agency in

1962 (Shin, 1989). The birth of Manbosa Advertising in 1969 led the market entry of

'COKE' and 'Pepsi' and helped to shape agency recognition and the commission system

(Kim, 1996).

The first Korean Advertising agency, Hap-Dong, appeared in 1967 as a technical

"tie-up" with a Japanese advertising agency, McCann-Erickson Hakuhodo. However, it

was a failure. A successful Korean advertising agency did not appear until the Korean

advertising agency, Cheil Communications, started in 1973. With the success of Cheil

Communications, several advertising agencies such as Oricom and Yon-Hap Advertising

began their business in 1973. The progress of Cheil Communication's annual billings,

South Korean per capital GNP, and total expenses for advertising in South Korea

increased drastically as shown in Table 6.

The South Korean advertising industry has grown very fast, primarily as a result of

the democratic reforms of 1987, the development of a free press and the effects of the

1988 Summer Olympics held in Seoul. South Korea has a population of about 44 million,

a per capital GNP of US $10,994 (1996), and spent US $6,184 million on advertising in

1995 (AD DATA, 1995).








Table 6. The progress of the South Korean advertising industry

Year Cheil Billings Per Capita GNP Total Expenses For
(US $) (US $) Advertising
(Us $)

1973 1,100,000 361 28,875,000
1975 5,000,000 532 81,250,000
1983 61,000,000 2,014 706,625,000
1994 575,492,500 8,483 5,355,000,000
1995 750,125,000 10,096 6,184,071,250
(US $1 = 800 Won)



Advertising Media in South Korea


In South Korea, newspaper, television, radio, and magazines are the four major

advertising media. Newspaper advertising has grown very fast for the last 10 years and it

covers approximately 56.4 percent of current total advertising expenses in South Korea.

There are about 90 newspapers that can be used as advertising media. The cost for each

1/3 page advertisement with four major national newspapers is about $900 to $2,400 per

advertisement.

TV is the second most widely used medium. About one third of total advertising

expenses (34.3%) are spent on television advertising. Currently, there are three network

television channels that can be used as advertising media in South Korea. Like the United

States, South Korean television advertising is categorized as program advertising and spot

television advertising. Program advertisements are 15, 20, and 30 second in length. The

cost depends on the popularity of each television program and the time when it is on the








air. In South Korea, most program advertisements are 15 seconds and the cost range is

$8 to $46 per second. On the other hand, spot television advertisements are on the air

between two programs. Most spot television advertisements are 30 seconds and cost

about $60 to $240 per second.

There are approximately 450 magazines that can be used as advertising media in

South Korea (Park, 1995). Among them, women's magazines and news magazines for

men are the most widely used advertising media. The estimated cost for a one page

advertisement ranges from $588 to $2,500 in South Korea. The advertising media, total

expenses and costs in South Korea are summarized in Table 7.


Table 7. South Korean advertising media, total expenses, and costs

Newspapers TVNetworks Magazines Radio Stations


No. of Advertising 90 3 450 7
Media

Total expenses in
1995 (US $) 2,676,323,000 1,627,854,000 220,170,000 216,734,000


Cost Range $588 2,400 Program Ads. $540- Program Ads.
for one 1/3 $8 ~ 46 2,400 $3 ~ 14
page with per second for per second
major national one page.
newspapers Spot TV Ads. Spot Radio
$60 240 Ads.
per second $6 ~ 25
per second

(US 1$= 800 Won)








Recent Changes in the South Korean Advertising Industry


Despite its growing market size, the South Korean advertising industry prohibited

foreign investment until the 1980s. It was the United States that estimated the market

potential and began to address market liberalization. During the 1985 American South

Korean Economic Consultations, liberalization of the South Korean advertising industry

first emerged as a trade issue. In 1986, the American trade representatives became much

more aggressive and then, during the ninth American-South Korean Economic

Consultations in 1988, South Korea announced liberalization of its advertising industry

(Kim, 1990).

Along with the liberalization of the economy and the increased economic power,

South Korean mass media started to be expanded dramatically to satisfy diverse consumer

interests. The number of print media outlets was doubled and the number of weekly

newspapers was increased about five times between 1987 and 1990. Daily newspapers

also increased their editorial and advertising pages (1996, Kim). Broadcasting media

outlets also have increased in number. The new private broadcast network, SBS, opened

in 1991 and emerged as a new media channel.

Another potential medium for advertising space and time is cable television. The

South Korean CATV began service in March 1995 with 15 to 20 channels and plans to

increase the channels up to 40 within the next five years. Therefore, the coming CATV

era in the next five years in South Korea will provide much more opportunities for

advertising (Kim, 1996).








The opening of the South Korean advertising industry to foreign companies also

brought both the establishment of an audit bureau of circulation and the development of

television rating services that were awaited by the South Korean advertising industry. In

1989, although it has been experiencing difficulties, the Korean Audit Bureau of

Circulation (KABC) was finally established and started to organize circulation audits of

print media. In addition to the KABC, the Korean advertising industry also prompted the

need for a television rating system. Along with the establishment of the Korea Survey

Gallup Polls (KSG) in 1990, three additional media-rating companies began operating

such as Media Service Korea (MSK), Lee's PR, and Korean Marketing Research (KMR)

(Park, 1995).

The opening of the South Korean advertising industry also prompted advertising

agencies to turn to market research and audience studies. Since the leading South Korean

advertising agency, Cheil Communications, established the first marketing research

institute in 1991, other major large agencies, such as LGAd, Daehong, KeumKang, Korad,

Oricom, and Seoul DMB&B opened their research institutes (Kim, 1996).



Studies on Measuring and Forecasting Advertising Effects in South Korea


Although the studies on measuring and forecasting advertising effects have been

active in the advanced countries, such as the United States and Japan, since the early

1900s, there has been little systematic study of evaluating advertising effects in South

Korea. It is assumed that some South Korean advertising agencies have developed their

own advertising evaluation models that are applicable to the South Korean advertising








environment since the late 1980s. However, those models are proprietary and therefore

are not available for access from outside of the agencies (Kim, 1996).

In South Korea, the first study in measuring and forecasting advertising effects was

conducted by Sung in 1989. In his study, "The establishment of a measurable model for

pre-estimation and evaluation of advertising effects in television commercials," Sung tried

to develop a model of predicting and measuring advertising effects that is applicable to the

South Korean environment. Based on the Japanese JNN Data Bank (JDB) model, as

shown in Figure 3, he suggested that GRPs or exposure to advertisements lead to

awareness that may affect brand purchase through brand preference. Using regression

analysis, he examined the relationship between GRPs and advertising exposure, GRPs and

brand awareness, awareness and preference, preference and purchase, awareness and

purchase, and GRPs and purchase. The results of this study showed that all these six

relationships are statistically significant and different regression equations are suggested.

The independent and dependent variables used in his study and regression equations are

shown in Table 8.

However, his study identified problems for future studies. Sung assumed that all

the other intermediate factors, such as competitors' promotional activities and quality of

advertising message that may mediate the relationship between independent and dependent

variables, are the same for all brands studied. His model also may not be applicable to

established brands because he used only new products.

As explained earlier, although some South Korean advertising agencies have been

trying to develop their own models to evaluate and predict advertising effects, most of

them are not published. However, in 1991, LGAd, which is one of the top five advertising








agencies in South Korea, developed an advertising effectiveness measuring model (AEM)

and the model was introduced by Lee and Park (1995). In addition to brand awareness,

preference, and purchase, this model also considers advertising message awareness and

advertising preference to predict advertising communication effects (Lee & Park, 1995).







GRPs Brand Preference Purchase
Awareness


-_F I ___

Exposure to
Advertisements.



Figure 3. Sung's model to estimate advertising effects



Table 8. Sung's independent and dependent variables and regression equations

Independent Dependent Regression Equations
Variables Variables for Food Products



GRPs Exposure to Ads. Y = 100 61.207128exp(-0.000125065X)
GRPs Brand Awareness Y = 100- 55.473exp(-0.0000135X)
Brand Awareness Brand Preference Y = 2.37916exp(0.032628345X)
Brand Preference Brand Purchase Y = 1.185111X + 1.85653
Brand Awareness Brand Purchase Y = 2.35262279exp(0.035743144X)
GRPs Brand Purchase Y = 100 95.2616282exp(-0.000091683X)








Although the South Korean advertising industry has been growing very fast, the

evaluation of advertising effects is still in its early stage. Therefore, it is necessary that a

new measurable model of advertising effects that fits to advertising environment of South

Korea should be developed to predict and measure advertising effects. From the two

previous Korean studies, it is also assumed that awareness, preference, and purchase or

purchase intention are the most widely used measures to predict or evaluate advertising

communication effects in South Korea.



Media Planning and Evaluation Procedures of South Korean Advertising Practitioners


Chapter II introduced concepts and studies dealing with the aspects of media

planning, such as the kind of communication effects that can be used to predict or measure

advertising effects, how effective reach is typically measured to achieve communication

effects, and the minimum and maximum levels of frequency required for target market

members to be effectively reached.

Studies of the development of media plans, media directors' perceptions of

effective reach/frequency, and the consideration of weighting and timing factors have been

frequently conducted in the US (Leckenby & Kishi, 1982b; Leckenby & Boyd, 1984;

Kreshel, Lancaster, & Toomey, 1985; Lancaster, Kreshel, & Harris, 1986; Leckenby, &

Kim 1994). However, in spite of the fast growth of the South Korean advertising

industry, no such research has been conducted in South Korea. Therefore, prior to the

application of a frequently used American model to the South Korean market, it may be

important to examine how media directors of South Korean advertising agencies view the








current state of reach/frequency estimation methods and which communication effects are

used to predict or measure potential advertising impact.



Purpose of the Study


No information was available about how the South Korean advertising industry

perceives reach/frequency estimation methods and how these concepts are used by the

South Korean advertising agencies. Consequently, this study shows the South Korean

advertising industry's 1) vehicle/message rating sources, 2) perceptions of the difference

between vehicle and message, 3) definition of reach and effective reach, 4) lower and

upper limit of effective reach, 5) and the use of communication effects, among other

factors.

In this study, the survey results of South Korean media practitioners also will be

compared with previous American studies to inspect similarities and differences between

the two advertising industries. The American studies, which are partially available for

comparison, are Kreshel, Lancaster, and Toomey (1985), Leckenby and Kim (1993), and

Falcheck (1995).



Methods


Currently there are about 120 advertising agencies that can deal with both

broadcasting and print advertising in South Korea. In this study, a mail survey was

conducted of the media directors of these 120 advertising agencies.








The questionnaire used in this study was adapted from the one published in 1986

by Lancaster, Kreshel, and Harris. The questions were modified in accordance with the

South Korean situation, such as vehicle rating sources, and translated into Korean by the

author. Both Korean and English versions of the questionnaire are shown in Appendix A.

The questionnaire was mailed to the media director of each agency, if there was any, with

a personalized cover letter, return envelope, and a return postage stamp for international

mail. For the agencies which do not have media directors, a questionnaire was mailed to

senior executives or owners of the agencies.

A follow-up letter, questionnaire, and return envelope were mailed to all non-

respondents approximately seven weeks after the initial mailing. After two mailings, a

total of 61 responses were received. SPSS PC+ for Windows was used to analyze the

data.



Results


About half(61) of 120 South Korean advertising agencies responded to this study

and 43 of them answered that they develop advertising media plans.

For the titles of the respondents, 36 percent were media directors or vice directors,

3.3 percent were executives, and 32.8 percent were others, not directors or vice-directors

but individuals who work for either media planning or media purchases, and owners of the

agencies. Approximately 28 percent of respondents did not provide their titles.

For the first question regarding sources for television program ratings, 57.4

percent of the South Korean advertising agencies receive the data from Media Service

Korea (MSK). Other sources used in the South Korean advertising industry were Lee's








PR (23%), and Korean Marketing Research (KMR) (26.2%). Since some of the agencies

receive the rating data from more than one source, the sum of percentages for each source

exceeds 100 percent.

For the TV advertising message ratings, 53.5 percent of the South Korean

advertising agencies also use MSK data. Lee's PR data are used by 8.3 percent of

agencies and 24.6 percent of agencies do not receive advertising message ratings.

This study also shows that the majority (70.5%) of the South Korean advertising

agencies develop media plans for their advertising campaigns. Among the agencies that

develop media plans, the most frequently used systems are non-computerized programs

(39.5%) and in-house computer programs (32.6%) (see Table 9).


Table 9. Systems used to develop advertising media plans (N = 43)

System Used (%)


In-house computer program 32.6

Foreign computer program 4.7

Non-computerized program 39.5

Others or no answer 23.3


For the next item examining the factors which were generally used to evaluate

advertising media plans, most of the South Korean advertising agencies that develop

media plans answered that they use reach (79.1%) and GRPs (Gross Rating Points)

(72.1%). Other frequently used factors are average frequency (79.1%), CPM (67.4%),








and CPRP (72.1%). Two previous American studies (Kreshel, Lancaster, & Toomey,

1985; Leckenby & Kim, 1994) also revealed that reach, GRPs, average frequency, and

CPM are the most widely used factors to evaluate media plans. Table 10 illustrates the

factors used in evaluating media plans in the South Korean advertising industry in

comparison with two American studies.


Table 10. Factors used in evaluating media plans


Factors US 85 US 94 Korea 96
(N =91) (N= 51) (N=43)
(%) (%) (%)

Reach 90.4 81.0 79.1

Average Frequency 87.2 73.0 79.1

GRPs (Gross Rating Points) 89.4 73.0 72.1

CPRP (Cost-Per-Rating Point) N/A N/A 72.1

Effective Reach 86.2 68.3 67.4

CPM (Cost-Per-Thousand) 88.3 77.8 67.4

Gross Impressions N/A N/A 2.3

Frequency Distributions 75.5 74.6 N/A

Quintile Distribution 43.6 58.7 N/A
N/A indicates that the factors are not used or no answer.



The evident differences discovered between South Korea and the United States are

the use of CPRP and Frequency Distribution. In spite of little or no use by American

practitioners, the majority (72.1%) of South Korean media practitioners use CPRP to








evaluate media plans. On the contrary, frequency distributions are highly used by only

American media practitioners. In other words, the introduction of frequency distributions,

which can provide planners with more pertinent information and a method of determining

the pattern of repetition, can help the South Korean media directors to evaluate their

media plans more effectively.


Definition of Reach and Effective Reach

Although the majority (91.8%) of the South Korean advertising practitioners

answered that the distinction between advertising vehicle and message is important for the

evaluation of advertising campaigns, only 39.1 percent answered that the most

representative definition of 'reach' is message exposure. Nearly half of the South Korean

advertising practitioners (53.5%) defined 'reach' as vehicle exposure. In other words,

53.5 percent of the South Korean advertising agencies surveyed use vehicle audiences as

a proxy of advertising audiences, while nearly 40 percent of the agencies attempt to

estimate advertising message exposure.

The definition of effective reach applied in the media plan of a particular product

or service can vary across a number of factors, including the media categories and time

frames used. When multiple media categories are evaluated simultaneously, the same

definition of effective reach and time frame must be used for all media categories. This

study found that only half (46.5%) of the media plans developed by the South Korean

advertising agencies include evaluations of schedule reach, frequency, and GRPs for

combinations of two or more media categories.








The South Korean advertising practitioners were then asked to note the general

definition of effective reach. The majority of the South Korean advertising practitioners

recognize that the definition of effective reach requires a minimum frequency level. About

93 percent answered that a minimum frequency of two or more should be the lower limit

of effective reach and 74.4 percent answered it requires a frequency of three or more,

while 11.6 percent answered that a frequency of four or more is needed.

For the upper frequency limit, 41.9 percent of the respondents recognize no upper

frequency limit to effective reach, while 37.2 percent set some kind of upper limit, ranging

from 9 to 11. In spite of the different environmental situation, two American studies

(Kreshel, Lancaster, & Toomey, 1985; Falcheck, 1995) also show similar results to this

study. Table 11 shows the South Korean results in comparison with the previous

American studies.


Communication Effects Used

The majority (90.3%) of the South Korean advertising practitioners answered that

they use communication effects to evaluate their advertising media plans. The two most

frequently used variables are exposure to advertising and awareness/recognition. Among

those who use communication effects to evaluate media plans, 80 percent answered that

they use advertising exposure. The second most frequently used variable is awareness or

recognition (46.5%). Since there is no distinct difference between awareness and

recognition in the Korean language, these two terms are treated as the same expression.








Table 11. Representative definitions of effective reach in comparison with two previous
American studies


US 85 US 95 Korea 96
Message Impact Required us 85 Uo
Message Impact Required (N = 91) (N = 50) (N = 43)
(%J (%) (%)

Vehicle Exposure 48.4 70.0 53.5
Message Exposure 31.9 13.3 39.5
Advertising Impact 16.5 13.3 7.0
Others and no answer 3.3 3.3


Number of Exposures Required Regardless of Message Impact Level
Lower Limit US 85 US 95 Korea 96
(%) (%) (%)

1+ 6.6 13.3 4.7
2+ 5.5 6.7 18.6
3+ 61.5 50.0 62.8
4+ 17.6 16.7 11.6
Others and no answer 8.8 3.3 2.3

Upper Limit US 85 US 95 Korea 96
(%) (%) (%)

No Upper Limit 58.2 46.7 41.9
9 4.4 6.7 4.7
10 6.6 10.0 20.9
11 2.2 0.0 11.6
Others and no answer 28.6 13.3 21.0








Table 12 illustrates the variables used to evaluate advertising communication

effects in the South Korean advertising industry in comparison with a previous American

study (Kreshel, Lancaster, & Toomey, 1985). Though the American study was conducted

about ten years before this study, the two studies show similar results. The only

observable differences are that Korean agencies use attitude toward advertisement or

brand more frequently than American agencies, while American agencies use recall more

than Korean agencies.


Table 12. Variables used to evaluate advertising communication effects

US 85 Korea 96 US 85 Korea 96
Variable (N=94) (N=61) Variable (N=94) (N=61)
(%) (%) (%) (%)

Exposure to 52.1 79.1 Preference 18.1 32.6
advertisement

Awareness/ 47.9 46.5 Willingness to 12.8 32.6
Recognition 24.5 purchase

Attitude toward 11.7 44.2 Purchase 31.9 23.3
Ads.

Recall 58.5 37 2 Comprehension 16.0 20.9

Interest 16.0 34.9 Conviction 8.5 14.0

Attitude toward 18.1 32.6 Others 9.6 4.7
brand




Message Weighting and Reasons of Not Weighting

The media practitioners of the South Korean advertising agencies that develop

media plans and evaluate their plans (about 70 percent of all South Korean agencies) were








asked whether they use such information to weight vehicle audience data to estimate

advertising message exposures or other desired communication effects.

Among 43 agencies that develop media plans, 34.9 percent indicated that the data

are used to weight vehicle audiences. More than half (56.3%) of the agencies which

weight audience data use individual judgements to derive weights, while only 18.8%

derive weights from a formula or an established standard.

The respondents who do not weight vehicle audiences gave approximately six

different reasons for not doing so. The most common of these are that media planning

situation for each brand or service is unique and weighting is too difficult to use

accurately. Lack of information about weighting procedures and not enough time are two

other common answers.


Needed Improvements

Almost all respondents (96.7%) answered that their agency's current procedures

for evaluating advertising media plans are in need of improvement.

The majority (82%) of South Korean advertising practitioners indicated that

recognition of the importance of media planning should be improved. Nearly 80 percent

of them also answered that training of media specialists is the second most important thing

to improve current procedures to develop and evaluate advertising media plans (78.7%).

The South Korean advertising media practitioners showed split responses to the

introduction of foreign media planning procedures, such as American reach and frequency

computer programs, for better media planning. About 34.9 percent answered that the








foreign procedures would be helpful for better media planning, while 32.6 percent replied

that the foreign procedures would not be helpful due to the different media environment.



Discussion and Implication for Further Research


In the United States, models for estimating reach and frequency distributions have

made increasing use of the concept of effective reach and frequency among both

practitioners and academicians (Kreshel, Lancaster, & Toomey, 1985; Lancaster, Pelati, &

Cho, 1991; Leckenby, & Kim, 1994). The results of this study clearly demonstrate that

the majority of the South Korean advertising agencies are also aware of the media

evaluation terms such as effective frequency, and more than half of the agencies actually

use reach/frequency estimation methods.

In spite of the different environments, the South Korean media practitioners'

representative definition of effective reach is also close to the American practitioners as

shown in Table 11. The majority of both American and South Korean media practitioners

answered that the most representative definition of effective reach has 3+ as the lower

limit and no upper limit.

In addition to the definition of effective reach, the majority of both American and

South Korean media practitioners also use communication effects to evaluate advertising

campaign effects and media plans. Among the common measures for advertising

communication effects, advertising exposure is the most frequently used measure in both

countries. Awareness is the second most widely used measure in South Korea and third in

the United States.








With the beginning of CATV in South Korea since 1995, the South Korean

advertising industry started to recognize the importance of media planning and evaluation

of media schedules. According to this study, about 96.7 percent of the South Korean

advertising agencies claim that their current procedures for developing and evaluating

media plans should be improved. The South Korean media directors also indicate that

systematic media evaluation methods that fit the South Korean environment are needed in

the near future.

Although only 34.9 percent of the South Korean media practitioners responded

that the introduction of foreign procedures would be helpful for better media planning and

campaign evaluation, 32.6 percent replied that the procedures would not be helpful. The

application of a frequently used American reach and frequency computer program to the

South Korean market will help develop a model that fits the advertising environment of

South Korea.

This study indicates that reach/frequency estimation is a good method to predict or

evaluate advertising communication effects in the South Korean advertising industry

because of the wide acceptance and actual use of reach and frequency as shown in Table

10. Furthermore, since the South Korean media practitioners also define the general

effective frequency level similar to American media directors as shown in Table 11, it is

assumed that in South Korea the effective frequency levels for common communication

measures, such as awareness, preference, and purchase intention, can be decided using the

procedures explained in Chapter II. Based on the optimal exposures for common

communication measures, which are used as bases to estimate the advertising





63


communication effects, the effective frequency level can be decided for each advertising

campaign goal with consideration of marketing, copy, and media factors.

A test market study is a good method for examining these assumptions.

Therefore, Chapter IV, Methodology, will introduce the brand used as a case study, along

with the normative theories, methods, and procedures required to evaluate it.














CHAPTER IV
METHODOLOGY





The primary purpose of this study is to demonstrate how accurately media

exposure distribution models can predict advertising communication effects. In particular,

one goal is to apply such a model to the South Korean market. In this test market study,

advertising campaign communication effects are "predicted" by a variation of the beta

binomial exposure distribution model based on the information available before the

advertising execution. Then, the impact of the advertising campaign on its target audience

is measured after a three-month execution. The predicted communication effects will be

compared with the actual target consumer responses to the advertising campaign to probe

the capability of this normative framework.

Generally, as shown in Table 13, regardless of the country or brand being studied,

particular items need to be gathered or defined to complete this type of study. These

items are 1) target audience definition, 2) vehicle data, 3) message data, 4) advertising

expenditures, 5) a media evaluation model, 6) a narrow time frame, 7) tracking study

results, 8) and the media plan used.

Advertising planners who wish to accomplish this type of research should follow

these procedures. After defining the target audience, advertising message and vehicle

ratings and expenditures should be gathered prior to the period of the campaign in order








to predict the campaign effects. It is also important to evaluate the plans over a narrow

time frame, such as monthly, due to target audience forgetting, competitive advertising,

and other factors that reduce advertising carry-over effects. For example, annual reach-

frequency analyses will grossly overstate likely communication effects because it fails to

account for these factors (Lancaster, Kreshel, & Harris, 1986).


Table 13. Normative framework necessary to predict advertising communication effects

Items General Meaning


Target Audience The desired or intended audience for an advertising campaign.
Definition Usually defined in terms of specific demographic, purchase, or
ownership characteristics.

Vehicle Data Ratings of particular component of a media class, such as a
particular TV program or magazine, used for an advertising
campaign.

Message Data Ratings for actual advertisements, not the vehicles.

Advertising Total media expenses for an advertising campaign.
Expenditures

Media Evaluation Computer program that embodies normative theories of media
Model planning, including vehicle and message ratings and duplication
plus corresponding exposure distributions.

Narrow Time Due to target audience forgetting, competitive advertising, and
Frame other factors that reduce advertising carry-over effects, a narrow
time frame for analysis, such as monthly, is preferred.

Tracking Study Actual target audience response to the advertising campaign in
Results terms of the advertising goals established.

Media Plan Used Advertising media schedules during the period of the campaign.








A media evaluation model such as the beta binomial exposure distribution model

can be used to predict the campaign communication effects. Then, consumer tracking

studies will show actual consumer response to the campaign.

In this chapter, after introducing the brand and advertising campaign used in this

case study, each item in Table 13 will be explained again with reference to the information

available for the case.



Case Study




ASIANA Airline


A South Korean airline company is selected to test the framework of predicting

and measuring advertising campaign effects introduced above. The South Korean

domestic and international air transportation business has grown rapidly (about a 12

percent annual increase in passenger transportation volume) since Korea hosted the 1988

Olympics. Seoul's growing importance in international air traffic, good international

geographic location, congested traffic on the highways, and drastic increase in GNP (see

Table 6 in Chapter III) since 1988 accelerated the growth of the airline business in South

Korea.

Currently, there are two airline companies in South Korea. The Korean airline

(KAL) had been the only air carrier in South Korea until ASIANA started its business in

1990. In 1993, both KAL and ASIANA carried 15,655,000 passengers on their domestic

lines and 11,651,000 on international flights. Cargo services also have grown impressively








since the opening of American routes in 1971. The number of employees of ASIANA

airline is about 7,900.

Domestically, KAL has maintained a 70 percent market share since 1990, while

ASIANA has held only 30 percent market share for the last 7 years. Internationally, the

market share of both Korean air carriers was only about 50 percent until 1991 because of

dynamic foreign airline companies, such as Delta Airlines, TWA, United Airlines, British

Airways, and Singapore Airline. Many foreign airlines have been attracted to Seoul's

growing importance and have opened regular services to and from Korea. As of 1993,

about 26 foreign airlines are operating in Korea. However, the international market share

of both Korean air carriers has been increasing since 1991. While the KAL holds its

international market share at about 47 percent, ASIANA airline has been growing fast for

the last 5 years. The domestic and international market shares for both air carriers are

shown in Table 14.

Current consumer research conducted by SangAm advertising agency shows that

the ASIANA airline company has maintained a good company image on its service

evaluation because of perceived flight attendant service and kindness. However, many

consumers still favor KAL because of safety, tradition, and mileage bonuses. It is

assumed that promotional activities such as advertising and public relations play a major

role in the recent successful growth of ASIANA. Since advertising is an important means

of promotion for ASIANA airline, it is necessary to conduct consumer research to

examine and predict its current advertising effects and to help develop a more efficient

advertising campaign in the future. This case study also provides an opportunity to test

and advance normative advertising theory.








Table 14. Domestic and international market shares

Domestic market share


1990 1991 1992 1993 1994 1995 1996
(%) (%) (%) (%) (%) (%) (%)

Korean Airline 70 71 71 70 69 69 69

ASIANA Airline 30 29 29 30 31 31 31

Total 100 100 100 100 100 100 100



International market share


1989 1990 1991 1992 1993 1994 1995 1996
(%) (%) (%) (%) (%) (%) (%) (%)

Korean 52.3 45.3 44.8 43.8 47.3 47.4 47.1 46.9
Airline

ASIANA 2.0 6.0 8.9 12.1 15.3 17.4 17.8
Airline

Total 52.3 47.3 50.8 52.7 59.4 62.7 64.5 64.7

Source: adaptedfrom "SangAm advertising agency.


Target Audience and Advertising Campaign


The annual expense for ASIANA advertising was about US $5,920,000 in 1994

and ASIANA is one of the 150 biggest advertisers in South Korea (ranked 101 in 1994 in

'Kwango Yeonkam 95').








In 1996, ASIANA airline completed its new advertising campaign, "Her name is

ASIANA," focusing on ASIANA's better service and kindness. Since the new campaign

was performed by SangAm, one of the major Korean advertising agencies, most of the

information needed for this study, such as consumer survey data, were supplied by

SangAm.

According to the research conducted by SangAm (1996), the primary target

market of this campaign was defined as 20 to 40 year old men who have used any air

carrier at least once. However, there are some difficulties in using this target market

definition. First, previous use of any air carrier makes it difficult to estimate the described

target market size. Second, the geographic area of consumer research conducted by

SangAm was limited to Seoul, the capital city of Korea, where one fourth of South

Korean people live. Consequently, although the primary target audience is defined as 20

to 40 year old men who have used any air carrier at least once, the consumer survey was

conducted with 20 to 40 year old Korean men who live in Seoul. The target market size

used for the consumer survey is about 3 million (2,884,983 in 1995).

ASIANA airline first advertised its new campaign of "Her name is ASIANA" from

February 17 to May 16 in 1996. Two different types of messages were advertised on

television and in magazines. One was called "piano scene" and the other was called

"poetry situation." Both types of the actual magazine advertisements are in Appendix B.

The summary of the advertising insertions by media category and expenses of the

campaign is shown in Table 15.








Table 15. Advertising insertions by media category and expenses of "Her Name is
ASIANA" campaign from February 17 to May 16, 1996

Media Insertions Total Expenses
(US $)

KBS-2TV 96 $414,373
MBC-TV 137 477,035
SBS-TV 105 259,441


Weekly Magazines 33 60,750
Monthly Magazines 9 19,875


Grand Total 380 $1,231,474
Source: adaptedfrom marketing department of SangAm and MSK.
(S $1 = 800 Won)


Competitor


The market leader, KAL, which started its business in 1969, is a strong competitor.

The first name of KAL was Korean National Airlines and was managed by the

government. Later the name was changed to KAL and the management was turned over

to private hands. Because of the long tradition and previous monopoly, KAL has

maintained its market share at about 70 percent domestically and 47 percent

internationally.

However, compared with ASIANA's rapid growth, the market share growth rate

of KAL has been almost zero for the last four years (see Table 14). KAL's annual

advertising expense was US $11,932,000 in 1994 and KAL has been one of the 50 biggest

advertisers in South Korea (ranked 39 in 1994) (Kwango Yeonkam 95). KAL's








advertising expenses were approximately three times higher than ASIANA during the

period of ASIANA'S first "Her name is ASIANA" campaign. Overall advertising GRPs

and expenses from February through April of 1996 are provided in Table 16.


Table 16. Competitor GRPs and advertising expenses

Media Month GRPs Expenses (US S)


TV February 969.58 $499,130
March 1158.18 549,560
April 1072.82 524,167
Total 3200.58 1,572,857

Radio February 272.89 90,395
March 273.54 97,407
April 322.18 93,980
Total 868.61 281,782

Newspaper February 742.62 889,327
March 236.86 309,484
April 261.58 267,725
Total 1241.06 1,466,536

Magazine February 89.73 63,625
March 82.10 46,813
April 45.63 35,500
Total 217.46 145,938

Total February 2074.82 1,542,477
March 1750.68 1,003,264
April 1702.21 921,372

Grand Total 5527.71 $3,467,113
Source: adaptedfrom MSK (Media Service Korea).
(US$1 800 Won).








Media Data



TV Ratings

Television program ratings are collected from Media Service Korea Inc. (MSK)

which is one of the major media research institutes for ratings in South Korea. SangAm,

the advertising agency for ASIANA airline, provides advertising message ratings that are

also collected from MSK. Media Service Korea Inc., founded in 1987, is the only media

research institute in South Korea that uses people-meters to collect its ratings from a total

of 1,200 panel members who live in Seoul. Audience panels were established by MSK and

the panel members were randomly selected based on four critical demographic variables,

such as age, occupation, education, and income. Each panel member's daily television

watching data are gathered in the MSK central computer from 6 A.M to 1 A.M and all the

television program and advertising message ratings are reported daily after aggregating

them.

The ratings are also provided by important variables, such as gender and age. It is

possible to gather demographic information with the use of separate remote controls

distributed by MSK. Each panel member is assigned to a number and is supposed to input

the assigned number while he/she is watching television. For example, if a panel

household consists of four family members, such as a father, mother, son and daughter,

each family member is assigned to a number from 1 to 4 and is instructed to push the

number button on the remote control. Therefore, in this study, program and advertising

message ratings were gathered for the target audience group. According to MSK, these

ratings are significant with 95 percent confidence levels (p < 0.05). The average sampling








error is within + 0.025. Television program ratings, advertising message ratings, and costs

provided by the 'SangAm' agency are listed in Appendix C.

Television program and advertising message ratings are also collected from Lee's

PR which is another popular research institute for ratings in South Korea. These are often

used in comparison to MSK. While MSK uses panel members and people-meters, Lee's

PR conducts face-to-face interviews once a month to gather television ratings from a total

of 1,300 sample members. However, since MSK data have been used by SangAm, MSK

data are used in this study unless otherwise noted. According to the South Korean media

practitioner survey conducted by the author and reported in Chapter III, MSK is also the

most frequently used vehicle and message source in South Korea.


Magazine Ratings

Magazine advertising expenses of $80,625 were 6.5 percent of total media

expenses of $1,231,474 for the ASIANA campaign. Eleven weekly magazines and three

monthly magazines were used for the ASIANA advertising campaign. The circulation and

ratings for these magazines were collected from each magazine company because of the

lack of independent information about magazine ratings in South Korea. A Korean

advertising practitioner called each magazine company and collected the ratings of these

14 magazines. The ratings may be higher than actual ratings because they were gathered

from each magazine publication company. In the future, the magazine ratings should be

gathered from an objective media research institute for better prediction of advertising

campaign effects. Magazine circulation, ratings, and costs are listed in Appendix C.








Tracking Data


The tracking data used in this study were gathered from the advertising agency of

ASIANA Airline, SangAm. The main purpose of the survey research was to evaluate the

impact of the new ASIANA advertising campaign, "Her name is ASIANA" and to prepare

the next ASIANA advertising campaign After the three month advertising campaign from

February 17 to May 16, a post test with a total of 200 samples of the target audience was

conducted in May, 1996.


Sampling

Since the target audience was defined as 20 to 40 year old men who live in Seoul,

Korea, the sample was randomly drawn among the target audience. In this survey

research, "stratified sampling" was used, which separates samples from each of several

subpopulations. The subpopulations were defined by SangAm in advance on the basis of a

critical subject variable, such as age, that may influence scores on the dependent measures.

Rather than relying on random sampling, the target population was divided into

subpopulations on the basis of age to create a total sample by selecting the appropriate

proportion of subjects from each of the subpopulations. For example, if 20 percent of the

target population is between ages 20 and 29, the number of subjects that would represent

20 percent of the total sample was selected from that subpopulation. A total sample of

200 individuals was drawn from three different age groups. Forty percent of the total

sample was selected from the 30s age subpopulation, and another forty percent was drawn

from the 40s age subgroup. The age groups, number of sample individuals in each age

group, and sample characteristics are shown in Table 17.








Table 17. Sample characteristics

Number in (%)
Sample
Age 20-29 40 20.0
30 39 80 40.0
40 49 80 40.0
Total 200 100.0

Education Up to High School Education 43 18.5
College Education or More 157 81.5
Total 200 100.0

Marital Status Single 37 21.5
Married 163 78.5
Total 200 100.0

Occupation Specialist or Freelancer 19 9.5
Management Staff 42 21.0
Office Worker 110 55.0
Self-employed 29 14.5
Total 200 100.0
Source: adaptedfrom the marketing department of SangAm.


Measurement

Since the purpose of this survey was to evaluate the new ASIANA advertising

campaign effects, only a post test was conducted. In this survey, a face-to-face interview

was used to gather target audience information about the advertiser image, advertising

campaign message awareness, evaluation of the advertising message quality, preference,

and willingness to use in comparison with competitors. A total of 18 interviewers, who

were college students or housewives, were temporarily hired by the advertising agency of

ASIANA Airline to conduct the personal interviews for seven days beginning on May 25,

1996. Three questions that were asked of the respondents and which are directly related

to this study are: 1) "Are you aware of the new ASIANA advertisement, 'Her name is








ASIANA'?" 2) "Which airline company do you prefer?" 3) "Which airline are you going

to use for your next trip?" After data were completely collected, 20 percent of randomly

selected responses were verified again before editing, coding, and analyzing the data.

Then, the SPSS PC+ program was used to show frequencies and cross-tabulations. These

tracking data will be used to examine how accurately the normative media evaluation

framework of this study can predict advertising communication effects.



Media Evaluation Model Used in This Study



Beta Binomial Exposure Distribution Model

A variety of computerized media evaluation models have been developed and used

in both the advertising industry and academia over the past thirty years. The availability of

computers in the 1960s led to the development of a number of mathematical media

planning models. The first level of sophistication in the quantitative evaluation of a media

plan involves estimating a media schedule's reach and frequency. Another related concept

is gross rating points (GRPs), which refer to the total number of exposures generated by

the schedule. In 1961, Agostini developed a simple reach estimation model and in 1964,

Metheringham tried to estimate unduplicated audience measurement (Craig & Ghosh,

1986). They provided the impetus for the development of media models. After that,

media evaluation models have become much more sophisticated, including the beta

binomial models, beta matrix method, heuristic models, and variations or extensions of the

beta binomial models (Rust, 1986).








Most recent models deal with the sophisticated calculation of the exposure

distribution by the number of exposures of a given media schedule and they provide all

information relevant to the advertising media schedule. The general goal of these models

is to evaluate advertising media schedules by estimating reach, frequency, effective

frequency, effective reach, and exposure distributions.

Exposure distribution models are divided into two groups in terms of estimation

dimensions that include a number of parameters, such as univariate probability distribution

models and multivariate probability distribution models. Univariate probability distribution

models treat all vehicles as one univariate and composite vehicle and then average all

vehicle insertions of a schedule (Ju et al., 1990). On the other hand, multivariate exposure

distribution models provide individual vehicle audience information by preserving as many

estimation parameters as vehicles in the schedule. Though the complexity means that

multivariate models require more data manipulation and computation time, they are

considered to provide more accurate estimation than do univariate models. However,

since true multivariate probability distribution methods that estimate broadcast audiences

have not been developed, most multivariate models are magazine audience estimation

models that are tested in schedules with no more than two insertions in each vehicle

(Leckenby & Kim, 1992).

More than fifteen univariate probability distribution models have been developed

since Agostini's simple reach estimation model. Some include the beta binomial

distribution (BBD), Markov binomial, Morgenzstern, and Hofians geometric distribution

(HGD) (Leckenby & Ju, 1990). Among them, one of the most widely used and studied

exposure distribution models is the beta binomial distribution (BBD) model because of the








simple, parsimonious, and relatively sound estimation (Ju & Leckenby, 1990; Kreshel,

Lancaster, & Toomey, 1985).

The BBD model, which was first introduced by Metheringham in 1964, is a

statistical procedure for estimating the likelihood of a target audience exposure to a

schedule a given number of times. The binomial distribution estimates the number of

different ways a target audience can be exposed to a schedule from one to N total

insertions. The beta distribution estimates the average probability of exposure to the

schedule from one to N times (Lancaster & Katz, 1988).

The BBD model has been used as a basis of estimating the exposure distribution of

not only univariate but also multivariate models. The notable univariate variations of the

beta binomial distribution are BBD-DE, Hofmans BBD, and BBD simulation. The

multivariate analog of the BBD is the Dirichlet distribution.

Although the BBD has been accepted to be fairly accurate across a wide variety of

advertising media circumstances, the use of the BBD to evaluate an entire media category

schedule has some limitations. One of the major problems is that the use of BBD treats all

vehicles and pairs of vehicles in a schedule as a single average vehicle. Therefore, using

the BBD alone can mislead the estimation of reach and exposure distributions, particularly

in the case of disparate insertions and/or ratings among the vehicles in a schedule

(Lancaster 1992). It is also known that the BBD model often yields an over-estimation in

reach (Leckenby & Rice, 1985; Leckenby & Kim, 1992). However, the BBD model has

been used as a mathematical basis to predict advertising message exposure distributions of

advertising schedules because of the wide acceptance, proven reliability, and possible

variations over more than two decades (Leckenby & Boyd, 1984; Ju et al., 1990). For








example, due to the lack of pair-wise duplication data such as vehicle self-pair and cross-

pair ratings, many studies use a variation of the BBD model which is the beta binomial

distribution with limited information (BBD-L). The BBD-L model, developed by

Lancaster and Martin (1988), does not require pair-wise duplication data and was shown

to be accurate in predicting pair-wise duplication among vehicles.


Model Selection

Based on the BBD model, a computerized media evaluation program, ADplus, is

used in this study because the program is comprehensive, widely available, and also can

handle vehicle and message reach and frequency.

ADplus, originally developed by Lancaster in 1990, is an easy-to-use program that

helps predict advertising campaign effects and evaluate advertising media plans. The

program is currently distributed worldwide by Telmar Information Service Corp. of New

York (http://www.telmar.com) and is used by thousands of media marketing professionals

worldwide. It is possible to evaluate all major advertising media categories and

subcategories with ADplus. It also helps advertising media planners find optimum

schedules involving multiple media categories.

As input data, ADplus needs advertising media vehicle ratings, target market size,

advertising cost, and number of insertions. As shown in Figure 4, ADplus results provide

media planners with frequency distributions, reach, effective reach, gross rating points

(GRPs), average frequency, gross impressions (GIs), cost-per-thousand (CPM), and cost-

per-rating point (CPP). ADplus also provides two types of frequency distributions,

including vehicle and message frequency distributions. For the magazine example in








Figure 4, the message effective reach (3+) is 0.1 percent. Therefore, if it is assumed that

consumers are aware of the new advertising message after three exposures to the

magazine advertisement (effective frequency), only 0.1 percent of the target audience is

estimated to have been exposed at this level.

Although the typical vehicle and message effective reach of 3+ is shown in the

ADplus results, media planners also can examine other definitions of effective frequency

and reach that may be more fitting with their situation including, for instance, effective

frequency and reach of 4+ or 5 through 10.

ADplus can overcome the problems of the BBD model which treats all vehicles

and pairs of vehicles in a schedule as a single average vehicle. With the estimation of

vehicle self-pair reach, ADplus can estimate separate vehicle exposure distributions. A

matrix of individual vehicle exposure distributions is then used to estimate multiple vehicle

exposure distributions in conjunction with vehicle cross-pair reach data. This procedure is

referred to as a beta binomial matrix exposure distribution (BBMD) model (Lancaster,

1993). The technical procedures used to implement the BBMD model are described in

detail by Lancaster (1993) and important dimensions of this model are presented in

Appendix D.

The ADplus program also helps media planners find an optimum combination of

media vehicles and media categories within desired budgets and recommends media

budgets for desired communication goals.











ADplus(TM) RESULTS: MAGAZINES

By: Hyunsoo Park
University of Florida --
For: ASIANA Airline

Typical monthly schedule
f
Target: 2,884,983
Korean men 20-49 years 0
old who live in Seoul 1


Message/vehicle


45.0%


Frequency (f) Distributions


Vehicle

% f % f+

65.1 100.0
28.4 34.9
5.8 6.5
0.7 0.8
0.1 0.1
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0


Message

% f % f+

82.6
15.9 17.4
1.4
0.1
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0
0.0 0.0


Summary Evaluation

Reach (1+) 34.9% 17 %
Effective reach (3+) 0.8% 0.1%)
Gross rating points (GRPs) 42.2 190
Average frequency (f) 1.2 1.1
Gross impressions (000s) 1,266.0 569.7
Cost-per-thousand (CPM) 21.23 47.17
Cost-per-rating point (CPP) 637 1,415

Vehicle List Rating Ad Cost CPM-MSG Ads Total Cost Mix


Sisa Journal
Han Kyurei 21
Jugan Chosun
News Plus
Jugan Hankuk
News Week
News people
News Maker
Jugan MaeKyun
Economist
HanKyung Busi
Shin Dong-Ah
WalKan Chosun
Win


2,250
2,000
2,250
1,875
1,500
2,250
1,500
1,500
1,750
1,750
1,625
1,875
1,875
2,875


25.25
33.67
30.86
49.60
69.44
59.52
55.56
65.36
81.02
129.63
240.74
26.21
26.21
177.47


2,250
2,000
2,250
1,875
1,500
2,250
1,500
1,500
1,750
1,750
1,625
1,875
1,875
2,875


8.4%
7.4
8.4
7.0
5.6
8.4
5.6
5.6
6.5
6.5
6.0
7.0
7.0
10.7


Totals: 47.17


26,875 100.0%


Figure 4. ADplus Results








SMdel Modification

As shown in Appendix C, the media evaluation model used in this study needs

regression equations for the estimation of R,, and R,,. R2, is reach as a percentage of the

target audience of one typical insertion in vehicle i (self-pair or two-use reach), while

R,, is cross-pair ratings in vehicle i and. The estimated regression equations of R2, and

R, for the South Korean target audience may be different from US equations. In 1992,

Korean Gallup reported the regression equations for the estimation of R2, (n = 528) and

R, (n = 560,292) with South Korean target audience groups (see table 18).

Table 18. R, and R, estimation equations for network television

US Equations (for overall audience)

R, Estimation R, = -0.0065 + 1.795 R, 0.331 (R,)2

R,, Estimation R, = -0.0015 + 1.0095 (R,+ R,)- 1.5R, R,

Korean Equations (for 20 49 year old men)

R, Estimation R2,= 0.000525 + 1.848325R,- 1.432(R,)2

R, Estimation R, = 0.00005 + 1.003 (R,+ R,)- 1.87755R, R,
American source: adapted from Kent Lancaster (1993).
South Korean source: adaptedfrom Korean Gallup (1992).



Since the Korean Gallup reported the estimated regression equations with gender

and age variables, in this study, the equations for only the target audience group of

ASIANA campaign are estimated. Then, the media evaluation model will be modified

with the regression equations calculated from South Korean target audience which is 20 -








49 year old men. The estimated American and South Korean regression equations are

shown in Table 18.



Error Estimation for the Media Evaluation Model


The BBD model has been widely accepted in both academia and industry because

of its simplicity, aggregating power, and good empirical fits. This approach has been

improved by the availability of household-level exposure records obtained from people-

meters and the increasing computation power of computers. However, this approach is

also assumed to exhibit error in its estimations. It is necessary to estimate how much error

the model is likely to produce.

Tests of the performance of media evaluation models have been undertaken in the

US. These studies were conducted to compare the performance of the various media

evaluation models. Recently, studies tried to compare the accuracy of media evaluation

models with different error definitions. Frequently used error definitions are average error

in reach (AER), average error in levels of exposure (ALE), and average percentage error

in exposure distribution (APE). The error terms are defined as the difference between

observed (tabulated) and estimated values in a schedule. For example, AER is the sum of

the absolute differences between observed and estimated reach in each schedule divided by

the total number of schedules. Sometimes, each error term produced different results that

made it difficult to decide which model's performance is superior to another. However,

AER, APE, and ALE have been frequently used in these studies.

Leckenby et al. (1990) compared four different models' performance in terms of

error estimation with about 500 magazine schedules tabulated from SMRB (Simmons








Market Research Bureau). This study showed that the BBD produces the smallest

percentage of error for reach estimation among the models used in the study. The study

also revealed that the larger the schedule size is, the more accurate the estimations made

with the models.

The beta binomial distribution (BBD) is an univariate probability distribution

model which treats all vehicles as one univariate and 'composite' vehicles and then

averages all vehicle insertions of a schedule. Therefore, the model often produces over-

estimations in reach. Because of the problem with univariate models, the development of

multivariate models has been attempted. However, because of the excessive

computational difficulties involved in using all of the vehicle numbers together, complete

multivariate models have not been developed. Ideally, multivariate media evaluation

models should be developed because they may allow the most realistic view of complex

media evaluation processes without simplifying or averaging assumptions which are

currently used with univariate models (Leckenby et al., 1993).

An alternative is "sequential aggregation." Although sequential aggregation is not

a true multivariate method, studies showed that the models using sequential aggregation

are better for estimating reach (n+) than those of univariate models. With sequential

aggregation, two-vehicle bivariate distributions are collapsed to form a new composite

vehicle which is then combined with a third vehicle and this procedure is continued until

all vehicles are aggregated into the distribution of the schedule. The BBMD model used

in this study uses sequential aggregation.

Leckenby et al. (1993) tried to test the performance of several media evaluation

models including model using sequential aggregation. In this study, about 480 magazine








schedules were tabulated from SMRB. The average error in reach (AER) with BBD

model was about 6.4 percent, while the AER with the model using sequential aggregation

based on the BBD model [Morgenszten Sequential Aggregation Distribution model

(MSAD)] was 3.1 percent. The error estimation of the model used in this study (BBMD)

is similar to MSAD because both models are based on the BBD and use sequential

aggregation.

According to Leckenby et al. (1993), the model which uses sequential aggregation

is also the most accurate in predicting each level of the exposure distribution. The average

total error percentage in the exposure distribution (APE) was about 15 percent with

MSAD, while that of the BBD was 33 percent. Therefore, it can be assumed that about 3

percent error is likely to be made for reach estimation and about 15 percent error is

expected for the total exposure distribution for models which use sequential aggregation

based on the BBD. According to a selected example of observed and estimated exposure

distributions with their large schedule sample in the study, the error estimation for reach

(3+) was about 1.9 percent which is smaller than the error estimation of reach (1+) of 3

percent.

However, these error estimations also have limits. First, all of the studies

mentioned above used vehicle data instead of advertising message data. Since the model

used in this study will predict reach/frequency estimation with advertising message data,

the error estimation with message data would be smaller than that of vehicle data.

Another important point to be considered is that all the studies mentioned above used only

magazine advertisements. No study has been conducted for the error estimation of media

evaluation models for TV advertising. Since the ASIANA campaign mostly used








television advertising (about 93.5 percent of total campaign expenses), the error

estimation for television advertising is suggested for a future study. Additionally, this type

of study should be conducted with the actual BBMD model for better error estimation and

it is also recommended that the error estimation studies should be conducted with South

Korean data because the error estimation with US data may be different from South

Korean error.

It also should be considered that all of the observed (tabulated) data of previous

studies are based on the sample estimates. Although the possible error can be very small

with proper sampling methods, all samples must have sampling errors. The studies

mentioned above seem to have ignore the possible sampling errors for observed

(tabulated) data to examine error estimation of media evaluation models. The application

of sampling errors should be considered for better error estimation of media evaluation

models.

The model used in this study will be examined to show how accurately the model

can predict the reach/frequency estimation. If a critical amount of error is involved with

the model estimation of reach (n+) in this study, the study results will not demonstrate

significant (accurate) predictions. Therefore, the error estimation may not be included to

test hypotheses in this study. Whatever amount of error occurs from the model used in

this study, the error should be a part of the model performance.

In addition to the error estimation of the model used in this study, additional errors

are likely to be produced because of the unknown factors in South Korea, such as carry-

over rate and magazine ratings. However, the normative framework combined with a




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EATWDW5SS_ZOA7SJ INGEST_TIME 2013-10-10T03:08:24Z PACKAGE AA00017634_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES