how to design a marketing project

7/30/2019 How to design a Marketing Project

1/41

Marketing research provides information that should help marketers to better

solve marketing problems. Therefore, it is important to define the marketing

problem precisely so that the information obtained from the marketing research

relates specifically to the key problem and provides information to help solve

that problem. This seems obvious. Yet, in practice, it is not unusual to find

research studies where it is very difficult to identify the key marketing prob-

lem or opportunity. When this happens, the information collected tends to be

unreliable or irrelevant for decision making purposes. Sometimes it appears to

the client that the problem is obvious, but the problem might seem poorly

defined to the consultant whose experience with the company and the industry

might be too brief to see the essence of the problem clearly. It is very impor-

tant for each party in the process to communicate openly so that less-informed

colleagues can clearly understand all project components.

Inadequate problem definition can result in the collection of information

that is not directly relevant to marketing decision making for the problem that

initiated the research. Unproductive research is typically due to the problem not

having been defined properly, the sample not having been chosen correctly, the

questionnaire not having been interpreted precisely, and conclusions not having

been made relevant to the problem at hand. In other words, if anything during

the project is not done correctly, the value of the end deliverable might be

severely compromised.

After completing this chapter, you

should be able to:

1. Understand why marketing researchproblems may not be defined pro-perly.

2. Understand the different types ofmarketing research problems.

3. Define the marketing problems andthe marketing research problems.

4. Distinguish between manifest andlatent variables.

5. Understand the differences betweenthe scope and the objectives of thestudy.

6. Understand research designs.

7. Identify exploratory, descriptive, andcausal information requirements.

8. Understand the difference betweenqualitative and quantitative research.

9. Know why the design of fieldmethodology matters.

10. Understand how to question a surveyresearch study.

How to Design a

Marketing Research Project

When I was a young man, I observed that nine out of ten things I did

were failures. I didnt want to be a failure, so I did ten times more work.

George Bernard Shaw

C H A P T E R 2

Marketing managers look in tactical and

strategic market management for answers to

the problems they face. When a marketing

manager defines a marketing problem, itwill typically include a need for marketing

and market information that can be provided

by marketing research.


2/41

Why Is Research Design Important?DEFINING THE PROBLEM

Research design is important because the marketing research is commissioned to pro-

vide a defensible basis on which to hinge marketing decisions. There is no other sound

reason for a company to pay for marketing research than to make better decisions that

will lead to an improved market and financial position for the organization.

There is an old saying that if you dont know where youre going, any destination is

just fine. While there is nothing wrong with just wandering around on your holidays,marketing research should produce very specific results that are usable for improving

the future of the organization. If you conduct a survey about the satisfaction of Toronto

Hydro customers with the companys service, the findings should fairly accurately pro-

vide that information if the study is well designed and well executed. However, if those

who designed the project did not focus on the core problem, the findings could be very

misleading.

If a company wants to increase its penetration of a specific market segment, then a

customer satisfaction study will, at best, provide a reading on current customers.

However, it may say nothing about prospective customers and might not deliver the

most important information on which a market penetration strategy can be built.

Sometimes marketing decisions are scheduled to be made at specific times during the

companys planning year. In other cases, a marketing decision is needed as soon as pos-

sible. This might mean immediately and then there would no marketing research that

could be produced or consulted. If there are even a few hours available prior to the deci-

sion time, at least sources of secondary research can be consulted. For example, even

research that was conducted two years ago might contain enough information to sup-

port a decision. Alternatively, perhaps all that is needed is some information about the

physical market. This information can either be purchased from Statistics Canada or

obtained for free from Statistics Canada reports that are published on the internet or

archived in selected public libraries and universities.

One of the first steps in a mar-

keting research project is to

design the overall project as

well as each step. Researchdesign involves specifying theactivities at each step of the

project so that the objectivesare met within the project

scope.

A market segmentation seeksto divide the brands market or

category market into groups.

Each of the market groups or

segments should be different

from other segments on some

key dimension(s) that help bet-

ter understand customers and

more effectively serve their

needs.

Customer satisfaction researchfocuses on measuring the satis-

faction level of brand cus-

tomers and understanding

which aspects of the brand and

the brand-buying experience

significantly affect that satis-

faction.

30 | CHAPTER 2

EXHIBIT 2-1 Steps to Failure of Marketing Research

Problem not

understood

Problem not

defined

properly

Incorrect

sample

chosen

Fieldworkdesign

improper

Findings

poorly

communicated

Questionnaire

design

flawed

Data analysis

misdirected

Problem

not solved


3/41

What Was I Thinking?It is very easy to get in over your head in many endeavours. People find themselves in

situations where they have some knowledge but not enough to manage their lives at 100

percent. Sometimes, project directors or marketing research staff think they are doing

the correct thing by making a statement to a client. This might happen when a market-

ing manager in a client organization provides partial information to the project direc-tor in the marketing research consulting firm, and the project director tries to satisfy the

client by stating something without adequate preparation or consideration. These

unfounded statements and promises can return to haunt those who made them.

For example, there have been occasions when a project director has listened to the

clients explanation of the marketing problem and then identified the solution as one of

customer satisfaction. However, if the client had actually wanted to segment their cus-

tomers based on customer loyalty, effective market segmentation could not be complet-

ed because the correct questions had not been asked of the respondents.

Why Problems Are Not Defined ProperlyThe primary reason for poorly defined marketing problems is lack of experience and

lack of discipline. It is necessary to deeply scrutinize the marketing situation within the

organization and to think clearly about the decision that will need to be made and the

information required for making that decision. The most important question to ask

when attempting to define the problem is, What decisions will be based on this

research? The range of answers can be extremely broad; consider the following:

I dont know. I just need to know more about my customers.

Maybe none. Well just stay the course unless something jumps out.

I want to segment the market based on key drivers of customer retention.

Most difficulties with regard to problem definition arise mainly from three basicfactors:

failing to distinguish between the marketing objectives and the marketing

research objectives; trying to identifylatent behaviour patterns by direct questioning; confusing the objectives with the scope of the study.

FAILURE TO DISTINGUISH MARKETING AND MARKETINGRESEARCH OBJECTIVES

Marketing decision makers think in terms of marketing opportunities to be exploited,

problems to be solved, and objectives to be achieved. Some of the challenges that they

face are:

How can I outsmart the competition? Why is the market share of my product going down? What can I do to increase the sales of my product?

After listening to such marketing objectives, a researcher may be tempted to define

the marketing research objectives in similar terms:

To assess the factors contributing to gaining a competitive edge. To find out why the market share of Product Xis going down. To find out how the sales of Product Xcan be increased.

Marketing objectives arethose key challenges of the

marketing problem that must

be achieved according to the

marketing team in order to

improve marketing of the

product.

Marketing research objectivesdefine the challenges for mar-

keting information that must

be achieved by the marketing

research project. The marketing

research objectives are a subset

of the marketing objectives.

Latent behaviour patterns arethose customer activities that

are not directly observable and

are important for better under-

standing how to serve customer

needs.

The scope of the study helpsto define the parameters of the

marketing research project per-

taining to its breadth and

depth. These can include the

definition of the relevant popu-

lation of customers, the depth

of statistical data analysis, and

other relevant aspects encoun-

tered during the execution of

the project.

How to Design a Marketing Research Project | 31


4/41

The main problem with such objectives is not that they are wrong but that they can-

not necessarily be solved by marketing research alone. Specific problems for a research

project must be defined in such a way that they can be answered directly by the study. If

this is not done, the research supplier might be held accountable for not solving a prob-

lem that could not be solved by marketing research alone.

For instance, in the example above, the research may or may not be able to find out

how the sales of ProductXcould be increased. The marketing research might identifythat ProductXs advertising is less likable than Product Ys commercials. If so, the total

advertising process would involve formulating the strategy and tactics to improve

Product Xs advertising, especially the likability, and then developing the creative ad

and placing it in the most effective and appropriate vehicles. This goes well beyond

marketing research. The objective defined must be specific enough that it can be

achieved solely through marketing research. If the researcher were to analyze the prob-

lem more closely, research objectives could be solely derived from marketing objectives.

(See Exhibit 2-2.)

EXHIBIT 2-2 A Marketing Problem of Market ShareMarketing problem: Why is our market share going down?

Factual information and analysis:

1. Consumers buy less of the brand than they did before.

2. This could be the result of fewer consumers buying the brand, or the same consumers buying

in smaller quantity, or both.

A research project might be designed to determine information on aspects such as:

whether consumers feel that the brand is inferior to competing brands and has become

more so during the recent past; the frequency with which consumers buy the product and the quantity per purchase

occasion;

the frequency with which consumers buy the brand and the quantity per purchase

occasion;

what other products tend to be bought with and be used together with the brand and

whether there have been any changes in these other products;

whether consumer brand perceptions have changed during the recent past and the

direction of those changes;

whether consumers tend to use another product as a substitute;

what proportion of buyers exist in different demographic groups;

whether there have been any changes in styles, government regulations, or the

environment that might have affected sales of the brand;

whether there is any change in claimed usage.

The aspects listed above can be answered directly by research. These, in effect, become the

marketing research objectives. Achieving these research objectives may or may not fully solve

the main marketing problem. The important point is that the research objectives must pertain

to the consumer, who is the subject of the research. While there may be other relevant variables,

the research will provide information to the decision maker as to the probable causes of his or

her marketing problem if the project is designed properly. Let us analyze another marketing

problem to see how it can be translated into manageable research objectives.

32 | CHAPTER 2


5/41

EXHIBIT 2-3 A Marketing Problemof Increased Profitability

Marketing problem: We would like to increase the price of our brand of soap, but we want to

make sure that this will result in increased profitability.

Factual information and analysis:

1. An increase in the unit price would increase the unit profit. This is purely a manoeuvre to

increase the profit margin and has nothing to do with the cost of production.

2. If people buy less or switch to other brands, then our total profitability will decrease rather

than increase. Therefore, we need to know how the consumersin particular the current

users of our brandwould react to a price increase.

A research project can be designed to determine the following:

whether consumers in this category are price conscious;

whether consumers in this category are price sensitive;

whether users of our brand are price sensitive;

the degree of loyalty to our brand among current users; the extent to which our brand users will remain loyal (continue to buy our brand) when

the price is increased;

whether the current users will switch to a competitive brand or simply consume less if

they find the price increase unacceptable.

These initial research objectives may be sharpened, extended, and modified depending on how

much information is already available from other sources. Another major consideration is the

availability of sufficient resources to accomplish all the research objectives. When resources are

limited, some research objectives (judged to be less important) may be sacrificed.

CONFUSING LATENT VARIABLES WITH MANIFEST VARIABLES

Now think about the challenge of measuring consumers attitudes. Attitudes are com-

prehensive, persistent, general evaluations of products, services, people, ideas, behav-

iours, and other constructs. Think about the difference between the challenges of

measuring a respondents age and a consumers attitude toward a brand of car. Variables

such as age, income, education, and purchase behaviour can be measured directly and

confirmed by the presence of the physical product. Brand purchase can be confirmed

physically through regular observations. These variables are called manifest variables.

However, an attitude can never be observed. A can of peaches can be placed into a shop-

ping cart and wheeled away. The consumers degree of liking the brand of peaches exists

only in that persons mind. One of the continuing major problems of marketing

research is how to measure peoples attitudes.

Theory holds that the physical act of purchasing an object depends on the buyers

attitudes as well as many other things. Since these attitudes can never be seen, they are

described as latent variables. If we were designing a research project to understand why

people buy very expensive, but outstanding, Sony plasma high-definition televisions or

luxurious European sport sedans, direct questioning of the consumers would hardly

provide us with any major insights. It is highly unlikely that users of these products, who

might be motivated by desires to increase their perceived social status through the use

of prestige brands, will admit to buying those products to enhance their social positions.

An even more serious threat to validity is that they may not even be aware of the real

reasons for their purchase behaviour.

Manifest variables are thoseaspects of the customer that

can be directly observed or

proven. Age, income, educa-

tion, and the number of boxesof cereal bought last month are

all observable and provable.

Latent variables are thoseaspects of behaviour that can-

not be directly observed such

as respondents attitudes.



6/41

Most marketing research projects include questions intended to measure both

manifest variables and latent variables. The values of manifest variables are obtained by

directly observing the behaviour of consumers or by direct, structured, and undisguised

questions. Manifest variables are associated with behaviour that is directly observable

and about which respondents can be expected to provide direct and accurate answers.

Examples of manifest variables include the age, education, income, marital status, sex of

a person, the model of car owned by a person, etc.

Accurately measuring latent variables continues to be a significant challenge to

marketing research. Many consumer characteristics cannot be directly observed. In

other instances, consumers cannot or will not answer questions about the characteris-

tics referred to as latent variables. The inherent nature of individuals attitudes is that

they are deeply held in the conscious and unconscious mind and will never be accurately

measurable in the same way that manifest variables are measured. Measurement of

latent variables through direct questioning only provides partial information.

HOW LATENT VARIABLES ARE RELATED TO LATENT CONSTRUCTS

The first type of latent variable that we considered was one related to a single question

where the subjects correct answer can never be observed or proven. This is often

referred to as a hypothetical construct (i.e., something we think exists but can never be

proven). Lets take this concept a little further and consider the topic of customer loyalty.

There are two key components of customer loyalty: behaviour and attitude. Behavioural

loyalty refers to the repeat purchase of a brand on some measured basis. For example,

some companies might say that a customer is loyal if four out of five purchases are of

the companys brand. Other companies might be very pleased to receive every other

purchase that a customer makes. Whatever the criterion for behavioural loyalty, it relies

on the observable and measurable action of buying a brand. This is a manifest variable

because it is provable.

Measurement of hypothetical latent constructs is typically attempted through the

measurement of several latent variables that are thought to capture the essence of the

latent construct. For example, brand loyalty, corporate reputation, and customer

satisfaction are all latent constructs.

A hypothetical construct per-tains to something that we

think exists but can neverprove. For example, customer

satisfaction and corporate rep-

utation are hypothetical con-

structs.

34 | CHAPTER 2

EXHIBIT 2-4 Examples of Latent Variables

OBJECTIVE

Identify reasons for consuming alcohol among upscale adults.

Respondent: To relax with friends.

Latent behaviour: Alcohol dependency

Better approach: Use an indirect research method, for example, projective techniques (see

Chapter 4)

OBJECTIVE

Determine the likelihood of medical doctors prescribing different brands of a pharmaceutical

product to treat Alzheimers disease.

Responding MD: I typically prescribe BrandX.

Latent behaviour: Difficult to ascertain likely future prescribing behaviour when new drugs are

available.

Better approach: Present MDs with several pharmaceutical products and ask them to choose the

one they would prescribe for a specific type of patient. Repeat this with different combinations

of products and different types of patients.


7/41

Attitudinal loyalty is a mental state where a customer is primarily motivated to sat-

isfy her or his needs provided by a product category by purchasing one particular brand.

This attitudinal loyalty is a hypothetical construct, the measurement of which might be

based on several variables rather than only one. For example, attitudinal loyalty could

be combined with behavioural loyalty and represented by the latent construct of brand

health, as in Exhibit 2-5. It is reasonable that brand health, being a latent construct, can-

not be measured directly. This is also true ofAffinityandAffect.Needs Met, Taste, Love

Brand, Upset if Gone, Customer Share, and Brand Consumerare all latent variables. In

these cases, the objective of the statistical analysis is to identify those latent and mani-

fest variables that are strongly related to the latent construct of brand health.

In summary, researchers can never be 100 percent certain that they correctly

measure latent variables or latent constructs. In fact, some attitudinal marketing con-

structs such as quality, customer satisfaction, and loyalty might need to be measured by

several variables.

CONFUSING PROJECT OBJECTIVES WITHTHE SCOPE OF THE STUDY

The third problem with clearly defining marketing research problems is confusing the

objectives with the scope of the study. The objectives specify the direct purposes for con-

ducting the study. At the end of the project, each objective must be addressed by the

study findings with marketing-oriented solutions provided. The scope of the study

relates to how the project will be conducted. The extent and context of the study pro-

vides parameters for its scope.

For example, if the objective is to identify the target market for a new brand of

detergent, the scope of the project specifies the extent of the geographic capture of

information, the range of people to include, the breadth of analyses to conduct, and the

time period during which the study is conducted. The scope of the study should relate

directly to the project objectives. If the objectives are poorly developed, the scope of the

actual study might not be fully relevant to solving the marketing research problem or

the marketing problem.


EXHIBIT 2-5 Measuring Brand HealthA Latent Construct

Needs Met

Love Brand

Taste

Upset if

Gone

BrandIncidence,

Last 5

Customer

Share

Brand

Consumer

Affinity

Affect

Brand

Health

Nestler, M. and K. Deal, "A New Brand Health Model," Advanced Research Techniques Forum, American

Marketing Association, 2002.


8/41

A Practical Guide to DefiningMarketing Research ProblemsMost marketing research objectives can be defined by using the four-step procedure

provided in Exhibit 2-6.

The marketing research objectives are to specify the information needed to advance

the understanding of the actual market situation and the marketing actions that can rec-tify the problem.

Steps in Defining the ProblemThe essence of defining the objectives for a marketing research study is identifying the

components of a marketing problem that can be answered by such a study. The scope of

the study is defined by the objectives and modified by the resources available. Exhibit 2-7

shows the steps in defining the marketing research problem. This process requires a

thorough understanding of the marketing problem, the information that is known from

past studies, secondary sources of information (see Chapter 3), and other organization-

al and environmental resources. Also of great importance are the hypotheses that needto be tested. Hypotheses are the decision makers understanding of the marketing envi-

ronment that need to be tested on data during the study.

The scope of the study describes the breadth of all the information in the study. For

example, a broad scope is a sample drawn from the population of all Canadians 18 years

of age or older across Canada. A narrow scope is drawing the sample from the popula-

tion of Canadians who own their own single detached homes in the lower mainland of

British Columbia. A narrow scope for a questionnaire is to ask respondents only about

the last purchase of a product; a broad scope is to ask about the last purchase, the last

five purchases, their current attitudes about the product, and their intentions regarding

upcoming purchases.

36 | CHAPTER 2

EXHIBIT 2-6 Defining Marketing Research Objectives

STEP 1. STATE THE MARKETING PROBLEM.

The sales for our brand are not increasing.

STEP 2. EXPLICITLY STATE THE FACTUAL INFORMATION IMPLIED IN THE MARKETING

PROBLEM.

The total sales of our brand is the same quantity as last year. There have been no significant

changes since last year either in total or among the geographic regions for which data is collected.

STEP 3. STATE THE POSSIBLE CAUSES THAT MIGHT HAVE LED TO THIS PROBLEM.

The market for the category has expanded but not to the benefit of our brand.

The market for the category has eroded since last year but not to the detriment of our

brand.

The market has shrunk, to our detriment.

The market is static and so are our sales and the sales of our main competitors .

STEP 4. LIST WHAT YOU WOULD LIKE TO KNOW ABOUT THE MARKET.

You may want to test one of the four states listed above or you may already know which one of

the four is the problem. Depending on the state of your knowledge, you may list the type of infor-

mation that can be obtained through marketing research.


9/41

Answering the following questions will help clarify the objectives and scope of the

project:

1. What is the marketing problem for which a solution is sought? (What mar-

keting decision will be based directly on the information collected?)

2. What is the marketing research problem?

3. What information is required to solve the problem?

4. What are the different sources that can provide the required information?

5. What part of this information can be obtained from consumers directly?

6. What part of the required information can be obtained from secondary

sources?

7. What resources are available for this purpose?

8. How diverse or narrow will each study component be?

9. Once this information becomes available, how can it be used to solve the key

marketing problem at hand?

10. What additional information can be collected? (The reason for collecting

additional information can be twofold: to augment the primary information

collected, as outlined in step 6 above; and to maximize the benefit derived

from the projectthe marginal cost of asking a few additional questions is

likely to be small compared to the total cost of the project.)

The objectives of the study are less negotiable than its scope. The objectives relate

to the basic purpose of the study. Sometimes a few of the objectives need to be cut due

to budget or time constraints or due to the limitations of respondents during a survey.

But, generally, the marketing manager will want to keep all or most of the objectives. If

the budget has a fixed limit that will not accommodate the full initial list of objectives

and scope, a decision will likely be made to limit the scope prior to a consideration of

eliminating objectives.


EXHIBIT 2-7 Steps in Defining theMarketing Research Problem

What

marketing

decisions

depend onthis research?

What

is known?

What

are the

unknowns?

What

is understood

about the

marketing

environment?

What

hypotheses

need to be

tested?

How broadly

does

information

need to be

collected?

What

information

must this

researchprovide?


10/41

The formalizing of information needs, as detailed above, clarifies the problem, sets

realistic goals for research, and defines the scope of research very clearly by weeding out

information that is not relevant to decision making.

CAN THIS STUDY PROVIDE ANSWERS TO THE PROBLEM?

Certain requirements must be met for a marketing research study to provide answers to

the problem. The two primary requirements are reliabilityand validity. While thesetwo words are often grouped together, they have two distinct meanings.

Reliability refers to the consistency of measurements. A study is reliable when a par-

ticular behaviour, object, trait, or construct is replicated using the same measurement

procedures among samples that are drawn from the same population and are compara-

ble. Reliability is a key indicator of the quality of the research measurement process. If

something is not well controlled among the measurement occasions, it will likely result

in measurements that are statistically different.

Reliability is necessary for a measurement instrument and process to be valid.

However, other requirements must be met as well. For example, the fastest way to lose

weight is to adjust your bathroom scale to read lighter. If you want to lose 5 kg quickly,

adjust your scale. If your actual weight stays exactly the same and if you have a goodscale, you will seem to weigh 5 kg less. The consistency of these weights indicates a reli-

able scale, but it is not a valid measurement.

Validity means that you are measuring exactly what you intended to measure. So, the

scale in the example above is not valid because it provides a reading lower than your true

weight. If the scale always provided your exact weight, it would be valid. It would also be

reliable. If a measurement instrument and process are valid, they are also reliable. If the

instrument is not reliable, it cannot be valid. If it is reliable, it might or might not be valid.

Naturally, clients and consultants intend that the research produces valid and, conse-

quently, reliable findings. The efforts of all parties should focus on building a high-quality

research mechanism that is accurate (valid) and produces consistent (reliable) findings.

The chart in Exhibit 2-8 shows an example of a study among six shopping malls in

Ontario that is not reliable and, consequently, cannot be valid. The study was intended to

measure the percentage of beverage drinkers who stated that one company put out a speci-

fic beverage when they were shown a bottle for a different beverage. The finding at

Queenston Mall in Hamilton (70 percent) is statistically different than the finding from Bay-

field Mall in Barrie (9.5 percent). A reliable research process would not produce measure-

ments such as these, which are significantly statistically different. It was stated in the report

that the samples were drawn from the same population and used the same methodology.

WHAT ARE MARKETING RESEARCH PROJECTS LIKE?

The research design specifies the path and the methodology for solving the marketing

research problem. It can be as simple as a plan for a basic descriptive study involving

seven-minute telephone interviews among the general population or as complex as an

elaborate experimental design for testing a marketing theory. The basic marketing

research path shown in Exhibit 2-9 accommodates all marketing research designs, from

the very simple to the tremendously complex.

The path begins with the steps covered earlier in this chapter: define the marketing prob-

lem, define the research problem, outline the information needs, specify the data required,

and decide on the study methodology. We will spend significant parts of this book on con-

tinuing to clarify these concepts and explaining the remaining parts of this research path.

There are three essential components to the marketing research design task. Exhibit

2-10 presents the different combinations of project type, information requirements, and

field methodology that might be employed in any one project. Naturally, for any given

38 | CHAPTER 2

Reliability in research projectsmeans that those projects canbe repeated with the same

questions on similar subjects

with similar findings obtained.

Validity in research projectsmeans that the projects meas-

ure what they intended to

measure.


11/41

problem, some designs are more appropriate than others. When deciding on the design

of marketing research projects, the following three components of the design task must

be fully specified:

1. the type of marketing research project

2. the information requirements

3. the field methodology

Marketing problems are often characterized as particular types of marketing

research projects. For example, if sales are down, we do a customer satisfaction study. If


There are several standard

types of marketing researchprojects such as usage andattitude studies, pricing

research, and market segmenta-

tion studies. These have very

similar components regardless

of where they are conducted.

The information requirementson a research project relate to

the objectives and needs of

the marketing research and

marketing problems; these are

typically listed as exploratory,

descriptive, and causal.

Define theMarketingProblem

Define theResearchProblem

Outline theInformation

Needs Specify theDate

Required

Decide onStudy

Methodology

ResolveSample

Definition

DeviseSampling

Plan

DesignQuestionnaireCollect

Data

AnalyzeData

InterpretFindings

PrepareReport &

Presentation

Present Findings,Make

Recommendations

Evaluate, Feedback& Payments

EXHIBIT 2-8 An Example of Unreliable Survey Findings

70% Reliability is the agreement between two or more efforts tomeasure the same trait through maximally similar methods.

The tests used were intended to be maximally similar butproduced dramatically different measurements. Consequently,this test cannot be reliable and, necessarily, is not valid.

Queenston,Hamiltonn = 20

Fairview,Toronton = 21

PercentageStating

thatCompanyB

PutsOuttheBeverageShow

n

Lawrence Sq.Toronton = 20

Westmount,Londonn = 20

Place d'Orleans,Ottawan = 22

Bayfield,Barrien = 21

Percentage of Respondents Shown the "A" Bottle Who Stated That

Company B Puts Out That BeverageAn Example of Unreliable Findings

42.9%40%

30%

13.6%9.5%

0

10

20

30

40

50

60

70

80

EXHIBIT 2-9 The Marketing Research Path


12/41

we need to evaluate various new product ideas, we do a type of new product research

called a conjoint analysis. If the type of project is identified correctly, then the critical

second step of the marketing research project has been successfully completed. As we

learned earlier, step one is correctly defining the problem and step two is identifying the

information needed to solve the problem. Some common types of marketing research

projects are presented in the first column of Exhibit 2-10.

Many problems can be translated quickly and accurately into one of the standardresearch types outlined in Exhibit 2-10. However, there are cases where premature

labelling of a marketing problem as a specific type of marketing research project results

in collecting the wrong type of information and the original need for information might

not be properly addressed.

Each of the types of research studies detailed in Exhibit 2-10 can sequentially step

through the three information phases of exploratory, descriptive, and causal research.

Although most studies tend to provide descriptive information, many would benefit

from more careful consideration of the exploratory stage and extension to the causal

research phase. In practice, the distinction between exploratory, descriptive, and causal

research is not always made. However, the fundamental distinctions still exist.

Some of the project types are typically conducted using a specific type of fieldwork;

for example, concept tests have traditionally been conducted in face-to-face interviews.

However, most of the research types are flexible enough to accommodate the fieldwork

procedure that is most appropriate given the specific domain of the study at hand.

Accurate research design is essential for the proper execution of the marketing research

project and for supplying the right information concerning the marketing problem.

Exhibit 2-10 shows another way to decide on the design for methods used to collect

information. Given a problem, the researcher will want to decide on the best possible

approach to answer the following questions:

Is the problem better solved by quantitative rather than qualitative methods?

Do I have to design a special study to solve the problem or can I make use of

an omnibus study or an ongoing panel?

40 | CHAPTER 2

EXHIBIT 2-10 Designing the Marketing Research Project

Type of Marketing Information

Research Project Requirements Field Methodology

Usage and Attitude Studies Exploratory In-Depth Interviews

Concept Tests Descriptive Focus Groups

Tracking Studies Causal Personal Interviews

Segmentation Studies Telephone Interviews

Product Testing Mall Intercepts

New Product Research Self-Administered Questionnaires

Test Marketing Internet Survey

Advertising Research Panel Study

Name Testing Omnibus Study

Package Research

Customer Satisfaction

Brand Loyalty

Pricing Research


13/41

Does the study call for personal interviews or can I use alternative, less expen-

sive methods like an internet or mail survey? Can the interviews be carried out in a shopping mall?

By answering questions like these, the researcher can arrive at one or more research

designs appropriate to the problem. An experienced researcher often arrives at suitable

research designs by mentally eliminating the inappropriate ones and evaluating the per-

tinent designs according to a set of constant criteria. The final design chosen dependson several factors, such as:

the suitability of the design for the given problem cost considerations the level of accuracy desired timing constraints other project-specific aspects that are variable

Although the designs are shown as alternatives in Exhibit 2-10, they are not mutu-

ally exclusive. Few problems require unique research designs. For example, a research

study may include a qualitative phase using focus groups followed by an ad hoc quanti-

tative stage using both internet surveys and computer surveys by respondents recruitedto central locations through telephone interviews.

STANDARD RESEARCH TYPES

One of the most critical things to understand in marketing research is that the primary

aim of a research design is to answer the research questions and, subsequently, the mar-

keting questions in an effective way while keeping the costs at a reasonable level. In prac-

tice, the marketing scope, the marketing research objectives, and the costs of the study

determine the research design. Listed in Exhibit 2-11 are the most commonly encoun-

tered marketing research project designs and some of the marketing problems that each

is intended to address.

LEVEL OF INFORMATION REQUIREMENTS

Although practical marketing research projects do not often characterize studies accord-

ing to their information requirements, it is often helpful to view projects from this per-

spective. A common way of referring to the information requirements of marketing

research projects is as follows:

exploratory studies descriptive studies causal studies

This classification usually represents a hierarchy or sequence of projects.

Exploratory projects are usually conducted first, followed by descriptive studies. Causal

studies require very sophisticated design, execution, and analysis and are typically con-

ducted after exploratory and descriptive studies have provided substantial information

about the market and consumers.

EXPLORATORY STUDIES Suppose that a company has identified a marketing problem

significant enough that the marketing research staff has been called to a meeting to dis-

cuss the options. It is important that someone at that first meeting recognize that the

marketing issue and the marketing research problem are not understood well enough to

enter immediately into a full-survey project. At the beginning of a new research project,

a company needs to explore the basic nature of the marketing issue that is facing it in

order to plan an effective fact-gathering mission (a survey). Since exploratory studies are

When little is known on a

topic, the research is called

exploratory. It is often

acknowledged that the initial

exploratory study will help torefine research methodology for

future projects. These studiesoften have a significant quali-

tative component.

Descriptive studies explainthe current state of the market

and marketing to a sector.

Causal studies attempt toexplain cause-and-effect rela-

tionships using some of the

most sophisticated design and

analysis techniques.



14/41

42 | CHAPTER 2

EXHIBIT 2-11 Marketing Problems and Research Techniques

Type of

Research Study

Usage and

Attitude Studies

Concept Tests

Product Testing

Segmentation

Studies

Tracking Studies

Advertising

Research

Test Marketing

Name Testing

Package Research

New Product

Research

Marketing Problem/Issue

Who and how many use the product category/brand?

What are the differences between heavy and light users?

What are the main features of their behaviour and

habitsfrequency of purchase, brand loyalty, what do

they mix it with?

What are their attitudes, beliefs, values to existing brands?

What are their product requirements?

What are the brands strengths and weaknesses?

Tell me everything I ever wanted to know about... !

Which of a number of alternative new product ideas has

the most appeal?

Is the idea easy to understand?

What expectations and associations does it generate?

Do you perceive distinct benefits for this product over

those products currently on the market?

Are the claims about this product believable?

Would you buy this product?

Would you replace your current brand with this new brand?

Would this product meet a real need?

What improvements can you suggest in various attributes

of the concept?

How frequently would you use the brand?

Who would use the brand?

What share potential does it have?

If I reposition my brand this way, will it work?

Which is the preferred prototype?

How much is the improved formulation liked?

Will consumers buy a cost-reduced product?

Is our product acceptable to consumers?

Is our product marginally or much better/worse than the

competition?

Can we get away with substituting carob for chocolate?

Does the product weve developed deliver its promise?

Which brand is betterours or P&Gs?

How does the market segment (demographics, psycho-

graphics, behaviour, attitudes, lifestyles)?

How large are the segments?

What is the potential of the segments?

How can each segment be reached with ads?

Can I improve my brands current positioning?

Are there any gaps in the market that can be exploited?

What are the levels of brand and advertising awareness

over time?

What are the progressing levels of trial and purchase?

Is there any conversion to occasional and regular usage?

Is the advertising communicating, compared to last quar-

ter, last year, etc?

How have attitudes toward the brand changed?

How has our brands image changed relative to that of

competitive brands?

How believable is our advertising?

How relevant is our advertising to the target consumers?

How much do the target consumers like our advertising?

How much power does our advertising have over the

receivers?

To what does our advertising appeal?

Does our advertising provide the information that the con-

sumers want?

Has our advertising shifted consumer attitudes?

How will our new brand fare in an actual market situation?

How well does the new brand gain retailer interest?

How does the new brand do against real competition?

Is the marketing plan strong enough to gain initial trial?

Does repeat purchase build as expected after initial trial?

Which is the best name out of the alternatives?

Which name connotes the benefits that consumers wantmost?

Is the name easily pronounced?

Does the name have any secondary meanings?Is the name remembered?

Do the target consumers like the name?

Which package communicates best on the shelf?

Which package is most visually appealing?

Which package is most useful to consumers?

Which package produces the best imagery of the brand to

consumers?

How effective is our advertising in building awareness of

the new brand?

Which brands will our new brand compete against?

How effective is our marketing plan in encouraging initial

trial?

Will the brand characteristics entice consumers to repeat

purchase?

How strong is the likability of the brand?

How can we answer all of the above as inexpensively as

possible?


15/41

conducted at the very beginning of a project, there is little more to rely upon than the

experience of the marketing managers, the firms marketing research manager, and the

consulting marketing research project director. The purposes of research designs are

shown in Exhibit 2-12.

There is always a great temptation to charge the marketing and research staff with

the task of defining the project completely and of getting on with the descriptive or

causal study. When this acceleration of the project occurs, the information gathered bythe study might not coincide with the real need to solve the marketing problem. The

research team might be so frantic to get something done that it loses its focus on the

problem. A project that is improperly designed at the early stages cannot be saved

through later manipulations. The design must be conducted thoughtfully and sensi-

tively right from the beginning.

The main output from an exploratory study is a solid understanding of the nature

of the marketing problem and a set of hypotheses that can be tested through survey

research. Classification of research designs is shown in Exhibit 2-13.

DESCRIPTIVE RESEARCH Descriptive research constitutes the bulk of marketing

research activities. A descriptive study begins with a good understanding of the market-

ing problem and of the marketing research problem. This understanding is based on

experience, knowledge of how the study will be executed, and on a set of hypotheses that

describe the relationships between the variables in the study. The difference between

exploratory and descriptive research is that descriptive studies have well-defined

hypotheses of relationships between variables while exploratory studies contain basic

intuitive hypotheses, at best.

In large firms, very often there is no lack of experience on the topic being researched

since the current project is sometimes a re-creation of some past study on the same

topic. At the very least, the marketing manager or especially the marketing research con-

sultants have executed similar studies on other brands many times in the past. For exam-

ple, concept tests are executed several times each year by most of the large packaged

goods companies in the vicinity of Toronto. Consequently, someone in the marketing

research department or among the marketing management staff has had experience

with that type of study. However, a particular firm might never have actually executed a

loyalty segmentation even though several people in the firm might have substantial

understanding of the nature and value of this project type. The expertise necessary to

conduct a loyalty segmentation, for example, is available from consultants at large

research firms and also from some smaller specialized research consultancies.


EXHIBIT 2-12 The Purposes of Research Designs

Exploratory Studies Causal StudiesDescriptive Studies

Formulate problems moreprecisely

Develop hypotheses

Establish priorities for research

Eliminate impractical ideas

Clarify concepts

Describe characteristics ofcertain groups

Estimate proportion of peoplein a population who behavein a certain way

Make specific predictions

Provide evidence regardingthe causal relationshipbetween variables by means of:

time order in which variablesoccur

elimination of other possibleexplanations

concomitant variation


16/41

The end report of a descriptive study provides a portrait of the type of people who

are heavy users of a product, for example. These descriptions would be fairly complete

and the relationships between some of the variables would be tested through standard

statistical hypothesis testing procedures.

In practice, the exploratory, descriptive, and causal stages are not singled out for

specific identification during a project. In fact, most projects would be called descrip-

tive. If asked, many researchers would say that standard marketing research projects

(i.e., descriptive studies) contain information that leads to conclusions regarding cause-

effect relationships. To some extent, they would be right. However, the term causal study

is reserved for more formal studies specifically designed to test causal hypotheses.

CAUSAL RESEARCH Causal research is directed at investigating and establishing the

cause of some phenomenon. The behavioural effects of consumers attitudes and states

as represented by several marketing variables are the concentration of many causal stud-

ies. For example, a soap manufacturer might be interested in whether the advertising or

the pricing for brands of soap is more instrumental in causing shifts in the demand for

the brands within the category. Alternatively, a dog food manufacturer might need to

determine whether media advertising or acceptability to the pet is more effective incausing shifts among category brands.

While causal research is potentially the most beneficial type of research, it is also the

most difficult to gain approval for and to execute. Causal research is usually the most

expensive type to carry out because of:

the great care needed in designing the project; the requirement for very highly qualified project directors, analysts, and inter-

preters of the findings; higher field costs resulting from the more intricate project designs.

Causal hypotheses are directed at proposing and testing relationships between more

44 | CHAPTER 2

EXHIBIT 2-13 Classification of Research Designs

SingleCross-

Sectional

MultipleCross-

Sectional

OmnibusTruePanel

ExperimentCausal

Modelling

ExperienceSurvey

Lit.Search

Cross-Sectional

Longitudinal

Exploratory

Research Design Classification

Conclusive

Descriptive Causal

CaseStudies

FocusGroups


17/41

variables than are descriptive hypotheses. The intention with causal hypotheses is to be

able to conclude specific cause-effect relationships that can be used to direct the highest-

level marketing decisions. The ideal would be to provide the advertising manager, for

example, with information that would allow for the achievement of specific sales figures

based on the total advertising budget, and the placement and timing of that budget.

Unfortunately, causal research based on traditional experimental designs has not

been spectacularly impressive in its contribution to marketing practice. One exceptionis the use of experimentation in the design, execution, and analysis of discrete choice

conjoint studies.

A related area called causal modelling or structural equation modelling has grown

in importance during the past 25 years. This procedure depends on developing a com-

plex set of hypothetical relationships and asking the questions so that the answers can

be analyzed with a particular type of statistical software. The end result is the develop-

ment of a set of statistically supported relationships that explain how a key operational

marketing result such as customer retention or brand health is achieved and is related

to other marketing activities, consumer attitudes, and personal characteristics. The dia-

gram in Exhibit 2-5 is the end result of an extensive analysis that included many more

variables. The final product is often said to establish weak causal relationships.

CAUSAL RESEARCH AND THE DESIGN OF EXPERIMENTS Establishing one or more vari-

ables as the cause of some phenomenon is one of the most challenging and complex

tasks in research. The focus of true causal research is on the design, execution, analysis,

and interpretation of experiments. Sometimes these experiments are conducted in lab-

oratories and other times they are set in actual market conditions.

A very simple experiment to understand the effect price has on soft drink sales

might be designed to be conducted in a laboratory. First, the range of price and the

number of different prices for the product would be identified. Lets say that four prices

were selected for Sprite, in 1.5 litre bottles. Second, a large sample would be randomly

selected and recruited from the relevant population of interest, perhaps young adults

between the ages of 18 and 25. Then, one-quarter of the respondents would be shown

the Sprite 1.5 litre bottle and asked how likely they would be to buy the soft drink at

$1.29. The second group, also one-quarter of the total sample, would be asked their pur-

chase likelihood at $1.49, the third group at $1.79, and the fourth group at $1.99. Other

questions could also be asked to identify purchase frequency and to understand the

demographic profile of the groups to ensure that they were comparable. The differences

in purchase likelihoods and frequency among the four groups could be attributed to the

price levels only because the products and the study environments would be the same

and the sample groups would be nearly identical in key attributes.

Now, think about moving the laboratory finding into the real world. Laboratory

administration has the benefit of eliminating stimuli other than the product from influ-

encing the subjects responses. However, the findings from laboratory experiments aresometimes difficult to generalize to the real marketing environment in which consumers

actually make their product choices. What else might compromise the findings? Of

course, no soft drink is without competition. The laboratory experiment presented only

Sprite to the subjects. But what would happen if 7-UP was priced lower or higher and

what about all the other brands?

Experiments can be very complicated to design, administer, and interpret. Yes,

experiments provide the best opportunity to test the effect of variation in product

attributes and other variables on key responses of consumers. A procedure called con-

joint analysis is an experiment-based procedure that has become very popular for this

type of research.



18/41

HOW IS DATA COLLECTION DESIGNED?

Regardless of the type of marketing problem, the methods used for obtaining information

may be quite diverse. Although marketing research is often equated with a survey, this is

not always the case. Sometimes qualitative information is best suited to the needs of a mar-

keting problem while, in other cases, a large cross-Canada survey is absolutely necessary.

It is often difficult to indicate where one project component begins and another ends.

Moreover, since projects can have all three information componentsexploratory,descriptive, and causalthe breadth of data collection design challenges can be extensive.

QUALITATIVE RESEARCH

Qualitative research is the term used to refer to studies that obtain information in the

form of verbatim comments resulting from unstructured conversations with a relatively

small number of respondents. The best-known types of qualitative research are focus

groups and in-depth personal interviews. While focus groups are normally conducted

with eight to twelve people, in-depth personal interviews are face-to-face conversations

between the researcher and one respondent. The findings from focus groups and in-

depth personal interviews are typically summaries of the comments of the respondents,

punctuated with their verbatim quotes.Qualitative research is used most often to obtain information for the exploratory

phase of studies. Qualitative research is sometimes the only research conducted on a

topic. However, occasionally qualitative research is used to obtain clarification of the

findings of descriptive and causal research. From time to time, quantitative findings

cannot be effectively assessed by the researchers. A focus group presented with those

same findings might illuminate the results and give them substantial usefulness.

Qualitative research is explored in considerable detail in Chapter 4.

WHEN IS QUALITATIVE RESEARCH MOST USEFUL? A qualitative research design is

chosen when the main interest for the project is in:

understanding how real customers talk about the brand; discovering the language and vocabulary that customers use when talking

about brands in the product category; exposing brand managers to frank discussions by their customers about the

companys brand and its competitors; obtaining background information; tracing salient behaviour patterns, attitudes, and motivations; exploring alternative concepts; investigating suspected problem areas; screening contenders for large-scale studies; piloting or amplifying quantitative studies;

generating hypotheses for further testing by quantitative research methods.

For some of the situations described above, it is also possible to use quantitative

research. Therefore, these guidelines cannot be used as an absolute set of criteria for

determining whether one should do a qualitative study. However, as one gains experi-

ence in research, it becomes easier to determine the appropriateness of qualitative

research for a given problem.

WHEN SHOULD QUANTITATIVE RESEARCH BE USED?

The bulk of marketing research uses quantitative techniques. The term quantitative

research is used to describe virtually any research project in which the interpretation

Qualitative research is focusedon obtaining respondents in-

depth comments on topics

through their verbatim discus-

sion of marketing issues.

Focus groups are discussionsorganized with six to twelve

people to fully explore a spe-

cific topic through oral discus-

sion.

In-depth personal interviewsare discussions that one mod-

erator has with one respondent

to qualitatively explore a spe-

cific topic in the words of the

respondent.

Quantitative research mainlyinvolves obtaining respondents

selection of one or several

answers to specific questions in

a survey, the quantification of

those answers over several

respondents, and the measure-

ment of the accuracy of those

answers.

46 | CHAPTER 2


19/41

depends on the numbers and percentages obtained from the investigation. Qualitative and

quantitative research complement and strengthen each other when used to solve well-

defined research problems. See Exhibit 2-14 for characteristics of quantitative studies.Quantitative research, when used in a formal sense, refers to studies that yield

results with known accuracy. For example, if the results indicate that 20 percent of the

consumers use BrandX, the analyst will be able to state the margin of error on this per-

centage along with the likelihood of error because these are aspects that can be mathe-

matically calculated. (This topic will be covered in Chapters 5 and 9.)

Compared to qualitative studies, quantitative studies use larger numbers of respon-

dents and generally ask questions to which the respondent can give a one- or two-word

answer rather than several sentences. Representative samples (especially in large-scale

studies), structured interviews, and formal questionnaires generally characterize quan-

titative studies. The interpretation of the results is based more on the numbers obtained

than on the intuition of the researcher.

Qualitative study interviewers (moderators) often have substantial experience in

quantitative marketing research as well. Thus, qualitative moderators are often capable

of interpreting the findings from a qualitative study and drawing relevant implications

for related quantitative projects. However, the typical interviewer in a quantitative study

specializes in survey interviewing and is not usually trained or experienced in other

aspects of research. Consequently, he or she might not appreciate the full importance of

qualitative research to the understanding of the marketing problem and the success of

the project.

Quantitative studies are used more for measuring and evaluating than for investi-

gating and exploring. Most research studies are assumed to belong to this category even

when they do not fulfil the stringent criteria. Most of this book, therefore, is devoted to

understanding the mechanics of doing what is loosely described as quantitative

research.

Quantitative studies collect information through interviews with samples of people

from the target audience or through investigating existing data obtained from various

sources. However, the method of collecting information may differ from study to study.

Most quantitative surveys use either telephone or personal interviews. Some studies

(e.g., mail surveys) use self-completion techniques by asking the respondents to fill in

their answers to questions on printed questionnaires. Others use observational tech-

niques where, for example, the behaviour of shoppers when selecting popcorn at super-

markets would be watched by hidden researchers. Once they have made their selections,

the subjects may then be interviewed.

Moderators of qualitative stud-ies lead respondents through

the research topics and encour-

age deep consideration and

discussion of the topics.


EXHIBIT 2-14 Characteristics of Quantitative Studies

The findings are presented in numerical fashion (e.g., in tables, graphs, or equations).

Numerical data are based on large numbers of data points.

They may be based on secondary sources of data or on surveys conducted with struc-

tured questionnaires. Formal findings from quantitative surveys can be stated with known accuracy.

Quantitative studies are used more for measuring and evaluating than for investigating

and exploring.

Most quantitative surveys use telephone or personal interviews or rely on mailed inter-

net questionnaires.


20/41

PERSONAL INTERVIEWS Personal interviews can take place in the home, on the street,

in shopping malls, in pre-arranged locations (e.g., a large meeting room where respon-

dents can participate in a taste test), or in places people congregate for a specific pur-

pose (e.g., a polling station or football game).

IN-HOME PERSONAL INTERVIEWS The method in which the interviewer personally

visits the homes of the respondents for the purpose of collecting information is calledthe in-home interview technique. In-home interviews allow for accurate selection of

respondents because the interviewer can ask to speak to a specific person. For this rea-

son, they are particularly well suited for large-scale quantitative studies. The main

advantages of in-home interviews are listed in Exhibit 2-15.

On the negative side, in-home interviews can be relatively more expensive, espe-

cially when carried out in nonurban areas. In some cases, due to restrictions on trans-

portation, availability of interviewers, and the availability of respondents at any given

time, personal interviews may be difficult to execute. Another problem associated with

in-home interviewing is that only a minimal degree of control can be exercised over

interviewers in remote locations. If an interviewer does not understand the proper ques-

tioning process, this might not be caught until several of the completed questionnairesare edited. Even then, subtle problems might never by identified. Also, when the subject

matter includes personally sensitive and potentially embarrassing topics, personal inter-

views might result in more distorted responses since there is no direct observation of an

interviewers diplomacy with regard to delicate matters and, therefore, no opportunity

to correct any interviewing faults.

MALL OR CENTRAL LOCATION PERSONAL INTERVIEWS In mall or central location

personal interviewing, the respondents are invited to a central location such as a shop-

ping mall, theatre, auditorium, or meeting room where an interview can be held.

Typically, respondents for these studies are recruited using shopping mall intercept pro-

cedures and, to a lesser extent, through telephone screening interviews. In the mall-intercept technique, potential respondents are intercepted in shopping malls by

interviewers for preliminary information gathering. If the respondents qualify, they are

invited into an auditorium, an interview room, a van, or a screened-off area to partici-

pate in further tests. The advantages are listed in Exhibit 2-16.

Many research houses have permanent interviewing facilities in shopping malls.

Although phrases such as ABC Research bought Brightside Mall are commonly used

by marketing research practitioners, this really means that ABC leased space in the mall

and has exclusive rights to conduct survey research in Brightside during the term of

In a quantitative marketing

research study methodology,

personal interviews occurwhen one person at a time is

led through the quantitative

questionnaire to elicit his or

her answers. These can be con-

ducted in a respondents home,

office, or in a central location.

In central location personalinterviewing, people arerecruited in a location such as

a shopping mall to be inter-

viewed for a quantitative study.

48 | CHAPTER 2

EXHIBIT 2-15 Main Advantages of In-Home Interviews

Compared to other forms of interviews, the respondents are, in general, more

co-operative.

Interviewers can establish a higher degree of rapport with the respondents.

A relatively more complex and lengthy questionnaire can be administered with greater

ease.

Prompting and probing can be done more effectively.

Personal interviews allow for the presentation of visual aids and stimuli during the

course of the interview.


21/41

their lease. Sometimes a research house will rent out its time in the mall when it has no

study being fielded. Also, some fieldwork firms generously allow student groups to use

their mall rights for short course projects.

The main advantages of central location interviews are that they offer all the advan-

tages of personal interviews discussed earlier, are less expensive than in-home inter-

views, can handle unwieldy stimulus materials and equipment, and can be relatively fast.Mall intercepts provide the flexibility of combining interviewer-administered questions

and self-completion questions within a cohesive framework and relatively short time

period.

The main disadvantages of this method are that the respondents are seldom repre-

sentative in a statistical sense, they may not want to participate in a lengthy test when

they are in the middle of their shopping, and the response rate can be low. The sampling

aspects of mall intercepts will be discussed later.

CURRENT AND FUTURE CHALLENGES FOR MALL INTERVIEWING Mall intercepts have

replaced much door-to-door interviewing, once a standard technique. Money is the rea-

son. While other techniques can yield more representative and projectable samples,

door-to-door interviewers are expensive to hire, and telephone interviewing precludes

the face-to-face interaction that is required in some product and advertising testing.

Ninety-five percent of all American households shop at malls at least once a year;

two-thirds go once every two weeks. Compared to total population proportions, how-

ever, mall customers tend to be more affluent, and on weekdays, more seniors and teens

are present.

Increasingly, critics maintain that mall shoppers are not representative of the

larger population, making survey results unprojectable. The sample might not be repre-

sentative of the people who shop in the mall, depending on selection criteria. Even with-

in a mall, more low-income respondents are found near entrances close to bus stops

than near entrances close to parking lots.

In textbooks, mall-shopper samples are considered nonprobability samples with

no theoretical basis for estimating population characteristics, so public opinion surveys

are seldom done in malls. Academic researchers have studied the biases and suggest

weighting frequent mall shoppers so that they are not over represented and randomly

placing interviewers proportionate to pedestrian traffic.

Practitioners seem to have fewer qualms than academics. Market Facts of Chicago

found in October 1988 that 89 percent of the 140 largest USA product and service com-

panies used mall intercepts and that 58 percent of these spent more on this technique

than on any other. Mall intercepts are said to be quick, easy, and inexpensive. Florence

Skelly says that enough data exist on specific malls shoppers across the USA to create

slightly biased random population samples.


EXHIBIT 2-16 Central Location Studies

Central location interviewing is particularly useful when:

the tests require bulky or elaborate material or use gadgets;

commercials are to be judged by respondents;

taste tests need to be conducted; the tests involve different sensory modalities;

statistical accuracy is not important;

the study involves simulated shopping behaviour.


22/41

Perhaps it is mall owners who will end mall interviewing. Managers are growing

skittish about allowing interviewers on their premises, apparently from a civil rights

point of view: if they allow non-lessee researchers onto their premises, are they bound

to accept religious groups? (Hammond, 1989)

TELEPHONE INTERVIEWS In recent years, telephone ownership has become sufficient-

ly high in some parts of the world to make the telephone an adequate medium for col-lecting representative opinions on many subjects. However, in-home wired telephone

access is still a luxury of the future to the majority of the population in many develop-

ing countries. Therefore, it is in modern countries that a large proportion of marketing

research depends on telephone interviews as a method of data collection. In addition

to being an alternative method of data collection, telephone interviews have some spe-

cial advantages, which are presented in Exhibits 2-17 and 2-18. Exhibit 2-18 also lists

some disadvantages of telephone intereviews.

Telephone surveys are typically conducted from central locations where scores of

interviewing stations can be used. This allows for the interviews to be monitored by field

supervisors. In addition, Central Location Telephone interviewing (CLT) is typically

computer assisted, which will be described in Chapter 8 (Computer Assisted Telephone

Interviewing or CATI). Most CATI software now provides for the supervisor and client

to sit in a room and listen to the interviews while observing the computer screen in front

of the interviewer being monitored. In fact, the most advanced software allows clients to

sit in their offices or homes and monitor the telephone interviews over the internet.

These connections provide both audio monitoring and visual observation of the inter-

viewers screen as the answers are entered.

MAIL AND HANDOUT SURVEYS The mail survey is a classic example of the self-

completion technique. In mail surveys, questionnaires are mailed to selected respon-

dents (or to the members of a panel) who are asked to complete and return them within

a specified time period. Of course, it is necessary to have a list of the names and address-

es of the target respondent audience.

Mail surveys can be fairly low cost and they seem deceptively simple to execute.

Consequently, many inexperienced people tend to write self-completion questionnaires

without realizing their true complexities and that there are major differences between

mail surveys and interviewer-administered surveys.

A great deal of care must be taken with the development of self-completion ques-

tionnaires. Remember, this is the only type of questionnaire that is actually seen by the

respondents. Consequently, not only must the questions and answers be perfect, but the

physical placement of the questions on the page and the overall visual appearance of the

questionnaire must motivate the respondent to begin, continue, complete, and return

the questionnaire. (The procedures for developing and executing successful mail surveys

will be presented in a subsequent chapter.)One advantage of mail surveys is that the questionnaires are all printed in the same

50 | CHAPTER 2

EXHIBIT 2-17 How Telephone Interviews Are Useful

contact screening for main studies

measuring the immediate recall of commercials and advertisements

assessing the immediate impact of a newly announced government policy

continuous political polling

The most common form of

quantitative fieldwork is the

telephone interviews wherethe interviewer asks respon-

dents questions during a tele-

phone conversation.

Telephone interviews often use

Computer Assisted Telephone

Interviewing (CATI) where thequestionnaire is programmedinto the computer so that the

interviewer reads questions off

the computer screen and enters

answers directly into the com-

puter.

The self-completion tech-nique relies on the respondentto read the questions and

respond manually. These are

typically administered using

paper-and-pencil question-

naires, computer administra-tion, and the internet.


23/41

format. Each respondent receives a questionnaire that is exactly the same as that received

by every other respondent. In addition, since no interviewers are involved, there is no

concern about interviewing inconsistencies and other fieldwork difficulties that might

somewhat alter what one respondent hears compared to what another might hear. The

fact that no interviewers or field supervisors are needed to carry out the study ensures

the respondents anonymity (just omit the name and address questions). In addition,

respondents can give considered responses at their own pace.

However, the lack of an interviewer is a major drawback as well as a major benefitof mail surveys. One of the main difficulties is that there is no interviewer present to

clarify, probe, and ensure that skip patterns (e.g., instructions for skipping from ques-

tion 4 to question 8 if the answer to question 4 is yes, but continuing to question 5 if

the answer to question 4 is no) are followed correctly, and that the information pro-

vided is internally consistent. One of the jobs of the field supervisor is to ensure that

interviewers are consistent in the way in which they read the questions, record answers,

probe, present visual and audio stimuli, and so on.

Most mail surveys (Exhibit 2-19), especially the ad hoc variety, suffer from several

shortcomings. Complex skip patterns cannot usually be accommodated and the

researcher cannot withhold anything from the respondents to be revealed at a specific

stage of the questioning. Brand awareness questions are most effectively used wherean unaided awareness question (e.g., Please tell me the names of all the brands of


EXHIBIT 2-18 Advantages and Disadvantagesof Telephone Interviews

Telephone interviews offer several advantages over other forms of interviews, in particular,

personal interviews.

Telephone interviews are less expensive. They are fast.

They are better controlled (in central locations).

They have (in practice) fewer restrictive sample frames.

When they are computer assisted, complex questionnaires can be administered with

relative ease.

Disadvantages associated with telephone interviews include:

problems of screening

easy refusals

limited rapport between interviewer and respondent

inability to do physical checks or present visual stimuli restrictions on open-ended questions due to constraints on the time available for

recording answers

EXHIBIT 2-19 Mail Surveys

Mail surveys are used to:

survey the membership of an organization

survey subscribers to magazines

study customers of a small retail outlet or a financial institution

study buyers of a given product or service

Skip patterns are directionsfor moving from one question

to another in a questionnaire

where some questions might be

skipped, depending on answers

to previous questions.

A field supervisormanagesthe fieldwork operations of

marketing research, often by

organizing and supervising the

operations of field interviewers

to maintain high quality

requirements.


24/41

chocolate candy bars that you are aware of) is followed by an aided awareness question

(e.g., Have you ever heard of Skor chocolate bars?). Unfortunately, this sequence cannot

be used in self-completion questionnaires because respondents will usually look ahead

and find the brand names to fill in the unaided question. Further problems arise when the

respondent consults with his or her family and friends before responding to a question.

The speed of completion of mail surveys is almost totally in the hands of the respon-

dents. Consequently, the total time span of the fieldwork can be quite long. However, in

most surveys, cut-off dates are used to limit the response period. Also, the response rate

in a mail survey (unless it is a panel) can be very poor. It can be as low as 20 to 30 per-

cent or as high as 60 to 70 percent. Low-interest surveys can have response rates lower

than 10 percent. See Exhibit 2-20 for field procedures to increase response rates.

INTERNET SURVEYS Compared to other survey field methodologies, conducting surveys

over the internet is a relatively new phenomenon. However, the use of internet surveys has

grown greatly and has become an accepted part of the marketing research toolbox.

There are two main ways in which the internet electronic medium can be used for

conducting surveys. The first and oldest is to develop a questionnaire and simply copy

it into the text of the email message or attach it to emails sent to prospective respon-

dents. The second is to develop the survey as an integrated internet instrument where

respondents would visit the website, answer questions, and then submit the completed

questionnaire.

While questionnaires attached to emails are relatively simple to develop, the com-

pletion of the survey is typically not as fluent and polished as are fully integrated websurveys. Very often the emailed questionnaire is just a self-administered questionnaire

rendered in Microsoft Word and attached to the email. In other cases, the questionnaire

can be translated to an Acrobat pdf document and developed so that it can be more eas-

ily completed.

The technology for constructing internet surveys has developed rapidly

during the past few years and has produced several computer applications for develop-

ing highly functional and attractive web questionnaires. Some of these are Confirmit

(www.confirmit.com), DASH (www.dash.com), VOXCO (www.voxco.com), Sawtooth

Software (www.sawtoothsoftware.com), SurveyMonkey (www.surveymonkey.com),

Vovici (www.vovici.com), and Zoomerang (www.zoomerang.com).

52 | CHAPTER 2

EXHIBIT 2-20 Field Procedures to IncreaseResponse Rates in Mail Studies

The following procedures can coax the response rates above 50 percent and can accelerate the

response time:

The survey is of special interest to the respondents (most marketing research does notfall under this category).

There is negative reinforcement for not responding (e.g., Your name will not be included

in the membership directory unless you return the questionnaire before October 1).

There is a high-interest, positive reinforcement for responding (e.g., All those who

respond before October 1 will be eligible to enter a draw for a two-week vacation in

Acapulco).

Financial incentives, usually a dollar, are enclosed with the questionnaires.

The respondents are followed up through reminders (this is particularly effective if it is

done by telephone).


25/41

PERCEPTUAL METHODS Perceptual methods include the use ofeye-movement cam-

eras, tachistoscopes, psychogalvanometers, and related technological equipment. They

appear quantitative in their orientation as a result of all the gadgetry and readings

involved in their implementation. Yet, there is no clear-cut scientific proof, apart from

their apparent face validity, that they are effective. However, there has been a recent rein-

vigoration of eye-movement cameras for advertising research.

Perceptual methods involveobserving and measuring some

aspect of human behaviour.

Eye-movement cameras trackthe movement of respondents

pupils as they look at a docu-

ment, picture, or advertise-

ment.


EXHIBIT 2-21 Research Trends for Internet Surveys

The following is from a paper presented by John Visser at the 2003 MIRA (formerly PMRS) Annual

Conference in Vancouver and used with permission.

Canada continues to lead the world with one of the most Internet-connected populations.

Studies have confirmed that Canadians are at or near the top in the world in overall Internetadoption, online banking and music downloading. Today 68 percent of Canadian adults have

Internet access.1 Further, while the US matches Canada with respect to inciden

how to design a marketing project

Documents