12 Sampling: Design

and Procedures

Source: Lynne Shulhafer.

LEARNING OBJECTIVES

After reading this chapter, the student should be able to: 5. Explain the various probability sampling techniques and
1. Understand how a sample differs from a census, and when they are used.
the conditions that favor the use of a sample over a 6. Describe the conditions that favor the use of nonprobability
census. sampling versus probability sampling.
2. Explain the steps involved in the sampling design process. 7. Understand how online samples should be selected.
3. Describe the classification of sampling techniques and the 8. Explain how sampling techniques are used in international
difference between nonprobability and probability sampling marketing research.
techniques. 9. Describe how the representativeness of social media
Social
4. Discuss the various nonprobability sampling techniques and Media samples can be improved.
when they are used. 10. Discuss the ethical issues involved in sampling design.
 “A data.
good sampling design, carefully executed, is a key to obtaining high quality
Too often this is taken for granted, and the results can be profound.
”
David Fruend
Director and Manager, Market Research,
Progress Energy, Inc., Raleigh, North Carolina

Old Spice Red Zone Invisible Solid Spices Growth

in the Deodorant Market
After spending millions to develop invisible solid and controlled-release scent technologies, Procter & Gamble (P&G,
www.pg.com) conducted consumer research on the best way to introduce a new antiperspirant/deodorant—Old Spice
Red Zone Invisible Solid—based on these technologies. The sampling procedures played a critical role in revealing
important findings that guided the development of the new product and its introduction. The company defined the
population of interest, the target population, as male heads of households. It then developed a questionnaire to
measure preferences for solid deodorants versus other forms, such as sprays and roll-ons, and to compare the invisible
solid to conventional solids. The questionnaire was administered via computer-assisted telephone interviewing.
Given the large size of the target population and limited time and money, it was clearly not feasible to interview the
entire population, that is, to take a census. So a sample was taken, and a subgroup of the population was selected for
participation in the research. The basic unit sampled was households, and within the selected households the male heads
of households were interviewed. Probability sampling, where each element of the population has a fixed chance of being
selected, was chosen, because the results had to be generalizable to the larger population—that is, projectable to all male
heads of U.S. households. Simple random sampling was used to select 1,000 households. This sample size was selected
based on qualitative considerations, such as the importance of the decision, nature of the research, statistical analyses that
would be required, resource constraints, and the sample sizes used in similar studies P&G had conducted on new-product
development. Simple random sampling was selected, because efficient computer programs were available to randomly
generate household telephone numbers and to minimize waste due to nonexistent household telephone numbers. An
initial call was made to recruit the respondents and to explain the study. Those who qualified and agreed to participate
were mailed a sample of the new invisible solid deodorant with instructions to use it for 2 weeks. A follow-up telephone
interview was then conducted.
The results indicated that 56 percent of men preferred solid deodorants. Moreover, invisible solids were preferred to
conventional solids on a variety of attributes: gliding on without dragging, ease of application, and no white residue on the
skin or clothing. Based on these findings, P&G introduced Old Spice Red Zone Invisible Solid. The launch was successful,
and Old Spice Red Zone Invisible Solid accelerated the retail value growth of the entire antiperspirant/deodorant market.
Sound sampling procedures adopted in this research resulted in clear-cut findings that aided in the development and launch
of a successful product. As of 2011, Old Spice Red Zone Invisible Solid antiperspirant and deodorant was being marketed
in 12 scents including After Hours, Aqua Reef, Artic Force, Pure Sport, and Showtime (www.oldspice.com).1
334 PART 2 • RESEARCH DESIGN FORMULATION

Overview
Sampling design issues are a part of the research design process. By this point in the research
process, the researcher has identified the information needs of the study as well as the nature of
the research design (exploratory, descriptive, or causal) (Chapters 3 through 8). Furthermore, the
researcher has specified the scaling and measurement procedures (Chapters 9 and 10) and has
designed the questionnaire (Chapter 11). The next step is to design suitable sampling procedures.
Figure 12.1 briefly explains the focus of the chapter, the relation of this chapter to the previous
ones, and the steps of the marketing research process on which this chapter concentrates.
Five basic questions are addressed in the sample design phase: (1) Should a sample be
taken? (2) If so, what process should be followed? (3) What kind of sample should be taken?
(4) How large should it be? (5) What can be done to adjust for incidence—the rate of occurrence
of eligible respondents—and completion rates?
This chapter addresses the first three questions of the sample design phase. Chapter 13
addresses the last two questions. The chapter discusses sampling in terms of the qualitative
considerations underlying the sampling design process. The question of whether or not to
sample is addressed and the steps involved in sampling are described. Next, nonprobability and
probability sampling techniques are presented. These issues were introduced in the opening
vignette, in which P&G used a probability sampling scheme to select the respondents.
element
Sampling techniques in international marketing research are described. The chapter also
Objects that possess the information
the researcher seeks and about
discusses sampling in social media and identifies the relevant ethical issues and Internet
which the researcher will make applications. Figure 12.2 provides an overview of the topics discussed in this chapter and how
inferences. they flow from one to the next.

population
The aggregate of all the elements, Sample or Census
sharing some common set of
characteristics, that comprise the In sampling, an element is the object (or person) about which or from which the information is
universe for the purpose of the desired. In survey research, the element is usually the respondent. A population is the total of all
marketing research problem. the elements that share some common set of characteristics. Each marketing research project has

FOCUS OF THIS RELATIONSHIP TO RELATIONSHIP TO MARKETING

CHAPTER PREVIOUS CHAPTERS RESEARCH PROCESS

• Sampling Design • Research Design

Problem Definition
Process Components
(Chapter 3)

• Nonprobability
Approach to Problem
Sampling
Techniques

• Probability Research Design

Sampling
Techniques

Fieldwork

FIGURE 12.1 Data Preparation

and Analysis
Relationship of Sampling
Design to the Previous
Chapters and the
Report Preparation
Marketing Research and Presentation
Process
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 335

OPENING VIGNETTE

Sample or Census

Table 12.1

Sampling Design Process

Figures
12.3–12.5 Table 12.2

A Classification of Sampling Techniques

Figure 12.6
Be a Nonprobability Sampling Techniques
DM! Figures 12.7
and 12.8
• What
Be an Convenience Judgmental Quota Snowball Would
MR! You
• Probability Sampling Techniques
Do?
Experiential Figures Tables 12.3
12.9–12.11 and 12.4
Learning
Simple Random Systematic Stratified Cluster

Choosing Nonprobability vs. Probability Sampling

Table 12.5

Internet Sampling

Figure 12.12

APPLICATION TO CONTEMPORARY ISSUES

Figure 12.13

International Social Media Ethics

FIGURE 12.2
Sampling Design
a uniquely defined population that is described in terms of its parameters. The objective of most Procedures: An Overview
marketing research projects is to obtain information about the characteristics or parameters of a
population. The proportion of consumers in the target market that are loyal to a particular brand
of toothpaste is an example of a population parameter. In the opening vignette, the percentage of
male heads of all U.S. households that preferred a solid deodorant was a population parameter.
This parameter was estimated via sampling to be 56 percent. The researcher can obtain informa-
tion about population parameters by taking either a census or a sample. A census involves a census
complete count of each element in a population. A sample is a subgroup of the population. A complete enumeration of the
Table 12.1 summarizes the conditions favoring the use of a sample over a census. elements of a population or study
The primary considerations favoring a sample are budget and time limits. Conducting a objects.
census is costly and time-consuming. In research studies involving large populations, such as sample
users of consumer products, it is generally not feasible to take a census, as illustrated in the A subgroup of the elements of the
opening vignette. Therefore, a sample is the only viable option. However, business-to-business population selected for participation
research involving industrial products typically involves a much smaller population. A census in the study.
becomes not only possible, but often desirable, in such situations. For example, while inves-
tigating the use of certain machine tools by U.S. automobile manufacturers, a census would
be preferable to a sample, because the population of automobile manufacturers is small.
A census also becomes more attractive when there are large variations in the population.
336 PART 2 • RESEARCH DESIGN FORMULATION

TABLE 12.1 Choosing a Sample Versus a Census

Conditions Favoring the Use of
Factors Sample Census
Budget Small Large
Time available Short Long
Population size Large Small
Variance in the characteristic Small Large
Cost of sampling error Low High
Cost of nonsampling errors High Low
Nature of measurement Destructive Nondestructive
Attention to individual cases Yes No

For example, large differences in machine-tool usage from Ford to Honda would suggest the
need for taking a census instead of sampling. In this case, if a sample was taken, it is unlikely
to be representative of the population, because the machine-tool usage of automobile manu-
facturers not included in the sample is likely to be substantially different than those in the
sample. Small populations, which vary widely in terms of the characteristics of interest to
the researcher, lend themselves to a census.
The cost of sampling error (e.g., omitting a major manufacturer such as Ford from the
machine-tool study) must be weighed against nonsampling error (e.g., interviewer errors).
In many business-to-business studies, concerns involving sampling error support the use of a
census. In most other studies, however, nonsampling errors are found to be the major contributor
to total error. Although a census eliminates sampling errors, the resulting nonsampling errors
might increase to the point that total error in the study becomes unacceptably high. In these
instances, sampling would be favored over a census. This is one of the reasons the U.S. Bureau
of the Census checks the accuracy of its census by conducting sample surveys.2
Sampling also is preferable if the measurement process results in the destruction or
consumption of the product. In this case, a census would mean that a large quantity of the
product would have to be destroyed or consumed, greatly increasing the cost. An example
would be product usage tests that result in the consumption of the product, such as a new brand
of cereal. Sampling might also be necessary to focus attention on individual cases, as in the case
of depth interviews. Finally, other pragmatic considerations, such as the need to keep the study
secret (an important consideration for firms such as Coca-Cola), might favor a sample over a
census. In the opening vignette, P&G chose sampling over a census because the size of the
population of male heads of households in the United States is too large to make a census
feasible, particularly given limited time and money. Also, as is generally the case in consumer
research, the cost of sampling errors was small as compared to the cost of nonsampling errors.
As explained in Chapter 3, sampling error is the error due to the particular sample selected
being an imperfect representation of the population of interest. In contrast, nonsampling errors
result from a variety of causes, including errors in problem definition, approach, scales,
questionnaire design, survey methods, interviewing techniques, and data preparation and
analysis. Evidence shows that in consumer research the cost of sampling error is small
compared to the cost of nonsampling errors.

Be an MR! Be a DM!
Boeing: Spreading Its Wings

Search the Internet—as well as your library’s online As the CEO of Boeing, how would you use information about
databases—to determine the population of all airlines the future plans of airlines to purchase/lease airplanes to
operating in the United States. If a survey of airlines is to be formulate your marketing strategy?
conducted to determine their future plans to purchase/lease
airplanes, would you take a sample or a census? Explain.
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 337

DEFINE THE POPULATION FIGURE 12.3

Sampling Design Process

DETERMINE THE SAMPLING FRAME

SELECT SAMPLING TECHNIQUE(S)

DETERMINE THE SAMPLE SIZE

EXECUTE THE SAMPLING PROCESS

The Sampling Design Process

The sampling design process includes five steps, which are shown sequentially in Figure 12.3. Each
step is closely related to all aspects of the marketing research project, from problem definition to
presentation of the results. Therefore, sample design decisions should be integrated with all other
decisions in a research project.

Define the Target Population

Sampling design begins by specifying the target population. The target population is the target population
collection of elements or objects that possess the information the researcher is seeking. It is The collection of elements or objects
essential that the researcher precisely define the target population if the data generated are that possess the information the
to address the marketing research problem. Defining the target population involves translating researcher seeks and about which
the researcher will make inferences.
the research problem into a precise statement of who should and who should not be included
in the sample. In the opening vignette, the target population was defined as all male heads of
households in the United States.
The target population should be defined in terms of elements, sampling units, extent, and
time frame. As stated earlier, an element is the object (or person) about which or from which the
information is desired. The respondent is an example. A sampling unit might be the element sampling unit
itself, or it might be a more readily available entity containing the element. Suppose that Revlon The basic unit containing the
wanted to assess consumer response to a new line of lipsticks and wanted to sample women over elements of the population to be
18 years of age. In this study, Revlon’s element would be women over 18 years of age. It might sampled.
be possible to sample them directly, in which case the sampling unit will be the same as an
element. More typically, Revlon would use a sampling unit such as households, interviewing
one woman over age 18 in each selected household. Here, the sampling unit and the population
element are different. As another illustration, in the opening vignette, the element was male head
of household, and the sampling unit was a household.
Extent refers to geographical boundaries. In our example, Revlon is interested only in the
domestic U.S. market. The time frame is the time period of interest. Revlon may be interested in
studying lipstick demand for the upcoming summer market. This target population is defined
in Figure 12.4.

Determine the Sampling Frame

A sampling frame is a representation of the elements of the target population. It consists of a list sampling frame
or set of directions for identifying the target population. A sampling frame can come from the A representation of the elements of
telephone book, a computer program for generating telephone numbers, an association directory the target population. It consists of a
listing the firms in an industry, a mailing list purchased from a commercial organization, a city list or set of directions for identifying
the target population.
directory, or a map. If a listing is not readily available, it must be compiled. Specific instructions
for identifying the target population should be developed, such as procedures for generating
random telephone numbers of households mentioned in the opening vignette.
The process of compiling a list of population elements often is difficult and imperfect, leading
to sampling frame error. Elements may be omitted, or the list might contain more than the desired
population (see Figure 12.5).
338 PART 2 • RESEARCH DESIGN FORMULATION

FIGURE 12.4 EXTENT:

Defining the Target TIME FRAME: Domestic
Upcoming Summer United States
Population

SAMPLING UNIT: ELEMENT:

Households with females Females over 18 years old
over 18 years old

FIGURE 12.5 TARGET POPULATION: SAMPLING FRAME:

Sampling Frame Error Single-parent households List supplied by a
in Chicago commercial vendor

Sampling
Frame Error

For example, the telephone book often is used as a sampling frame for telephone surveys.
However, at least three sources of sampling frame error are present in the telephone book: (1) It does
not contain unlisted numbers; (2) it does not contain the telephone numbers of people who have
moved into the area after the telephone book was published; and (3) it lists the inactive telephone
numbers of people who have moved out of the area since the telephone book was published. Small
differences between the sample frame and the population can be ignored. In most cases, however, the
researcher should recognize and treat the sampling frame error. The researcher has three options:
1. The population can be redefined in terms of the sampling frame. When the telephone book
is used as the sampling frame, the population can be defined as households with a correct
listing in the telephone book in a given area. This approach is quite simple and eliminates
any misinterpretation of the definition of the population under study.3
2. The representativeness of the research frame can be verified during the data collection process.
Basic demographic, product familiarity, product usage, and other relevant information can be
collected to ensure that the elements of the sampling frame satisfy the criteria for the target
population. Although inappropriate elements can be identified and eliminated from the sample
in this way, this procedure does not correct for elements that have been omitted.
3. The data can be statistically adjusted by weighting under- or overrepresented segments to
achieve a more representative sample. Although sample frame error can be minimized
through this type of adjustment, it assumes the researcher has accurate knowledge of the
makeup of the target population.
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 339

SAMPLING FIGURE 12.6

TECHNIQUES Classification of
Sampling Techniques

Nonprobability Probability
Sampling Sampling
Techniques Techniques

The researcher can adopt any combination of these adjustments. The important point is
to recognize and attempt to eliminate sampling frame error, so that inappropriate population
inferences can be avoided.

Select a Sampling Technique

Selecting a sampling technique involves choosing nonprobability or probability sampling (see
Figure 12.6). Nonprobability sampling relies on the personal judgment of the researcher, rather nonprobability sampling
than chance, in selecting sample elements. The researcher might select the sample arbitrarily, Sampling techniques that do not use
based on convenience, or make a conscious decision about which elements to include in the chance selection procedures, but
sample. Examples of nonprobability sampling include interviewing people at street corners, that instead rely on the researcher’s
in retail stores, or in malls. Although nonprobability sampling produces good estimates of personal judgment and/or
convenience.
population characteristic, these techniques are limited. It is not possible to objectively evaluate
the precision of the sample results. Precision refers to the level of uncertainty about the charac- precision
teristic being measured. The greater the precision, the smaller the sampling error. Suppose the The level of uncertainty about the
researcher wanted to determine how much an average household spends on Christmas shopping characteristic being measured.
and surveyed people in the malls. Due to the convenience nature of the sample, there would be no Greater precision implies smaller
way of knowing how precise the results of this survey are. The probability of selecting one sampling error.
element over another is unknown. Therefore, the estimates obtained cannot be projected to the
population with any specified level of confidence.
In probability sampling, elements are selected by chance, that is, randomly. The probability probability sampling
of selecting each potential sample from a population can be prespecified. Although every potential A sampling procedure in which each
sample need not have the same probability of selection, it is possible to specify the probability of element of the population has a
selecting a particular sample of a given size. Confidence intervals can be calculated around the fixed probabilistic chance of being
sample estimates, and it is meaningful to statistically project the sample results to the population— selected for the sample.
that is, to draw inferences about the target population. Because projectability of the sample results
to the population of all male heads of households was important, P&G used probability rather than
nonprobability sampling in the opening vignette. We will discuss the various nonprobability and
probability sampling techniques later in this chapter.

Determine the Sample Size

Sample size refers to the number of elements to be included in the study. Determining the sample size
sample size involves both qualitative and quantitative considerations. The qualitative factors The number of elements to be
are discussed in this section; the quantitative factors are discussed in Chapter 13. Important included in a study.
qualitative factors that the researcher should consider in determining the sample size are
(1) the importance of the decision, (2) the nature of the research, (3) the number of variables,
(4) the nature of the analysis, (5) sample sizes used in similar studies, and (6) resource
constraints.
As a general rule, the more important the decision, the more precise the information must be.
This implies the need for larger samples. The need for greater precision must be weighed
against the increase in cost that comes with the collection of information from each additional
element.
The nature of the research also has an impact on the sample size. Exploratory research, such
as a focus group, employs qualitative techniques that are typically based on small samples.
Conclusive research, such as a descriptive survey, requires large samples. As the number of
variables in a study increases, the sample size must grow accordingly. For example, problem
identification surveys that measure a large number of variables typically require large samples of
1,000 to 2,500 (see Table 12.2).
340 PART 2 • RESEARCH DESIGN FORMULATION

TABLE 12.2 Samples Sizes Used in Marketing Research Studies

Type of Study Minimum Size Typical Range

Problem-identification research (e.g., market potential) 500 1,000–2,500

Problem-solving research (e.g., pricing) 200 300–500
Product tests 200 300–500
Test-marketing studies 200 300–500
TV/radio/print advertising (per commercial or ad tested) 150 200–300
Test-market audits 10 stores 10–20 stores
Focus groups 2 groups 10–15 groups

The type of analysis planned also influences the sample size requirements. Sophisticated
analysis of the data using advanced techniques or analysis at the subgroup rather than the total
sample level requires larger samples, as in the case of P&G in the opening vignette.
Prior studies can serve as a guide for estimating sample sizes. Table 12.2 gives an idea of
sample sizes used in different marketing research studies. These sample sizes have been deter-
mined based on experience and can serve as rough guidelines, particularly when nonprobability
sampling techniques are used, but they should be applied with caution. Finally, the sample size
decisions are guided by money, personnel, and time limitations. In any marketing research
project, resources are limited, in turn limiting the sample size. In the opening vignette, the
sample size of 1,000 was based on the following considerations: P&G would use the results to
make an important decision involving the introduction of invisible solid deodorants; quantitative
analysis would be conducted; and this sample size had been adequate in similar studies
conducted by P&G in the past.

Execute the Sampling Process

Execution of the sampling process refers to implementing the various details of the sample
design. The population is defined, the sampling frame is compiled, and the sampling units are
drawn using the appropriate sampling technique needed to achieve the required sample size. If
households are the sampling unit, an operational definition of a household is needed. Procedures
should be specified for vacant housing units and for callbacks in case no one is home.
Sometimes it is necessary to qualify the potential respondents to make sure they belong
to the target population. In such a case, the criteria used to qualify the respondents should be
specified, and the qualifying questions should be asked at the beginning of the interview (see
Chapter 11). In a survey of heavy users of online services at home, a criterion for identifying
heavy users is needed (e.g., those using online services for more than 30 hours per week at
home). Detailed information must be provided for all sampling design decisions. This is
illustrated by a survey done for the Florida Department of Tourism.

Research in Action
Tourism Department Telephones Birthday Boys and Girls

A telephone survey was conducted for the Florida Department of All household members meeting these four qualifications
Tourism related to the travel behavior of in-state residents. The were eligible to participate in the study. The person with the next
sampling unit was the household. Probability sampling was birthday in the household was selected. Repeated callbacks were
used. Data were collected using a stratified random sample from made to reach that person. The following are the components of
three regions: north, central, and south Florida. To participate the sampling design process:
in the study, respondents had to meet the following qualifying
1. Target population: Adults meeting the four qualifications
criteria:
(element) in a household with a working telephone number
• Age 25 years or older (sampling unit) in the state of Florida (extent) during the sur-
• Live in Florida at least 7 months of the year vey period (time)
• Have lived in Florida for at least 2 years 2. Sampling frame: Computer program for generating random
• Have a Florida driver’s license telephone numbers
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 341

3. Sampling technique: Probability sampling of computerized random-digit dialing. List all members
4. Sample size: 868 in the household who meet the four qualifications. Select
one member of the household using the next-birthday
5. Execution: Allocate the sample among the north, central,
method.4
and southern strata. Use the probability sampling technique

Experiential Learning
The New York Yankees: Targeting Families

The New York Yankees is one of America’s favorite baseball teams. assist the marketing manager of the N.Y. Yankees to target
families.
1. As the marketing manager of the N.Y. Yankees, what market-
ing programs will you design to target families? 3. The N.Y. Yankees want to conduct a telephone survey to
determine how to attract more families to the Yankees’ Social
2. Search the Internet and social media—as well as your Media
games. Design the sampling process.
library’s online databases—to obtain information that will

A Classification of Sampling Techniques

Sampling techniques can be broadly classified as nonprobability or probability (see Figure 12.6).
Commonly used nonprobability sampling techniques include convenience sampling, judgmental
sampling, quota sampling, and snowball sampling (see Figure 12.7). The important probability
sampling techniques are simple random sampling, systematic sampling, stratified sampling, and
cluster sampling.

Nonprobability Sampling Techniques

Figure 12.8 presents a graphical illustration of the various nonprobability sampling techniques.
The population consists of 25 elements; a sample size of 5 is needed. A, B, C, D, and E represent
groups and can also be viewed as strata or clusters.

Convenience Sampling
Convenience sampling, as the name implies, involves obtaining a sample of elements based convenience sampling
on the convenience of the researcher. The selection of sampling units is left primarily to the A nonprobability sampling technique
interviewer. Respondents often are selected because they happen to be in the right place at that attempts to obtain a sample of
the right time. Examples of convenience sampling are (1) use of students, church groups, and convenient elements. The selection
members of social organizations; (2) mall-intercept interviews conducted without qualifying the of sampling units is left primarily to
the interviewer.
respondents; (3) department stores using charge account lists; (4) tear-out questionnaires
included in a magazine; (5) “people on the street” interviews; and (6) Internet browsers.
Convenience sampling has the advantages of being both inexpensive and fast. Additionally,
the sampling units tend to be accessible, easy to measure, and cooperative. Despite these advan-
tages, this form of sampling has serious limitations. Primary among them is that the resulting
sample is not representative of any definable target population. This sampling process suffers
from selection bias, which means the individuals who participate in a convenience sample might
have characteristics that are systematically different than the characteristics that define the target
population. Because of these limitations, it is not theoretically meaningful to generalize to any
population from a convenience sample.

NONPROBABILITY SAMPLING TECHNIQUES FIGURE 12.7

Nonprobability Sampling
Techniques

Convenience Judgmental Quota Snowball

Sampling Sampling Sampling Sampling
342 PART 2 • RESEARCH DESIGN FORMULATION

FIGURE 12.8 1. CONVENIENCE SAMPLING

A Graphical Illustration A B C D E Group D happens to assemble at a convenient
of Nonprobability 1 6 11 16 21
time and place. So all the elements in this group
are selected. The resulting sample consists of
Sampling Techniques elements 16, 17, 18, 19, and 20.
2 7 12 17 22
NOTE: No elements are selected from groups
3 8 13 18 23 A, B, C, and E.
4 9 14 19 24
5 10 15 20 25

2. JUDGMENTAL SAMPLING
A B C D E The researcher considers groups B, C, and E to
be typical and convenient. Within each of these
1 6 11 16 21
groups one or two elements are selected based
2 7 12 17 22 on typicality and convenience. The resulting
sample consists of elements 8, 10, 11, 13, and 24.
3 8 13 18 23 NOTE: No elements are selected from groups A
4 9 14 19 24 and D.

5 10 15 20 25

3. QUOTA SAMPLING
A B C D E A quota of one element from each group, A to E,
is imposed. Within each group, one element is
1 6 11 16 21
selected based on judgment or convenience.
2 7 12 17 22 The resulting sample consists of elements 3, 6,
13, 20, and 22.
3 8 13 18 23 NOTE: One element is selected from each
4 9 14 19 24 column or group.

5 10 15 20 25

4. SNOWBALL SAMPLING
RANDOM
SELECTION REFERRALS
A B C D E Elements 2 and 9 are selected randomly from
groups A and B. Element 2 refers to elements
1 6 11 16 21
12 and 13. Element 9 refers to element 18. The
2 7 12 17 22 resulting sample consists of elements 2, 9, 12,
13, and 18.
3 8 13 18 23 NOTE: No element is from group E.
4 9 14 19 24

5 10 15 20 25

Convenience samples are not appropriate for descriptive or causal research where the aim is
to draw population inferences. Convenience samples are useful, however, in exploratory research
where the objective is to generate ideas, gain insights, or develop hypotheses. They can be used
for focus groups, pretesting questionnaires, or pilot studies. Even in these cases, caution should
be exercised in interpreting the results. Despite these limitations, this technique is sometimes
used even in large surveys, as in the example that follows.

Research in Action
Starbucks Bucks Up on Ice Cream

When Starbucks wanted to assess the brand equity behind its market for the line extensions would consist of existing
name in order to consider possible line extensions, researchers Starbucks customers.
used convenience sampling to select a group of people patron- A survey was administered that obtained information on the
izing its various coffee shops and retail outlets. Using conven- respondents’ preference and loyalty for Starbucks coffee and for other
ience sampling in this manner was justified because the core products they would like to see branded with the Starbucks name.
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 343

Results of this survey revealed a tremendous potential to extend

the Starbucks name to ice creams. Accordingly, Starbucks lever-

Source: Jordan Strauss/Stringer/WireImage.

aged the strong equity of its coffee and loyal customer base to cre-
ate a line of ice cream based on its coffee. The product is a rich,
creamy ice cream that delivers on the intense flavor of Starbucks
coffee. According to SymphonyIRI (www.symphonyiri.com) data, it
became the nation’s number-one brand of coffee ice cream
3 months after it was introduced. As of 20011, Starbucks was
selling seven coffee and noncoffee flavors of ice cream: Java Chip,
Coffee, Caramel Macchiato, Mocha Frappuccino, Signature Hot
Chocolate, Strawberry & Crème Frappuccino, and Vanilla Bean
Frappuccino.5

Judgmental Sampling
Judgmental sampling is a form of convenience sampling in which the population elements are judgmental sampling
selected based on the researcher’s judgment. The researcher chooses the sampling elements A form of convenience sampling in
because she or he believes they represent the population of interest. Common examples of which the population elements are
judgmental sampling include (1) test markets selected to determine the potential of a new purposively selected based on the
product; (2) purchase engineers selected in industrial marketing research because they are judgment of the researcher.
considered to be representative of the company; (3) bellwether precincts selected in voting-
behavior research; (4) expert witnesses used in court; and (5) department stores selected to test a
new merchandising display system.
Judgment sampling has appeal because it is low cost, convenient, and quick. It is subjec-
tive, however, relying largely on the expertise and creativity of the researcher. Therefore,
generalizations to a specific population cannot be made, usually because the population is
not defined explicitly. This sampling technique is most appropriate in research when broad
population generalizations are not required. For example, Carl’s Jr. test-marketed its Foot-
long Cheeseburger in 50 Carl’s Jr. restaurants in Southern California and 50 Hardee’s units
in Indiana. These locations were selected because the researcher felt they would provide a
good general indicator of initial response to the new product. Note that Carl’s Jr. was
only looking to evaluate the initial response to Foot-long Cheeseburger, not to make projec-
tions of how much it would sell nationally. An extension of judgment sampling involves
the use of quotas.

Quota Sampling
Quota sampling introduces two stages to the judgmental sampling process. The first stage quota sampling
consists of developing control categories, or quotas, of population elements. Using judgment A nonprobability sampling technique
to identify relevant categories such as age, sex, or race, the researcher estimates the distribu- that is a two-stage restricted
tion of these characteristics in the target population. For example, white women ages 18 to judgmental sampling. The first stage
35 years might be considered the relevant control category for a study involving cosmetic consists of developing control
categories or quotas of population
purchases. The researcher would then estimate the proportion of the target population falling
elements. In the second stage,
into this category based on past experience or secondary information sources. Sampling
sample elements are selected based
would then be done to ensure that the proportion of white women ages 18 to 35 years in on convenience or judgment.
the target population would be reflected in the sample. Quotas are used to ensure that the
composition of the sample is the same as the composition of the population with respect to
the characteristics of interest.
Once the quotas have been assigned, the second stage of the sampling process takes place.
Elements are selected using a convenience or judgment process. Considerable freedom exists in
selecting the elements to be included in the sample. The only requirement is that the elements
that are selected fit the control characteristics. This technique is illustrated in the following
example.
344 PART 2 • RESEARCH DESIGN FORMULATION

Research in Action
How Timely Is Time Magazine’s Readership?

A readership study was conducted for Time magazine using a define the makeup of the sample. Based on the composition of the
quota sample. One thousand adults living in a metropolitan area of adult population in that community, the quotas were assigned as
500,000 people were selected. Age, sex, and race were used to follows:

Population Sample
Composition Composition

Control
Characteristic Percentage Percentage Number

Sex
Male 48 48 480
Female 52 52 520
100 100 1,000
Age (Years)
18–30 27 27 270
31–45 39 39 390
46–60 16 16 160
Over 60 18 18 180
100 100 1,000
Race
White 59 59 590
Black 35 35 350
Other 6 6 60
100 100 1,000

By imposing quotas proportionate to the population reflected the composition of the metropolitan area in terms
distribution, the researcher was able to select a sample that of sex, age and race.

In the preceding example, proportionate quotas were assigned so that the composition of the
population in the community was reflected in the sample. In certain situations, however, it is
desirable either to under- or oversample elements with certain characteristics. For example,
heavy users of a product might be oversampled in order to examine their behavior in greater
detail. Although this type of sample is not representative, it might nevertheless be quite relevant.
A number of potential problems are associated with this sampling technique. Relevant
characteristics might be overlooked in the quota-setting process, resulting in a sample that does
not mirror the population on relevant control characteristics. Because the elements within each
quota are selected based on convenience or judgment, many sources of selection bias are poten-
tially present. Interviewers might be tempted to select areas where they believe they will have
success in soliciting participants. They might avoid people who look unfriendly, who are not
well dressed, or who live in undesirable locations. Quota sampling also is limited in that it does
not permit assessment of sampling error.
snowball sampling Quota sampling attempts to obtain representative samples at a relatively low cost. Quota
A nonprobability sampling technique samples are also relatively convenient to draw. With adequate controls, quota sampling obtains
in which an initial group of results close to those for conventional probability sampling.
respondents is selected randomly.
Subsequent respondents are
Snowball Sampling
selected based on the referrals or
information provided by the initial In snowball sampling, an initial group of respondents is selected, usually at random. After being
respondents. This process may be interviewed, these respondents are asked to identify others who belong to the target population
carried out in waves by obtaining of interest. This process is continued, resulting in a snowball effect, as one referral is obtained
referrals from referrals. from another. Thus, the referral process effectively produces the sampling frame from which
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 345

respondents are selected. Although this sampling technique begins with a probability sample, it
results in a nonprobability sample. This is because referred respondents tend to have demo-
graphic and psychographic characteristics that are more similar to the person referring them than
would occur by chance.
Snowball sampling is used when studying characteristics that are relatively rare or difficult to
identify in the population. For example, the names of users of some government or social services,
such as food stamps, are kept confidential. Groups with special characteristics such as widowed men
under 35 years old or members of a scattered minority population might be impossible to locate
without referrals. In industrial research, snowball sampling is used to identify buyer–seller pairs.
The major advantage of snowball sampling is that it substantially increases the likelihood of
locating the desired characteristic in the population. It also results in relatively low sampling
variance and costs. Snowball sampling is illustrated by the following example.

Research in Action
Survey Snowball

To study the demographic profile of marketing research interview- through a referral system. Each responding interviewer was asked
ers in Ohio, a sample of interviewers was generated using a varia- for the names and addresses of other interviewers. Eventually, this
tion of snowball sampling. Interviewers were initially contacted by process identified interviewers from many communities through-
placing classified advertisements in the newspapers of seven out the state who had not seen the original newspaper notices.
major metropolitan areas. These notices asked experienced mar- Only 27 percent of returned questionnaires resulted from the clas-
keting research interviewers willing to answer 25 questions about sified notices; the remainder could be traced to referrals and to
their job to write to the author. These responses were increased referrals from referrals.6

In this example, the initial groups of respondents were contacted using a nonrandom selection
technique, through classified advertisements. In this instance, the procedure was more efficient
than random selection. In other cases, random selection of respondents through probability
sampling techniques is more appropriate.

Be a DM! Be an MR!
Unisex Shirts: Sampling Gender Inequalities Affirmatively

As the vice president of marketing for Polo Ralph Lauren, Visit www.polo.com and search the Internet and social
what information would you need to determine whether the media—as well as your library’s online databases—to obtain
Social
company should launch nationally a new line of unisex shirts information on Polo Ralph Lauren’s marketing strategy. Media
it has developed? Polo Ralph Lauren would like to determine initial consumer
reaction to a new line of unisex shirts it has developed.
If nonprobability sampling is to be used, which sampling
technique would you recommend and why?

Probability Sampling Techniques

The probability sampling techniques consist of simple random, systematic, stratified, and cluster
sampling (see Figure 12.9). Probability sampling techniques vary in terms of sampling efficiency.
Sampling efficiency is a concept that reflects a trade-off between sampling cost and precision.

PROBABILITY SAMPLING TECHNIQUES FIGURE 12.9

Probability Sampling
Techniques

Simple Random Systematic Stratified Cluster

Sampling Sampling Sampling Sampling
346 PART 2 • RESEARCH DESIGN FORMULATION

Costs increase with improved precision. The trade-off comes into play as researchers balance the
need for greater precision with higher sampling costs. The efficiency of a probability sampling
technique can be assessed by comparing it to that of simple random sampling. Figure 12.10
presents a graphical illustration of the various probability sampling techniques. As in the case of
nonprobability sampling, the population consists of 25 elements, and a sample size of 5 must be
simple random sampling selected. A, B, C, D, and E represent groups and can also be viewed as strata or clusters.
A probability sampling technique in
which each element in the
Simple Random Sampling
population has a known and equal
probability of selection. Every In simple random sampling (SRS), each element in the population has a known and equal
element is selected independently of probability of selection. Furthermore, each possible sample of a given size (n) has a known and
every other element, and the sample equal probability of being the sample actually selected. The implication in a random sampling
is drawn by a random procedure procedure is that each element is selected independently of every other element. Placing the
from a sampling frame. names in a container, shaking the container, and selecting the names in a lottery-style drawing is

FIGURE 12.10
1. SIMPLE RANDOM SAMPLING
A Graphical Illustration
of Probability Sampling A B C D E Select five random numbers from 1 to 25. The
resulting sample consists of population
Techniques 1 6 11 16 21 elements 3, 7, 9, 16, and 24.
2 7 12 17 22 NOTE: There is no element from group C.
3 8 13 18 23

4 9 14 19 24

5 10 15 20 25

2. SYSTEMATIC SAMPLING
A B C D E Select a random number between 1 to 5, say 2.
The resulting sample consists of population 2,
1 6 11 16 21
(2 + 5 = 7), (2 + (5 × 2) = 12), (2 + (5 × 3) =
2 7 12 17 22 17), and (2 + (5 × 4) = 22).
NOTE: All the elements are selected from a
3 8 13 18 23 single row.
4 9 14 19 24

5 10 15 20 25

3. STRATIFIED SAMPLING
A B C D E Randomly select a number from 1 to 5 for each
stratum, A to E. The resulting sample consists of
1 6 11 16 21 population elements 4, 7, 13, 19, and 21.
2 7 12 17 22 NOTE: One element is selected from each
column.
3 8 13 18 23

4 9 14 19 24
5 10 15 20 25

4. CLUSTER SAMPLING (TWO-STAGE)

A B C D E Randomly select 3 clusters, B , D, and E. Within
each cluster, randomly select one or two
1 6 11 16 21 elements. The resulting sample consists of
2 7 12 17 22 population elements 7, 18, 20, 21, and 23.
NOTE: No elements are selected from clusters
3 8 13 18 23 A and C.
4 9 14 19 24
5 10 15 20 25
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 347

an example of a random sampling procedure. To draw a simple random sample, the sampling
frame is compiled by assigning each element a unique identification number. Random numbers,
generated using a computer routine or random number table (see Table 1 in the Appendix of
Statistical Tables), are then used to determine which element to select.
In telephone surveys, the random-digit-dialing (RDD) technique often is used to generate a
random sample of telephone numbers. RDD consists of randomly selecting all 10 digits of a
telephone number (area code, prefix or exchange, suffix). Although this approach gives all
households with telephones an approximately equal chance of being included in the sample, not
all the numbers generated in this way are working telephone numbers. Several modifications
have been proposed to identify and eliminate the nonworking numbers. This makes the use of
SRS in telephone surveys quite attractive, as illustrated in the case of P&G in the opening
vignette and the following GM example.

Research in Action
GM Refocuses on Hybrid Technology

Lagging behind the global industry leader in hybrid technology The survey found that Toyota had a 65 percent favorable
(Toyota), GM (www.gm.com) funded a study examining consumers’ rating; Honda, 64 percent; GM, 60 percent; Ford, 56 percent; and
preferences and thoughts about energy-efficient cars. The poll was con- Chrysler, 47 percent. In addition, among top priorities for
ducted by Peter D. Hart Research Associates (www.hartresearch.com) automakers, 36 percent of respondents wanted them to improve
and was carried out as a telephone survey of 1,004 adults nationwide. gas mileage and safety features, and 32 percent urged the industry
The respondents were selected using RDD software. The use of RDD to find alternatives to gas engines. Encouraged by these findings,
resulted in a sample that was representative of automobile buyers. GM introduced six hybrid models in 2010 that relied on two
The study had a margin of error of 3.1 percent. innovative technologies to efficiently power the vehicles.7

SRS has many benefits. It is easily understood and produces data that are representative of a
target population. Most statistical inference approaches assume that random sampling was used.
However, SRS suffers from at least four significant limitations: (1) Constructing a sampling
frame for SRS is difficult; (2) SRS can be expensive and time-consuming because the sampling
frame might be widely spread over a large geographical area; (3) SRS often results in lower
precision, producing samples with large standard error; and (4) samples generated by this
technique might not be representative of the target population, particularly if the sample size is
small. Although samples drawn will represent the population well on average, a given simple
random sample might grossly misrepresent the target population. For these reasons, SRS is not
widely used in marketing research. Procedures such as systematic sampling are more popular.

Systematic Sampling
In systematic sampling, the sample is chosen by selecting a random starting point and then systematic sampling
picking every ith element in succession from the sampling frame. The frequency with which the A probability sampling technique
elements are drawn, i, is called the sampling interval. It is determined by dividing the population in which the sample is chosen by
size N by the sample size n and rounding to the nearest integer. For example, suppose there are selecting a random starting point
100,000 elements in the population and a sample of 1,000 is desired. In this case, the sampling and then picking every ith element
in succession from the
interval, i, is 100. A random number between 1 and 100 is selected. If, for example, this number
sampling frame.
is 23, the sample consists of elements 23, 123, 223, 323, 423, 523, and so on.8
The population elements used in systematic sampling typically are organized in some
fashion. If the telephone book is used as the sampling frame, the elements are organized alpha-
betically. In some cases, this order might be related to some characteristic of interest to the
researcher. For example, credit-card customers might be listed in order of outstanding balance
or firms in a given industry might be ordered according to annual sales. When the population
elements are organized in a manner related to the characteristics under study, systematic
sampling can produce results quite different from SRS. Systematic sampling from a list of indus-
trial firms, organized in increasing order by sales, will produce a sample that includes small and
large firms. In comparison, a simple random sample might be less representative. For example,
only small firms or a disproportionate number of small firms might be drawn in SRS.
348 PART 2 • RESEARCH DESIGN FORMULATION

Sampling frames that are organized in a cyclical pattern tend to be less representative when
systematic sampling is used. To illustrate, consider the use of systematic sampling to generate a
sample of monthly department-store sales from a sampling frame containing monthly sales for
the last 60 years. If a sampling interval of 12 is chosen, the resulting sample would not reflect the
month-to-month variation in sales.
Systematic sampling is less costly and easier than SRS because random selection is done
only once. Systematic sampling can also be applied without knowledge of the makeup of the
sampling frame. For example, every ith person leaving a department store or mall can be inter-
cepted. For these reasons, systematic sampling often is employed in consumer mail, telephone,
mall-intercept, and Internet surveys, as illustrated by the following example.

Research in Action
Tennis Magazine’s Systematic Sampling Returns a Smash
Tennis magazine conducted a mail survey of its subscribers to
gain a better understanding of its market. A systematic sample
was drawn from the subscription list to produce a sample of
1,472 subscribers. The list was ordered according to the duration
of subscription to ensure that recent as well as long-time
subscribers would be included. If we assume that the subscriber
list had 603,520 names, the sampling interval would be 410
(603,520/1,472). A starting point was selected between 1 and 410.
Suppose this was 237. The sample would then be collected by
selecting every 410th subscriber, that is, subscribers numbered
237, 647, 1057, 1467, and so on. Systematic sampling was
chosen because a convenient sampling frame was available and
this procedure is easier to implement than SRS.
An incentive in the form of a brand-new dollar bill was

Source: Courtesy of Tennis Magazine.

included in the mail survey to boost participation in the study. The
respondents were sent an alert postcard a week before the survey.
A second, follow-up questionnaire was sent to the entire sample
10 days after the initial questionnaire. Of the 1,472 questionnaires
mailed, 76 were returned with no forwarding address, so the
net effective mailing was 1,396. Six weeks after the first mailing,
778 completed questionnaires were returned, yielding a response
rate of 56 percent.9

Be an MR! Be a DM!
Dell: Targeting Fortune 500 Companies

Visit www.fortune.com and find the list of Fortune 500 As the marketing director for Dell, how would you target the
companies. A systematic random sample of 25 firms is to Fortune 500 companies for computers, servers, and storage
be selected from the list of Fortune 500 firms, numbered from and networking equipment?
1 to 500. Determine the sampling interval, and specify the
firms that should be selected.

Stratified Sampling
stratified sampling
Stratified sampling involves a two-step sampling process, producing a probability rather than a
A probability sampling technique
that uses a two-step process to
convenience or judgment sample. First, the population is divided into subgroups called strata.
partition the population into Every population element should be assigned to one and only one stratum, and no population
subpopulations, or strata. Elements elements should be omitted. Second, elements are then randomly selected from each stratum.
are selected from each stratum by a Ideally, SRS should be used to select elements from each stratum. In practice, however, systematic
random procedure. sampling and other probability sampling procedures can be used.
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 349

A major objective of stratified sampling is to increase precision without increasing cost.

The population is partitioned using stratification variables. The strata are formed based on four criteria:
homogeneity, heterogeneity, relatedness, and cost. The following guidelines should be observed:
 Elements within strata must be similar or homogeneous.
 Elements between strata must differ or be heterogeneous.
 The stratification variables must be related to the characteristic of interest.
 The number of strata usually varies between two and six. Beyond six strata, any gain in
precision is more than offset by the increased costs.
Stratification offers two advantages. Sampling variation is reduced when the research
follows these listed criteria. Sampling costs can also be reduced when the stratification variables
are selected in a way that is easy to measure and apply. Variables commonly used for stratifica-
tion include demographic characteristics (as illustrated in the example for quota sampling), type
of customer (credit-card versus non–credit-card customer), size of firm, or type of industry.
Stratified sampling improves the precision of SRS. Therefore, it is a popular sampling technique,
as illustrated in the BMW 5 Series survey.

Research in Action
Stratifying the Success of BMW 5 Series

Extensive marketing research went into the design of the BMW 5 Control and Dynamic Brake Control). The results? The models
Series models (www.bmwusa.com). When a survey of luxury car were a roaring success. As of 2011, the new 5 Series was being
buyers was conducted to project preferences for specific luxury car marketed with an extended line consisting of five Sedan models
features, stratified random sampling was used. This procedure was and four Gran Turismo models (www.bmwusa.com).10
selected because it includes all the important subpopulations and
results in good precision. The variables chosen for stratification were
age and income, variables known to correlate with the purchase of
luxury cars. The results indicated that luxury car buyers put a pre-
mium on performance, handling, engineering, and, of course, luxury.
Based on this feedback, the new 5 Series advanced the
concept of a four-door sedan that embodies performance, easy
handling, active and passive safety engineering, and luxury. In the
Source: Lynne Shulhafer.

U.S. models, the engine was larger, delivering greater torque for
more effortless performance. For the first time, automatic climate
control was standard in 5 Series models offered in the United
States. Extensive innovations and new features were incorporated
into virtually every part of the vehicle (e.g., Dynamic Stability

Cluster Sampling cluster sampling

In cluster sampling, the target population is first divided into mutually exclusive and A two-step probability sampling
collectively exhaustive subpopulations, or clusters. Then a random sample of clusters is selected technique. First, the target
population is divided into mutually
based on a probability sampling technique, such as SRS. For each selected cluster, either all the
exclusive and collectively exhaustive
elements are included in the sample or a sample of elements is drawn probabilistically. If all
subpopulations called clusters. Then,
the elements in each selected cluster are included in the sample, the procedure is called one-stage a random sample of clusters is
cluster sampling. If a sample of elements is drawn probabilistically from each selected cluster, selected based on a probability
the procedure is two-stage cluster sampling (see Figure 12.11). Cluster sampling increases sampling technique, such as simple
sampling efficiency by decreasing cost. random sampling. For each selected
A number of key differences between cluster and stratified sampling are summarized in cluster, either all the elements are
Table 12.3. included in the sample or a sample
One common form of cluster sampling is area sampling. Area sampling relies on clustering of elements is drawn probabilistically.
based on geographic areas, such as counties, housing tracts, or blocks. Sampling can be achieved area sampling
using either one or more stages. Single-stage area sampling involves sampling all the elements A common form of cluster sampling
within a particular cluster. For example, if city blocks were used as the clusters, then all the house- in which the clusters consist of
holds within the selected blocks would be included within a single-stage sample. In the case of geographic areas, such as counties,
two-stage area sampling, only a portion of the households within each block would be sampled. housing tracts, blocks, and so forth.
350 PART 2 • RESEARCH DESIGN FORMULATION

FIGURE 12.11 DIVIDE POPULATION INTO CLUSTERS

Types of Cluster
Sampling
RANDOMLY SAMPLE CLUSTERS

ONE-STAGE TWO-STAGE

Randomly
Include All Elements
Sample Elements
from Each Selected
from Each Selected
Cluster
Cluster

TABLE 12.3 Cluster Sampling Versus Stratified Sampling

Cluster Sampling Stratified Sampling

Only a sample of the subpopulations All of the subpopulations (strata) are

(clusters) is selected for sampling. selected for sampling.
Within a cluster, elements should be Within a strata, elements should be
different (heterogeneous), whereas homogeneous with clear differences
homogeneity or similarity is maintained (heterogeneity) between the strata.
between different clusters.
A sampling frame is needed only for the A complete sampling frame for the entire
clusters selected for the sample. stratified subpopulations should be drawn.
Increases sample efficiency by Increases precision.
decreasing cost.

Cluster sampling has two major advantages: feasibility and low cost. Because sampling
frames often are available in terms of clusters rather than population elements, cluster
sampling might be the only feasible approach. Given the resources and constraints of the
research project, it might be extremely expensive, and perhaps infeasible, to compile a list of
all consumers in a population. Lists of geographical areas, telephone exchanges, and other
clusters of consumers can be constructed relatively easily. Cluster sampling is the most cost-
effective probability sampling technique. This advantage must be weighed against several
limitations. Cluster sampling produces imprecise samples in which distinct, heterogeneous
clusters are difficult to form. For example, households in a block tend to be similar rather than
dissimilar. It can be difficult to compute and interpret statistics based on clusters. The
strengths and weaknesses of cluster sampling and the other basic sampling techniques are
summarized in Table 12.4.

Be a DM! Be an MR!
Herbal Essences: Introducing New Products Is of the Essence

As the marketing chief of Herbal Essences, how would you Search the Internet—as well as your library’s online
determine which new shampoos should be introduced into databases—to determine the size of the shampoo market in
the market? the United States. Herbal Essences would like to determine
the demand for a new shampoo. If a survey is to be conducted
using probability sampling, which sampling technique
should be used and why?
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 351

TABLE 12.4 Strengths and Weaknesses of Basic Sampling Techniques

Technique Strengths Weaknesses

Nonprobability Sampling
Convenience sampling Least expensive, least Selection bias, sample not
time-consuming, most representative, not
convenient recommended for descriptive
or causal research
Judgmental sampling Low cost, convenient, not Does not allow generalization,
time-consuming subjective
Quota sampling Sample can be controlled for Selection bias, no assurance of
certain characteristics representativeness
Snowball sampling Can estimate rare characteristics Time-consuming
Probability Sampling
Simple random Easily understood, results Difficult to construct sampling
sampling(SRS) projectable frame, expensive, lower
precision, no assurance of
representativeness
Systematic sampling Can increase representativeness, Can decrease
easier to implement than SRS, representativeness
sampling frame not necessary
Stratified sampling Includes all important Difficult to select relevant
subpopulations, precision stratification variables, not
feasible to stratify on many
variables, expensive
Cluster sampling Easy to implement, Imprecise, difficult to compute
cost-effective and interpret results

Choosing Nonprobability Versus

Probability Sampling
Choosing between nonprobability and probability samples is based on considerations such as the
nature of the research, the error contributed by the sampling process relative to nonsampling
error, variability in the population, and statistical and operational considerations such as costs
and time (see Table 12.5). For example, in exploratory research, the findings are treated as
preliminary and the use of probability sampling might not be warranted. However, in conclusive
research where the researcher wishes to generalize results to the target population, as in estimating
market shares, probability sampling is favored. Probability samples allow statistical projection of
the results to a target population. For these reasons, P&G decided to use probability sampling as
described in the opening vignette.

TABLE 12.5 Choosing Nonprobability Versus Probability Sampling

Conditions Favoring the Use of

Factors Nonprobability Sampling Probability Sampling

Nature of research Exploratory Conclusive

Relative magnitude of sampling Nonsampling errors are larger Sampling errors are larger
and nonsampling errors
Variability in the population Homogeneous (low) Heterogeneous (high)
Statistical considerations Unfavorable Favorable
Operational considerations
Time Favorable Unfavorable
Cost Favorable Unfavorable
352 PART 2 • RESEARCH DESIGN FORMULATION

When high levels of sampling accuracy are required, as is the case when estimates of population
characteristics are made, probability sampling is preferred. In these situations, the researcher needs
to eliminate selection bias and calculate the effect of sampling error. To do this, probability sampling
is required. Even with this added sampling precision, probability sampling will not always result in
more accurate results. Nonsampling error, for example, cannot be controlled with probability
sampling. If nonsampling error is likely to be a problem, then nonprobability sampling techniques,
such as judgment samples, might be preferable, allowing greater control over the sampling process.
When choosing between sampling techniques, the researcher must also consider the similarity
or homogeneity of the population with respect to the characteristics of interest. For example,
probability sampling is more appropriate in highly heterogeneous populations, in which it
becomes important to draw a representative sample. Probability sampling also is preferable from
a statistical viewpoint, because it is the basis of the most common statistical techniques.
Although probability sampling has many advantages, it is sophisticated and requires statisti-
cally trained researchers. It generally costs more and takes longer than nonprobability sampling.
In many marketing research projects, it is difficult to justify the additional time and expense,
and thus operational considerations favor the use of nonprobability sampling. In practice, the
objectives of the study often exert a dominant influence on which sampling method will be used,
as illustrated in the following example.

Research in Action
The Alluring Charm of Allure Fragrance
Chanel, Inc. (www.chanel.com) wanted to launch a major new were looking for a perfume that would enable them to express
fragrance in department and specialty stores that would reach and themselves. Based on these findings, the company launched Allure, a
sustain at least the sales volume of Chanel No. 5. The target popula- unique and innovative fragrance created by Chanel perfumer Jacques
tion was defined as women 18 years and older who would appreciate Polge and packaged in a signature beige carton and sleek bottle.
a unique and broadly appealing fragrance. In conducting a survey to Diversity and individuality were successfully communicated through
determine women’s preferences for perfumes, a nonprobability multi-image ads in a breakthrough print and TV campaign. The
sampling scheme making use of quota sampling was adopted in advertising portrayed many different women, because every woman
conjunction with mall-intercept interviews. Quotas were assigned for expresses Allure in her own way. After its launch, Allure achieved a
age, income, and marital status. A nonprobability sampling scheme top-10 ranking in major retail accounts and sales volume equivalent
was chosen because the main objective of this exploratory research to Chanel No. 5. The success of Allure could be attributed in no small
was qualitative: to understand what constituted a desirable, classic, part to the carefully designed sampling procedures that resulted in
yet contemporary, fragrance that women would choose to wear. important survey findings on which the new perfume was developed
The results of the survey indicated a strong preference for a and marketed. As of 2011, Allure was being marketed as “Clean and
perfume that was fresh, clean, warm, and sexy. Moreover, women sheer, warm and sexy. Difficult to define. Impossible to resist.”11

Experiential Learning
U.S. Department of Transportation Survey

Enlightened U.S. public-policy makers seek citizen input to the 1. What was the target population in this study?
development of the country’s infrastructure. One example of this is 2. How was the sampling frame derived in this study?
the requirement for the U.S. Department of Transportation (DOT)
3. Was a nonprobability or probability sampling technique used?
to conduct an annual survey about citizen satisfaction with all
transportation modes in the country. The Bureau of Transportation 4. What were the two variables used for stratifying the sample?
Statistics (BTS) is charged with conducting this survey in order 5. How many implicit strata were employed in the design of the
to provide insights about citizen satisfaction with travel and sample?
transportation to members of Congress, to employees of the
6. The study designers wanted to secure 2,000 completed
U.S. DOT, as well as to U.S. citizens themselves.
interviews. To obtain this number of completed interviews,
Go to the textbook’s Web site and download the file U.S.DOT
how many phone numbers will need to be screened to
Survey. This is the actual record of how the BTS and Battelle
determine if the phone number is a working residential phone
(a research organization hired to assist in the survey) designed
number?
and administered a nationally representative study called the
Omnibus Survey. Read the first two pages of the Introduction and 7. What do users of the survey results gain from the BTS using
Background (pages 3 and 4), and then Appendix F: The Final this sampling design instead of using a convenience sample
Sampling Plan (pages 124–131 or F1–F8). of 2,000?
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 353

Internet Sampling
We will discuss the salient issues in online sampling as well as the commonly used online
sampling techniques.

Issues in Online Sampling

As discussed in Chapter 7, Internet surveys (and sampling) offer many advantages. The
respondents can complete the survey at their convenience. Internet surveys can incorporate
automatic skip patterns, consistency checks, and other intelligent features, as is also true for
computer-assisted surveys [computer-assisted telephone interviewing (CATI) and computer-
assisted personal interviewing (CAPI)]. The data collection can be fast and inexpensive. A major
issue related to Internet sampling is representativeness, because many U.S. households do
not have computers with Internet access. Internet access is even more restricted in foreign
countries. Furthermore, heavy users of the Internet might have a disproportionately higher
probability of being included. Unrestricted Internet samples in which any visitor can participate
are convenience samples and suffer from self-selection bias in that the respondents can initiate
their own selection.
Sampling potential respondents who are surfing the Internet is meaningful if the sample that
is generated is representative of the target population. More and more industries are meeting this
criterion. In many technical fields, it is rapidly becoming feasible to use the Internet for sampling
respondents for quantitative research, such as surveys. For internal customer surveys, where the
client’s employees share a corporate e-mail system, an intranet survey is practical even if
workers have no access to the external Internet. Sampling on the Internet is not yet practical for
many non–computer-oriented consumer products. For example, if P&G were to do a survey of
housewives to determine their preferences and usage of laundry detergents, an Internet survey
would not be a good choice, because an Internet sample is unlikely to be representative of the
target population.
To avoid sampling errors, the researcher must be able to control the pool from which the
respondents are selected. Also, steps must be taken to ensure that the respondents do not
respond multiple times (“stuff the ballot box”). These requirements are met by e-mail surveys,
for which the researcher selects specific respondents. Furthermore, the surveys can be encoded

In Internet surveys, the

researcher must be able to
control the pool from which the
respondents are selected in order
to avoid sampling errors.
Source: © Dennis MacDonald/Alamy.
354 PART 2 • RESEARCH DESIGN FORMULATION

to match the returned surveys with their corresponding outbound e-mailings. This can also be
accomplished with Web surveys by e-mailing invitations to selected respondents and asking
them to visit the Web site where the survey is posted. In this case, the survey is posted in a
hidden location on the Web, which is protected by a password. Hence, uninvited Web surfers
are unable to access it.

Online Sampling Techniques

The sampling techniques commonly used on the Internet can be classified as online intercept
(nonrandom and random), online recruited, and other techniques, as shown in Figure 12.12.
Online recruited sampling can be further classified as panel (recruited or opt-in) or nonpanel
(opt-in list rentals).
In online intercept sampling, visitors to a Web site are intercepted and given an opportunity
to participate in the survey. The interception can be made at one or more Web sites, including
high-traffic sites such as Yahoo! In nonrandom sampling, every visitor is intercepted. This can
be meaningful if the Web site traffic is low, the survey has to be completed in a short time, and
no incentive is being offered. This results in a convenience sample, however. Quotas can be
imposed to improve representativeness. In random intercept sampling, the software selects
visitors at random and a “pop-up” window asks whether the person wants to participate in the
survey. The selection can be made based on simple random or systematic random sampling.
If the population is defined as Web site visitors, then this procedure results in a probability
sample (simple random or systematic, as the case may be). If the population is other than Web
site visitors, then the resulting sample is more similar to a nonprobability sample. Nevertheless,
randomization improves representativeness and discourages multiple responses from the same
respondent.
Internet panels function in ways similar to non–Internet panels discussed in Chapters 3 and 5,
and they share many of the same advantages and disadvantages. In recruited panels, members
can be recruited online or even by traditional means (mail, telephone). Based on the researcher’s
judgment, certain qualifying criteria can be introduced to prescreen the respondents. They are
offered incentives for participation, such as sweepstake prizes, redeemable points, and other
types of Internet currencies. Members typically provide detailed psychographic, demographic,
Internet usage, and product consumption information at the time of joining. Opt-in panels
operate similarly except that members choose to opt-in as opposed to being recruited. To select a
sample, the online company sends an e-mail message to those panelists who qualify based
on sample specifications given by the researcher. All of the sampling techniques can be imple-
mented using both types of Internet panels. The success of probability sampling techniques
depends on the extent to which the panel is representative of the target population. Highly
targeted samples can be achieved (e.g., teenage girls who shop in malls more than twice a
month). For example, respondents for Harris Poll Online (HPOL) surveys are drawn from the
multimillion-member HPOL database (www.harrisinteractive.com). E-mail addresses for
respondents in the database have been obtained from a number of sources, including the HPOL
FIGURE 12.12
A Classification of
Internet Sampling
INTERNET SAMPLING

ONLINE INTERCEPT SAMPLING ONLINE RECRUITED SAMPLING OTHER TECHNIQUES

NONRANDOM RANDOM PANEL NONPANEL

Recruited Opt-in Opt-in List

Panels Panels Rentals
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 355

registration site, HPOL banner advertisements, and Epinion registrations. To maintain the
reliability and integrity in the sample, the following procedures are used:
 Password protection. Each invitation contains a password that is uniquely assigned to
that e-mail address. A respondent is required to enter the password at the beginning of the
survey to gain access into the survey. Password protection ensures that a respondent
completes the survey only one time.
 Reminder invitations. To increase the number of respondents in the survey and to
improve overall response rates, up to two additional reminder invitations are typically
e-mailed at 2- to 4-day intervals to those respondents who have not yet participated in
the survey.
 Summary of the survey findings. To increase the number of respondents in the survey and
to improve overall response rates, respondents often are provided with a summary of some
of the survey responses via the Internet.
Nonpanel recruited sampling methods can also be used, which involves requesting
potential respondents to go online to answer a survey. To illustrate, a consumer electronics
store, such as Best Buy, might hand its customers a flier that directs them to a password-
protected site to respond to a questionnaire. If the population is defined as the company’s
customers, as in a customer satisfaction survey, and a random procedure is used to select
respondents, a probability sample will be obtained. Other nonpanel approaches involve the use
of e-mail lists that have been rented from suppliers. Presumably, these respondents opted in or
gave permission for their e-mail addresses to be circulated. Offline techniques such as short
telephone-screening interviews also are used for recruiting Internet samples. Several companies
routinely collect e-mail addresses in their customer relationship databases by obtaining that
information from customer telephone interactions, product registration cards, on-site registra-
tions, special promotions, and so on.
A variety of other online sampling approaches also are possible. For example, a survey
invitation might pop up every time a visitor makes a purchase. Furthermore, the Internet can be
used to order and access samples generated by marketing research suppliers, such as Survey
Sampling International (www.surveysampling.com).

Be an MR! Be a DM!
Is Online Sampling for Amazon?

Visit www.amazon.com and search the Internet and social As the marketing chief for Amazon, what marketing strategies
media—as well as your library’s online databases—to obtain would you adopt to enhance customer satisfaction and Social
information on consumer Internet shopping behavior. loyalty? Media
Amazon wants you to conduct an Internet survey to determine
customer satisfaction. How would you select the sample?

Summary Illustration Using the Opening Vignette

In the Old Spice Red Zone Invisible Solid antiperspirant/deodorant survey, the target population
was defined as male heads of households. Given the large size of the target population and
limited time and money, P&G decided to take a sample rather than a census. The sampling unit
was households, and the elements or the respondents were male heads of households. The
sampling frame consisted of telephone numbers of all U.S. households, randomly generated by
computer programs. The researchers chose probability sampling because the results had to
be generalizable—that is, projectable to all male heads of households. The technique of
simple random sampling was selected because computer programs were available to randomly
generate household telephone numbers and to minimize wastage due to nonexistent household
telephone numbers. Simple random sampling was used to select 1,000 households. This sample
size was selected based on qualitative considerations, such as the importance of the decision,
the nature of the research, resource constraints, and the sample sizes used in similar studies
356 PART 2 • RESEARCH DESIGN FORMULATION

SAMPLING TECHNIQUES

selection based on
selection based on
convenience/judgment
chance

NONPROBABILITY SAMPLING PROBABILITY SAMPLING

sample based sample based two-stage subsequent each element has every ith a random sample of a random sample
on convenience on judgment restricted selection based a known and equal element is elements selected of clusters selected
judgmental on referrals probability of selection selected from each stratum in a based on a
sampling two-step process two-step process

CONVENIENCE JUDGMENTAL QUOTA SNOWBALL SIMPLE RANDOM SYSTEMATIC STRATIFIED CLUSTER

SAMPLING SAMPLING SAMPLING SAMPLING SAMPLING SAMPLING SAMPLING SAMPLING

FIGURE 12.13
A Concept Map for
Sampling Techniques conducted by P&G. Figure 12.13 provides a concept map for sampling techniques. In this figure,
the major concepts are connected by bold arrows using descriptors. Thus, sampling techniques in
which selection is based on judgment/convenience are called nonprobability sampling. The rest
of the figure can be interpreted similarly.

International Marketing Research

Implementing the sampling design process in international marketing research is seldom an easy
task. Several factors should be considered in defining the target population. Identification and
access to the relevant sampling elements vary widely across countries. Additionally, a reliable
sampling frame might not be available. In many studies, the researcher is interested in analyzing the
attitudes and behaviors of the household decision maker. This individual responsible for making or
influencing decisions also varies across countries. For example, in the United States, children play
an important role in the purchase of children’s cereals. In countries with authoritarian child-rearing
practices, the mother might be the dominant decision maker for that purchase. For an international
researcher, this means the mother, rather than the child, would be the element of interest.
Identification of the decision maker and the relevant respondent might have to be done on a
country-by-country basis.
Developing an appropriate sampling frame also is a difficult task. In the United States,
researchers can rely on government census data or commercially available lists for their
sampling frames. These types of high-quality secondary data, however, often are not
available in international markets. Government data in many developing countries might be
unavailable or highly biased. Population lists might not be available commercially. For
example, elections are not held in Saudi Arabia, and there is no officially recognized census
of population. Hence, there are no voter registration records or accurate maps of population
centers. In such a case, the time and money required to compile this information might be
prohibitive. Alternate sampling techniques, such as instructing the interviewers to begin at
specified starting points and to sample every nth dwelling, could be used until the specified
number of units has been sampled.
Equivalence of samples is a key issue in marketing research studies extending beyond the
home country. Researchers must minimize the effects of sample differences that are not relevant
to the main purposes of the study. For example, if American Express wanted to compare
cultural differences in consumer attitudes toward saving in the United States and Mexico,
differences in saving behavior could not be attributed to cultural differences alone if the U.S.
respondents averaged 55 years of age while the Mexican respondents averaged 30 years of age.
The life stage differs greatly for these two groups. The U.S. group would likely be saving for
retirement years, whereas the Mexican group would be saving little in order to establish their
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 357

households and to pay for the upbringing of their children. Thus, in order to compare the
effects of cultural differences, it would be necessary to hold the other variables constant across
the two countries.
Given the lack of suitable sampling frames; the inaccessibility of certain respondents,
such as women in some cultures; and the dominance of personal interviewing over other
survey methods, probability sampling techniques are uncommon in international marketing
research. These constraints have led to a reliance on quota sampling for both consumer and
industrial surveys.
Likewise, the use of Internet sampling in international marketing research must consider the
possibility that Internet availability and use vary markedly across countries. Many countries lack
servers, hardware, and software, and this lack of technical infrastructure makes Internet research
difficult and expensive. In addition, there may be cultural differences in attitude toward and usage
of the Internet. For example, in Latin America the attitude toward Internet research is not positive,
due to the respondents missing social interaction with others while participating in the research.
Sampling techniques and procedures vary in accuracy, reliability, and cost from country
to country. If the same sampling procedures are used in each country, the results might not be
comparable.12 To achieve comparability in sample composition and representativeness, it might
be desirable to use different sampling techniques in different countries, as the following example
illustrates.

Research in Action
Achieving Sample Comparability Through Diversity

Research in the United States found that most consumers feel that social customs prohibited spontaneous personal interviews.
a purchase is accompanied by a degree of risk when they choose Comparability in sample composition and representativeness was
among alternative brands. A study was conducted to compare the achieved by adapting the sampling procedures to suit the social
U.S. results with those from Mexico, Thailand, and Saudi Arabia. customs and information restrictions of each country.13
The targeted respondent in each culture was an upper–middle-
income woman residing in a major city. Reaching that target
respondent presented challenges that required an adjustment in
the sampling process across the countries in this study.
In the United States, random sampling from the telephone
directory was used. In Mexico, judgmental sampling was used by
Source: © Novel Creative/Alamy.

having experts identify neighborhoods where the target respondents

lived. Homes were then randomly selected for personal interviews.
In Thailand, judgmental sampling also was used, but the survey took
place in major urban centers, and a store-intercept technique was
used to select respondents. Finally, in Saudi Arabia, convenience
sampling employing the snowball procedure was used, because
there were no lists from which sampling frames could be drawn and

Be a DM! Be an MR!
PC Usage in USA and India

As the global marketing manager for Intel, the computer Search the Internet and social media—as well as your
microprocessor manufacturer, would you adopt the same library’s online databases—to obtain information on the Social
or different marketing strategies in the United States and usage or nonusage of personal computers by households in Media
India? the United States and India. Intel wants to conduct cross-
cultural research to determine the similarities and differences
in the usage or nonusage of personal computers by house-
holds in the United States and India. Should the same or
different sampling techniques be used in the two countries,
and which technique(s) should be used?
358 PART 2 • RESEARCH DESIGN FORMULATION

Social
Marketing Research and Social Media
Media General social media content available in the public domain may not be representative or even
appropriate in all cases. The sampling frame is biased and limited in that only consumers who
are online and engaged in social media are represented. Even so, consumers who are heavy users
of social media have a greater probability of being sampled. Yet, there are ways in which one
could improve the representativeness of information gleaned from general social media analysis
and monitoring.
 Instead of targeting an entire site, select sections of sites that suit the brand’s profile.
Careful screening can result in a more targeted and representative sample.
 Narrow your search results by designing search queries that mine social media content with
consumer-, category-, or brand-related terms.
 Use text analysis that detects age, gender, geography, or other characteristics that distinguish
different types of voices and then filter the results to more accurately reflect your target
population. This information may be obtained from publicly available social media profile
information of the individuals posting comments.

Research in Action
Simple Random Sampling Results in Sophisticated Market Segmentation

Umbria, now a part of J.D. Power and Associates (www. the selected comments to arrive at six segments: Fit Finders,
jdpower.com), used its text-mining technology to segment the Self-Expressives, Bargain Seekers, Label Whores, Style Gurus,
market for an apparel client. A large number of comments and Dissenters. These labels were quite descriptive of the
pertaining to apparel were first collected from a variety of social segments. For example, Fit Finders are Generation Xers who want
media sites. Simple random sampling (SRS) was then used to jeans that complement their changing physiques. Self-Expressives
select a manageable but representative subset. SRS was chosen want to customize their jeans by distressing them or adding
because the sampling frame was readily available and the results patches and embroidery. The client used these findings to design
were projectable. The selection of a representative sample facili- and market products tailor made for each segment, resulting in a
tated a more detailed analysis than might have been possible on sophisticated segmentation scheme and a successful marketing
all the comments collected. Text-mining was used to analyze strategy.14

Ethics in Marketing Research

The researcher has the ethical responsibility to formulate a sampling design that is consistent
with the objectives of the research and adequately controls the sampling error. Probability sam-
pling techniques should be used whenever the results are to be projected to the population.
Projecting findings based on nonprobability sampling to the population is misleading and there-
fore unethical. As the following example demonstrates, appropriate definitions of the population,
sampling frame, and sampling technique are essential if the research is to be conducted and the
findings are to be used ethically.

Research in Action
Systematic Sampling Reveals Systematic Gender Differences in Ethical Judgments

In an attempt to explore differences in research ethics judgments Respondents were asked to indicate the degree of ethicalness of
between male and female marketing professionals, data were various ethical scenarios and promised a copy of the results in an
obtained from 420 respondents. The population was defined as attempt to boost response rates.
marketing professionals (elements), who also comprised the The study revealed that female marketing professionals, in
sampling unit. The sampling frame was the American Marketing general, demonstrated higher levels of research ethical judgments
Association directory, and the extent was the United States. The than did their male counterparts. Similar findings have been
respondents were selected based on a systematic sampling plan obtained in other studies. Thus, the systematic random sampling
from the directory. Systematic random sampling was chosen over plan was not only efficient, but resulted in findings that were also
simple random sampling because it was easier to implement. generalizable to the target population.15
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 359

When conducting research with small populations, as in business-to-business marketing or

employee research, researchers must be sensitive to preserving the respondents’ anonymity.
The results should be presented in such a way that respondents cannot be identified or linked to
specific comments and findings.

WHAT WOULD YOU DO ?

MTV: The World’s Most Widely Distributed TV Network
The Situation 2. Discuss how the marketing management-decision action that you recommend to
Judy McGrath is influenced by the sampling technique suggested earlier and by
As of 2011, MTV (www.mtv.com) sites included favorites such as MTV, MTV Dance,
the findings of that research.
MTV Tr3s, MTV2, MTVU, and other networks around the world. MTV uses a free-
flowing corporate culture and a group of power brands to earn big profits for parent
Viacom (www.viacom.com). What Judy McGrath Did
Tom Freston, CEO of MTV, actually took MTV as a small start-up company and MTV launched music hours shown live from Times Square. New shows, such as the
turned it into a cultural behemoth. Judy McGrath was named chair and CEO of MTV animated Daria, were created, and the Video Music Awards were revamped. VeeJays
Networks in July 2004. Although MTV has become a global icon, with millions of became more authentic, and not as TV-pretty. The objective was to give MTV a cleaner,
viewers all over the world glued to its programming, MTV is constantly trying to keep more pensive image to keep its viewers happy. The 18-to-24-year-old crowd is
up with what is popular and keep its viewers watching. important to the network, because younger teens and the 25-to-35-year-old crowd
MTV is always trying to find new ways to engage its target audience, which is looks to it for image and style ideas.
18- to 24-year-olds. This has not always been an easy task. For example, the MTV is still striving to make its programming fresh and hip. TV shows such as
channel had fading hits with shows like Beavis & Butt-head and The Real World. The Osbournes and Punk’d have gotten great reviews. Also, on MTV’s detailed
Ratings began to slip as users complained that there was no longer music on MTV. Web site (www.mtv.com) viewers can do everything from requesting music
Telephone surveys were conducted with the target audience. The sample was videos to chatting with fellow MTV fans. On July 8, 2010, MTV Networks acquired
selected by choosing households based on computerized random-digit dialing. Social Express, a social gaming development company, marking the company’s
If the household had multiple 18- to 24-year-olds, one person was selected using first entry into the social gaming space. MTV is continuing to grow and expand,
the next-birthday method. If no person between these ages resided in the house, a and with the help of marketing research it should be able to stay hip for
new random number was selected. The results of this survey showed that MTV generations to come.16
needed a makeover.

The Marketing Research Decision

1. As trends come and go, it is important for MTV to stay in touch with its audience
members and know their wants. Judy McGrath especially wants to keep in touch
with the 18- to 24-year-olds through periodic surveys. What sampling technique
would you recommend?
a. Convenience sampling
b. Quota sampling
c. Systematic sampling
d. Stratified sampling
e. Simple random sampling
2. Discuss the role of the type of sampling technique you recommend in enabling
Judy McGrath to keep in touch with the 18- to 24-year-olds.

The Marketing Management Decision

Source: Paul Chappells/Newscom.

1. What should Judy McGrath do to attract the 18- to 24-year-old audience group
to MTV?
a. Launch music hours shown live from popular places (e.g., Times Square)
b. Make VeeJays more authentic
c. Feature more celebrities
d. Focus more on music programming
e. All of the above
360 PART 2 • RESEARCH DESIGN FORMULATION

Summary
Researchers can obtain information about the characteristics of a population by conducting
either a sample or a census. Samples tend to be preferred because of budget and time limits, large
population sizes, and small variances in the characteristics of interest. Sampling also is preferred
when the cost of sampling error is low, the cost of nonsampling error is high, the nature of
measurement is destructive, and attention must be focused on individual cases. The opposite
conditions favor the use of a census.
Sampling design begins by defining the target population in terms of elements, sampling
units, extent, and time. Then the sampling frame should be determined. A sampling frame is a
list of the elements of the target population. Directions for constructing the sampling frame
should also be included. Sampling techniques are applied to the sample frame to develop the
eventual sample. The sample size is determined based on both quantitative and qualitative
considerations. Finally, execution of the sampling process requires detailed specifications for
each step in the process.
Sampling techniques can be classified as nonprobability and probability techniques.
Nonprobability sampling techniques rely on the researcher’s convenience and/or judgment.
Consequently, they do not permit an objective evaluation of the precision of the sample results,
and the estimates cannot be projected to the larger population. The commonly used nonprobability
sampling techniques include convenience sampling, judgmental sampling, quota sampling, and
snowball sampling.
In probability sampling techniques, sampling units are selected by chance. Each sampling
unit has a nonzero chance of being selected, and the researcher can prespecify every potential
sample of a given size that could be drawn from the population, as well as the probability of
selecting each sample. It also is possible to determine the precision of the sample estimates and
inferences and make projections to the target population. Probability sampling techniques
include simple random sampling, systematic sampling, stratified sampling, and cluster sampling.
The choice between probability and nonprobability sampling should be based on the nature of
the research, degree of error tolerance, relative magnitude of sampling and nonsampling errors,
variability in the population, and statistical and operational considerations. Nonprobability
as well as probability sampling techniques can be implemented on the Internet, with the respon-
dents being prerecruited or tapped online.
When conducting international marketing research, it is desirable to achieve comparability
in sample composition and representativeness even though this might require using different
sampling techniques in different countries. There are several ways to improve the representative-
ness of social media samples. It is unethical and misleading to treat nonprobability samples as
probability samples and project the results to a target population.

Key Terms and Concepts

Element 334 Nonprobability sampling 339 Snowball sampling 344
Population 334 Precision 339 Simple random sampling 346
Census 335 Probability sampling 339 Systematic sampling 347
Sample 335 Sample size 339 Stratified sampling 348
Target population 337 Convenience sampling 341 Cluster sampling 349
Sampling unit 337 Judgmental sampling 343 Area sampling 349
Sampling frame 337 Quota sampling 343
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 361

Suggested Cases and Video Cases

Running Case with Real Data
1.1 Hewlett-Packard
Comprehensive Critical Thinking Cases
2.1 American Idol 2.2 Baskin-Robbins 2.3 Akron Children’s Hospital
Comprehensive Cases with Real Data
3.1 Bank of America 3.2 McDonald’s 3.3 Boeing
Comprehensive Brief Harvard Business School Cases
4.1 TruEarth Healthy Foods: Market Research for a New Product Introduction (4065)
4.2 The Springfield Nor’easters: Maximizing Revenues in the Minor Leagues (2510)
4.3 Metabical: Positioning and Communications Strategy for a New Weight-Loss Drug (4240)
4.4 Saxonville Sausage Company (2085)
Online Video Cases
12.1 Motorola 13.1 Subaru 14.1 Intel 19.1 Marriott

Live Research: Conducting a Marketing Research Project

1. A census might be feasible in a business-to-business 3. Probability sampling techniques are more difficult and
project where the size of the population is small, but it is time-consuming to implement, and their use might not be
infeasible in most consumer projects. warranted unless the results are being projected to a
2. Define the target population (element, sampling unit, population of interest.
extent, and time), and discuss a suitable sampling frame.

Acronym
The sampling design process and the steps involved may be represented by the acronym SAMPLE:
S ampling design process
A mount: sample size determination
M ethod: sampling technique selection
P opulation definition
L ist: sampling frame determination
E xecution of the sampling process

Review Questions
1. What is the major difference between a sample and a 3. Describe the sampling design process.
census? 4. How should the target population be defined?
2. Under what conditions is a sample preferable to a census? 5. What is a sampling unit? How is it different from the
When is a census preferable to a sample? population element?
362 PART 2 • RESEARCH DESIGN FORMULATION

6. What qualitative factors should be considered in 12. Describe the procedure for selecting a systematic random
determining the sample size? sample.
7. How do probability sampling techniques differ from 13. Describe stratified sampling. What are the criteria for the
nonprobability sampling techniques? selection of stratification variables?
8. What is the least expensive and least time-consuming of 14. Describe the cluster sampling procedure. What is the
all sampling techniques? What are the major limitations key distinction between cluster sampling and stratified
of this technique? sampling?
9. What is the major difference between judgmental and 15. What factors should be considered in choosing between
convenience sampling? probability and nonprobability sampling?
10. What is the relationship between quota sampling and 16. Social How can the representativeness of social media
judgmental sampling? Media samples be improved?

11. What are the distinguishing features of simple random

sampling?

Applied Problems
1. Define the appropriate target population and the sampling a. Identify the population and sampling frame that could
frame in each of the following situations: be used.
a. The manufacturer of a new cereal brand wants to b. Describe how a simple random sample can be drawn
conduct in-home product usage tests in Chicago. using the identified sampling frame.
b. A national chain store wants to determine the c. Could a stratified sample be used? If so, how?
shopping behavior of customers who have its store d. Could a cluster sample be used? If so, how?
charge card. e. Which sampling technique would you
c. A local TV station wants to determine households’ recommend? Why?
viewing habits and programming preferences. 3. Ben & Jerry’s Homemade Ice Cream would like to
d. The local chapter of the American Marketing conduct a survey of consumer preferences for premium
Association wants to test the effectiveness of its ice cream brands in Los Angeles. Stratified random
new-member drive in Atlanta. sampling will be used. Visit www.census.gov to identify
2. A manufacturer would like to survey users to information that will be relevant in determining income
determine the demand potential for a new power press. and age strata.
The new press has a capacity of 500 tons and costs 4. Using a computer program, generate a set of 1,000
$225,000. It is used for forming products from light- random numbers for selecting a simple random sample.
and heavyweight steel and can be used by automobile, 5. Visit Gallup (www.gallup.com). Examine the Internet
construction equipment, and major appliance surveys being conducted. Write a report about the
manufacturers. sampling plans being used.

Group Discussion
1. “Given that the U.S. Bureau of the Census uses sampling 2. “Because nonsampling errors are greater in
to check on the accuracy of various censuses, a magnitude than sampling errors, it really does not matter
constitutional amendment should be passed replacing which sampling technique is used.” Discuss this
the decennial census with a sample.” Discuss as a statement.
small group.
 CHAPTER 12 • SAMPLING: DESIGN AND PROCEDURES 363

Hewlett-Packard Running Case

Review the Hewlett-Packard (HP) case, Case 1.1, and question- 3. HP wants to conduct a telephone survey to determine how
naire given toward the end of the book. it can attract more families to HP PCs and notebooks.
Design the sampling process.
1. As the marketing manager of HP personal computers, what
4. Social Discuss the advantages and disadvantages of HP
marketing programs will you design to target families? Media using social media samples to obtain information
2. Social Search the Internet and social media—as well as
Media your library’s online databases—to obtain
on consumers’ preferences for notebook
computers.
information that will assist you in targeting families.

ONLINE VIDEO CASE 12.1

MOTOROLA: Projecting the Moto Lifestyle
Visit www.pearsonhighered.com/malhotra to read the video case and view the accompanying video. Motorola: Projecting the Moto
Lifestyle highlights the use of marketing research findings to create the Moto campaign, which established Motorola as a chic and aspirational
brand that helped it overcome its poor consumer image and branding problems. The case can be used to evaluate the mall-intercept interviewing
conducted by Motorola and to design a sampling plan for such survey research. Specific marketing research questions on this and the previous
chapters are posed in the video case.
 Naresh K. Malhotra, Basic Marketing Research: Integration of Social Media, 4e. Copyright © 2012 Pearson Education, Inc. Publishing as Prentice Hall.
ONLINE VIDEO CASE 12.1
MOTOROLA: Projecting the Moto Lifestyle
Starting in 1928 as the Galvin Manufacturing Corporation, Motorola (www.motorola.com) hands-free driving, in October 2007 the company introduced MOTOROKR T505
has evolved into a worldwide company with more than $22.04 billion in revenue in 2009. Bluetooth, the In-Car Speakerphone and Digital FM Transmitter, its first road-ready,
Today, it is a leading manufacturer and provider of wireless, semiconductor, broadband, music-oriented ROKR accessory and the latest addition to its portfolio of in-car
and automotive products and services. With the wireless division, Motorola knew it solutions.
needed to change. It had found through focus groups and survey research that many Given the high costs associated with an advertising campaign, it was well
customers and potential customers saw Motorola’s phone models as dependable, researched and backed with strong supporting evidence and data. Marketing profes-
but also as dull, predictable, and boring. With the mobile phone market being flashy and sionals need to substantiate their spending on advertising and brand building with
consumer driven, Motorola needed answers on how to become more mainstream research data that spells out the rationale for that spending. The Moto campaign,
and popular. instead of inducing customers to buy Motorola phones because of their features,
To find these answers, Motorola turned to marketing research and an adver- appealed to consumers’ lifestyle choices. The campaign positioned Motorola
tising agency named Ogilvy and Mather. Motorola and Ogilvy and Mather conducted phones as aspirational products that embodied a certain attitude. This positioning
focus groups, depth interviews, and mall-intercept surveys. Although focus groups created an emotional connection with the consumers and targeted people’s desires
generated some innovative ideas, depth interviews enabled the probing of emotions to be associated with products that stand for qualities that they consider to be
related to mobile phones. Mall intercepts were chosen because the respondents “cool”—that is, fashionable and worthy of being identified with.
could be shown models of Motorola and competing brands. They found from this Motorola sold its wireless equipment unit to Nokia Siemens in July 2010.
research that customers buying mobile phones did not buy the phone based on Motorola planned to split into two separate publicly traded companies in the first
technical schematic selling points. Customers buy phones based on how they quarter of 2011, with each retaining the Motorola name. The company would spin
emotionally feel about the brand of phone and the particular style of the phone. Most off Motorola Mobility, its mobile device and set-top box business. That would leave
customers do not understand the technical parame- only Motorola Solutions, which will focus on government and commercial clients
ters of the different phone models enough to make a with products such as police radios and bar-code scanners, as well as the division
decision based on them. So they are choosing that makes iDEN equipment. This separation would enable Motorola Mobility to
among cell phones based on whether the phone focus on its digital entertainment and mobile converged devices, providing instant
MOTOROLA “fits” into their lives or by considering “Is this phone access to television, phone, and Internet.
me?” This research challenged the company’s
management to think of cell phones not so much Conclusion
as engineered functional devices but as fashion
Based on marketing research findings, the Moto campaign established Motorola as
accessories that help consumers make statements
a chic and aspirational brand that helped it overcome its poor consumer image and
about who they are. It pointed Motorola into developing a marketing strategy that
branding problems. The Moto campaign projects Motorola’s core values and its
developed the brand name instead of pushing the features of the phone. Moreover,
lifestyle appeals to consumers across the world. Possessing a Moto was no longer
the brand name had to be a global one based on universal principles. Marketing
possessing a cell phone, but having a product whose core values represented the
research also revealed that consumers were looking for “intelligence everywhere,”
type of lifestyle the user of the phone desired and lived. With this dependence on
and therefore the brand had to be developed in that environment.
marketing research in the forefront of Motorola’s actions, it is certain to remain a
Ogilvy and Mather sought to develop the Motorola brand to represent a set of
global contender in the mobile market for years to come.
universal principles—a set of core principles that defined the brand—and then
send out this idea to every country and have a localized interpretation for the idea.
The result of this is that the core ethos of the brand is preserved while at the same Questions
time offering local offices the flexibility to mould the brand according to local 1. Discuss the role that marketing research can play in helping Motorola further
conditions and develop the brand such that the people of that country can relate to build the Moto brand.
and identify with it. For Motorola, Ogilvy developed the core idea of “intelligence 2. Management would like to continue rebuilding Motorola. They feel this can
everywhere.” This core idea is used as the framework for all Motorola businesses be best accomplished by increasing Motorola’s U.S. marketing share. Define
around the world. the management-decision problem.
This was accomplished by creating the Moto, which is a cute name for 3. Define an appropriate marketing research problem based on the
Motorola’s global-branded cell phone. The name is easy to pronounce, and it does management-decision problem you have identified.
not mean anything bad or weird anywhere in the world. It also carries a part of the 4. Use the Internet to determine the market shares of the major cell-phone
Motorola name, a strong positive brand name that reminds consumers of the handset manufacturers (Nokia, Motorola, Samsung, Sony/Ericsson, etc.) for
company’s heritage. Motorola’s advertising agency, Ogilvy and Mather, created a the last calendar year.
Moto lifestyle image. Knowing from research that customers wanted to relate to the 5. What type of syndicate data will be useful to Motorola?
phone and brand on a personal level, Motorola’s Moto lifestyle showed the public 6. Discuss the role of qualitative research in helping Motorola expand its
fast, upbeat, and flashy people living and using Motorola products in an intelligent market share.
way. This created an emotional connection with customers, as they were almost 7. Do you think that the mall-intercept interviewing conducted by Motorola was
saying to themselves, “That’s the way I want to live.” This was followed by the best method of administering the survey? Why or why not?
Motodextrous ads in 2004 that projected a perfect balance between design and 8. Discuss the role of experimentation in helping Motorola design handsets
technology to enable people to live the Moto lifestyle with the slogan of “Intelligence that are preferred by consumers.
Everywhere.” This campaign, an obvious success, was possible due only to the 9. Develop a questionnaire for assessing consumer preferences for cellular
marketing research conducted to find why customers buy certain brands and handsets.
models of mobile phones. When marketing research indicated a big need for 10. What sampling plan should be adopted for the survey of question 7?
11. What are the advantages and disadvantages of Motorola sampling References
Social
Media social media to determine consumer preferences for cellular
1. www.motorola.com, accessed February 15, 2011.
handsets?
2. www.hoovers.com, accessed February 15, 2011.
12. If Motorola were to conduct marketing research to determine consumer
3. http://www.wikinvest.com/wiki/Motorola_(MOT), accessed October 27, 2010.
preferences for cellular handset manufacturers in Asia, how would the
4. www.technologyreview.com/read_article.aspx?id=17663, accessed June 4, 2007.
research process be different?
5. Soo Youn, “Motorola Chips Away at Nokia’s Lead in Cell-Phone Market,” Knight
13. Discuss the ethical issues involved in researching consumer preferences for
Ridder Tribune Business News, June 9, 2004, p. 1.
cellular handset manufacturers.