LESSON 9: SELECTING THE SAMPLE

International markets are measured in hundreds of millions of people, national markets comprise millions of
individuals, and even local markets may constitute hundreds of thousands of households. To obtain information from
every single person in a market is usually impossible and obviously impractical. For these reasons, marketing
researchers make use of a sample.

BASIC TERMS

 Population is defined as the entire group under

study as specified by the objectives of the
research project.
 Census is defined as an accounting of the
complete population.
 SAMPLE AND SAMPLE UNIT
 Sample - is a subset of the population that
suitably represents that entire group.
 Sample unit is the basic unit of
investigation
 Sample frame is a master source of sample units
in the population.

Sampling error is any error in a survey that occurs because a sample is used. Sampling error is caused by two factors.
First, there is the method of sample selection, which, as you will learn later in this chapter, includes sample frame
error. In Figure 9. 1 , the sample shape is inside the sample frame shape, but it includes some area outside the
population shape. This type of sample error results when the sample frame is not completely faithful to the
population definition.

Reasons for Taking a Sample

1. There are practical considerations, such as cost and population size, that make a sample more desirable than a
census.
2. Typical research firms or the typical researcher cannot analyze the huge amounts of data generated by a census.

PROBABILTY AND NONPROBABILITY SAMPLING

Probability samples
 are samples in which members of the population have a known chance (probability) of being selected into
the sample.
 The essence of a “known” probability rests in the sampling method rather than in knowing the exact size of
the population.
 Probability sampling methods are those that ensure that, if the exact size of the population were known for
the moment in time that sampling took place, the probability of any member of the population being
selected into the sample could be calculated. In other words, this probability value is really never calculated
in actuality, but we are assured by the sample method that the chances of any one population member
being selected into the sample could be computed.

Nonprobability Samples
 are samples where the chances (probability) of selecting members from the population into the sample are
unknown.
 With nonprobability methods there is no way to determine the probability even if the population size is
known because the selection technique is subjective. As one author has described the difference,
nonprobability sampling uses human intervention, whereas probability sampling does not.
 this is sometimes called “haphazard sampling” because it is prone to human error and even subconscious
biases.
TYPES OF PROBABILITY SAMPLING

1. Simple Random Sampling. With simple random sampling , the probability of being selected into the sample is equal
for all members of the population. This probability is expressed by the following formula:

Probability of selection = sample size/population size

With simple random sampling, if the researcher is surveying a population of 100,000 recent Apple Watch buyers with
a sample size of 1,000 respondents, the probability of selection of any single population member into this sample
would be 1,000 divided by 100,000, or 1 out of 100, calculated to be 1%.
A. Random device method - involves using an apparatus or a procedure to ensure that every member of the
population has the same chance of being selected into the sample. Familiar examples of the random
device method include the flipping of a coin to decide heads or tails, lottery numbers selected by
numbered balls, a roulette wheel in a casino, and a hand dealt in a poker game.

B. Random Numbers Method All of our random device examples involve small populations that are easily
accommodated by the physical aspects of the device. With large populations, random devices become
cumbersome (just try shuffling a deck of 1,000 playing cards). A tractable and more sophisticated
application of simple random sampling is to use computer-generated numbers based on the concept of
random numbers , which are numbers whose chance nature is assured. Computer programs can be
designed to generate numbers without any systematic sequence to the numbers whatsoever—that is,
they are random. A computer easily handles datasets of hundreds of thousands of individuals; it can
quickly label each one with a unique number or designation, generate a set of random numbers, and
match the random numbers with the unique designations of the individuals in the dataset to select or
“pull” the sample.

2. Systematic Sampling
Using a sample frame that lists members of the population, the researcher selects a random starting point
for the first sample member. A constant skip interval , calculated by dividing the number of population members in
the sample frame by the sample size, is then used to select every other sample member from the sample frame. This
procedure accomplishes the same end as simple random sampling, and it is more efficient.
Skip interval = population list size/sample size
Systematic sampling is probability sampling because it employs a random starting point, which ensures there
is sufficient randomness in the systematic sample to approximate an equal probability of any member of the
population being selected into the sample. In essence, systematic sampling envisions the list as made up of the skip
interval number of mutually exclusive samples, each one of which is representative of the listed population. The
random starting point guarantees that the selected sample is selected randomly.

3. Cluster Sampling
 The sample frame is divided into groups called clusters, each of which must be highly similar to the others.
The researcher can then randomly select a few clusters and perform a census of each one (one stage). Alternatively,
the researcher can randomly select more clusters and take samples from each one (two stage). This method is
desirable when highly similar clusters can be easily identified, such as subdivisions spread across a wide geographic
area.
Area sampling (As a form of Cluster Sampling)- the researcher subdivides the population to be surveyed into
geographic areas, such as census tracts, cities, neighborhoods, or any other convenient and identifiable geographic
designation. The researcher has two options at this point: a one-step approach or a two-step approach.
In the one-step area sample approach, the researcher may believe the various geographic areas (clusters) to
be sufficiently identical to allow concentrating his or her attention on just one area and then generalizing the results
to the full population. But the researcher would need to select that one area randomly and perform a census of its
members.
In the two-step area sample approach, the first step, the researcher could select a random sample of areas,
and then for the second step, he or she could decide on a probability method to sample individuals within the chosen
areas. The two-step area sample approach is preferable to the one-step approach because there is always the
possibility that a single cluster may be less representative than the researcher believes. But the two-step method is
more costly because more areas and time are involved.
Area grid sampling is a variation of the area sampling method. To use it, the researcher imposes a grid over
a map of the area to be surveyed. Each cell within the grid then becomes a cluster. The difference between area grid
sampling and area sampling lies primarily in the use of a grid framework, which cuts across natural or artificial
boundaries, such as streets, rivers, city limits, or other separations normally used in area sampling.

4. Stratified Sampling
If the population is believed to have a skewed distribution for one or more of its distinguishing factors (e.g.,
income or product usage), the researcher identifies subpopulations in the sample frame called strata . A simple
random sample is then taken of each stratum. Weighting procedures may be applied to estimate population values,
such as the mean. This approach is better suited than other probability sampling methods for populations that are not
distributed in a bell-shaped pattern.
A Skewed Population deviates greatly from the bell-shaped distribution that is assumed to be the case in the
use of simple random, systematic, or cluster sampling. If any of these methods is used to draw the sample from a
skewed distribution, it most certainly would be inaccurate.
With stratified random sampling, one takes a skewed population and identifies the subgroups, or strata ,
contained within it. Simple random sampling, systematic sampling, or some other type of probability sampling
procedure is then applied to draw a sample from each stratum because we typically believe that the individual strata
have bell-shaped distributions. In other words, it is a “divide and conquer” approach to sampling.

TYPES OF NONPROBABILITY SAMPLING

1. Convenience Sampling
 The researcher or interviewer uses a high-traffic location, such as a busy pedestrian area or a shopping
mall, as the sample frame from which to intercept potential respondents. Sample frame error occurs in the
form of members of the population who are infrequent users or nonusers of that location. Other error may
result from any arbitrary way the interviewer selects respondents from the sample frame.
2. Purposive Sampling
 The researcher uses his or her judgment or that of some other knowledgeable person to identify who will
be in the sample. Subjectivity and convenience enter in here; consequently, certain members of the
population will have a smaller chance of selection than will others.
3. Chain Referral Sampling
 Sometimes called “snowball samples”
 Respondents are asked for the names or identities of others like themselves who might qualify to take part
in the survey. Members of the population who are less well known or disliked or whose opinions conflict
with the selected respondents have a low probability of being selected.
4. Quota Sampling
 The researcher identifies quota characteristics, such as demographic or product use factors, and uses these
to set up quotas for each class of respondent. The sizes of the quotas are determined by the researcher’s
belief about the relative size of each class of respondent in the population. Often, quota sampling is used as
a means of ensuring that convenience samples will have the desired proportion of different respondent
classes.

Online Sampling Techniques

1. Online panel samples - are comprised of individuals who have agreed to participate in online surveys. Normally,
they have registered with a panel company and have agreed to participate in surveys with some sort of
compensation, such as points that can be redeemed for products and services. Panel members have
divulged large amounts of information about themselves to the panel company, so it is easy for the
company to select panelists who satisfy population membership criteria specified by clients, such as age
range, education level, and household size.
2. River sample is created via the use of banners, pop-ups, or other online devices that invite website visitors to
take part in the survey. The “river” is the steady stream of website visitors, and these invitations figuratively
dip respondents out of the Internet river. Of course, the online questionnaire may have screening questions
so that only qualified prospects are allowed to take part in the survey. The sample frame of a river sample is
the stream of visitors visiting the site issuing the invitation, and river samples are considered random
samples of these sample frames as long as the invitations are not unusual in duration, appearance, or
relevance.
3. Email list samples are those purchased or otherwise procured from someone or some company that has
compiled email addresses of opt-in members of the population of interest. The vendor company can pull
random samples and may have the ability to satisfy selection criteria specified by the client company. The
list company may sell the list, or it might issue email invitations to maintain the propriety of email addresses
on the list. Obviously, the master source list is the sample frame, so if the list company has been diligent,
the email list sample will be a good representation of the population. However, if the list company has not
done due diligence, there will be sample frame error in the email list.

DEVELOPING A SAMPLE PLAN

STEP 1 : Define the population. Create a precise description of the group under investigation
using demographics, buyer behavior, or other relevant constructs.
STEP 2 : Obtain a sample frame. Gain access to some master source that uniquely identifies all
the units in the population with minimal sample frame error.
STEP 3: Decide on the sample method. Based on survey objectives and constraints, endeavor to
select the best probability sample method, or alternatively, if appropriate, select the best
nonprobability sample method
STEP 4: Decide on the sample size. If a probability sampling plan is selected, use a formula ;( to
be covered in the following chapter)
STEP 5: Draw the sample. Using the chosen sample method, apply the necessary steps to select
potential respondents from the sample frame
STEP 6: Validate the sample. Inspect some relevant characteristics of the sample (such as
distribution of males and females, age ranges, etc.) to judge how well it matches the
known distribution of these characteristics in the population