1. Marketing 6 Chapter 14
Sampling Fundamentals
Submitted by:
Franklin K. Go
Raymond A. Gonzales
Angelo R. Cantor
Submitted to:
Mr. Abelito T. Quiwa, MBA
2. Learning Objectives:
1. Distinguish between a census and a sample.
2. Know the difference between sampling and non-sampling errors.
3. Learn the concept of sampling process.
4. Describe probability and non-probability sampling procedures.
5. Determine sample size with ad-hoc methods.
6. Learn to deal with non-response problems.
7. Understand sampling in international context.
Sample or Census
A researcher is interested in the characteristics of the population wherein his potential target
market may reside. The respondents are asked to input necessary information through surveys
called census.
When a Census is Appropriate
A census is appropriate if the sample size is small. Every individual may be included in the
population.
When a Sample is Appropriate
A sample is appropriate if the population is large, the time allotted in obtaining the information
from the population is long and the cost is high.
Error in Sampling
Error in sampling occurs if there is a gap between the true value and the observed value of the
variable interest of the population.
• Sampling Error occurs if the error lies in the sample parameter and the sample statistics
because of sampling.
• Non-sampling Error occurs if the error lies in the population.
3. Sample Process
Factors that should be considered when the decision to use a sample is made:
• Various steps included in sampling:
• Major activities with sampling process:
1. Identify the target population.
2. Determine the sampling frame.
3. Resolving the Differences.
4. Selecting a sampling procedure.
5. Determining the relevant sample size.
6. Obtaining information from respondents.
7. Dealing with the non-response public.
8. Generating information for decision-making purposes.
Determining the Target Population
4. Sampling is necessary to get information about the population. Inaccurate definition of the
population results to inaccurate data collection. This leads to the wrong question being
answered. For instance, in Houston, in order for car dealers to know the prospective car buyers,
the population should consist of adults with driver’s licenses. In order for toy stores to know
their potential customers, they need to identify the number of children in every household of the
population in.
• How do you define children? Are they below 10 years, 13 years or 16 years old?
• How do you define Houston? Does it include only the metropolitan area or are the
suburban areas included as well?
• Who in the household is going to provide the information?
The following should be considered in determining the target population:
1. Look to the research objectives
2. Consider alternatives
3. Know your market
4. Consider the appropriate sampling unit
5. Specify clearly what is excluded
6. Don’t overdefine
7. Should be reproducible
8. Consider convenience
Determining the Sampling Frame
It is important to distinguish the difference between the population and the sample frame. The
sampling frame is a list of the population with which the sample is taken from. They may be
magazine subscribers, retail stores or college students. Even maps may serve as a list.
Tasks to do in determining the sample frame:
1. Create lists
The hardest way to obtain a sample is creating a list. The researcher needs to
identify which among the population in an area should be included.
2. Create lists for telephone interviewing
Telephone directories are often used to generate samples because the population
is narrowed down to one entry per household and also eliminates those who do
not have telephones.
3. Dealing with population sampling frame difference
The following are three types of problems when dealing with population sampling
frame difference:
1. Subset problem
Occurs when sampling frame is smaller than the population.
2. Superset problem.
Occurs when the sampling frame is larger than the population.
3. Intersection problem.
Occurs when some elements of the population are omitted from
the sampling frame.
5. Selecting a Sampling Procedure
There are many ways of selecting a sampling procedure. First, the researcher must select
between Bayesian Sampling Procedure and Traditional Sampling Procedure. Then, the
researcher must decide whether the sample may have a replacement or not.
Probability Sampling
Probability sampling involves four considerations. First, the target population must be specified.
Second, the methods for selecting sample needs to be developed. Third, the sample size must be
determined. Last, the non-response problem must be addressed.
Selecting the Probability Sample
Methods used to select a probability sample:
1. Simple Random Sampling.
Each population member has equal probability of being selected.
2. Accuracy Cost Trade-off
The ratio of accuracy over cost. In general, the higher the cost, the higher the
accuracy.
3. Stratified Sampling
Similar to the Random sampling but increases the accuracy higher than the cost
increase.
4. Proportional Stratified Sampling
The number of sample units is proportional to the number of population.
5. Directly Proportional Stratified Sampling
The population is grouped based on categories like a population of 600 is divided
into 400 for brand loyalty and 200 for variety seeking. Samples are taken from
each division. If a sample size of 60 is desired, a 10% directly proportional
stratified random sampling is employed.
10% Directly
Consumer Group Proportional
Type Size Stratified Random
Size
Brand
400 40
Loyal
Variety
200 20
Seeking
Total 600 60
6. Inversely Proportional Stratified Sampling
If in a population of 600, 200 are heavy drinkers and 400 are light drinkers, If the
researcher values the opinion of heavy drinkers more than the light drinkers, more
6. people shall be sampled from the desired group. In such cases, the researcher can
use inversely proportional stratified sampling. If a sample size of 60 is desired, a
10% inversely proportional stratified random sampling is employed.
10% Inversely
Consumer Group Proportional
Type Size Stratified Random
Size
Brand
400 40
Loyal
Variety
200 20
Seeking
Total 600 60
7. Disproportional Stratified Sampling
The sample size is not proportional to the group size. This usually occurs when
multiple groups are compared.
8. Cluster Sampling
This is done by decreasing cost at a faster rate than accuracy.
9. Systematic Sampling
This is done by systematically spreading the sample size through the list of
population members.
Multistage Design
It is often used when samples in multiple areas are desired.
Cities in Ajax Country
Cumulative
City Population
Population
Concorde 15000 1-15000
Mountain 15001-
10000
View 25000
25001-
Filmore 60000
85000
85001-
Austin 5000
90000
90001-
Cooper 2000
92000
92001-
Douglas 5000
97000
Rural 97001-
3000
Area 100000
7. Non-probability Sampling
The cost and trouble of developing a sampling frame are eliminated unlike the probability
sampling wherein probability and theory guides the researcher in obtaining data from
samples. The following are stages where non-probability sampling is used:
1. Exploratory stage of a research project
2. Pretesting a questionnaire
3. Dealing with homogeneous population
4. When researcher lacks statistical knowledge
5. When operational ease is required
The following are the four types of non-probability sampling:
1. Judgment sampling
Expert use of judgment to identify representative samples.
The situations where judgment sampling is useful are, first, probability
sampling is not feasible and is expensive. Second, if the sample size is
very small. Third, it is useful to obtain a deliberately biased example.
2. Snowball Sampling
Appropriate for small, specialized population.
3. Convenience Sampling
Used to obtain information quickly and inexpensive.
4. Quota Sampling
It is a judgmental sampling with constraints which includes a minimum
number from each specified subgroup in the population. The following
are the characteristics of Quota Sampling:
1. It is often based on demographic data such as geographic location,
age, sex and income.
2. It eliminates gross biases that could be part of judgment sampling.
3. There are serious biases that are not controlled by the quota.
Determining the Sample Size
How large should the sample be? Although this question is direct, answering it is not easy.
Web-Based Samples – frequently used because of the ease that it brings in collecting
data. It has several benefits such as speed, flexibility and economy.
Non-response Problems
Sampling is done to obtain necessary data from the population. Unfortunately, some of the
populations are not cooperative. The following are reasons why some of the population become
non-respondents:
1. Refuse to respond
2. Lack of ability to respond
3. Not at home
8. 4. Inaccessible
What can be done about the non-response problems?
1. Improve the research design to reduce non-responses.
• For telephone interviews, gain initial interest through interviewer skills and design
the proper placements of questions.
• For mail surveys, motivate respondents through incentives.
2. Repeat the contact one or more times to reduce non-responses (callbacks).
• It is necessary to do as much callbacks as possible to reduce the number of non-
responses.
3. Attempt to estimate the non-response bias.
• Make extra effort to interview the sub sample to reduce non-response bias.
• Give incentives such as worthwhile gifts to entice respondents.
Shopping Center Sampling
Shoppers are intercepted and interviewed personally or through surveys. This method is called
store-intercept interviews.
Shopping Center Selection
Selecting a shopping center is essential because the respondents will be those who live in the
nearby area. The standard of living of the neighborhood will affect how the sampling will take
place and the types of samples that will be obtained.
Sample Locations within a Center
This is randomly selecting samples through shopping center visits.
Time Sampling
Time segments are devised when obtaining sample data because of the fact that the time people
go shopping vary from one another.
Sampling People versus Shopping Visits
Researchers should ask the respondents on how often they shop. The respondents are weighted
based on the frequency of visits when getting the average. Those who visit twice are weighed
1/2, those who visit thrice are weighted 1/3 and so on. Another approach is to use quotas. For
example, shoppers aged 25 to 45 tend to make more visits than those who are younger and older.
Another is the employment status. Unemployed people tend to shop more than those who are
employed. Quotas are set so that the number of samples is proportional to the number of the
population.