SAMPLING
(CH. 11)
Agenda and objectives
2
Sampling Procedure
Population vs. sample
Review the basic types of sampling
Distinguish probability samples from
nonprobability samples
Where we are in the marketing researc
h process:
Stage 4: Sampling and Data Collection
3
Formulate
Problem
Determine Research
Design
Design Data-Collection Method
and Forms
Sampling and Data Collection
Analyze and Interpret
Example: Are students satisfied with
4
NJTU?
Who would you survey?
Why not survey all students?
If some students can represent all students,
surveying them can be enough.
Who should be included in the sample?
How should students be chosen?
Example: Are students satisfied with
5 NJTU?
Student FRES SOP JUNIO SENI NDE TOTAL
Level H. H. OR
2017 7,24 7,35 7,91 9,13 647 32,294
Enrollm 5 6 1 5
ent
N = number of the population
n = number of the sample
To address the question, you plan to ask 20
(200 or 2000) undergraduate students
enrolled in the College of Business.
What would be N and n?
Any issues or problems?
What is a sample?
6
Population vs. Sample
Population
Any COMPLETE group
shares some common set
of characteristics
Census
complete survey of
an entire population
Sample
A portion of the population
(must be representative of
population)
Population vs. Sample
8
3 factors that influence sample
representativeness
Sampling procedure
Sample size
Participation (response rate)
When might you sample the entire
population?
When your population is very small （ less
than 500 ）
When you have extensive resources (time
and money)
Sampling – why?
9
Sampling
The process of getting
samples from a population
Benefits of sampling
Money
Time
Possibility
Accuracy?
By studying the sample,
we may generalize results
to the population
6-Step Procedure for Drawing a Sample
10
Define the Target
Population
Identify the
Sampling Frame
Select a Sampling
Procedure Steps 1~3:
today Steps
4~5: next class
Determine the
Step 6: next
Sample Size
week
Select the Sample Elements
Collect the Data from the
Designed Elements
Step 1 - Defining Target Populat
11
ion
( Target) population
The collection of elements or objects
that possess the information the
researcher is seeking
Depends on your research question
Examples: What’s the target population?
12
RQ: What are the most three preferred
car models of college students in NanJing?
Target population: all college students in NanJing
RQ: Which brand(s) of office appliances
do Chinese accounting firms use?
Target population: all Chinese accounting firms
RQ: How much do Latino American parents
whose annual income are above $50,000
spend for children’s toys per month?
Target population: all Latino American parents
whose annual income are above $50,000
Step 2 – Sampling Frame
13
Sampling frame
The listing of the elements from which the
actual sample is drawn
Examples: telephone book, customer list, etc.
What are the possible problems with telephone book?
Why?
Because we do not always know all the
elements in the population
Step 2 – Sampling Frame
14
Examples:
survey method course students: student roster
NanJing residents: telephone book
If no such list exists, you have to make
your own list of sampling frame.
Examples: a list of all restaurants of NJTU
campus, a list of all department stores at
NanJing.
Practice: What’s the sampling frame
15
?
Target population: NJTU college student
Sampling frame: NJTU college student roster
(from NJTU directory?)
Target population: Italian
restaurants of Chicago
Sampling frame: a list of all of Italian
restaurants of Chicago (from Yelp list?)
Target population: grocery stores of
NJTU
Sampling frame: a list of all grocery stores of NJTU
that customers may go to (from Yelp? Or
Sampling frame vs. Population
16
Ideally, a perfect sampling frame should
cover the target population only.
But realistically, a sampling frame is usually
an estimation of the target population, which
might result in sampling frame bias.
Populati Sampling
on frame
An example of biased sampling frame
17
…
In 1936, Literary Digest conducted a postcard
US Presidential election poll
Sample size: 10 million Americans (2 million
returned)
Sampling based on telephone books and
automobile registrations
Survey result: Alf Landon (R) > Franklin
Roosevelt (D)
Election result: Alf Landon (R) <
Franklin Roosevelt (D)
What was the problem?
Step 3 – Select sampling proced
18
ure
Define the Target
Population
Identify the
Sampling Frame
Select a Sampling
Procedure Probability Samples
• Simple Random
Nonprobability
sampling
Samples • Stratified sampling
• Convenience Samples
• Proportionate
• Judgment Samples
• Disproportionate
• Quota Samples
• Cluster sampling
• Snowball samples
• Systematic
• Area
Non-probability versus probability sampl
1 ing
9
Nonprobability sampling Probability sampling
Probability of being Probability of
chosen unknown, being chosen
including zero known with
chance of being
nonzero chance
included
Planned and
Cheaper,
convenient but systematic
researcher’s bias approach, less
and lack of bias with more
representativenes
Whether generalizability
each element in the sample has a
s
known, nonzero
chance of being selected or not
Non-probability sampling
20
Convenience
sampling
Judgment
sampling
Quota
sampling
Snowball
Non-probability
21
sampling:
Convenience
Convenience sampling
sampling
Attempts to obtain a sample of convenient
elements
Respondents were selected because they
happen to be in the “right place at the right
time”
Volunteer to participate…
Examples
Mall interception
Department stores use charge account lists
“People on the street” interview
Friends of researchers
Example of convenience sampling
22
On a 6:00pm broadcast, a local TV
station posed the following question
to its viewers:
"Do you think the drinking age in New
York should be lowered from 21 to 18?"
Almost 4,000 people called in that night
and 78% were in favor of lowering the
age requirement.
Conclusi
on?
Problem
Non-probability
sampling: Judgment
sampling
23
Judgment sampling
An experienced individual selects the sample
based on his experience about appropriate
characteristics required of the sample
member
researcher uses judgment to decide elements
to sample
Examples
Expert witness
Purchase engineers selected in industrial
marketing research
Non-probability
24
sampling: Quota
sampling
Quota sampling
researcher ensures that proportion possessing a
certain characteristic is approximately the same
as the proportion in the population
Two stages: 1st stage, figure out quotas of population
elements.
2nd stage: sample elements based on convenience or judgment
Populati Sample
on composit
Control composit ion
Characteri ion
Percenta Percenta Numb
stic Sex ge ge er
Male 4 4 48
Fema 8 8 0
le 5 5 52
10
2 10
2 100
0
0 0 0
Non-probability
2
sampling: Snowball
5
sampling
An initial group of respondents is
selected, usually at random
After being interviewed, these
respondents are asked to identify
others who belong to the target
population of interest
Subsequent respondents are
selected based on the referrals.
Types of Probability Sampling
26
Simple Random
Sampling
Stratified
Sampling
Proportionate Sample
Disproportionate
Sample
Cluster
Sampling
Systematic Sampling
Area Cluster Sampling
Probability
27
Sampling: Simple
Random
Sampling
Simple random sampling - Using sampling
procedures so that each element has equal
chance of being selected in the sample
Examples
Table of random numbers
Computer generated random numbers
Random digit dialing
•This is a basic method of other probability sampling,
But it can not be widely used, it’s difficult to obtain a
complete, current, and accurate list of the population's
elements.
Example: Are students satisfied with
NJTU?
Who would you survey?
All of NJTU students
Sampling frame:
NJTU student directory
Who to be included in the
sample?
Population: 45,000
Sample: 450
By simple random sampling
Example - Simple Random Sampling
2
9
Probability
Sampling:
31
Stratified
Stratified Sampling
Sampling
Divide population into mutually exclusive
and exhaustive subsets (a.k.a. strata)
A simple random sample of
elements is chosen independently
from each strata
Guarantees representation of
Marketing Research - University of
small/rare segments
Illinois
Probability
32
Sampling:
Stratified
Types of stratified sampling:
Sampling
Proportionate: # of respondents sampled
from each stratum reflects number of
persons in each stratum in population
Disproportionate: Sample sizes of strata
reflect relative
variability in each stratum in population
e.g., if expect more individual differences in male
respondents than in females, sample more males to
try to pin down estimates and increase precision
([Link]=var/n)
Stratified Sampling
Revisit Example: Are students satisfied
with NJTU?
3
3
Student Level FRESH. SOPH. JUNIOR SENIOR NDEG TOTA
L
2015 7,24 7,35 7,91 9,13 64 UGR
Enrollm 5 6 1 5 7 32,2
ent 94
Student
roster
Fresh Sophom Juni Seni Nondegr
men ore or or ee
Rando Rando Rando Rando Rando
m m m m m
sampli sampli sampli sampli sampli
ng ng ng ng ng
Stratified Sampling – Proportional
3
4
Student FRES SOP JUNIOR SENI NDEG TOTA
Level H. H. OR L
2017 7,24 7,35 7,91 9,13 64 UGR
Enrollm 5 6 1 5 7 32,2
ent (22 (22 (25 (28 (2 94
%) %) %) %) %) (100
Sample %)
Size
(n=100)
Fresh Sophom Juni Seni Nondegr
men ore or or ee
2 2 2 2 2
2 2 5 5
Stratified Sampling – Disproportion
3
5
al
Student FRES SOP JUNIOR SENI NDEG TOTA
Level H. H. OR L
2013 7,24 7,35 7,91 9,13 64 UGR
Enrollm 5 6 1 5 7 32,2
ent (22 (22 (25 (28 (2 94
%) %) %) %) %) (100
Sample %)
Size
(n=100)
Fresh Sophom Juni Seni Nondegr
men ore or or ee
1 2 2 3 6
8 0 3 3
Why Use Disproportional?
36
Why use disproportional?
If we expect more variance (spread, within-
strata difference) in a particular group?
If a particular strata is small? (std. error)
7 Satisfac
6.
6 2 tion
5. 5.
5 5 5
5
4.
4 2
3
2
1
Freshm Sophom Juni Seni ND
Probability
37
sampling:
Cluster
sampling
Divide population into subsets
called clusters
Randomly select a few subsets
(Area) Cluster Sampling
Section 1 Section 2
Section 3
Section 5
Section 4
Multi-stage (Area) Cluster Sampling
3
9
Northe Midwe
ast st
Iow India Illino
a na is
Lak Coo Will Champai Kan
e k Coun gn e
Coun Coun ty County Coun
ty ty ty
Cluster sampling
4
0
Clusters should be heterogeneous
within and homogeneous across. Ideally,
each cluster should be a small-scale
representation of the population.
Advantage: Economically efficient
Disadvantage: Statistically inefficient
Difference between cluster and stratified
sampling?
Stratification vs. Clustering
4
1
Stratification
Stratification Clustering
• Divide population • Divide population
into groups different into comparable
from each other: groups: schools,
sexes, races, ages cities
• Sample • Randomly sample
randomly from some of the groups
each group • More error
• Less error compared to simple
compared to random
simple random • Reduces costs to
sample only some
areas or
Example – Cluster sampling
42
Cluster sampling
Cluster all NJTU students by
departments
Randomly sample from each
department
Systematic Random Sampling
43
An easy cluster sample
A random sampling process in which every
kth (e.g. every 5th element) or member of the
population is selected for the sample after a
random start is determined
Population (N) = 2000, sample size (n) =
50, k=N/n, so k = 2000/50 = 40
With list of the 2000 subjects in the sampling
frame, go to the starting point, and select every
40th name on the list until the sample size is
reached.
intercept shoppers as probability sampling
Systematic Random Sampling – every 5th s
4 hopper
4
Example: Are students satisfied with NJTU
?
Who would you survey?
All of NJTU students
Sampling frame:
NJTU student directory
Who to be included in the
sample?
Population: 45,000
Sample: 450
By the systematic random
sampling
Example – systematic random sam
46
pling
K = 45,000/450 = 100
A random start
Choose Every 100th student from the
NJTU student directory
Which Sampling Method to Use
4
7
Cost/ease Statistical efficiency
NonprobabilitySampling
very low very poor
Convenience
Judgment
slightly more very poor
slightly more very poor
Quota
slightly more very poor
Snowball
Probability Sampling
SRS moderate fair
costly can be very good
Stratified
less than fair to good
Cluster stratified
Sampling in Practice
48
Often a non-random selection of
basic sampling frame (city,
organization etc.)
Fit between sampling frame and
research goals must be evaluated
Sampling frame as a concept is
relevant to all kinds of research
(including nonprobability)
Nonprobability sampling means
you cannot generalize beyond the
sample
Probability sampling means you can
Revisit Example:
49
Are students satisfied with
NJTU?
Who would you survey?
Why not survey all students?
If some students can represent all
students, surveying them can be enough.
Who should be included in the sample?
How should students be chosen?
Stratified Sampling
Revisit Example: Are students satisfied
with NJTU?
5
0
Student FRES SOP JUNIOR SENI NDEG TOTA
Level H. H. OR L
2015 7,24 7,35 7,91 9,13 64 UGR
Enrollm 5 6 1 5 7 32,2
ent 94
Stude
nt
Rost
er
Fre Sop Juni Seni NDE
sh h. or or G
Rando Rando Rando Rando Rando
m m m m m
sampli sampli sampli sampli sampli
1. Bed and breakfast inn
The owners of a popular bed and breakfast inn in Kauai, Hawaii, noticed a decli
ne in the number of tourists and length of stay during the past 3 years. An over
view of industry trends indicated that the overall tourist trade was expanding a
nd growing rapidly. The managers decided to conduct a study to determine pe
ople’s attitudes toward the particular activities that were available at and arou
nd the inn. Because they wanted to cause the minimum amount of inconvenie
nce to their guests, the owners devised the following plan. Interview request c
ards, which were available at the Chamber of Commerce office, the Visitor Info
rmation Center, and three of the more popular restaurants, indicated the natur
e of the study and encouraged visitors to participate. Visitors were asked to re
port to a separate room at either the Chamber of Commerce office or the Visit
or Information Center. Personal interviews, lasting 20 minutes, were conducte
d at these locations.
50
1. Bed and breakfast inn
a. What type of sampling method was used?
b. Critically evaluate the method used.
51
1. Bed and breakfast inn
The owners of a popular bed and breakfast inn in Kauai, Hawaii, noticed a decli
ne in the number of tourists and length of stay during the past 3 years. An over
view of industry trends indicated that the overall tourist trade was expanding a
nd growing rapidly. The managers decided to conduct a study to determine pe
ople’s attitudes toward the particular activities that were available at and aro
und the inn. Because they wanted to cause the minimum amount of inconveni
ence to their guests, the owners devised the following plan. Interview request
cards, which were available at the Chamber of Commerce office, the Visitor Inf
ormation Center, and three of the more popular restaurants, indicated the nat
ure of the study and encouraged visitors to participate. Visitors were asked to
report to a separate room at either the Chamber of Commerce office or the Vi
sitor Information Center. Personal interviews, lasting 20 minutes, were conduc
ted at these locations.
52
1. Bed and breakfast inn
a. What type of sampling method was used?
volunteers were asked to come forward
convenience sampling
b. Critically evaluate the method used.
Convenience samples may or may not be representative.
Volunteers = helpful and kind; might provide a biased or favorable view
Alternative: a random sample
nonresponse bias
Alternative:
randomly select the local attractions or eating establishments
and attempt to conduct personal interviews with the visitors
53
Summary
51
Sampling Procedure
Population vs. sample
Basic types of sampling
Probability samples vs.
nonprobability samples