Educational Research
Chapter 5
Selecting a Sample
Gay, Mills, and Airasian
10th Edition
Topics Discussed in this Chapter

Quantitative sampling



Selecting random samples
Selecting non-random samples
Qualitative sampling

Selecting purposive samples
Quantitative Sampling


Purpose – to identify participants from
whom to seek some information
Issues



Nature of the sample
Size of the sample
Method of selecting the sample
Quantitative Sampling

Terminology

Population: all members of a specified group





Target population – the population to which the
researcher ideally wants to generalize
Accessible population – the population to which the
researcher has access
Sample: a subset of a population
Subject: a specific individual participating in a
study
Sampling technique: the specific method used to
select a sample from a population
Quantitative Sampling

Important issues

Representation – the extent to which the sample
is representative of the population




Demographic characteristics
Personal characteristics
Specific traits
Generalization – the extent to which the results of
the study can be reasonably extended from the
sample to the population
Quantitative Sampling

Important issues (continued)

Sampling error



The chance occurrence that a randomly
selected sample is not representative of the
population due to errors inherent in the
sampling technique
Random nature of errors
Controlled by selecting large samples
Quantitative Sampling

Important issues (continued)

Sampling bias




Some aspect of the researcher’s sampling design creates
bias in the data
Non-random nature of errors
Controlled by being aware of sources of sampling bias
and avoiding them
Examples


Surveying only students who attend additional help
sessions in a class
Using data returned from only 25% of those sent a
questionnaire
Quantitative Sampling

Important issues (continued)

Three fundamental steps



Identify a population
Define the sample size
Select the sample
Quantitative Sampling

Important issues (continued)

General rules for sample size



As many subjects as possible
At least thirty (30) subjects per group for
correlational, causal-comparative, and true
experimental designs
At least ten (10) to twenty (20) percent of the
population for descriptive designs
Quantitative Sampling

Important issues (continued)

General rules for sample size (continued)

See Table 4.2 (see NEXT SLIDE) for additional
guidelines for survey research





The larger the population size, the smaller the percentage
of the population needed to get a representative sample
For population of less than 100, use the entire population
If the population is about 500, sample 50%
If the population is about 1,500, sample 20%
If the population is larger than 5,000, sample 400
Qualitative Sampling
Selecting Random Samples



Known as probability sampling: Everyone has
probability of getting chosen
Best method to achieve a representative
sample
Four techniques




Random
Stratified random
Cluster
Systematic
Selecting Random Samples

Random sampling


Selecting subjects so that all members of a
population have an equal and independent chance
of being selected
Advantages




Easy to conduct
High probability of achieving a representative sample
Meets assumptions of many statistical procedures
Disadvantages


Identification of all members of the population can be
difficult
Contacting all members of the sample can be difficult
Selecting Random Samples

Random sampling (continued)

Selection process







Identify and define the population
Determine the desired sample size
List all members of the population
Assign all members on the list a consecutive number
Select an arbitrary starting point from a table of random
numbers and read the appropriate number of digits
If the number corresponds to a number assigned to an
individual in the population, that individual is in the
sample; if not, ignore the number
Continue until the desired number of subjects have been
selected
Selecting Random Samples

Random sampling (continued)

Selection issues

Use a table of random numbers (page 562)




Need to list all members of the population
Ignore duplicates and numbers out of range when sampled
Potentially time consuming and frustrating
Use SPSS-Windows or other software to select a random
sample


Create a SPSS-Windows data set of the population or their
identification numbers
Pull-down commands
 Data, select cases, random sample, approximate or
exact
Selecting Random Samples

Stratified random sampling


Selecting subjects so that relevant subgroups in
the population (i.e., strata) are guaranteed
representation
A strata represents a variable on which the
researcher would like to see representation in the
sample



Gender
Ethnicity
Grade level
Selecting Random Samples

Stratified random sampling (continued)

Proportional and non-proportional (i.e., equal size)

Proportional – same proportion of subgroups in the
sample as in the population


If a population has 45% females and 55% males, the
sample should have 45% females and 55% males
Non-proportional – different, often equal, proportions of
subgroups

Selecting the same number of children from each of the
five grades in a school even though there are different
numbers of children in each grade
Selecting Random Samples

Stratified random sampling (continued)

Advantages




More precise sample
Can be used for both proportional and non-proportional
samples
Representation of subgroups in the sample
Disadvantages


Identification of all members of the population can be
difficult
Identifying members of all subgroups can be difficult
Selecting Random Samples

Stratified random sampling (continued)

Selection process




Identify and define the population
Determine the desired sample size
Identify the variable and subgroups (i.e., strata)
for which you want to guarantee appropriate
representation
Classify all members of the population as
members of one of the identified subgroups
Selecting Random Samples

Stratified random sampling (continued)

Selection process (continued)

For proportional stratified samples


Randomly select a number of individuals from each
subgroup so the proportion of these individuals in
the sample is the same as that in the population
For non-proportional stratified samples

Randomly select an equal number of individuals from
each subgroup
Selecting Random Samples

Stratified random sampling (continued)

Selection process for proportional samples





Identify and define the population
Determine the desired sample size
Identify the variable and subgroups (i.e., strata) for
which you want to guarantee appropriate representation
Classify all members of the population as members of
one of the identified subgroups
Randomly select an equal number of individuals from
each subgroup
Selecting Random Samples

Cluster sampling

Selecting subjects by using groups that have
similar characteristics and in which subjects can
be found


Clusters are locations within which an intact group of
members of the population can be found
Examples




Neighborhoods
School districts
Schools
Classrooms
Selecting Random Samples

Cluster sampling (continued)

Multistage sampling involves the use of
two or more sets of clusters



Randomly select a number of school districts
from a population of districts
Randomly select a number of schools from
within each of the school districts
Randomly select a number of classrooms from
within each school
Selecting Random Samples

Cluster sampling (continued)

Advantages




Very useful when populations are large and spread over
a large geographic region
Convenient and expedient
Do not need the names of everyone in the population
Disadvantages


Representation is likely to become an issue
Assumptions of some statistical procedures can be
violated (you don’t need to know which ones in this class)
Selecting Random Samples

Cluster sampling (continued)

Selection process








Identify and define the population
Determine the desired sample size
Identify and define a logical cluster
List all clusters that make up the population of clusters
Estimate the average number of population members per
cluster
Determine the number of clusters needed by dividing the
sample size by the estimated size of a cluster
Randomly select the needed numbers of clusters
Include in the study all individuals in each selected
cluster
Selecting Random Samples

Systematic sampling


Selecting every Kth subject from a list of the
members of the population
Advantage


Very easily done
Disadvantages


Susceptible to systematic exclusion of some subgroups
Some members of the population don’t have an equal
chance of being included
Selecting Random Samples

Systematic sampling (continued)

Selection process







Identify and define the population
Determine the desired sample size
Obtain a list of the population
Determine what K is equal to by dividing the size of the
population by the desired sample size
Start at some random place in the population list
Take every Kth individual on the list
If the end of the list is reached before the desired
sample is reached, go back to the top of the list
Selecting Non-Random Samples




Known as non-probability sampling
Use of methods that do not have random
sampling at any stage
Useful when the population cannot be
described
Three techniques



Convenience
Purposive
Quota
Selecting Non-Random Samples

Convenience sampling

Selection based on the availability of
subjects



Volunteers
Pre-existing groups
Concerns related to representation and
generalizability
Selecting Non-Random Samples

Purposive sampling


Researcher believes that this is a representative
sample or an appropriate sample.
Selection based on the researcher’s experience
and knowledge of the individuals being sampled

Usually selected for some specific reason




Knowledge and use of a particular instructional strategy
Experience
Need for clear criteria for describing and
defending the sample
Concerns related to representation and
generalizability
Selecting Non-Random Samples

Quota sampling



Selection based on the exact
characteristics and quotas of subjects in
the sample when it is impossible to list all
members of the population
Example: “I need 35 unemployed mothers
and 35 employed mothers.”
Concerns with accessibility, representation,
and generalizability
Qualitative Sampling

Unique characteristics of qualitative research






In-depth inquiry
Immersion in the setting
Importance of context
Appreciation of participant’s perspectives
Description of a single setting
The need for alternative sampling strategies
Qualitative Sampling

Purposive techniques – relying on the
experience and insight of the researcher
to select participants

Intensity – compare differences of two or
more levels of the topics


Students with extremely positive and extremely
negative attitudes
Effective and ineffective teachers
Qualitative Sampling

Purposive techniques (continued)



Homogeneous – small groups of
participants who fit a narrow homogeneous
topic
Criterion – all participants who meet a
defined criteria
Snowball – initial participants lead to other
participants
Qualitative Sampling

Purposive techniques (continued)


Random purposive – given a pool of
participants, random selection of a small
sample
Inherent concerns related to
generalizability and representation
Qualitative Sampling

Sample size


Generally very small samples given the nature of
the data collection methods and the data itself
Two general guidelines



Redundancy of the information collected from
participants: Once you are hearing the same thing from
everyone, you are done collecting that data.
Representation of the range of potential participants in
the setting. Make sure that you select someone from
every part of the population that you want to examine.
More subjects does not mean “better.” More than 20 is
often unusual.
Generalizability

Probability sampling


Begins with a population
and selects a sample
from it
Generalizability to the
population is relatively
easy

Non-probability and
purposive sampling



Begins with a sample
that is NOT selected from
some larger population
Must consider the
population hypothetical
as it is based on the
characteristics of the
sample
Generalizability is often
very limited