RESEARCH METHODOLOGY

Sampling

Introduction
Our Knowledge, Attitudes and Actions are based to a

large extent on samples.

This is equally true in every day life and scientific

research.

Introduction (Contd)
A persons opinion of an institution that conducts

thousands of transaction every day is often determined

by one or two encounters he has had with the
institution in the course of several years.
In science and human affairs alike we lack the

resources and time to study more than a fragment of

the phenomena/population that might advance our
knowledge.

Objectives of the presentation

Enhance knowledge, skills and abilities of
participants in Sampling theory and
practice
Aims to:
Correct common misconceptions about sampling.
Develop an understanding of the various types of
random and non-random sampling
provide the advantages and disadvantages of
each.

Questions addressed
What is sampling?
Basic Definitions and concepts in Sampling?
Why is sampling necessary?
Random (Probability) sampling?
Non-Random (Non-Probability) sampling?
Advantages and disadvantages of Random sampling

and Non-Random sampling?

What is Sampling?
Sampling involves the selection of a number of study

elements/units from a defined study population.

Sampling is the process of selecting elements from the

study population in such a way that the elements

selected represent the population

Concepts in Sampling
A population (target population) is the entire collection of all

the elements that are of interest in a particular investigation.

A single member of the population is referred to as a

population element

Sampled/Study Population is an aggregation of elements from

which the sample is actually drawn.

A sample is a collection of elements (subset) drawn from the

study population.

An Illustration of a Sample
Who do you want to

generalize to?
What Population can you
get access to?
How can you get access to
them?
Who is under study
Who is in you study?

The Theoretical

Population
The Study Population
The Sampling Frame
The Sample
The Sub-sample of the

Sample

Concepts (contd)
Sampling Unit: Sets of elements considered for selection

in some stage of sampling.

In single stage sampling, sampling units are the same as

elements.

Sampling Frame: A complete listing of units/elements of a

study population from which a sample is drawn.

A census: is the complete enumeration of all the elements

in the population.

Concepts (contd)
Representative sample: This is a sub- group of a

population in which every major criteria/characteristic

of the study population from which the sample is
drawn is present.

Observation unit: This is an element or aggregate of

elements from which information is collected

Response Rate: Proportion of the sample elements from

which complete or partial data is actually collected.

Concepts (Contd)
Parameter: A summary description of a fine variable in a

population. For example the mean Income of all families

in the city and the age distribution of the citys population
are parameters. Scientific research involves the
estimation of population parameters on the basis of
sample observation.

Statistic: A summary description of a given variable in

samples. Thus a mean income computed from a sample

and age distribution within a given sample studied
constitute statistic. Sample statistics are used to make
estimates of population parameters.

Concepts (contd)
Sampling error: Probability sampling methods do

not often provide statistics exactly equal to the

parameter that they are used to estimate. However,
probability theory enables us to estimate the degree
of Error to be expected for a given sampling design.
Sampling error refers to the differences between the

attributes of the sample and those of the population

from which it is drawn.

Why use Samples

To reduce the time and cost of data collection. It is cheaper

to collect data from a sample than from an entire

population. Also, sample information can be collected and
summarised more quickly since fewer elements are
involved.
By concentrating on fewer elements, more comprehensive

data can be obtained from the sample than in a census.

In investigations where data collection leads to the

destruction of the elements observed (e.g. in testing the

lifespan of bulbs or in blood tests) , there is no alternative
but to sample.

Sample Selection
When we have to draw a sample from the total population,
we are confronted with the following key questions:
What is the group of people/things/elements (Study population)

from which we want to draw a sample?

How many people/elements do we need in our sample?(sample

size)

How will these be selected? (Sampling Methods)

Types of Sampling Methods

Random Sampling (Probability or Statistical Sampling)
A random sample is one in which the units of analysis have

been selected from the sampling frame (or sometimes

directly from the population) by a process involving the
use of random numbers. Random sampling is structured
so that all units within a population (or subset of the
population) have an equal chance of being selected.
Random sampling is presumed to be done without

replacement meaning that once a unit is selected to be in

the sample, it is not placed back in the sampling frame and
eligible to be sampled for second or third time.

Random Sampling (Contd)

Note:
Random sampling is not the same as chance

sampling, such as interviewing the first thirty

patients who happen to visit the health clinic
during your site visit.
It is also not the same as deliberately selecting

units to be representative of the population.

Both of these are non-random samples and
will be discussed later in this presentation

Probability Sampling Theory

The purpose of sampling is to select a set of elements in a given

population in such way that the description of those elements

accurately portrays the characteristics of the total population from
which they are selected.
Random selection is key to a probability sample. In random selection
each element has an equal chance of selection, which is independent of
any other event in the selection process.
Why Random Selection?
Serves as a check against conscious or unconscious bias on the part of
the researcher

It offers access to the body of probability theory, which in turn

provides the estimates of population characteristics as well as estimates
of Error.

Types of Sampling Design

Simple Random Sampling
A Simple Random Sample is a sample that is selected in

such a way that samples of the same size have equal

chances of being selected.
Consequently, all individual elements in the population

would have the same chance of being selected.

Simple Random Samples are selected using a table of

random numbers and requires to assign numbers to all

elements/units in the sampling frame before the sample
can be drawn.

Types of Sampling Design

(Contd)
Systematic Sampling
A Systematic Sample is a sample obtained by randomly

selecting one element from the first k elements in the frame

and every kth element thereafter.
Systematic sampling provides an alternative to Simple

Random Sampling especially if a good sampling frame is

not available. It is also easier to perform in the field hence
is less subject to selection errors by field workers.
Note: When the population is repetitive or cyclic in nature

(e.g. rainfall, sale of luxurious items etc), systematic

sampling should not be used.

Types of Sampling Design

(Contd)
Stratified Random Sampling
The sample is drawn by first dividing the population

into homogeneous sub-populations (strata) and then

drawing simple random samples from each of the subpopulation (stratum).
It ensures proportionate representation of key

characteristics when drawing a sample from a

heterogeneous population and hence increases the
quantity of information for a given cost.

Stratified Random Sampling (Contd)

Procedure 1: Sort all the numbered units of the sampling frame into their

appropriate strata. The strata can be defined by more than one

dimension for instance recent and older projects, and also small and
large investments.
Decide the size of the sample you want to draw from each stratum. Then
sample from each stratum exactly as for simple random sample.

Procedure 2: Make a tally sheet to represent each stratum. Decide the size

of the sample that you want from each stratum and record and circle that
number in the appropriate cell of the tally sheet. Sample from the full
sampling frame, just as for a simple random sample, but after drawing
each unit check its strata characteristics and record the units number in
the cell of the tally sheet that corresponds with those characteristics.

Types of Sampling Design

(Contd)
Disproportionate Sampling
Disproportional stratified random samples are used to

improve the precision of inferences made to an

individual stratum and comparisons among the strata.

It comes into play when one wishes to sample some

population disproportionately to ensure sufficient

number of cases from each group for analysis.

Types of Sampling Design

(Contd)
Cluster Sampling
A Cluster Sample is obtained by first grouping the elements of

the population into clusters and then simple random

sampling or other type of sampling is used to select the
clusters.

This type of sampling is used when a sampling frame cannot

be prepared for individual units in the population but can be

prepared for some cluster of them or when substantial time
or expense can be saved by collecting data from a modest
number of clusters.

Cluster Sampling (Contd)

Cluster sampling is similar to stratification in that both involve

partitioning the population into sub-groups. However, while in stratified

sampling the elements within a stratum are relatively homogeneous and
cannot be used as a representative subset of the population, in cluster
sampling, the elements in each cluster are relatively heterogeneous and
each cluster can be used as a representative subset of the population.
In cluster sampling, all elements of the sampled clusters are observed.
Since elements within a cluster are often physically together, they tend to

have similar characteristics. This means the amount of information

about a population parameter may not be increased substantially by
taking many measurements within a cluster. Hence, a cluster sample
consisting of many small-sized clusters is likely to yield better
information than one with few large size clusters.

Cluster Sampling (Contd)

When drawing a cluster sample, the first task is to specify

appropriate clusters. In doing so, consideration has to be

made about the level of heterogeneity of elements within
clusters. If clusters are generally heterogeneous, then few
large clusters may be selected to constitute the sample but if
they are homogeneous, then many small-sized clusters
should be used.

Types of Sampling Design

(Contd)
Multi Stage Cluster Sampling
In this type of sample, a random cluster sample is

drawn, then units within the clusters are randomly

selected. The second phase of sampling can be a
cluster sample (e.g. sub-counties within a district), a
stratified random sample or a simple random
sample. More than two stages of sampling are
possible, such as from regions, districts, subcounties, schools to classrooms.

NON-RANDOM SAMPLING
A non-random sample (Non-Probability Sample) is one in which the

units of analysis have been selected from the sampling frame by

convenience (such as the women at a market on the morning that the
Researcher/evaluator makes a visit there) or for the purposes of
representing certain characteristics of the population that are of
particular interest.
Non-random samples are usually used in qualitative research

methods. They are used in instances where the elements of the

population do not have a well-defined chance of being selected.
Here, the representativeness of the sample is not a major issue.

Non-Random Sample (contd)

Rules of the Game
Non-random samples serve legitimate purposes, but

those purposes need to be kept in mind when making

inferences from the sample.
Non-random samples should not be used in infer

generalizations about the whole population. In addition,

statistical testing and confidence intervals should never
be computed from non-random samples because those
procedures assume random samples.

Types of Non-Random Samples

Convenience (haphazard, accidental sample)
The most easily available units from the population are

selected because they are convenient for the

Researcher/evaluator.
Units could be selected because they show up at a

convenient time or place for data collection. They may

be selected because you encounter the units first, or
because the units volunteer to provide needed
information. The researcher may simply stand at the
entrance of a market or a street corner and interview
people who pass by.

Types of Non-Random Samples

(contd)
Snowball (Chain Sample)
When the boundaries of the population are unknown, and

there is no sampling frame, sometimes a few encountered

units of analysis can lead the researcher/evaluator to other
units.
For example, in a commission of inquiry or in a criminal
investigation, interrogation of a few suspects/witnesses
usually leads to a chain of other suspects/witnesses who can
also be interrogated. Researchers also commonly snowball
the bibliography of their background research, checking
found documents for references to other relevant documents.

Types of Non-Random Samples

(contd)
Quota Sample:
In this type of sampling, units are selected proportionally (but
not randomly) from given strata of interests.
Begins with a matrix describing the characteristics of the target

population such as sex, age, and education composition. When

such matrix has been created and a proportion assigned to
each cell in the matrix, the researcher collects data from a
person having all the desired characteristics. All the persons in
the given cell are then assigned a weight appropriate to their
portion of the population when all the sample elements are so
weighed.

Types of Non-Random Samples

(contd)
Maximum Variation (heterogeneity) Sample
Units

are drawn to represent the range of some

characteristic of interest.

The researcher/evaluator draws units to represent the

range of some characteristics that is thought likely to affect

the outcomes of interest. The characteristics can be
measured numerically or categorically. (height in inches or
brown eyes)

Types of Non-Random Samples

(contd)
Confirming and Disconfirming Cases Sample
Units

are drawn from cases known to be

successful and those known to be unsuccessful to
try to determine what accounts for the differences.

Units are drawn from project sites known to

confirm and contradicts some established or

developing theory, and then examined to
determine what accounts for the differences and
refine the theory.

Types of Non-Random Samples

(contd)
Extreme Cases Sample
Units are drawn from the extreme cases of some

characteristics of interest, such as oldest and most

recent cases.
Units are drawn from both extremes of some

characteristics that is thought likely to affect project

success. Generally this can be done only for
characteristics measured

Types of Non-Random Samples

(contd)
Criterion sample
All units in the population that meet a specific criterion are

selected, such as all senior officer in UPDF or all

participants collar pimps, Student in A ; O or University
The

researcher/evaluator selects every unit in

population that has a certain specified characteristic.

Advantages of Random Samples

Random samples are almost always preferable when

the intent is to generalize from the sample to an entire

population. They eliminate the chance that the
evaluator will bias the selection of the sample- either
consciously or unconsciously.

Random samples entail procedures for estimating the

extent of sampling error. These procedures are applied

each time one conducts a test of statistical significance
or computes confidence intervals.

Advantages of Non-Random Samples

Several non-random samples have very legitimate objectives, but

they generally create the risk of evaluator bias in the selection

process unless the population is so small that no discretion is
needed in selecting among eligible units in the population.

One can always compare the prevalence of the characteristics of

concern in the population with their prevalence in the sample, but

that only assures that sample is representative of characteristics
known at the time of sampling, whereas sampling is undertaken to
collect data on additional characteristics not yet known.

END OF PRESENTATION

U
O
Y
K
N
A
N
TH
R
O
I
O
T
F
A
P
I
C
I
T
R
A
P
R
U
YO