SAMPLING

DAY 4
Learning Objectives

By the end of this lecture, you should be able to:

• define sampling concepts, including: population, sample, sampling frame,
sample design, parameter, statistic, and random sample
• discuss the differences between random and nonrandom samples
• list and define the types of random samples, including:simple, stratified,
and cluster
• list and define the types of non-random samples, including: quota,
accidental, snowball, judgmental, and convenience
• describe the role of statistics to determine confidence and precision
• describe the guidelines to determine the size of sample that is needed.
Define sampling concepts

o population: The total set of units. It could be all the citizens in a country,
all farms in a region, or all children under the age of five living without
running water in a particular area.
o Population is an accessible group of people who meets a well-defined
set of eligibility criteria
o It consists of the totality or aggregate of the observations with which the
researcher is concerned
o The utmost importance in selecting a population is that the population
should be clearly defined so that the sample can be accurately identified.
The specific population types are:
-Target population which is a group of individuals who meet the
criteria.

-Subject or respondent population which refers to a group of

individuals participating in the study.

-Strata or stratum which is described as a mutually exclusive

segment of a population established by one or more
characteristics.
Sample
o A subset of units selected from a larger set of the same units.
o Also referred to as subjects or respondents of the evaluation
o A sample must be representative, meaning that it must be like
the population in as many ways as possible
o It is selected through following an eligibility criteria that should
be established by the evaluator and this defines which
elements should be included or excluded from the population.
Such criteria may include age, marital status, education level,
etc.
Sampling
o Refers to the process of choosing a representative portion of the
entire population
What is sampling?

 When we begin planning our data collection strategy, we have to decide whether it is
possible to collect data from the entire population we intend to study
 If we are able to do that, we can then make an accurate report
 If we collect all the data accurately and reliably, then there is little chance of error.
 However, most often we are unable to collect data from the entire population. It takes
too much time and costs too much.
 Instead, we take a sample − a subset of the entire population.
 If we select a sample, we may be able to draw inferences (extrapolations,
interpretations) about a population based on our sample results; that is, we can
estimate what the population is like based on our sample results.
 We call this “generalizing to a population.”
 We use samples all the time. For example, when we have a blood test to check on our
health, the laboratory takes a sample rather than all our blood.
Sampling theory:
o This is developed to determine mathematically the most effective way to
acquire a sample that would accurately reflect the population under
study.
The key concepts of a sampling theory include:
1. A sampling unit which refers to a specific place or location which can be
used during the sampling process
2. A sampling frame which describes the complete list of sampling units
from which you can select your sample.
The steps involved in sampling include:

o Identifying the target population

o Identifying the subject or respondent population
o Specifying the criteria for subject or respondent selection
o Specifying the sampling design
o Recruiting the subjects
What is sampling?

 When we begin planning our data collection strategy, we have to decide whether it is
possible to collect data from the entire population we intend to study
 If we are able to do that, we can then make an accurate report
 If we collect all the data accurately and reliably, then there is little chance of error.
 However, most often we are unable to collect data from the entire population. It takes
too much time and costs too much.
 Instead, we take a sample − a subset of the entire population.
 If we select a sample, we may be able to draw inferences (extrapolations,
interpretations) about a population based on our sample results; that is, we can
estimate what the population is like based on our sample results.
 We call this “generalizing to a population.”
 We use samples all the time. For example, when we have a blood test to check on our
health, the laboratory takes a sample rather than all our blood.
How Large a Sample Do You Need?
 Sample size is a function of size of the population of interest,
the desired confidence level, and level of precision.
 You can calculate a formula to determine the appropriate
sample size or you can use a tool such as the one shown in the
slide below, a Guide to Minimum Sample Size.
 This table shows the sample size needed when estimating a
population percentage (or proportion) at the 95%
 confidence level and a + or – 5 percentage point confidence
interval.
 As you can see, the smaller the population, the higher
proportion of cases you will need.
Sample size.
o Prior to the selection of the sampling technique, the evaluator must first
determine the sample size.
o A sample size can be determined using the Slovin’s (1960) formula,
which is as follows:

n
n = ………………..
1 + Ne2

o n is the sample size

o N is the population size
o E is the margin of error
o 1 is the constant value
Population Size Sample Size Population Size Sample Size
10 10 550 226
20 19 600 234
40 36 700 248
50 44 800 260
75 63 900 269
100 80 1000 278
150 108 1200 291
200 132 1300 297 Source: R. V. Krejcie, and
250 152 1500 306 D. W. Morgan
300 169 3000 341
350 184 6000 361
400 196 9000 368
450 207 50000 381
500 217 100000 +385
Summary of Sampling Size
• Accuracy and precision can be improved by increasing your sample size.
In other words: By increasing sample size, you increase accuracy and
reduce margin of error.
• The standard you should aim for is a 95% confidence level and a margin
of error of +/- 5%.
• The larger the margin of error, the less precise your results will be.
• The smaller the population, the larger the needed ratio of the sample size
to the population size.
Types of samples

There are two kinds of samples, random (Probability) and non-random (non-
probability).

 Random samples are samples in which each unit in the population has an equal
chance of being selected.
 You can take a random sample of files, roads, farms, or people.
 An appropriately sized random sample should be representative of the population as a
whole, enabling you to generalize to the population from which the sample was
drawn.
 A complete list of every unit in the population of interest, called a sampling frame, is
needed to select a random sample.
 These units are selected using a random schedule; typically, we would use a table of
random numbers and select every unit until we reach the sample size we set.
 Random numbers can be generated using any major spreadsheet program.
Types of Random Samples

There are four types of random samples. They are:

• simple random samples
• stratified random samples
• multi-stage samples
• cluster samples
• combination random samples.
Summary of Random Sampling Process

1. Obtain a complete listing of the entire population

2. Assign each case a unique number.
3. Randomly select the sample using a random numbers table.
4. When no numbered listing exists or is not practical to create,
use systematic random sampling:
• make a random start
• select every nth case.
Simple Random Samples
 A simple random sample is the simplest sample. We establish a
sample size and then proceed to randomly select units until we reach
that number.
 Let us say we want to select 100 files from a population of 500.
 All the files have been consecutively numbered from 001 to 500 and are
filed in numerical order from 1-500.
 We could then use a random numbers table, mentally block it off into
three-digit numbers, and then select the first 100 numbers that fall
between 001 and 500. These are the files we select for our study.
Stratified Random Samples

 Sometimes we want to make sure specific groups are included that might otherwise
be missed by using a simple random sample; those groups are usually a small
proportion of the population.
 In this case, we would divide the population into strata based on some meaningful
characteristic.
 This kind of sample is called a stratified random sample.
 For example, you may want to make sure you have enough people from rural areas in
your study.
 If selected by a simple random sample, you may not get enough people from rural
areas if they are a small proportion of all the people in the area.
 This is especially important if you want to have sufficient numbers in each stratum so
you can make meaningful comparisons.
Cluster and Multi-stage Samples
Cluster sampling is another form of random sampling. A “cluster” is any
naturally occurring aggregate of the units that are to be sampled.
 Cluster samples are most often used when:
 You do not have a complete list of everyone in the population of interest
but do have a complete list of the clusters in which they occur, or
 You have a complete list of everyone, but they are so widely disbursed
that it would be too time consuming and expensive to send data
collectors out to a simple random sample.
 In a cluster sample, the cluster is randomly sampled (such as towns or
household) and then data is collected on all the target units. For
instance, if the evaluation needs to collect data on the height and weight
of children ages 2-5 in the program sites scattered across a large rural
region, the evaluators might randomly sample 20 villages from the 100
villages receiving the program, and then collect data on all the children
ages 2-5 in those villages.
Multi-stage random sampling

 Multi-stage random sampling combines two or more forms of

random sampling. It applies a second form of random sampling
to the results of a first form of random sampling.
 Most commonly, it begins with random cluster sampling and
then applies simple random sampling or stratified random
sampling.
 The drawback of multi-stage and cluster samples is that they
may not yield an accurate representation of the population.
Combination Random Samples
 Sometimes combinations of methods are used.
 The group may be divided into strata; all the people in one
stratum might be selected (as a census) and a random sample
selected from the other strata.

Examples of selecting random samples

For example, you want to observe classroom activities to measure
the amount of time spent doing hands-on learning
activities. You can:
• randomly select classrooms
• randomly select times of day
• randomly select days of the week.
Non-Random Sampling
When random sampling is not possible, there is need to use a different
approach. Non-random sampling approaches enable you to use a group
smaller than the total population.
While there are many names for different non-random sampling techniques,
they all are limited in their ability to generalize your findings back to a larger
population.
Even when you have a non-random sample (limited to a particular school,
for example), you can still use random sampling within it to make your
evaluation more credible.
Types of Non-Random Samples
quota: a sample in which a specific number of different types of units are
selected. For example, you may want to interview 10 teachers and decide
that five will be men and five will be women.
snowball: this type of sampling is used when you do not know who or what
you should include. Typically used in interviews.
Types of Non-Random Samples

judgmental: in this kind of sample, selections are made based on

pre-determined criteria that, in your judgment, will provide the data
you need.
For example, you may want to interview primary school principals
and decide to interview some from rural areas as well as some
from urban areas (but no quota is established).
convenience: in this type, selections are made based on the
convenience to the evaluator. Principals from local schools may be
selected because they are near where the evaluators are located.
Combinations

 Random and non-random methods can be combined. For

example, you may be collecting data on schools.
 You can select two schools from the poorest communities and
two from the wealthiest communities.
 Then from these four schools, you can randomly select students
for your data collection.
Confidence levels and precision
 Even when you use a random sample, there is some possibility of error.
It is possible that your sample will be different from the population. This is
where statistics come in.
 The first thing you want to do is decide how confident you wish to be
that your sample results are an accurate estimate of what is true for the
entire population.
 Your next choice is about how precise you need your estimates to be.
 This is sometimes called sampling error or margin of error.
 The more precise you want to be, the larger your sample will need to be.