SAMPLING

1.0 INTRODUCTION
In data collection, an ideal situation is to canvas or collect data from all elements of a population.
This is taking a census (or complete “count”) of a population. However, this is not always
practical for many reasons which include.
Cost: censuses are vast undertakings requiring large amounts of equipment, materials, and
human resource.
Time: Censuses require time in planning, data Collection and analysis.
Depth of data: Because of the size of the undertaking, it is not possible to ask in depth questions,
instead of a census, a sampling is used.
A sample is a group of selected elements of the population.
Sampling refers to the process of selecting the elements from the population.
For many enquiries, sampling is preferred because of costs are lower compared to censuses, it
takes shorter time and in-depth data can be collected. However, sampling has the problem of
representation, and accuracy. These can be mitigated by using an appropriate sampling technique
and increasing the sample size (the number of items selected into a sample.)
1.1 IMPORTANT TERMS
1.1.1 Population
A population is the collection of all people or items with the characteristics that one wishes to
understand in an enquiry.
Example
For a survey collecting data on family size in a district. The populations are all the families in the
district.
1.1.2 Population size
The numbers elements in the population. This number is often denoted by N.
1.1.3 Sample
A sample is a collection of items selected from the population. A sample is therefore a part of the
population.
1.1.4 Sampling frame
This is an itemized list of the items from which a sample is to be drawn. The term sampling
frame is different from population because it refers to a situation where the elements of a
population can be and have been individually identified for purposes of sampling. With a
sampling frame it is possible to state exactly the proportion of the population/frame which has
been sampled. Not all populations render themselves to the construction of a sampling frame.
1.1.5 Sample size
This is the number of items in a sample. Sample size is often denoted by n. For example, given a
population has 250 element and 25 items are selected into a sample, then the sample size n is 25.
1.2 SAMPLING TECHNIQUES
1.2.1 Major groupings of sampling techniques.
a) Random or scientific techniques:
These are techniques where sample elements have a defined chance of being selected into the
sample.
b) Non-Random or non-scientific:
Sample items do not have a defined probability of being selected.
1.2.2 Sampling techniques.
Overview
The most common sampling designs are simple random sampling, stratified random sampling,
and
multistage random sampling.
The simple random sampling (SRS)
Simple random sampling is a random (scientific) sampling technique where each individual is
chosen entirely by chance and each member of the population has an equal chance of being
included in the sample.
A simple illustration of simple random sampling is what happens with raffle ticket draws.
Several tickets are crumpled thrown into a drum and reshuffled. A person is blind folded and
asked to pick a ticket or tickets. The tickets so picked are picked purely at random
In the case of sample surveys random numbers are used to achieve this randomness. Random
numbers are number often generated by computer and they are such that one cannot predict any
particular number from the set. They do not have a pattern nor is it possible to fit a formula to
them.
Example 1:
Given a list of people in the following table clearly identified and numbered from 01 to 20, select
a simple random sample.
Sampling frame

From the random number table, going vertically, the following are the first 5 random numbers
between 01 and 20: 12, 08, 14, 19, and 09
Thus, the elements of the sampling frame selected into the sample are:
12 M Mitepa*
08 A Lwanda*
14 M Chalola*
19 E Kankodo*
09 J Sikwese*
Simple random sampling has the following advantages and disadvantages:
Advantages
i. It minimizes bias
ii. Because of low bias the sample is usually representative.
iii. The method is generally accepted by layman as fair
Disadvantages
i. The sample is venerable to sampling error
ii. The elements selected may not reflect the population.
iii. it may be Cumbersome and tedious.
iv. it will always require a sampling frame.
v. There is chance that certain significant attributes of the population are over or under
represented Systematic sampling
Systematic sampling is a technique where sample items are selected at a regular interval from a
random start. The interval at which the population elements are picked is called a sampling
interval. That is sampling is carried out by picking say every 5th, 8th, etc item depending on the
sampling interval determined.
The sampling interval is calculated as follows
Let k = the sampling interval
N = the population size
n = the sample size
Then k = N/n
Example 2:
Select a sample of 5 people from the list shown in 3.2.1 above using systematic sampling
Solution
N = 20, n = 5 the sampling interval k = 20/5 = 4
In such a case, sampling proceeds by taking every 4th item from a random start in the first 4
elements of the sampling frame (the first interval).
Using the same random number table, the first number between 01 and 04 is 01.
This means starting with item 01 selecting every 4th element from the population, we have:
01, 05, 09, 13, and 17
The names are:
1 D Magwera
5 P Chisambo
9 J Sikwese
13 I Jemu
17 J Nyirenda
Advantages
i. The method is practical and easy to carry out on the ground.
ii. It does not require a strict sampling frame. It only requires the arrangement of the elements in
the population.
Disadvantage
i. Systematic sampling can produce biased results if the population has periodicity in the
elements. That is the arrangement forms a pattern and the sampling interval resonates with that
pattern then only elements of a similar type will be selected.
ii. The method is not truly random since (once a random starting point has been selected) all
subjects are pre-determined.
Stratified sampling
This is a sampling technique where the population is first divided into categories according to
pre
existing characteristics in the population. The categories are called “strata” (Single stratum).
A sample is then drawn from each stratum and results combined later. For example, a survey on
industrial turn over industries can be stratified as follows:
Stratum 1 Large industries
Stratum 2 medium
Stratum 3 small
Stratified sampling techniques are generally used when the population is heterogeneous, or
dissimilar, and certain homogeneous, or similar, sub-populations can be isolated (strata). Simple
random sampling is most appropriate when the entire population from which the sample is taken
is homogeneous. Any appropriate sampling technique can then be used in the individual stratum:
Th general procedure for taking a stratified sample is:
i. Stratify the population, defining a number of separate partitions.
ii. Calculate the proportion of the population lying on each partition.
iii. Split the total sample size up into the above proportions.
iv. Take a separate sample (normally simple random) from each partition, using the sample sizes
as defined in (iii).
v. Combine the results to obtain the required stratified sample.
Advantages
i. It enables researchers to draw inferences about specific sub groups. That may be lost in a more
general random sample.
ii. It leads to better statistical estimates.
iii. Stratified sampling makes data more readily available for pre-existing subgroup within a
population.
iv. If strata are treated independently different sampling technique can be used depending on the
data required and a situation on the ground.
Disadvantages
i. The technique requires selection of a relevant clarification criteria.
ii. It can be expensive due to extra time and manpower required for the organization and
implementation of the sample.
iii. It requires an extensive sampling frame.
iv. Strata levels of importance can only be selected subjectively.
v. It is not useful when there are no homogeneous subjects.
Multistage sampling
A multistage sample is constructed by taking a series of samples in stages starting with wider
area or definition of units to narrower and more specific units. The method is used where a
population is spread over a relatively wide geographical area and SRS would require travelling to
all parts of the area.
The method is as follows:
i. Split the area up into a number of regions
ii. Randomly select a small number of regions
iii. Confine sub-samples to these regions alone with size of each sub-sample proportional to the
size of the area.
iv. The above procedure can be repeated for sub-regions within regions.
Once the final regions have been selected the final sampling technique could be random or
systematic depending on the existence or otherwise of a sampling frame.
Example 3:
A researcher wants to estimate the proportion of household above a poverty line in terms of
income in rural Kenya. Help the researcher design a sample.
Solution:
Taking the country as a whole the sample design would as follow:
Stage 1 Pick districts using SRS
Stage 2 From each district select a number of locational Areas
Stage 3 From each locational Area select a number of Village areas
Stage 4 From each Village area select the required number of households.
The number of units (districts, locations, villages and households) being sampled at each stage
depends on the accuracy required. Larger samples tend to give better accuracy.
Advantages
i. Cost of canvassing is low
ii. Data can be collected with relative speed due to ease of organization the technique can give.
iii. It has convenience of finding the survey sample
iv. Normally more accurate than cluster sampling for the same size sample
Disadvantages
i. Is not as accurate as SRS if the sample is the same size
ii. More testing is difficult to do
iii. Possible bias if a very small number of regions is selected.
iv. No member of the population in any other region can be selected.
Cluster sampling
Cluster sampling is used where it is difficult or impossible to isolate the target responding units
(elements of the population from which data is collected). In such cases it is more cost-effective
to select respondents in groups ('clusters'). Grouping or clustering is often based on geography,
or by time periods.
Data is then collected from all respondents in a selected cluster.
Example 4
An NGO has read in the papers that very few people use mosquito nets in Bondo-Kisumu. To
confirm this, the NGO would like to collect first hand data necessary for intervention.
However, the Director is at pains as to what sampling method can be used because the dwelling
units are not numbered and they are not arranged in proper rows. He has turned to you for help in
the matter.
Suggest a sampling method.
Solution
The proper sampling technique is cluster sampling. Much as the dwelling units are not numbered
or arranged in rows, there are small access roads which form a network. The roads form
boundaries of groups or clusters of houses. These clusters can be identified and an SRS can be
picked and data
collected from every household in a selected cluster.
Example 5:
A manager in Peoples Trading Centre would like to find out from shoppers whether or not they
like the new arrangement of goods. Help him draw a sample of the shoppers.
Solution
It is difficult to identify all shoppers individually beforehand and therefore a sampling frame
cannot be constructed. The best way is to cluster shoppers on time. Possible clusters are:
Treating shoppers that enter the shop starting at 08:00 can be grouped in periods of 15 minutes
up to say close of the shop at 20:00. A systematic sample of these groupings can be made. One
possible such sample is to pick the groupings every two hours as follows
08:00 – 08:15
10:00 – 10:15
12:00 – 12:15
14:00 – 14:15
16:00 – 16:15
18:00 – 18:15
Note that a SRS sample can also be drawn since the time clusters can be uniquely identified.
Advantages
i. Clustering can reduce travel and administrative costs. An interviewer can make a single trip to
visit several responding units in one block, rather than having to drive to a different block for
each unit.
ii. It also means that one does not need a sampling frame listing all elements in the target
population.
Instead, clusters can be chosen from a cluster-level frame, with an element-level frame created
only for the selected clusters.
Disadvantages
i. Cluster sampling requires a larger sample than SRS to achieve the same level of accuracy
because of the variability among clusters.
ii. The fact that the method is not random sampling then the selection bias could be significant.
Quota sampling
Quota sampling is one of the non-scientific techniques of sampling. The researcher decides on
the number of sampling units (i.e., sample size) according to the required coverage (purely from
judgement). The data collector then collects data from any responding units in the population up
to the number arrived at earlier.
Advantages
i. The technique is simple and convenient
ii. There is no non-response
iii. The cost of data collection is very low,
iv. Quota sampling does not require a sampling frame.
Disadvantage
i. The technique produces very biased results because of the human interference in the sample
selection.
ii. Severe interviewer bias can be introduced into the survey by inexperienced or untrained
interviewers, since all the data collection and recording rests with them

ASSIGNMENT 1
1 Define and give examples of the following terms
a) Population
b) Sampling frame
c) Sampling unit
d) Random numbers
e) Responding unit
2 Compare and contrast probability and non-probability sampling techniques.
3 Systematic sampling may produce a biased sample. Discuss
4 How can you collect a sample using the following sampling techniques
Simple random sampling
Cluster sampling
Multistage sampling
Stratified sampling
5 Suggest a suitable sampling technique in each of the following situations Describe the
technique
and state reason for its suitability.
(i) In an audit you wish to form an opinion from a number of invoices received throughout
the year.
(ii) You are trying to estimate maize field over a 5-hectare field.
(iii) You want to gather opinions from spectators on the strength of two soccer teams just
before a match.