STATISTICS

FC301

Week 11

Sampling methods
 OBJECTIVES

• Explain what is meant by a representative sample.

• Describe simple random sampling.
• Describe systematic random sampling.
• Describe cluster sampling
• Describe stratified sampling.
 TERMINOLOGY
Population A collection of individuals or Students at Kaplan
items. Colleges in the UK
Sample A selection of individuals from A particular group of
given population (we use a students
sample to help us make
generalisations about the whole
population).
Census A collection of data relating to a Kaplan’s students
whole population. records
Sampling unit An individual member of a Particular student
sample. from the sample
Sampling frame A list of all the items in the List of all the
population, from which the students at Kaplan
sample can be selected. Colleges
 THE PURPOSE OF SAMPLING (1)
There may be various reasons why we need to take a sample of
a population rather than asking the whole population.
q Cost
Any research tends to be costly. Imagine the cost of creating the
materials, paying the researchers, etc, to interview the entire
population.
q Time
If we were to survey the whole population, this could be very time-
consuming. Taking a representative sample of the population allows
us to spend more time on each sampling unit, and perhaps obtain
more detailed information.
 SELECTING THE MEMBERS OF A SAMPLE

§ A great deal of thought goes into selecting an

appropriate sample.
§ How closely a sample is likely to approximate the
characteristics of the population depends on the sample
process.
§ There are many different types of samples.
§ Which one is chosen may depend upon a variety of
factors.
 PROBABILITY AND NON-PROBABILITY
SAMPLES
qProbability sampling
•Any sampling method in which all population members
have an equal chance of being selected.
• This requires a sampling frame listing all members of the
population.
q Non-Probability sampling
• Samples from which some of the population are excluded.
• Only used when there is no appropriate sampling frame
 NON-RANDOM SAMPLING METHODS
q Non-Probability sampling
§ Convenience sampling: found through observation, e.g.
waiting outside a shop and asking people as they walk
past
§ Quota sampling: requires some information on the
population and choosing only specific members of the
population who meet some criteria.
§ Self-selected sampling: there are volunteers for the
sampling.
§ Snowball sampling: respondents find other respondents.
 Convenience Sampling

Data is chosen based on convenience.

Advantage
´ Cuts down the cost of preparing a
sampling frame as it is less time-
consuming.
Disadvantage
´ Bias may exist, as it does not represent
the population well.
 Quota Sampling
The sample has the same proportions of individuals as
the entire population for known characteristics, traits
or focused phenomenon.
Advantages

´ Relatively easy to administer. Can be performed

quickly
´ Cost-effective accounts for population proportions
Disadvantages

´ Sample selection is not random

´ Selection could be biased, leading to an
unrepresentative population
´ Items that cannot be categorised into certain
groups
 PROBABILITY SAMPLING
§ Requires a sampling frame, which is the full set of
individuals that have a chance of being included in the
sample procedure

§ A sample will only be representative of the population

included in the sample frame

§ Sample frames should be evaluated for:

ü Comprehensiveness
ü Probability of selection
ü Efficiency
 PROBABILITY SAMPLING METHODS

§ Simple random sampling: people randomly selected from

the population.

§ Systematic sampling: randomly select an item and then

take evenly spaced people from a list.

§ Stratified sampling: the population is broken into

subgroups, e.g. religious denomination, age, type of worker

§ Cluster sampling: we put the population into clusters

 Simple Random Sampling
Random sampling is a sample chosen from a population
using a random number table or generator. Each member
of the population has an equal, independent and known
chance of being selected.
Advantages
´ Easy to implement.
´ Each member of the population has an equal chance
of being chosen.
´ Free from bias.
Disadvantages
´ If the sampling frame is large, random sampling may be
impractical.
´ A complete list of the population may not be available.
´ Minority subgroups within the population may not be
present in the sample.
 SIMPLE RANDOM SAMPLING
Simple random sampling: A sampling procedure for which
each possible sample of a given size is equally likely to be
the one obtained.

Simple random sample: A sample obtained by simple

random sampling.

You essentially just close your eyes and spin around and
then see who you’re pointing at and take them in your
sample. Then repeat until you have the number of people
you wanted in the sample. You can also use a random
number generator to pick people in you sample
 Systematic Sampling
All data is sequentially numbered, and every piece of
data is chosen. The number n is chosen by n=size of
population desired population size.=size of population
desired population size.
Advantages
´ Easy to select.
´ Identified easily.
´ Evenly spread over the entire population.
Disadvantages
´ May be biased where the pattern used for the
samples coincides with a pattern in the population.
 Stratified Sampling
The population is divided into subgroups (strata) based
on specific characteristics like age, gender or race.
Within the strata, random sampling is used to choose the
sample.
Advantages
´ Strata can be proportionally represented in the final
sample.
´ It is easy to compare subgroups.
Disadvantages
´ Information must be gathered before dividing the
population into subgroups.
 16

Example: In a report on conditions in the workplace, it was

decided to conduct a survey in three factories, A, B and C,
in a particular area, employing 180, 126 and 144 workers
respectively. Explain how we can obtain a sample of 30
workers, using stratified sampling with proportional
allocation.
STEP 1 Divide the population into strata
Strata 1 Strata 2 Strata 3
A B C
180 126 144

STEP 2 From each stratum obtain a simple random

sample of size proportional to the size of
the stratum; that is:

THE SAMPLE SIZE FOR A STRATUM EQUALS THE TOTAL

SAMPLE SIZE TIMES THE STRATUM SIZE DIVIDED BY THE
POPULATION SIZE
Number of workers in Strata 1 180
18 ´ Total sample size = ´ 30 = 12
Total number of workers 450

Number of workers in Strata 2 126

´ Total sample size = ´ 30 = 8.4 » 8
Total number of workers 450

Number of workers in Strata 3 144

´ Total sample size = ´ 30 = 9.6 » 10
Total number of workers 450

STEP 3 Use all the members obtained is Step 2 as

the sample
Remember to
round to a
whole number
Sample size = 12 + 8 + 10 = 30 of people!
Cluster Sampling
Data is divided into clusters, and random sampling is
used to select whole clusters. The sample will be
obtained from a collection of entire cluster groups. It is
usually used with naturally occurring groups of
individuals, for example, classrooms, city blocks or
postcodes.
Advantages
´ Cuts down the cost and time by collecting data from
only a limited number of groups.
´ Can show grouped variations.
Disadvantages
´ It is not a genuine random sample.
´ The sample size is smaller, and thus, the sample will
likely be less representative of the population.
 20

Example: London city council wants to install bike paths in the

city, so they decided to take a sample of London residents and
cluster sampling is a good idea as it doesn’t privilege any
particular region of the city.
STEP 1 Divide the population into groups (clusters)
The residential portion
of the city was divided
into 5 blocks (SE, SW,
N, E and NW), each
containing ~10000
homes.
STEP 2 Obtain a simple random sample of the clusters

The planner numbered the clusters from 1 to 5 and then

used a table of random numbers to obtain a simple random
sample of 2 of the 5 blocks. Let’s say it’s SE and N.

STEP 3 Use all the members of the clusters obtained

in Step 2 as the sample.
The sample consisted of all the homes in the SE and N, so
that is about:
2 regions ´ 10000 homes per block = 20000 homes
So the people in these 20000 homes form our cluster
sample
 Summary

´ Primary data is data collected yourself

´ Secondary data is collected by others and presented as a summary or
in tables
´ When collecting data, always keep notes of how it was collected,
precise definitions of variables
´ Various types of random sampling
´ Simple
´ Stratified
´ Cluster
´ All the above can be combined in Multi-stage Sampling
´ Type of sampling affects the size of standard errors
´ Quota sampling is non-random sampling and cheaper
´ Sampling frame is the list from which the sample is
taken.
´ Careful interview techniques needed to ensure
reliable answers
 Task

Find the Gross Domestic Product

(GDP) for the United Kingdom
and the United States from 2000
to 2020. Obtain both series at
constant prices.