Inferential Statistics:
Sampling
Objectives
By the end of this lesson, you should be able to;
• Describe sampling
• Describe the Types of sampling
• State the advantages of sampling
Terms
• Inferential statistics
• Population
• Sample
• Sampling
Central Limit theorem
• As samples become larger, the shape of the
sample distribution becomes more
normalised.
Inferential Statistics
• Defined as statistical methods whereby a small
number of participants are chosen to represent the
population.
A sample
Population
Sampling definition
• Sampling is defined as the selection of a sample from the population.
• The population is referred to as the entire participants
• While a sample is a small group of participants obtained from the population
The general A sample
population
of 12
regions Infer the parameters to
the population Calculate the parameters of
your interest eg Frequency
mean, risk, prevalence,
incidence
Advantages of sampling
Types of sampling
There are two types of sampling;
• The probability sampling
• The non probability/Purposeful sampling: any
sampling that is not in keeping with EPSEM
criteria.( Equal probability of selection sampling)
Probability sampling
• Probability sampling is defined as the sampling
whereby the members of the sample selected have
the same chance/probability of being selected.
• 5 types
• Simple random sampling
• Cluster sampling
• Stratified sampling
• Multistage sampling
• Systematic sampling
Simple Random Sample(SRS)
• Simple random sampling(SRS) is a method of choosing
cases within a population where all cases have a equal
chance of being selected.
E.g. If you have a population of say 4 people like a, b, c and
d. The probability of selection of ;
a= ¼ and this should be equal to the probability of
selection of
b=1/4 or
c=1/4 or
d=1/4.
Simple Random Sampling (SRS)
To do the SRS one has to have;
• All members of the population (sampling frame) available and known.
• Decide on how many (sample size) should be selected from the population.
• Obtain the sample without bias and by chance.
i.e. either by;
i. lottery method or
ii. using the table of random numbers
iii. Random number generator
1. SRS by Lottery method.
• Take all members in the sample frame in one pot, do lottery to choose the
members. Every time you select the member, the selected member should be
noted down and replaced to the lottery pot so that the probability of selecting
the members should be the same.
…SRS
2. SRS by use of table of random numbers
• Label and table
• List all population and number them.
• Determine how many number digits you have.
• How many people do you want to sample?
• Choose which way to go. i.e. left, right, up or down
• Without looking at the table, choose where to start.
• Obtain the number in your list, follow the way you choose until all
participants are obtained.
Exercises
Use a table of random numbers to select 10 students from the
classroom.
…SRS
3. SRS using the computer software
• Computer software can be used to select the
participants at random.
• Its convenient and reliable for a large population
What if the list(sampling frame) isn't
known/available?
• Simple random sampling requires a sampling frame
(list of all participants) to be known.
• However, there are some cases where the exact
sampling frame either isn’t known or is not available
to make SRS.
E.g.
i. Sampling of people in the households for
household survey of malnutrition
ii. Sampling of children less than 5 years at the
clinics for the study of prevalence of malnutrition
…Sampling frame not known
• In this case, it is convenient then to group
participants.
• The groups can have 2 properties.
i. The members within the group can have similar
characteristics as possible(Homogenous) or
different as possible (Heterogeneous) groups.
ii. The groups can be similar to each other as
possible(Homogenous) or different as
possible(Heterogeneous)
Cluster Sampling
• In cluster sampling, geographic units are sampled and cases are
randomly sampled within those geographic units. The members
within the cluster should be heterogeneous as possible, while
between clusters should be homogenous as possible.
• E.g. hospital wards—Pediatric, Gynecology and Psychiatric wards
are all wards intended to admit patients(homogenous). However
within the ward like Pediatric ward, children have different
diseases like Flu / Fracture /HIV/TB etc (Heterogeneous).
How to do cluster sampling?
• Group or assign the groups with numbers/names/letters.
• Choose the clusters you wants at random using SRS.
Stratified Sampling
• Strata (subgroups) are the sampling units.
• Intra strata should be as homogenous as
possible.
• Inter strata should be as heterogeneous as
possible.
• E.g. religion, occupation, age groups.
Multistage Sampling
Systematic sampling
Non probability sampling
• Selection is done without the rule of chance.
• Equal probability of selection sampling not
adhered to.
• Examples:
- Convenience sampling; sample obtained
conveniently. Sample unrepresentative.
- quota sampling= fixed number of people. A
fixed, predetermined sample size is obtained
from a population.
Exercises
Exercise
The researcher is expected to collect the representative
sample using the probability method. Suggest(to him) the
sampling method(s) to be done in selecting the following;
i. Prevalence of HIV in Botswana.
ii. Association between marijuana use and HIV risky
factors among adolescents.
iii. Alcohol use in college students in Gaborone
iv. The use of Blackboard in students at Botho University
v. Prevalence of Diabetes Mellitus in Bot
