Learning objectives
• By the end of the session, the students will be able to:
• Define sampling and discuss the importance of sampling in research
• Distinguish between probability and non probability sampling
• Discuss the relative advantages and disadvantages of each sampling method
Important terms to note
• Sampling – process of selecting the elements from a defined study
population of interest so as to study the sample and therefore generalise
the findings to the population
• A population - a set or group of items or elements of interest
• Target Population – population to which the researcher intends to
generalise his/her results
• Study population – population to be studied or the actual sampling frame
from which the sample is drawn
• Should clearly defined for example sex, age, residence, villages etc
2
Cont’
• Study/sampling unit – the smallest unit from which the sample is selected
from and these are units on which measurements are made
• Compose the study population
• Depends on the research questions and objectives to be addressed
• Sample – a subset of a population or collection of sampling units from the
eligible population
• Sampling frame – list of all the sampling units from which sample is drawn
3
• Sampling method – technique used to select sampling units from sampling
frame
Cont’
4
Target
population
Sample
Study
population
Why sample?
5
• Unable to study all members of a population
• Reduce bias
• Save time and money
• Measurements may be better in sample than in entire population
• Feasibility
Cont’
6
• When can you study the entire population?
• If population of interest is too small
• Adequate resources are available
• If you do not anticipate a very high response
Representativeness
• The aim of any sample is to represent the characteristics of
the sample frame
7
• Factors that influence the representativeness of the sample
• Sampling procedure
• Sample size
• Participation/response rate
Probability sampling methods
• Each sampling unit has an equal and nonzero probability of being included
in the sample
8
• A probability sample commonly is more accurate than a census because
there are more rigorous controls that reduce nonsampling errors
• Probability samples – findings can be generalised from the sample to the
population
• Precision of estimates can be computed from the sample
Cont’
• Probability sampling procedures include
• Simple random sampling
9
• Stratified random sampling
• Systematic random sampling
• Cluster sampling
• Multistage sampling
Simple random sampling
• Each sampling unit has an equal chance of being selected for the sample
• Applicability – if population is small, homogeneous and readily available
10
• The researcher prepares a complete list of all elements of the population of
interest (sampling frame)
• From the list, a sample of units is drawn such that each unit has an equal chance of
being selected into the sample
• Use a table of random number (computer) or lottery system to determine which units are to be
selected
• Samples are drawn either with or without replacement
• In most surveys, sampling without replacement is most common
Cont’
11
• Advantages
• Simple and easy to understand
• Easy calculation of estimates
• Disadvantages
• Not most efficient method, that is, not the most precise estimate for the cost
• Requires knowledge of the complete sampling frame
• Cannot always be certain that there is an equal chance of selection • Non-
respondents or refusals
Cont’
12
• Example: Estimate haemoglobin levels in patients with sickle cell
anemia in Mulago hospital in Uganda
• Determine sample size
• Obtain a list of all patients with sickle cell anaemia in the hospital
• Patient is the sampling unit
• Use a table of random numbers to select units from the sampling frame
• Measure haemoglobin in all patients
• Compute mean and standard deviation of sample
Stratified random sampling
13
• The sampling frame is organised into mutually exclusive homogenous
groups or strata based on certain characteristics of interest to the
research
• From each stratum, an independent simple random sample of units is
drawn
• Every unit in a stratum has the same chance of being selected
Cont’
• Advantages
14
• Ensures representation of that certain subgroups in a sample
• Allows estimation of parameters in different strata
• There is more precision of estimates of the parameters because strata are more
homogeneous
• Strata of interest can be sampled most intensively
• Disadvantages
• Precision is lost if small number of units is selected from strata
Cont’
• Example: To assess dietary intake in adolescents
15
• Compute sample size
• Stratify by age groups 11-13, 14-16, 17-19 and sex
• Obtain list of children in this age range from schools
• Randomly select children from each of the 6 strata until sample size is obtained
• Measure dietary intake
Systematic random sampling
16
• Systematic sampling depends on arranging the target population in some
ordering format
• At regular intervals, the elements are selected from the ordered list
• It involves a random start and then proceeds with the selection of every kth
element from then onwards
• In this case, k=(population size/sample size).
• The starting point is not automatically the first on the list, but is randomly
selected from within the first to the kth element in the list
Cont’
17
• Example:
• You have a list of 10,000 people and you want a sample of 1,000.
• To create a sample includes three steps:
• Divide number of the population size by the desired sample size. Dividing 10 ,000
by 1,000 gives a value of 10
• Select a random number between one and the value attained – 10 to get the
random start which could be 5
• Then start with case number chosen – 5, and select every tenth number (5,
15 , 25, etc. )
Cont’
18
• Advantages
• Sampling frame does not need to be defined in advance
• Easier to implement
• If there are unrecognized trends in the sample frame, systematic sample ensure
coverage of the spectrum of units
• Disadvantages
• Difficult to compute precision of estimates unless assumptions are made
Cluster sampling
19
• Divide the population into clusters of homogeneous units, commonly geographical or
administrative units like villages, parishes etc
• A simple random sample of clusters is first drawn
• Within each cluster,
• If all of the elements of the clusters selected are sampled - one-stage cluster sampling
• If a random sample of the elements of each selected cluster is drawn – two-stage cluster sampling
• Done when researchers cannot compile a comprehensive list of all the elements in the
population of interest
20
21
Cont’
• Advantages
• The entire sampling frame does not need to be enumerated in advance
• More economical in terms of resources than simple random sampling
• Often used to evaluate vaccination coverage in EPI
• Disadvantages
• Precision is lost, i.e. wider variance, however can be accounted for with higher
number of clusters
22
Important to note
• Cluster sampling is used in most large scale surveys
• Cluster sampling may be used with stratification, clusters within strata
• For a particular sample size, the number of cluster samples are less
accurate than other sampling methods because the parameters will
have a greater variability than simple-, stratified- and systematic
random samples
23
Cont’
• Example: To estimate the prevalence of dental caries in school children
• Among the schools in the catchment area, list all of the classrooms in each school
• Take a simple random sample of classrooms, or cluster of children
• Examine all children in a cluster for dental caries
• Estimate prevalence of caries within clusters than combine in overall estimate
Multistage sampling
24
• Similar to cluster sampling in which two or more stages or levels of
sampling events
• Primary units are randomly selected
• Individual units within primary units randomly selected for measurement
• It is used frequently when a complete list of all elements of the
population does not exists
• Multistage sampling avoids large sample sizes and unnecessary costs
associated with traditional cluster sampling
Cont’
25
• For example:
• 1st stage, a random number of districts is chosen from all regions
• 2nd stage, a random number of parishes is selected
• 3rd stage units, a random number of villages is selected
• Either all or a sample of households are randomly selected for the survey
Cont’
• Example: To estimate the prevalence of dental carries in school children in
Jinja district
26
• Among the schools in the catchment area, list all of the classrooms in each school
• Take a simple random sample of classrooms, or cluster of children
• Enumerate the children in each classroom
• Take a simple random sample of children within the classroom
• Examine all children in a cluster for dental caries
• Estimate prevalence of caries within clusters than combine in overall estimate
Non-probability sampling methods
27
• Sampling method where elements of population have no chance of selection or
where the probability of selection is unknown
• It is cheaper, easy to execute but findings cannot be generalised to the population
• Subject to biases
• The researcher cannot compute estimates that indicate precision of sampling •
These include:
• Convenience sample
• Quota sample
28
• Snowballing
• Purposive
Convenience/Availability sampling
• A non-random collection of sampling units from a part of the population that is
readily available and convenient
• Most useful for pilot testing
• Advantages
• Convenient and easy to perform
29
• Inexpensive to create samples - money and time needed is not as much as that for other
probability sampling methods
• Easily collectible samples – sample is easily accessible to the researchers and.
• Disadvantages
• The researcher cannot generalise their findings
Quota sampling
• The population is first divided into mutually exclusive sub-groups or
quotas
30
• Using a specified proportion, units or subjects are selected from each
sub-group
• In quota sampling the selection of the sample is non-random
• It is this second step which makes the technique one of nonprobability
sampling methods.
Snowballing
• This method is used when the study is focused on participants with specific
characteristics of interest
31
• It relies on the study subjects being used to recruit more subjects with the
characteristics of interest into the sample
• The researcher first identifies a subject of the population of interest, then speaks to
him/her and then requests that person to identify others in the population that the
researcher might speak to
• This method is not representative but is useful, especially where the groups
in the research are not socially organised
Cont’
• Advantages
• It is more efficient in finding the sample
32
• It is a convenient approach and therefore not as costly like other sampling methods
• Researchers can trace individuals who not ordinarily participate in research studies
• Disadvantages
• Sampling bias and margin of error are if concern because people refer those whom they
know and have similar traits this sampling method
• Lack of cooperation – referrals may not be cooperative and refuse to participate in the
research studies.
• It is not representative but is useful, especially where the groups in the research are not
socially organised
Purposive sampling
33
• The researcher uses his or her own ‘expert’ judgment to decide who is
to be included in the sample frame
• Prior knowledge and research skills are used in sampling the
respondents or elements
• Drawback - Statistics that measure the precision of the estimates
cannot be computed
Consecutive sampling
34
• Samples are picked at the ease of a researcher more like convenience
sampling, only with a slight variation
• Researcher selects a sample or group of people and conduct research
over a period of time, collect results, and then moves on to another
sample.
• Researcher has a chance to work with multiple samples to fine tune
his/her research work to collect vital research insights
Cont’
• Advantages
35
• The researcher has many options for sample size and sampling schedule.
• It enables the researcher to make changes and adjustments at the start of
the research to avoid considering research bias due to its repetitive nature.
• Very little effort is needed from the researcher’s end to carry out the
research.
• Disadvantages
• It does not yield a representative sample. However, if large sample sizes are
used to represent the populations, then it tries to get closer to
representativeness.
• Findings and conclusions are not representative
36
Bias in sampling
• This is a systematic error in sampling procedures that results into distortion
of study results
• Improper sampling can also introduce bias which leads to poor sample
representativeness
• Several possible sources of bias in sampling include:
• Non-response
• Studying volunteers only
37
• Sampling of registered patients only
• Seasonal bias
• Tarmac bias
End!!!!!!
38
