Definition of sampling

population
are selected as representatives of the entire
Procedure by which some members of the population
National Institute of Epidemiology NIeCer 101 National Institute of Epidemiology NIeCer 101
Chennai HEALTH RESEARCH FUNDAMENTALS Chennai HEALTH RESEARCH FUNDAMENTALS

[Link]
[Link]
2/1/2016

1
 Population

Study population

the results of the study will be inferred

Sample

The study population is the population to which

National Institute of Epidemiology NIeCer 101 National Institute of Epidemiology NIeCer 101
Chennai HEALTH RESEARCH FUNDAMENTALS Chennai HEALTH RESEARCH FUNDAMENTALS

[Link]
[Link]
2/1/2016

2
 2/1/2016

The study population depends upon the

research question

HEALTH RESEARCH FUNDAMENTALS

• How many injections do people received each
year in India?

NIeCer 101
• Study population: Population of India
• How many needle-sticks health care workers

National Institute of Epidemiology

Chennai
experience each year in India?
• Study population: Health care workers of India
• How many hospitals have a needle-sticks
prevention policy in India
• Study population: Hospitals of India
[Link]

The sample needs to be representative

of the population in terms of time

HEALTH RESEARCH FUNDAMENTALS

• Seasonality
NIeCer 101
• Day of the week
• Time of the day
National Institute of Epidemiology
Chennai

[Link]

3
 2/1/2016

The sample needs to be representative

of the population in terms of place

HEALTH RESEARCH FUNDAMENTALS

• Urban

NIeCer 101
• Rural

National Institute of Epidemiology

Chennai
[Link]

The sample needs to be representative

of the population in terms of persons

HEALTH RESEARCH FUNDAMENTALS

• Age
NIeCer 101
• Sex
• Other demographic characteristics
National Institute of Epidemiology
Chennai

[Link]

4
 2/1/2016

Definition of sampling terms

HEALTH RESEARCH FUNDAMENTALS

• Sampling unit (Basic sampling unit, BSU)
• Elementary unit that will be sampled

NIeCer 101
• People
• Health care workers

National Institute of Epidemiology

Chennai
• Hospitals
• Sampling frame
• List of all sampling units in the population
• Sampling scheme
• Method used to select sampling units from the
sampling frame [Link]

Why do we sample populations?

HEALTH RESEARCH FUNDAMENTALS

• Obtain information from large populations
NIeCer 101
• Ensure the efficiency of a study
• Obtain more accurate information
National Institute of Epidemiology
Chennai

[Link]

5
 2/1/2016

Population Sample

HEALTH RESEARCH FUNDAMENTALS

NIeCer 101
National Institute of Epidemiology
Chennai
•Infinite/finite size
•Characterized by unknown
parameters

[Link]

Practical example

HEALTH RESEARCH FUNDAMENTALS

• The Ministry of Health of a country X wants to
estimate the proportion of children in NIeCer 101

elementary schools who have been immunized

National Institute of Epidemiology
Chennai

against childhood infectious diseases

• The task must be completed in one month
• The objective is to estimate the proportion of
immunized children
[Link]

6
 2/1/2016

Type of samples

HEALTH RESEARCH FUNDAMENTALS

• Non-probability samples
• Probability of being selected is unknown

NIeCer 101
• Convenience samples

National Institute of Epidemiology

Chennai
• Biased
• Best or worst scenario
• Subjective samples
• Based on knowledge
• Time/resources constraints
• Probability samples
[Link]

Type of samples

HEALTH RESEARCH FUNDAMENTALS

• Non-probability samples
NIeCer 101
• Probability samples
• Every unit in the population has a known probability of
National Institute of Epidemiology
Chennai

being selected
• Only sampling method that allows to draw valid
conclusions about population

[Link]

7
 2/1/2016

Random sampling in probability

samples

HEALTH RESEARCH FUNDAMENTALS

• Removes the possibility of bias in selection of
subjects

NIeCer 101
• Ensures that each subject has a known

National Institute of Epidemiology

Chennai
probability of being chosen
• Allows application of statistical theory

[Link]

Sampling error

HEALTH RESEARCH FUNDAMENTALS

• No sample is a perfect mirror image of the
population NIeCer 101

• Magnitude of error can be measured in probability

samples
National Institute of Epidemiology
Chennai

• Expressed by standard error of mean, proportion,

differences…
• Function of:
• Sample size
• Variability in measurement [Link]

8
 2/1/2016

Methods used in probability samples

HEALTH RESEARCH FUNDAMENTALS

1. Simple random sampling

NIeCer 101
2. Systematic sampling
3. Stratified sampling

National Institute of Epidemiology

Chennai
4. Cluster sampling
5. Multistage sampling

[Link]

1. Simple random sampling

HEALTH RESEARCH FUNDAMENTALS

• Principle
• Equal chance for each sampling unit
NIeCer 101
• Procedure
• Number all units
National Institute of Epidemiology
Chennai

• Randomly draw units

• Advantages
• Simple
• Sampling error easily measured
• Disadvantages
• Need complete list of units
• Does not always achieve best representation
[Link]

9
 2/1/2016

Example of simple random sampling

Numbers are selected at random

HEALTH RESEARCH FUNDAMENTALS

1 Albert D. 25 Monique Q.
2 Richard D. 26 Régine D.
3 Belle H. 27 Lucille L.
4 Raymond L. 28 Jérémy W.
5 Stéphane B. 29 Gilles D.

NIeCer 101
6 Albert T. 30 Renaud S.
7 Jean William V. 31 Pierre K.
8 André D. 32 Mike R.
9 Denis C. 33 Marie M.
10 Anthony Q. 34 Gaétan Z.

National Institute of Epidemiology

Chennai
11 James B. 35 Fidèle D.
12 Denis G. 36 Maria P.
13 Amanda L. 37 Anne-Marie G.
14 Jennifer L. 38 Michel K.
15 Philippe K. 39 Gaston C.
16 Eve F. 40 Alain M.
17 Priscilla O. 41 Olivier P.
18 Frank V.L. 42 Geneviève M.
19 Brian F. 43 Berthe D.
20 Hellène H. 44 Jean Pierre P.
21 Isabelle R. 45 Jacques B.
22 Jean T. 46 François P.
23 Samanta D. 47 Dominique M.
24 Berthe L. 48 Antoine C.
[Link]

2. Systematic sampling

HEALTH RESEARCH FUNDAMENTALS

• Principle
• A unit drawn every k units
• Equal chance of being drawn for each unit NIeCer 101

• Procedure
• Calculate sampling interval (k = N/n)
National Institute of Epidemiology
Chennai

• Draw a random number ( k) for starting

• Draw every k units from first unit
• Advantages
• Ensures representativity across list
• Easy to implement
• Disadvantage
• Dangerous if list has cycles
[Link]

10
 2/1/2016

Example of systematic sampling

HEALTH RESEARCH FUNDAMENTALS

NIeCer 101
National Institute of Epidemiology
Chennai
Every eighth house is selected
[Link]

3. Stratified sampling

HEALTH RESEARCH FUNDAMENTALS

• Principle
• Classify population into homogeneous subgroups (strata)
NIeCer 101
• Draw sample in each strata
• Combine results of all strata
National Institute of Epidemiology
Chennai

• Advantage
• More precise if variable associated with strata
• All subgroups represented, allowing separate conclusions about
each of them
• Disadvantages
• Sampling error difficult to measure
• Loss of precision if small numbers sampled in individual strata
[Link]

11
 2/1/2016

Example of stratified sampling

HEALTH RESEARCH FUNDAMENTALS

• Estimate vaccination coverage in a country

NIeCer 101
• One sample drawn in each region
• Estimates calculated for each stratum

National Institute of Epidemiology

Chennai
• Each strata weighted to obtain estimate for
country

[Link]

4. Cluster sampling

HEALTH RESEARCH FUNDAMENTALS

• Principle
• Random sample of groups (“clusters”) of units
NIeCer 101
• All or proportion of units included selected clusters
• Advantages
National Institute of Epidemiology
Chennai

• Simple: No list of units required

• Less travel/resources required
• Disadvantages
• Imprecise if clusters homogeneous (Large design
effect)
• Sampling error difficult to measure
[Link]

12
 2/1/2016

Cluster sampling

HEALTH RESEARCH FUNDAMENTALS

• The sampling unit is not a subject, but a group
(cluster) of subjects.

NIeCer 101
• It is assumed that:

National Institute of Epidemiology

Chennai
• The variability among clusters is minimal
• The variability within each cluster is what is observed
in the general population

Sampling techniques [Link]

The two stages of a cluster sample

HEALTH RESEARCH FUNDAMENTALS

1. First stage: Probability proportional to size
• Select the number of clusters to be included
• Compute a cumulative list of the populations in each unit with a NIeCer 101

grand total
Divide the grand total by the number of clusters and obtain the
National Institute of Epidemiology
Chennai

•
sampling interval
• Choose a random number and identify the first cluster
• Add the sampling interval and identify the second cluster
• By repeating the same procedure, identify all the clusters
2. Second stage
• In each cluster select a random sample using a sampling frame of
subjects (e.g. residents) or households
[Link]

13
 2/1/2016

5. Multistage sampling

HEALTH RESEARCH FUNDAMENTALS

• Principle
• Several chained samples

NIeCer 101
• Several statistical units
• Advantages

National Institute of Epidemiology

Chennai
• No complete listing of population required
• Most feasible approach for large populations
• Disadvantages
• Several sampling lists
• Sampling error difficult to measure
[Link]

Key issues

HEALTH RESEARCH FUNDAMENTALS

• We cannot study the whole population so we
sample it NIeCer 101

• Taking a sample leads to sampling error, which is

measurable
National Institute of Epidemiology
Chennai

• Good design and quality assurance ensure

validity and while appropriate sample size will
ensure precision
• Probability samples are the only one that allow
use of statistics as we know them
[Link]

14