SAMPLING TECHNIQUES AND

SAMPLE SIZE ESTIMATION

Ms. Irene
(BSc Ehs, MSc Epi)
Department of Medical Science and Technology

1
 LEARNING OBJECTIVES

After successful completion of this session, you should be

able to:
• Distinguish a sample from a population

• Define the key terms used in sampling (eg representative

sample, study unit, sampling frame etc)
• Distinguish between probability and non-probability
sampling
• Explain the importance of random sampling

• Distinguish different probability and non probability

sampling techniques
 WHAT IS SAMPLING?

Sampling is the process of selecting a small number of

elements from a larger defined target group
(Population) such that the information gathered from
the small group will allow judgments to be made about
the larger groups.
 DEFINITIONS
 Target population
The entire set of subjects we would like to say
something about
 Study population
The population from which the sample is chosen
 Sample
A subset of the study population
 Observation/study units
We collect data on these units (objects or people)


4
 Example: If you are studying the health effects of a new
drug for adults aged 18–65 living in the U.S.
 Target population would be all adults aged 18–65 in the

U.S
 The study population would be the adults aged 18–65

living in California.
 Observation units would be the individual adults aged

18–65 living in California (subset of the study

population)
 DEFINITIONS CONT...
 Sampling unit
 Unit actually sampled (can be different from observation unit)
 For example
 Households as sampling units (individuals as observation units)

 Health facilities as sampling units (health workers as observation

units)
 Sampling frame
 List of sampling units
 Examples
 Individuals

 Households

 Street addresses

 Villages
RESEARCH PROCESS

o Target population
o Identify the Study
population
o Select a sample
o Study the sample
o Infer
o Conclude
SAMPLING: A PICTORIAL VIEW

Sampled
(Study)
Population Sample

Target Population

Target Population  Sampled Population  Sample 8

A GOOD SAMPLE

Should be:
• Representative

• Unbiased
A REPRESENTATIVE SAMPLE
 A representative sample has all the important characteristics
of the study population from which it was drawn

UNBIASED SAMPLE
 Responses are not influenced by selection of subjects
 A biased sample leads to distortion of results (under- or over-
estimates
SAMPLING METHODS

Probability Nonprobability
sampling sampling
METHODS USED IN NON
PROBABILITY SAMPLING

 Non probability Sampling techniques include:

1. Convenient Sampling
2. Quota sampling
3. Purposive sampling
4. Snowball sampling

12
CONVINIENT SAMPLING
 Sampling is done Conveniently
 Selection of the most rapid available subjects
 Example, hospital patients, sampling customers at
a shop

PURPOSIVE SAMPLING
 Select sample units that are most useful to the
purposes of research
 Example: sampling of the disabled students at the

university
QUOTA SAMPLING
o Specify characteristics of your study population e.g. location,
age, income level, education, gender, ethnicity,
religion etc
o Each of the pre-specified characteristics is called a quota

o Population is divided into important sub groups

(characteristics) in the population (quota)

o No. of subjects to be included from each subgroup

determined before.
o Recruitment of subjects is done within each subgroup

conveniently until the quota is reached.

Example
o You are conducting a study about the rising food prices. You study

population is both men and women buying a selling food. 14

o
 SNOWBALL (CHAIN REFERRAL)
SAMPLING

o Used to recruit participants via other participants, if the

population is hard to access
o After the interview, the researcher requests each of them to
suggest additional individuals to be interviewed
 Useful method in reaching the “hidden” or “hard-to-reach” populations e.g.
 Female sex workers
 Injecting drug users
 Men who have sex with men
 Example: You are conducting a study to understand how and why female sex
workers decide which clients to accept and which ones to turn down
 You get in touch with one female sex worker you know or one you are referred
to by a person they trust, interview them, and ask them to recommend another
potential responded
ADVANTAGES AND DISADVANTAGES
OF NON-PROBABILITY SAMPLES
• May be more convenient and less expensive to execute
(sometimes this is the only feasible option) – does not require the
identification of a sampling frame

• May have strong bias

• The results cannot be generalized

• Requires judgment and caution when interpreting the results

16
PROBABILITY SAMPLES

 Selection of units based on chance

 All individuals (elements) in the population have a known
probability of being selected.
 A sampling frame, whether of individual elements or
clusters of elements, from which the sample is to be
drawn should be available.
 Results over several repeated samples will be similar.
 Statistical inferences can be drawn.

17
PROBABILITY SAMPLING METHODS
 Simple Random Sampling (SRS)
 Systematic Random Sampling (SYRS)
 Stratified Random Sampling (STRS)
 Cluster Random Sampling (CLRS)
 Multi-stage (cluster) Sampling (MSRS)
(1) SIMPLE RANDOM SAMPLING
(SRS)
 Simplest of random sampling
 Basis of all other sampling methods
 ALL units have EQUAL chance of selection
 Count all units of a population and create a

sampling frame
 Assign identification number to units and
determine sample size
 Select unit from sampling frame by lottery or

Table of random numbers (created by random

number generator softwares) until the sample size
is reached
 Example: sampling registered students
LOTTERY METHOD

 The lottery method is a simple, manual technique used for random

selection.
 How it works:

 Step 1: Assign each individual in the population a unique identifier

(e.g., numbers or names).

 Step 2: Write down these identifiers on slips of paper (or some

other medium).
 Step 3: Place all the slips of paper into a container (like a hat or

box).
 Step 4: Mix the slips thoroughly and randomly draw a specified

number of slips to form the sample.

(2) SYSTEMATIC RANDOM
SAMPLING (SYRS)
Sample units obtained systematically
Count all units of a population and create a
sampling frame (N)
Determine sample size (n)
Establish sampling interval( k): k=N/n

First unit (n ) should be between 1st and kth of N

i

Subsequent unit (nt)=ni + k

Example: sampling of houses
NOTE:

 There should be no order in which subjects are

arranged

 Better than SRS: more conveniently

 More representative than SRS

(3) STRATIFIED RANDOM SAMPLING
(STRS)

 Populationdivided into subgroups based on

desired characteristics
 Example, occupation, HW cadre, age, sex, etc
 High variability between strata
(heterogeneous)
 Low variability within strata (homogeneous)
 Sample within strata
NOTE:

 Similar to quota sampling

 Stratification done before

 Sex: guarantee membership of M or F

 Aim: to enhance representativeness

 Problem: attributes may not be available for

stratification
 Unequal size of strata -> PSS
(4) CLUSTER SAMPLING (CLS)

 Sometimes, not possible to obtain listing

 Example, under-fives in Tanzania

 SRS, SYS or STS not possible

(4) CLUSTER SAMPLING (CLS)

 Sampling unit = collection of units (cluster)

 Examples, schools, villages, wards, etc
 Sampling using SRS; cluster = a unit
 Homogeneous clusters (between)
 Heterogeneous units (within)
(4) CLUSTER SAMPLING (CLS)

Stages of Sampling
 “One-stage cluster sampling”: all elements in a groups
sampled
 “Two-stage cluster sampling”: elements within selected
group sampled
 Multi-stage cluster sampling
STRATIFIED VS CLUSTER SAMPLING

4
4

STS: Take a sample in each stratum CLS: Fully study all units in a selected
cluster
(5) MULTI-STAGE SAMPLING (MSS)
 Many stages of sampling (clusters)
 Select clusters by random sampling

 = Sampling at each stage

 Possibility of different techniques

 First-stage -> second stage -> third stage, etc

 Example: A study conducted on school performance in a country

X
 Stage 1: Primary Clusters (Regions or Districts)

 Stage 2: Secondary Clusters (Schools)

 Stage 3: Tertiary Clusters (Classes or Students)

EXERCISE ON SAMPLING

 The Council Health Management Team (CHMT)

in district X is planning a study to determine the
proportion of underfive children who are
malnourished: What is :
A) The Target population?
B) The study population?
C) The best sampling technique?

30
EXERCISE:

A study is designed to assess knowledge of emergency

contraception among Semester undergraduate
students at MUST.
a) What is the target population?
b) What is the study population?
c)Use SRS to select a sample of 100 students to include
in this study.
EXERCISE ON SAMPLING
 You are assigned a task to conduct a study to
assess prevalence of schistosomiasis among
school age children in Rufiji district.

1. What is the target population

2. What is the study population
3. What sampling technique will be most
appropriate?

32
 THANKYOU FOR YOUR ATTENTION

33
QUIZ (5 MARKS)
 Mention five things that you have learnt in
today’s session