Você está na página 1de 41

SAMPLING STRATEGIES

HE MISIRI
BLANTYRE, 14 JANUARY,2014

POPULATION
This is any collection of objects
animate or inanimate- that we intend
to study.

EXAMPLES
1. A woodlot comprising 1000 trees.
2. A litter of piglets
3. A fleet of cars produced in a day by
mass production
4. Patients in a ward
5. People in a village
6. Bees in a hive

FINITE AND INFINITE


POPULATIONS
A finite population is a countable
population .
An infinite population is a population
which due to some reasons is not
countable. For example, a swarm of bees.
This is the total number of subjects in a
population.

EXAMPLES
1. If the maximum capacity of a hospital
ward is 20 and all beds are occupied ,
then the population of patients in that
ward is of size 20.
2. If the total number of students at a
college of medicine is 100, then the
population of students at that campus
is of size 100.

STATISTICAL NOTATION
The population size is usually
denoted by upper case letter N.
Thus in the above two examples, we
have:
N = 20 and
N=100 respectively.

TARGET POPULATION
This is the basic finite set of
individuals that we are interested in.
For example if we are interested in
studying pregnant mothers attending
a prenatal clinic, then our target
population is all pregnant mothers
attending that clinic regardless of
their age, parity and trimester.

STUDY POPULATION
This is the specific set of individuals
we want to study. For example, if we
want to study pregnant mothers who
are in their first trimester of
pregnancy, then this group is our
study population.
The study population is a subset of
the target population if not the target
population itself.

HETEROGENEITY AND
HOMOGENEITY
A population is said to be homogeneous
with respect to some attribute if all the
members of the population possess
that attribute.
A population is said to be
heterogeneous with respect to a certain
attribute if not all members of the
population has that attribute or if they
have varying degrees of that attribute.

EXAMPLES-HOMOGENEITY
Literacy
If all members of the population are
literate, then we say that that
population is homogeneous with
respect to literacy.
Employment:
If all members of a population have
blue-collar jobs then that population is
homogeneous with respect to that.

EXAMPLES-HETEROGENEITY
Education
A heterogeneous population may
have people who are educated and
those who are not.
Alternatively, people may be
educated but the levels of education
may be diverse from one subject to
another.

CENSUS AND SAMPLE


A census is the complete enumeration
of all members of the population.
A sample is a collection of members
from the population.
This is just a set of individuals and not
the whole population.
The size of a sample is usually
denoted by lowercase n.

EXAMPLE
If my population comprises 18 first
year medical students, then a group
of students of size less that 18 is a
sample from this population.

SAMPLING UNIT
This is a potential member of a
sample.
Sometimes, a sampling unit may be
a collection of several units. For
instance, if you want to study children
under 20 years of age in a village,
then you will sample households.

SAMPLING FRAME
This is a finite collection of all sampling
units.
This is, in essence, a collection of all
population members.
This can be a list of all population members.
Alternatively, if you want to sample areas,
the whole area is divided into several parts
which are numbered. Your sampling frame
is a map divided into numbered areas.

REASONS FOR SAMPLING

To reduce cost/expenses
To reduce time
Utility
Accessibility of Members of a
Population
To increase the scope of the study
To increase the precision of
estimates

Replacement of sampled
subjects
Sampling With Replacement:
Is sampling where sampled subjects are
replaced to the population. There is a
high possibility of sampling them again.
Sampling Without Replacement:
Is sampling where sampled members
are not replaced to the population

ACCESSIBILITY SAMPLING
Only is subjects who are accessible to
the researcher are selected to be
members of a sample.
There is a problem of selection bias.
People may be accessible because of
being friendly, agreeable, pleasant,
and presentable, among other things.
In general, accessible subjects do not
necessarily represent the population.!

JUDGMENTAL SAMPLING
The researcher uses his judgement to
choose what he thinks is a
representative group of subjects.
If his judgement is expert judgement,
the sample may be representative.
If he is a novice, the resulting sample
may be nonrepresentative.
Sampling bias is inevitable.

QUOTA SAMPLING
Market researchers and opinion poll
experts often use these methods.
It is not probablistic
It can be outlined as below:

SAMPLING PROCEDURE
1. The categories of people to be sampled are
known in advance. E.g. Political Party
Affiliation.
2. The number of people to be interviewed for
each category is specified in advance. This is
known as a quota.
3. The interviewer then has the mandate to
choose whom he wants to interview for each
category.
4. The process of interviewing people is carried
out until all the categories are filled.

SAMPLING PROCEDURE
i. The categories of people to be sampled are
known in advance. E.g. Political Party
Affiliation.
ii. The number of people to be interviewed
for each category is specified in advance.
This is known as a quota.
iii. The interviewer then has the mandate to
choose whom he wants to interview for each
category.
iv. The process of interviewing people is
carried out until all the categories are filled.

MAJOR DRAWBACKS
1. There is bias in the selection of
subjects selection bias.
2. There is no way of measuring the
precision of estimates obtained from
surveys using such methods of
sampling.

PROBABILITY-BASED
METHODS

SIMPLE RANDOM SAMPLING


SYSTEMATIC SAMPLING
STRATIFIED SAMPLING
CLUSTER SAMPLING

SIMPLE RANDOM SAMPLING


This is a method of sampling where
each member of the population has
an equal chance of selection.
This can be done with or without
replacement.
For sampling without replacement,
the procedure is as follows:

i. Compile a sampling frame. Numbers from


1 to N will index the population members.
ii. Specify the size of your sample which is
usually denoted by n.
iii. Using random number tables (A
computer program like EPI INFO can
produce them for you) draw a sample of n
numbers from 1 to N inclusive.
iv. Use these sampled numbers to identify
your sample members.

This method is handy.


It produces representative samples.
However, if the population is very
big, this method is laborious and
expensive as the sample produced
may be staggered in the whole
population.

SYSTEMATIC SAMPLING
This technique relies on the
existence of a sampling frame.
A subject is chosen from the
population at random.
Every kth member after the first is
chosen as a subsequent member of
your sample. k is known as the
sampling interval.

EXAMPLE
Suppose the sample size is 23 and
the sampling interval is 5.
Draw a systematic sample of size 5:

POSSIBLE SYSTEMATIC
SAMPLES
1ST sample 2nd sample

3rd sample

4th sample

5th sample

10

11

12

13

14

15

16

17

18

19

20

21

22

23

ADVANTAGES
It is easy
A sample can be drawn on site
Intuitively speaking, it produces a sample
which is spread more evenly over the
population.
If the list is random, this method is
equivalent to simple random sampling
If the sampling frame is periodic, it produces
a better (representative) sample than when
simple random sampling is used.

DISADVANTAGES
In the case of a periodic frame, if the
sampling interval is equal to the
period of the frame or a multiple of
it, it will not yield good results.

It yields samples whose size may not


be equal to the desired sample size.

CLUSTER SAMPPLING
A cluster is a non-overlapping group
of population members.
Clusters may occur naturally. A good
example is a village. Aletrnatively,
clusters may be an artefact of the
researcher.
Clusters may be either of equal sizes
or of different sizes.

EXAMPLES
A village is a collection of households
A household may comprise several
people
A traditional authority may comprise
several villages

SAMPLING PROCEDURE

Identify clusters.
Compile a list of clusters
Obtain a random sample of clusters.
Study members of clusters.

MULTI-STAGE SAMPLING
If a sample of clusters is obtained and
every member of each sampled cluster is
included in the study then this is known
as a first-stage cluster-sampling scheme.
Sometimes this is done for several
stages.
This is known as multistage sampling.

EXAMPLE
Sample several districts.
From each sampled district, sample
several schools.
From each sampled school, sample
100 pupils.

STRATIFICATION
In some cases, the population is divided
into several divisions based on some
attribute e.g. sex, race, religion, district,
tribe etc.
Within each subdivision, members are
homogeneous and between subdivisions,
members are heterogeneous.
The process of dividing populations into
such categories is called stratification

REASONS FOR
STRATIFICATION
To achieve gain in precision.
Administrative convenience
To collect data of known precision for
certain divisions of the population.
To reduce sampling costs: Sampling
problems may differ in different parts
of the population.

STRATIFIED RANDOM
SAMPLING
Random samples of equal or different
sizes are each drawn from each
stratum.

THE END

Você também pode gostar