Probability sampling

 Simple random sampling

1
Simple Random sampling-I
 Every unit has equal chance to be included in the
sample.
 Process
i) Identify all units of population
ii) Select the number of units to be included in the sample.
iii) Select the sampling method.
 Methods
i) Card
ii) Lottery
iii) Random Numbers
iv) Computer assisted
 Type
i) With replacement
ii) Without replacement

2
Simple random sampling-II
 When to use simple random sampling
 Less variability in the population.
 Units are not scattered over a large geographical
area.
 When enough resource is available.
 When statistical treatment of data and
generalization is planned.

3
Systematic random sampling
 Selecting every nth number of equally divided parts
of the population
 Process
i) List all units of population.
ii) Determine the number of sample.
iii) Divide the population by the number of sample.
iv) Select a sample randomly from the 1st part (sub-population).
v) Select the same number from all following parts.
 When to Use
i) When the units likely to follow order throughout the population.
ii) Geographical concentration of units.
iii) To ensure the sample spread over all parts of the population.
iv) When a list of units of population is available.
4
Stratified random sampling
 Stratify the population according to specific characteristic(s) and then
randomly select specified number of sample from each strata.
 Process
i) Identify the characteristics important for the study, e.g., gender.
ii) Divide the population by the characteristics. Each division is called a
stratum.
iii) Select the specified units from every strata using simple random sampling
method.
 Types
i) Equal size: Same number of sample; irrespective of the number of units in
the strata.
ii) Proportionate: Number of sample from each strata is proportionate to the
number of unit of that particular strata.
 When to use
i) Maximum variability in the population.
ii) To ensure inclusion of units with particular characteristics.
iii) When characteristics of population is known.
iv) When use of random sampling might not include all variabilities of
population.
5
Cluster sampling
 Instead of a single unit, number of units considered together as a
unit. These units are called cluster.
 All clusters are homogenous but units in a cluster are different.
 The total number of cluster is treated as population.
 From population, determined number of clusters selected randomly.
 All units of the selected cluster are supposed to be studied.
 Process
i) Divide the population into clusters.
ii) Listing the clusters.
iii) Select specified number of cluster randomly.
iv) Include all units of the selected clusters in the sample.
 When to Use
i) Listing of all units not possible.
ii) If maximum variability in the population exists.
iii) To ensure inclusion of units with particular characteristics.
iv) When characteristics of population is known.
v) Use of random sampling may not include all variabilities of
population.
6
Multistage sampling
 Original unit of population is reached through step by step. In
every step samples are supposed to be selected randomly.
 Process
 Suppose, whole Bangladesh is the study area and household is
the unit of analysis
First: Divide whole Bangladesh into 8 divisions.
Second: Divide each division into districts and select specified number
of District(s) from each division randomly.
Third: Divide each selected District into Upazilas and select specified
number of Upazilas from each District randomly.
Fourth: Divide each selected Upazila into Unions and select specified
number of Unions from each Upazila randomly.
Fifth: Divide each selected Union into Village and select specified
number of Villages from each Union randomly.
Sixth: Select specified number of households from each selected
Village randomly.
 When to use
 When population spreads over a large geographical area. 7
Non-probability sampling methods-I
Convenience/Accidental sampling: Select those easily
available and fulfilling the sample criteria. Also called
accidental sampling, opportunity sampling or grab
sampling.
Quota sampling: Keep going until the sample size is
reached. It is often used as stratified manner and may be
proportional or non-proportional.
Snowball sampling: Get sampled people to nominate
others.
Judgment sampling: Selecting what seems like a good
enough sample.

8
Sample size determination
 It is difficult to decide; number of sample depends on-
1) Homogeneity or heterogeneity of the population
2) Number of classes proposed in the analysis
3) Nature of study: Short or long time, usual or
technical
4) Practical considerations
5) Standard of accuracy
6) Size of the questionnaire or schedule
7) Location of cases to be contacted
8) Type of sampling used
9) Type of statistical technique used
10) Availability of money, time & resources
9