Sampling Techniques

and Sampling Design

DR. UROOJ A. SIDDIQUI

 Significance of Sampling

1. When population is large - data is unlimited and more

scattered / inaccessible
2. A large number of units can be studied with small
representative sample.
3. Economy of time/ efforts/ resources
4. Useful for a study of homogenous universe.
5. Higher degree of accuracy with use of scientific method on a
smaller number of units
6. Inferences more reliable with data collected by well trained
investigators.
7. Easier for tabulation and analysis.
 Basic Terms

Population or Universe: represents the entire group of

units which is the focus of the study. It is group from which
the sample is to be selected.
Sample: The segment of the population that is selected for
investigation.
Sample Unit: Unit of analysis/research => units of samples
Sampling Frame: The listing of all units in the population
from which the sample will be selected.
Census: The enumeration of an entire population
 Basic Terms (contd.)

Representative Sample: A sample that reflects the population

accurately.
Sampling Bias: A distortion in the representativeness of the
sample that arises when some members of the population stand
little or no chance of being selected for inclusion in the
sample.
1. Sampling Error
2. Non-sampling Error (Systematic Bias)

Non-response: A source of non-sampling error that

occurs whenever some members of the sample refuse to
cooperate, cannot be contacted, or for some reason cannot
supply the required data
 Systematic Bias

Error due to deficiencies in the sampling approach, Causes:

1 ) Inappropriate Sampling Frame
2) Non Response
3) Defective Measurement – problems in measurer / device /
interviewer / items
4) Indeterminacy Principle – people behave differently when
kept under observation
5) Natural Bias – people tend to give what they think should be
correct instead of revealing truth – Upward / Downward Bias
 Systematic Bias can be reduced by detecting and
correcting the cause
 Larger the sample size larger the Systematic Bias
 Sampling

 When field of inquiry (population) is large – we

select a few units/items – respondents
 Selected respondents constitute sample
 Selection process of respondents is sampling
technique
 The survey so conducted is called as sample survey
 Population – N ; Sample – n
 Sample should be representative of population
 Sample Design

 Type of Universe/Population: all objects to be studied.

Finite – population of city, factory workers, flats in colony
Infinite – TV/Radio audience; stars in sky
 Sampling Unit: city / village/ locality / family / individual

 Source List: Sampling Frame

 Sample Size: optimum; neither too large nor small
 Parameters of Interest: variables to be measured

 Budget: cost and convenience

 Sampling Procedure: Process of selecting units of sample –

sampling technique
 Sampling Techniques

Sampling Types
Probability Sample: A sample that has been selected
using random selection so that each unit in the population
has a known/equal chance of being selected.
Non-probability Sample: A sample that has not been
selected using a random selection method.
 Sampling Techniques

Probability Sample: Random/Chance Sampling Design

 Simple Random Sampling (RS)
 Systematic RS
 Stratified RS
 Cluster Sampling
 Multi-stage Sampling
Non-probability Sample: Non-Random Sampling Design
 Convenience
 Purposive / Judgmental
 Quota
 Snowballing
 1. Simple Random Sampling

Simple Random Sampling: It is a sample

randomly drawn from the population, in which
each unit has equal probability of inclusion in
the sample.
a. Define Population (N) (e.g DU students)
b. Devise a Comprehensive Sampling Frame
c. Decide your sample size (n)
d. List whole population assigning numbers 1 to N.
e. Using a table of random numbers (Manual or
computer generated), select (n) different random
numbers that lie between 1 and N
f. The units with the selected n random numbers
constitute the sample
 Random Numbers

44 38 67 30 75 68 98 3 36 96 30 35 16 35 19 8 49 35
13 6 18 44 98 26 18 36 50 59 29 75 9 31 92 85 12 63
60 82 69 89 22 68 26 100 69 90 39 90 33 18 20 41 10
28 61 89 19 96 74 34 52 54 69 27 89 1 32 9 39 17
66 6 27 16 56 63 96 86 41 81 21 34 30 63 26 49 44
31 34 100 68 24 88 70 70 44 67 14 34 4
 2. Systematic Random Sampling

Systematic Random Sampling: It is a random sample

systematically drawn from the population, in which
each unit has equal probability of inclusion in the
sample.
a. Define Population (N) (e.g DU students)
b. Devise a Comprehensive Sampling Frame
c. Decide your sample size (n)
d. List whole population assigning numbers 1 to N.
e. Find out sampling fraction: s = N/n
f. Use random numbers to select the first sample (nₒ)
between 1 and s.
g. Select units in the following sequence: nₒ, nₒ+ s, nₒ+2s …..
.. [nₒ+(n-1)s]
h. The students with the selected n numbers constitute the
sample.
 3. Stratified Random Sampling

Stratified Random Sampling is a sample drawn

randomly and/or systematically from the
population, after stratification using a stratifying
criterion, in which each unit has equal
probability of inclusion in the sample.
a. Define Population (N) (e.g DU students)
b. Decide your sample size (n)
c. Find out sampling fraction: s = N/n
d. Stratify the population based on a stratifying
criterion
e. List whole population in each strata.
f. Use random or systematic random sampling method
to select units from each strata, such that the selected
units constitute the sample (n).
 4. Cluster Sampling

Cluster Sampling is a sample drawn from all

units of randomly selected clusters out of a
population divided into clusters using a
clustering criterion.
a. Divide the population into clusters using a clustering
criterion
b. Use random or systematic random sampling method
to select (c) clusters from the total number of clusters.
c. The sample constitutes all the units from each
selected cluster z = c₁ + c₂ + ….+cn
 5. Multi-Stage Sampling

Multi-stage sampling refers to sampling plans where

the sampling is carried out in stages using a particular
sampling method to select smaller and smaller
sampling units at each stage.

In a two-stage sampling design, a sample of primary

units is selected using a particular sampling method
and then a sample of secondary units is selected within
each primary unit using the same or some other
sampling method.
 e.g. Multi-Stage Cluster Sampling

Multi-Stage Cluster Sampling is a sample drawn

randomly and/or systematically from randomly
selected clusters out of a population divided into
clusters using a clustering criterion.
a. Divide the population into clusters using a clustering
criterion
b. Use random or systematic random sampling method to
select (c) clusters from the total number of clusters.
c. Decide your sample size (n)
d. The sample selected from each cluster will be z = n/c
e. Use random or systematic random sampling method to
select z units from each cluster, such that the selected
units constitute the sample (n).
Non-Probability Sampling Techniques

1. Convenience Sampling : It is a sample constituted of

units simply available to the researcher by virtue of its
accessibility.
2. Purposive or Judgmental Sampling : when sampled
units are selected on basis of some specific purpose
relevant to the research.
3. Quota Sampling: It is a sample generated from a fixed
quota of units from different categories of the population
– gender / caste / group / area
4. Snowball Sampling: It is a sample generated from the
contacts of a small group of people relevant to the
research topic. Every contact gives 1 or 2 other contacts
 Criteria of Selecting Sample

Two costs in sampling analysis

 Cost of data collection
 Cost of incorrect inference

Causes of Incorrect Inferences

 Sampling Errors
 Systematic Bias (non-sampling errors)
 Systematic Bias

Error due to deficiencies in the sampling approach, Causes:

1 ) Inappropriate Sampling Frame
2) Non Response
3) Defective Measurement – problems in measurer / device /
interviewer / items
4) Indeterminacy Principle – people behave differently when
kept under observation
5) Natural Bias – people tend to give what they think should be
correct instead of revealing truth – Upward / Downward Bias
 Systematic Bias can be reduced by detecting and
correcting the cause
 Larger the sample size larger the Systematic Bias
 Sampling Error
Error in the findings due to random variations in sample
statistic around the true population parameter
(mean/variance)
 Random variations – chance variations equally likely in
both directions
 Sampling error can be calculated for a given sample
design and size
 Sampling error decreases with increase in sample size
 It is smaller in homogenous population
 Measurement of error is precision of sample
 Sampling Error

 Inference meaning – Estimating population

mean/variance (SD) with sample mean/variance (SD)
 Are chances of catching a fish with a spear
higher or with using a net?

 Sample – Point Estimate: X

 Sample – Range: X ± a

 Precision – Confidence Level & Confidence Interval

 Sampling Error

 Confidence Level (CL) – Selected / Chosen

 How much are we confident that population mean will
under this range?
 Confidence Interval (CI) – Calculated (distribution)
 Generally accepted CL Level – 95% (90% , 99%)
 95% CL means if we collect 100 samples then 95 samples
will have population mean within a particular range of
sample means i.e. Confidence Interval (range of mean)
 Margin of Error/ Significance Level = 100 – CL
 If CL = 95% then Error = 5%
 Generally selected SL = 5% (10% , 1%)
 Sampling Error

Confidence Interval (CI) calculation

 CI is calculated as per the distribution that the sample
statistic follows
 Sample characteristic - statistic
 Population characteristic - parameter
 Sample statistic may be mean / variance / correlation
Distributions
 Mean – z (normal) or t distribution
 Difference of mean – z or t distribution
 Variance – chi square distribution
 Ratio of Variance – F distribution (ANOVA)
 Sample Size

Sample Size (n) Considerations

 Population Size (N) – proportionate
 Population Variance (square of Standard Deviation)
 Homogenous – small size
 Heterogeneous – large size
 Sampling error (decreases with n)
 Systematic Bias (increases with n)
 Cost/Budget
 Sample Size Finite N

n = Nx E.g. Population Size = 20,000

(N-1)E² + x Confidence Level = 95%
x = Z(c/100)² . r(100-r) Margin of Error = 5%
E = √(N-n)x/n(N-1) Response Distribution = 50%

n is sample size, Recommended Sample

Size = 377
N is population size,
For 90% n will be 263
E is Error;
For 99% CL n will be 643
r is fraction of responses;
(Source: Raosoft sample size
Z(c/100) is the critical value calculator)
for the confidence level c.
 Sample Size Infinite Population

Experts calculate the sample size on the basis of techniques

employed in data analysis
 In research studies where various abstract concepts /
unobserved variables are measured the sample size is
calculated in accordance with the number of total number
of indicators / items in the study
 Thumb rule n = 5 to 10 times the number of indicators
 E.g. 5 Concepts/unobserved variables with 4 items each;
the total items will be 20, so
 n = 5x20 to 10x20 = 100 to 200
 Observational Design / Data Collection

 Sources of Data Collection

1. Secondary Data – already existing data /
information collected in past by some other/s
person for some other purpose
 Secondary Data Sources
 Earlier researches / papers / articles in journals, thesis,
books, magazines, news papers / project reports
 Organizational Publications like news letters, annual
reports, special reports, periodicals etc.
 Govt. publications like census, economic surveys etc
 Observational Design / Data Collection

 Sources of Data Collection

2. Primary Data – fresh / first hand data collected
by the researcher for one’s own purpose
Sources of Primary Data
 Observation
 Experiment
 Interview
 Survey - collection of data on many variables (many
questions) from large set of respondents
 Observational Design / Data Collection

 Survey Tools
1. Questionnaire – the questions are given in
printed form / online to the respondents
 Respondents select options/ write the answers
2. Schedule (Interview Schedule)– questions are
asked verbally by researcher/ enumerator to the
respondents
 Enumerator mark options / write the answers
 Questionnaire

1. Open ended question: What was your experience of RM

course? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
2. Closed ended questions : Fixed answers from which
answers are to be chosen. Eg: What is your education
Level?
a) Less than X/XII b) X/XII c) UG d) PG
3. Converting open ended into closed (Coding of variables)
Employment: ______________
Agriculture – 1, Teacher – 2, Any other - 3
 Types of Questions

1. Personal factual questions – age, occupation,

marital status, income, social group, religion etc
2. Factual questions about others – Household
income, organisation practices
3. Questions about attitudes – Likert Scale is best
(psychological)
4. Questions about beliefs – True or false / Yes or no
5. Questions about knowledge – Awareness, facts
 Rules for designing questions (Rules of thumb)

1. Remember in mind your research question

(Do not miss out critical questions)

2. What is it that you want to know?

(Do you have a computer? – can mean own or
access)

3. How would you answer it?

(Put yourself in the place of respondent)
 Avoid the followings

 Avoid ambiguous terms in questions - How often do

you usually visit the cinema?
 Ambiguous: Very often; quite often, not very often , not at all
 Clarity: once a week, 2 or 3 times a week…

 Avoid long questions – Keep it short and to one point

 Avoid double barrelled questions
 How satisfied are you with the MBA course and campus infra?

 Avoid very general questions – General questions

lack frame of reference – How satisfied are you with
the course? (Content, methodology, presentation etc)
 Avoid leading questions
 Would you agree to cutting taxes though it might reduce
the welfare programmes of the govt?
 Avoid questions that are actually asking same
things twice
 Which political party did you vote in the last general
elections?
 Did you vote in the last election? Yes – No; If yes, which
political party did you vote for?
 Avoid questions that include negatives - although
sometimes unavoidable
 try not to use – easy to miss out ‘not’
 Do you agree that students should not carry mobile to the
schools?
 Avoid technical terms
 Do you think that this product has high Brand Equity?