FACULTY OF COMPUTER AND MATEMATHICAL SCIENCES

UNIVERSITI TEKNOLOGI MARA

Chapter 1: Introduction
1. Sampling
2. Technical Terms
3. Sample Survey Design
4. Data Collection Methods
5. Types of Sampling Techniques
in Survey
6. Sources of Errors in Surveys
7. Reducing Errors in Survey

10/12/2020 2
 Sampling

10/12/2020 3
 Sampling
Sampling is the process of selecting a
few number of subjects from all the
subjects in a particular group or
"universe.“

In other words, it is the process of

deciding, out of the entire population
of interest, from whom you will collect
data.

10/12/2020 4
 Sampling …(cont.)
. . . involves selecting a relatively
small number of elements
(sample) from a larger defined
group (population) and
expecting the information
gathered from the small group
will enable judgments about the
larger group.

10/12/2020 5
 Sampling..(cont.)
 Sampling is the process of selecting units such
as people, organizations, or items from a
population of interest.
 By studying the sample, we may fairly generalize
our results and make inference on the population
from which they were chosen.
 For the results of a survey to be accurate and
unbiased, certain sampling rules need to be
accurate and unbiased, certain sampling rules
need to be followed.

10/12/2020 6
 Sampling..(cont.)
 Each observation or item taken from the population
contains a certain amount of information about the
population parameter.
 Too little information prevents the experiment from
making good estimates.
 Too much information results in a waste of money.
 The quantity of information obtained in the sample
depends on the number of items sampled and on
the amount of variation in the data.

10/12/2020 7
 Why do Sampling?
 We sample because we cannot measure every
person in the population.
 Sampling is done because you usually cannot
gather data from the entire population.
 Even in relatively small populations, the data
may be needed urgently, and including
everyone in the population in your data
collection may take too long.

10/12/2020 8
 Purpose of Sampling
A random sample is selected from the population so that
inference could be made from the sample about the population
 It saves cost and time
 It is easier to collect data from a sample rather than from
Reasons for sampling

the whole population

 Since data are collected in a short time, the data can be
processed and analyzed more quickly
 It's easier to control and organize a small team of
workers.
 Some observations require the destruction of the sample
– such as the destruction of electrical appliances by
lighting
 Sometimes a certain geographical area is not accessible
such as the swampy land. In this case sampling
appropriate.

10/12/2020 9
 Advantages of the Sampling Method

10/12/2020 10
 Sampling Process
Specifying a Determining
Defining the
sampling the sampling
population
frame design

Select actual
Conduct field Determining
sampling
work sample size
units

10/12/2020 11
 Characteristics of a good sample

10/12/2020 12
 Technical Terms

10/12/2020 13
 Sampling Terminology
Some basic sampling terminologies that need to be
clearly understood are:
 An element is an object on which a
measurement is taken.
 A population is a group of all people or items
that possess the characteristics one wishes to
study.
 Sampling units are nonoverlapping collections
of elements of the population that cover the
entire population.

10/12/2020 14
 Sampling Terminology...(cont.)
 The sampling frame is any list of all the
sampling units in the target population (list of
individuals or items from which the sample
can be obtained)
 Example of sampling frames:

✓ Telephone directory
✓ Car ownership list
✓ Credit card customer list
✓ Voter list
✓ Students list
10/12/2020 15
 Sampling Terminology...(cont.)
 A sample is a selected portion of the
predetermined population, characteristics of
interest of these elements will be measured
or observed.
 The census is the actual measurements or
observations of all possible elements from
the population or a study conducted which
measure a variable for every unit in the
population.

10/12/2020 16
 Sampling Terminology...(cont.)
 Parameter is a characteristic of the population,
such as the population mean, standard deviation
or population proportion.
 A statistic is a characteristic of the sample such
as the sample means, standard deviation or
sample proportion . This value is used as an
estimate of the value of the corresponding
population parameter.
 A sample survey, involves a subgroup (or
sample) of a population being chosen and
questioned on a set of topics.

10/12/2020 17
 Sampling Terminology ...(cont.)
 A Pilot Study is a study done before the
actual fieldwork is carried out.
 The uses of the pilot study are
✓ to identify possible problems and
difficulties that the researcher may
encounter when actual study is carried
out.
✓ to test out questionnaires and to improve
in terms of flow, question design,
language and clarity.

10/12/2020 18
 Population and Sample
❑ Population ❑ Sample
 A collection of all  A subset of the
object/ people/ population
events in which  Selected to
information is represent the
desired. population by
 Mostly impossible sampling technique.
to measure.  “n” represent the
 “N” represent the total of a sample of
total of the population.
population.
10/12/2020 19
 Survey

10/12/2020 20
 Survey
A survey is any activity that
collects information in an
organised and methodical
manner about characteristics of
interest from some or all units of
a population using well-defined
concepts, methods and
procedures, and compiles such
information in a useful summary
form.
10/12/2020 21
 Survey...(cont.)
Survey

Sample Census
survey Survey
Sample are collected
Sample are collected
for only a fraction of
for all units in the
units of the
population
population

10/12/2020 22
 Sample Vs Census
Conditions Favoring the Sample Census
Use of
1. Budget Small Large
2. Time available Short Long
3. Population size Large Small
4. Variance in the characteristic Small Large
5. Cost of sampling errors Low High
6. Cost of nonsampling errors High Low
7. Nature of measurement Destructive Nondestructive
8. Attention to individual cases Yes No

10/12/2020 23
 Planning a Survey
1 • Formulation of the statement of objectives
2 • Define the target population
3 • The sampling frame
4 • Sample Design
5 • Method of measurement (Data collection)
6 • Measurement instrument (etc. questionnaire)
7 • Selection and training of field-workers
8 • The pilot/pre-test
9 • Organization of fieldwork
10 • Organization of data management
11 • Data Analysis
10/12/2020 24
 Data Collection Methods
 There are several methods used to collect
data in survey research. Each has its own
advantages and disadvantages.

 The choice of methods will depend upon

the:

✓ Type and size of the audience

✓ The purpose of the study
✓ Timeline
✓ Budget & staff available

10/12/2020 25
 Data Collection Methods…(cont.)
1. Personal Interview (Face to face interviews)
- Respondents will be interviewed based on the
questionnaire or the questionnaires will be given to
them.
- Make sure the appropriate questions are answered.

2. Telephone Interview
- Interviewer will ask the respondents based on the
questionnaire

3. Mail Questionnaires
- Questionnaires will be posted to the respondents

4. Direct Observations
- Observer/researcher will observe the respondent
regarding the matter of interest.

10/12/2020 26
 Designing a Questionnaire

Questionnaire is a written
instrument that contains a series
of questions or items that
attempt to collect information on
a particular topic.

10/12/2020 27
 Designing a Questionnaire…(cont.)
Steps to be consider when designing questionnaire

 Questions asked should be short and easy.

 Languages used should be simple and non-
technical.
 Questions should be arranged in logical
sequence
 Provide clear and distinct answers to the
objective questions.
 Try to avoid private and personal matters
 Try to avoid calculations and tests of memory
 Try to avoid long answers
 Units used should be precise
10/12/2020 28
 Major Concerns in Designing a
Questionnaire

1 • Question Ordering

2 • Open Versus closed Questions

3 • Response Options

4 • Wording of Questions

10/12/2020 29
 Types of Sampling
Techniques in
Surveys

10/12/2020 30
 Sampling Techniques
 Sampling techniques are scientific methods of
selecting samples from population of interest.
 The sampling technique used in each study
depends on the nature of the population of
interest.
 This includes factors such as:
✓ homogeneity (or heterogeneity) of the
population
✓ the availability of the sampling frame
✓ the research budget
✓ the important of the research

10/12/2020 31
 Sampling Techniques...(cont.)
Sampling techniques can be classified as
probability and nonprobability sampling

 Probability sampling is a sampling process

that ensures each person or element in the
population has a known chance of being
included in the sample. However, each person
or element may or may not have an equal
chance of being selected. This method is used
when generalization regarding the population
is required.

10/12/2020 32
 Sampling Techniques...(cont.)
 Probability sampling techniques are Simple
Random Sampling, Systematic Sampling,
Stratified Sampling, Cluster Sampling and
Multistage Sampling.
 These sampling techniques have two things in
common:
i. Every element has an equal chance of being
sampled and
ii. Involves random selection at some point.

10/12/2020 33
 Sampling Techniques...(cont.)
 Nonprobability sampling is a sampling
process where some elements of the
population have no chance of selection.
 The selection of elements is done based on
assumptions regarding the population of
interest. These assumptions become the
criteria for selection.

10/12/2020 34
 Sampling Techniques...(cont.)
 The selection of elements is nonrandom as
compared to probability sampling.
 This method of sampling is primarily used in
research studies where no inferences is to
be made from the sample to the population.
In other words no generalization regarding
the population is needed.

10/12/2020 35
 Sampling Techniques

Non – Probability
probability
Simple Random Systematic
sampling sampling
Convenience Judgemental
sampling sampling Stratified Multistage
sampling sampling
Snowball Quato
sampling sampling
Cluster sampling

10/12/2020 36
 Probability Sampling
Simple Random Sampling → a sample of size n is
drawn from a population of size N in such a way
that every element has the same chance of being
selected
 Can be done using random number table or
random numbers generated by computers
 Requires a complete and updated sampling
frame
 Population is homogeneous in nature (units are
of similar characteristics.
 Not suitable for a large population

10/12/2020 37
 Probability Sampling

Systematic Sampling
 Random selection of one unit from the first k
units (r) then followed by the selection of every
kth unit thereafter.
 Requires a random sampling frame
 Population must be homogeneous

10/12/2020 38
 Probability Sampling
Stratified Sampling → a sample obtained by
separating the population elements into non-
overlapping groups, called strata and then
selecting a random sample from each stratum.

 Appropriate when there is a large variation

within the population
 To ensure that certain significant attributes are
present in each stratum or group
 Populations that are stratified are
heterogeneous

10/12/2020 39
 Probability Sampling
Cluster Sampling → a sample in which each unit
is a collection of elements from selected
clusters
 Can be applied to a large study area
 Cost of obtaining information is reduced
 Higher sampling error

Multistage Sampling
 An extension of cluster sampling, several
stages of sampling involved in the process of
obtaining a sample. Normally the surveys
cover large geographical areas.
 This method is widely used in surveys
10/12/2020 40
 Probability Strategies
Type of Sampling When to Use It Advantages Disadvantages
Simple random When the population Ensures a high degree Time consuming and
sampling members are similar to of representatives tedious
one another on
important variables

Systematic sampling When the population Ensures a high degree Less random than
members are similar to of representatives and simple random
one another on no need to use a table sampling
important variables of random numbers

Stratified random When the population is Ensure a high degree of Time consuming and
sampling heterogeneous and representatives of all tedious
contains several the strata or layers in
different groups, some the population
of which are related to
the topic of study
Cluster sampling When the population Easy and convenient Possibly members of
consists of units rather units are different from
than individuals one another ,
decreasing the
technique’s
effectiveness
10/12/2020 41
 Non-probability Sampling
Convenience Sampling
 Selection of elements or sampling units is
left primarily to the interviewer
 The sample selected is not representative of
the population
 Least cost and least time – consuming
compared to any other sampling techniques.
Judgemental Sampling
 Sampling units are selected based on the
judgement of the researcher

10/12/2020 42
 Nonprobability Sampling
Quota Sampling
 Similar to convenience sampling except the
numbers allocated for each group of
respondents depends on the population

Snowball Sampling

 Initial group of respondents is selected at

random, after being interviewed these
respondents are asked to identified others
who belong to the population.

10/12/2020 43
 Non-probability Sampling Strategies
Type of When to Use It Advantages Disadvantages
Sampling

Convenience When the Convenient and Degree of

sampling members of the inexpensive generalizability
population are is questionable
convenient to
sample
Quota sampling When strata are Ensures some Degree of
present and degree of generalizability
stratified representativen is questionable
sampling is not ess of all the
possible strata in the
population

10/12/2020 44
 When to use non probability sampling
research does not aim to generate results that will be used to
 entire population.
create generalizations pertaining to the

the researcher aims to do a qualitative, pilot or exploratory

study.

demonstrating that a particular trait exists in the population.

randomization is impossible like when the population is

almost limitless.

the researcher has limited budget, time and workforce.

in an initial study which will be carried out again using a

randomized, probability sampling.
10/12/2020 45
 Probability vs Non-probability Sampling
Techniques
Comparison Probability Non-probability
Factors sampling sampling
List of the Complete list Not necessary
Population necessary
Elements
Time Requirement Time consuming Low time consumption
Estimates of Unbiased Biased
Population
Parameters
Measurement of Statistical measures No true measure
Sampling Error available
Sample Good, assured Suspect,
Representativeness undeterminable
10/12/2020 46
 Probability vs Non-probability Sampling
Techniques... (cont.)
Probability Non-probability
Allows use of statistics, tests Exploratory research,
hypotheses generates hypotheses

Can estimate population Population parameters are

parameters not of interest
Eliminate bias Adequacy of the sample
can’t be known
Must have random selection Cheaper, easier, quick to
of units carry out

10/12/2020 47
 Choose point in sampling design
Is representiveness of sample
critical for the study?

Yes No

Assessing differential
Choose one of the non-
Generalizibility parameters in subgroups of
probability sampling design
population

Choose
Choose
Choose cluster If the purpose of
simple
systematic sampling if all subgroups
random study mainly is:
sampling not enough
sampling
budget

To obtain
To obtain information
quick even if
Yes No relevant to and available
unreliable
only with certain groups
information

Choose Choose
proportionate disproportiona Choose Looking for Need response of
stratified te stratified convenience information that special interest
random random sampling only few experts minority groups?
sampling sampling can provide?

Choose
Choose quota
judgmental
10/12/2020 sampling
48
sampling
 Sources of Errors
in Surveys

10/12/2020 49
 Source of Errors in Surveys

. . . factors which reduce the

accuracy of a survey
estimate.

10/12/2020 50
 Sources of Errors in Surveys…(cont.)
Errors

Nonobservation Observation

Sampling Interviewers

Respondents
Non-response
Measurement Instrument
Coverage
Methods of Data
Collection
10/12/2020 51
ERRORS OF NONOBSERVATION
SAMPLING ERROR
 The deviation between an estimate from an ideal
sample and true population.
 Another way to look at this is: the differences in
results for different samples (of the same size) is
due to sampling error.

10/12/2020 52
ERRORS OF NONOBSERVATION

 E.g.: Two samples of size 10 of 1,000 households.

If we happened to get the highest income level
data points in our first sample and all the lowest
income levels in the second, this is a
consequence of sampling error.
 Increasing the sample size will reduce this type of
error.

10/12/2020 53
 How to minimize the sampling error
• Using appropriate
Sampling sampling technique when
technique drawing the sample from
its population

• Increasing the size of

Size of sample or respondents to
sample be obtained from its
population

10/12/2020 54
ERRORS OF NONOBSERVATION
COVERAGE ERROR
 Occur when some part of the population cannot be
included in the sample, the sampling frame does not
match up perfectly with the target population.
 To be precise, Groves specifies three different
populations:
1. The population of inference is the population that
the researcher ultimately intends to draw
conclusions about.
2. The target population is the population of inference
less various groups that the researcher has chosen
to disregard.
10/12/2020 55
ERRORS OF NONOBSERVATION
3. The frame population is that portion of the
target population that can be counted prior to
the selection of the sample.
 The survey sample then consists of those
members of the sampling frame that were
chosen to be surveyed and coverage error is
the difference between the frame population and
the population of inference.

10/12/2020 56
ERRORS OF NONOBSERVATION
 The two most common approaches to reducing
coverage error are:
a) obtaining as complete a sampling frame as
possible (or employing a frameless sampling
strategy in which most or all the target
population has a positive chance of being
sampled), and
b) post-stratifying to weight the survey sample to
match the population of inference on some
observed key characteristics.

10/12/2020 57
ERRORS OF NONOBSERVATION
NONRESPONSE (most serious)
 Nonresponse errors refers to error (or bias)
introduced when responses are not obtained
from some members of the sample, i.e. the
sample observations that are collected may not
be representative of the target population.
 Occur when data is not collected on either entire
respondents (unit nonresponse) or individual
survey questions (item nonresponse).

10/12/2020 58
ERRORS OF NONOBSERVATION
 The response rate, which is the ratio of the
number of survey respondents to the number
sampled, is often taken as a measure of
goodness.
 Higher response rates are taken to imply a
lower likelihood of nonresponse bias, where
such a bias exists if there is a substantial
difference between the observed survey results
and what would have resulted if the entire
sample had responded to the survey.

10/12/2020 59
ERRORS OF NONOBSERVATION
 Non-response can occur because:
1) the sampling frame is out of date and the person
that is trying to be contacted no longer lives at
the stated address;
2) the person is not available to take part in the
survey; or
3) the person refuses to participate in the survey.
 Non-response error can be reduced through good
questionnaire design, assurances of confidentiality, re-
contacts and regularly updating the sampling frame.

10/12/2020 60
Approach to deal with nonresponse
 Prevent it. Design the survey so that nonresponse is low.
This is by find the best method.
 Take a representative subsample of the nonrespondents; use
that subsamples to make inferences about the other
nonrespondents.
 Use a model to predict values for the nonrespondents.
Weighting class adjustment methods implicitly use a model
to adjust for unit nonresponse. Imputation often adjusts for
item nonresponse, and parametric models may be used for
either type of nonresponse.
 Ignore the nonresponse

10/12/2020 fskm.../sept2013 61
ERRORS OF OBSERVATION
INTERVIEWERS
 An interviewer’s manner, appearance, gender,
accent and opinions can affect the way in which
the respondent answers the questions. This is
known as interviewer bias.
 Further sources of interviewer bias include the
way in which the interviewer asks the survey
questions, prompts for an answer, makes
assumptions about the respondent’s answers
and records the answers.

10/12/2020 62
ERRORS OF OBSERVATION
RESPONDENTS

 Response bias is the error introduced into the

survey through inaccurate answers being given
by the respondent to the questions asked. There
are many reasons why this may happen.
1) The questionnaire may include questions
that are loaded, misleading or ambiguous,
hence the respondent provides the wrong
answer.

10/12/2020 63
ERRORS OF OBSERVATION
2) The questionnaire is designed badly with
poorly worded questions; hence the
respondent misunderstands the question,
providing a wrong answer.
3) Through protecting their personal privacy,
integrity and interests, respondents may not
answer a question truthfully. They may give
what they perceive to be socially desirable
responses, particularly to questions of a
sensitive nature.

10/12/2020 64
ERRORS OF OBSERVATION
MEASUREMENT INSTRUMENT
 Measurement bias results from poorly
measuring the outcome you are measuring.
 Items to be measured must be precisely defined
and be unambiguously measurable.
 In any measurement question, the unit of
measurement must be clearly defined i.e.
inches, kg, cm and meter.

10/12/2020 65
ERRORS OF OBSERVATION
METHODS OF DATA COLLECTION

 Personal Interviews
 Telephone Interviews
 Self-administered Questionnaires
 Direct Observation

10/12/2020 66
 Reducing Errors
in Surveys

10/12/2020 67
 Approach to reduce measurement error
• Test your questions
• Interview potential respondents to see if
1 they interpret the questions as you intend.

• Write clear questions.

• If a respondent does not know how to answer a question,
the interviewer is likely to have more influence on the
response in a self-administered survey, unclear questions
2 can lead to more variability or bias in the responses.
Open-ended questions may be more susceptible to
interviewer effects than closed questions.

• Provide training and supervision for interviewers so they

act consistently,.
• Interviewers should read the questions exactly as written
3 and should not indicate that noneresponse is preferred
over another.
10/12/2020 68
REDUCING ERRORS IN SURVEYS
 Callbacks
 Rewards and Incentives
 Trained Interviewers
 Data Checks
 Questionnaire Construction

10/12/2020 fskm.../sept2013 69
THANK
YOU

10/12/2020 Shamsunarnie... JTMSK 70