Chapter 4.

5,6,7
Sampling and Sample Design
 How and why sampling relate to
business research?

 The world is large and full of people

 We wanted to find out things about people
 Sampling is a practical way of studying people and
their activities, thoughts, attitudes, abilities,
relationships in relation to business
 Note that sample must be representative of the
population from which it is taken
 Sampling: Important terms

 Population: is the total set of units in which a researcher is

interested; Can be finite or infinite population
 Examples: All employees of an organisation to study the reasons of
employee turnover
 Element/case: a single member of the population.
 Census: includes all the elements in the population
 Two conditions are appropriate for census study:
 when the population is small ( for populations under 50 it is usually more sensible to
collect data from the entire population)
 When the variability is high (when the elements are quite different from each other) and
when the size is manageable

 Sampling: is the process of selecting units into a sample from a

larger set of the same units (Population)
3
 Sampling: Important terms
 Sampling frame: a listing of all the elements in the
population from which the sample is drawn
 For example the list of employees found in personnel department to
get information on employee turnover
 Unit of analysis: the type of object whose characteristics the
researcher wants to measure and study.
 For example: If data are collected on Employees, the unit of
analysis is employee.
 Is the object that the hypothesis describes.
 All variables in a hypothesis must be operationalized for
the same unit of analysis.

4
 Sampling: Important terms
 Sampling unit: a unit or set of units considered for selection at
a stage of sampling.
 Sampling unit may or may not be the same as a unit of
analysis. It is possible to include several units of analysis.
 For example, if the researcher wants to interview senior managers in
the public sector, the senior managers become the unit of analysis and
the public organisations across the country become sampling unit.
 Parameter: is a characteristics of the population about which
researchers are interested to find out.
 Example: The average income of all families in a city or the
age distribution of the city’s population.

5
 Sampling: Important terms
 Sample bias: misrepresentation of the population by the sample.
Caused by the flaw in the design or in the implementation of
sampling procedures.
 Sampling fraction: % of population selected for the sample
 Sample size: the number of elements selected for the sample to
represent the population.
Sample size determination is influenced by:
 the purpose of the study,
 population size,
 the risk of selecting a "bad“ sample,

6
 Sample Design

 Sample design: the set of procedures for selecting the

units from the population that are to be in the
sample.
Two major types of sample design
 probability sampling:
 non probability sampling

7
 I. Probability sampling
Four types of probability sampling
 simple random sampling:
 Systematic random sampling
 Stratified random sampling
 Cluster sampling
Simple Random Sampling
 Each unit in the population has equal chance of being
selected.
 Can be lottery method or a random number table
 It requires a complete list of the study population. The
researcher assigns each member of sampling frame a number
before selecting sample units
8
 1.1. Simple random sampling
 Helps to eliminate the inadvertent introduction of sample
bias.
Example: assume there are 150 employees (with BA degree and above) in
the organization with the problem of high employee turnover. If the
sample size is 35 employees. Use lottery and random number table. to
select the sample elements.
Procedures:
For both lottery and random table case
1. Identify the population: All employees with BA degree and
above in the organisation
2. THE sampling frame: The list of employees with BA degree
and above, names are sequentially numbered from 001 to
150
9
 1.2. Simple random sampling
For lottery case
3, Prepare numbered balls equal to the number of the population
4. thoroughly mix up the balls and then
5. Draw 35 balls from the 150 balls.
For random number table case
3. Open Excel file and put =Round(rand()*150, 0) in the first cell and
copy this value to the other 34 cells that follow the first cell.
4. Block again all the 35 cells, select copy, select paste special and
select values.
5. Sort the values and the whole numbers are the sample units
selected for your analysis.
6. In case if there are duplicated whole numbers, replace them with
new whole numbers drawn randomly until the sample size is
equal to 35.

10
 1.2. Systematic Random sampling
Requires the complete list of population, Reduces the amount of effort
required to draw a sample and provides adequate results
Applicable when the researcher wants to pick households in the sample from
the population of consecutive households found along a street/road.
Procedures:
 Population has N units. Plan to sample n units and then
 The sampling interval/skip= N/n------K
 Line-up all N units and Randomly select a number between 1 and K
 Select the randomly selected unit and every kth unit after that
Example: the list contains 10,000 element and you want a sample of 1,000:
 Sampling interval = Population size/Sample size=10
 Randomly select a number between 1 and 10. Assume the first element in
the sample is number 7, then the selection of elements continue as 7, 17, 27
…, 9987, 9997
disadvantage
 Does not result in a truly random sample or suffers from the problem of
periodicity.
11
 1.3. Stratified sampling

 Involves a process of stratification or segregation, followed by

random/purposive/sample from each stratum.
 1st: divide or classify the population into strata, or groups, on the
basis of some common characteristics such as sex, race, or
institutional affiliation, level of management, or income, etc.
 Mutually exclusive groups: the classification should be done so
that every member of the population is found in one and only
one stratum.
 Separate samples are drawn from each stratum.
(proportionately or disproportionately).
 It ensures homogeneity within each stratum, but heterogeneity
between strata

12
 1.3. Stratified Sampling Example
 Problem: The researcher Educ level No. Samplin No. in
wants to study about the of g sample
empl fraction
satisfaction level of
employees.
PhDs 85 10% 9
 Population: 800 employees
MAs 150 10% 15
 sampling frame: all employees
and the list of employees in BA Degree 200 10% 20
personnel department. Diploma 130 10% 13
Holders
 Sample size : 80
Highschool 235 10% 23
 sampling fraction : 10%
or below
 Procedure: stratify the Total 800 10% 80
population based on education
level

13
 Stratified Sampling: disproportionate

 Problem: Assessing the

transportation needs of Bole House No of Proporti Samplin Number
Sub-city. hold House on of g in
(Stratum holds total fraction sample
 Population: 75000 people in )
15,000 house holds
 sampling frame: the list of Zone 1 6000 40.00% 6% 360
households in the four Zones
Zone 2 4500 30.00% 6% 270
of Bole Sub-city office.
 Sample size : 1000 Zone 3 3500 23.33% 6% 210
heads/member/ households
Zone 4 1000 6.67% 10% 100
 Procedure: stratify the
population in zones Total 15000 100.00% 940

14
 A researcher splits the population of interest into three
subsets based on their age groups:
 Subset A (16–25): 120,000
 Subset B (26–35): 80,000
 Subset C (36–45): 100,000
 Disproportionate stratified sampling means the researcher
randomly chooses members of the sample from each group.
So, you could have 60,000 participants from the first group
and 20,000 and 17,000 from others, respectively. There's no
clear-cut method for choosing the variables for the research
sample.
 A key advantage of disproportionate sampling is it allows
you to collect responses from minority subsets whose
sample size would otherwise be too low to allow you to draw
any statistical conclusions.

Tesfaye D 11/15/2012 15
 Cluster sampling

 It involves division of elements of a population into groups-the groups are

termed clusters
 Recommended when:
 it is necessary to study a large geographical area and
 It is difficult to identify the sampling frame
 The geographical distribution of the members is scattered
Stages in cluster sampling
1. The sampling frame is the complete list of clusters rather than individuals
2. Select a few clusters, normally using simple random sampling technique.
3. then collect data from the cases within the selected clusters either using
census or by taking sample.
Note: Cluster sampling can also be done at several stages, and then called
multistage cluster sampling.

16
 Cluster sampling

Example:
 The Problem: The AA city administration wants to assess the problem of
transportation in AA
 Population: all households in Addis Ababa
 Sampling Frame: List of sub cities, list of Woredas, and list of Kebeles (Villages), List of
households
 Procedure: construct a four-stage cluster sampling
1. Randomly select a sub city from the lists of sub cities
2. Obtain lists of Woredas from the selected subcity and randomly select a Woreda
3. Obtain lists of kebeles/subworedas from the selected woreda and randomly select a
kebele/subworeda
4. Obtain lists of households from the selected kebeles/subworedas and obtain
information from all households or by randomly selecting the sample elements.

17
 Cluster sampling
 Different from stratified sampling because Every cluster is
not sampled where as every stratum is sampled in the case
of stratified sampling.
 Saves time and money
Disadvantage:
 it may require larger sample than other methods for the
same level of accuracy
 Susceptible for the loss of key information as a result of
random selection and re-selection process of groups.
 (better to use weight, based on the number of people living in
the cluster, in the random selection process.

18
 II. Non probability sampling

 Four types
 Convenient sampling
 Purposive sampling (Expert sampling)
 Quota sampling
 Snow ball (referral Sampling)
Non-probability sampling designs
 Can work well for exploratory studies
 Useful if it is not important to obtain accurate estimates of population
characteristics
 The units are selected at the discretion of the researcher
 Cheaper and easier to carry out than probability designs
Some of the disadvantages of non-probability sampling:
 one cannot estimate parameters from sample statistics
 Such samples would not be a representative of the population : does
not rely on random sampling

19
 Convenience sampling

 also called haphazard or Accidental sampling

 Involves collecting information from members of the
population who are conveniently available to provide it.
 For example: collecting information from Volunteers
 Criteria: The availability/ the ease of obtaining/ and
willingness to respond
 convenient and economical to sample employees in a
nearby area
 During election times TV channels often present man-on-the-street
interviews to reflect public opinion.

20
 Quota sampling
 selecting a quota of individual units with defined characteristics in
the same proportion as they exist in the population.
 address the issue of representativeness (gender: two categories:
male, female; Class level: graduate and undergraduate, social-
economic class: upper, middle, lower)
 A type of stratified sample in which selection of cases within
strata is entirely non-random.
 Is called Dimensional sampling If all dimensions of the
population are considered in quota sample
 Example: A researcher is interested to assess the attitudes of employees
towards working condition. male are 60 percent and female are 40% in the
organizations: If Sample size is 30 employees, then 18 conveniently available
male and 12 female workers will be sampled

21
 Purposive sampling

 Is judgmental/ deliberate sampling

 It invites the researcher to identify and target individuals who are
believed to be typical of the population being studied.
 The researcher uses his own judgment about which respondents to
choose, and picks only those best meet the purposes of the study.
Expert sampling: involves selecting persons with known experience or
expertise in an area.
 With purposive sampling the sample is ‘hand picked’ for the research
 Example: A Local government uses purposive sampling when it seeks
information from cities with a reputation for excellent administration
(about their experiences, with outsourcing services, what performance
measures they use, how they monitor citizen satisfaction).

22
 Snowball sampling/referral sampling

 Snowball: Individuals are discovered initially, and then each

individual is used to locate others (the names & addresses)
who possess similar characteristics and who, in turn, identify
others.
 Used when members of a population cannot be located
easily by other methods and where the members of a
population know each other.
 Example: we may want to sample very small populations
who are not easily distinguishable from the general
population or who do not want to be identified, example
drug users, homeless people
23
4.7 Data Collection
instruments &
Writing Research
Proposal
 Understand methods of primary data
collection techniques and tools;
 know methods of secondary data collection

techniques and tools;

 have the skill of proposal writing

25
1. What are the Types of Data?
2. What are the ethical issues in Data collection?
3. What are the methods of primary data collection?
4. Questionnaire:
a. Closed ended: Advantages and disadvantages
b. Open ended Advantages and disadvantages
c. What are the do’s and don'ts in developing questionnaires
5. Interview
a. Types of interview
b. Advantages and disadvantages
6. Observation
a. Types of observations
b. Advantages and disadvantages
7. Focus group discussion
a. Moderator
8. Secondary data
a. Sources
b. Advantages and disadvantages

26
 Ethical issues: protection from harm, informed consent, right to
privacy and honesty with professional colleagues
 Data can be primary or secondary

1.1. The primary data:

 are data collected for the first time and hence they are original
 Major sources of primary data are diaries of eyewitness, tape-
records, films, letters and autobiographies
1.2. The secondary data:
 are those which have already been gathered by someone else and
which have already been passed through the statistical process
 These include historical studies based on the actual data, statistical
research based on census data, newspaper reports of an event,
company records, government publications

27
 The questionnaire is usually mailed, administered
personally or electronically
 Preparation of questionnaire can be tedious,

involving several drafts and more than one pretests

Questions can be asked to gather information on:
 Facts: help to get objective information from

respondents. Examples are gender, age, marital

status, education, income, etc.
 Behaviour: behaviour questions help to get

information about what people do. Examples: “Have

you ever attended any public lecture”

28
 Opinions: asking people what they think about
specific issue or event
 Attitudes: help to get information on the
underlying belief of the respondent or the way
people look at things.
 Motives: asking people why people behave in a
particular manner or hold certain opinions or
attitudes.
 Knowledge: It helps to obtain information about
the extent of knowledge an individual has and to
what extent the information is accurate.

29
Types of Questionnaires: open-ended or closed-ended
Closed ended questions allow only answers which fit into
categories that have been established in advance by the
researcher.
◦ Open-ended - unrestricted type of questionnaire, free response
in the respondent's own words; leave the respondent to decide
the wording and the length of the answer and the kind of matters
to be raised in the answer.
Advantages of closed- ended questions:
 The answers are standards, and can be compared from person to
person.
 The answers are much easier to code and analyze
 They are easier for a respondent to answer as he or she merely
choose a category,
Disadvantages of closed-ended questions:
 guesses or randomly answers if a respondent does not

know the answer or has no opinion

 There is no opportunity for the respondent to clarify or

qualify his or her answer

 Differences in interpretation of what was meant by the

question may go undetected

 Variations in answers among the different respondents may

be eliminated artificially by forced- choice responses

 A likelihood of a clerical error as the respondent circles

31
The advantages of open-ended questions:
 They can be used when not all of the possible answer categories

are known and to see what the respondent views as appropriate

answer categories
 To allow the respondent to answer adequately, in all detail he or

she likes, and to clarify and qualify his or her answer

 They can be used when there are too many potential answer

categories to list on the questionnaire – you can not list too many
questions in a given questionnaire
 They allow the respondent to have more opportunity for creativity

or self-expression

32
Disadvantages of open-ended questions
 The possibility of collecting worthless and irrelevant information

 Data are often not standardized from person to person, making

comparison or statistical analysis difficult

 Coding is often very difficult and subjective

 require a lot of time for analysis

 may require more of the respondent's time and effort, and may
engender a high refusal rate /reduce respondents’ willingness to
take part in the research.
 possibly discouraging some respondents who do not wish to
answer a lengthy questionnaire.
 make each question brief and the wording clear and concise with
minimal use of jargon
 keep the length of the questionnaire to a minimum: a maximum
of around 20 questions is probably a good guide for most
surveys.
 Make all definitions, assumptions, and qualifiers clearly
understood
 Avoid making significant memory demands
 make the questions simple to answer
 Keep it interesting – don’t make it monotonous, do have a logical
sequence to the question
 Avoid biased, loaded, leading, or sensitive questions.
◦ Example: ‘are you a heavy smoker?’ Instead use ranges:. Please indicate
your smoking habit: less than 10 cigarettes a day, between 10 and 20,
over 20
◦ Leading: what are your views on the level of VAT in Ethiopia? Is better than
‘ do you agree that the level of VAT in Ethiopia is too high?

34
 start with simple questions such as gender, leaving
more complicated questions to be answered late in
the questionnaire.
 avoid asking personal questions
 avoid asking the same questions in a different
fashion
◦ Abortion should be legalized - agree / disagree) at one
point and Abortion should not be legalized
(agree/disagree)
 Don’t ask two questions in one/ double barrelled
questions- with two purposes or interpretations/
◦ Example: Is your job interesting and well paid? Is unlikely to
be answered with a simple yes or no
 Don’t ask hypothetical question such as winning the
National Lottery).

35
I. Introductory statement of a Questionnaire
state the purpose of the study and its significance;
explain who the data collector is, the basis of its authority/the
sponsor-under whose auspices is the research being undertaken?,
and why it is conducting the study;
tell how and why the respondents were selected;
explain why their answers are important;
tell how to complete the format and list the person to call if help
is needed to complete the form;
provide assurance of confidentiality and anonymity when
appropriate;
explain how the data will be used;
explain who will have access to the information; and
present the response efforts as a favor and thank the respondents
for their cooperation.

36
II. Format of the questionnaire
Cleanly format your questionnaire: margins,
spacing, font, etc.
Proofread your questionnaire carefully for
grammar and spelling errors
conduct a pilot survey on a small sample so that
to test the reliability and validity of your measure.
Schedule vs questionnaire method
 Schedule is a device consisting of a set of
questions, which are asked and filled in by the
enumerators who are specially trained for the
purpose

37
 Could be conducted: face-to-face/in-person interview; or
telephone call, through internet/online
 Types of Interview: structured , unstructured and semi-
structured
a) Structured interview
 Set of predetermined questions and of highly standardized
recording technique (audio or digital recorders)
 The same types of questions are presented in the same order
to each interviewee
 The interviewer has no freedom to rephrase or change the
order of questions to be presented
 quite often used in the case of descriptive studies

38
Unstructured interview
 The interviewer has a general plan of inquiry but not a specific

set of questions that must be asked in particular words and in a

particular order
 characterized by flexibility

 The interviewer is largely free to arrange, rephrase, modify,

and add some new questions

 Very important for exploratory research studies

Semi-structured interview
 Shares the nature of both structured and unstructured

interview

39
 Prepare for interview, self-presentation:
objective of the study, its background, how the
respondent was selected, the confidential nature
of the interview, the beneficial values of the
research findings
 Conduct the interview – use your communication
skills here (attentive, non-judgmental), ask the
questions properly, probe meaningfully
 Recording of the interview; record the responses
accurately, record response as they occur; use
some shorthand system (abbreviating words, key
words)

40
Advantages:
 More and in-depth information can be obtained
There is greater flexibility
Personal information can be obtained easily
high response rate
The language of the interview can be adapted to the ability
the person interviewed
Disadvantages
 Expensive including Cost of selecting, training and
supervising the field-staff
Bias of interviewer and the respondent - presence of the
interviewer on the spot may over-stimulate the respondent -
may give imaginary information
Important officials or executives may not be easily
approachable
More-time-consuming, when calling the respondents

41
 observing what is occurring in some real - life situation , without
asking questions of respondents
 It is valuable instrument in a wide range of research studies.

◦ Examples: Cultural study, traffic counts, direction of traffic flows

Planning and execution of observation
 Selecting an appropriate group of subjects to observe

 Identifying the specific activities or units of behavior to be observed

and focusing attention on same at the time of implementation

 Proper arrangement of specific conditions for the subject(s) to be

observed
 Assuming the proper role or physical positions for observing

 Handling well the recording instruments to be used

42
a) Direct versus Indirect observation
Direct observation: the observer is physically present
and personally monitors what takes place
Very flexible - the observer can react to events
The observer is free to shift places, change the focus
of observation, or concentrate on unexpected events
weakness - the observers' perception may become
overloaded as events move quickly; they must later try
to reconstruct what they are not be able to record
Observer fatigue, boredom, and distracting events can
reduce the accuracy and completeness.

43
Indirect observation
 The recording is done by mechanical/adjusted instruments

◦ Example: a special camera that takes one frame every

second is mounted in a department of a large store to
study customer and employee movement
Less flexible but much less biasing, less unpredictable or

erratic in accuracy
The permanent record can be analyzed to include any different

aspects of an event

44
b) Disguised Vs undisguised observation
The role of the observers should be disguised in situations

where people behave differently if they know they were being

observed
Often technical means are used such as one-way mirrors,

hidden cameras, or microphones

Reduce the risk of observer bias but bring up a question of

ethics
◦ Hidden observation is a form of intelligence work
A modified approach - the presence of the observer is not

concealed, but the observer´s real purpose and subject of

interest are hidden

45
c) Participant Vs non-participant observation
 Participant observation: The observer becomes one of the
groups under observation
Non-participant observation: Observer takes position where
his presence is not disturbing the group.
d) Structured Vs. unstructured observation
Structured observation is systematic and has a high level of
predetermined steps
Objective: To quantify behavior (your focus is to determine
how often things happen rather than why they happen. Ex:
Time and motion study
Unstructured observation: The observer has no definite
ideas of the particular aspects that need focus.
Observing events that are happening may also be a part of
the plan as in many qualitative studies.

46
 the observer must take utmost care to minimize the
influence of his biases, attitudes and values on the
observation report
Advantages:
 Useful in locating data about non-readers, young
children, people with mental disorders, and
laboratory animals
 The data obtained through observation of events as
they normally occur are generally more reliable and
free from respondent bias.
Disadvantages:
 It is time consuming
 It is costly to collect data.
 The data may reflect observers’ bias

47
 a special type of interview that offers opportunity to
interview a number of people at the same time.
 Made by a panel of 8 to 12 respondents led by a trained
moderator
 The moderator uses group dynamics principles to focus or
guide the group in an exchange of ideas, feelings, and
experiences on a clearly understood topic
 Good for exploratory research
 Qualities of a moderator: (Kindness with firmness,
Tolerance, Involvement, understanding, Encouragement,
Flexibility, Sensitivity / emotional response)
 Benefits of FGD : (Synergism , Snowballing, Stimulation,
Security, Spontaneity—/natural behavior/,
Serendipity/discovery of something fortunate,
Specialization, Scientific Scrutiny, Structure, Speed)

48