BRM Module-3 2022

Business Research
Module-3 Measurement

Prepared by
Dr. Arya Kumar
Asst. Professor
SCHOOL OF SOCIAL, FINANCIAL & HUMAN SCIENCES, KIIT DU
Bhubaneswar

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 1

 BRM Module-3 2022

What is a scale?

A scale suggests a systematic process of allocating the numbers or symbols such process is
called Scale of measurement.

A measurement scale may be represented from a highest point to lowest points. Such may
also have the sub-points in between it. The measuring instruments are scales.

Essentiality of measurement in research:

It is used to quantify the variable/ construct/ an object in respect to certain properties. This
may be books, people, tables, plants, chairs and etc. or certain attributes like anger, aptitude,
beauty, personality and etc.

However, a researcher doesn’t consider the objects or its relevant properties for analysis
rather they consider the indicants of object’s properties. There are several characters that are
not easy to quantify like level of hard work, dealing with issue or etc.

Types of measurement scales as per Data

A researcher designs certain rule for mapping and then transfer the observation to the
property indicants using such rule. As per the rules, a qualitative and quantitative data can be
presented as per the construct/ variable that are feasible to measure. The allocation of the
numbers or symbols is assigned under certain assumption from the real world observation.

Mapping rules:

 Order: Ordering of numbers

 Distance: stating about a range i.e. stating about two set of points
 Classification: Grouping or classifying the numbers
 Origin: The number should have unique origin. It may be zero.

Classification of Scale

a) Nominal Scale

This is one of the least powerful, simple and common form of measurement of scale. The
scales are just the numbers and restrictive among all the scales. The scaling is done on all the
facts/ choices or various categories that are exclusive in nature. This scale shows all the
categories of response into numbers of mutually exclusive categories. If there exist no
relationship between the categories then there is no need for ordering or ranking. Generally in
nominal scaling the responses are yes or no, like or dislikes male or female and etc.

For Example:
Gender
1. Male
2. Female

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 2

 BRM Module-3 2022

Another example between two dimensions:

Investment decision
1. Saving Deposit
2. Equity Share
3. Mutual Fund
4. Fixed Deposit

Annual Income
1. 1,50,000-2,50,000
2. 2,50,000-5, 00,000
3. 5,00,000-10,00,000
4. 10,00,000 and above

b) Ordinal Scale
It is a simple measuring scale usually used in social science and humanities based research.
However it is more powerful in comparison to nominal scale. The ordinal scale is useful in
ranking the categories of similar variable. The respondents are expected to rank all the
attributes/benefits of certain products.
This scale has a limitation that the degree of difference between the ranks cannot be
measured. It is useful to implement the positional statistics like mean, and mode for analysis
and interpretation.
Example- Rank the following attributes from 1 to 5, (stating 1 as the highest rank and 5 as the
lowest rank) on the prospective of qualities required for a good team work.
a) Leadership
b) Communication
c) Secrecy
d) Positive attitude
e) Loyalty
In this type of scale numbers are not only assigned certain symbol or letters can be also
been allotted. However, this scale fails in identifying the degree of favourability of each
rank.
A researcher can ask questions from more than one attributes like University, firms or
industry and call for likes and dislikes. For example, the respondents are satisfied with
public transport?

 Strongly like it
 Like it
 Neutral
 Dislike it
 Strongly dislike it
c) Interval Scale

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 3

 BRM Module-3 2022

It was observed that nominal and ordinal scale has several issues which were taken care of in
the interval scale. The interval scale has a regular interval between the points that may also
start from the point zero. It no way affects the statistical operations as the addition,
subtraction or any kind of computation can be done by identifying the mean and mode.
Generally the ordinal and nominal scale consider the measurement of attitude but the interval
scale provides them a range so as to make the data more appropriate for statistical operation.
Example:
A survey is conducted on the preference of buying a two wheeler (Put the tick mark, 1 as
lowest rate and 5 as highest rate).

Company Name 1 2 3 4 5
Design 1 2 3 4 5
Availability of spare 1 2 3 4 5
Mileage 1 2 3 4 5
Engine Capacity 1 2 3 4 5

Here, we can see the distance between two rates are similar from each other i.e. distance from
1 to 2 is similar to that of 4 to 5. But it won’t be interpreted that 4 is twice stronger than 2.
This result can be considered for considered for calculating the mean scores of every variable
for each respondent. From here the standard deviation can also be measured.

d) Ratio Scale-
This is a scale which has some complexity due to which it has less popularity in measurement
of scale. This scale need a base item upon which the comparison can be made. For instance,
in the Interval scale, a score 4 is not the twice of score 2 but in ratio scale it will be
considered as score 4 is two times stronger than score 2. A ratio scale denotes zero as the
natural point and any numbers placed should have equal appearing interval.

The ratio scales are used generally in physical scenarios like height, distance and etc. This
scale gets its popularity for the data that can be easily quantified. This scale is similar to
interval scale but its numbers are more meaningful for interpretations. This kind of data can
be collected mainly through open ended questionnaire where the respondents are allowed to
response without any force of option.
For example:
Which Cricket team is going to win the World Cup?
In which company you would like to invest your savings?
To answer the above question one can use an unambiguous starting point i.e. zero. As there is
no force in choosing the point so a ratio analysis can be made on it. For example, the
dividend announced by Company B in last year was 4% and company A had announced 2%
so one would like to invest in company B as it gives twice the amount of dividend of
company A.

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 4

 BRM Module-3 2022

A comparison Table of Scale

Basis for Nominal Ordinal Interval Ratio
Comparison
CharacteristicsIdentity of Assigning Difference
Zero as fixed
number and numbers for betweenand ratios can
classification objects as per be used for
of objects their position comparison
Statistics Percentage Percentile and Range, Mean, Geometric
implementation and Mode Median Standard Mean and
Deviation Harmonic
Mean
Example Cricket Ranking of Income or age Weight,
player’s students in an group height
numbering examination distance.
Classification Of measurement of Scale
The scales for measurement can be divided in respect to its response i.e.
 Rating Scale
 Ranking Scale

Stapled
Scale
Rati
phic

Scaling Technique
Gra

Non Comparative
ng
Ranking

Differe
Seman

ntial
Comparative

tic
Sum
Con
stan
t

Scale
Stap

Scal

Thurst

Scale
le

one
Seman

Differ
Measurement of Scales

ential
tic

Likert
Scale
Classification of

Guttm

Scale
Ranking Scale

an
Multi

choic

Liker

Forced
ple

Contin
Scaling Technique

Rating
Scale
uous
Comparative

Q-Sort
Scale
t type
Liker

scale
ratin
g
choic

Const

Scale
Mult

Mult

Sum
iple

ant
e

Comparison
Rating

choic
Mult

singl

Paired
iple

Order
Scale
Rank
e
Cate
gory
Sim
ple

compa
rison
Pair-
wise

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 5

 BRM Module-3 2022

Rating Scale- This scale is observed when a respondent response on certain object
without any comparison with other objects.
For example: Identifying a person’s knowledge on a particular field can be asked through
rating by himself. Instead of that as per the objective of research a researcher can design
a list of attribute that defines similar kind of objective which varies with certain degree
from the other attributes considering such a respondent will rate as per its convenience.
For example- Strongly like, like, neutral, dislike or strongly dislike.
As it is a rating scale the decision will be made from the list of given rates amongst 5 or
7 scaling points. However, the scale should not be too finely or coarsely. If there lies an
odd scaling then it is advisable to put the middle point as a neutral option for the
respondents.

Ranking Scale- Unlike rating scale, a ranking scale gives an opportunity to a respondent
to make a comparison between the objects that is prepared on the basis of degree of
difference between each other.

A) Rating Scale- This is a scale used for the judging of properties of an object. Generally,
the rating scale is used in social science or management research. Some of the terms used
in this scales are, yes/no, likes/dislikes, approve/disapprove or may be further
classification. A rating scale is used by a person to rate any product or brand.
However, there is no rule of thumb for designing the scaling points a researcher may
design with two set of points or more than it to provide a wide range of opportunity for a
greater accuracy.

The examples of rating scales are as follows:

1. Simple Category Scale- (Nominal Data)

It relates to questions based on dichotomous that states about two options for
selection like yes/no, true or false, likes or dislikes.
For example, would you like to work after completion of your graduation? Yes/No.
2. Multiple Choice Single Response- (Nominal Data) [ select any one]
For example: In which company you would like to invest money?
 TATA
 JINDAL
 RELIANCE
 SAIL
 NALCO
3. Multiple Choice multiple response- (Nominal Data)[select multiple option]
For Example- which toppings on pizza is your favourite?
 Anchovies
 Artichoke hearts
 Mushroom
 Pepperoni
 Sausage
4. The Likert Type rating Scale- (Interval Data)
A rating is done by a respondent that states about a well-structured space between the
ranges. It can help to interpret the direction and intensity of attitudes.
For Example: I would like suggest my friends to join my college

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 6

 BRM Module-3 2022

 Very likely
 Likely
 Neutral
 Not likely
 Very unlikely
5. Multiple choice likert type scale- (Ordinal and Interval data)
A series of question with similar answer scale
For example- Your experience on our service centre

Strongly Agree Neutral Disagree Strongly

Agree Disagree
A well behaved Staffs
Need not to wait for
longer time
Immediate response by the
supervisor
No additional charges
charged for work

6. Semantic Differential Scale- (Data Interval)

It is a rating scale that allows the respondents to rate a particular brand or products
upon seven point rating scale. The scales are rate under bipolar adjectives i.e. one
side states a positive adjective and other side states about negative adjective. The
respondent needs to rate in between that.
For example

Beauty 1 2 3 4 5 6 7 Ugly
Liberal 1 2 3 4 5 6 7 Rigid
Simple 1 2 3 4 5 6 7 Complicated
Stylish 1 2 3 4 5 6 7 Out Dated

7. Staple Scale- (Ordinal and Interval Data)

It is an another form of rating scale that allows the respondents to rate a brand/
product/ service on the basis of certain characteristics from -5 to 5. This help to
clarify how well the characteristics define a service or products.
For example: The growth of India in last 10 years- Significant growth or in
significant growth.

Significant -5 -4 -3 -2 -1 0 1 2 3 4 5 Insignificant

8. Constant Sum Scale- (Data Ratio)

It is a scale useful during the ratio measurement data. It is most powerful among all
measurement scales. This measurement scale includes an equal interval scale and
zero point scale.
Here the respondents are asked to mention the degree of preference by stating percent
i.e. 100 as the highest. It not only highlights the like or dislike of an object rather it
helps in identifying intention or strength to shift, stay or buy and etc.
For example:

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 7

 BRM Module-3 2022

Out of 100 rate each attributes that motivates you to buy a new home
 Location
 Price
 Furnished home
 Distance from Hospital
 Distance from school

The total sum of all the values allotted must be equal to 100%.

9. Graphic Rating Scale-

It is a scale designed with certain graphical or pictorial representation that establishes a
clear difference among all the scale. This is used to clarify the difference between each
point between the scale for those who hardly able to understand the difference in degree
between each point. Generally, other scale the measures length, height or weight are
difficult to quantify as it needs to have a predefined level for comparison. But here it
creates a clear interval of data the helps in coding and decoding of responses. The
graphic rating scale uses icons, pictures and etc. for rating.
These are very popular among the children.

For example: Did you like the taste of the chocolate milk shake?

B) Ranking Scale
It is a scale that allows ratters to make a comparison between two or more objects and
select the best amongst the all. A simple multiple choices tends to select the those that
the respondents feels like appropriate but a ranking scale gives them an opportunity to
select all as per their priority which makes this scale more powerful. A rank order
scale provides direction and relative position but not absolute difference. A rank order
defines the order but does not tell about how much an item is preferred by another.

Some of the example of ranking order scale which are as follows:

1. Paired Comparison scale- (ordinal data)

It is a scale of choosing the between two objects unambiguously.
For example- among the list of compared fruits select those fruits you want to
buy.
Mangoes Cherries
Apples Peers

Mangoes Apples
Cherries Peers

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 8

 BRM Module-3 2022

Mangoes Cherries
Peers Apples

2. Forced ranking Scale- (Ordinal data)

This is a phenomenon of listing the attributes that are relatively related to each
other. The process is faster in comparison to paired comparison scale, this scale
motivates the respondents to response.
For example:
Rank the list of books given below as per your preference. Jot rank 1 to the most
preferred book and number 2 as your second preference and so on.

 Twilight
 Lord of rings
 Harry Potter
 The lion, the witch and the wardrobe
But this scale has a limitation that the respondent may become more careless
in ranking such a huge number of items.

3. Comparative Scale (Ordinal Data)-

It is a scaling technique where a benchmark is set prior to collection of responses,
basing upon such benchmark upon such all the objects, brands, programs will be
compared. This scale is useful for those respondents who are familiar with the
objects or its attributes and have the idea about the benchmark.
For example: Refined oil is better than Olive Oil
Superior Same Inferior
1 2 3 4 5

A comparative scale is further divided into two sub-groups that is useful in the
field of social science/ business research
a) Comparative Scale technique
b) Non-Comparative technique

a) Comparative Scaling Technique- it is a techniques that allows in comparing every

single object or items that are available within the scale. A comparative scale should
be in ordinal or rank order. As a result it is often called as non-metric scaling. This
technique is simple and clear.
b) Non-Comparative Scaling Technique- unlike comparative scale each object and items
has independent scale which are not compared with each other. This scale are
normally ratio or interval scale. It is a technique that is widely used in marketing
research.

a) Comparative Scaling Technique

i) Paired Comparison Scaling:

It is a scaling method where a respondent is asked to select an object/ items and
choose one from the list of option. Here, data are normally ordinal in nature. For
Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 9
 BRM Module-3 2022

instance if there lies 4 items then this will lead to 6 comparison between the items.
Here in each paired the respondent is asked to select the on amongst the two as per
its preference.
For example: Select which Bollywood film star is a better actor (Shahrukh Khan,
Salman Khan, Amir Khan and Akshaya Kumar)

Salman Khan Amir Khan

Shahrukh Khan Akshaya Kumar

Salman Khan Amir Khan

Akshaya Kumar Shahrukh Khan

Akshaya Kumar Amir Khan

Shahrukh Khan Salman Khan

Here, when respondent select one and rejects the other then such responses are
gathered and represented by adding all the numbers for respective options. After the
addition the researcher will make an average for every option from the total number
of respondents. However there are several ways for comparing the data in paired
comparison analysis.
But this comparison is possible when the number of items is limited as it needs
direct comparison with the objects.

ii) Rank Order Scale-

It is a scaling method where the respondents are asked to rank the objects under
certain criteria. This is usually suitable in in measuring for brand preference or
attributes.
Suppose the survey was conducted on the best Bollywood film actors in order of
preference. From the following list they have to assign the ranks like rank 1 is
most preferable then rank 2 comes is for 2nd most preferable and so on till the
option get exhausted.
Ranking scale is more complex than rating scale as there lays a comparison
between each alternative available due to which more attention is required.

Sl. No. Actors Ranks

1. Akshaya Kumar
2. Amir Khan
3. Ranveer Singh
4. Salman Khan
5. Shahrukh Khan

iii) Constant Sum Scaling-

The respondents in these scaling techniques are asked to mention certain numbers
against the objects whose sum total will be same for all the respondents which is
fixed by the researcher. A respondent is also allowed to assign zero for any
attributes they wish to. For instance, the researcher fix a total value for the entire
attribute is 100 then the respondents are allowed to assign numbers to the attribute
as per their preferences.

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 10

 BRM Module-3 2022

For example: a group of three has respondent on the basis of preference regarding
buying a shaving cream/ foam.
Average Responses
[Link]. Attributes Group-1 Group-2 Group-3
1. Price 20 18 29
2. Fragrance 35 26 22
3. Lather 10 33 15
4. Packing 5 18 9
5. Brand 30 5 25
Total 100 100 100

The advantage of this scale is it helps in fine discrimination of the attributes and helps
in making decision.
iv) Q-Sort Scaling-
It is a scale that helps in discriminating a large number of object’s characteristics
with certain attributes. The objects are sorted with certain piles as per the similar
criteria.
For example a respondent is provided 100 card containing various attribute/object/
characters. They were asked to pile or group them into 10 piles starting from
higher preference to lower preference.

The researcher will fix the number of attributes to be placed in each pile. This
practice helps in normal distribution of data. After the pile is prepared the
respondent is asked to rank the attribute from the most preferred object or
characteristics.

v) Guttman Scaling-
This scaling technique was founded by Louis Guttman during 1944 which is
sometimes called as Cumulative scaling or scalogram analysis. This scaling
technique provides a uni-dimension concept basing upon which respondent will
respond. The scaling technique is designed in such a way that the respondent if
agrees with item 3 it automatically means the person is agreed on item 1 and 2.
Thus the respondents who agree with a particular item have a higher score than
those who disagree with it. In practical sense, this scaling technique has high
popularity as it is helpful in making complex analysis that has a cumulative
pattern. So this technique also help in understanding that how close each items are
towards each other.
Following steps are to be followed in preparing Guttman Scale:
a. Define focus-
The first step is to identify the area that needs to be focused
b. Develop the items-
Identify a large number of items that defines the concept. Several books,
journal or expert advices are required to identify the attributes.
c. Rate the items-
Allow the respondents after preparation to identify which attributes has
highest value and favourable in explaining the concept. They were asked
to respond by yes/no.
d. Develop cumulative

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 11

 BRM Module-3 2022

Here, a matrix table is to be prepared that will assign all the responses that
are collected from respondents. The assignment of attributes in matrix will
be prepared as such those attributes gets highest respond will be placed at
top and those that get lower respond will be placed at bottom. If the
response is found to be of similar agreements then they will be placed
from left to right from those that most agreed to fewest agreed to.
Guttman scales have cumulative meaning for the researcher not to the
respondent’s response to a single object. So we need to consider the scale
items and scale scores associated with them.
For example
Scologram
Score 3 2 1 0
Agree 3 √ x √ √
with 2 √ x x √
1 √ √ √ x
Disagree 3 x √ x x
with 2 x √ √ x
1 x x x √
Note: For each agreed respond it gives a score as 1 which will be
cumulated to identify the highest preference amongst all.

b) Non- Comparative Scaling Technique

i) Continuous rating scale

It is a scale that sometimes called as graphic rating scale. The respondents in this scale
are asked to put a mark on the proper place on a given line. The extreme ends of the
line contain certain variable that are completely opposite to each other. It gives an
advantage to the respondent of not selecting predefined points rather they are free to
move to any extreme side or in between it to give their response. Here the respondent
will mention a mark which will not be compared to any numbers or point in between
it.
For example:
How much will you rate your present employer?

Highly 10 20 30 40 50 60 70 80 90 100 Less

Friendly Friendly

After the response the researcher can divide the line into several groups and assigns
scores as per the groups in which the rating tends to fall. It is easy to construct but the
scoring may be unreliable.
ii) Likert Scale
A likert scale is a scale of psychometric (attitude based). This scale is mostly
found in researches based on questions that are related to social sciences or
business research. This is a scale that sets several points for any particular objects
Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 12
 BRM Module-3 2022

from agree to disagree or like to dislike. This scale asks the respondents to
respond as per their agreement level for every set of questions. The scale is
developed during 1932 by Rensis Likert. It should have the continuous
representative sample of all possible attitude or opinions about an object. The
scale should be formulated in such a way that half of it will show favourable and
remaining half as unfavourable. In case, the scales have odd point then the middle
point will be considered as neutral.
The development of point is made as per the degree of favours or un-favours
result to which the points has five points or seven points or for more accuracy nine
points. Sometime this scale is prepared with force i.e. there will be no middle
point that states about neutrality in preference.
This scale should be presented in bipolar scaling i.e. one side should be positive
while other as negative.

At last the response from the questionnaire will be summed together to prepare the
test score as a result it is otherwise called as summative scales.

For example:
Football can be considered as the most effective outdoor game
 Strongly agree
 Agree
 Neutral
 Disagree
 Strongly disagree

iii) Thurstone Scale-

It is a scale developed during the 1928 by Louis Leon Thurstone. It is a scale of
measuring people attitudes that is designed by asking whether they agree or
disagree to a large set of questions. The statements are designed parallel to each
other but some of them are set at one end while some at the other end indicating a
slightly different attitude from each other.

This scale is different from likert scale. A likert scale ask the respondents to
respond on the basis of the intensity of an agreement or disagreement while
Thurstone scale provides a huge set of concepts or range that vary from highly
favourable to highly unfavourable keeping a neutral respond in between. They
were asked to select those attribute that they feel is very close to their own
attitude.

Some also call it as method of equal appearing intervals as the objective of the
scale was to obtain uni-dimensional scale with interval properties.

Steps involved in preparing Thurstone scales:

1) Prepare a large set of statements (100 +) that can be presented with two ends
i.e. positive and negative ends towards the attitude along with neutral point in
between.
2) After preparation of statement it will be confirmed regarding the degree of
favourableness and unfavourableness from an judges or knowledgeable
person. This is usually prepared with 11 bipolar points.

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 13

 BRM Module-3 2022

3) The values are then collected and mean/median value is calculated for each
statement considering the number of judges who has rated the statements.
4) After such standard deviation (σ) or quartile deviation (QD) is computed if the
judges tends to give high rating then σ or QD will be low and if the response
are poor then σ or QD will be high.
5) After the calculation of σ or QD for each statement the result are positioned in
the favourable and unfavourable point from 1 to 11. Any responses if found as
high variability then such response will be rejected. If the response for tw or
more statements are found to be close or same then the one with minimum σ
or QD will selected for interpretation.
6) Finally, it should be kept in mind of not showing the responses of respondents
as this may bias others while giving the ratings.
For example:
If a respondent agrees on a statements that gives a score like 2.4, 1.9, 3.2 and
2.9 then the attitude score is:
2.4 + 1.9 + 3.2 + 2.9 = 2.6
4
Here, only those responses are taken into consideration that brings a full
responds to the statement.

4.3.5 Comparison between Likerts and Thurstone scales

Likert Scales Thurstone Scales

It is simpler to construct It is complicated in comparison to
Likert
It gives scores for interpretation Give more absolute figure for
interpretation
It gives the degree of favourableness The participation of Judges is must
of attitude towards a particular topic who might fail difficult in rating the
items

iv) Semantic Differential scale:

It is a scale that identifies the people’s reaction to a stimulus words or concepts
that is represented in bipolar scales. The bipolar scales are contrasting to each
other
For example

Sad 3 2 1 0 1 2 3 Happy

Note: 0 denotes neutral, 1 slightly, 2 quiet and 3 denotes extreme

This scale is beneficiary in measuring the direction of any particular reaction.

People were asked to rate it on a number of such as per the interest or any
particular concept. Various combinations are considered to identify person
feelings or emotions.

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 14

 BRM Module-3 2022

This scale is beneficiary to measure the psychological attribute of any particular

object. Like other rating scale the respondents are asked to rate a point guided
with two extreme ends but the points describe the degree of any of the extreme
ends or objects. Some researcher is free to use one sided objects for response
which is called unipolar scales. In unipolar scale the middle point may be as angry
or not angry but in bipolar there exist neutral. There is no rule of thumb to
mention the point for scaling it may be as many point as a researcher wants to
mention.

After all this steps the total score can be generated by totalling all rating scores
and finding the average of it. This process helps in identifying the meaning behind
the objects and the reaction/ perception to the object.

This scale not only provides an interval scaling but it is a scale that is highly
efficient and very convenient in collecting a large number of response based on
attitude. It is highly versatile rating scale due to which it gains its popularity in
social science and business research.

For example- Semantic scales for (selecting an admission into an University)

Low Course Fee - - - - High Course fee

Average Placement record - - - - Good Placement Record
Remote Location - - - - Well located
Average passing record - - - - Good passing records

v) Stapel scale-
It is a scale that called as unipolar rating scale with the scaling point of 10 which
is numbered from -5 to 5 including 0 as a neutral point. This scale is named after
Jan Stapel. The respondents are asked to rate an object from the numerical
response category that may be presented horizontal or vertical. The higher the
score in positive the higher the adjective of the object if one moves towards the
lowest point of scale it relates to less adjective for an object.
For example: rating for the selection of an University for admission

-5 -5 -5
-4 -4 -4
-3 -3 -3
-2 -2 -2
-1 -1 -1
Teaching 0 Price 0 Placement 0
1 1 1
2 2 2
3 3 3
4 4 4
5 5 5

The data thus obtained from here can be interpret like that od Semantic
differential scale as the result that is produced from this is similar to semantic

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 15

 BRM Module-3 2022

differential scale. The reason that makes it unique from Semantic differential
scale is that it needs no bipolar adjective to define the attitude of an object.

5.5 Basic concepts in Sampling

A. Sample Design-

A sample design can be defined by following two factors

Sampling method- a rules and procedures through which some of the elements of population
are found to be a part of sample. Some of the widely used techniques are convenience
sampling, stratified sampling, random sampling, cluster sampling and etc.

Estimator- it states about the estimation process for calculating the collected sample. If the
sample methods are changed then sample estimation will also tends to change. For example
calculating a mean for simple random sample is different from clustering sample. Result to
this all the variables like standard error or quartile deviation will also tends to change.

To confirm the best sample design it is better to design as per the objective and survey
resources. If the researcher has less time, then it may select a design that takes less time even
if a researcher need accuracy of the sample and result then they may use a design that gives
cent percent result without worrying about time, energy or money.

5.5.1 Steps in Sampling

 Defining Target Population

 Identifying sampling frame
 Identifying sampling unit
 Selecting sampling method
 Confirming of sample size
 Identifying sampling plan
 Selecting the sample.

1. Defining the Target Population-

In the process of sampling the primary objective was to define the population interest
for business research. A target population is called those sampling unit, element,
people or objects that are purely relevant to the research topic or objective. For
example- a sport bike manufacturer would like to open a showroom, prior to that he
should collect the data through survey for those who can afford it and mainly from the
younger boys between the age of 22 to 30. As the topic survey is vast so maximum
data can be collected from the population described i.e. the target population and such
group from where data is collected is known as sample. This is a process which is
highly essential as it reduces the whole universe to a smaller set of group. This kind of
refinement can be done into various ways as per the research objective.
This process can reduce the probability as it reaches the real respondents for whom
the research is conducted. For instance if the research is under taken from the whole
Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 16
 BRM Module-3 2022

population then there is a chance that responses from non-target samples will
discourage and change the decision for opening a new showroom.

2. Specifying the Sample Frame-

After identifying the target population, the next work is to design the sample frame. A
frame for sample suggests listing of all the elements out of which the sample may be
drawn. For instance, a survey on selling of automatic chimney for such the survey or
sample frame will be a family with a monthly income of Rs. 30,000 the families
whose one of the member spend maximum time in cooking so this are some of the
frames that a researcher should design while selecting the sample. However, it is not
so easy to identify the accurate sample frame that suits to your research objective.
Researcher prefers the convenient way of selecting the sample that is easily available
like cell phone users, credit card users and etc. there are also several agencies those
who provide database on different demographic and economic variables. A researcher
can also use maps and aerial pictures as a mean for sample frames. Hence, sample
frames are those that represent the total elements which are available in the sample
from the entire population.
A sample frame also has certain issues when
 When the sample frame is over represented
 When sample frame is not represented accurately of the total population.

3. Identifying the sample Unit-

A sample unit signifies an unit which contains a group of elements or a single
elements out of the sampled population. By this approach:
 Universities can be called as sampling unit and
 The faculties in commerce can be sample element.

If the researcher able to identify the exact target population then each sample element
can be considered for sample unit by this way the researcher can be more focus to its
sample for collection data as per the research objective. So in a safe side it is wise to
select the universities as sample unit and every individual commerce faculty as the
sample element. This can be considered as secondary sample unit:

i) Individual unit
ii) Construction unit- house, building, apartment, flat
iii) Geographical unit- rural, urban, town, city, state, district
iv) Special Group unit- team, family, universities

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 17

 BRM Module-3 2022

Sampling
Unit

Analytical Unit Enumeration Unit

 Rate  Hypothetical Unit

 Ratio  Sample Unit
 Coefficient  Composite Unit

Application of sampling Unit

i) An experimental statistical unit is considered as one of the set of the object that is
initially equivalent with each object then subjected to one of several experimental
treatments.
ii) Universal example of a unit can be a single person, an entity, trees, animals or any object
that belongs to a larger group out of which the research is to be conducted.
iii) The sampling unit is characteristically thought of as an entity that has been sampled as of
statistical unit.
iv) This is a term frequently used during poll or survey of sampling
v) A sample may not be from the units observed from any meaningful population but
consider as a convenience sample or may represent the entire population of interest.
vi) In this statistical units descriptively or we may learning their dynamics above time.
vii) In extra or additional complex data sets multiple dimensions are made for each one in the
entire units.
viii) In easy data sets the unit is in on to on connection with the data values
ix) At the same time as units is frequently lowest level at which explanation are made,
further cases, a unit can be some rotten as a statistical gathering.

4. Selection of the sampling method

It discuss about the techniques for selection of sample units. The preference for selection of
the sample units is directly related to the objective of research, financial constraints, time
availability and the nature of problem to be identified. All the groups of sampling method can
be divided in to two groups:

a. Probability Sampling
b. Non-Probability Sampling

5. Determining the sample Size

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 18

 BRM Module-3 2022

It is one of the crucial stages of confirming the size of a sample. There are various ways
of defining the sample that is to be considered for analysis or deciding the size. Few
things that are considered during the decision on sample size:

a. Fixed or sequential sampling

b. Bayesian or traditional methods

In non- probability sampling technique some of the rules that are considered are as
follows:

a. Rule of thumbs
b. Nature of analysis
c. Importance of the decision
d. Availability of finance
e. Number of group and sub-group
f. Incidence rates
g. Number of variables
h. Time duration

In probability sampling technique there exist few formulas basing upon which the sample
size are calculated. The selections are done only after the confirmation of level of
acceptable errors and level of confidence. The details that determine the sample size are as
follows:
6. Specifying the sampling Plan-
It is a stage that defines the decision about the execution of the research process. For
instance, the sectors in a town are considered as sample unit and the houses are
considered as sampling element. This stage states about the nature of sampling plan of
selecting the houses on the basis of certain characteristics. This stage deals with
certain questions like:
a) How the interview is to be conducted?
b) Who will conduct the interview?
c) What an interviewer will do if no one is available for interview?
Similarly, various questions are to be framed and answered to make the data
collection process easy. These act as a guideline for the researcher that will help
the researcher at every step of research. Sometimes, the data are collected by a
third party who might face various obstacles in data collection so this plan can
provide various specifications for smooth function in interviewing the sample
element.

7. Selecting Sample
The last step in the process of sampling, here the actual work is carried out i.e.
selection of sample elements. Here, the interviewer and researcher should not be
lenient towards selection of huge number of sample as this may give a spurious result.
So a strict procedure should be followed to avoid such practice and get the true facts

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 19

 BRM Module-3 2022

of the research. This is a stage where the sample for survey is considered by following
the sample plan.

5.6 Sampling Error

An error in survey or sampling is a normal activity which may occur due to random
sampling or due the sources (other than random) i.e. non sampling errors like non
response of the respondents, specification errors, processing errors and etc. These two
sources can affect both estimate and its mean square errors.

Types of Errors

Random Sampling Errors Non- Sampling Errors

Errors of Coverage Non-response Observational Processing

Specification errors errors errors errors

5.7.1 Types of Sampling

Types of Sampling

Probability Sampling Non- Probability Sampling

Simple Stratified Cluster Systematic Convenience

Random Random

Quota Purposeful Judgemental Accidental Incidental Multi Snow

stage Ball

A. Probability Sampling-
It is a kind of sampling process where the entire samples considered are calculable
and non-zero probability is considered in the sample then such sample is termed as
probability sampling. The methods for selection of sample from the population may
not be similar for all the cases. A probability sample is highly desirable in the case

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 20

 BRM Module-3 2022

when the research is made for a conclusion or prediction that affects the whole
population.

1. Simple Random Sampling

It is a sampling process where all the members in the population have a probability of
considering as a sample. For instance, if a wide range of sample is to be considered
from a city consisting of 1, 10,000 people then each people that are selected will be a
target sample. Such people will be assigned with a number out of which a sample will
be prepared randomly out of it. In small populations, random sampling is done
without the instance of a unit being sampled more than once.

This method is appropriate when the population size is small, frame of sampling is
clear, homogenous and when the details of the population are not known. The main
advantage here is that the population has a minimum knowledge and there is a free of
classification. There are other benefits that there is no human bias or any kind of
dependency on the availability of element. But its application is bit complex so it has
gained less popularity among the researcher as it fails to list every item in the
population prior to sampling.

2. Stratified Random Sampling

It is a type of sampling where the population is initially divided into a number of
strata. The design of strata depends on single criterion (e.g. Income of a family,
yielding a number of strata related to different places of educational attainment) or on
a combination or more criteria (e.g., gender & age). In this sampling, a random
sampling is considered from each of the strata and such sub-samples are brought
together to form the total sample.

Two forms of stratified sampling are as follows:

Proportionate Stratified sampling Disproportionate stratified

sampling
A method of considering the sample When the sample is prepared
proportion in respect to the irrespective to the proportion of
proportion of members available in member available within the
the universe universe.
For example- It is known that in the For example- It is known that in the
universe the population of male be universe the population of male be
65% and female is 35% then the 65% and female is 35% then the
sample is to be selected in the sample need not follow a similar
proportion of 65:35. proportion it may be 50:50, 70:30 or
40:60.

3. Cluster Sampling-
Here, the whole member of the population will be assigned to one group. This each
group is called as a cluster. A cluster sampling is selected by using probability

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 21

 BRM Module-3 2022

method. Then one individual among the cluster is considered for survey. This cluster
may be a town, village, geographical area, qualification or any kind of social units.
The sampling of clusters from the population may be done through stratified or simple
random sampling.

4. Systematic Sampling-
Here all the samples from the sample frame are laid down out of which the every k th
element from the sampling frame is to select. Here the k element states about the skip
interval or a systematic gap between the selected samples. This is selected by
implementing a formula:

Skip Interval (k) = Population size (PS)

Sample Size (SS)

This method is accepted as a substitute to the simple random sample, it is a way of

unit selection from the list by implementing skip interval (k) so that each k th element
on the list follow a random order starting from 1 to k within the sample.

For instance: if the k th element is found as 3 and the sample will start from 3 then the
sample would include:

3rd, 6th, 9th, 12th element from the sample frame and so on.

In case, the result of skip interval (k) is not a whole number then the nearest whole
number can be considered as skip interval unit. This type of sampling is mainly
popular in industrial sectors to identify the machines or equipment in the production
line are functioning properly. The manufacturer can check every k th item to verify the
quality is maintained as per the specification. But it has a limitation that the researcher
has to decide the skip interval for every sample frame separately. This method is more
suitable the researcher have less fund and time.

5. Convenience Sampling-
This is a method of selecting the samples as per the easy availability or accessibility
of the researcher. This type of sampling for data collection is easy and cost effective.
For example: a company is planning to manufacture sports shoes for which they
prefer to collect the data from their own employees this activity of data collection is
considered as a perfect example of convenient sampling. But this type of survey has a
limitation that the result of it cannot be made generalized to the whole population as
the sample selection are not the direct representative of target population. Adding to it
there is a chance of personal bias as there is not proper technique for selecting the
sample. As in the example it can be said that the employees may support the
manufacturing of the sports shoes as this will help them to continue their job for a
longer period of time.

B. Non- Probability Sampling

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 22
 BRM Module-3 2022

It is a process of selecting the units as per certain factors other than random. This
sampling is completely purposive in nature. In general, all the process of selection of
sample that does not come under probability sampling will be under non-probability
sampling.

1. Quota Sampling-
The whole sample is divided into two groups of mutually exclusive. The number
of respondents (quota) that are selected from the several pre-defined categories the
final selection will be carried on by the interviewer who will continue the process
till the quota response is complete.
This type of sampling is widely used in commercial based research where the
purpose was to select the representatives’ who are from different categories of the
population. For instance the survey is carried and the category is gender then the
respondents will be only two i.e. male and female. If the researcher wants a data
of 100 people then it should be made 50 from male and remaining 50 as female.

Quota sampling has a chance of interviewer bias that may affect the result in:
a) Respondents (quota) reflecting as superficial characteristics.
b) Sampling was conducted as per the availability not as per the suitability.

Quota sampling has three steps:

I. Classification of population in respect to properties known or expected to
be relevant to the characteristics being studied.
II. The amount of the population falling into each group is confirmed on the
basis of the known, assumed or pre-defined arrangement of the population.
III. Lastly, each observes or interviewer is assigned a quota of respondents.
Then this people will take the charge for final selection.

2. Judgemental Sampling-

When a selection of sampling is carried on the judgement of an experienced

researcher then such is known as Judgemental or purposive sampling. The
selection of sample units is based on the population’s parameter. In this kind of
research a researcher keeps on changing the sample frame and categories as per
the objective of research. The researcher is well aware about the responses from
the respondents which might not solve the purpose of research. For instance, a
researcher wants to make a survey on the preference of selecting a deodorant. The
sample will be collected form a shopping mall where researcher can find a wide
range of respondents of different section of societies.

3. Accidental-
Here the researcher selects a sample without any pre-defined process. It simply
picks up those respondents that fall to hand. By this the researcher continues till it
finds a desired level of sample size and responses.

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 23

 BRM Module-3 2022

For instance, a researcher collects information from 100 persons to whom he meet
on the footpath that are willing to respond. It has certain limitations that the
respondent may not be aware not educated about the purpose of research. They
might not be the target population for survey. But it is an easy, time effective and
cost effective process.

4. Multistage Sampling-

In some of the research activity it is preferable to follow several phrases of data

collection. This is done after making several fraction of the sample while other
sample activities allows collection of information from whole sample. This
sampling technique allows stratification of sub-sample so the collection of
information from the first stage sample can be collected from the following sub-
sample. It means the researcher need not to follow a systematic order for data
collection they can collect information from sub sample at any point of time.

5. Snow Ball-
It is a process of selecting additional respondents for research activities. This is
useful for those populations that are rare and have fewer incidences. The process
of sampling always face several problems, as it is difficult to identify the sample
from the huge population so the better option is following a chain system of
referrals. For example an International MBA institute makes an open survey on its
website regarding the interest of selecting an MBA institute for admission.
This is very difficult to collect the response like such so the institute consider the
first 100 visitors of the website and request them to provide some names those
who are looking for taking admission in MBA college. The will be requested
mention at least three names. After the collection of name the institute will try to
interact with them personally and collect the information in details.
This is a method that reduces the costly affairs and saves time but the
disadvantage of this method is biasness. The process of collecting names is the
initial stage the process may continue to follow as much as it is possible for the
process of research. However, this sampling technique may not represent the cross
section of the total population. It may happens that visitor to the site might not
aware of any of the sample like or reluctant to share the names and details.

Dr. Arya Kumar, [Link]@[Link], 09853422575 Page 24