You are on page 1of 25

BIET MBA Programme :Compiled by BVN Page 1

Quantitative Methods 1
I Semester MBA : 2012-14

Module 1

Business Research

Introduction

Organizations across the globe prosper due to their constant efforts towards solving the problems being
faced in managing their business exposed to many external forces which are not under direct control. But
still growing organizations chose the path of continual improvement and constantly upgrade themselves
from one layer to the other towards success. In doing so, organizations constantly search for the scope to
grow along with the problems in the path of growth. It is this situation organizations sit aside and chalk
out a plan to take on solving their business problems with focused approach which paves a way for
business research and its methods.

Business Research is defined as the systematic and objective process of gathering, recording and
analyzing data to make business decisions.

The objective of Business research is to facilitate the managerial decision making process for all aspects
of a business. It generates and provides the necessary qualitative and quantitative information upon which
the decisions are based. It reduces the uncertainty of the decisions thereby reducing the risk of making
wrong decisions.

Basic & Applied Research

Research is broadly classified into two categories viz., Basic Research and Applied Research

Basic Research attempts to expand the limits of knowledge. It does directly involve the solution to a
particular business problem.

Examples:

Is executive success correlated with high need for achievement?
Are members of highly cohesive work groups more satisfied than members of less cohesive work
groups?
Do consumers experience cognitive dissonance in low-involvement situations?

Applied research is conducted when a decision must be made about a specific real-life problem or to
make decisions about a particular course of action or policy.

Examples:

Should McDonalds add Italian pasta dinners to its menu?
What should be the suitable location & price to launch new product?
Are the customers satisfied with the existing after sales service of Aqua guard from Eureka
Forbes?
What needs to be done in order to retain good engineers at Infosys?

BIET MBA Programme :Compiled by BVN Page 2

Business research is an applied research to find out the solution for business and management problems
faced by organizations

Determinants of Business research

Time constraint
Availability of data
Nature of decision
Benefits Vs Costs


Types of Research

Decision making is the process of resolving a problem or choosing among the alternative opportunities.
Depending on the decision making situation and the information availability in the continuum of
certainty, uncertainty and ambiguity, the business research is classified into three types.





When the researcher certainly has all information about business problem, it may lead him to choose
descriptive or causal research
Descriptive research seeks to determine the characteristics of the population and answers to who,
what, when, where and how questions.
It helps segment and target markets.
Descriptive studies are based on some previous understanding of the nature of the research
problem.
For example if researcher wants to study the market penetration of a product, he will try to see the
variables such as sales volume, price competitiveness and other characteristics of a product like
quality, durability, customer satisfaction etc.,.

Researcher resort to causal research when the situation under research is certain and he knows about some
sort of inter-dependency of the variables under study.

Causal research is conducted when there is a cause and effect relationship between the variable
sunder study.
Using causal research, one can establish a relationship or dependency of one variable on the
other.
For example, if HR manager wants to see the change in productivity as an effect of providing
training to his employees, training is a cause for change in productivity.

Under the cases of ambiguity, where the nature of the problem to be solved is unclear, objectives vague
and the alternatives are difficult to define, exploratory research is used.

Exploratory research is conducted as a preliminary exercise to analyze the ambiguous situation
and to find further direction to the research.
Exploratory research does not provide conclusive evidence to the research.
For example if the company wish to launch a product in a new market, it has to study through
exploratory research, the buying behavior of the people in that location.
Completely
Certain
Uncer
tainty
Abso
lute
Amb
iguit
y
Descriptive or
Causal Research
Exploratory
Research
BIET MBA Programme :Compiled by BVN Page 3



Research Process

The stages of the research process are.,

1. Problem definition/formulation: Allows the researcher to set the proper research objectives. If
the purpose of the research is clear, the chances of collecting the necessary and relevant
information will be high. To be efficient, business research must have clear objectives and
effective designs. Exploratory research, Experience survey, Secondary information and literature
review, pilot studies in the form of focus group interview are some techniques used by the
researcher in precisely defining the research problem with specific objectives.

2. Planning a research design: It is a master plan specifying the methods and procedures for
collecting and analyzing the needed information. It also specifies the sources of information, the
research method, the sampling methodology, the schedule and cost of the research. Surveys,
experiments, secondary data reviews and observation are the basic tools which help the researcher
in selecting appropriate design for the research.

3. Selection of sampling design: It should clearly specify the sampling units of the population,
sample size and the most appropriate sampling procedure.

4. Data collection: Once the research design has been finalized, the process of collecting data for
the research begins with pre-determined tools like interviews, questionnaires, schedules etc.,
ensuring minimal sampling error.

5. Data analysis: Once the data is collected, classification, tabulation, editing and coding of the data
is done as per the requirement for further analysis. Analysis of the data is determined by
appropriate technique as required by the researcher.

6. Conclusion and reporting: The final stage of research is to interpret the information and draw
conclusions relevant to managerial decisions. The research report should communicate research
findings effectively keeping in mind the managerial audience.


Problem Definition / Formulation

Defining a research problem involves the following steps.
1. Understand the objectives of the researcher: Research problem should be so framed that it
comprehensively takes care of the existing business problem and finds specific solutions through
focused objectives.
Iceberg Principle : It is defining the research problem without understanding its depth. Such
business problems are like the submerged portion of the iceberg in a sea, which is neither visible
to nor understood by sailor and hence decisions based on such research are dangerous.
Often exploratory research helps researchers to clarify their objectives and decision.

2. Understand the background of the problem : One should always have fair knowledge about the
situation in which the research is being carried out. This can be done by talking to experienced
people in the area of research, going through the past literature which give some clue as to what
needs to be addressed in the research.

BIET MBA Programme :Compiled by BVN Page 4

3. Identify the problem rather than its symptoms: Narrow down the process of research through
specific objectives.
4. Determine the unit of analysis: The researcher must specify whether the level of investigation will
focus on data about the entire organization, departments, work groups, individuals or objects.

5. Determine the relevant variables: Researcher should know the possible set of variables required
to be used in research. Later these variables play very important role in data analyses and
conclusion. For eg., if the researcher wants to study the market share of a product, he should
focus on the total sales of his product, competitors product, demand, supply, benefits to the
retailer, price, quality etc.,.

6. State research questions and research objectives: At the end of the problem definition stage of the
research process, researchers should prepare a written statement that clarifies any ambiguity
regarding research outcomes.

Management Problem Vs. Research Problem

Management problems

Those problems which need solutions leading to tangible / intangible benefits to the organization.
Continuously effect the organization
Focus on process up-gradation or policy changes directed towards better efficiency
Specific in nature

Research Problems
Those problems which are derived out of thorough analysis of the earlier efforts which seek
solution in the current situation
Lead to focus on the reasons which influence business problem
Can be addressed through planned academic approach in deriving the solutions
Out-comes of research problem may lead to direct or indirect solution to the existing business
problem

Importance of Literature Review in Problem definition

Literature review (LR) is the secondary information review regarding the present research under
discussion during the initial phase of problem definition and setting up of its objectives.

It gathers information regarding the earlier research done on the issues of present research.
It helps in understanding the past approaches towards problem, methodology adopted, variables
used, key issues already addressed etc.,. which further helps in setting up new objectives and
approach for the present research.
It gives insight to the researcher in finding out research gap and the areas of further research in
the related topic.
The comprehensive LR about the related topic gives direction to the researcher in making his
present research more interesting and novel.
LR avoids taking up the research of known things through earlier research thereby ensuring
newness in the research.



BIET MBA Programme :Compiled by BVN Page 5



Business Research Design: Steps involved in a research design.

Research design is the master plan specifying the methods and procedures for collecting and analyzing
the needed information. A well charted research design makes the research process more clear and self
directing.

Following are the steps involved in setting up research design..

Statement of objectives of research
Specifying the sources of information
Indicating the appropriate research method (i.e., Surveys, experiments, secondary data review or
observation)
Specifying the sampling methodology, schedule and cost of research


Experimental Research

Type of research where the researcher know about the behavior of one variable by having control over the
other variables. For example, if the investigator wants to know most efficient complementary offer which
ensure better sales, he has to go on testing the customer response for different offers such as discount, gift
hampers, gift coupons, customer cards etc.,.

Experimental designs

These are the tested models / frameworks which help researcher to carry out experimental research,
allowing investigator to control or manipulate some independent variables and study its effect on some
dependent variable so that causal relationship between variables may be evaluated.

Basic issues in experimental design

1. Manipulation of independent variable
2. Selection and measurement of the dependent variable
3. Selection and assignment of test units
4. Control over extraneous variables

Evaluation / Validity of experimental designs

Experimental design is said to be valid if it satisfies the conditions of internal and external validity.

In an experimental research, if observed change in the dependent variable is the sole result of the
independent variables being manipulated, then the experimental design used in such experiment is said to
possess internal validity.

If research findings through an experimental design can be generalized/applicable to the external
environment, such experimental design is said to possess external validity.



BIET MBA Programme :Compiled by BVN Page 6



Extraneous variables affecting internal validity of experimental design.

The variables which are not considered under the experimental design and are likely to influence the final
outcome of the research without the knowledge of researcher are called extraneous variables. Following
are six major types of extraneous variables which affect internal validity of experimental design.

History effect : refers to the effect of specific event on respondents which may alter their behavior or
response to the research tool / questionnaire. Further cohort effect is a special case of history effect where
responses collected from two group differ because of their exposure to different environments

Maturation effect : is an effect on the results of an experiment caused by changes in the responses of the
respondents due to the changes in circumstances, taste, behavior, attitude as they get matured in their
thinking during the process of data collection

Pre-testing effect : is the change in the validity of an experiment that occurs when initial testing gives
clue about nature of the experiment which in turn may influence the response of the respondent when real
data is collected.

I nstrumentation effect : It refers to the effect caused by changing the wordings of the questionnaire,
change in interviewers, or a change in the other procedures to measure the dependent variable.
Selection: It is a sample bias resulting from differential selection of respondents for the comparison of
groups. It is also referred to as sample selection error.

Mortality effect : It occurs when respondents start retiring from the experiment due to nay reason caused
by the situation or self.

Classification of experimental design

Experimental design is said to be basic if only one variable is manipulated. Otherwise if researcher wishes
to investigate effect of more than one variable, the required experimental design is called complex or
statistical experimental design.

Quasi-experimental designs
(do not qualify as true experimental designs as they do not adequately account for loss of internal and
external validity)

1. One-shot design (After -Only Design)

Symbolically represented as X O
1


X stands for exposure of a group to an experimental treatment
O
1
Stands for one observation made after exposing the group to an experimental treatment

Eg. If one want to see the effect of price hike on sales for a product, he/she can test it by exposing
few customers to the new price and observing their buying behavior.

2. One group pretest-posttest design

Symbolically represented as O
1
X O
2
BIET MBA Programme :Compiled by BVN Page 7


Here two observations of an experimental group are made; one before and one after the
experiment

Eg. If HR manager wants to measure effectiveness of a training program on employee
productivity, he can do it by measuring productivity before and after exposing the employees to a
training program. (O
2
-O
1
) gives measure of effectiveness.


3. Static group design

Symbolically represented as Experimental Group : X O
1

Control Group : O
2


Here one group is exposed to a new treatment and compared against control group, which is not
exposed to any treatment. (O
2
-O
1
) gives the effectiveness of new treatment.

Eg. Suppose that an agricultural scientist wants to see the change in yield with application of new
fertilizer; he can observe the yield of a crop with (O
1
) and without (O
2
) application of new
fertilizer. (O
2
-O
1
) tells about the change in yield due to the application of new fertilizer.

Basic Experimental Designs

1. Pretest-Posttest control group design

Experimental Group : O
1
X O
2


Control Group : O
3
O
4

Stands for randomization of samples. The effect of experimental treatment is (O
2
-O
1
) - (O
4
-
O
3
).

2. Posttest only control group design

Experimental Group :

X O
1


Control Group : O
2


(O
2
-O
1
) measures treatment effect.

3. Solomon Four Group Design

Experimental Group 1 :

O
1
X O
2


Control Group 1 : O
3
O
4


Experimental Group 2 :



X O
5


Control Group 2 : O
6



Compromise designs
R

R

R R

R

R
R

R
R

R
R

R
R

R
R

BIET MBA Programme :Compiled by BVN Page 8


In many instances of business research, true experimentation is not possible., then what best the
researcher can do is approximate an experimental design. Such designs are called compromise designs.



Time series designs

When researcher wants to study a variable or phenomenon over a period of time before and after
application of an experimental treatment, it is called time series design.

O
1
O
2
O
3
X O
4
O
5
O
6


Eg. If fund manager of a financial services company wants to know the investment behavior of his
customers towards a particular mutual fund before and after declaration of dividend; he can do that by
observing the investment pattern at different time points O
1
, O
2
, O
3
before and at time points O
4
, O
5
, O
6
after dividend declaration.

Complex experimental designs

1. Completely Randomized Design (CRD)
In this experimental design, effect of one treatment (independent variable) at different levels on
the dependent variable is studied ensuring randomization of samples exposed to the treatment
under study.

Eg. Effect of application of new fertilizer at varied quantities on yield of the crop can be studied
using CRD through one-way ANOVA

2. Randomized Block Design (RBD)
In this experimental design, effect of two treatments (independent variables) at different levels on
the dependent variable is studied ensuring randomization of samples exposed to the treatment
under study.

Eg. Effect of application of new fertilizer at varied quantities and type of soil on yield of the crop
can be studied using RBD through two-way ANOVA

3. Latin Square Design (LSD)
It attempts to control or block out the effect of two or more confounding extraneous factors. It
manipulates one independent variable and controls for two additional sources of extraneous
variation by restricting randomization with respect to row and column effects. It assumes that
there is no interaction effect.

4. Factorial Designs
These are the most sophisticated experimental designs in which effect of two or more
independent variables at different levels on the dependent variable can be studied with a scope for
measuring interaction effect and confounding of extraneous variables.


BIET MBA Programme :Compiled by BVN Page 9

Module 2 : Sampling and Data collection

Introduction

Sampling design is the crucial part of any business research. It is this phase of research upon which the
validity of remaining stages rests. As the behavior of the population under research largely depends on the
quality of samples collected, it is very important to see that the samples are collected with utmost care so
that the error due to sampling is as least as possible. In this section, we learn about those various sampling
techniques suited to different situations which ensure collection of best possible samples out of a
population under study.

Sampling

It is a process of collecting samples from the given population using scientific approach, which ensure
minimum sampling error and better estimates of population characteristics.

Sampling distribution

Distribution of sample statistics is called sampling distribution.
(detailed explanation will be covered under Module 7}

Sampling Error

The difference between actual value of the population characteristic (eg. mean, s.d etc.,) when compared
to the estimated value based on sample observations.

This error is attributed to sampling fluctuations. This error can not be eliminated completely due to the
fact that population parameters are estimated based on only few sample observations. However it can be
minimized by applying most suitable sampling technique for data collection.

For example, if the actual value of the average salary of 2000 employees of an organization is Rs. 15000
and the estimated value of the average salary based on 100 sample observations is Rs. 14650, then the
difference 15000 - 14650 = 350 signifies sampling error.

Sampling error may creep in due to
Faulty selection of the sample (/sampling)
Substitution of convenient sample in place of actual sample
Faulty demarcation of sampling units
Error due to bias in the estimation method

Non Sampling Errors

Non sampling errors in data collection are not attributed to chance and are due to certain causes which can
be traced and may arise at any stage of the enquiry viz., planning, execution, collection, processing and
analysis.

Following are the common reasons for non sampling errors
Faulty planning, vague and imperfect questionnaire, defective methods of data collection
Lack of trained investigators, incomplete responses
Improper coverage, compiling errors and publication errors

BIET MBA Programme :Compiled by BVN Page 10


A good sampling design is the one in which sampling and non-sampling errors are minimized.


Census (Complete) Vs. Sample (Partial) Enumeration

Census Survey Sample Survey
1. Here the researcher resort to 100%
inspection of the population
2. Provide more accurate and exact
information as the data is collected from
each and every unit of the population.
3. Affords to more extensive and detail
study
4. Very expensive, time & manpower
consuming.
5. Take much time for data analysis if the
population is vast
6. Sampling errors are almost nil.
7. Non sampling errors can be controlled

1. It is inspection of representative part of the
population
2. Yields near to accurate results when
administered scientifically
3. Offer quick leads towards the required area
of research

4. Cost effective and less time consuming
5. Limited manpower can administer sample
survey

6. Sampling errors are always present.
7. Non sampling errors can be controlled



Types of sampling

The choice of an appropriate sampling design is of paramount importance in the execution of a sample
survey and is generally made keeping in view the objectives and scope of the enquiry and the type of the
population to be sampled. The sampling techniques may be broadly classified as

1. Non-Probability sampling (purposive/convenience/ judgment sampling)
2. Probability sampling
3. Mixed sampling

Non-probability sampling : In this method, a desired number of sample units are selected purposely
depending upon the object of the enquiry so that only the important items representing the true
characteristics of the population are included in the sample.

For example, if one wishes to know through sample survey, the customer satisfaction towards Eureka
Forbes products, he has to collect the needed information only from Eureka Forbes customers.

Non-probability sampling is very subjective in nature and depends entirely on the convenience of the
investigator. Hence there is a larger scope of bias. Convenience sampling is appreciated when research is
limited to small sample size or in the cases where finding sample units is very difficult and /or expensive.

Probability sampling

Probability sampling provides a scientific technique of drawing samples from the population according to
the laws of probability. The probability sampling scheme will have following characteristics.
(i) Each sample unit has an equal chance of being selected
(ii) Sampling units have varying probability of being selected
(iii) Probability of selection of a unit is proportional to the sample size
BIET MBA Programme :Compiled by BVN Page 11


Mixed sampling

Sampling design in which the sample units are selected partly according to some probability laws, and
partly according to a fixed sampling rule (non-probability) is known as mixed sampling.

Some of the mixed sampling schemes are

1. Simple random sampling
2. Systematic sampling
3. Stratified sampling
4. Cluster sampling
5. Multistage sampling
6. Quota sampling

Simple random sampling (SRS) : It is a technique in which sample is so drawn that each and every unit
in the population has equal chance of being included in the sample.

If the unit selected in first draw is not replaced back into the population before drawing the next sample, it
is called simple random sampling with replacement (srswor) and if it is replaced back before making the
next draw, then the sampling plan is called simple random sampling with replacement (srswr).

Selection of simple random sample can be done by (i) lottery method (ii) use of random number tables.

SRS eliminates bias due to personal judgment. Samples selected using SRS will be more representative
than that of judgment sampling. Samples using SRS will be most reliable at least cost, less time and labor.
However if the population is vast, fairly large sample size is required to ensure better estimates of
population characteristics.


Systematic Sampling : Here the population units are numbered in serial order. Samples are collected by
following an order such that first sample is randomly selected by randomly assuming a number between 1
to 9 and there afterwards required number of samples are selected in a definite sequence at equal spacing
from one another.

For example, if 10 sample bags are to be selected out of 100 bags of Basmati rice for quality inspection;
all 100 bags are numbered serially from 1 to 100. Then first sample is selected as random number
between 1 to 9 say 4. Then bag nos. 4,14,24,34,44,54,64,74,84,94 are the units to be sampled using
systematic sampling technique.

Stratified Sampling : When population is heterogeneous with respect to the variable or characteristic
under study, then the technique of stratified sampling random sampling is used to obtain more efficient
results. It involves dividing the population into groups (strata) having homogeneous characteristics. Then
random samples are drawn from each stratum according to size of each stratum.

For example, if we are interested in understanding the shopping behavior of people of different age
groups at Big Bazaar, we can stratify the customers according to their age groups and then pick
proportionate number of respondents belonging to different age groups randomly according to the size of
each stratum to study their shopping behavior.

BIET MBA Programme :Compiled by BVN Page 12

Cluster Sampling : In this case, total population is divided, depending on problem under study, into some
recognizable sub divisions called clusters and simple random samples of these clusters are drawn.

For example, if one wants to conduct opinion survey of people in an election constituency, the whole city
may be divided into different blocks i.e., clusters of manageable size and then households from each of
these clusters are selected randomly.

Multistage Sampling : In multistage sampling, instead of selecting all samples from different clusters,
clusters are further subdivided to get better and efficient samples.

For example, if we are interested in obtaining a sample of say n households from a particular state, the
first stage units may be districts, the second stage units may be villages in the districts and third stage
units will be households in the villages.


Quota Sampling : Quota sampling is similar to that of stratified sampling with pre-defined sample size
(quota) in each stratum. Investigator has to collect designated number of samples from each stratum.
However investigator is given a liberty to choose sample units on his discretion if some sample units are
missing from the stratum at the time of data collection.

Qualitative techniques of data collection.

When the data is qualitative in nature, the usual methods of data collection viz., through observation,
interview, questionnaire and schedule can not be used for data collection. In such situations, data may be
collected by:

(i) Structured and direct form of interview
(ii) Unstructured and direct form of interview
(iii) Structured and indirect mode of interview
(iv) Unstructured and indirect way of interview

Structured and direct form of interview : Here formal questionnaire consists of structured and direct
form of questions and the respondents are asked to respond to definite questions through direct interview
method.
Unstructured and direct form of interview : In this method, interviewer is given general instructions on
the type of information desired. Interviewer is left to ask direct questions to obtain this information using
the wording and the order that seems more appropriate in the context of each interview.

Depth interview and Delphi technique are the popular methods of data collection through unstructured
and direct form of interview.

In Depth interview, interviewer will continue to ask probing questions to the respondent until the required
information is gathered to his satisfaction. It requires lengthy duration to collect information and depends
on the ability and intelligence of the interviewer

In Delphi technique, discussions about a particular issue or business problem is conducted among the
group of experts from respective fields and the necessary information is gathered to sort out the problem
or to make better decision using the inputs given by the experts panel on the issue.

I ndirect interviews : Under this method, the respondent is given non-personal, ambiguous situation and
asked to describe it. The respondent will tend to interpret the situation in terms of his own needs, motives
BIET MBA Programme :Compiled by BVN Page 13

and values. Hence these techniques are also known as Attitude Measurement Techniques. These
include
1. Projective Techniques
a. Association techniques : Here respondents are asked to associate themselves with the
situation and asked to give their inputs. Eg. Group discussion
b. Construction techniques : Respondents are given a task of constructing a story or
strategy based on the explained situation with constraints.
Eg. If you are selecting a marketing executive for your company, he may be asked to design a
strategy for the imaginative problem being faced by the company.
c. Completion techniques : Here respondents are exposed to some incomplete situations
and are asked to respond with their skills and attitude towards meaning solution.
Eg. Students are asked to solve the business cases to assess whether they can apply the theoretical
concepts for solving real life problems.
d. Choice of ordering techniques : Here the respondents are asked to place some objects or
constructs in their order of choice, which in turn helps in understanding their attitude
towards a particular goal.
Eg. If you are interviewing a person for HR executive post and wish to assess the value system of
the candidate: You can ask him to place the values like honesty, sincerity, commitment, loyalty,
integrity, timeliness in his order of preference.
e. Expressive techniques : Here, respondents are exposed a situation and asked to respond
to it through their expression. Eg. Play techniques, role playing etc.,.
2. Thematic Appreciation Test
In this method, respondents are exposed to a theme in the form of a story or picture; and asked to
review and respond to the story or situation depicted in the pucture.
3. Focus Group Interview
Here a group of people jintly participate in an unstructured interview conducted by moderator.
Moderator attempts to focus the discussion on the problem areas in a relaxed, non-directed
manner. The interaction among the grop members during the interview will possibly lead to
spontaneous discussions and the disclosure of attitudes, opinions and information.
Scaling techniques like Likert Scale and Semantic Differential Scale are used to measure attitude.


BIET MBA Programme :Compiled by BVN Page 14

Module 3

Measurement and scaling techniques

Scaling Techniques

These are used to measure different psychological aspects such as attitudes, perceptions and preference of
people with the help of certain predefined set of stimuli or instructions. The stimuli may be certain brands
of product, alternative advertising themes, package designs and so on. The response may involve how
consumers judge the brands, which brand looks more pleasing for them, which package is more attractive
etc.,.

Scale of Measurement

Scaling procedures can be classified as nominal, ordinal, interval and ratio type.

Nominal Scale: Here symbols/numbers are assigned to objects just for the sake of identification.

Eg.1. Roll numbers are assigned to students just for the sake of identification. These numbers do not
reveal strength, rank or superiority of the student./goodness of the student behavior.
2. Numbers are assigned to buses for identification of routes only.
3. Question regarding profession with options like Worker, Businessman, Student, Govt. employee,
Pvt. Employee, Former may be assigned numbers 1,2,3,4,5,6 respectively.

Ordinal Scale : In this type of measurement, numbers are assigned to objects to indicate some
orderliness based on relative importance of the response to certain question.

Eg.1. Rank list of students show some order in which the number 1 stands for the student who has topped
in the examination results, number 2 stands for the student who has scored II highest marks in the exam
and so on. i.e., in a rank list, the numbers assigned to the student signify students ability/strength in
passing the exam.
2. The response to a question, Name the three companies that you would like to join in the order of
your preference. Will generate answers in the form of three Cos in which the first co. has highest
importance, second co. will have importance lesser than that of the first and so on.

Interval Scale : This type of measurement involves comparison between different categories of response.
In marketing research, interval scale is often chosen to measure attitude towards brand, culture,
management practice etc.,.

Eg.1. Data regarding Income of people captured on the scale of 10000-15000, 15000-20000 etc.,

Ratio Scale: This measurement consider absolute magnitude of numbers. Typical issues like quantity
sold, number of consumers, probability of purchase are the examples of data recorded on ratio scale.

Methods of scale construction

There are number of ways in which scales may be constructed. They are,
Paired comparison method
Ranking
Ordered category sorting
BIET MBA Programme :Compiled by BVN Page 15

Rating Techniques

In paired comparison method, respondent has to respond by
comparing two attributes for their desired characteristic.

Eg. If six tyre brands are to be compared on road grip quality
by comparing two at a time, the following tables give paired
comparison responses.

1 implies that row brand is better than column brand
0 implies that column brand is better than row brand
X implies invalid comparison


In Ranking method, respondent is given an option to rank the objects according to the desired
characteristic of the object.

Eg. For the above example, if the respondent is asked to rank the tyre brands according to their road grip
quality, the resulting response in terms of brands being ranked as 1,2,3.represent ranking method.
MRF(1), Ceat(2), Dunlop(3) etc.,.

In Ordered category sorting method, respondent will be instructed to give his response by placing the
objects in desired categories as per his experience and perception.

Eg. For the same example of tyres, respondent may be asked to categorize the tyre brands as Top class,
Average class and Poor class

Rating Techniques are the most widely used scaling techniques in marketing research. They are most
popular as they allow easy flow of data collection. In all rating scale researchers devise a number of
statements that relate to the product, attribute or service quality. Respondents have to indicate their
response with the help of numerical, graphical or verbal scale values. Likert scale and Semantic
Differential Scales are the popular rating techniques used in all social / marketing research.

Likert scale: Here, respondents are asked to indicate on a five point numerical scale their degree of
agreement of disagreement with some pre designed set of statements. The researcher is free to choose his
own way to label the agree/disagree response. Care has to be exercised in labeling the response categories
for favorable vis--vis unfavorable statements. That is obviously for unfavorable statements direction of
response scores will be opposite that of a favorable statements.

Eg. Give your opinion about voting rights to the shareholders in choosing Managing Director of the PSU.
on the five point scale as Highly Important(5), Important(4) Neutral(3), Not important(2), Not at all
important(1).

Semantic Differential Scale (SDS) : It is also one of the mostly used rating techniques in corporate
product and brand image studies. In SDS, the respondents are asked to indicate their choice among a set
of bi-polar phrases for adjectives that best describe their feelings towards an issue or object. A set of
statements are constructed and the responses consisting set of opposite meaning phrases at the two
extreme ends of the scale. Semantic scale assumes that the centre position of the response is somewhat
neutral and the intermediate positions are defined by two extreme poles.

Brand A B C D E F
A X 0 1 1 1 1
B 1 X 1 1 1 1
C 0 0 X 0 0 0
D 0 0 1 X 0 0
E 0 0 1 1 X 1
F 0 0 1 1 0 X
BIET MBA Programme :Compiled by BVN Page 16

Eg. In a study of a brand of two cough drops viz., Vicks and Halls, the following form of semantic
differential scale was developed.

Cooling effect : Weak - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -Strong
Strength of flavor : Mild - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Strong
Color : Not pleasing - - - - - - - - - - - - - - - - - - - - - - - - - - Pleasing
Clears Blocked nose : Ineffective - - - - - - - - - - - - - - - - - - - - - - - - - - - Effective
Wrapper : Bad - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Good
After effect : Short Lived - - - - - - - - - - - - - - - - - - - - - - - - - - - Long Lasting

Source: Marketing Research Authored by Ramanuj Majumdar, New Age International Publishers, 1e, 2005.




BIET MBA Programme :Compiled by BVN Page 17

Module 6

Probability & Theoretical Probability Distributions (Covered for B & C Sections)

Probability in statistics is a measure of chance associated with activities which are random in nature. The
following few examples make us understand the necessity of probability in our day to day life / making
better business decisions.

Eg. It may rain today ; production of wheat this year will be high ; we may experience steep drop in
recruitments due to US recession ; Capital requirement to start a new business will be roughly more than
Rs. 2 bn in the next year.

In all the above examples, no information or expression is certain to happen but it is an estimate based on
the chance factor associated with respective situations. If we could explore and quantify this chance
factor, it is remarkable in making better decisions.

In this module, we try to learn about basic concept of probability and probability distributions which help
us to apply probability rules in solving business problems for better decision making under uncertainty. It
is for this reason, probability finds its wide space in statistics, serving various needs in the fields of
science, social science as well as business.

In order to understand probability concepts , knowledge of following terminology is essential.

Random Experiment (r.e.) : An experiment resulting in random outcomes
Eg., Tossing a coin results in two random outcomes Head & Tail

Trail : Each repetition of a random experiment is called trial
Eg., Tossing coin for two times means 2 trials of tossing a coin

Event : It is an outcome of a random experiment
Eg., In tossing a coin, H & T are the two events

Sample Space (S) : Set of all outcomes of a random experiment
Eg., When a coin is tossed, S ={H,T}; When two coins are tossed, S ={TT, HT, TH, HH}

Mutually exclusive events : Two events are said to be mutually exclusive when happening of one event
prevents the happening of the other.

Eg., When a coin is tossed, H & T are called mutually exclusive events; i.e., if Head turns up, it is to
understand that head is preventing Tail to turn up or vice-versa.

Equally Likely events : Two events are said to be equally likely, when chance of occurrence of those
events are same.

Eg., : In tossing a coin, chance of getting Head = = chance of getting tail

Exhaustive events : Two events are said to be exhaustive, when total chance of their occurrence is 1.
Eg. When a coin is tossed, total chance of (occurrence of head + occurrence of tail) = + =1



BIET MBA Programme :Compiled by BVN Page 18

Mathematical Definition of Probability
Let there be a random experiment with n mutually exclusive, equally likely and exhaustive events. Let
m of these outcome be favorable for happening of event A. Then probability of event A is given by,
P(A) = (No. of favorable outcomes of A) / (Total No. of outcomes)
= m/n
For example, if we wish to find probability of getting Head out of tossing a coin, P(H) =
Similarly, in rolling a dice experiment, if we wish to get only even nos. P(getting even nos.) = 3/6
( as there are only 3 even nos. out of 6 numbered faces of a dice. i.e., 1,2,3,4,5,6)

BIET MBA Programme :Compiled by BVN Page 19

Numerical Examples on Z Test (Solved in all 3 sections)


(Equality of mean in single population)

1. A random sample of 100 students have mean weight of 58 kg and s.d. 4 kg. Test the hypothesis
that the mean weight in the population is 60 kg at 5% and 1% level of significance (l.o.s.).

2. Daily sales figures of 40 shop keepers showed that their average sales and s.d. were Rs. 528 and
Rs. 60 respectively. Can you conclude that daily average sales is Rs. 400 at 5% l.o.s.

3. An educator claims that the average IQ of American college students is at most 110, and that in a
study made to test his claim, 150 students, selected at random had an average IQ of 111.2 with a
standard deviation of 7.2. Use l.o.s. of 0.01 to test the claim of the educator.

(Equality of two means in two populations)

4. The average number of articles produced by two machines per day are 200 and 210 with standard
deviations 20 and 15 respectively. On the basis of records of 25 day production, can we regard
both machines equally efficient?

5. A company claims that its light bulbs are superior to those of a competitor on the basis of a study
which showed that a sample of 40 of its bulbs had an average life time of 628 hours with s.d. of
27 hours, while a sample of 30 bulbs made by the competitor had an average life time of 619
hours with s.d. 25 hours. Check at 5% l.o.s. whether this claim is justified?

6. The mean scores of two samples of 1000 and 2000 students are 67.5 and 68 respectively. Can the
samples be regarded as drawn from the same population of standard deviation 2.5 inches?

(Equality of proportion in single population)

7. A candidate at an election claims 90% of support of all voters in a locality. Verify his claim, if a
random sample of 400 voters from the locality, 320 supported his candidature. Use 5% l.o.s.

8. In a city, 325 men out of 600 men were found to be smokers. Does this information support the
conclusion that the majority of men in this city are smokers?

(Equality of proportions in two population)

9. In a year there are 956 births in town A, of which 52.5 % were males, while in town B 450 births
result in 43.4% males. Is there any significant difference in the proportion of male births in two
towns?

10. In a days production, 16 out of a random sample of 500 bolts were found to be defective. After
fine tuning the machine, another sample of 100 bolts contained 2 defective bolts. Is there any
significant difference in the proportion of defectives produced? 5% l.o.s..
BIET MBA Programme :Compiled by BVN Page 20

t-test

Applicable when sample size is small (i.e., n<30)
Assumptions of t-test
1. The parent population from which the samples are drawn is normally distributed
2. The samples are random and independent of each other
Application
1. To test the equality of single mean Similar to that of Z test, but for small samples.
2. To test the equality of two means
3. To test the equality of paired observations ( paired t-test)

Numerical Examples (Solved in class)
1. A machine is designed to produce insulating washers for electrical devices of average thickness
of 0.025 cm. A random sample of 10 washers was found to have an average thickness of 0.024
cm with standard deviation of 0.002 cm. Test the significance of deviation in thickness.
(t
5%, 9 d.f.
=2.262)

2. The mean weekly sales of the chocolate bar in a candy stores was 146.3 bars per store. After an
advertising campaign the mean weekly sales in 22 stores for a typical week increased to 153.7
and showed a s.d. of 17.2. Was the advertising campaign successful?

3. The mean life of a sample of 10 electric bulbs was found to be 1456 hours with s.d. 423 hrs. A
second sample of 17 bulbs chosen from a different batch showed a mean life of 1280 hours wth
s.d. of 398 hours. Is there a significant difference between the mean lives of the two batches of
bulbs?

4. A group of seven week old chickens fed on a high protein diet weigh 14 ounces and second group
of five chickens fed on low protein diet weigh 11 ounces. Test whether there is a significant
evidence that additional protein has increased the weight of the chickens?

5. CET test paper was given to 10 students before and after coaching and the scores were recorded
as follows. Test whether the coaching is effective.

Before Coaching 35 39 47 53 27 19 36 46 8 17
After Coaching 41 37 45 56 31 21 47 41 5 12


BIET MBA Programme :Compiled by BVN Page 21

Chi-square Test
(
2
- Test)

Application
1. To test goodness of fit
2. To test independence of attributes
3. To test the equality of variance These applications out of syllabus
4. To test the equality of several proportions


Conditions for the validity of
2
- Test
1. The total frequency N should be large (> 50)
2. The sample observations should be independent
3. Theoretical frequencies / cell frequencies should be more than 5 (otherwise pooling to be done).


F- Test
Application
1. To test equality of several population means using ANOVA
2. To test for equality of population variances
3. To test the significance of observed sample multiple correlation These applications out of syllabus
4. To test the significance of observed sample correlation ratio

Assumptions for F-Test
1. Samples are drawn randomly
2. Samples are independent of each other
3. Samples are drawn from normal population



BIET MBA Programme :Compiled by BVN Page 22

Testing of Hypothesis
(Summary of important terms)
Parameter & Statistic
The value (mean, s.d. etc.,) calculated based on whole population is called parameter.
To estimate value of the parameter using sample observations is called statistic.
Sampling distribution
The distribution of sample statistics is called sampling distribution.
Hypothesis is an assumption or statement made about population value /characteristic, that can be verified based on
sample observations.
Eg. Average score of SSLC students of year 2012 in Udupi district is 65% is a statement made regarding all
SSLC passed students of 2012; it can be verified for its correctness by considering sample scores of students from
Udupi district.
Null Hypothesis is the hypothesis which is stated for its possible rejection.
Alternative Hypothesis is the complementary statement to the null hypothesis, which is accepted when null
hypothesis is rejected.
Errors in testing of hypothesis
In testing of hypothesis using sample data, an experimenter may commit two types of errors viz., Type 1 error, Type
2 error.
Type 1 error is the error committed by rejecting the null
hypothesis even when it is true.
Type 2 error is the error committed by accepting the null
hypothesis when it is false.
Level of Significance (l.o.s)
The probability of committing type 1 error is called level of significance, it is denoted by ; it is usually fixed at 5%
or 1%.
Power of the test is the ability of the test procedure to give correct decision = 1-Prob(committing type 2 error)
= 1-
A test with alternative hypothesis of the type H
1
:=
0
is called two tailed test

A test with alternative hypothesis of the type H
1
:<
0
or H
1
:>
0
is called single tailed test

The set containing sample statistics of a sampling distribution which favor acceptance of null hypothesis is called
acceptance region and set of those statistics which favor for rejection of null hypothesis is called critical region.

The value of the test statistic which separates critical region and acceptance region is called critical / significant
value of the test statistic.

Accept H
0
Reject H
0
H
0
is true Correct Decision Type 1 Error
H
0
is false Type 2 Error Correct Decision
BIET MBA Programme :Compiled by BVN Page 23

Steps in the Procedure of Testing Hypothesis

1. State the Null and alternative hypothesis
2. Choose desired level of significance
3. Choose the most suitable test statistic (i.e., Z, t,
2
or F)
4. Obtain the sample statistics and other information about known population parameters
5. Calculate the value of the test statistic
6. Decide about acceptance or rejection of hypothesis afer comparing the calculated value of the test statistic
with the table value(critical value of the test statistic)



Module 8
Non Parametric Tests


Statistical tests that are learnt in the previous module are parametric tests, which assume that the form of the
population distribution is known and the test concerning a parameter is to be made. However there are many
situations where one or more assumptions made in the case of parametric tests can not be met. In such cases we
make use of non-parametric tests.
Non-Parametric tests are those, for which the distribution from which samples are drawn is not known and testing
of hypothesis is not concerned with parameters.
Advantages of Non-Parametric tests
1. No or less stringent assumptions are required in non-parametric tests as compared to parametric tests
2. More suitable when data is ranked, scaled or rated.
3. Easier to perform as they involve simple calculations.
List of some non-parametric tests
1. Sign test for paired data
2. Runs test for randomness
3. Mann-Whitney U-test
4. Signed rank test (Wilcoxon Matched Pairs Test)
5. Kruskal-Wallis Test

Multivariate Analysis (Write a note on Multivariate Analysis techniques.)
Multivariate analysis techniques are applicable when two or more variables are considered under study. These are
usually applied to analyze the dependence or interdependence between two or more than two variables.
Dependence multivariate techniques are
ANOVA
Multiple regression
Discriminant analysis,
Conjoint Analysis,
MANOVA and
BIET MBA Programme :Compiled by BVN Page 24

Canonical correlations.

Interdependence multivariate techniques include
Factor analysis,
Cluster analysis and
Multi-dimensional scaling

Multiple regression is used when dependence of one variable on many other variables to be studied. For example
yield of a crop depends on weather condition, fertilizer used, amount of rain, fertility of land, care of farmer etc.,.
Using Multivariate analysis, one can understand the extent of dependence of these variables for one variable under
study that is yield.
ANOVA is used when two or more variables/factors are to be studied for their equality under controlled situations.
For example, if sales efficiency of 4 salesmen are to be studied on new marketing locations during the same time,
one can make use of ANOVA, which compare between two sets of data using the variance explained between and
within each group of data.
Discriminant Analysis is used to establish a linear relationship between one dependent and many independent
variables. For example if one wants to predict the sale of insurance policies based on Income, savings, psychological
and demographic factors of the respondents, discriminant analysis gives a linear function to estimate sales using the
relationship between sales of insurance with income, savings and given psychological and demographic factors.
Conjoint Analysis is used when an investigator makes respondents preference judgment about a new concept and
test it using the actual response of the respondant.
Canonical Correlation is applicable when an experimenter wishes to relate two factors determined by two linear
models (constructed with interdependence of many variables), which generate canonical variables in two sets;
further they are tested for correlation.
Factor analysis is carried out when the investigator wants to define those factors which influence a change in
dependent variable, based on the data collected using likert scale for many questions framed on attitudes and
behaviors of the respondents.
Cluster analysis helps in separating the population of data into different homogenous groups based on common
characteristics suited to each group. For example if investigator wants to classify the customers into different
categories of brand loyalty using sales, frequency of visit, duration of relationship etc., one can make use of cluster
analysis.
Multidimensional Scaling(MDS) is used when liking towards different brands are studied based on the perceptions
of customers towards their brands.






BIET MBA Programme :Compiled by BVN Page 25

Report Writing (Write a note on various steps of report Writing). Elaborate on the following steps(*****)
1. Title
2. Executive Summary
Major findings, conclusion and recommendations as a summary in one paragraph
3. Problem Definition
Background of the problem
Literature review
Statement of the problem with specific objectives
4. Research Design
Type of research, Data collection methods, Scaling techniques used, Questionnaire, Sampling,
Fieldwork
5. Data collection, Analysis
6. Results
7. Conclusions and Recommendations


Answer to Qn 4b in model paper. (it is based on Harmonic Mean)

Airplane covers distance of first 100km with speed 100kmph, Second 100km with 200kmph, third 100km with
300kmph and fourth 100km with speed.

We know that velocity = Distance/time

Time taken = distance/velocity
= (100/100)+(100/200)+(100/300)+(100/400)=1+0.5+0.33+0.25=2.08

Therefore Avg. Speed = total distance/total time taken = 400/2.08 = 192.08kmph


Do not neglect theory part of your paper.. it is hardly about 20 pagesAll the best.

You might also like