rm 07 quant

8/3/2019 rm 07 quant

http://slidepdf.com/reader/full/rm-07-quant 1/31

Slide no. 1

April 27, 2012

Author: Dilip Thosar

RM : Research Methodology

Research Methodology

IntroductionAll progress is born of inquiry.

Doubt is often better than overconfidence, for it leadsto inquiry and inquiry leads to invention

8/3/2019 rm 07 quant


Slide no. 2

April 27, 2012



What is Research?• Systematic and planned investigation in search or

verification of facts or knowledge as yet unknown

• Recommended Text :

– Research Methodology : Kothari, New Age, – Marketing Research : Nargundkar, TMH, 978070220874

8/3/2019 rm 07 quant


Slide no. 3

April 27, 2012



Objective of Research• Explorative: gain familiarity with and insights into a

phenomenon of interest• Descriptive: Discover and portray characteristics of

something of interest

• Diagnostic: Measure statistics of some occurrences of

interest and their association with other visible factors

• Hypothesis testing: Test out if a causative relation assumed(hypothesis) is true in practice or reality

8/3/2019 rm 07 quant


Slide no. 4

April 27, 2012



Types of Research• Descriptive v/s Analytical

• Applied v/s Fundamental

• Quantitative v/s Qualitative

• Conceptual v/s Empirical

• Conclusion- v/s Decision- oriented

8/3/2019 rm 07 quant


Slide no. 5

April 27, 2012



The Research Process• Define Research Problem

• Review existing knowledge related to problem

• Formulate the hypothesis

• Research Design

•Laboratory experimentation / Field survey

• Data Collection

8/3/2019 rm 07 quant


Slide no. 6

April 27, 2012



Defining the Research Problem• Defining the problem type

– Exploring states of nature

– Deduce relationships between variables

• Coming to existing level of knowledge

– Existing Literature

– Domain study

– Pilot Survey

– Discussions with experts

• Defining and re-phrasing the Problem – Relation with business objective

– Gives direction to the research process

8/3/2019 rm 07 quant


Slide no. 7

April 27, 2012



Group Exercise• Identify a Business Problem

• List out one point of strategy in solution of above

• Locate the need for Research

• Define a Research Objective

• Define the Research Problem

• Document all of above for detailing the Research Design

8/3/2019 rm 07 quant


Slide no. 8

April 27, 2012



Research Design• Detailing the process and environment for inquiry

• Chain of variables for resolving the research problem fromdata that can be collated/ created by experiment

• Source of data and process of procuring data

• Questions:

– What (information) is the study trying to find out?

– What data is required to infer this?

– How will that data lead up to the

– Where/from whom will this data be procured

8/3/2019 rm 07 quant


Slide no. 9

April 27, 2012



Research Plan• Specification of Research Objective and Problem

• Identification of what information is required to enable solvethe problem

• Defining Operationally the variables required: Quantification,source, method of procurement, conditions of measurement,

instruments, assumptions.• Defining how the variable values procured will compute up

to the problem solution

8/3/2019 rm 07 quant


Slide no. 10

April 27, 2012



Group Exercise: research plan• For the Research objective defined in earlier group

exercise, delineate the research plan including the researchdesign

8/3/2019 rm 07 quant


Slide no. 11

April 27, 2012



Inductive and Deductive Logic• Deductive Logic:

– Testing a hypothesis – Top-down

• All iiebm students study• Rahul is an iiebm student’ => Rahul studies

– Think:• Rohan is not an iiebm student• Does Rohan study?

• Some iiebm students play chess• Does Rahul play chess?

• People who don’t play the game as I suggest, lose the game

• You lost the game• Did you listen to me?

Deduction

8/3/2019 rm 07 quant


Slide no. 12

April 27, 2012



Inductive and Deductive Logic• Induction

– Bottom-up – educated guess

– Derive a rule with a probability

– Think• I saw three Japanese tourists

• I could not distinguish one from the other

All Japanese look the same

ORAll Japanese look very similar (so similar that there is a

low (10%) chance of anyone (I) differentiating onefrom another.

Induction

8/3/2019 rm 07 quant


Slide no. 13

April 27, 2012



Ethics in Research• Why is Ethics very important in Research

• Liability to Research sponsor : public

• Honesty, Integrity, Transparency

• Integrity and Carefulness

• Confidentiality and IP

• Credit to source of data

• Legality

• Privacy

8/3/2019 rm 07 quant


Slide no. 14

April 27, 2012



Sampling• Universe/Population, Sample, Census, Sample Survey

• Sampling Design – Universe type: finite/infinite

– Sampling Frame (Source list): representative of universe

– Sampling Unit

– Characteristics/ parameters of interest• Watch out:

– The Indeterminacy Principle

– Defective Measuring Device

– Biased Sampling – Non-respondents

• Sampling Error and Systemic Bias

8/3/2019 rm 07 quant


Slide no. 15

April 27, 2012



Sampling Design• Definition of Universe and the Sampling Unit

• Fixing Sampling frame (source): representative of universe

• Group parameters of interest (average of an attribute, …)

• Budgetary considerations

• Procedure of sampling

• Quality of Sampling Design

• “Random” choice of sample respondents

• Sampling Methods

– Systematic Sampling – Stratified Sampling: homogenous strata, pro-rata nos.

– Cluster Sampling : single/multi-stage

8/3/2019 rm 07 quant


Slide no. 16

April 27, 2012



Exercise (local) : Sampling• No. of stores in 15 cities. Want to sample 10 stores

• Systematic Sampling: Step of 50 (500/10) starting random 12 gives

• Stratified Sampling:

• Cluster Sampling: city as cluster, choose 10 “largest” cities, 1store/city

City 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 15

No. of Stores 35 17 10 32 70 28 26 19 26 66 37 44 33 29 28 500

Serial nos 35 52 62 94 164 192 218 237 263 329 366 410 443 472 500

sample piece 1 2 3 4 5 6 7 8 9 10

store no. 12 62 112 162 212 262 312 362 412 462

city no 1 3 5 5 7 9 10 11 13 14

strata=city 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

city/total 7% 3% 2% 6% 14% 6% 5% 4% 5% 13% 7% 9% 7% 6% 6%

prorata 0.7 0.34 0.2 0.64 1.4 0.56 0.52 0.38 0.52 1.32 0.7 0.9 0.7 0.6 0.6

chosen city 3 6 9 10 1 4 2 5 7 8

store no. 1 63 165 193 264 330 367 411 444 473

chosen city 5 7 1 10 2 4 3 6 8 9 402

city/total 9% 8% 17% 7% 16% 9% 11% 8% 7% 7% 100%

No. of Stores 1 1 2 1 2 1 1 1 1 1 12

chosen city 1 1 2 2 1 1 1 1 10

store no. 18 79 130 297 348 389 427 458

8/3/2019 rm 07 quant


Slide no. 17

April 27, 2012



Group Exercise : Sampling• For the Research Design planned in earlier exercise, define

the sampling plan

8/3/2019 rm 07 quant


Slide no. 18

April 27, 2012



Variables• Contextual position of variables

– Dependent

– Independent

– Extraneous

8/3/2019 rm 07 quant


Slide no. 19

April 27, 2012



Measurement Errors• Sources of errors

– Respondent

– Situation

– Measurer

– Instrument

• Tests of good measurement

– Validity

– Reliability

– Practicality

8/3/2019 rm 07 quant


Slide no. 20

April 27, 2012



Measurement And Scales• Attribute “valuation”

• Value domains: quantity, set membership, name, serial no.• Types of Data:

– Nominal: Names. eg. “Anil”, ”2001” (year), “8” (room no.)

– Ordinal: Ranking. Eg. 1,2,3; Employee grade M1, M2, M3

– Interval: ranking on scale with equidistant points. Eg. ºC – Ratio: measuring on scale with an absolute zero. Eg.

Length 1.1, 1.45, mtrs.

• Think: Which operations work on each of above datatypes?

– Commutation: if op(a,b) and op(b,c) then op(a,c) – Difference: is (a-b) or (a-b)>(b-c) meaningful?

– Ratio: a is twice, 10% or even just plain more than b?

8/3/2019 rm 07 quant


Slide no. 21

April 27, 2012



Scaling Techniques: Rating• Rate items under study

• Multiple Choice Questions

– Eg. Q. How do you rate this slide for utility: “E,VG,G,A,P”

• Graphic (Adjective)

• Itemized (Descriptive)

8/3/2019 rm 07 quant


Slide no. 22

April 27, 2012



Scaling Techniques: Ranking• Compare pairs• Eg. Q. Whom do you prefer than the other as a friend?

– Khan, Singh, Sheila, James• N = Sample size = 100• n = no. of candidates = 4• Q = No. of comparisons (questions) = n(n-1)/2 = 6• C= no. from sample preferring a candidate• Mp = Column Mean = (C+N/2)/nN• Z : Z-score for cumulative area = Mp on a standard normal curve*

http://www.danielsoper.com/statcalc/calc19.aspx

Preferred-> Khan Singh Sheila James Khan - 65.00 32.00 20.00

Singh 40.00 - 38.00 42.00

Sheila 45.00 50.00 - 70.00

James 80.00 20.00 98.00 -

no. preferring C 165.00 135.00 168.00 132.00

Rank order 2.00 3.00 1.00 4.00

Column Mean Mp 0.54 0.46 0.55 0.46Z-factor for Mp Zj 0.09 (0.09) 0.11 (0.11)

Positive Zj Rj 0.20 0.02 0.22 0.00

0.0 0.1 0.2 0.3 0.4 0.5

James Singh Khan Sheila

http://www.danielsoper.com/statcalc/calc19.aspx

8/3/2019 rm 07 quant


Slide no. 23

April 27, 2012



FU Scaling Methods• The Likert Scales

– Measuring Attitude

– 1-5 (or 3/7) degrees on a spectrum

– Eg.

• The Semantic Deifferential Scale

– Measure between bipolar evaluations

– Antonym adjectives as ends with 5/3 points in between

– Eg. Q: How do you find the taste of Acme oranges?

1 2 3 4 5

StronglyDisagree

Disagree Undecided AgreeStronglyAgree

-2 -1 0 1 2

Sweet Sour

8/3/2019 rm 07 quant


Slide no. 24

April 27, 2012



Experimental Design• Informal

– Before-and-after without control – After-only with control

– Before-and-after with control

• Formal

– Completely randomized

– Randomized Block design

– Latin Square design

– Factorial Designs

• Groups

– Experimental

– Control

• Treatment of Sample

8/3/2019 rm 07 quant


Slide no. 25

April 27, 2012



Methods of Data Collection• Observation/ Experiment

• Study / Interview / Questionnaire

• Primary Data collection best-practice points

– Don’t ask “leading questions”

– Explain questions/options, but without leading a bias

– Don’t stress the respondent’s memory/intellect

– Keep the length of questionnaire optimal

– Provide understandable options to closed ended qs

• Secondary Data – Reliability, suitability, Adequacy

– Cross-check and Cross-link

8/3/2019 rm 07 quant


Slide no. 26

April 27, 2012



Data Analysis• Cleaning Data

• Coding data

• Classification

• Tabulation

• Frequently Occurring Problems

– Handling “Don’t know” or “no response” responses

• Understanding and use of Statistics

• Types of Analyses

– Univariate, Bivariate and Multivariate analysis – Causal and Correlational analysis

8/3/2019 rm 07 quant


Slide no. 27

April 27, 2012



Data Input Format

INPUT Variables Age

(Ratio)

Gender

(Nominal)

Income

(Interval)V4 V5

Respondents Data

Rahul 21 M 3 1 2

Rajiv 22 F 4 3 3

R3 21 M 3 1 1

R4 23 F 2 5 3

R5 24 F 3 3 2

R6 20 M 4 5 5

R7 22 M 3 7 6

8/3/2019 rm 07 quant


Slide no. 28

April 27, 2012



Cross Tabulation

Sales Region N E W S

Brand 849 305 132 255 157

1 123 42 23 43 15

2 279 154 22 78 25

3 131 22 21 56 32

4 184 55 34 44 51

5 132 32 32 34 34

8/3/2019 rm 07 quant


Slide no. 29

April 27, 2012



Statistical Analysis• Univariate Analysis (Frequency distribution: nominal/ordinal)

• Bivariate/ Multivariate Analyses : – Correlation (Chi-squared test for 2 variables)

– Regression

– Variance

– ANOVA

– Factor Analysis

– Cluster Analysis

– Discriminant Analysis

– Multidimensional Scaling

– Conjoint Analysis

• Hypothesis Testing

– Hypothesis: statement or opposite, of what we desire/ hope to prove

– p-value of t-test < 5% => Reject H0)

8/3/2019 rm 07 quant


Slide no. 30

April 27, 2012



Basic Statistics• Percentage, Percentile (Quartile)

• Mean, Median, Mode• Variance

• Standard Deviation

• Co-Variance

i x1 x2 v1=x1-x1bar sqr(v1) v2=x2-x2bar sqr(v2) v1*v2

1 1 8 -3.5 12.25 3.5 12.25 -12.25

2 2 7 -2.5 6.25 2.5 6.25 -6.25

3 3 6 -1.5 2.25 1.5 2.25 -2.25

4 4 5 -0.5 0.25 0.5 0.25 -0.25

5 5 4 0.5 0.25 -0.5 0.25 -0.25

6 6 3 1.5 2.25 -1.5 2.25 -2.25

7 7 2 2.5 6.25 -2.5 6.25 -6.25

8 8 1 3.5 12.25 -3.5 12.25 -12.25

n=count (i) 8 8 8 8 8 8 8

s=sum(xi) 36 36 0 0 42 -42

xbar=s/n 4.5 4.5

sd=sum(sqr(v)) 42 42 -42

covar(x1,x2)=sd/(n-1) 6 6 -6

i x1 x2 x1 percentile x1 percentile1 1 8 0% 100%2 1 7 0% 86%

3 3 6 29% 71%4 4 4 43% 43%5 4 4 43% 43%6 6 3 71% 29%7 7 2 86% 14%8 8 1 100% 0%

n=count (i) 8 8s=sum(xi) 34 35mean=s/n 4.25 4.375

median 4 4mode 1 4

8/3/2019 rm 07 quant


Slide no. 31

April 27 2012



Statistical Packages• SPSS (IBM), SAS (SAS), R (FSF/GNU), …

rm 07 quant

Documents