:: Survey Research
:: Medium:
Personal (response rate = 30-35%)
Telephone (25-30%)
Computer (0.5%)
Mail/post (2%)
:: Survey Research
EXPERIMENTAL RESEARCH
:: Experimentation:
Experimentation is the manipulation of one or more
variables to measure their effect on one or more
variables. That is, cause and effect studies
:: Types of variables:
Independent
Dependant variable
:: Treatment group:
Ample exposed to manipulation of independent variable
RM (Research Methodology)
EXPERIMENTAL RESEARCH
:: Control group:
The independent variable is unchanged.
To ensure that the change in independent variable is
caused by independent variable, other variables need to
be controlled/measured in terms of their effects.
Randomization
Matching
EXPERIMENTAL RESEARCH
:: Symbols::
MB = Pre-measurement
MA = Post-measurement
X = Treatment
R = Random selection
RM (Research Methodology)
EXPERIMENTAL RESEARCH
:: Basic designs :
:: Statistical design:
They permit measurement of effect of more than 1
independent variable. It allows researchers to control for
specific extraneous variables that may affect results. It is
more economical
:: Types :
Randomized Blocks Design
Latin Square Design
Factorial Design
RM (Research Methodology)
11 IV
IV 1 DV
11 EV
EV
RM (Research Methodology)
Q1. Hoppers Hop has 800 outlets all over India. They are
planning a major sales promotion during the peak Diwali
buying period.
To determine the costing and maximized profitability, they
have identified three different price levels – P1, P2, P3 at
which they can offer their goods. The sales volumes at their
outlets vary from Rs.3,00,000 pm to Rs.30,00,000 pm. They
are apprehensive about trying out all the 3 prices at each
location as it may disturb their customer base.
Hence, they have approached you to design an ext which
would minimize customer disturbance, and at the same time
provide solution profit maximization during Diwali. Additional
information indicates that sales of the product specified are a
function of the total sales at that particular store. You have
identified 27 stores for your experiment and the process
used is RBD.
RM (Research Methodology)
Q2. GKK Ltd. markets a face cream, Hair & Lovely thru 3
types of shops – medical sore, super market, grocery
(kirana) store. Sales of the product vary based on the
store type. Sales also vary based on seasonality, which
could be summer, monsoon, winter. The company desires
to test out the impact of 3 different price discounts so as
to rationalize its product pricing all India. For this purpose
they have asked you to conduct an experiment which can
be completed in 6-7 months. The process used is LSD.
RM (Research Methodology)
Process :
Each extraneous variable is divided into equal number of
blocks/levels. Independent variable is divided into same
number of levels. Extraneous variables make up rows and
columns of the table. IV is assigned to each cell such that
each IV level appears only once in each row/column.
Then use basic design on each cell.
3 x 3 Latin Square:
Store Type
Time period Drug store Super market Grocery store
1 Price 2 P3 P1
2 P1 P2 P3
3 P3 P1 P2
RM (Research Methodology)
Interaction effect:
Sum of combined effect
<is not equal to>
Effect of IV1 + effect of IV2
Process :
Each level of 1 IV is row, and the other is the column. A
separate cell is for each combination. Each
combination/cell is rated and average rating value is put
into the cell using basic design). Analysis of Variance
(ANOVA) is used to analyze favorable outcome and
interaction.
Carbonation Sweetness
1 2 3 4 5
1 2 4 7 10 12
2 2 3 4 7 8
3 4 6 8 5 5
4 10 15 11 6 4
5 13 9 6 3 2
RM (Research Methodology)
EXPERIMENTAL ENVIRONMENT
:: LAB Design :
This is widely used in package pretesting, ads, products
and product concepts
EXPERIMENTAL ENVIRONMENT
:: FIELD Experiments
Advertising copy field tests (using Day After Recall)
Test marketing :
- Standard Test Marketing (thru manufacturer
distribution)
- Controlled Test Marketing (thru marketing
research firm)
- Electronic Test Marketing (using electronic test
marketing tools)
- Simulated Test Marketing (pre-qualify users)
RM (Research Methodology)
MEASUREMENT
:: Definition:
Assignment of numbers to characteristics of objects,
persons, states or events, acc to some rules.
:: Concept :
A concept is an invented name for an object, person,
state or event.(The above characteristics are called
concept)
eg : Sales (measured by volume or by value), Market
share (Own sales / Total sales . Eg: For Bajaj Chetak,
market share is calculated as sales of B.Chetak / total
sales of all motorcycles/scooters. Therefore, be careful
when using the term M.S bcoz the Denominator may
vary. So always specify "MS of what??"), Brand Loyalty
(cud be towards product, store, manufacturer)
RM (Research Methodology)
MEASUREMENT
MEASUREMENT
:: Measurement Scales :
There are 4 types of scales :
(a) NOMINAL –
Numbers used to characterize objects or events (numbers
used as symbols – such as 0 for male & 1 for female)
(b) ORDINAL –
Numbers/letters/symbols used to rank items (Rank 1, 2 etc.
But the problem is that 1st rank may have got 99% and 2nd
rank might have got 88%)
(c) INTERVAL –
Numerically equal distance on the scale represents equal
distance in property being measured. Location of 'zero point'
not fixed. Eg : body temperature, CGPA scores
(d) RATIO –
Interval scale with a meaningful zero. Eg : sales (if A sold
10 units & B sold 20, B sold twice as much as A)
RM (Research Methodology)
MEASUREMENT
:: Measurement Accuracy :
M = C+E
M - measurement
C - correct value
E - errors
MEASUREMENT
:: Measurement Development :
MEASUREMENT
:: QUESTIONNAIRE DESIGN :
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
Preliminary decisions -
- Required information
QUESTIONNAIRE DESIGN
Question content:
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
Response format
QUESTIONNAIRE DESIGN
Response format
Multiple choice:
Advantages :
Easy to administer
Reduces interviewer bias
Tabulation & analysis is simple
Disadvantages :
Development needs a lot of effort
Issues: 1. Number of alternatives (each should appear only
once, all possible alternatives should be included) 2.
Alternative to be balanced <<very bad, bad, ok, good, very
good>> 3. Remove position bias by using split-ballot technique
(if alternatives are read out, preference is for last alternative) (if
alternatives are given together, preference is for first)
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Response format
Dichotomous:
Issues:
1. May include 3rd option (don’t know, no opinion, can’t say)
2. Form may influence results, so use split ballot technique.
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Question sequence
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
Pretest & revision
1. Mandatory for all questionnaires
2. Decisions involved are:
- What items to pretest? <<ALL>>
- How should it be conducted (std administration,
debriefing, protocol analysis)
- Who should conduct it
(Field exec, research associate, research head)
- Which respondents should be involved
(known people, unknown people, part of sample)
- How many respondents?
(Field exec should go to as many respondents as he can cover in one
working day, research associate should go to as many respondents
as required until further pre-testing does not result in any change in
the questionnaire, research head should pre-test it with 5-8
respondents outside the office)
3. Language issues: (Especially in India). Do back-translation
RM (Research Methodology)
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
Rating scales:
Attribute of an object being marked on a numerically valued
continuum or in numerically ordered category.
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Rating
Scale
Non-comparative
Comparative
(monadic)
Paired
comparison
Rank order
Constant
sum
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Non-comparative (monadic):
Respondent not provided with a standard to use in the
rating. Only one product/brand is evaluated (hence,
monadic). Commonly used at product concept testing stage
in new product development process
Eg: Intention to buy, using a Juster scale (11 pt)
10. Certain/practically certain (99 in 100)
9. Almost sure (9 in 10)
8. Very possible (8 in 10)
7. Probable (7 in 10)
6. Good possibility (6 in 10)
5. Fairly good possibility (5 in 10)
4. Fair possibility (4 in 10)
3. Some possibility (3 in10)
2. Slight possibility (2 in 10)
1. Very slight possibility (1 in 10)
0. No chance/almost no chance (1 in 100)
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Non-comparative (monadic):
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
Types:
(i) D-T scale (Delighted – Terrible) scale.
Eg: How do u feel after seeing the movie KKHH?
Scale varies as “Delighted, Pleased, Mixed, Unhappy, Terrible”.
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
You have to find out what soft drink the guy usually drinks
(ref pt). Otherwise it is like non-comparative scale.
RM (Research Methodology)
QUESTIONNAIRE DESIGN
A B C D E
A
B
C
D
E
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
Price : 1
Economy : 2
Reliability : 3
Safety : 4
Comfort : 5
Style : 6
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Economy:
Style:
Price:
Comfort:
Reliability:
Safety:
TOTAL = 100
RM (Research Methodology)
QUESTIONNAIRE DESIGN
3 groups emerge:
Group A Group B Group C
Price 35 20 65
Eco 30 18 9
Rel 20 17 8
Saf 10 16 7
Comf 3 15 6
Style 2 14 5
QUESTIONNAIRE DESIGN
Attitude scales::
scales:
Forms -
1. Semantic differential scale
2. Stapel scale
3. Likert scale
RM (Research Methodology)
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
QUESTIONNAIRE DESIGN
2. Stapel scale:
Simplified version of the SD. Standard is unipolar, 10
interval rating scale with values from +5 to -5. Any number
of values can be used. Values can be all positive.
QUESTIONNAIRE DESIGN
Eg:
XYZ is one of the most attractive stores (SA,A,N,D,SD)
Service at XYZ is not satisfactory (SA,A,N,D,SD)
Service is very important to me (SA,A,N,D,SD)
RM (Research Methodology)
QUESTIONNAIRE DESIGN
Depends on:
• Information requirements of the problem
• Characteristic of response
• Means of administration
• Cost
In general,
- Use multiple measures since summation of
measurements
is more accurate.
RM (Research Methodology)
SAMPLING
Objective of MR is
- estimation
- hypothesis testing.
SAMPLING
Census Vs Sample
Sample Design
Sample Size
Estimation of
sampling costs
Execute sampling
process
RM (Research Methodology)
SAMPLING
Population definition:
Definition should include -
- element
- sampling unit
- extent
- time
SAMPLING
Census vs Sample:
Census includes the entire population. Sample is a
representative part of the population.
Census may not be feasible all the time. Hence use sample.
Advantages of sampling: speed/timelines & costs
However we need to use census in cases when the population
is very small.
SAMPLING
Sample design:
Sample frame: This is a means of representing elements of the
population.
Eg: telephone directory. It is not used for non-probability sampling.
SAMPLING
competing, but are being sold in the same way. Eg: shampoo &
toothpaste)
RM (Research Methodology)
SAMPLING
SAMPLING
Probability sampling:
Sampling unit is selected by chance. Chance of each unit being
selected is known.
Designs:
(i) Simple random sampling (SRS): Each sample unit has a
known and equal probability of selection. It is drawn by
random procedures from the sample frame.
SAMPLING
SAMPLING
Sample size :
Unaided judgment:
- Arbitrary approach
- Precision not clear
- Costing not taken into account
- Used commonly, nut should be avoided
RM (Research Methodology)
SAMPLING
Sample size :
SAMPLING
Sample size :
1250!
Then total possible samples =
(1250 – 50!) 50!
This is a very high number. Hence (take for eg) 500 samples.
Determine mean of each sample.
RM (Research Methodology)
SAMPLING
Value of mean Frequency of Rel Freq of sample
occurrence mean
38 – 39.99 1 .002
40 – 41.99 2 .004
42 – 43.99 17 .034
44 – 45.99 39 .078
46 – 47.99 52 .104
48 – 49.99 85 .170
50 – 51.99 110 .220
52 – 53.99 77 .154
54 – 55.99 64 .128
56 – 57.99 37 .074
58 – 59.99 10 .020
60 – 61.99 4 .008
62 – 63.99 2 .004
TOTAL 500 1.00
RM (Research Methodology)
SAMPLING
SAMPLING
SAMPLING
N = 1250
N = 50
N = s2z2 /E2
RM (Research Methodology)
SAMPLING
SAMPLING
DATA REDUCTION
Steps involved:
(i) Field controls
(ii) Editing
(iii) Coding
(iv) Transcribing
(v) Generating new variables
(vi) Summary statistical calculation
RM (Research Methodology)
DATA REDUCTION
(iii) Coding
Coding involves establishing categories and assigning data to
them. Categories established as per QD procedures. Categories
shown by numbers as much as possible. Data assignment is
relatively simple for close-ended questions. For open-ended,
create a code-book to indicate system of coding used. That
ensures uniformity.
RM (Research Methodology)
DATA REDUCTION
(iv) Transcribing
This is the process of physically transferring data from
measurement instrument to computer storage. At times may be
DATA REDUCTION