Download Marketing Research: Conjoint Analysis and Direct Value Method - Prof. Ali A. Yassine and more Study notes Engineering in PDF only on Docsity!
Marketing Research –
Conjoint Analysis &
Direct Value Method
Ali Yassine
Department of Industrial & Enterprise Systems Engineering University of Illinois at Urbana Champaign [email protected]
Introduction to Systems &
Entrepreneurial Engineering
Today
z Marketing Research
- Preference Conjoint Method
- Direct Value Method
z Examples
Marketing Research
z Value is related to key product attributes
z Managers must find a way to provide value
to consumers through those attributes.
z The goal of marketing research is to provide
facts and direction to make more informed
decisions about how to create value for
consumers.
Steps in Marketing Research
z Define the Market
z Research Design
z Design Surveys
z Collect Data
z Analyze Data
Industry & Market
z Industries
- Automotive
- Computer Software
- Aerospace
- Agriculture
- Steel
- Jewelry
z Market
- Light Truck
- Laptop
- Desktop
- Operating Systems
- Diamonds
- Wheat, soy-beans
Defining the Market
z Demographics
z Attitudes
z Preferences with Respect to Product Benefits
z Price Sensitivity
z Decision Rules
z Usage
z Product Form
z Competitive Products
z Multiple Perspectives
Source: Urban and Hauser 1993
Research Design
z List attributes
z Determine the range for each attribute
z Determine the number of levels
Experimental Designs
z Full-Factorial Design
z Fractional-Factorial Design
z Orthogonal Arrays
The analyst must be very careful in
choosing the right array for the occasion
Design the Survey
z Dillman (1978)— Mail and Telephone
Surveys: The Total Design Method (book)
z Make sure to include questions that can be
used to determine if individual is in market
under investigation
z The survey should not take too long
z The survey should not be taxing
Conjoint Study
z A Popular Marketing Research Technique
- determine ideal combination of attributes
conjointly
- determine the relative values of each attribute
- determine the expected value of a product to the
customer
Select One
Select One
Select One
$16,
$16,
$16,
$17,
$19,
$21,
Mid-sized Sedans
Conjoint Example
Vehicl
e 1
Rank
_____
0 to 60 mph
12 sec
Interior noise
68 dBA
Fuel economy
20 mpg
Price
Multi-factor Conjoint Evaluation
z Make cards with each feasible combination
z Ask the respondent to put cards in order
z Gets complicated with more attributes
z Psychological stress
34 =81combinations!!!
Levels Attributes
Comparison $ Part Worths to Values
Values were
computed from
value curves
using a
baseline value
of V0=$50,
(twice base
price for E2=1).
V W X Y
Ordinal $ Part Worths Value
Attribute Part Worths vs Baseline Change
Baseline 1.
10 sec 2.33 $ 3,500$ 876
8 sec 4.67 $ 7,000$ 1,
66 dBA 1.67 $ 2,500$ 770
62 dBA 2.33 $ 3,500$ 2,
25 mpg 1.00 $ 1,500$ 1,
30 mpg 2.00 $ 3,000$ 1,
$ 27,000 0.33 $ 500 N/A
$ 30,000 -2.33 $ (3,500)N/A
=MMULT(XS,InverseRank)
X10 =W10*$V$
$ 1,500= $ per part worth
Bar Graph Comparison
10 sec 8 sec 66 dBA 62 dBA 25 mpg 30 mpg
Attributes (acceleration, interior noise, fuel economy)
$ Part Worths
$ Added Value
There is poor
agreement because
demand is not an
analytic function of
part worths
(preference rankings).
Rather involved
simulation programs
are used to forecast
demand using part
worths.
Values and prices
can be combined in a
straightforward,
transparent manner
to forecast demand.
Critiques
z Limited Capacity
z Human error in Selecting Attributes
z Incomplete “Maps”
z Unrealistic Choice Model
z No consistent baseline
Additional Critique
z Great for getting preferences
z Poor at predicting demand and ideal price
because demand is not an analytic function
of part worths (preference rankings)
z Rather involved simulation programs are
needed to forecast demand using part worths
Direct Value Method
z Developed at University of Illinois
z A re-application of the conjoint concept
z Relates value, demand, and price in a more useful
way to aid with decision making
z Uses a conjoint survey as the basis for finding the
value for each attribute
DVM
z Direct Value (DV) Method
_1. Demand theory
- Survey method (developed_
around demand theory)
Direct Value Survey Example
Value of a Single Feature
Prospect theory has shown the importance of comparing an alternative to
a consistent, specific baseline.
Select one
Select one
Select one
Without 4th door With 4th door $20,
$20,000^ Select^ one $21,
$20,000 Select one $22,
×
×
×
×
×
DV Method Plot
The fraction (f) is
plotted versus the
price of the
alternative to
determine neutral
price.
f [%]
Price
Value of 4th Door =$21,225-$20,000 = $1,
DVM Survey
Select One
Select One
Select One
Select One
Select One
Assume you are in the market to buy a new mid-sized car. You have a choice of the baseline car on the left or the alternative on the right. The vehicles are identical except for price and acceleration time from 0 to 60 mph. For each of the six paired comparisons, select either the baseline or the alternative.
Select One
Baseline 0 to 60 mph 12 sec
Alternate 0 to 60 mph 8 sec Price Price
Example of a Direct Value
Method survey for
determining the value of
reducing 0 to 60 mph time
from 12 to 8 sec.
DVM Survey
Example of a Direct
Value Method
(conjoint) survey for
determining the value
of a change in
multiple attributes.
Baseline 0 to 60: 12 sec Interior noise: 68 dBA 20 mpg
Alternate 0 to 60: 12 sec Interior noise: 66 dBA 25 mpg
Conjoint Trial 2 using DV Method
Price Select One
Select One
Select One
Select One
Select One
Select One
DV Method Plot: Reducing Noise Level
from 66 to 62 dB(A)
The logit form
Ln(f/[1-f]) can be
used instead of f
to improve
linearity of plot.
P − P 0
V − V 0 = $1896 ± 173
Ln
f
1 − f
Source: reprinted with permission from SAE paper 980621 Copyright Society of Automotive Engineers, Inc.
Value Curve Equations
( ) ( ) (^) V
V g v g f V
γ = =
[ ] [ ]
[ ] [ ]
I C I
V
I C I
g g g g
f
g g g g
⎢ −^ −^ − ⎥
[ ] [ ]
[ ] [ ]
( ) (^) I C I
I C I
V g g^ g^ g^ g Ln Ln V (^) g g g g
⎛ ⎞ ⎡^ − − − ⎤
⎜ ⎟ =^ ⎢^ ⎥ ⎝ ⎠ ⎢ −^ −^ − ⎥ ⎣ ⎦
Dimensionless value coefficient
Parabolic function
Linear plot equation used to determine GAMMA:
y slope x
