Measuring Consumer Preference:

Conjoint Analysis

Li Yang
Dec 2019
 Measure Consumer Preferences
• Why do conjoint?

• How to do conjoint?

• What can we get with conjoint results?

– Apple vs. Samsung: The $2 Billion Case
 Example
Which of the following tablets do you prefer?
A. Windows, with cellular network
connectivity, 12 hour battery
life, 7 inch screen, for $300

B. iOS (Apple), without cellular

network connectivity, 8 hour
battery life, 10 inch screen, for
$100
0% 0%
...
..

llu
r.
la

ce
llu

t
ou
ce

ith
ith

w
w

,
s,

e)
w

pl
do

Ap
in

S(
W

iO
Measuring Preferences for Tablets
 Price:  Screen:
• $100 • 7 inches
• $300 • 10 inches
• $500
 Battery life:  Cellular network access:
• 4 hours • No
• 8 hours • Yes
• 12 hours
 OS: What do you want to know?
• iOS (Apple)
• Windows
• Android
 Questions We Want to Answer
• What are the attributes most important to consumers?

• Within the attributes:

– How much do people prefer one level to another?
• Apple vs. Android?
• 8 vs. 12 hour battery life?
– How much are people willing to pay for those attribute levels?

• What combination of attributes

– Do customers like best?
– Will gain the most market share?
– Are there segments of people with distinct preferences
 Managerial Uses of Conjoint
• Understand the relative contributions of each attribute and
each level to the overall evaluation of the product/service
• Select optimal combination of features for a product /service
• Predict market shares among products with different sets of
features
• Identify market potential for product concepts not yet
available
• Define benefit segments
• Estimate brand equity
 Did You Use Conjoint?
1. I never heard of it
2. I learned it but never really used it
3. I used it but never really learned it
4. I learned it and used it
5. I could teach this session 0% 0% 0% 0% 0%

it

n
it
..

io
.
of

al

l..

ed

ss
re
rd

ly

us

se
er

al
ea

is
re
ev

an
h

th
er
tn
er

it

h
ev
ev

bu

ac
ed
tn
In

te
rn
it

bu

ld
ed

ea

ou
Il
it
rn

Ic
d
ea

se
Il

Iu
 What is Conjoint Analysis
• Multivariate technique used to understand how consumers
make tradeoffs among attributes or characteristics of
products or services
• Based on the premise that consumers evaluate the utility of a
product by combining the separate amount of utility provided
by each attribute
• Respondents perform a realistic task – provide overall
evaluations of products
• Attributes are considered jointly

– Subjects make trade-offs for attribute levels

Corporate Uses of Conjoint
 Conceptual Underpinnings
• Consumers differ in their utilities for attribute levels
• Utility for a product = sum of utilities of its attribute levels

U(Tablet) = U(Screen) +U(Cell) +U(Price) +U(Battery) +U(OS)

• Utilities can be measured by consumer evaluation of overall

product profiles
 Why Not Simply Ask People Directly?
• Difficult for people to answer questions like: “How much do
you value a 7-inch screen on your tablet?”

• Less accurate
– People tend to rate all attributes as (equally) important

30%
Actual
25%
response data
20% from CKMBA13
15% survey
10%

5%

0%
Screen size Cell network OS Battery life Price

Stated importance Revealed importance (Conjoint)

 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

We show the procedure for the simplest Conjoint
Rating-based Conjoint
 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

 Stage 1 – Identify Attributes
 Price:  Screen:
• $100 • 7 inches
• $300 • 10 inches
• $500
 Battery life:  Cellular network access:
• 4 hours • No,
• 8 hours • Yes
• 12 hours
 OS:
• iOS (Apple),
• Windows,
• Android
 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

 Stage 2 – Identify Levels
 Price:  Screen:
• $100 • 7 inches
• $300 • 10 inches
• $500
 Battery life:  Cellular network access:
• 4 hours • No
• 8 hours • Yes
• 12 hours
 OS:
• iOS (Apple)
• Windows
• Android
 Stages 1 & 2: Attribute & level
Selection Guidelines
• Attributes in conjoint should be
– Clear and unambiguous
– Relevant to choice
– Actionable by the firm

• The total number of attributes should be kept low

– 5-6 is the average, most studies fall between 4 and 8

• Levels should span the realistic range of possibility

• Use qualitative research and pretests to decide on
attributes and levels
 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

Apple + 7” Screen + Cell NO + 12 hour battery + $100
Apple + 7” Screen + Cell NO + 12 hour battery + $300
Apple + 7” Screen + Cell NO + 12 hour battery + $500

Apple + 7”  Screen:
Screen + Cell YES + 4 hour
Price: battery + $100
Apple + 7” Screen + Cell YES + 4 hour battery + $300
• $100 • 7 inches
Apple + 7” Screen + Cell YES + 4 hour battery + $500
• $300 • 10 inches
Apple + 7”• Screen
$500 + Cell YES + 8 hour battery + $100
Apple + 7” Screen + Cell YES + 8 hour battery + $300
Apple + 7”
Battery
Screen +life:  Cellular
Cell YES + 8 hour batterynetwork
+ $500 access:
• 4 hours • No,
Apple + 7”• Screen
8 hours+ Cell YES + 12 hour battery + $100
• Yes
Apple + 7” Screen + Cell YES + 12 hour battery + $300
• 12 hours
Apple + 7” Screen + Cell YES + 12 hour battery + $500
 OS:
…..
• iOS (Apple),
…..
….. • Windows,
• Android
Android +10” Screen + Cell YES + 12 hour battery + $500
 Stage 3: Design study
• How many combinations?
– 3 OS X 2 screens X 2 cell connectivity X 3 batteries X 3 prices
– Totally 108 combinations!!

• Data collection needs simplification

– Theory of experimental designs can be used to choose a subset of
profiles for the questionnaire
– Efficient subset = “fractional factorial design”

• Where to obtain fractional factorial designs?

– Computer programs – e.g., SAS MktEx macro
– Catalogs – e.g. http://support.sas.com/techsup/technote/ts723.html
 Example: Experimental Design
Product definition:
Attribute è 1 2 3 4 5
Level ê Screen (in) Cell Price ($) Battery (hr) OS
1 7 N 100 4 iOS
2 10 Y 300 8 Windows
3 500 12 Android

Supplied design for 2x2x3x3x3 attribute levels: Resulting survey:

Screen Battery
Profile # Attribute 1 Attribute 2 Attribute 3 Attribute 4 Attribute 5 Profile # Cell Price ($) OS
(in) (hr)
1 1 2 2 3 2 1 7 Y 300 12 Windows
2 1 2 1 2 2 2 7 Y 100 8 Windows
3 2 2 3 3 3 3 10 Y 500 12 Android
4 1 2 2 1 3 4 7 Y 300 4 Android
5 1 1 2 2 1 5 7 N 300 8 iOS
6 2 1 2 3 2 6 10 N 300 12 Windows
7 1 1 3 3 3 7 7 N 500 12 Android
8 2 1 2 2 3 8 10 N 300 8 Android
9 1 1 3 1 2 9 7 N 500 4 Windows
10 2 2 1 3 1 10 10 Y 100 12 iOS
11 2 2 2 1 1 11 10 Y 300 4 iOS
12 1 1 1 3 1 12 7 N 100 12 iOS
13 2 2 3 2 2 13 10 Y 500 8 Windows
14 2 1 3 1 1 14 10 N 500 4 iOS
15 2 1 1 1 2 15 10 N 100 4 Windows
16 2 1 1 2 3 16 10 N 100 8 Android
17 1 2 3 2 1 17 7 Y 500 8 iOS
18 1 2 1 1 3 18 7 Y 100 4 Android
 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

Stage 4: Collect Data
Stage 4: Collect Data
Different Conjoint Methods
 Tradeoffs for Response Formats
Format Ease of Ease of Analysis Task Realism
Completing the
Task

Rating Intermediate Best (OLS regression) Worst

Ranking Most Difficult Intermediate (Linear Intermediate
programming)
Choice Easiest Worst (Logit regression) Best

• Ranking/choice require more sophisticated analysis

• Choice requires larger sample size/least information per
respondent
• All method estimate part-worth for each individual
• Most advanced techniques these days are adaptive
 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

 Stage 5: Data analysis
• With ratings data, compute part-worths with regression (OLS)

• Steps (for each individual) in Excel:

– Arrange profile ratings (Y) in a column
– Prepare regressor matrix (X): code experimental design with dummy
variables
• Create one column for every attribute level and fill in with zeros
• For each profile (row), set cell value of the column to 1 if the profile has that
attribute level
• For each attribute, delete one of the columns representing its attribute levels

– Perform regression using Data Analysis  Regression

 Preparing Data for Regression
Dummy
Dummy variable
variable coding
coding
Scr:
Scr: Scr: Cell:
Scr: Cell: Cell:
Cell: Px:
Px: Px: Px: Px:Px: Batt:
Batt: Batt:
Batt:Batt:
Batt:OS:OS: OS:OS:OS:OS: Subject
Subject
Profile
Profile## Profile
Profile description
description 7"
7" 10"
10" NN YY $100
$100 $500
$300 $600
$500 4h4h 8h8h 12h12h iOSiOS WinWinAndAnd ratings ratings
11 7,Y,300,12,Windows
7,Y,300,12,Windows 01 00 00 0 1 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 1 0 0 3 3
22 7,Y,100,8,Windows
7,Y,100,8,Windows 01 00 00 01 01 00 00 0 0 0 1 0 0 0 0 0 1 0 0 4 4
33 10,Y,500,12,Android
10,Y,500,12,Android 00 01 00 01 00 00 01 0 0 0 0 0 1 0 0 0 0 0 1 4 4
44 7,Y,300,4,Android
7,Y,300,4,Android 01 00 00 01 00 01 00 0 1 0 0 0 0 0 0 0 0 0 1 2 2
55 7,N,300,8,iOS
7,N,300,8,iOS 01 00 01 00 00 01 00 0 0 0 1 0 0 0 1 0 0 0 0 4 4
66 10,N,300,12,Windows
10,N,300,12,Windows 00 01 01 00 00 01 00 0 0 0 0 0 1 0 0 0 1 0 0 4 4
77 7,N,500,12,Android
7,N,500,12,Android 01 00 01 00 00 00 01 0 0 0 0 0 1 0 0 0 0 0 1 3 3
88 10,N,300,8,Android
10,N,300,8,Android 00 01 01 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 1 4 4
99 7,N,500,4,Windows
7,N,500,4,Windows 01 00 00 01 00 00 01 0 1 0 0 0 0 0 0 0 1 0 0 1 1
10
10 10,Y,100,12,iOS
10,Y,100,12,iOS 00 01 00 01 01 00 00 0 0 0 0 0 1 0 1 0 0 0 0 7 7
11
11 10,Y,300,4,iOS
10,Y,300,4,iOS 00 01 00 01 00 01 00 0 1 0 0 0 0 0 1 0 0 0 0 4 4
12
12 7,N,100,12,iOS
7,N,100,12,iOS 01 00 01 00 01 00 00 0 0 0 0 0 1 0 1 0 0 0 0 6 6
13
13 10,Y,500,8,Windows
10,Y,500,8,Windows 00 01 00 01 00 00 01 0 0 0 1 0 0 0 0 0 1 0 0 2 2
14
14 10,N,500,4,iOS
10,N,500,4,iOS 00 01 01 00 00 00 01 0 1 0 0 0 0 0 1 0 0 0 0 3 3
15
15 10,N,100,4,Windows
10,N,100,4,Windows 00 01 01 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 4 4
16
16 10,N,100,8,Android
10,N,100,8,Android 00 01 01 00 01 00 00 0 0 0 1 0 0 0 0 0 0 0 1 6 6
17
17 7,Y,500,8,iOS
7,Y,500,8,iOS 01 00 00 01 00 00 01 0 0 0 1 0 0 0 1 0 0 0 0 4 4
18
18 7,Y,100,4,Android
7,Y,100,4,Android 01 00 00 01 01 00 00 0 1 0 0 0 0 0 0 0 0 0 1 5 5

Regression X Regression Y
matrix matrix
 Preparing Data for Regression
Dummy variable coding
Scr: Cell: Px: Px: Batt: Batt: OS: OS: Subject
Profile # Profile desc
10" Y $300 $500 8h 12h Win And ratings
1 7,Y,300,12,Windows 0 1 1 0 0 1 1 0 3
2 7,Y,100,8,Windows 0 1 0 0 1 0 1 0 4
3 10,Y,500,12,Android 1 1 0 1 0 1 0 1 4
4 7,Y,300,4,Android 0 1 1 0 0 0 0 1 2
5 7,N,300,8,iOS 0 0 1 0 1 0 0 0 4
6 10,N,300,12,Windows 1 0 1 0 0 1 1 0 4
7 7,N,500,12,Android 0 0 0 1 0 1 0 1 3
8 10,N,300,8,Android 1 0 1 0 1 0 0 1 4
9 7,N,500,4,Windows 0 1 0 1 0 0 1 0 1
10 10,Y,100,12,iOS 1 1 0 0 0 1 0 0 7
11 10,Y,300,4,iOS 1 1 1 0 0 0 0 0 4
12 7,N,100,12,iOS 0 0 0 0 0 1 0 0 6
13 10,Y,500,8,Windows 1 1 0 1 1 0 1 0 2
14 10,N,500,4,iOS 1 0 0 1 0 0 0 0 3
15 10,N,100,4,Windows 1 0 0 0 0 0 1 0 4
16 10,N,100,8,Android 1 0 0 0 1 0 0 1 6
17 7,Y,500,8,iOS 0 1 0 1 1 0 0 0 4
18 7,Y,100,4,Android 0 1 0 0 0 0 0 1 5
Regression
Regression X matrix
Y matrix
 Regression Output
SUMMARY OUTPUT

Regression Statistics
Multiple R 0.98
R Square 0.95
Adjusted R Square 0.91
Standard Error 0.44
Observations 18

ANOVA
df SS MS F Significance F
Regression 8 36.05 4.51 23.48 0.00
Residual 9 1.73 0.19
Total 17 37.78

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 4.98 0.33 15.21 0.00 4.24 5.72 4.24 5.72
Scr: 10" 0.69 0.21 3.26 0.01 0.21 1.17 0.21 1.17
Cell: Y 0.11 0.22 0.51 0.62 -0.39 0.62 -0.39 0.62
Px: $300 -1.83 0.25 -7.25 0.00 -2.41 -1.26 -2.41 -1.26
Px: $500 -2.52 0.26 -9.86 0.00 -3.10 -1.94 -3.10 -1.94
Batt: 8h 0.85 0.26 3.33 0.01 0.27 1.43 0.27 1.43
Batt: 12h 1.35 0.26 5.29 0.00 0.77 1.93 0.77 1.93
OS: Win -1.69 0.26 -6.60 0.00 -2.26 -1.11 -2.26 -1.11
OS: And -0.67 0.25 -2.64 0.03 -1.24 -0.09 -1.24 -0.09

• Intercept captures utility of “baseline” profile

– Baseline profile corresponds to all omitted attribute levels
• Part-worths measure incremental utility relative to the baseline
– Part-worths of included attribute levels = regression coefficients
– Part-worths of omitted attribute levels = 0 by definition
 Part-worth Plots
Screen Cellular network connec on
1.5 1.5
1 1
0.5 0.5
Part-worth

Part-worth
0
-0.5 7" 10" -0.5 N Y
-1 -1
-1.5 -1.5
-2 -2
-2.5 -2.5
-3
Price -3
1.5
1
0.5
Part-worth

0
-0.5 $100 $300 $500
-1
-1.5
-2
-2.5
OS -3
Ba ery
1.5 1.5
1 1
0.5 0.5
Part-worth

Part-worth
0 0
-0.5 iOS Windows Android -0.5 4 hours 8 hours 12 hours
-1 -1
-1.5 -1.5
-2 -2
-2.5 -2.5
-3 -3
 Calculating Attribute Importances
• An attribute’s importance is proportional to the range (taken
over levels) of its part-worths

• For each attribute:

– Range = max(part-worth) – min(part-worth)
– Importance = range / (sum of ranges)

Attribute Range Importance

Screen 0.69-0=0.69 0.69/6.36 = 11%
Cellular network 0.11 2%
Price 2.52 40%
Battery 1.35 21%
OS 1.69 26%
Total 6.36 100%
 Stages in Conjoint Analysis
1. Identify a set of relevant product attributes
2. Define reasonable levels for those attributes

3. Design the study

4. Collect data

5. Analyze the data for each respondent

6. Simulate market outcomes

 Stage 6: Simulate Market Outcomes

• Consider a hypothetical market

– New product(s) compete against existing product(s)

• Compute expected utility for each of the products,

for each subject in the sample

• Predict market share – Max utility choice rule

• Choose product with the highest expected utility
• Product share = fraction of subjects who choose that product
Utility Computation
Market Share Prediction
 Apple VS. Samsung
 In 2012, Apple Inc. sued Samsung Electronics for $2.2
billion, contending that Samsung violated a number
of its patents related to the iPhone and iPad
 The patents covered device features such as “pinch-
to-zoom” and “double-tap-to-zoom”
 How did Apple Arrive at the $2.2B Valuation?

 Apple’s expert witness (Prof. John Hauser, MIT) ran

a conjoint analysis study to estimate the value of
each patented feature

 The six features included in the study were:

 Touch screen capabilities (allegedly infringed features)
 Size and weight
 Camera
 Storage/memory
 Connectivity (cellular and wifi)
 Price

 Each feature is described by four levels

 Choice-based Conjoint
• 455 smartphone users made 16 choices among 4 phones at a
time
 Apple vs. Samsung: Samsung’ Rebuttal

 Samsung’s claimed that the Apple’s conjoint design

emphasized relatively unimportant features while excluding
many important features (e.g. brand, battery life).

 Apple’s conjoint design artificially elevated the importance of

features that “are not even in the radar screen of
consumers.”
 Prof. Tulin Erdem (NYU), Samsung expert witness

 The conjoint design was similar to conducting a study of cars

and asking people what type of cup-holder they would prefer
rather than asking about the brand
 Prof. David Reibstein (Wharton), Samsung expert witness
Apple vs. Samsung: Court Case Verdict
• Samsung was found guilty of infringing several of the patents
in question, having “willfully infringed” on five of those
• Apple was awarded over a billion dollar in damages across
multiple trials
• Over the past three years, Apple and Samsung have filed over
40 patent lawsuits against each other
• As of August 2014, they dropped all patent lawsuits outside of
the U.S. but continue their legal battle in the U.S.
Applicability of Conjoint Analysis
Wealth Management Products

Conjoint analysis can be applied

Applicability of Conjoint Analysis
Paintings by Vincent Van Gogh

Conjoint analysis NOT applicable

 Commercial Conjoint Services
• Sawtooth Software

• Quatrics

• IBM SPSS Conjoint

 Summary
Conjoint analysis is a powerful technique for quantifying
consumers’ preferences
– Widely used
– Analysis is relatively easy
– Enables market share simulations for desirable new products
– Benefit segmentation using cluster analysis
– Willingness to pay (legal cases)
– Product must be realistically decomposable into basic
attributes
– Don’t be too ambitious in numbers of attributes/levels
 Next Time …
• Starbucks case study
– Presentations by Groups 3, 6

• Marketing strategy: Segmentation

Appendix Slides
Digital Point-and-Shoot Camera
 Conjoint: Step One
Define attributes and levels
– A product is a combination of attributes at various
(categorical) levels

Attributes Levels

An 18M Camera Brand Name Sony, Canon

Price ￥1269, ￥1369, ￥1469

Color Black, White, Pink, Red

Style Lovely, Fashionable

Function Scenic, Portrait

Type Travel, Home-use

 Conjoint: Step Two
Determine combination of attributes for the conjoint test
– Full factor design
2x3x4x2x2x2= 192 possible combinations. Way too many!
– Use computer program to generate 16 “orthogonal designs”
– Why 16? Small enough to get accurate responses, and large enough to
allow sufficient information

Attributes Levels
An 18M Camera
Brand Name Sony, Canon
Price ￥1269, ￥1369, ￥1469
Color Black, White, Pink, Red
Style Lovely, Fashionable
Function Scenic, Portrait
Type Travel, Home-use
 Conjoint: Step Three
Code Dummy Variables
Attributes Levels
An 18M Camera
Brand Name Sony, Canon
Price ￥1269, ￥1369, ￥1469
Color Black, White, Pink, Red
Style Lovely, Fashionable
Function Scenic, Portrait
Type Travel, Home-use

Utility = Intercept + 𝛽1 Sony + 𝛽2 Canon + 𝛽3 ￥1269

+ 𝛽4 ￥1369 + 𝛽5 ￥1469 + 𝛽6 Black + 𝛽7 White
+ 𝛽8 Pink + 𝛽9 Red + 𝛽10 Lovely + 𝛽11 Fashionable
+ 𝛽12 Scenic + 𝛽13 Portrait + 𝛽14 Travel + 𝛽15 Homeuse
+ε What’s Wrong?
 Dummy Variables
Incorporate categorical data with dummy variables
– When the independent variable is nonmetric and have two
or more categories
– Each dummy variable represents one category and only k-1
dummies are needed to represent k categories
• Otherwise redundant dummies

– How to interpret the coefficient?

 Conjoint: Code Dummy Variables
Attributes Levels
An 18M Camera
Brand Name Sony, Canon
Price ￥1269, ￥1369, ￥1469
Color Black, White, Pink, Red
Style Lovely, Fashionable
Function Scenic, Portrait
Type Travel, Home-use

Utility = Intercept + 𝛽1 Sony + 𝛽2 ￥1269 + 𝛽3 ￥1369

+ 𝛽4 Black + 𝛽5 White + 𝛽6 Pink + 𝛽7 Lovely + 𝛽8 Scenic
+ 𝛽9 Travel + ε

• How to interpret the coefficient?

– Part-worth 𝛽1 : the utility contribution if brand is Sony
 Conjoint: Step Four
Construct conjoint survey
– 16 test product profiles based on orthogonal design
Rate each product on a 1-to-7-integer scale
 Orthogonal Profile Design
Tell SAS, SPSS, Stata, R, etc., that you need certain number of
attributes with certain number of level under each attribute

Attributes Levels
An 18M Camera
Brand Name Sony, Canon
Price ￥1269, ￥1369, ￥1469
Color Black, White, Pink, Red
Style Lovely, Fashionable
Function Scenic, Portrait
Type Travel, Home-use
 Orthogonal Profile Design
Translate the 0/1 dummies into actual levels of each attribute

Canon, ￥1369,
white, fashionable,
portrait, home-use

Sony, ￥1269,
white, lovely,
portrait, travel
Attributes Levels
An 18M Camera
Brand Name Sony, Canon
Price ￥1269, ￥1369, ￥1469
Color Black, White, Pink, Red
Style Lovely, Fashionable
Function Scenic, Portrait
Type Travel, Home-use
 Conjoint: Step 5
Collect data from respondents

Attributes Levels
An 18M Camera
Brand Name Sony, Canon
Price ￥1269, ￥1369, ￥1469
Color Black, White, Pink, Red
Style Lovely, Fashionable
Function Scenic, Portrait
Type Travel, Home-use
 Example: Digital Camera
Estimation Results from Sophia
Attributes Part-worth Values Range Importance Weights
Brand
Sony -2.0 2.0 39%
Canon 0
=2.0/(5.125)
Price
￥1269 0.375 0.375 7.3%
￥1369 0.25
￥1469 0
Color
Black -0.5 1.0 19.5%
White 0.25
Pink -0.75 =0.25-(-0.75)
Red 0
Style
Lovely -0.5 0.5 9.8%
Fashionable 0
Function
Scenic -1 1.0 19.5%
Portrait 0
Type
Travel -0.25 0.25 sum 4.9%
Home-use 0 5.125
 Understanding Sophia
• She likes Canon better
• White is her favorite color
Did we ask her
• She’s no fan of lovely camera direct questions?

• She prefers portrait function

• Camera is for home use
• Brand is what she cares most
• Holding everything else constant, if you cut $200 from $1469,
she would give up home-use camera for travel camera
 Forecast Choices and Market Share
• Objective
– Forecast market share of a new product concept based on
conjoint estimates of utilities

• Method
– Calculate the expected utility of each product for each
respondent
• First choice rule: assume highest utility profile is chosen by each
respondent
 Camera Preference Study

• Camera preference Conjoint Analysis

study (N=8 from last class)

• The study includes the following

attributes:
– Brand – Sony, Canon
– Price - ￥1269, ￥1369, ￥1469
– Color: Black, White, Pink, Red
– Style: Lovely, Fashionable
– Function: Scenic, Portrait
– Type: Travel, Home-use
 Market Share Simulation
Consider the following products:

Sony Black Canon White Canon Pink

￥1369 ￥1269 ￥1369
Fashionable Lovely Lovely
Scenic Scenic Portrait
Travel Travel Home-use

What is Sony black’s market share? 25%, 25%, 50%

Would shares change if Canon pink raises its price to ￥1469 ? 25%, 50%, 25%
Would shares change if Sony black switches to pink? 0%, 37.5%, 62.5%