Conjoint Analysis
◼
◼
◼
Conjoint analysis attempts to determine the relative
importance consumers attach to salient attributes and
the utilities they attach to the levels of attributes.
The respondents are presented with stimuli that
consist of combinations of attribute levels and asked
to evaluate these stimuli in terms of their desirability.
Conjoint procedures attempt to assign values to the
levels of each attribute, so that the resulting values or
utilities attached to the stimuli match, as closely as
possible, the input evaluations provided by the
respondents.
What is Conjoint Analysis?
◼
◼
◼
◼
Research technique developed in early 70s
Measures how buyers value components of a
product/service bundle
Dictionary definition-- “Conjoint: Joined together,
combined.”
Marketer’s catch-phrase-- “Features CONsidered
JOINTly”
How Does Conjoint Analysis Work?
◼
◼
◼
◼
We vary the product/service features (independent variables) to
build many (usually 12 or more) product concepts
We ask respondents to rate/rank or choose among a subset of
those product concepts (dependent variable)
Based on the respondents’ evaluations of the product concepts,
we figure out how much unique value (utility) each of the
features (attributes) added
(Regress dependent variable on independent variables;
estimated betas equal to part worth utilities.)
Traditional, Less-Effective Questions
◼
How important is horsepower to you in a vehicle?
◼
How important is fuel efficiency to you in a vehicle?
◼
Which is more important to you, horsepower or fuel
efficiency?
What’s So Good about Conjoint?
◼
More realistic questions:
Which product would you prefer . . .
- 210 horsepower
or
- 140 horsepower
- 17 MPG
◼
◼
- 28 MPG
If choose left, you prefer Power. If you choose right, you
prefer Fuel Economy
Rather than ask directly whether you prefer Power over
Fuel Economy, we present realistic tradeoff scenarios
and infer preferences from your product choices
What’s So Good about Conjoint? (cont.)
◼
◼
When respondents are forced to make difficult
tradeoffs, we learn what they truly value
These values (utility scores) are associated with
specific and actionable attribute levels relevant
to the problem at hand
Building a Model
◼
Inputs:
◼ Attributes
◼ Levels
◼ Respondents
◼ Prior Knowledge
◼ External Data
◼ Experimental
Design
◼ Conjoint Method
◼
Outputs:
◼
◼
◼
Utility Scores for
each level
Importance
Scores for each
attribute
Ability to perform
Simulations
Defining Attributes
◼
◼
◼
◼
◼
Attributes are independent aspects of a product or a service (Brand, Price, Size,
Color etc.)
How many attributes?
-Depends on research objectives
◼
One rule of thumb was that no more than 6 0r 7 attributes is too much
◼ May cause respondents to simplify, looking only at 2-3 most important
Attributes should be independent, mutually exclusive
◼
Brand, quality and product life expectancy may all measure the same thing
Each attribute has varying degrees, or “levels”
◼
Cost: $1, $2, $3
◼
◼
Biodiversity Loss: 10, 50, 100
Each level is assumed to be mutually exclusive of the others (a program has one
and only one level of that attribute)
Rules for Formulating Attribute Levels
◼
Attributes are assumed to be mutually exclusive
◼
◼
◼
◼
◼
Attribute: Add-on features
Level 1= Sun roof
Level 2= GPS system
Level 3=DVD player
If you define levels in this way, you cannot determine the
value of providing 2 or 3 of these features at the same time
(or none of them)
Solutions
◼
◼
8 level Attribute:
Features
◼
◼
◼
◼
◼
◼
◼
◼
None
Sunroof
GPS system
DVD Player
Sunroof, GPS
Sunroof, DVD
GPS, DVD
Sunroof, GPS, DVD
◼
3 Binary Attributes:
◼
Sunroof:
◼
◼
◼
GPS System
◼
◼
◼
None
Sunroof
None
GPS
DVD Player
◼
◼
None
DVD Player
Rules for Formulating Attribute Levels
◼
◼
◼
◼
Levels should have concrete/unambiguous meaning
“very expensive” vs “ costs $575”
“weight: 5-7 kilos” vs “ weight 6 kilos”
-One description leaves meaning up to individual
interpretation, while the other does not
Rules for Formulating Attribute Levels
◼
Don’t include too many levels for any one attribute
◼
◼
The usual number is about 3-5 levels per attribute
Make sure levels from your attributes can combine freely
with one another without resulting in utterly impossible
combinations (very unlikely combinations OK)
Statistics and Terms Associated with
Conjoint Analysis
◼
◼
◼
◼
◼
Part-worth functions. The part-worth functions, or
utility functions, describe the utility consumers attach to
the levels of each attribute.
Relative importance weights. The relative importance
weights are estimated and indicate which attributes are
important in influencing consumer choice.
Attribute levels. The attribute levels denote the values
assumed by the attributes.
Full profiles. Full profiles, or complete profiles of brands,
are constructed in terms of all the attributes by using the
attribute levels specified by the design.
Pairwise tables. In pairwise tables, the respondents
evaluate two attributes at a time until all the required
pairs of attributes have been evaluated.
Statistics and Terms Associated with
Conjoint Analysis
◼
◼
◼
◼
Cyclical designs. Cyclical designs are designs
employed to reduce the number of paired comparisons.
Fractional factorial designs. Fractional factorial
designs are designs employed to reduce the number of
stimulus profiles to be evaluated in the full profile
approach.
Orthogonal arrays. Orthogonal arrays are a special
class of fractional designs that enable the efficient
estimation of all main effects.
Internal validity. This involves correlations of the
predicted evaluations for the holdout or validation
stimuli with those obtained from the respondents.
Conducting Conjoint Analysis
Formulate the Problem
Construct the Stimuli
Decide the Form of Input Data
Select a Conjoint Analysis Procedure
Interpret the Results
Assess Reliability and Validity
Conducting Conjoint Analysis
Formulate the Problem
◼
◼
◼
◼
◼
Identify the attributes and attribute levels to be used in
constructing the stimuli.
The attributes selected should be salient in influencing
consumer preference and choice and should be
actionable.
A typical conjoint analysis study involves six or seven
attributes.
At least three levels should be used, unless the attribute
naturally occurs in binary form (two levels).
The researcher should take into account the attribute
levels prevalent in the marketplace and the objectives
of the study.
Conducting Conjoint Analysis
Construct the Stimuli
◼
◼
◼
In the pairwise approach, also called two-factor
evaluations, the respondents evaluate two attributes at a
time until all the possible pairs of attributes have been
evaluated.
In the full-profile approach, also called multiple-factor
evaluations, full or complete profiles of brands are
constructed for all the attributes. Typically, each profile
is described on a separate index card.
In the pairwise approach, it is possible to reduce the
number of paired comparisons by using cyclical designs.
Likewise, in the full-profile approach, the number of
stimulus profiles can be greatly reduced by means of
fractional factorial designs.
Sneaker Attributes and Levels
Attribute
Number
Level
Description
Sole
3
2
1
Rubber
Polyurethane
Plastic
Upper
3
2
1
Leather
Canvas
Nylon
Price
3
2
1
$30.00
$60.00
$90.00
Full-Profile Approach to Collecting
Conjoint Data
Example of a Sneaker Product Profile
Sole
Made of rubber
Upper
Made of nylon
Price
$30.00
Pairwise Approach to Conjoint Data
Sole
Rubber
U
p
p
e
r
Sole
Polyurethane
Rubber
Plastic
P
r
i
c
e
Leather
Canvas
$30.00
$60.00
$90.00
Nylon
Price
$ 30.00
Leather
U
p
p
e
r
Canvas
Nylon
$60.00
$90.00
Polyurethane
Plastic
Conducting Conjoint Analysis
Construct the Stimuli
◼
◼
A special class of fractional designs, called
orthogonal arrays, allow for the efficient estimation
of all main effects. Orthogonal arrays permit the
measurement of all main effects of interest on an
uncorrelated basis. These designs assume that all
interactions are negligible.
Generally, two sets of data are obtained. One, the
estimation set, is used to calculate the part-worth
functions for the attribute levels. The other, the
holdout set, is used to assess reliability and validity.
Conducting Conjoint Analysis
Decide on the Form of Input Data
◼
◼
◼
◼
◼
For non-metric data, the respondents are typically required
to provide rank-order evaluations.
In the metric form, the respondents provide ratings, rather
than rankings. In this case, the judgments are typically
made independently.
In recent years, the use of ratings has become
increasingly common.
The dependent variable is usually preference or intention
to buy. However, the conjoint methodology is flexible and
can accommodate a range of other dependent variables,
including actual purchase or choice.
In evaluating sneaker profiles, respondents were required
to provide preference.
Sneaker Profiles and Ratings
Attribute Levels
Profile No.
1
2
3
4
5
6
7
8
9
Sole
1
1
1
2
2
2
3
3
3
a
Upper Price
1
1
2
2
3
3
1
2
2
3
3
1
1
3
2
1
3
2
Preference
Rating
9
7
5
6
5
6
5
7
6
Conducting Conjoint Analysis
Decide on the Form of Input Data
The basic conjoint analysis model may be represented by the
following formula:
m
U(X ) = 
i =1
ki
a x
j =1
ij
ij
Where:
U(X)
a ij
xjj
ki
m
= overall utility of an alternative
= the part-worth contribution or utility associated with
the j th level (j, j = 1, 2, . . . ki) of the i th attribute
(i, i = 1, 2, . . . m)
= 1 if the j th level of the i th attribute is present
= 0 otherwise
= number of levels of attribute i
= number of attributes
Conducting Conjoint Analysis
Decide on the Form of Input Data
The importance of an attribute, Ii, is defined in terms of the range
of the part-worths, a ij, across the levels of that attribute:
The attribute's importance is normalized to ascertain its importance
relative to other attributes, Wi:
Wi =
I
I
i
m
i =1
i
m
So that
W i = 1
i =1
The simplest estimation procedure, and one which is gaining in popularity,
is dummy variable regression (see Chapter 17). If an attribute has ki
levels, it is coded in terms of ki - 1 dummy variables (see Chapter 14).
Other procedures that are appropriate for non-metric data include LINMAP,
MONANOVA, and the LOGIT model.
Conducting Conjoint Analysis
Decide on the Form of Input Data
The model estimated may be represented as:
U = b0 + b1X1 + b2X2 + b3X3 + b4X4 + b5X5 + b6X6
Where:
X1, X2 = dummy variables representing Sole
X3, X4 = dummy variables representing Upper
X5, X6 = dummy variables representing Price
For Sole the attribute levels were coded as follows:
Level 1
Level 2
Level 3
X1
X2
1
0
0
0
1
0
Sneaker Data Coded for Dummy
Variable Regression
Preference
Ratings
Y
9
7
5
6
5
6
5
7
6
Sole
X1
X2
1
1
1
0
0
0
0
0
0
0
0
0
1
1
1
0
0
0
Attributes
Upper
X3
X4
1
0
0
1
0
0
1
0
0
0
1
0
0
1
0
0
1
0
Price
X5
X6
1
0
0
0
0
1
0
1
0
0
1
0
1
0
0
0
0
1
Conducting Conjoint Analysis
Decide on the Form of Input Data
The levels of the other attributes were coded similarly. The
parameters were estimated as follows:
b0
b1
b2
b3
b4
b5
b6
= 4.222
= 1.000
= -0.333
= 1.000
= 0.667
= 2.333
= 1.333
Given the dummy variable coding, in which level 3 is the base
level, the coefficients may be related to the part-worths:
a11 - a13 = b1
a12 - a13 = b2
Conducting Conjoint Analysis
Decide on the Form of Input Data
To solve for the part-worths, an additional constraint is necessary.
a11 + a12 + a13 = 0
These equations for the first attribute, Sole, are:
a 11 - a 13 = 1.000
a 12 - a 13 = -0.333
a11 + a12 + a13 = 0
Solving these equations, we get:
a11
a12
a13
= 0.778
= -0.556
= -0.222
Conducting Conjoint Analysis
Decide on the Form of Input Data
The part-worths for other attributes reported in Table
21.6 can be estimated similarly.
For Upper we have:
a 21 - a 23 = b3
a 22 - a 23 = b4
a21 + a22 + a23 = 0
For the third attribute, Price, we have:
a 31 - a 33 = b5
a 32 - a 33 = b6
a31 + a32 + a33 = 0
Conducting Conjoint Analysis Decide
on the Form of Input Data
The relative importance weights were calculated based on ranges
of part-worths, as follows:
Sum of ranges
of part-worths
= (0.778 - (-0.556)) + (0.445-(-0.556))
+ (1.111-(-1.222))
= 4.668
Relative importance of Sole
Relative importance of Upper
Relative importance of Price
= 1.334/4.668 = 0.286
= 1.001/4.668 = 0.214
= 2.333/4.668 = 0.500
Results of Conjoint Analysis
Attribute
Sole
Upper
Price
Level
No.
Description
Utility
3
2
1
Rubber
Polyurethane
Plastic
0.778
-0.556
-0.222
3
2
1
Leather
Canvas
Nylon
0.445
0.111
-0.556
0.214
3
2
1
$30.00
$60.00
$90.00
1.111
0.111
-1.222
0.500
Importance
0.286
Conducting Conjoint Analysis
Interpret the Results
◼
◼
◼
For interpreting the results, it is helpful to plot the
part-worth functions.
The utility values have only interval scale properties,
and their origin is arbitrary.
The relative importance of attributes should be
considered.
Conducting Conjoint Analysis
Assessing Reliability and Validity
◼
◼
◼
◼
The goodness of fit of the estimated model should be
evaluated. For example, if dummy variable regression is used,
the value of R2 will indicate the extent to which the model fits
the data.
Test-retest reliability can be assessed by obtaining a few
replicated judgments later in data collection.
The evaluations for the holdout or validation stimuli can be
predicted by the estimated part-worth functions. The predicted
evaluations can then be correlated with those obtained from
the respondents to determine internal validity.
If an aggregate-level analysis has been conducted, the
estimation sample can be split in several ways and conjoint
analysis conducted on each subsample. The results can be
compared across subsamples to assess the stability of conjoint
analysis solutions.
Part-Worth Functions
0.0
0.0
Utility
-1.0
-1.5
-0.4
-0.8
-1.2
Leather
-2.0
Rubber Polyureth. Plastic
Canvas
Sole
Nylon
0.0
-0.5
Sole
-1.0
Utility
Utility
-0.5
-1.5
-2.0
-2.5
-3.0
$30
$60
Price
$90
Assumptions and Limitations of
Conjoint Analysis
◼
◼
◼
◼
◼
Conjoint analysis assumes that the important attributes of
a product can be identified.
It assumes that consumers evaluate the choice
alternatives in terms of these attributes and make
tradeoffs.
The tradeoff model may not be a good representation of
the choice process.
Another limitation is that data collection may be complex,
particularly if a large number of attributes are involved
and the model must be estimated at the individual level.
The part-worth functions are not unique.