TOPIC 9

MULTIDIMENSIONAL SCALING AND

CONJOINT ANALYSIS

Group 6 presentation
MULTIDIMENSIONAL
SCALING (MDS)
 What is multidimensional
scaling analysis?

A class of procedures for representing perceptions and preferences of

respondents spatially by means of a visual display.
Perceived or psychological relationships among stimuli are represented as
geometric relationships (often called spatial maps) among points in a
multidimensional space.
MDS has been used in Marketing to identify:
The number and nature of dimensions
consumers use to perceive different brands
in the marketplace
The positioning of current brands on these
dimensions
The positioning of consumers’ ideal brand
on these dimensions
Example
A study of the automobile industry found
that consumers perceived cars along two
dimensions: Price and Sportiness.
Four market segmentations: economy
sports cars, luxury sports cars, economy
conservative cars, and luxury
conservative cars.
When to use?
MDS is particularly useful in situations where
you need to visualize and understand the
structure or relationships in your data. It is
suitable in the following scenarios:

Understanding Similarities or Differences

Visualizing High-Dimensional Data
Non-Metric Data
Pattern Discovery
Comparison of Objects or Individuals
Conducting
Multidimensional Scaling
Business question that can be answered
Market segmentation Product Positioning Competitor Analysis Customer Satisfaction
Identify clusters of Determine the Visualize the Map customer
products or brands perceived competitive perceptions to
based on customer similarities or landscape and identify areas of
perceptions, differences between understand where strength and
helping with products, aiding in your brand stands weakness in your
targeted marketing strategic decisions relative to others service or
strategies. about product from the customer
positioning and perspective of experience, guiding
portfolio customers. improvement
management. efforts.
CONJOINT ANALYSIS
Conjoint Analysis
A technique that attempts to determine the
relative importance consumers attach to
salient attributes and the utilities they
attach to the levels of attributes. This
information is derived from consumers’
evaluations of brands, or brand profiles
composed of these attributes and their levels.
 VARIABLE
Dependent variable: ordinal, are the preferences or
choices made by the study participants.

Independent variables: nominal, are being

evaluated. These attributes might include things like
price, quality, features, design, or brand name.
 PURPOSE
Determining the relative importance of attributes in the consumer choice process

Estimating market share of brands that differ in attribute levels.

Determining the composition of the most preferred brand

Segmenting the market based on similarity of preferences for attribute levels

Statistics and Terms Associated with
Conjoint Analysis
Part-worth functions or utility functions describe the utility consumers attach to the levels of each attribute.
The relative importance weights are estimated and indicate which attributes are important in influencing
consumer choice.
Attribute levels: denote the values assumed by the attributes
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 where the respondents evaluate two attributes at a time until all the required pairs of
attributes have been evaluated.
Cyclical designs: are designs employed to reduce the number of paired comparisons.
Fractional factorial designs: are designs employed to reduce the number of stimulus profiles to be evaluated
in the full profile approach.
Orthogonal arrays: are a special class of fractional designs that enable the efficient estimation of all main
effects.
Internal validity: involves correlations of the predicted evaluations for the holdout or validation stimuli with
those obtained from the respondents.
CONDUCTING CONJOINT ANALYSIS
DIFFERENCES BETWEEN MULTIDIMENSIONAL SCALING
AND CONJOINT ANALYSIS
Multidimensional
Conjoint analysis
scaling (MDS)

combinations of attribute
Stimuli products or brands levels determined by the
researcher

develop a spatial map seeks to develop the part-

Goal depicting the stimuli in a worth or utility functions
multidimensional describing the utility
perceptual or preference consumers attach to the
space levels of each attribute
Business question that can be answered
Feature Prioritization Pricing Strategy Brand Positioning Market Share Prediction
Identify the Determine the price Understand the Predict market
relative importance sensitivity of unique value share based on
of different customers and proposition of your different scenarios,
features, helping optimize pricing brand and how it helping in strategic
prioritize based on perceived stacks up against planning.
development value. competitors.
efforts.
 Stage 1: Developing product profiles to collect data on consumers’ preferences
R&D team of a company has been developing a new product whose attributes can be customized
to meet the demands of different consumers. Specifically, the new product has 5 key attributes that
have different options as follows:
 Product package: there are 3 options including A style, B style and C style.
 Brand image: there are 3 options including luxury (LUX), common (COM) and cheap (CHE).
 Price: there are 3 levels of price including 200$, 500$ and 700$.
 Delivery: Consumers can choose home delivery service for this product or buy it at company store.
 Money-back guarantee: Consumers can choose whether to participate in money-back guarantee
program for this product or not.
Before launching this product to the market, marketing team tries to investigate how product
profiles are different in terms of the utility they provide to the consumers. Thus, they need to identify
possible product profiles that will be used to collect consumers’ preference. Help the team to
identify the number of product profile that is feasible to collect data.
 Stage 1: Developing product profiles to collect data on consumers’ preferences
Research Consider all the possible combinations of the 5 key attributes of the product.

1. Product package options: 3 ( A style, B style, C style)

Find the Problem 2. Brand image options: 3 ( Luxury, Common, Cheap)
3. Price options: 3 ( 200$, 500$, 700$)
4. Delivery options: 2 ( Home delivery, Buy at company store)
5. Money-back guarantee options: 2 ( Participate, Not participate)

Calculate Multiply the number of options for each attribute:

Total product profiles = 3 x 3 x 3 x 2 x 2 = 108

Conclusion There are 108 feasible product profiles that can be used to collect data on consumers’
preferences.
Each profile represents a unique combination of the product’s attributes, which marketing
teams can use to gather information about what consumers prefer.
Stage 1: Developing product profiles to collect data on consumers’ preferences
Stage 1: Developing product profiles to collect data on consumers’ preferences

The Card list table is a

summary table that displays
sequential numbers for all
product profiles, while the
Profiles table breaks down
each product profile and
shows details about their
attributes.
 Stage 2: Launching new product to the market
After having the preference data, ranking from the most preferred (1) to the least preferred (xx) (File:
ProductDesign_prefs) for the feasible product profiles (File: ProductDesign_plan), the team needs to
know which product profiles should be offered to the consumers. Help the team to identify the
product profiles that can potentially meet consumers’ demands.

Tip: The syntax is as follows:

CONJOINT PLAN="address where the plan file is located"
/DATA="address where the data file is located"
/SEQUENCE=PREF1 TO PREF22
/SUBJECT=ID
/FACTORS=PACKAGE BRAND (DISCRETE)
PRICE (LINEAR LESS)
DELIVERY (LINEAR MORE)
MONEY (LINEAR MORE)
/PRINT=SUMMARYONLY.
 Stage 2: Launching new product to the market
In which:
• SEQUENCE: Specify the ranking preferences from the most to the least. Must be exact variable
names in the data file and no blanks in between. (PREF1 TO PREF22)
• SUBJECT: Specify the ID of participants in the data file.
• FACTORS: Specify the factors (exact factor names in the data file).
• DISCRETE: DISCRETE is assumed if a factor is not labelled with one of the four alternatives
(DISCRETE, LINEAR, IDEAL, ANTIIDEAL).
• LINEAR: LINEAR is used for the factors indicating that the data are expected to be linearly related
to the preference.
• LESS: LESS indicates an expected direction for the relationship between the factor and
preference: higher preference for lower factor. For instance: higher preference for lower price.
• MORE: MORE indicates an expected direction for the relationship between the factor and
preference: higher preference for higher factor. For instance: Higher preference for higher quality.
THE PROCESS
File => New => Syntax to open Syntax window.
Write the syntax.
Then click Run Selection.
 RESULTS
A. UTILITIES

⇨ This table shows the utility (part-worth) scores

and their standard errors for each factor level.
Higher utility values indicate greater preference.

⇨ As expected, there is an inverse relationship

between price and utility, with higher prices
corresponding to lower utility (larger negative
values mean lower utility)
 RESULTS
B. COEFFICIENTS

⇒ This table shows the linear regression

coefficients for those factors specifies as
LINEAR (for DEAL and ANTIIDEAL
models, there would also be a quadratic
term). The utility for a particular factor
level by the coefficient.
 RESULTS
C. IMPORTANCE VALUES
⇨ The results show that the product package has the
most influence on overall preference. This means that
there is a large difference in preferences between product
profiles containing the most desired packaging and those
containing the least desired packaging.

⇨ The results also show that a money-back guarantee

plays the least important role in determining overall
preference. Price plays a significant role but not as
significant as package design. Perhaps subject, and the
results are then averaged over all of the subjects
 RESULTS
D. CORRELATIONS

⇒ This table displays 2 statistics, Pearson’s R and Kendall’s tau, which provide
measures of the correlation between the observed and estimated preferences.
 RESULTS
E. NUMBER OF REVERSALS

The "number of reversal" is a count of how many

times these inconsistent preferences occur across all
the choice sets or rating tasks in the conjoint analysis.

⇒ A higher number of reversals may indicate that the

respondent's preferences are less stable or more
difficult to interpret (potentially affecting the accuracy
of the conjoint analysis results.)

⇨ This table displays the number of reversals for

each factor and for each subject.
 RESULTS
F. PREFERENCE PROBABILITIES OF SIMULATIONS

This table gives the predicted probabilities of choosing each of the simulation cases as
the most preferred one, under three different probability-of-choice models.
⇒ Across the 10 subjects in this study, all three models indicated that simulation
profile 2 would be preferred.
 Multidimensional Scaling
Identify Direct and Indirect competitors

1. Data description
2. Process on SPSS Visualize the position of competitors
3. Result & Conclusion based on their similarity degree

Identify the strengths and weaknesses

of competitors
 Unichoice Paired Comparison
DATA DESCRIPTION

Purpose: See the similarity between universities in order to identify

the direct and indirect competitors.
Data: distance between the two universities (ratio)
Sample questions: how similar pairs of university are in their
perception (1 – very dissimilar to 5 – very similar).
 Unichoice Paired Comparison
PROCESS ON SPSS

1. Analyze > Scale > PROXSCAL Keep the original setting

Unichoice Paired Comparison
PROCESS ON SPSS

Choose all variables

Unichoice Paired Comparison
PROCESS ON SPSS
 Unichoice Paired Comparison
PROCESS ON SPSS

A sharp bend occurs indicates an

appropriate number of dimensions. Change dimension min and max to 2
 Unichoice Paired Comparison
PROCESS ON SPSS

Output: Common space coordinates; Distance; Multiple stress measure

Plot: Common space
Unichoice Paired Comparison
RESULTS & CONCLUSIONS

S-Stress: 0 < 0.00447 < 0.025

Dispersion Accounted For: 0.99776
Tucker's Coefficient of Congruence: 0.99888

--> Quite perfect fit

 Unichoice Paired Comparison
RESULTS & CONCLUSIONS

Identify Direct and Indirect competitors

Visualize the position of competitors

based on their similarity degree
 Unichoice Likert
DATA DESCRIPTION
Purpose: Find out the strengths of each Factors including:
university, based the factors impacting ·URE: University Reputation
students’ university choice ·LEC: Lecturer
·PRO: Training Program
The students are from six universities (Code)
·COS: Cost
including: FTU, IU, RMIT, OU, UEH and UEL.
·FAC: Facilities
Likert 5-point (1 to 5) scale was used to
·CDE: Career Development
measure all the key attributes
·ENV: Environment
·EBE: Extra Benefits
The numbers presented in the dataset are the
·CON: Consulting services
means of students’ evaluation on each
·INF: Influence of other people
attribute for each university.
·SUC: Satisfaction with the choice
Unichoice Likert
DATA DESCRIPTION
Switch the first column and row
position to make the universities
become variables.
 Unichoice Paired Comparison
PROCESS ON SPSS

1. Analyze > Scale > PROXSCAL Choose “Create proximities from data”
Unichoice Likert
PROCESS ON SPSS
 Unichoice Likert
PROCESS ON SPSS

Output
Common space coordinates
Distance
Multiple stress measure
Plot: Common space
Unichoice Likert
PROCESS ON SPSS

S-Stress: 0 < 0.00484 < 0.025

Dispersion Accounted For: 0.99770
Tucker's Coefficient of Congruence: 0.99885

-> Quite perfect fit

Unichoice Likert
PROCESS ON SPSS

• Both IU and RMIT are quite far, so there

are just a few similarities
• OU and UEL are in the same second
quadrant and quite near, so there are
many similarities between them
• UEH and FTU are in the third quadrant
and the distance between them is the
nearest, so they are the most similar with
each other among three couples
For Your Attention