Perceptual mapping by multidimensional scaling MDS

Embed Size (px)

Citation preview

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    1/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    2/70

    RRRRESEARCHESEARCHESEARCHESEARCHRRRREPORTS INEPORTS INEPORTS INEPORTS INCCCCONSUMERONSUMERONSUMERONSUMERBBBBEHAVIOREHAVIOREHAVIOREHAVIOR

    These analyses address issues of concern to marketing and advertising professionals and to

    academic researchers investigating consumer behavior. The reports present original research

    and cutting edge analyses conducted by faculty and graduate students in the Consumer-

    Industrial Research Program at Cleveland State University.

    Subscribers to the series include those in advertising agencies, market research organizations,

    product manufacturing firms, health care institutions, financial institutions and other

    professional settings, as well as in university marketing and consumer psychology programs.

    To ensure quality and focus of the reports, only a handful of studies will be published each

    year.

    ProfessionalSeries - Brief, bottom line oriented reports for those in marketing and

    advertising positions. Included are both B2B and B2C issues.

    How To Series - For marketers who deal with research vendors, as well as for

    professionals in research positions. Data collection and analysis procedures.

    Behavioral Science Series - Testing concepts of consumer behavior. Academically

    oriented.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    3/70

    AVAILABLE PUBLICATIONS:

    Professional Series

    Lyttle, B. & Weizenecker, M. Focus groups: A basic introduction, February, 2005.

    Arab, F., Blake, B.F., & Neuendorf, K.A. Attracting Internet shoppers in the Iranianmarket, February, 2003. Liu, C., Blake, B.F., & Neuendorf , K.A. Internet shopping in Taiwan and U.S.,

    February, 2003. Jurik, R., Blake, B.F., & Neuendorf, K.A. Attracting Internet shoppers in the

    Austrian market, January, 2003. Blake, B.F., & Smith, L. Marketers, Get More Actionable Results for Your Research

    Dollar!, October, 2002.

    How To Series

    Blake, B.F., Valdiserri, J., Neuendorf, K.A., & Nemeth, J. Validity of the SDS-17measure of social desirability in the American context, November, 2005.

    Blake, B.F., Dostal, J., & Neuendorf, K.A. Identifying constellations of websitefeatures: Documentation of a proposed methodology, February, 2005.

    Saaka, A., Sidon , C., & Blake, B.F. Laddering: A How to do it manual with anote of caution, February, 2004.

    Blake, B.F., Schulze, S., & Hughes, J.M. Perceptual mapping by multidimensionalscaling: A step by step primer, July, 2003.

    Behavioral Science Series

    Shamatta, C., Blake, B.F., Neuendorf, K.A, Dostal, J., &Guo, F. Comparing websiteattribute preferences across nationalities: The case of China, Poland, and the USA,

    October, 2005. Blake, B.F., Dostal, J., & Neuendorf, K.A. Website feature preference constellations:

    Conceptualization and measurement, February, 2005. Blake, B.F., Dostal, J., Neuendorf, K.A., Salamon, C., & Cambria, N.A. Attribute

    preference nets: An approach to specifying desired characteristics of an innovation,

    February, 2005. Blake, B.F., Neuendorf, K.A., Valdiserri, C.M., & Valdiserri, J. The Online

    Shopping Profile in the cross-national context: The roles of innovativeness and

    perceived newness, February, 2005. Blake, B.F., & Neuendorf , K.A. Cross-national differences in website appeal: A

    framework for assessment, July, 2003.

    Blake, B.F., Neuendorf , K.A., & Valdiserri , C.M. Appealing to those most likely toshop new websites, June, 2003.

    Blake, B.F., Neuendorf , K.A., & Valdiserri , C.M. Innovativeness and variety ofinformation shopping, April, 2003.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    4/70

    RRRRESEARCHESEARCHESEARCHESEARCHRRRREPORTS INEPORTS INEPORTS INEPORTS INCCCCONSUMERONSUMERONSUMERONSUMERBBBBEHAVIOREHAVIOREHAVIOREHAVIOR

    EDITORIAL DIRECTOR: DR.BRIAN BLAKE

    Dr. Brian Blake has a wide variety of academic and professional experiences.

    His early career... academically, rising from Assistant Professor to tenured Professor at Purdue

    University, his extensive published research spanned the realms of psychology (especially

    consumer, social, and cross-cultural), marketing, regional science, sociology, community

    development, applied economics, and even forestry. Professionally, he was a consultant to the

    U.S. State Department and to the USDA, as well as to private firms.

    Later on...on the professional front, he co-founded a marketing research firm, Tactical Decisions

    Group, and turned it into a million dollar organization. After merging it with another firm to

    form Triad Research Group, it was one of the largest market research organizations based in

    Ohio. His clients ranged from large national firms (e.g., Merck and Co., Dupont, Land o Lakes)

    to locally based organizations (e.g., MetroHealth System, American Greetings, Progressive

    Insurance, Liggett Stashower Advertising). On the academic side, he moved to Cleveland State

    University and co-founded the Consumer-Industrial Research Program (CIRP). Some of

    Clevelands best and brightest young marketing research professionals are CIRP graduates.

    In the last few years...academically, he is actively focusing upon establishing CIRP as a center

    for cutting edge consumer research. Professionally, he resigned his position as Chairman of

    Triad and is now Senior Consultant to Action Based Research and consultants with a variety of

    clients.

    EDITOR(2003):JILLIAN HUGHES

    Currently a CIRP graduate student, she graduated Magna Cum Laude from Mount Union

    College, where she majored in Psychology, with a focus on Consumer Behavior, and minored in

    Sociology. Among her many research interests; she focuses on Internet buying behavior, and the

    effects of Social Desirability Bias on Innovativeness Scales. She had the honor of presenting

    research concerning age differences in brand labeling at the Ohio Undergraduate Psychology

    Conference in April of 2001 at Kenyon College. She also presented another piece of original

    research on Internet buying behavior of college students at the Interdisciplinary Conference for

    the Behavioral Sciences hosted by Mount Union College in April, 2001.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    5/70

    Forward

    Perceptual Maps are widely used by market researchers, e.g., to portray a brands

    image or consumers reactions to product features. Although the major statistical

    techniques have been available for several decades, these are still many questions about

    those techniques among practicing professionals.

    This report is intended for on the job professionals who are fairly unfamiliar

    with the concrete procedures used to generate and to interpret such maps

    In overview, three types of maps are especially popular among professional

    researchers:

    perceptual maps that identify the images of brands, products, services, etc.

    preference maps that estimate differences among segments or individuals in

    the appealorattractiveness of brands, products, services, features.

    hybrid maps which portray both images and appeal.

    A variety of statistical techniques can be used to generate each type of map.

    Perceptual maps are usually constructed via multidimensional scaling - multiple

    discriminant function correspondence analysis. Preference maps are typically

    developed by a form of multidimensional unfolding. Hybrid maps are composed

    by first devising a perceptual map and then introjecting preferences as ideal points

    or as vectors.

    This paper focuses upon one mapping technique, multidimensional scaling

    (MDS), and executes it via a program package that is widely used by market

    researchers, SPSS.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    6/70

    Abstract

    The overall objective of this report was to document the applications of two

    widely applied forms of perceptual mapping, Classical and Weighted Multidimensional

    Scaling. A step-by-step guide is provided for the use of these mapping techniques. It is

    anticipated that this report will be valuable to the professional market researcher who is

    new to perceptual mapping and to others looking for a detailed reference source for

    performing the basics of these techniques.

    The illustrative data pertain to the images of particular hospitals in the Northeast

    Ohio area. The data were gathered from a convenience sample of family, friends, and

    acquaintances of the researchers. A total of 107 took part in the study.

    The techniques were Classic Multidimensional Scaling (CMDS) and Weighted

    Multidimensional Scaling (WMDS). The statistical software program SPSS was used,

    but the ideas can be generalized to other statistical packages and programs.

    I. Overview of the Three Mapping Procedures

    Before describing each technique in detail, let us present them in overview.

    1) Classic Multidimensional Scaling (CMDS)

    To begin, the data for CMDS and WMDS are indicators of the degree of

    similarity among objects, brand names, etc. Here, the data are ratings of the degree of

    perceived similarity among the twelve stimuli- four hospitals identified by name, four

    unnamed hospitals described by taglines, and four unnamed hospitals described by

    advertisements.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    7/70

    The four named hospitals in our illustration were Fairview Hospital, Parma

    Community Hospital, Southwest General Hospital, and MetroHealth System. The four

    taglines were 1) In the hands of doctors 2) Its a people thing 3) When its your

    health, experience counts and 4) Your partner in good health. For this report, let us

    abbreviate them as 1) doctors hands 2) people thing 3) experience counts and 4)

    partner. The four advertisements were labeled 1) magic bullet 2) heart surgery 3)

    diet and exercise and 4) heart center. The advertisements are shown on pages 3-6.

    Respondents rated the similarity of pairs of these items on a 0-10 point scale, higher

    numbers meaning more similarity. For the CMDS example on the next page, we consider

    just the four named hospitals and the four hospitals described by advertisements.

    CMDS analyzed ratings to produce the positioning map on page 7. The goal of

    mapping is to portray the respondents perceptions of similarity among the items along a

    given number of dimensions or yardsticks. The closer together on the map are the items,

    the

    more similar they are perceived to be. In the figure on the next page, the following

    abbreviations are used for the named hospitals: fairview (Fairview Hospital), parmacom

    (Parma Community Hospital), swgener (Southwest General Hospital), and metro

    (MetroHealth System). The following abbreviations were used for the hospitals

    described by advertisements: magbull (magic bullet), hrtsurg (heart surgery),

    dietexe (diet and exercise), and hrtcent (heart center).

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    8/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    9/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    10/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    11/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    12/70

    Here, the hospital described in the advertisement Heart Center is perceived to be

    similar to Parma Community Hospital. However, the advertisement Heart Center is

    seen as quite distinct from MetroHealth System. Thus, respondents as a group perceived

    that the advertisement Heart Center fits Parma Community Hospital better than

    MetroHealth System.

    Among other uses, this kind of mapping can identify how well an advertisement

    or tagline can fit a hospital, or even be incompatible with a particular brand name.

    2) Weighted Multidimensional Scaling (WMDS)

    In Weighted Multidimensional Scaling (WMDS), we are again considering the

    perceived similarity among stimuli, but here we identify differences in perception among

    specific segments of individuals. WMDS calculates the differences among the groups of

    respondents on a given number of dimensions. Each dimension is essentially the same

    Classic Multidimensional Scaling CMDS

    Hospitals by Advertisements

    Dimension 1

    1.51.0.50.0-.5-1.0-1.5-2.0

    Dim

    ension2

    2.0

    1.5

    1.0

    .5

    0.0

    -.5

    -1.0

    -1.5

    swgener

    parmacom

    metro

    fairview

    hrtcent

    dietexe

    hrtsurg

    magbull

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    13/70

    for all segments (e.g. world-wide reputation). In contrast, if we were to do a CMDS

    separately for each segment, we could conceivably develop maps in which a dimension in

    one group would have to be interpreted as different from the dimension found in other

    groups (e.g. personal attention, location close by).

    Respondents rated the same 12 objects (4 hospitals/ 4 advertisements/ 4 taglines)

    by similarity for selected pairs of these items. Also, WMDS shows the importance of a

    dimension to a particular segment when that segment perceives the stimuli in question.

    One map is produced for each segment. The closer together the stimuli are on the map,

    the more similar is the weight assigned to an object on a particular dimension. WMDS

    analyzed these ratings to produce the positioning map illustrated on the next page. The

    data were separated into male and female segments, and the maps can be seen on the next

    two pages.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    14/70

    Magic BulletHeart Surgery

    Diet and Exercise

    Heart Center

    Fairview

    Parma Community

    SW General

    Metrohealth

    WMDS Hospitals by Ads (Males)

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    15/70

    Magic Bullet

    Heart Surgery

    Heart Center Parma Community

    Diet and Exercise

    SW General

    Fairview

    Metrohealth

    WMDS Hospitals by Ads (Females)

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    16/70

    In the male WMDS map, the hospital described in the advertisement Heart

    Center is perceived to be similar to Parma Community Hospital in the eyes of males.

    However, the advertisement Heart Center is perceived to be much different from

    Southwest General Hospital. Therefore, male respondents as a group perceive that the

    advertisement Heart Center fits Parma Community Hospital better than Southwest

    General Hospital.

    This kind of perceptual mapping can aid in determining how appropriate or

    inappropriate an advertisement or tagline can be for a hospital depending on the target

    market segment of interest. Furthermore, this mapping technique can identify the most

    appropriate advertisement or tagline for a particular brand name in general.

    The information used for more detailed explanation in subsequent sections of this

    paper was drawn from:

    Young and Harris (19XX), Chapter 7 Multidimensional Scaling.

    Hair, Anderson, Tatham, and Black (1998), Multivariate Data Analysis,

    5th

    Ed., Chapter 10.

    Meyers (1996), Segmentation and Positioning for Strategic Marketing

    Decisions. Perceptual Positioning Maps, Chapter 8.

    Lecture notes from Professor Brian Blakes Advanced Consumer Research

    course PSY 620.

    George and Mallery (2001), SPSS for Windows Step by Step: A Simple

    Guide and Reference 10.0 Update 3rd

    Ed., Multidimensional Scaling

    Chapter 19.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    17/70

    II. CLASSIC MULTIDIMENSIONAL SCALING (CMDS)

    A. Goals/ Objectives of CMDS

    A researcher can use CMDS to reach two general goals or objectives. First,

    CMDS can estimate the relative importance of the dimensions that respondents use to

    judge the degree of similarity or dissimilarity among the stimuli. Second, the degree of

    similarity among all of the stimuli on those dimensions can be assessed. To empirically

    understand or to label the nature of a given dimension requires analyses above and

    beyond CMDS itself.

    B. CMDS General Rationale

    Classic Multidimensional Scaling (CMDS) is a statistical technique created to

    transform data indicating the degree of rated similarity or dissimilarity of objects to

    scores indicating distances among the objects. Then, a map is constructed showing the

    distances among the objects. Objects closer together on the map are perceived as more

    similar and objects further apart are perceived as more dissimilar. The same unit of

    measurement is used for all of the distances among the objects. One matrix of data is

    used, displaying the perceptions of one person or the average persons answers in the

    group of respondents in question.

    C. Alternatives/ Options

    CMDS can produce perceptual maps that portray disaggregated results (which

    show evaluations of a single individual) or aggregated results (which show the combined

    assessments of many individuals). If a single respondents evaluations are desired, the

    researcher should input the respondents judgments into a matrix where the units used for

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    18/70

    comparison are listed across the columns and rows as shown in Table 1 on the next page.

    On the contrary, if cumulative respondent answers are desired, the average or mean

    evaluation can be computed in SPSS under the descriptives option. The researcher then

    inputs the resulting mean scores into a matrix.

    In this matrix, the hypothetical data below are the similarity ratings between the

    two items in each pair. The higher the number, the more similar respondents perceived

    the two items to be.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    19/70

    A B C D E F G H

    A 0

    B 6 0

    C 8 10 0D 5 6 9 0

    E 3 8 10 4 0

    F 5 6 3 7 5 0

    G 4 7 6 9 2 3 0

    H 6 6 10 7 4 8 2 0

    TABLE 1

    In the resulting positioning map, CMDS presents the distances between objects on

    the dimensions along which the distances are calculated. The researcher must specify

    the number of dimensions, which is usually two or three for ease of interpretation.

    CMDS doesnt directly label these dimensions and the researcher must indirectly

    estimate names for the dimensions based, for example, on the rank order of the objects on

    a given dimension.

    Specifically, each unit or object is illustrated by a plotted location in

    multidimensional space on a positioning map. These plotted locations illustrate the

    similarities that are perceived by respondents. CMDS determines the distances among all

    combinations of pairs of the objects and plots the objects accordingly. Briefly, the

    perceptual space is usually generated by a 2 or 3 dimensional Euclidean model. The

    Euclidean geometrical representation is a multidimensional generalization of the

    Pythagorean theorem we all learned in high school geometry class. The formula is: the

    squared distance between the two points on Dimension Xplus the squared distance

    between the two points on Dimension Y equals the squared straight line distance between

    the two points. For example, we can figure the Euclidean straight-line distance on two

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    20/70

    dimensions between A and C if we know their coordinates on the X and Y axes

    (dimensions). As shown below, if the plotted coordinates (locations) for A are 1 on X

    and 3 on Y, and Cs coordinates are 4 on X and 2 on Y, then the distance between A & C

    can be calculated using the formula. The difference between A and C on Dimension 1

    for X is 4-1=3 and on Dimension 2 for Y is 3-2=1. Therefore, the Euclidean distance

    formula would be (1)2 + (3)2 = 10 or 3.33. The Euclidean distance between A and C is

    3.33.

    D. Illustrative Case

    Twelve presentation boards were presented to respondents representing twelve

    different hospitals. The respondents were informed that the hospitals represented may or

    may not be located in the Northeast Ohio area, and that they may or may not be familiar

    with the hospitals portrayed. Four of the hospitals were described by taglines, four of the

    hospitals were described by advertisements, and four of the hospitals were described by

    their respective hospital names.

    Euclidean Distance

    0

    0.5

    1

    1.5

    2

    2.5

    3

    3.5

    0 1 2 3 4 5

    X

    YA

    C

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    21/70

    First, respondents were asked about their familiarity with a tagline. The

    respondents were instructed to assume that the tagline was an accurate description of the

    hospital represented. An illustrative question stated, Now, focus on tagline In the hands

    of doctors. What do you know about this tagline? The respondents options were: A)

    I have never seen it before, and know nothing about it, B) I have seen it before, but dont

    know much about it, C) I have seen it before, and know quite a bit about it, or D) I have

    seen it before, and know that its a tagline for _______ hospital. This questioning was

    repeated for the other three taglines, which again were Its a people thing, When its

    your health, experience counts, and Your partner in good health.

    Respondents rated a total of six pairs of the four taglines. Ratings were made on a

    scale of 0-10, where 0 indicated very different and 10 indicated very similar. The

    interviewer presented the taglines and a question example stated, Rate how similar/

    dissimilar are the two hospitals described by the two taglines. Use a scale of 0 to 10, with

    0 meaning very different and 10 meaning very similar. The first two hospitals are the

    ones described by In the hands of doctors and the one described by Its a people

    thing.

    Next, respondents were asked about their familiarity with an advertisement. The

    respondents were instructed to assume that the advertisement was an accurate description

    of a hospital. An illustrative question stated, Now, focus on advertisement that was

    labeled magic bullet. What do you know about this ad? The respondents options

    were: A) I have never seen it before, and know nothing about it, B) I have seen it

    before, but dont know much about it, C) I have seen it before, and know quite a bit about

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    22/70

    it, or D) I have seen it before, and know that its a advertisement for _______ hospital.

    This questioning was repeated for the other three advertisements.

    Then, respondents rated a total of six pairs of the four advertisements. Ratings

    were made using the same scale of 0-10, where 0 indicated very different and 10

    indicated very similar. The interviewer then presented the advertisements and a question

    example stated, Rate how similar/ dissimilar are the two hospitals described by the two

    advertisements. Use a scale of 0 to 10, with 0 meaning very different and 10 meaning

    very similar. The first two hospitals are the ones described by the advertisements labeled

    magic bullet and people thing.

    Finally, hospital names were presented to respondents to assess what the

    respondents knew about the hospital names. An illustrative question stated, What do

    you know about Parma Community Hospital? A respondents options were: A) I know

    practically nothing about it, B) I have heard the name, but dont know much about it, C) I

    have heard the name and know a little about it, or D) I have heard the name and know

    quite a bit about it.

    Next, respondents rated a total of six pairs of the four hospital names. Ratings

    were made on the same scale of 0-10, again where 0 indicated very different and 10

    indicated very similar. The interviewer presented the hospital names and a question

    example stated, Rate how similar/ dissimilar are the two hospitals named. Use a scale of

    0 to 10, with 0 meaning very different and 10 meaning very similar. The respondent

    was then presented with each of the six pairs of hospitals: Fairview & MetroHealth

    System, Fairview & Parma Community, etc. This questioning was repeated for the

    other three hospital names.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    23/70

    Respondents were then asked to indicate the degree of similarity or dissimilarity

    of hospitals when taglines were paired with hospital names. The same 0-10 similarity

    scale was employed. There were 16 pairs of hospitals and taglines. An illustrative

    question stated, Rate how similar/ dissimilar are the two hospitals using a scale of 0 to

    10, with 0 meaning very different and 10 meaning very similar. The respondent was

    then presented with the pairs doctors hands and Fairview, etc.

    Respondents were then asked to indicate the degree of similarity or dissimilarity

    of advertisements paired with hospital names using the same 0-10 scale. There were 16

    pairs of hospitals and advertisements. An illustrative question stated, Rate how similar/

    dissimilar are the two hospitals. Use a scale of 0 to 10 with 0 meaning very different and

    10 meaning very similar. The respondents were then given the pairs magic bullet &

    Fairview, etc.

    E. Data Needed

    The data analyzed in CMDS are displayed in a single matrix showing the degree

    of rated dissimilarity between hospitals in a pair. The matrix consists of rows and

    columns listing each object paired with all other objects. The measurement level,

    shape, and conditionality of the data can vary among various CMDS analyses and

    must be specified for any MDS to be conducted.

    A) Measurement level-- The data can be ordinal, interval, or ratio.

    1) Ordinal data arranges objects in a rank order from high to low on a

    dimension.

    2) Interval data pertains to numbering in which one number is a fixed

    amount more or less than another number. For example, the amount of

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    24/70

    the dimension (here, similarity) between the number 3 and 4 on a scale

    is assumed to be the same as the amount between 4 and 5, and that

    amount is the same as between 5 and 6 on the scale. In other words,

    an interval scale is one in which the intervals between consecutive

    numbers are equal.

    3) Ratio data has a true zero point (absence of the factor) and allows one

    to calculate ratios or proportions. One cannot do this with an interval

    scale because there is no zero starting point for the numbers.

    However, a ratio scale has a true zero point, and so one can calculate

    ratios, such as one item is twice as much as another.

    B) Shape-- The shape of the data for CMDS is square, where rows and

    columns in the matrix represent the same set of units or objects.

    C) Conditionality-- Data used in CMDS are typically matrix conditional

    indicating that in the analysis all numbers in the data matrix can be compared

    to each other, no matter what the row or column involved. The data matrix is

    devised by averaging (mean) similarity ratings for all respondents for a given

    pair of stimuli. In the illustrative case, this yields an 8 X 8 matrix.

    By default, SPSS assesses higher numbers as more dissimilar, so it was necessary

    to recode our values into higher numbers representing more dissimilarity and lower

    numbers representing more similarity. Keep in mind that if the questionnaire uses higher

    numbers to indicate dissimilarity and lower numbers to indicate similarity, then it is not

    necessary to recode the values.

    F. SPSS Specific Steps

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    25/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    26/70

    and are listed here only for explanatory purposes. These options include the S-Stress

    convergence, which represents the lowest possible level of improvement that the

    algorithm will configure. The criteria box also specifies the minimum S-Stress value,

    which by default is .001 and the maximum number of iterations that the program will

    attempt in finding the best possible solution for the data. The default for the latter is 30.

    G. SPSS Output

    After analyzing the ratings and specifying separate 2 and 3 dimensional solutions

    for both taglines by hospitals and advertisements by hospitals, a 3 dimensional solution

    was found to have the best fit. However, a 2 dimensional solution is shown for

    illustrative purposes because a two dimensional solution is used far more frequently in

    professional practice. It is important to note that the fit of the 2 dimensional solution

    obtained in this data set would be too low to be considered useful in practice.

    The output generated in SPSS for a CMDS solution provides an abundance of

    information. First, the SPSS output provides the unscaled means of the similarity ratings

    of the objects among the aggregated respondents. The ratings are shown in SPSS Output

    1.Each unit or object used for the paired comparisons is shown on the left-hand side as

    rows numbered 1-8 and across the top of the matrix as columns numbered 1-8.

    Raw (unscaled) Data

    1 2 3 45

    1 Magic Bullet .0002 Heart Surgery 4.690 .0003 Diet & Exercise 4.440 4.020 .0004 Heart Center 4.740 3.710 5.920 .0005 Fairview General .000 5.120 4.230 4.040

    .000

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    27/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    28/70

    Iteration history for the 2 dimensional solution (insquared distances)

    Young's S-stress formula 1 is used.

    Iteration S-stress Improvement

    1 .372952 .34050 .032463 .33447 .006034 .33378 .00069

    Iterations stopped becauseS-stress improvement is less than

    .001000

    Stress and squared correlation (RSQ) indistances

    RSQ values are the proportion of variance of the scaleddata (disparities)

    in the partition (row, matrix, or entire data)which

    is accounted for by their correspondingdistances.

    Stress values are Kruskal's stress formula 1.

    For matrixStress = .20196 RSQ = .75785

    SPSS OUTPUT 2The display of the stimulus coordinates on each dimension is provided next. The

    coordinates of each object are the coordinates used to create the plots in the map.

    Configuration derived in 2 dimensionsStimulus Coordinates

    Dimension

    Stimulus Stimulus 1 2Number Name

    1 MAGIC BULLET 1.1437 .74852 HEART SURGERY -1.3325 -.7954

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    29/70

    3 DIET & EXERCISE .9450 -1.25164 HEART CENTER -.8620 1.11935 FAIRVIEW .8389 .89956 METROHEALTH -1.3411 -.72827 PARMA COMMUNITY .9433 -1.0970

    8 SW GENERAL -.3353 1.1048SPSS OUTPUT 3

    In SPSS Output 4 on the next page, the CMDS procedure presents a matrix of the

    optimally scaled data for subject 1 (the aggregated respondents) in the aggregated

    matrix. The data reflect the original ratings of respondents considered as a group. These

    are the distances among the hospitals and advertisements in two-dimensional space.

    Optimally scaled data (disparities) forsubject 1

    1 2 3 45

    1 Magic Bullet .0002 Heart Surgery 2.137 .0003 Diet & Exercise 1.033 1.033 .0004 Heart Center 3.370 3.010 3.280 .0005 Fairview General 2.843 2.651 2.330 2.715

    .000 6 MetroHealth 1.842 1.996 1.791 1.3933.203

    7 Parma Comm. 2.997 2.997 2.959 4.2302.060

    8 SW General 1.996 2.728 2.407 2.8051.816

    6 7 8

    6 MetroHealth .0007 Parma Comm. 3.601 .0008 SW General 3.010 1.945 .000

    SPSS OUTPUT 4

    The perceptual map is then presented and shown on the next page in SPSS Output

    5. The interpretation of this perceptual map indicates that respondents perceive the

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    30/70

    advertisement diet & exercise to fit Southwest General Hospital. On the contrary,

    respondents perceive the advertisement diet & exercise to be quite distinct from

    Fairview.

    The researcher should estimate the nature of the two dimensions. The

    dimensions can be interpreted as yardsticks or criteria people use to judge the similarity

    of the items. Respondents may differentiate the hospitals/ advertisements in regard to

    where the hospitals are located, the quality of the care, the prestige of the hospital, etc.

    We can develop a feel for the nature of a dimension by looking at where the hospital is

    located on a dimension. Other and better ways of labeling dimensions involve more

    complex statistical procedures beyond the goals of this report.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    31/70

    Magic Bullet

    Fairview

    Parma Community

    Diet and Exercise

    Metrohealth

    Heart Surgery

    SW General

    Heart Center

    CMDS Hospitals by Ads

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    32/70

    Finally, the CMDS Output 6 provides the scatterplot of fit between the scaled

    input data (horizontal axis) against the distances (vertical axis). That is, this diagram

    represents the fit of the distances with the data. It is important to examine the scatter of

    the objects along a perfect diagonal line running from the lower left to the upper right to

    assess the fit of the data to the distances. Ideally, when there is a perfect fit, the

    disparities and the distances will show a straight line of points. As the points diverge

    from the straight line, the fit or accuracy of the map decreases. When stress levels are

    very low, the points are close to the straight line. The worse the fit (and the higher the

    stress), the more the points diverge from the straight line. In SPSS Output 6, the

    scatter of the objects shows that the objects are not a very good fit.

    SPSS OUTPUT 6

    Scatterplot of Linear Fit

    Euclidean distance model

    Disparities

    3.02.52.01.51.0.5

    Distances

    3.5

    3.0

    2.5

    2.0

    1.5

    1.0

    .5

    0.0

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    33/70

    Besides looking at statistical indicators of fit, one should eyeball the matrix of

    raw (unscaled) input data against the perceptual maps distances. It is important to

    ensure that the input data match the resulting perceptual map, especially for critically

    important objects (e.g. an advertisement under evaluation, the client hospital, etc.). For

    example, looking again at the CMDS map on page 24 and the raw data matrix below, if

    the client hospital is Southwest General Hospital, the closest advertisement is diet &

    exercise. On the contrary, the advertisement magic bullet is farther away from

    Southwest General Hospital. The input data should convey the same message through

    the numbers. Therefore, the two matrices should be juxtaposed together to verify that

    both show the same pattern.

    MagicBullet

    HeartSurgery

    Diet &Exe.

    HeartCenter

    Fairview

    Metro

    SWGeneral

    Parma

    MB 0

    HS 4.69 0

    DE 4.44 4.02 0

    HC 4.74 3.71 5.92 0

    Fair 0 5.12 4.23 4.04 0

    Met 5.12 0 5.43 4.97 4.06 0SWG

    4.23 5.43 0 4.14 4.96 4.96 0

    Par 4.04 4.97 4.14 0 4.18 4.75 4.50 0

    CMDS Summary

    In summary, the use of CMDS has both advantages and disadvantages to the

    researcher.

    First, it is especially useful in finding unique brand images and distinctive

    product concepts.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    34/70

    Second, it is easy to determine the fit, or lack of fit, of advertisements to

    brands.

    Third, CMDS can also identify the competitors of a brand (if by

    competitor, we mean a brand perceived to be comparable).

    Fourth, it is relatively simple to understand the output.

    Overall, CMDS shows the uniqueness of an object based on specific dimensions, which

    represent distinguishing attributes.

    However, it can also present obstacles for the researcher.

    First, the researcher doesnt know the nature of the dimensions unless

    additional analyses are conducted to label the dimensions.

    Second, CMDS does not directly show any differences in individual

    respondents or segments because it aggregates everyone.

    Third, the program you are using may not show the goodness of fit for a

    single stimulus object, although it estimates for the objects as a group.

    Fourth, it does not inform the researcher whether differing from another

    brand in the set is good or bad for the brands image because CMDS does

    not incorporate respondents preferences into the map.

    Fifth, there is a problem of actionability. In many applications, it cannot

    be the sole guide to strategy because it does not provide information on

    how to change a brands image.

    III. Weighted Multidimensional Scaling (WMDS)

    A. General Rationale

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    35/70

    WMDS is based upon CMDS, but extends the simpler CMDS to allow for

    individual segment differences.

    WMDS generates a group space, a mapping that pertains in general to

    all individuals/ segments. The group space (or common space) does not

    show the uniqueness of a specific individual/ segment.

    Separate spaces (maps) are produced for each individual or segment.

    The group space mapping is adjusted (through stretching or shrinking of

    the dimensions) in an attempt to capture the uniqueness of the judgments

    of each individual/ segment.

    The more an individual/ segment is estimated to differentiate among

    objects on a given dimension, the more important is that dimension

    assumed to be to that individual/ segment.

    The spaces (maps) for the various individuals/ segments must have the

    same dimensions. That is, the rank order of objects on a given dimension

    (e.g. Dimension 1) is the same for each individual/ segment. So, for

    example, Fairview is the highest of all the hospitals/ ads/ taglines on

    Dimension 1. In the map of each and every individual/ segment, it will be

    the highest on Dimension 1. The maps of the various individuals/

    segments differ, though, in how much the objects are spread out on a

    dimension. For example, Fairview may be higher than Parma Community

    on Dimension 1 in all maps. But the distance between the two hospitals

    on Dimension 1 may be very small for one individual/ segment and be

    very great for another individual/ segment.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    36/70

    WMDS can be run for individuals, in which case a separate data matrix is

    required for each individual. Or, more popularly, WMDS can consider

    differences among preselected segments. Individuals can be grouped

    together based on a wide variety of factors.

    For simplicity in our example, respondents are grouped together based on their

    gender. Persons can also be grouped together based on comparability of their individual

    level maps. The latter would be a three phase analysis: (a) do a WMDS, in which each

    individual is treated separately; (b) in the resulting solution, group together those persons

    who have similar maps into a reasonable number of segments; (c) do a second WMDS

    assessing differences among the segments. In keeping with the primer goals of this

    report, we only note this application in passing.

    B. Data Needed

    Demographics and general background questions were asked in the questionnaire.

    These questions pertained to respondent age, income, level of education, adults in

    household, gender, and ethnicity. One of these options can be used to separate into

    segments. WMDS can place more or less weight on the variable (for example, income)

    depending on the goals/ objectives of the analysis. We used gender as the criterion to

    divide the respondents into segments for the WMDS solution. The data needed for

    WMDS (measurement level, shape, and conditionality) is the same as the data needed for

    CMDS.

    C. SPSS Specific Steps

    The analysis proceeded in the following steps. First, averaging across all males,

    the mean for each paired comparison was entered into each cell of the first matrix. Again,

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    37/70

    the data below are hypothetical and represent the dissimilarity ratings between each pair

    combination. The higher the number, the more dissimilar respondents perceived the two

    items to be. The data in SPSS for the second matrix of the next segment (females) should

    begin immediately following the conclusion of the preceding matrix as shown below.

    A B C D E F G H

    A 0

    B 5 0

    C 6 3 0

    D 3 2 6 0

    E 10 1 4 2 0

    F 9 4 5 4 2 0G 8 5 10 3 8 7 0

    H 7 8 8 6 5 4 5 0

    A 0

    B 2 0

    C 5 9 0

    D 6 8 7 0

    E 3 3 8 4 0

    F 2 3 6 6 4 0

    G 5 6 4 8 6 6 0

    H 7 6 9 2 5 2 6 0

    The first step in conducting a WMDS analysis is under theAnalyze option. The

    steps are exactly the same as above if you were conducting a CMDS solution; however,

    the only difference is under theModeltab, in which Individual Differences Euclidean

    Distance should be selected. Also, under the options tab, the researcher should specify

    group plots, the data matrix, and the model and options summary. Individual subject

    plots need not be selected, as it would be in a CMDS solution. Individual subject plots

    show separate plots of each subjects data transformation for ordered categorical (ordinal)

    data only.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    38/70

    D. SPSS Output

    Again, after analyzing the ratings and specifying separate 2 and 3 dimensional

    solutions for advertisements by hospitals, the data in the two matrices was found to have

    the best fit with a 3 dimensional solution. However, again we present a 2 dimensional

    solution for illustrative purposes. The fit of the 2 dimensional solution would not be used

    in practice because of the high stress and the low Pearson R correlation.

    The output created by SPSS for a WMDS solution first shows the iteration history

    of the solution. Youngs S-Stress formula, Kruskals stress formula, and the R squared

    correlation are shown below in SPSS Output 8 for the WMDS solution.

    Iteration history for the 2 dimensional solution (insquared distances)

    Young's S-stress formula 1 is used.

    Iteration S-stress Improvement

    0 .273081 .272512 .25375 .018763 .25254 .001224 .25236 .00017

    Iterations stopped becauseS-stress improvement is less than

    .001000

    Stress and squared correlation (RSQ) indistances

    RSQ values are the proportion of variance of the scaled

    data (disparities)in the partition (row, matrix, or entire data)

    whichis accounted for by their corresponding

    distances.Stress values are Kruskal's stress formula 1.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    39/70

    Matrix Stress RSQ Matrix StressRSQ

    1 .243 .644 2 .197.777

    Averaged (rms) over matricesStress = .22105 RSQ = .71037

    SPSS OUTPUT 8

    In SPSS Output 8, the program provides the separate S-Stress and R Squared

    values for both male and female matrices as well as the combined S-Stress and R Squared

    value, which represents the average subject. In our output, it is shown that the Stress

    value indicates a bad solution. The R Squared value is also low. Again, we use this

    example only for illustration. In practice, we would want a lower stress value and a

    higher R Squared value.

    Next, the display of the stimulus coordinates on each dimension is provided in

    SPSS Output 9 below. The coordinates of each object are the coordinates used to create

    plots in the combined map, or group space.

    Configuration derived in 2 dimensionsStimulus Coordinates

    Dimension

    Stimulus 1 2Name Number

    1 MAGIC BULLET -.3858 1.12612 HEART SURGERY -.6418 1.00863 DIET & EXERCISE -.3909 .91224 HEART CENTER -1.0139 -1.86315 FAIRVIEW 1.0387 -.77986 PARMA COMMUNITY -1.1670 -.34137 SW GENERAL 1.7995 .55378 METROHEALTH .7612 -.6164

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    40/70

    SPSS OUTPUT 9

    The combined WMDS solution (group space) for both segments, male and

    female, is provided next. The computer algorithm uses the group space coordinates to

    plot the stimuli accordingly. Whatever dimensions both male and females use to

    differentiate among the stimuli are shown in SPSS Output 10 on the next page.

    This group space can be interpreted in the same manner as the CMDS, (i.e. in

    terms of what points are close (similar) to an ad or to a hospital).

    Keep in mind that this group space is derived by giving equal emphasis to each

    segment. Thus, both male and female segments have equal say in devising the group

    space. Even though there may be more males or more females in the sample, by devising

    just two matrices (produced by averaging across everyone with the same gender), the

    WMDS gives equal weight to the two matrices. If we would want to give more weight

    (emphasis) to one gender rather than to another, we would have to do the additional step

    of weighting the two matrices.

    Hence, the group space in WMDS is not the same solution (map) as one would

    generate by combining all respondents, male and female, into one matrix and then doing

    a CMDS. The CMDS will give more emphasis to whatever gender has the larger sample.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    41/70

    SPSS OUTPUT 10

    The derived subject weights are plotted in SPSS Output 11. The program plots

    the weights according to their location on the two dimensions. The space or map is

    computed for each segment or individual. This is done by combining the subject (i.e.

    individual/ segment) weights and the coordinates of the items in group space.

    WMDS Euclidean Distance

    Males and Females Group Space

    Dimension 1

    2.01.51.0.50.0-.5-1.0-1.5D

    imension2

    1.5

    1.0

    .5

    0.0

    -.5

    -1.0

    -1.5

    -2.0

    metro

    swgener

    parmacom

    fairvw

    hrtcent

    dietexehrtsurgmagbull

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    42/70

    SPSS OUTPUT 11

    Next, the subject weights and weirdness index are provided by SPSS. Subject

    weights measure the importance of each dimension to each individual/ segment. The

    higher the weight, the more stretched out is a dimension; the smaller the weight

    conversely, the more is that dimension shrunk for that individual/ segment. The

    weirdness index reflects the atypicality of an individual/ segments space. It has values

    between 0 and 1, where subjects with weights that are proportional to the average weights

    has a weirdness of 0. A subject with one large weight and many low weights has a

    weirdness near 1. A subject with exactly one positive weight also has a weirdness of 1.

    The subject weights and weirdness index can be seen in SPSS Output 12.

    Subject WeightsDimension

    Subject Weirdness 1 2

    1 .0759 .6155 .5146

    WMDS

    Subject Weights

    Dimension 1

    .74.72.70.68.66.64.62.60D

    imension2

    .52

    .51

    .50

    .49

    .48

    2

    1

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    43/70

    2 .0714 .7347 .4871Overall importance of

    each dimension: .4594 .2510

    SPSS OUTPUT 12

    Using the stimulus coordinates in group space (Output 9) and the subject weights

    (Output 12), Euclidean distance was figured by multiplying the square root of the weight

    for each segment in weight space by the stimulus coordinate location on each

    dimension from their shared common space. For example, the square root of the weight

    for males on Dimension 1 was .6155

    and on Dimension 2 was .5146. These weights

    were multiplied by the stimulus coordinate location for Magic Bullet on both dimensions,

    which were -.3858 & 1.1261. Therefore, (.6155*-.3858)

    = -.3027 was the male

    segment space for Magic Bullet on dimension 1 and (.5146*1.1261)

    = .8078 was the

    male segment space for Magic Bullet on dimension 2. This procedure was repeated for

    all stimuli for males. Next, the female segment was figured in the same manner.

    However, the female weights for dimensions 1 and 2 were .7347

    and .4871. The

    following table on the next page was found for each of the segments.

    Males Dimension1

    Dimension2

    Females Dimension1

    Dimension2

    Magic Bullet -.3027 .8078 Magic Bullet -.3307 .7859HeartSurgery

    -.5035 .7235 HeartSurgery

    -.5501 .7039

    Diet/Exercise

    -.3067 .6544 Diet/Exercise

    -.3351 .6366

    HeartCenter

    -.7954 -1.3365 Heart Center -.8691 -1.3003

    Fairview .8149 -.5594 Fairview .8903 -.5442ParmaComm.

    -.9156 -.2448 ParmaComm.

    -1.0003 -.2382

    SW General 1.4118 .3972 SW General 1.5424 .3864MetroHealth .5972 -.4422 MetroHealth .6525 -.4302

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    44/70

    Next, the male and female maps were plotted and are shown on the next two

    pages. Each of the two maps is interpreted in the same way as the CMDS maps.

    The two maps appear to be quite comparable, but not exactly the same. How

    comparable are the two maps? To determine their comparability, one can correlate the

    interpoint distances in one map with the interpoint distances in the other maps. That is,

    we would calculate the distance between each of the possible pairs of points in one map

    and then correlate that with the corresponding distance on the other map.

    We first calculate the Euclidean distance separating all points on a map. There

    are 28 pairs of the 8 items, so there are 28 interpoint distances on each map.

    Magic BulletHeart Surgery

    Diet and Exercise

    Heart Center

    Fairview

    Parma Community

    SW General

    Metrohealth

    WMDS Hospitals by Ads (Males)

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    45/70

    Magic Bullet

    Heart Surgery

    Heart Center Parma Community

    Diet and Exercise

    SW General

    Fairview

    Metrohealth

    WMDS Hospitals by Ads (Females)

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    46/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    47/70

    For example, pair 1 was Magic Bullet and Heart Surgery. As shown in the

    previous table, the subtracted distance between Magic Bullet (-.3027) and Heart Surgery

    (-.5035) was .2008 for dimension 1. The subtracted distance for the same items (.8078

    and .7235) was .0843 for dimension 2. The Euclidean formula was then used to

    determine the straight line distance between the pairs. According to the example, this

    would be ((.2008)2 + (.0843)2), which is .2178. This was repeated for all 28 pairs of the

    8 stimuli and the following table was generated.

    PAIRMALE

    DISTANCESFEMALE

    DISTANCES

    Magic Bullet and Heart Surgery 0.2178 0.2342Magic Bullet and Diet and Exercise 1.535 0.1494Magic Bullet and Heart Center 2.002 2.1546Magic Bullet and Fairview 1.7659 1.8055Magic Bullet and ParmaCommunity 1.218 1.2236Magic Bullet and SouthwestGeneral 1.763 1.9152Magic Bullet and Metrohealth 1.5402 1.5638Heart Surgery and Diet andExercise 0.2086 0.2253

    Heart Surgery and Heart Center 2.0806 2.0294Heart Surgery and Fairview 1.8396 1.9059Heart Surgery and ParmaCommunity 1.0523 1.0441Heart Surgery and SW General 1.9429 2.1165Heart Surgery and Metrohealth 1.6032 1.653Diet and Exercise and HeartCenter 2.05 2.0092Diet and Exercise and Fairview 1.6527 1.7017Diet and Exercise and ParmaCommunity 1.086 1.099Diet and Exercise and SW General 1.7376 1.8941Diet and Exercise and Metrohealth 1.4211 1.4538Heart Center and Fairview 1.788 1.915Heart Center and ParmaCommunity 1.0983 1.0702Heart Center and SW General 2.8067 2.9428Heart Center and Metrohealth 1.655 1.7528Fairview and Parma Community 1.7589 1.9152

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    48/70

    Fairview and SW General 1.1276 1.1363Fairview and Metrohealth 0.2472 0.2637Parma Community and SWGeneral 2.4143 2.6183Parma Community and

    Metrohealth 1.5256 1.6639SW General and Metrohealth 1.1697 1.2078

    Next, a simple Pearson R correlation was calculated between the male and female

    groups. If a Pearson R correlation is high, it can be concluded that the two spaces (male

    and female maps) are comparable. If a Pearson R correlation is low, it can be concluded

    that there is a huge difference between the two matrices. In our data, it was found that

    the two matrices were highly correlated at .10 and significant at the .942 level.

    As in the CMDS solution, the WMDS Output 13 on the next page provides the

    scatterplot of fit between the scaled input data (horizontal axis) against the distances

    (vertical axis). This diagram represents the fit of the distances with the data. Again,

    when the stress levels are low, the points are close to the straight line running from the

    lower left-hand corner to the upper right-hand corner. The worse the fit, the more the

    points diverge from the straight line.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    49/70

    SPSS OUTPUT 13

    The scatter of the objects shows that the objects are running along the straight

    line, but not necessarily a good fit. In a professional or academic setting, we would want

    to use a map only if it had less scatter than is displayed here.

    The researcher can now interpret the separate maps of each segment and draw

    action implications specific to each segment.

    WMDS Summary

    In summary, WMDS has both advantages and disadvantages to the researcher.

    Let us first consider the advantages of using WMDS:

    It is especially useful for comparing sectors of the population or

    market in terms of the way they see particular objects.

    WMDS

    Scatterplot of Fit

    Disparities

    3.02.52.01.51.0.50.0Distances

    3.0

    2.5

    2.0

    1.5

    1.0

    .5

    0.0

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    50/70

    In WMDS, the dimensions on the maps are exactly the same for all

    segments. If a CMDS were to be calculated independently for each

    segment, the dimensions may have completely different meanings for

    each segment. This is because WMDS calculates the separate

    segment solutions using the same dimensions whereas CMDS does

    not.

    The ease of interpretability is evident through the use of WMDS due to

    the dimensions meaning the same thing for all segments.

    Actionability is easier because it clarifies the orientations of different

    segments of the population.

    Finally, interpretation is the same as CMDS because all interpoint

    distances between the objects are on the same scale of distance

    between each other.

    However, WMDS has its disadvantages as well.

    First, WMDS cannot be used as a scaling technique if there are

    dramatic differences between the matrices. It may be difficult for

    WMDS to find common dimensions that work for the groups.

    Next, WMDS indicates the perceived similarity of the stimuli, but

    doesnt necessarily explain the basis of the perceived similarity

    (dimensions/ attributes). The researcher will need additional

    information in the survey to determine labels for the dimensions. One

    can guess at the dimensions, but it is not advisable.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    51/70

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    52/70

    APPENDIX A:

    SPSS OUTPUT

    Classic Multidimensional Scaling: Two Dimensions

    Alscal Procedure Options

    Data Options-

    Number of Rows (Observations/Matrix). 8Number of Columns (Variables) . . . 8Number of Matrices . . . . . . 1Measurement Level . . . . . . . Interval

    Data Matrix Shape . . . . . . . SymmetricType . . . . . . . . . . . DissimilarityApproach to Ties . . . . . . . Leave TiedConditionality . . . . . . . . MatrixData Cutoff at . . . . . . . . .000000

    Model Options-

    Model . . . . . . . . . . . EuclidMaximum Dimensionality . . . . . 2

    Minimum Dimensionality . . . . . 2Negative Weights . . . . . . . Not Permitted

    Output Options-

    Job Option Header . . . . . . . PrintedData Matrices . . . . . . . . PrintedConfigurations and Transformations . PlottedOutput Dataset . . . . . . . . Not CreatedInitial Stimulus Coordinates . . . Computed

    Algorithmic Options-

    Maximum Iterations . . . . . . 30Convergence Criterion . . . . . .00100Minimum S-stress . . . . . . . .00500Missing Data Estimated by . . . . Ulbounds

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    53/70

    Raw (unscaled) Data for Subject 1

    1 2 3 45

    1 .0002 4.690 .0003 4.440 4.020 .0004 4.740 3.710 5.920 .0005 .000 5.120 4.230 4.040

    .0006 5.120 .000 5.430 4.970

    4.0607 4.230 5.430 .000 4.140

    4.9608 4.040 4.970 4.140 .000

    4.180

    6 7 8

    6 .0007 4.960 .0008 4.750 4.500 .000

    _

    Iteration history for the 2 dimensional solution (in

    squared distances)

    Young's S-stress formula 1 is used.

    Iteration S-stress Improvement

    1 .372952 .34050 .032463 .33447 .006034 .33378 .00069

    Iterations stopped becauseS-stress improvement is less than

    .001000

    Stress and squared correlation (RSQ) indistances

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    54/70

    RSQ values are the proportion of variance of the scaleddata (disparities)

    in the partition (row, matrix, or entire data)which

    is accounted for by their corresponding

    distances. Stress values are Kruskal's stress formula 1.

    For matrixStress = .20196 RSQ = .75785

    _

    Configuration derived in 2 dimensions

    Stimulus Coordinates

    Dimension

    Stimulus Stimulus 1 2Number Name

    1 MAGBULL 1.1437 .7485

    2 HRTSURG -1.3325 -.79543 DIETEXE .9450 -1.25164 HRTCENT -.8620 1.11935 FAIRVW .8389 .89956 METRO -1.3411 -.72827 PARMACOM .9433 -1.09708 SWGENER -.3353 1.1048

    _

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    55/70

    Optimally scaled data (disparities) forsubject 1

    1 2 3 45

    1 .0002 2.317 .0003 2.229 2.081 .0004 2.335 1.971 2.751 .0005 .662 2.469 2.155 2.088

    .0006 2.469 .662 2.578 2.416

    2.0957 2.155 2.578 .662 2.123

    2.4128 2.088 2.416 2.123 .662

    2.137

    6 7 8

    6 .0007 2.412 .0008 2.338 2.250 .000

    Classical Multidimensional Scaling CMDS

    Hospitals by Advertisements

    Dimension 1

    1.51.0.50.0-.5-1.0-1.5

    Dimension2

    1.5

    1.0

    .5

    0.0

    -.5

    -1.0

    -1.5

    swgener

    parmacom

    metro

    fairvwhrtcent

    dietexe

    hrtsurg

    magbull

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    56/70

    Scatterplot of Linear Fit

    Euclidean distance model

    Disparities

    3.02.52.01.51.0.5

    Distances

    3.5

    3.0

    2.5

    2.0

    1.5

    1.0

    .5

    0.0

    Classic Multidimensional Scaling: Three Dimensions

    Alscal Procedure Options

    Data Options-

    Number of Rows (Observations/Matrix). 8Number of Columns (Variables) . . . 8Number of Matrices . . . . . . 1Measurement Level . . . . . . . IntervalData Matrix Shape . . . . . . . SymmetricType . . . . . . . . . . . DissimilarityApproach to Ties . . . . . . . Leave TiedConditionality . . . . . . . . MatrixData Cutoff at . . . . . . . . .000000

    Model Options-

    Model . . . . . . . . . . . EuclidMaximum Dimensionality . . . . . 3Minimum Dimensionality . . . . . 3Negative Weights . . . . . . . Not Permitted

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    57/70

    Output Options-

    Job Option Header . . . . . . . PrintedData Matrices . . . . . . . . PrintedConfigurations and Transformations . Plotted

    Output Dataset . . . . . . . . Not CreatedInitial Stimulus Coordinates . . . Computed

    Algorithmic Options-

    Maximum Iterations . . . . . . 30Convergence Criterion . . . . . .00100Minimum S-stress . . . . . . . .00500Missing Data Estimated by . . . . Ulbounds_

    Raw (unscaled) Data for Subject 1

    1 2 3 45

    1 .0002 4.690 .000

    3 4.440 4.020 .0004 4.740 3.710 5.920 .0005 .000 5.120 4.230 4.040

    .0006 5.120 .000 5.430 4.970

    4.0607 4.230 5.430 .000 4.140

    4.9608 4.040 4.970 4.140 .000

    4.180

    6 7 8

    6 .0007 4.960 .0008 4.750 4.500 .000

    _

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    58/70

    Iteration history for the 3 dimensional solution (insquared distances)

    Young's S-stress formula 1 is used.

    Iteration S-stress Improvement

    1 .120732 .11583 .004903 .11551 .00032

    Iterations stopped becauseS-stress improvement is less than

    .001000

    Stress and squared correlation (RSQ) indistances

    RSQ values are the proportion of variance of the scaleddata (disparities)

    in the partition (row, matrix, or entire data)which

    is accounted for by their correspondingdistances.

    Stress values are Kruskal's stress formula 1.

    For matrixStress = .10963 RSQ = .95017

    _

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    59/70

    Configuration derived in 3 dimensions

    Stimulus Coordinates

    Dimension

    Stimulus Stimulus 1 2 3Number Name

    1 MAGBULL -.6025 -.6484 1.43472 HRTSURG 1.5085 .9025 -.04353 DIETEXE -1.2598 1.2315 -.30784 HRTCENT .4510 -1.2801 -1.09875 FAIRVW -.3146 -.7253 1.44256 METRO 1.5989 .8474 .33097 PARMACOM -1.3418 .8997 -.69778 SWGENER -.0398 -1.2272 -1.0604

    _

    Optimally scaled data (disparities) forsubject 1

    1 2 3 45

    1 .0002 2.824 .0003 2.732 2.577 .0004 2.843 2.462 3.279 .0005 1.089 2.983 2.654 2.584

    .0006 2.983 1.089 3.098 2.928

    2.5917 2.654 3.098 1.089 2.621

    2.924

    8 2.584 2.928 2.621 1.0892.636

    6 7 8

    6 .0007 2.924 .0008 2.847 2.754 .000

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    60/70

    Derived Stimulus Configuration

    Euclidean distance model

    metro

    Dimension 2

    fairvw

    2.0 2.0

    magbull

    hrtsurg

    -1.0

    -.5

    1.51.5

    0.0

    .5

    1.01.0

    1.0

    1.5

    hrtcent

    .5.5

    Dimension 3Dimension 1

    swgener

    0.00.0 -.5 -.5-1.0-1.0

    dietexe

    parmacom

    Scatterplot of Linear FitEuclidean distance model

    Disparities

    3.53.02.52.01.51.0

    Dis

    tances

    3.5

    3.0

    2.5

    2.0

    1.5

    1.0

    .5

    0.0

    Weighted Multidimensional Scaling: Two Dimensions

    Alscal Procedure Options

    Data Options-

    Number of Rows (Observations/Matrix). 8Number of Columns (Variables) . . . 8Number of Matrices . . . . . . 2

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    61/70

    Measurement Level . . . . . . . IntervalData Matrix Shape . . . . . . . SymmetricType . . . . . . . . . . . DissimilarityApproach to Ties . . . . . . . Leave TiedConditionality . . . . . . . . Matrix

    Data Cutoff at . . . . . . . . .000000

    Model Options-

    Model . . . . . . . . . . . IndscalMaximum Dimensionality . . . . . 2Minimum Dimensionality . . . . . 2Negative Weights . . . . . . . Not Permitted

    Output Options-

    Job Option Header . . . . . . . PrintedData Matrices . . . . . . . . Not PrintedConfigurations and Transformations . PlottedOutput Dataset . . . . . . . . Not CreatedInitial Stimulus Coordinates . . . ComputedInitial Subject Weights . . . . . Computed

    Algorithmic Options-

    Maximum Iterations . . . . . . 30Convergence Criterion . . . . . .00100

    Minimum S-stress . . . . . . . .00500Missing Data Estimated by . . . . Ulbounds_

    Iteration history for the 2 dimensional solution (insquared distances)

    Young's S-stress formula 1 is used.

    Iteration S-stress Improvement

    0 .273081 .272512 .25375 .018763 .25254 .001224 .25236 .00017

    Iterations stopped because S-stress improvement is lessthan .001000

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    62/70

    Stress and squared correlation (RSQ) indistances

    RSQ values are the proportion of variance of the scaleddata (disparities)

    in the partition (row, matrix, or entire data)whichis accounted for by their corresponding

    distances.Stress values are Kruskal's stress formula 1.

    Matrix Stress RSQ Matrix StressRSQ

    1 .243 .644 2 .197.777

    Averaged (rms) over matricesStress = .22105 RSQ = .71037

    _

    Configuration derived in 2 dimensions

    Stimulus Coordinates

    Dimension

    Stimulus Stimulus 1 2Number Name

    1 MAGBULL -.3858 1.12612 HRTSURG -.6418 1.00863 DIETEXE -.3909 .91224 HRTCENT -1.0139 -1.86315 FAIRVW 1.0387 -.77986 PARMACOM -1.1670 -.34137 SWGENER 1.7995 .55378 METRO .7612 -.6164

    _

    Subject weights measure the importance of each dimension toeach subject.Squared weights sum to RSQ.

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    63/70

    A subject with weights proportional to the average weightshas a weirdness ofzero, the minimum value.A subject with one large weight and many low weights has a

    weirdness near one.A subject with exactly one positive weight has a weirdnessof one,the maximum value for nonnegative weights.

    Subject Weights

    DimensionSubject Weird- 1 2Number ness

    1 .0759 .6155 .51462 .0714 .7347 .4871

    Overall importance ofeach dimension: .4594 .2510_

    Flattened Subject Weights

    Variable

    Subject Plot 1Number Symbol1 1 -1.00002 2 1.0000

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    64/70

    WMDS Euclidean Distance

    Males and Females Group Space

    Dimension 1

    2.01.51.0.50.0-.5-1.0-1.5

    Dimension2

    1.5

    1.0

    .5

    0.0

    -.5

    -1.0

    -1.5

    -2.0

    metro

    swgener

    parmacom

    fairvw

    hrtcent

    dietexehrtsurg

    magbull

    WMDSSubject Weights

    Dimension 1

    .74.72.70.68.66.64.62.60

    Dim

    ension2

    .52

    .51

    .50

    .49

    .48

    2

    1

    WMDS

    Scatterplot of Fit

    Disparities

    3.02.52.01.51.0.50.0

    Distances

    3.0

    2.5

    2.0

    1.5

    1.0

    .5

    0.0

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    65/70

    Flattened Subject Weights

    Individual differences (weighted) Euclidean di

    One Dimensional Plot

    .6.4.2-.0-.2-.4-.6

    Variable1

    1.5

    1.0

    .5

    0.0

    -.5

    -1.0

    -1.5

    2

    1

    Weighted Multidimensional Scaling: ThreeDimensions

    Alscal Procedure Options

    Data Options-

    Number of Rows (Observations/Matrix). 8Number of Columns (Variables) . . . 8Number of Matrices . . . . . . 2Measurement Level . . . . . . . IntervalData Matrix Shape . . . . . . . SymmetricType . . . . . . . . . . . DissimilarityApproach to Ties . . . . . . . Leave TiedConditionality . . . . . . . . MatrixData Cutoff at . . . . . . . . .000000

    Model Options-

    Model . . . . . . . . . . . IndscalMaximum Dimensionality . . . . . 3Minimum Dimensionality . . . . . 3Negative Weights . . . . . . . Not Permitted

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    66/70

    Output Options-

    Job Option Header . . . . . . . Printed

    Data Matrices . . . . . . . . Not PrintedConfigurations and Transformations . PlottedOutput Dataset . . . . . . . . Not CreatedInitial Stimulus Coordinates . . . ComputedInitial Subject Weights . . . . . Computed

    Algorithmic Options-

    Maximum Iterations . . . . . . 30Convergence Criterion . . . . . .00100Minimum S-stress . . . . . . . .00500Missing Data Estimated by . . . . Ulbounds_

    Iteration history for the 3 dimensional solution (insquared distances)

    Young's S-stress formula 1 is used.

    Iteration S-stress Improvement

    0 .185621 .184672 .17557 .009113 .17387 .001704 .17235 .001525 .17094 .001426 .16971 .001227 .16876 .00095

    Iterations stopped because

    S-stress improvement is less than.001000

    Stress and squared correlation (RSQ) indistances

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    67/70

    RSQ values are the proportion of variance of the scaleddata (disparities)

    in the partition (row, matrix, or entire data)which

    is accounted for by their corresponding

    distances. Stress values are Kruskal's stress formula 1.

    Matrix Stress RSQ Matrix StressRSQ

    1 .104 .853 2 .104.884

    Averaged (rms) over matricesStress = .10424 RSQ = .86887

    _

    Configuration derived in 3 dimensions

    Stimulus Coordinates

    Dimension

    Stimulus Stimulus 1 2 3

    Number Name

    1 MAGBULL .0752 1.1438 1.12472 HRTSURG -.6683 1.0738 -.96303 DIETEXE -.2696 1.1741 -.04474 HRTCENT -1.2907 -1.6781 .37955 FAIRVW .7337 -1.0397 -1.18136 PARMACOM -1.2219 -.1611 .78807 SWGENER 1.7006 -.0265 -1.40068 METRO .9412 -.4863 1.2974

    _

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    68/70

    Subject weights measure the importance of each dimension toeach subject.Squared weights sum to RSQ.

    A subject with weights proportional to the average weights

    has a weirdness ofzero, the minimum value.A subject with one large weight and many low weights has aweirdness near one.A subject with exactly one positive weight has a weirdnessof one,the maximum value for nonnegative weights.

    Subject Weights

    DimensionSubject Weird- 1 2 3Number ness

    1 .1742 .5804 .5666 .44202 .2027 .7548 .5105 .2326

    Overall importance ofeach dimension: .4533 .2908 .1247_

    Flattened Subject Weights

    VariableSubject Plot 1 2Number Symbol1 1 -1.0000 1.00002 2 1.0000 -1.0000

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    69/70

    Derived Stimulus Configuration

    Individual differences (weighted) Euclidean distance

    metro

    Dimension 2

    2.0 1.5

    magbull

    -1.5

    -1.0

    1.01.5

    swgener

    -.5

    0.0

    fairvw

    .5

    .51.0

    hrtcent

    1.0

    parmacom

    1.5

    0.0.5

    dietexe

    Dimension 3Dimension 1

    -.50.0 -.5 -1.0

    hrtsurg

    -1.5-1.0

    Derived Subject WeightsIndividual differences (weighted) Euclidean distance

    Dimension 2

    .8 .5

    .51

    .522

    .53

    .54

    .55

    1

    .56

    .7 .4

    .57

    Dimension 3Dimension 1.6 .3

    Scatterplot of Linear Fit

    Individual differences (weighted) Euclidean dis

    Disparities

    3.02.52.01.51.0.5

    Distances

    3.5

    3.0

    2.5

    2.0

    1.5

    1.0

    .5

  • 8/7/2019 Perceptual mapping by multidimensional scaling MDS

    70/70

    Flattened Subject Weights

    Individual differences (weighted) Euclidean di

    Variable 1

    1.51.0.50.0-.5-1.0-1.5

    Variable2

    1.5

    1.0

    .5

    0.0

    -.5

    -1.0

    -1.5

    2

    1