spss conjoint™ 17 - university of sussex spss statistics 17.0 is a comprehensive system for...

63
i SPSS Conjoint 17.0

Upload: lydieu

Post on 12-Mar-2018

221 views

Category:

Documents


2 download

TRANSCRIPT

i

SPSS Conjoint™ 17.0

For more information about SPSS Inc. software products, please visit our Web site at http://www.spss.com or contact

SPSS Inc.233 South Wacker Drive, 11th FloorChicago, IL 60606-6412Tel: (312) 651-3000Fax: (312) 651-3668

SPSS is a registered trademark and the other product names are the trademarks of SPSS Inc. for its proprietary computersoftware. No material describing such software may be produced or distributed without the written permission of theowners of the trademark and license rights in the software and the copyrights in the published materials.

The SOFTWARE and documentation are provided with RESTRICTED RIGHTS. Use, duplication, or disclosure bythe Government is subject to restrictions as set forth in subdivision (c) (1) (ii) of The Rights in Technical Data andComputer Software clause at 52.227-7013. Contractor/manufacturer is SPSS Inc., 233 South Wacker Drive, 11thFloor, Chicago, IL 60606-6412.Patent No. 7,023,453

General notice: Other product names mentioned herein are used for identification purposes only and may be trademarksof their respective companies.

Windows is a registered trademark of Microsoft Corporation.

Apple, Mac, and the Mac logo are trademarks of Apple Computer, Inc., registered in the U.S. and other countries.

This product uses WinWrap Basic, Copyright 1993-2007, Polar Engineering and Consulting, http://www.winwrap.com.

Printed in the United States of America.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means,electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher.

Preface

SPSS Statistics 17.0 is a comprehensive system for analyzing data. The Conjointoptional add-on module provides the additional analytic techniques described in thismanual. The Conjoint add-on module must be used with the SPSS Statistics 17.0 Basesystem and is completely integrated into that system.

Installation

To install the Conjoint add-on module, run the License Authorization Wizard using theauthorization code that you received from SPSS Inc. For more information, see theinstallation instructions supplied with the Conjoint add-on module.

Compatibility

SPSS Statistics is designed to run on many computer systems. See the installationinstructions that came with your system for specific information on minimum andrecommended requirements.

Serial Numbers

Your serial number is your identification number with SPSS Inc. You will need thisserial number when you contact SPSS Inc. for information regarding support, payment,or an upgraded system. The serial number was provided with your Base system.

Customer Service

If you have any questions concerning your shipment or account, contact your localoffice, listed on the Web site at http://www.spss.com/worldwide. Please have yourserial number ready for identification.

iii

Training Seminars

SPSS Inc. provides both public and onsite training seminars. All seminars featurehands-on workshops. Seminars will be offered in major cities on a regular basis.For more information on these seminars, contact your local office, listed on the Website at http://www.spss.com/worldwide.

Technical Support

Technical Support services are available to maintenance customers. Customers maycontact Technical Support for assistance in using SPSS Statistics or for installationhelp for one of the supported hardware environments. To reach Technical Support,see the Web site at http://www.spss.com, or contact your local office, listed on theWeb site at http://www.spss.com/worldwide. Be prepared to identify yourself, yourorganization, and the serial number of your system.

Additional PublicationsThe SPSS Statistical Procedures Companion, by Marija Norušis, has been published

by Prentice Hall. A new version of this book, updated for SPSS Statistics 17.0,is planned. The SPSS Advanced Statistical Procedures Companion, also basedon SPSS Statistics 17.0, is forthcoming. The SPSS Guide to Data Analysis forSPSS Statistics 17.0 is also in development. Announcements of publicationsavailable exclusively through Prentice Hall will be available on the Web site athttp://www.spss.com/estore (select your home country, and then click Books).

iv

Contents

1 Introduction to Conjoint Analysis 1

The Full-Profile Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2An Orthogonal Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2The Experimental Stimuli . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Collecting and Analyzing the Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Part I: User's Guide

2 Generating an Orthogonal Design 6

Defining Values for an Orthogonal Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Orthogonal Design Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9ORTHOPLAN Command Additional Features . . . . . . . . . . . . . . . . . . . . . . . . . 10

3 Displaying a Design 11

Display Design Titles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12PLANCARDS Command Additional Features . . . . . . . . . . . . . . . . . . . . . . . . . 13

v

4 Running a Conjoint Analysis 14

Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14Specifying the Plan File and the Data File. . . . . . . . . . . . . . . . . . . . . . . . 15Specifying How Data Were Recorded . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Optional Subcommands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Part II: Examples

5 Using Conjoint Analysis to Model Carpet-CleanerPreference 21

Generating an Orthogonal Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22Creating the Experimental Stimuli: Displaying the Design . . . . . . . . . . . . . . . 26Running the Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29Utility Scores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Relative Importance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Reversals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35Running Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35Preference Probabilities of Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

vi

Appendix

A Sample Files 38

Bibliography 52

Index 54

vii

Chapter

1Introduction to Conjoint Analysis

Conjoint analysis is a market research tool for developing effective product design.Using conjoint analysis, the researcher can answer questions such as: What productattributes are important or unimportant to the consumer? What levels of productattributes are the most or least desirable in the consumer’s mind? What is the marketshare of preference for leading competitors’ products versus our existing or proposedproduct?The virtue of conjoint analysis is that it asks the respondent to make choices in the

same fashion as the consumer presumably does—by trading off features, one againstanother.For example, suppose that you want to book an airline flight. You have the choice

of sitting in a cramped seat or a spacious seat. If this were the only consideration, yourchoice would be clear. You would probably prefer a spacious seat. Or suppose youhave a choice of ticket prices: $225 or $800. On price alone, taking nothing else intoconsideration, the lower price would be preferable. Finally, suppose you can takeeither a direct flight, which takes two hours, or a flight with one layover, which takesfive hours. Most people would choose the direct flight.The drawback to the above approach is that choice alternatives are presented on

single attributes alone, one at a time. Conjoint analysis presents choice alternativesbetween products defined by sets of attributes. This is illustrated by the followingchoice: would you prefer a flight that is cramped, costs $225, and has one layover, or aflight that is spacious, costs $800, and is direct? If comfort, price, and duration are therelevant attributes, there are potentially eight products:

Product Comfort Price Duration1 cramped $225 2 hours2 cramped $225 5 hours3 cramped $800 2 hours4 cramped $800 5 hours

1

2

Chapter 1

Product Comfort Price Duration5 spacious $225 2 hours6 spacious $225 5 hours7 spacious $800 2 hours8 spacious $800 5 hours

Given the above alternatives, product 4 is probably the least preferred, while product 5is probably the most preferred. The preferences of respondents for the other productofferings are implicitly determined by what is important to the respondent.Using conjoint analysis, you can determine both the relative importance of each

attribute as well as which levels of each attribute are most preferred. If the mostpreferable product is not feasible for some reason, such as cost, you would know thenext most preferred alternative. If you have other information on the respondents,such as background demographics, you might be able to identify market segmentsfor which distinct products can be packaged. For example, the business traveler andthe student traveler might have different preferences that could be met by distinctproduct offerings.

The Full-Profile Approach

Conjoint uses the full-profile (also known as full-concept) approach, whererespondents rank, order, or score a set of profiles, or cards, according to preference.Each profile describes a complete product or service and consists of a differentcombination of factor levels for all factors (attributes) of interest.

An Orthogonal Array

A potential problem with the full-profile approach soon becomes obvious if more thana few factors are involved and each factor has more than a couple of levels. The totalnumber of profiles resulting from all possible combinations of the levels becomes toogreat for respondents to rank or score in a meaningful way. To solve this problem, thefull-profile approach uses what is termed a fractional factorial design, which presentsa suitable fraction of all possible combinations of the factor levels. The resulting set,called an orthogonal array, is designed to capture the main effects for each factorlevel. Interactions between levels of one factor with levels of another factor areassumed to be negligible.

3

Introduction to Conjoint Analysis

The Generate Orthogonal Design procedure is used to generate an orthogonal arrayand is typically the starting point of a conjoint analysis. It also allows you to generatefactor-level combinations, known as holdout cases, which are rated by the subjectsbut are not used to build the preference model. Instead, they are used as a check onthe validity of the model.

The Experimental Stimuli

Each set of factor levels in an orthogonal design represents a different version of theproduct under study and should be presented to the subjects in the form of an individualproduct profile. This helps the respondent to focus on only the one product currentlyunder evaluation. The stimuli should be standardized by making sure that the profilesare all similar in physical appearance except for the different combinations of features.Creation of the product profiles is facilitated with the Display Design procedure. It

takes a design generated by the Generate Orthogonal Design procedure, or entered bythe user, and produces a set of product profiles in a ready-to-use format.

Collecting and Analyzing the Data

Since there is typically a great deal of between-subject variation in preferences, muchof conjoint analysis focuses on the single subject. To generalize the results, a randomsample of subjects from the target population is selected so that group results canbe examined.The size of the sample in conjoint studies varies greatly. In one report (Cattin and

Wittink, 1982), the authors state that the sample size in commercial conjoint studiesusually ranges from 100 to 1,000, with 300 to 550 the most typical range. In anotherstudy (Akaah and Korgaonkar, 1988), it is found that smaller sample sizes (lessthan 100) are typical. As always, the sample size should be large enough to ensurereliability.Once the sample is chosen, the researcher administers the set of profiles, or cards, to

each respondent. The Conjoint procedure allows for three methods of data recording.In the first method, subjects are asked to assign a preference score to each profile.This type of method is typical when a Likert scale is used or when the subjects areasked to assign a number from 1 to 100 to indicate preference. In the second method,subjects are asked to assign a rank to each profile ranging from 1 to the total numberof profiles. In the third method, subjects are asked to sort the profiles in terms of

4

Chapter 1

preference. With this last method, the researcher records the profile numbers in theorder given by each subject.Analysis of the data is done with the Conjoint procedure (available only through

command syntax) and results in a utility score, called a part-worth, for each factorlevel. These utility scores, analogous to regression coefficients, provide a quantitativemeasure of the preference for each factor level, with larger values corresponding togreater preference. Part-worths are expressed in a common unit, allowing them to beadded together to give the total utility, or overall preference, for any combination offactor levels. The part-worths then constitute a model for predicting the preferenceof any product profile, including profiles, referred to as simulation cases, that werenot actually presented in the experiment.The information obtained from a conjoint analysis can be applied to a wide variety

of market research questions. It can be used to investigate areas such as product design,market share, strategic advertising, cost-benefit analysis, and market segmentation.Although the focus of this manual is on market research applications, conjoint

analysis can be useful in almost any scientific or business field in which measuringpeople’s perceptions or judgments is important.

Part I:User's Guide

Chapter

2Generating an Orthogonal Design

Generate Orthogonal Design generates a data file containing an orthogonal main-effectsdesign that permits the statistical testing of several factors without testing everycombination of factor levels. This design can be displayed with the Display Designprocedure, and the data file can be used by other procedures, such as Conjoint.

Example. A low-fare airline startup is interested in determining the relative importanceto potential customers of the various factors that comprise its product offering. Price isclearly a primary factor, but how important are other factors, such as seat size, numberof layovers, and whether or not a beverage/snack service is included? A survey askingrespondents to rank product profiles representing all possible factor combinations isunreasonable given the large number of profiles. The Generate Orthogonal Designprocedure creates a reduced set of product profiles that is small enough to include in asurvey but large enough to assess the relative importance of each factor.

To Generate an Orthogonal Design

E From the menus choose:Data

Orthogonal DesignGenerate...

6

7

Generating an Orthogonal Design

Figure 2-1Generate Orthogonal Design dialog box

E Define at least one factor. Enter a name in the Factor Name text box. Factor namescan be any valid variable name, except status_ or card_. You can also assign anoptional factor label.

E Click Add to add the factor name and an optional label. To delete a factor, select it inthe list and click Remove. To modify a factor name or label, select it in the list, modifythe name or label, and click Change.

E Define values for each factor by selecting the factor and clicking Define Values.

Data File. Allows you to control the destination of the orthogonal design. You can savethe design to a new dataset in the current session or to an external data file.

Create a new dataset. Creates a new dataset in the current session containing thefactors and cases generated by the plan.Create new data file. Creates an external data file containing the factors and casesgenerated by the plan. By default, this data file is named ortho.sav, and it is savedto the current directory. Click File to specify a different name and destination forthe file.

8

Chapter 2

Reset random number seed to. Resets the random number seed to the specified value.The seed can be any integer value from 0 through 2,000,000,000. Within a session,a different seed is used each time you generate a set of random numbers, producingdifferent results. If you want to duplicate the same random numbers, you should set theseed value before you generate your first design and reset the seed to the same valueeach subsequent time you generate the design.

Optionally, you can:Click Options to specify the minimum number of cases in the orthogonal designand to select holdout cases.

Defining Values for an Orthogonal DesignFigure 2-2Generate Design Define Values dialog box

You must assign values to each level of the selected factor or factors. The factor namewill be displayed after Values and Labels for.Enter each value of the factor. You can elect to give the values descriptive labels.

If you do not assign labels to the values, labels that correspond to the values areautomatically assigned (that is, a value of 1 is assigned a label of 1, a value of 3 isassigned a label of 3, and so on).

9

Generating an Orthogonal Design

Auto-Fill. Allows you to automatically fill the Value boxes with consecutive valuesbeginning with 1. Enter the maximum value and click Fill to fill in the values.

Orthogonal Design OptionsFigure 2-3Generate Orthogonal Design Options dialog box

Minimum number of cases to generate. Specifies a minimum number of cases for theplan. Select a positive integer less than or equal to the total number of cases that can beformed from all possible combinations of the factor levels. If you do not explicitlyspecify the minimum number of cases to generate, the minimum number of casesnecessary for the orthogonal plan is generated. If the Orthoplan procedure cannotgenerate at least the number of profiles requested for the minimum, it will generate thelargest number it can that fits the specified factors and levels. Note that the design doesnot necessarily include exactly the number of specified cases but rather the smallestpossible number of cases in the orthogonal design using this value as a minimum.

Holdout Cases. You can define holdout cases that are rated by subjects but are notincluded in the conjoint analysis.

Number of holdout cases. Creates holdout cases in addition to the regular plancases. Holdout cases are judged by the subjects but are not used when theConjoint procedure estimates utilities. You can specify any positive integer lessthan or equal to the total number of cases that can be formed from all possiblecombinations of factor levels. Holdout cases are generated from another randomplan, not the main-effects experimental plan. The holdout cases do not duplicatethe experimental profiles or each other. By default, no holdout cases are produced.Randomly mix with other cases. Randomly mixes holdout cases with theexperimental cases. When this option is deselected, holdout cases appearseparately, following the experimental cases.

10

Chapter 2

ORTHOPLAN Command Additional Features

The command syntax language also allows you to:Append the orthogonal design to the active dataset rather than creating a new one.Specify simulation cases before generating the orthogonal design rather than afterthe design has been created.

See the Command Syntax Reference for complete syntax information.

Chapter

3Displaying a Design

The Display Design procedure allows you to print an experimental design. You canprint the design in either a rough-draft listing format or as profiles that you can presentto subjects in a conjoint study. This procedure can display designs created with theGenerate Orthogonal Design procedure or any designs displayed in an active dataset.

To Display an Orthogonal Design

E From the menus choose:Data

Orthogonal DesignDisplay...

Figure 3-1Display Design dialog box

E Move one or more factors into the Factors list.

E Select a format for displaying the profiles in the output.

11

12

Chapter 3

Format. You can choose one or more of the following format options:Listing for experimenter. Displays the design in a draft format that differentiatesholdout profiles from experimental profiles and lists simulation profiles separatelyfollowing the experimental and holdout profiles.Profiles for subjects. Produces profiles that can be presented to subjects. Thisformat does not differentiate holdout profiles and does not produce simulationprofiles.

Optionally, you can:Click Titles to define headers and footers for the profiles.

Display Design TitlesFigure 3-2Display Design Titles dialog box

Profile Title. Enter a profile title up to 80 characters long. Titles appear at the top ofthe output if you have selected Listing for experimenter and at the top of each newprofile if you have selected Profiles for subjects in the main dialog box. For Profiles for

subjects, if the special character sequence )CARD is specified anywhere in the title, theprocedure will replace it with the sequential profile number. This character sequence isnot translated for Listing for experimenter.

Profile Footer. Enter a profile footer up to 80 characters long. Footers appear at thebottom of the output if you have selected Listing for experimenter and at the bottom ofeach profile if you have selected Profiles for subjects in the main dialog box. For Profiles

for subjects, if the special character sequence )CARD is specified anywhere in the

13

Displaying a Design

footer, the procedure will replace it with the sequential profile number. This charactersequence is not translated for Listing for experimenter.

PLANCARDS Command Additional Features

The command syntax language also allows you to:Write profiles for subjects to an external file (using the OUTFILE subcommand).

See the Command Syntax Reference for complete syntax information.

Chapter

4Running a Conjoint Analysis

A graphical user interface is not yet available for the Conjoint procedure. To obtain aconjoint analysis, you must enter command syntax for a CONJOINT command into asyntax window and then run it.

For an example of command syntax for a CONJOINT command in the context of acomplete conjoint analysis—including generating and displaying an orthogonaldesign—see Chapter 5.For complete command syntax information about the CONJOINT command, seethe Command Syntax Reference.

To Run a Command from a Syntax Window

From the menus choose:File

NewSyntax...

This opens a syntax window.

E Enter the command syntax for the CONJOINT command.

E Highlight the command in the syntax window, and click the Run button (theright-pointing triangle) on the Syntax Editor toolbar.

See the Base User’s Guide for more information about running commands in syntaxwindows.

RequirementsThe Conjoint procedure requires two files—a data file and a plan file—and thespecification of how data were recorded (for example, each data point is a preferencescore from 1 to 100). The plan file consists of the set of product profiles to be rated

14

15

Running a Conjoint Analysis

by the subjects and should be generated using the Generate Orthogonal Designprocedure. The data file contains the preference scores or rankings of those profilescollected from the subjects. The plan and data files are specified with the PLAN andDATA subcommands, respectively. The method of data recording is specified with theSEQUENCE, RANK, or SCORE subcommands. The following command syntax shows aminimal specification:

CONJOINT PLAN='CPLAN.SAV' /DATA='RUGRANKS.SAV'/SEQUENCE=PREF1 TO PREF22.

Specifying the Plan File and the Data File

The CONJOINT command provides a number of options for specifying the plan file andthe data file.

You can explicitly specify the filenames for the two files. For example:CONJOINT PLAN='CPLAN.SAV' /DATA='RUGRANKS.SAV'

If only a plan file or data file is specified, the CONJOINT command reads thespecified file and uses the active dataset as the other. For example, if you specify adata file but omit a plan file (you cannot omit both), the active dataset is used asthe plan, as shown in the following example:CONJOINT DATA='RUGRANKS.SAV'

You can use the asterisk (*) in place of a filename to indicate the active dataset, asshown in the following example:CONJOINT PLAN='CPLAN.SAV' /DATA=*

The active dataset is used as the preference data. Note that you cannot use theasterisk (*) for both the plan file and the data file.

Specifying How Data Were Recorded

You must specify the way in which preference data were recorded. Data can berecorded in one of three ways: sequentially, as rankings, or as preference scores. Thesethree methods are indicated by the SEQUENCE, RANK, and SCORE subcommands.You must specify one, and only one, of these subcommands as part of a CONJOINTcommand.

16

Chapter 4

SEQUENCE Subcommand

The SEQUENCE subcommand indicates that data were recorded sequentially so thateach data point in the data file is a profile number, starting with the most preferredprofile and ending with the least preferred profile. This is how data are recorded if thesubject is asked to order the profiles from the most to the least preferred. The researcherrecords which profile number was first, which profile number was second, and so on.

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/SEQUENCE=PREF1 TO PREF22.

The variable PREF1 contains the profile number for the most preferred profile outof 22 profiles in the orthogonal plan. The variable PREF22 contains the profilenumber for the least preferred profile in the plan.

RANK Subcommand

The RANK subcommand indicates that each data point is a ranking, starting with theranking of profile 1, then the ranking of profile 2, and so on. This is how the data arerecorded if the subject is asked to assign a rank to each profile, ranging from 1 to n,where n is the number of profiles. A lower rank implies greater preference.

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/RANK=RANK1 TO RANK22.

The variable RANK1 contains the ranking of profile 1, out of a total of 22 profilesin the orthogonal plan. The variable RANK22 contains the ranking of profile 22.

SCORE Subcommand

The SCORE subcommand indicates that each data point is a preference score assignedto the profiles, starting with the score of profile 1, then the score of profile 2, and soon. This type of data might be generated, for example, by asking subjects to assigna number from 1 to 100 to show how much they liked the profile. A higher scoreimplies greater preference.

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/SCORE=SCORE1 TO SCORE22.

The variable SCORE1 contains the score for profile 1, and SCORE22 containsthe score for profile 22.

17

Running a Conjoint Analysis

Optional Subcommands

The CONJOINT command offers a number of optional subcommands that provideadditional control and functionality beyond what is required.

SUBJECT Subcommand

The SUBJECT subcommand allows you to specify a variable from the data file tobe used as an identifier for the subjects. If you do not specify a subject variable,the CONJOINT command assumes that all of the cases in the data file come fromone subject. The following example specifies that the variable ID, from the filerugranks.sav, is to be used as a subject identifier.

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/SCORE=SCORE1 TO SCORE22 /SUBJECT=ID.

FACTORS Subcommand

The FACTORS subcommand allows you to specify the model describing the expectedrelationship between factors and the rankings or scores. If you do not specify a modelfor a factor, CONJOINT assumes a discrete model. You can specify one of four models:

DISCRETE. The DISCRETE model indicates that the factor levels are categorical andthat no assumption is made about the relationship between the factor and the scores orranks. This is the default.

LINEAR. The LINEAR model indicates an expected linear relationship between thefactor and the scores or ranks. You can specify the expected direction of the linearrelationship with the keywords MORE and LESS. MORE indicates that higher levels of afactor are expected to be preferred, while LESS indicates that lower levels of a factorare expected to be preferred. Specifying MORE or LESS will not affect estimates ofutilities. They are used simply to identify subjects whose estimates do not matchthe expected direction.

IDEAL. The IDEAL model indicates an expected quadratic relationship between thescores or ranks and the factor. It is assumed that there is an ideal level for the factor,and distance from this ideal point (in either direction) is associated with decreasingpreference. Factors described with this model should have at least three levels.

18

Chapter 4

ANTIIDEAL. The ANTIIDEAL model indicates an expected quadratic relationshipbetween the scores or ranks and the factor. It is assumed that there is a worst level forthe factor, and distance from this point (in either direction) is associated with increasingpreference. Factors described with this model should have at least three levels.

The following command syntax provides an example using the FACTORS subcommand:

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/RANK=RANK1 TO RANK22 /SUBJECT=ID/FACTORS=PACKAGE BRAND (DISCRETE) PRICE (LINEAR LESS)

SEAL (LINEAR MORE) MONEY (LINEAR MORE).

Note that both package and brand are modeled as discrete.

PRINT Subcommand

The PRINT subcommand allows you to control the content of the tabular output. Forexample, if you have a large number of subjects, you can choose to limit the outputto summary results only, omitting detailed output for each subject, as shown in thefollowing example:

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/RANK=RANK1 TO RANK22 /SUBJECT=ID/PRINT=SUMMARYONLY.

You can also choose whether the output includes analysis of the experimental data,results for any simulation cases included in the plan file, both, or none. Simulationcases are not rated by the subjects but represent product profiles of interest to you. TheConjoint procedure uses the analysis of the experimental data to make predictionsabout the relative preference for each of the simulation profiles. In the followingexample, detailed output for each subject is suppressed, and the output is limited toresults of the simulations:

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/RANK=RANK1 TO RANK22 /SUBJECT=ID/PRINT=SIMULATION SUMMARYONLY.

19

Running a Conjoint Analysis

PLOT Subcommand

The PLOT subcommand controls whether plots are included in the output. Liketabular output (PRINT subcommand), you can control whether the output is limited tosummary results or includes results for each subject. By default, no plots are produced.In the following example, output includes all available plots:

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/RANK=RANK1 TO RANK22 /SUBJECT=ID/PLOT=ALL.

UTILITY Subcommand

The UTILITY subcommand writes a data file in SPSS Statistics format containingdetailed information for each subject. It includes the utilities for DISCRETE factors,the slope and quadratic functions for LINEAR, IDEAL, and ANTIIDEAL factors, theregression constant, and the estimated preference scores. These values can then beused in further analyses or for making additional plots with other procedures. Thefollowing example creates a utility file named rugutil.sav:

CONJOINT PLAN=* /DATA='RUGRANKS.SAV'/RANK=RANK1 TO RANK22 /SUBJECT=ID/UTILITY='RUGUTIL.SAV'.

Part II:Examples

Chapter

5Using Conjoint Analysis to ModelCarpet-Cleaner Preference

In a popular example of conjoint analysis (Green and Wind, 1973), a companyinterested in marketing a new carpet cleaner wants to examine the influence offive factors on consumer preference—package design, brand name, price, a GoodHousekeeping seal, and a money-back guarantee. There are three factor levels forpackage design, each one differing in the location of the applicator brush; three brandnames (K2R, Glory, and Bissell); three price levels; and two levels (either no or yes)for each of the last two factors. The following table displays the variables used in thecarpet-cleaner study, with their variable labels and values.Table 5-1Variables in the carpet-cleaner study

Variable name Variable label Value labelpackage package design A*, B*, C*brand brand name K2R, Glory, Bissellprice price $1.19, $1.39, $1.59seal Good Housekeeping seal no, yesmoney money-back guarantee no, yes

There could be other factors and factor levels that characterize carpet cleaners, butthese are the only ones of interest to management. This is an important point in conjointanalysis. You want to choose only those factors (independent variables) that youthink most influence the subject’s preference (the dependent variable). Using conjointanalysis, you will develop a model for customer preference based on these five factors.This example makes use of the information in the following data files:

carpet_prefs.sav contains the data collected from the subjects, carpet_plan.savcontains the product profiles being surveyed, and conjoint.sps contains the command

21

22

Chapter 5

syntax necessary to run the analysis. For more information, see Sample Files inAppendix A on p. 38.

Generating an Orthogonal Design

The first step in a conjoint analysis is to create the combinations of factor levels that arepresented as product profiles to the subjects. Since even a small number of factors anda few levels for each factor will lead to an unmanageable number of potential productprofiles, you need to generate a representative subset known as an orthogonal array.The Generate Orthogonal Design procedure creates an orthogonal array—also

referred to as an orthogonal design—and stores the information in a data file. Unlikemost procedures, an active dataset is not required before running the GenerateOrthogonal Design procedure. If you do not have an active dataset, you have theoption of creating one, generating variable names, variable labels, and value labelsfrom the options that you select in the dialog boxes. If you already have an activedataset, you can either replace it or save the orthogonal design as a separate data file.

To create an orthogonal design:

E From the menus choose:Data

Orthogonal DesignGenerate...

23

Using Conjoint Analysis to Model Carpet-Cleaner Preference

Figure 5-1Generate Orthogonal Design dialog box

E Enter package in the Factor Name text box, and enter package design in the FactorLabel text box.

E Click Add.

This creates an item labeled package ‘package design’ (?). Select this item.

E Click Define Values.

24

Chapter 5

Figure 5-2Generate Design Define Values dialog box

E Enter the values 1, 2, and 3 to represent the package designs A*, B*, and C*. Enterthe labels A*, B*, and C* as well.

E Click Continue.

You’ll now want to repeat this process for the remaining factors, brand, price, seal,and money. Use the values and labels from the following table, which includes thevalues you’ve already entered for package.

Factor Values Labelspackage 1, 2, 3 A*, B*, C*brand 1, 2, 3 K2R, Glory, Bissellprice 1.19, 1.39, 1.59 $1.19, $1.39, $1.59seal 1, 2 no, yesmoney 1, 2 no, yes

25

Using Conjoint Analysis to Model Carpet-Cleaner Preference

Once you have completed the factor specifications:

E In the Data File group, leave the default of Create a new dataset and enter a datasetname. The generated design will be saved to a new dataset, in the current session,with the specified name.

E Select Reset random number seed to and enter the value 2000000.

Generating an orthogonal design requires a set of random numbers. If you want toduplicate a design—in this case, the design used for the present case study—you needto set the seed value before you generate the design and reset it to the same value eachsubsequent time you generate the design. The design used for this case study wasgenerated with a seed value of 2000000.

E Click Options.

Figure 5-3Generate Orthogonal Design Options dialog box

E In the Minimum number of cases to generate text box, type 18.

By default, the minimum number of cases necessary for an orthogonal array isgenerated. The procedure determines the number of cases that need to be administeredto allow estimation of the utilities. You can also specify a minimum number of cases togenerate, as you’ve done here. You might want to do this because the default numberof minimum cases is too small to be useful or because you have experimental designconsiderations that require a certain minimum number of cases.

E Select Number of holdout cases and type 4.

Holdout cases are judged by the subjects but are not used by the conjoint analysis toestimate utilities. They are used as a check on the validity of the estimated utilities.The holdout cases are generated from another random plan, not the experimentalorthogonal plan.

26

Chapter 5

E Click Continue in the Generate Orthogonal Design Options dialog box.

E Click OK in the Generate Orthogonal Design dialog box.

Figure 5-4Orthogonal design for the carpet-cleaner example

The orthogonal design is displayed in the Data Editor and is best viewed by displayingvalue labels rather than the actual data values. This is accomplished by choosingValue Labels from the View menu.The variables in the data file are the factors used to specify the design. Each case

represents one product profile in the design. Notice that two additional variables,CARD_ and STATUS_, appear in the data file. CARD_ assigns a sequential number toeach profile that is used to identify the profile. STATUS_ indicates whether a profile ispart of the experimental design (the first 18 cases), a holdout case (the last 4 cases), ora simulation case (to be discussed in a later topic in this case study).The orthogonal design is a required input to the analysis of the data. Therefore, you

will want to save your design to a data file. For convenience, the current design hasbeen saved in carpet_plan.sav (orthogonal designs are also referred to as plans).

Creating the Experimental Stimuli: Displaying the Design

Once you have created an orthogonal design, you’ll want to use it to create the productprofiles to be rated by the subjects. You can obtain a listing of the profiles in a singletable or display each profile in a separate table.

27

Using Conjoint Analysis to Model Carpet-Cleaner Preference

To display an orthogonal design:

E From the menus choose:Data

Orthogonal DesignDisplay...

Figure 5-5Display Design dialog box

E Select package, brand, price, seal, and money for the factors.

The information contained in the variables STATUS_ and CARD_ is automaticallyincluded in the output, so they don’t need to be selected.

E Select Listing for experimenter in the Format group. This results in displaying the entireorthogonal design in a single table.

E Click OK.

28

Chapter 5

Figure 5-6Display of orthogonal design: Single table layout

The output resembles the look of the orthogonal design as shown in the DataEditor—one row for each profile, with the factors as columns. Notice, however, thatthe column headers are the variable labels rather than the variable names that you seein the Data Editor. Also notice that the holdout cases are identified with a footnote.This is of interest to the experimenter, but you certainly don’t want the subjects toknow which, if any, cases are holdouts.Depending on how you create and deliver your final product profiles, you may want

to save this table as an HTML, Word/RTF, Excel, or PowerPoint file. This is easilyaccomplished by selecting the table in the Viewer, right clicking, and selecting Export.Also, if you’re using the exported version to create the final product profiles, be sure toedit out the footnotes for the holdout cases.Perhaps the needs for your survey are better served by generating a separate table

for each product profile. This choice lends itself nicely to exporting to PowerPoint,since each table (product profile) is placed on a separate PowerPoint slide.

29

Using Conjoint Analysis to Model Carpet-Cleaner Preference

To display each profile in a separate table:

E Click the Dialog Recall button and select Display Design.

E Deselect Listing for experimenter and select Profiles for subjects.

E Click OK.

Figure 5-7Display of orthogonal design: Multitable layout

The information for each product profile is displayed in a separate table. In addition,holdout cases are indistinguishable from the rest of the cases, so there is no issue ofremoving identifiers for holdouts as with the single table layout.

Running the Analysis

You’ve generated an orthogonal design and learned how to display the associatedproduct profiles. You’re now ready to learn how to run a conjoint analysis.

30

Chapter 5

Figure 5-8Preference data for the carpet-cleaner example

The preference data collected from the subjects is stored in carpet_prefs.sav. The dataconsist of responses from 10 subjects, each identified by a unique value of the variableID. Subjects were asked to rank the 22 product profiles from the most to the leastpreferred. The variables PREF1 through PREF22 contain the IDs of the associatedproduct profiles, that is, the card IDs from carpet_plan.sav. Subject 1, for example,liked profile 13 most of all, so PREF1 has the value 13.Analysis of the data is a task that requires the use of command syntax—specifically,

the CONJOINT command. The necessary command syntax has been provided in thefile conjoint.sps.

CONJOINT PLAN='file specification'/DATA='file specification'/SEQUENCE=PREF1 TO PREF22/SUBJECT=ID/FACTORS=PACKAGE BRAND (DISCRETE)PRICE (LINEAR LESS)SEAL (LINEAR MORE) MONEY (LINEAR MORE)

/PRINT=SUMMARYONLY.

The PLAN subcommand specifies the file containing the orthogonal design—inthis example, carpet_plan.sav.The DATA subcommand specifies the file containing the preference data—in thisexample, carpet_prefs.sav. If you choose the preference data as the active dataset,you can replace the file specification with an asterisk (*), without the quotationmarks.The SEQUENCE subcommand specifies that each data point in the preference datais a profile number, starting with the most-preferred profile and ending with theleast-preferred profile.

31

Using Conjoint Analysis to Model Carpet-Cleaner Preference

The SUBJECT subcommand specifies that the variable ID identifies the subjects.The FACTORS subcommand specifies a model describing the expected relationshipbetween the preference data and the factor levels. The specified factors refer tovariables defined in the plan file named on the PLAN subcommand.The keyword DISCRETE is used when the factor levels are categorical and noassumption is made about the relationship between the levels and the data. Thisis the case for the factors package and brand that represent package design andbrand name, respectively. DISCRETE is assumed if a factor is not labeled withone of the four alternatives (DISCRETE, LINEAR, IDEAL, ANTIIDEAL) or is notincluded on the FACTORS subcommand.The keyword LINEAR, used for the remaining factors, indicates that the data areexpected to be linearly related to the factor. For example, preference is usuallyexpected to be linearly related to price. You can also specify quadratic models (notused in this example) with the keywords IDEAL and ANTIIDEAL.The keywords MORE and LESS, following LINEAR, indicate an expected directionfor the relationship. Since we expect higher preference for lower prices, thekeyword LESS is used for price. However, we expect higher preference for eithera Good Housekeeping seal of approval or a money-back guarantee, so the keywordMORE is used for seal and money (recall that the levels for both of these factorswere set to 1 for no and 2 for yes).Specifying MORE or LESS does not change the signs of the coefficients or affectestimates of the utilities. These keywords are used simply to identify subjectswhose estimates do not match the expected direction. Similarly, choosing IDEALinstead of ANTIIDEAL, or vice versa, does not affect coefficients or utilities.The PRINT subcommand specifies that the output contains information for thegroup of subjects only as a whole (SUMMARYONLY keyword). Information foreach subject, separately, is suppressed.

Try running this command syntax. Make sure that you have included valid paths tocarpet_prefs.sav and carpet_plan.sav. For a complete description of all options, seethe CONJOINT command in the Command Syntax Reference.

32

Chapter 5

Utility ScoresFigure 5-9Utility scores

This table shows the utility (part-worth) scores and their standard errors for each factorlevel. Higher utility values indicate greater preference. As expected, there is an inverserelationship between price and utility, with higher prices corresponding to lower utility(larger negative values mean lower utility). The presence of a seal of approval ormoney-back guarantee corresponds to a higher utility, as anticipated.Since the utilities are all expressed in a common unit, they can be added together

to give the total utility of any combination. For example, the total utility of acleaner with package design B*, brand K2R, price $1.19, and no seal of approval ormoney-back guarantee is:

utility(package B*) + utility(K2R) + utility($1.19)+ utility(no seal) + utility(no money-back) + constant

or

1.867 + 0.367 + (−6.595) + 2.000 + 1.250 + 12.870 = 11.759

If the cleaner had package design C*, brand Bissell, price $1.59, a seal of approval,and a money-back guarantee, the total utility would be:

0.367 + (−0.017) + (−8.811) + 4.000 + 2.500 + 12.870 = 10.909

33

Using Conjoint Analysis to Model Carpet-Cleaner Preference

CoefficientsFigure 5-10Coefficients

This table shows the linear regression coefficients for those factors specified asLINEAR (for IDEAL and ANTIIDEAL models, there would also be a quadratic term).The utility for a particular factor level is determined by multiplying the level by thecoefficient. For example, the predicted utility for a price of $1.19 was listed as −6.595in the utilities table. This is simply the value of the price level, 1.19, multiplied by theprice coefficient, −5.542.

Relative Importance

The range of the utility values (highest to lowest) for each factor provides a measure ofhow important the factor was to overall preference. Factors with greater utility rangesplay a more significant role than those with smaller ranges.

Figure 5-11Importance values

This table provides a measure of the relative importance of each factor known as animportance score or value. The values are computed by taking the utility range foreach factor separately and dividing by the sum of the utility ranges for all factors. Thevalues thus represent percentages and have the property that they sum to 100. The

34

Chapter 5

calculations, it should be noted, are done separately for each subject, and the resultsare then averaged over all of the subjects.Note that while overall or summary utilities and regression coefficients from

orthogonal designs are the same with or without a SUBJECT subcommand, importanceswill generally differ. For summary results without a SUBJECT subcommand, theimportances can be computed directly from the summary utilities, just as one cando with individual subjects. However, when a SUBJECT subcommand is used, theimportances for the individual subjects are averaged, and these averaged importanceswill not in general match those computed using the summary utilities.The results show that package design has the most influence on overall preference.

This means that there is a large difference in preference between product profilescontaining 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 indetermining overall preference. Price plays a significant role but not as significant aspackage design. Perhaps this is because the range of prices is not that large.

CorrelationsFigure 5-12Correlation coefficients

This table displays two statistics, Pearson’s R and Kendall’s tau, which providemeasures of the correlation between the observed and estimated preferences.The table also displays Kendall’s tau for just the holdout profiles. Remember that

the holdout profiles (four in the present example) were rated by the subjects but notused by the Conjoint procedure for estimating utilities. Instead, the Conjoint procedurecomputes correlations between the observed and predicted rank orders for theseprofiles as a check on the validity of the utilities.In many conjoint analyses, the number of parameters is close to the number of

profiles rated, which will artificially inflate the correlation between observed andestimated scores. In these cases, the correlations for the holdout profiles may give abetter indication of the fit of the model. Keep in mind, however, that holdouts willalways produce lower correlation coefficients.

35

Using Conjoint Analysis to Model Carpet-Cleaner Preference

Reversals

When specifying LINEAR models for price, seal, and money, we chose an expecteddirection (LESS or MORE) for the linear relationship between the value of the variableand the preference for that value. The Conjoint procedure keeps track of the numberof subjects whose preference showed the opposite of the expected relationship—forexample, a greater preference for higher prices, or a lower preference for a money-backguarantee. These cases are referred to as reversals.

Figure 5-13Number of reversals by factor and subject

This table displays the number of reversals for each factor and for each subject. Forexample, three subjects showed a reversal for price. That is, they preferred productprofiles with higher prices.

Running Simulations

The real power of conjoint analysis is the ability to predict preference for productprofiles that weren’t rated by the subjects. These are referred to as simulation cases.Simulation cases are included as part of the plan, along with the profiles from theorthogonal design and any holdout profiles.The simplest way to enter simulation cases is from the Data Editor, using the value

labels created when you generated the experimental design.

36

Chapter 5

To enter a simulation case in the plan file:

E On a new row in the Data Editor window, select a cell and select the desired value fromthe list (value labels can be displayed by choosing Value Labels from the View menu).Repeat for all of the variables (factors).

E Select Simulation for the value of the STATUS_ variable.

E Enter an integer value, to be used as an identifier, for the CARD_ variable. Simulationcases should be numbered separately from the other cases.

Figure 5-14Carpet-cleaner data including simulation cases

The figure shows a part of the plan file for the carpet-cleaner study, with two simulationcases added. For convenience, these have been included in carpet_plan.sav.The analysis of the simulation cases is accomplished with the same command

syntax used earlier, that is, the syntax in the file conjoint.sps. In fact, if you ran thesyntax described earlier, you would have noticed that the output also includes resultsfor the simulation cases, since they are included in carpet_plan.sav.You can choose to run simulations along with your initial analysis—as done

here—or run simulations at any later point simply by including simulation cases inyour plan file and rerunning CONJOINT. For more information, see the CONJOINTcommand in the Command Syntax Reference.

37

Using Conjoint Analysis to Model Carpet-Cleaner Preference

Preference Probabilities of SimulationsFigure 5-15Simulation results

This table gives the predicted probabilities of choosing each of the simulation casesas the most preferred one, under three different probability-of-choice models. Themaximum utility model determines the probability as the number of respondentspredicted to choose the profile divided by the total number of respondents. For eachrespondent, the predicted choice is simply the profile with the largest total utility.The BTL (Bradley-Terry-Luce) model determines the probability as the ratio of aprofile’s utility to that for all simulation profiles, averaged across all respondents.The logit model is similar to BTL but uses the natural log of the utilities instead ofthe utilities. Across the 10 subjects in this study, all three models indicated thatsimulation profile 2 would be preferred.

Appendix

ASample Files

The sample files installed with the product can be found in the Samples subdirectory ofthe installation directory. There is a separate folder within the Samples subdirectory foreach of the following languages: English, French, German, Italian, Japanese, Korean,Polish, Russian, Simplified Chinese, Spanish, and Traditional Chinese.

Not all sample files are available in all languages. If a sample file is not available in alanguage, that language folder contains an English version of the sample file.

Descriptions

Following are brief descriptions of the sample files used in various examplesthroughout the documentation.

accidents.sav. This is a hypothetical data file that concerns an insurance companythat is studying age and gender risk factors for automobile accidents in a givenregion. Each case corresponds to a cross-classification of age category and gender.adl.sav. This is a hypothetical data file that concerns efforts to determine thebenefits of a proposed type of therapy for stroke patients. Physicians randomlyassigned female stroke patients to one of two groups. The first received thestandard physical therapy, and the second received an additional emotionaltherapy. Three months following the treatments, each patient’s abilities to performcommon activities of daily life were scored as ordinal variables.advert.sav. This is a hypothetical data file that concerns a retailer’s efforts toexamine the relationship between money spent on advertising and the resultingsales. To this end, they have collected past sales figures and the associatedadvertising costs..

38

39

Sample Files

aflatoxin.sav. This is a hypothetical data file that concerns the testing of corn cropsfor aflatoxin, a poison whose concentration varies widely between and within cropyields. A grain processor has received 16 samples from each of 8 crop yields andmeasured the alfatoxin levels in parts per billion (PPB).aflatoxin20.sav. This data file contains the aflatoxin measurements from each of the16 samples from yields 4 and 8 from the aflatoxin.sav data file.anorectic.sav. While working toward a standardized symptomatology ofanorectic/bulimic behavior, researchers (Van der Ham, Meulman, Van Strien, andVan Engeland, 1997) made a study of 55 adolescents with known eating disorders.Each patient was seen four times over four years, for a total of 220 observations.At each observation, the patients were scored for each of 16 symptoms. Symptomscores are missing for patient 71 at time 2, patient 76 at time 2, and patient 47 attime 3, leaving 217 valid observations.autoaccidents.sav. This is a hypothetical data file that concerns the efforts of aninsurance analyst to model the number of automobile accidents per driver whilealso accounting for driver age and gender. Each case represents a separate driverand records the driver’s gender, age in years, and number of automobile accidentsin the last five years.band.sav. This data file contains hypothetical weekly sales figures of music CDsfor a band. Data for three possible predictor variables are also included.bankloan.sav. This is a hypothetical data file that concerns a bank’s efforts to reducethe rate of loan defaults. The file contains financial and demographic informationon 850 past and prospective customers. The first 700 cases are customers whowere previously given loans. The last 150 cases are prospective customers thatthe bank needs to classify as good or bad credit risks.bankloan_binning.sav. This is a hypothetical data file containing financial anddemographic information on 5,000 past customers.behavior.sav. In a classic example (Price and Bouffard, 1974), 52 students wereasked to rate the combinations of 15 situations and 15 behaviors on a 10-pointscale ranging from 0=“extremely appropriate” to 9=“extremely inappropriate.”Averaged over individuals, the values are taken as dissimilarities.behavior_ini.sav. This data file contains an initial configuration for atwo-dimensional solution for behavior.sav.

40

Appendix A

brakes.sav. This is a hypothetical data file that concerns quality control at a factorythat produces disc brakes for high-performance automobiles. The data file containsdiameter measurements of 16 discs from each of 8 production machines. Thetarget diameter for the brakes is 322 millimeters.breakfast.sav. In a classic study (Green and Rao, 1972), 21 Wharton School MBAstudents and their spouses were asked to rank 15 breakfast items in order ofpreference with 1=“most preferred” to 15=“least preferred.” Their preferenceswere recorded under six different scenarios, from “Overall preference” to “Snack,with beverage only.”breakfast-overall.sav. This data file contains the breakfast item preferences for thefirst scenario, “Overall preference,” only.broadband_1.sav. This is a hypothetical data file containing the number ofsubscribers, by region, to a national broadband service. The data file containsmonthly subscriber numbers for 85 regions over a four-year period.broadband_2.sav. This data file is identical to broadband_1.sav but contains datafor three additional months.car_insurance_claims.sav. A dataset presented and analyzed elsewhere (McCullaghand Nelder, 1989) concerns damage claims for cars. The average claim amountcan be modeled as having a gamma distribution, using an inverse link functionto relate the mean of the dependent variable to a linear combination of thepolicyholder age, vehicle type, and vehicle age. The number of claims filed canbe used as a scaling weight.car_sales.sav. This data file contains hypothetical sales estimates, list prices,and physical specifications for various makes and models of vehicles. The listprices and physical specifications were obtained alternately from edmunds.comand manufacturer sites.carpet.sav. In a popular example (Green and Wind, 1973), a company interested inmarketing a new carpet cleaner wants to examine the influence of five factors onconsumer preference—package design, brand name, price, a Good Housekeepingseal, and a money-back guarantee. There are three factor levels for packagedesign, each one differing in the location of the applicator brush; three brandnames (K2R, Glory, and Bissell); three price levels; and two levels (either no oryes) for each of the last two factors. Ten consumers rank 22 profiles defined bythese factors. The variable Preference contains the rank of the average rankingsfor each profile. Low rankings correspond to high preference. This variablereflects an overall measure of preference for each profile.

41

Sample Files

carpet_prefs.sav. This data file is based on the same example as described forcarpet.sav, but it contains the actual rankings collected from each of the 10consumers. The consumers were asked to rank the 22 product profiles from themost to the least preferred. The variables PREF1 through PREF22 contain theidentifiers of the associated profiles, as defined in carpet_plan.sav.catalog.sav. This data file contains hypothetical monthly sales figures for threeproducts sold by a catalog company. Data for five possible predictor variablesare also included.catalog_seasfac.sav. This data file is the same as catalog.sav except for theaddition of a set of seasonal factors calculated from the Seasonal Decompositionprocedure along with the accompanying date variables.cellular.sav. This is a hypothetical data file that concerns a cellular phonecompany’s efforts to reduce churn. Churn propensity scores are applied toaccounts, ranging from 0 to 100. Accounts scoring 50 or above may be looking tochange providers.ceramics.sav. This is a hypothetical data file that concerns a manufacturer’s effortsto determine whether a new premium alloy has a greater heat resistance than astandard alloy. Each case represents a separate test of one of the alloys; the heat atwhich the bearing failed is recorded.cereal.sav. This is a hypothetical data file that concerns a poll of 880 people abouttheir breakfast preferences, also noting their age, gender, marital status, andwhether or not they have an active lifestyle (based on whether they exercise atleast twice a week). Each case represents a separate respondent.clothing_defects.sav. This is a hypothetical data file that concerns the qualitycontrol process at a clothing factory. From each lot produced at the factory, theinspectors take a sample of clothes and count the number of clothes that areunacceptable.coffee.sav. This data file pertains to perceived images of six iced-coffee brands(Kennedy, Riquier, and Sharp, 1996) . For each of 23 iced-coffee image attributes,people selected all brands that were described by the attribute. The six brands aredenoted AA, BB, CC, DD, EE, and FF to preserve confidentiality.contacts.sav. This is a hypothetical data file that concerns the contact lists for agroup of corporate computer sales representatives. Each contact is categorizedby the department of the company in which they work and their company ranks.Also recorded are the amount of the last sale made, the time since the last sale,and the size of the contact’s company.

42

Appendix A

creditpromo.sav. This is a hypothetical data file that concerns a department store’sefforts to evaluate the effectiveness of a recent credit card promotion. To thisend, 500 cardholders were randomly selected. Half received an ad promoting areduced interest rate on purchases made over the next three months. Half receiveda standard seasonal ad.customer_dbase.sav. This is a hypothetical data file that concerns a company’sefforts to use the information in its data warehouse to make special offers tocustomers who are most likely to reply. A subset of the customer base was selectedat random and given the special offers, and their responses were recorded.customer_information.sav. A hypothetical data file containing customer mailinginformation, such as name and address.customers_model.sav. This file contains hypothetical data on individuals targetedby a marketing campaign. These data include demographic information, asummary of purchasing history, and whether or not each individual responded tothe campaign. Each case represents a separate individual.customers_new.sav. This file contains hypothetical data on individuals who arepotential candidates for a marketing campaign. These data include demographicinformation and a summary of purchasing history for each individual. Each caserepresents a separate individual.debate.sav. This is a hypothetical data file that concerns paired responses to asurvey from attendees of a political debate before and after the debate. Each casecorresponds to a separate respondent.debate_aggregate.sav. This is a hypothetical data file that aggregates the responsesin debate.sav. Each case corresponds to a cross-classification of preference beforeand after the debate.demo.sav. This is a hypothetical data file that concerns a purchased customerdatabase, for the purpose of mailing monthly offers. Whether or not the customerresponded to the offer is recorded, along with various demographic information.demo_cs_1.sav. This is a hypothetical data file that concerns the first step ofa company’s efforts to compile a database of survey information. Each casecorresponds to a different city, and the region, province, district, and cityidentification are recorded.demo_cs_2.sav. This is a hypothetical data file that concerns the second stepof a company’s efforts to compile a database of survey information. Each casecorresponds to a different household unit from cities selected in the first step, and

43

Sample Files

the region, province, district, city, subdivision, and unit identification are recorded.The sampling information from the first two stages of the design is also included.demo_cs.sav. This is a hypothetical data file that contains survey informationcollected using a complex sampling design. Each case corresponds to a differenthousehold unit, and various demographic and sampling information is recorded.dietstudy.sav. This hypothetical data file contains the results of a study of the“Stillman diet” (Rickman, Mitchell, Dingman, and Dalen, 1974). Each casecorresponds to a separate subject and records his or her pre- and post-diet weightsin pounds and triglyceride levels in mg/100 ml.dischargedata.sav. This is a data file concerning Seasonal Patterns of WinnipegHospital Use, (Menec , Roos, Nowicki, MacWilliam, Finlayson , and Black, 1999)from the Manitoba Centre for Health Policy.dvdplayer.sav. This is a hypothetical data file that concerns the development of anew DVD player. Using a prototype, the marketing team has collected focusgroup data. Each case corresponds to a separate surveyed user and records somedemographic information about them and their responses to questions about theprototype.flying.sav. This data file contains the flying mileages between 10 American cities.german_credit.sav. This data file is taken from the “German credit” dataset inthe Repository of Machine Learning Databases (Blake and Merz, 1998) at theUniversity of California, Irvine.grocery_1month.sav. This hypothetical data file is the grocery_coupons.sav data filewith the weekly purchases “rolled-up” so that each case corresponds to a separatecustomer. Some of the variables that changed weekly disappear as a result, andthe amount spent recorded is now the sum of the amounts spent during the fourweeks of the study.grocery_coupons.sav. This is a hypothetical data file that contains survey datacollected by a grocery store chain interested in the purchasing habits of theircustomers. Each customer is followed for four weeks, and each case correspondsto a separate customer-week and records information about where and how thecustomer shops, including how much was spent on groceries during that week.guttman.sav. Bell (Bell, 1961) presented a table to illustrate possible social groups.Guttman (Guttman, 1968) used a portion of this table, in which five variablesdescribing such things as social interaction, feelings of belonging to a group,physical proximity of members, and formality of the relationship were crossedwith seven theoretical social groups, including crowds (for example, people at a

44

Appendix A

football game), audiences (for example, people at a theater or classroom lecture),public (for example, newspaper or television audiences), mobs (like a crowd butwith much more intense interaction), primary groups (intimate), secondary groups(voluntary), and the modern community (loose confederation resulting from closephysical proximity and a need for specialized services).healthplans.sav. This is a hypothetical data file that concerns an insurance group’sefforts to evaluate four different health care plans for small employers. Twelveemployers are recruited to rank the plans by how much they would prefer tooffer them to their employees. Each case corresponds to a separate employerand records the reactions to each plan.health_funding.sav. This is a hypothetical data file that contains data on health carefunding (amount per 100 population), disease rates (rate per 10,000 population),and visits to health care providers (rate per 10,000 population). Each caserepresents a different city.hivassay.sav. This is a hypothetical data file that concerns the efforts of apharmaceutical lab to develop a rapid assay for detecting HIV infection. Theresults of the assay are eight deepening shades of red, with deeper shades indicatinggreater likelihood of infection. A laboratory trial was conducted on 2,000 bloodsamples, half of which were infected with HIV and half of which were clean.hourlywagedata.sav. This is a hypothetical data file that concerns the hourly wagesof nurses from office and hospital positions and with varying levels of experience.insure.sav. This is a hypothetical data file that concerns an insurance company thatis studying the risk factors that indicate whether a client will have to make a claimon a 10-year term life insurance contract. Each case in the data file represents apair of contracts, one of which recorded a claim and the other didn’t, matchedon age and gender.judges.sav. This is a hypothetical data file that concerns the scores given by trainedjudges (plus one enthusiast) to 300 gymnastics performances. Each row representsa separate performance; the judges viewed the same performances.kinship_dat.sav. Rosenberg and Kim (Rosenberg and Kim, 1975) set out toanalyze 15 kinship terms (aunt, brother, cousin, daughter, father, granddaughter,grandfather, grandmother, grandson, mother, nephew, niece, sister, son, uncle).They asked four groups of college students (two female, two male) to sort theseterms on the basis of similarities. Two groups (one female, one male) were askedto sort twice, with the second sorting based on a different criterion from the firstsort. Thus, a total of six “sources” were obtained. Each source corresponds to a

45

Sample Files

proximity matrix, whose cells are equal to the number of people in a sourceminus the number of times the objects were partitioned together in that source.kinship_ini.sav. This data file contains an initial configuration for athree-dimensional solution for kinship_dat.sav.kinship_var.sav. This data file contains independent variables gender, gener(ation),and degree (of separation) that can be used to interpret the dimensions of a solutionfor kinship_dat.sav. Specifically, they can be used to restrict the space of thesolution to a linear combination of these variables.mailresponse.sav. This is a hypothetical data file that concerns the efforts of aclothing manufacturer to determine whether using first class postage for directmailings results in faster responses than bulk mail. Order-takers record how manyweeks after the mailing each order is taken.marketvalues.sav. This data file concerns home sales in a new housing developmentin Algonquin, Ill., during the years from 1999–2000. These sales are a matterof public record.mutualfund.sav. This data file concerns stock market information for various techstocks listed on the S&P 500. Each case corresponds to a separate company.nhis2000_subset.sav. The National Health Interview Survey (NHIS) is a large,population-based survey of the U.S. civilian population. Interviews arecarried out face-to-face in a nationally representative sample of households.Demographic information and observations about health behaviors and statusare obtained for members of each household. This data file contains a subsetof information from the 2000 survey. National Center for Health Statistics.National Health Interview Survey, 2000. Public-use data file and documentation.ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Datasets/NHIS/2000/. Accessed2003.ozone.sav. The data include 330 observations on six meteorological variables forpredicting ozone concentration from the remaining variables. Previous researchers(Breiman and Friedman, 1985), (Hastie and Tibshirani, 1990), among others foundnonlinearities among these variables, which hinder standard regression approaches.pain_medication.sav. This hypothetical data file contains the results of a clinicaltrial for anti-inflammatory medication for treating chronic arthritic pain. Ofparticular interest is the time it takes for the drug to take effect and how itcompares to an existing medication.

46

Appendix A

patient_los.sav. This hypothetical data file contains the treatment records ofpatients who were admitted to the hospital for suspected myocardial infarction(MI, or “heart attack”). Each case corresponds to a separate patient and recordsmany variables related to their hospital stay.patlos_sample.sav. This hypothetical data file contains the treatment records of asample of patients who received thrombolytics during treatment for myocardialinfarction (MI, or “heart attack”). Each case corresponds to a separate patient andrecords many variables related to their hospital stay.polishing.sav. This is the “Nambeware Polishing Times” data file from the Data andStory Library. It concerns the efforts of a metal tableware manufacturer (NambeMills, Santa Fe, N. M.) to plan its production schedule. Each case represents adifferent item in the product line. The diameter, polishing time, price, and producttype are recorded for each item.poll_cs.sav. This is a hypothetical data file that concerns pollsters’ efforts todetermine the level of public support for a bill before the legislature. The casescorrespond to registered voters. Each case records the county, township, andneighborhood in which the voter lives.poll_cs_sample.sav. This hypothetical data file contains a sample of the voterslisted in poll_cs.sav. The sample was taken according to the design specified inthe poll.csplan plan file, and this data file records the inclusion probabilities andsample weights. Note, however, that because the sampling plan makes use ofa probability-proportional-to-size (PPS) method, there is also a file containingthe joint selection probabilities (poll_jointprob.sav). The additional variablescorresponding to voter demographics and their opinion on the proposed bill werecollected and added the data file after the sample as taken.property_assess.sav. This is a hypothetical data file that concerns a countyassessor’s efforts to keep property value assessments up to date on limitedresources. The cases correspond to properties sold in the county in the past year.Each case in the data file records the township in which the property lies, theassessor who last visited the property, the time since that assessment, the valuationmade at that time, and the sale value of the property.property_assess_cs.sav. This is a hypothetical data file that concerns a stateassessor’s efforts to keep property value assessments up to date on limitedresources. The cases correspond to properties in the state. Each case in the datafile records the county, township, and neighborhood in which the property lies, thetime since the last assessment, and the valuation made at that time.

47

Sample Files

property_assess_cs_sample.sav. This hypothetical data file contains a sample ofthe properties listed in property_assess_cs.sav. The sample was taken accordingto the design specified in the property_assess.csplan plan file, and this data filerecords the inclusion probabilities and sample weights. The additional variableCurrent value was collected and added to the data file after the sample was taken.recidivism.sav. This is a hypothetical data file that concerns a government lawenforcement agency’s efforts to understand recidivism rates in their area ofjurisdiction. Each case corresponds to a previous offender and records theirdemographic information, some details of their first crime, and then the time untiltheir second arrest, if it occurred within two years of the first arrest.recidivism_cs_sample.sav. This is a hypothetical data file that concerns agovernment law enforcement agency’s efforts to understand recidivism ratesin their area of jurisdiction. Each case corresponds to a previous offender,released from their first arrest during the month of June, 2003, and recordstheir demographic information, some details of their first crime, and the dataof their second arrest, if it occurred by the end of June, 2006. Offenders wereselected from sampled departments according to the sampling plan specified inrecidivism_cs.csplan; because it makes use of a probability-proportional-to-size(PPS) method, there is also a file containing the joint selection probabilities(recidivism_cs_jointprob.sav).rfm_transactions.sav. A hypothetical data file containing purchase transactiondata, including date of purchase, item(s) purchased, and monetary amount ofeach transaction.salesperformance.sav. This is a hypothetical data file that concerns the evaluationof two new sales training courses. Sixty employees, divided into three groups, allreceive standard training. In addition, group 2 gets technical training; group 3,a hands-on tutorial. Each employee was tested at the end of the training courseand their score recorded. Each case in the data file represents a separate traineeand records the group to which they were assigned and the score they receivedon the exam.satisf.sav. This is a hypothetical data file that concerns a satisfaction surveyconducted by a retail company at 4 store locations. 582 customers were surveyedin all, and each case represents the responses from a single customer.screws.sav. This data file contains information on the characteristics of screws,bolts, nuts, and tacks (Hartigan, 1975).

48

Appendix A

shampoo_ph.sav. This is a hypothetical data file that concerns the quality control ata factory for hair products. At regular time intervals, six separate output batchesare measured and their pH recorded. The target range is 4.5–5.5.ships.sav. A dataset presented and analyzed elsewhere (McCullagh et al., 1989)that concerns damage to cargo ships caused by waves. The incident counts can bemodeled as occurring at a Poisson rate given the ship type, construction period, andservice period. The aggregate months of service for each cell of the table formedby the cross-classification of factors provides values for the exposure to risk.site.sav. This is a hypothetical data file that concerns a company’s efforts to choosenew sites for their expanding business. They have hired two consultants toseparately evaluate the sites, who, in addition to an extended report, summarizedeach site as a “good,” “fair,” or “poor” prospect.siteratings.sav. This is a hypothetical data file that concerns the beta testing of ane-commerce firm’s new Web site. Each case represents a separate beta tester, whoscored the usability of the site on a scale from 0–20.smokers.sav. This data file is abstracted from the 1998 National Household Surveyof Drug Abuse and is a probability sample of American households. Thus, thefirst step in an analysis of this data file should be to weight the data to reflectpopulation trends.smoking.sav. This is a hypothetical table introduced by Greenacre (Greenacre,1984). The table of interest is formed by the crosstabulation of smoking behaviorby job category. The variable Staff Group contains the job categories Sr Managers,Jr Managers, Sr Employees, Jr Employees, and Secretaries, plus the categoryNational Average, which can be used as supplementary to an analysis. Thevariable Smoking contains the behaviors None, Light, Medium, and Heavy, plusthe categories No Alcohol and Alcohol, which can be used as supplementary to ananalysis.storebrand.sav. This is a hypothetical data file that concerns a grocery storemanager’s efforts to increase sales of the store brand detergent relative to otherbrands. She puts together an in-store promotion and talks with customers atcheck-out. Each case represents a separate customer.stores.sav. This data file contains hypothetical monthly market share data fortwo competing grocery stores. Each case represents the market share data for agiven month.stroke_clean.sav. This hypothetical data file contains the state of a medicaldatabase after it has been cleaned using procedures in the Data Preparation option.

49

Sample Files

stroke_invalid.sav. This hypothetical data file contains the initial state of a medicaldatabase and contains several data entry errors.stroke_survival. This hypothetical data file concerns survival times for patientsexiting a rehabilitation program post-ischemic stroke face a number of challenges.Post-stroke, the occurrence of myocardial infarction, ischemic stroke, orhemorrhagic stroke is noted and the time of the event recorded. The sample isleft-truncated because it only includes patients who survived through the end ofthe rehabilitation program administered post-stroke.stroke_valid.sav. This hypothetical data file contains the state of a medical databaseafter the values have been checked using the Validate Data procedure. It stillcontains potentially anomalous cases.survey_sample.sav. This hypothetical data file contains survey data, includingdemographic data and various attitude measures.tastetest.sav. This is a hypothetical data file that concerns the effect of mulch coloron the taste of crops. Strawberries grown in red, blue, and black mulch were ratedby taste-testers on an ordinal scale of 1 to 5 (far below to far above average). Eachcase represents a separate taste-tester.telco.sav. This is a hypothetical data file that concerns a telecommunicationscompany’s efforts to reduce churn in their customer base. Each case correspondsto a separate customer and records various demographic and service usageinformation.telco_extra.sav. This data file is similar to the telco.sav data file, but the “tenure”and log-transformed customer spending variables have been removed and replacedby standardized log-transformed customer spending variables.telco_missing.sav. This data file is a subset of the telco.sav data file, but some ofthe demographic data values have been replaced with missing values.testmarket.sav. This hypothetical data file concerns a fast food chain’s plans toadd a new item to its menu. There are three possible campaigns for promotingthe new product, so the new item is introduced at locations in several randomlyselected markets. A different promotion is used at each location, and the weeklysales of the new item are recorded for the first four weeks. Each case correspondsto a separate location-week.testmarket_1month.sav. This hypothetical data file is the testmarket.sav data filewith the weekly sales “rolled-up” so that each case corresponds to a separatelocation. Some of the variables that changed weekly disappear as a result, and thesales recorded is now the sum of the sales during the four weeks of the study.

50

Appendix A

tree_car.sav. This is a hypothetical data file containing demographic and vehiclepurchase price data.tree_credit.sav. This is a hypothetical data file containing demographic and bankloan history data.tree_missing_data.sav This is a hypothetical data file containing demographic andbank loan history data with a large number of missing values.tree_score_car.sav. This is a hypothetical data file containing demographic andvehicle purchase price data.tree_textdata.sav. A simple data file with only two variables intended primarilyto show the default state of variables prior to assignment of measurement leveland value labels.tv-survey.sav. This is a hypothetical data file that concerns a survey conducted by aTV studio that is considering whether to extend the run of a successful program.906 respondents were asked whether they would watch the program under variousconditions. Each row represents a separate respondent; each column is a separatecondition.ulcer_recurrence.sav. This file contains partial information from a study designedto compare the efficacy of two therapies for preventing the recurrence of ulcers. Itprovides a good example of interval-censored data and has been presented andanalyzed elsewhere (Collett, 2003).ulcer_recurrence_recoded.sav. This file reorganizes the information inulcer_recurrence.sav to allow you model the event probability for each intervalof the study rather than simply the end-of-study event probability. It has beenpresented and analyzed elsewhere (Collett et al., 2003).verd1985.sav. This data file concerns a survey (Verdegaal, 1985). The responsesof 15 subjects to 8 variables were recorded. The variables of interest are dividedinto three sets. Set 1 includes age and marital, set 2 includes pet and news, and set3 includes music and live. Pet is scaled as multiple nominal and age is scaled asordinal; all of the other variables are scaled as single nominal.virus.sav. This is a hypothetical data file that concerns the efforts of an Internetservice provider (ISP) to determine the effects of a virus on its networks. Theyhave tracked the (approximate) percentage of infected e-mail traffic on its networksover time, from the moment of discovery until the threat was contained.

51

Sample Files

waittimes.sav. This is a hypothetical data file that concerns customer waiting timesfor service at three different branches of a local bank. Each case corresponds toa separate customer and records the time spent waiting and the branch at whichthey were conducting their business.webusability.sav. This is a hypothetical data file that concerns usability testing ofa new e-store. Each case corresponds to one of five usability testers and recordswhether or not the tester succeeded at each of six separate tasks.wheeze_steubenville.sav. This is a subset from a longitudinal study of the healtheffects of air pollution on children (Ware, Dockery, Spiro III, Speizer, and FerrisJr., 1984). The data contain repeated binary measures of the wheezing statusfor children from Steubenville, Ohio, at ages 7, 8, 9 and 10 years, along with afixed recording of whether or not the mother was a smoker during the first yearof the study.workprog.sav. This is a hypothetical data file that concerns a government worksprogram that tries to place disadvantaged people into better jobs. A sample ofpotential program participants were followed, some of whom were randomlyselected for enrollment in the program, while others were not. Each case representsa separate program participant.

Bibliography

Akaah, I. P., and P. K. Korgaonkar. 1988. A conjoint investigation of the relativeimportance of risk relievers in direct marketing. Journal of Advertising Research,28:4, 38–44.

Bell, E. H. 1961. Social foundations of human behavior: Introduction to the study ofsociology. New York: Harper & Row.

Blake, C. L., and C. J. Merz. 1998. "UCI Repository of machine learning databases."Available at http://www.ics.uci.edu/~mlearn/MLRepository.html.

Breiman, L., and J. H. Friedman. 1985. Estimating optimal transformations formultiple regression and correlation. Journal of the American Statistical Association,80, 580–598.

Cattin, P., and D. R. Wittink. 1982. Commercial use of conjoint analysis: A survey.Journal of Marketing, 46:3, 44–53.

Collett, D. 2003. Modelling survival data in medical research, 2 ed. Boca Raton:Chapman & Hall/CRC.

Green, P. E., and V. Rao. 1972. Applied multidimensional scaling. Hinsdale, Ill.:Dryden Press.

Green, P. E., and Y. Wind. 1973. Multiattribute decisions in marketing: A measurementapproach. Hinsdale, Ill.: Dryden Press.

Greenacre, M. J. 1984. Theory and applications of correspondence analysis. London:Academic Press.

Guttman, L. 1968. A general nonmetric technique for finding the smallest coordinatespace for configurations of points. Psychometrika, 33, 469–506.

Hartigan, J. A. 1975. Clustering algorithms. New York: John Wiley and Sons.

Hastie, T., and R. Tibshirani. 1990. Generalized additive models. London: Chapmanand Hall.

Kennedy, R., C. Riquier, and B. Sharp. 1996. Practical applications of correspondenceanalysis to categorical data in market research. Journal of Targeting, Measurement,and Analysis for Marketing, 5, 56–70.

McCullagh, P., and J. A. Nelder. 1989. Generalized Linear Models, 2nd ed. London:Chapman & Hall.

52

53

Bibliography

Menec , V., N. Roos, D. Nowicki, L. MacWilliam, G. Finlayson , and C. Black. 1999.Seasonal Patterns of Winnipeg Hospital Use. : Manitoba Centre for Health Policy.

Price, R. H., and D. L. Bouffard. 1974. Behavioral appropriateness and situationalconstraints as dimensions of social behavior. Journal of Personality and SocialPsychology, 30, 579–586.

Rickman, R., N. Mitchell, J. Dingman, and J. E. Dalen. 1974. Changes in serumcholesterol during the Stillman Diet. Journal of the American Medical Association,228, 54–58.

Rosenberg, S., and M. P. Kim. 1975. The method of sorting as a data-gatheringprocedure in multivariate research. Multivariate Behavioral Research, 10, 489–502.

Van der Ham, T., J. J. Meulman, D. C. Van Strien, and H. Van Engeland. 1997.Empirically based subgrouping of eating disorders in adolescents: A longitudinalperspective. British Journal of Psychiatry, 170, 363–368.

Verdegaal, R. 1985. Meer sets analyse voor kwalitatieve gegevens (in Dutch). Leiden:Department of Data Theory, University of Leiden.

Ware, J. H., D. W. Dockery, A. Spiro III, F. E. Speizer, and B. G. Ferris Jr.. 1984.Passive smoking, gas cooking, and respiratory health of children living in six cities.American Review of Respiratory Diseases, 129, 366–374.

Index

anti-ideal model, 31

BTL (Bradley-Terry-Luce) model, 37

)CARDin Display Design, 12

card_ variablein Generate Orthogonal Design, 6

coefficients, 33command syntaxCONJOINT command, 30

correlation coefficients, 34

data filesin Generate Orthogonal Design, 6

discrete model, 31Display Design, 3, 11, 26

)CARD, 12footers, 12listing format, 11saving profiles, 13single-profile format, 11titles, 12

factor levels, 2, 21–22factors, 2, 21–22footersin Display Design, 12

full-profile approach, 2

Generate Orthogonal Design, 3, 6, 22data files, 6defining factor names, labels, and values, 8holdout cases, 9minimum cases, 9random number seed, 6simulation cases, 10

holdout cases, 2in Generate Orthogonal Design, 9

ideal model, 31importance scores, 33importance values, 33

Kendall’s tau, 34

linear model, 31listing formatin Display Design, 11

logit model, 37

max utility model, 37

orthogonal array, 2orthogonal designsdisplaying, 11, 26generating, 6, 22holdout cases, 9minimum cases, 9

part-worths, 4Pearson’s R, 34

random number seedin Generate Orthogonal Design, 6

reversals, 35

sample fileslocation, 38

sample size, 3simulation cases, 4, 18, 35in Generate Orthogonal Design, 10

54

55

Index

simulation resultsBTL (Bradley-Terry-Luce) model, 37logit model, 37max utility model, 37

single-profile formatin Display Design, 11

status_ variablein Generate Orthogonal Design, 6

syntaxCONJOINT command, 30

titlesin Display Design, 12

total utility, 32

utility scores, 4, 32