Optional Workshops and Tutorials
Optional 2-Day Workshops
Monday 8 am - 5 pm & Tuesday 8 am - 5 pm	CBC CONSULTING CHALLENGE WORKSHOP (Sponsored by Survey Sampling International) Aaron Hill and Brian McEwan, Sawtooth Software, Andy Cargill, Microsoft, and Chris Chapman, Google
	Participants will create several conjoint surveys and analyze actual respondent data using a team-oriented case study approach. A Microsoft products study will be fielded (SSI Sample) the evening after the first day, with data analyzed the second day. Each team’s final presentation will be critiqued by the instructors, including Andy Cargill, Director at Microsoft Hardware and Chris Chapman, formerly of Microsoft and currently at Google. This is a tremendous opportunity! Andy is not only a former research director and familiar with Microsoft Hardware issues, but more importantly is one of Microsoft's visionaries in the world of consumer products and consumer experience. Cost is $1,350.
	MENU-BASED CHOICE (MBC) SOFTWARE WORKSHOP Bryan Orme and Walter Williams, Sawtooth Software
	A new type of flexible menu-based choice questionnaire is becoming more popular over the last few years. Instead of picking one concept from available options (as in CBC), MBC tasks allow respondents to pick from zero to multiple items from a menu. The software takes a data processing and analysis process that can take experienced analysts from one to two weeks to do and compresses the timeline to 1 to 2 days. Attendees receive an evaluation copy of the software that they may use for 90 days (for non-commercial studies and evaluation purposes only). Cost is $1,000.
Optional Half-Day Tutorials
	Tutorial workshops are being offered to provide opportunities for a more in-depth learning experience. Each tutorial will be led by an outstanding professional with pertinent research and teaching experience. These classes are offered only on Tuesday. Tutorials are optional and are an additional cost ($250). Please note that you must register separately for the tutorials.
Tuesday 8 am - Noon	ADVANCES IN MARKET SEGMENTATION ANALYSIS Joseph Retzer, Market Tools
	The focus of this tutorial will be on new unsupervised learning (Cluster Analysis) algorithms and models. It will begin with a short review of some well known methods while paying particular attention to the Cluster Ensemble approach. Unsupervised methods covered will include: K-means PAM (Partitioning Around Medoids) Finite Mixture Modeling Clustering Objects on Subsets of Attributes (COSA) Cluster Ensembles Supervised learning techniques (e.g. CART) will also be discussed with emphasis on the relatively new and highly regarded Random Forest algorithm. In addition, Semi-Supervised Learning (SSL) algorithms, which combine information from both supervised and unsupervised learning, will be reviewed. SSL can be useful in producing clusters of high quality that are also informative to pre-specified marketing strategy. Finally, an illustration of effective post hoc cluster profiling, and presentation of cluster analysis results in general, will be demonstrated. Example code will be provided, typically in R, but also in Sawtooth Software, and Latent Gold when appropriate.
Tuesday 8 am - Noon	CUSTOMER SATISFACTION RESEARCH: CHALLENGES AND OPPORTUNITIES Keith Chrzan, Maritz Research
	This tutorial begins with basic topics in Customer Satisfaction Research, including What is customer satisfaction research? How are customer satisfaction studies designed and analyzed? How end users benefit from customer satisfaction studies? How does customer satisfaction research relate to loyalty? In many ways customer satisfaction data is the least well-behaved kind of data we as researchers will ever face; part of the presentation covers ways to address vexing data problems that go beyond basic regression and correlation and PLS analyses. The presentation concludes with some more advanced topics, including empirical tests of satisfaction rating scales and newer models for modeling customer satisfaction, like non-compensatory models of satisfaction and models using a particular maxdiff scaling with natural zero points for attribute utilities.
Tuesday 1 - 5 pm	ADVANCES IN MARKET SEGMENTATION ANALYSIS (repeat session) Joe Retzer, Market Tools
Tuesday 1 - 5 pm	POWER TRICKS FOR SSI WEB USERS Justin Luster, Sawtooth Software
	Even though SSI Web is easy to begin using, there is an amazing degree of flexibility and power awaiting the adventurous and advanced user. The course will demonstrate a number of power tricks that will open your eyes to new possibilities to accomplish challenging tasks and impress your clients. You’ll see how you can take on new work and problems you previously thought could not very easily be done. Many of these tricks involve Perl, JavaScript, and CSS. We will teach you the basics of these technologies and how you can apply them to create powerful SSI Web surveys. Some of the topics will include: Free Format questions, custom JavaScript validation, advanced constructed lists, advanced formatting options with CSS, how to search and replace text across your whole study, how to include Flash, how to build drag and drop sliders, surveys for the iPhone, and how to include audio and video into your surveys. Sample study code will be shared.
Main Conference Sessions
Wednesday, March 21, 2012
Session 1
Christopher Chapman Microsoft and Edwin Love Western Washington University	Game Theory and Conjoint Analysis: Using Choice Data for Strategic Decisions
	We demonstrate how choice-based conjoint (CBC) market simulations may be combined with game theory (GT) models. This approach can clarify strategic understanding, communicate CBC results more effectively, and focus clearly on business decisions. We present two cases of CBC+GT in a retail product line, and discuss considerations to conduct such analysis properly.
David G. Bakken, Michaela Gascon, and Dan Wasserman KJT Group	Contrast Effects on Willingness to Pay: How Conjoint Design Affects Adult Day Care Preferences
	We present the results of an experiment testing the impact of contrast effects on estimates of willingness to pay obtained using choice-based conjoint analysis. We tested the effect of external comparisons and pre-conditioning information on respondent preference for options of adult day care.
Break
Session 2
Paul Johnson and Brent Fuller SSI	Optimizing Pricing of Mobile Apps with Multiple Thresholds in Anchored MaxDiff
	We expand the MaxDiff threshold model to contain two thresholds. Our MaxDiff experiment compares SSI’s mobile app to other mobile apps at two price levels. We establish thresholds with two methods and examine order effects. Our survey results are tested against live behavior data and we recommend best methods of implementation.
Jack Horne, Bob Rayner, Reg Baker, and Silvo Lenart Market Strategies International	Continued Investigation into the Role of the ‘Anchor’ in MaxDiff and Related Tradeoff Exercises
	Several researchers have recently developed methods for setting an anchor or threshold for MaxDiff. These anchoring techniques, as well as one of our own, have drawn considerable interest among our marketing-research clients. We investigate six anchored methods using survey results, and simulations. Our recommendations should be of interest to methodologists and market research professionals alike.
Ralph Wirth and Anette Wolfrath GfK Marketing Sciences	Assessing Large Sets of Items Using MaxDiff Scaling: Evaluation of an Alternative Approach
	We introduce an approach to MaxDiff with large item sets that is based on blocked designs and utilizes HB’s mechanism for inferring full individual parameter vectors. Results of both a simulation study and an empirical study give insights into the performance of this new approach (a) under different study settings, and (b) relative to the performance of alternative approaches suggested by Hendrix and Drucker (2007).
Lunch
Session 3
Nicole Huyghe and Anita Prinzie solutions-2	What’s in a Label? Business Value of ‘Soft’ vs ‘Hard’ Cluster Ensembles
	Cluster ensembles improve segmentations. Hard cluster ensembles, combining segment labels from multiple cluster solutions are popular. Combining soft segment membership probabilities is as simple but surprisingly less common. We assess the added business value of soft cluster ensembles and guide decision making between soft membership data and hard segment labels.
Jane Tang and Andrew Grenville Vision Critical	How Low Can You Go?: Toward a Better Understanding of the Number of Choice Tasks Required for Reliable Input to Market Segmentation
	CBC exercises are becoming shorter in the era of on-line panels, resulting in lower precision in HB estimates. Sample size can usually be increased to compensate. For market segmentations, this is more problematic. We hope to provide guidelines on how few tasks are required while still enabling a reliable segmentation analysis.
Peter Kurz TNS Infratest Forschung GmbH and Stefan Binner bms marketing research + strategy	‘The Individual Choice Task Threshold’ Need for Variable Number of Choice Tasks
	Too many choice tasks lead to worse data, and our research shows that the maximum or optimum number of choice tasks differs by respondent. We present an application of “individual choice task thresholds” aiming for better data and reduced interview time. We also evaluate conjoint designs with enhanced utility balance as a further step in this direction.
Break
Session 4
Kevin Karty Affinnova	Taking Nothing Seriously: A Review of Approaches to Modeling the ‘None’ Option
	In any form of choice study, respondents consistently understate their likelihood of not purchasing, using, or adopting a good or service. This results from both methodological and behavioral factors, and creates an array of challenges. The proposed research will examine and compare a variety of ways to measure preference for “None”.
Christopher Saigal and Ely Dahan UCLA	Translating Voice of the Customer into Conjoint Attributes and Levels
	We present a voice-of-the-customer-based method for identifying conjoint attributes and levels directly from customer interviews. This approach elicits non-obvious attributes, gives voice to customer concerns, and applies Agglomerative Hierarchical Clustering (AHC) to quantify attributes and levels that easily plug into conjoint. A real-world application highlights the method and its benefits.
Thursday March 22, 2012
Session 5: Joint Sawtooth Software & Conjoint in Health Care Session
Warren Kuhfeld SAS Institute, Inc. and John Wurst University of Georgia and Atlanta Marketing Sciences Consulting	An Overview of the Design of Stated Choice Experiments
	Choice modeling has been an important technique for marketing researchers for many years. In this paper, we review some of the competing methods for designing stated choice experiments. Generic and alternative-specific designs, design properties, and ways to ensure realism (for example, by adding restrictions or interactions) are discussed.
F. Reed Johnson Research Triangle Institute	In Defense of Imperfect Experimental Designs: Statistical Efficiency and Measurement Error in Choice-Format Conjoint Analysis
	For given sample size, efficient designs insure that researchers can minimize estimate variances. In many cases, however, measurement error may be a greater source of estimation imprecision than statistical error. This study reports the results of simulating effects of sample size and other study characteristics on model precision using resampled data from 33 conjoint studies.
Joel Huber Duke University	CBC Design for Practitioners: What Matters Most
	Designs can be characterized as leading to greater statistical precision or more respondent precision. In the past 20 years testing various designs has demonstrated that respondent precision trumps statistical precision, suggesting that simpler, clearer designs are preferred to more complex ones. In the light of the Kuhfeld and Wurst, and Johnson presentations, I will summarize practical knowledge about designs that work.
Break
Session 6
Ely Dahan UCLA	Adaptive Best-Worst Conjoint (ABC) Analysis
	Conjoint subjects choose the best and worst of four alternatives without much more effort than CBC, as Louviere and Orme showed. Our adaptive choice conjoint method utilizes best-worst full-profile questioning to instantaneously estimate utility at the individual level with 12-15 tasks. A real-world application highlights the method and its benefits.
Jeffrey Dotson Vanderbilt University Jeff Larson Brigham Young University and Mark Ratchford Vanderbilt University	Maximizing Purchase Conversion by Minimizing Choice Deferral: Examining the Impact of Choice Set Design on Preference for the No-Choice Alternative
	We develop a model to capture the effect of choice set design on preference for the no-choice alternative. This model allows us to simplify a consumer’s choice process by optimizing the presentation of products, thus increasing the likelihood of purchase. We illustrate the value of our model in the context of preference for hotels in an online travel marketplace.
Lunch
Session 7
Carlo Borghi, Paolo Cordella, Kees van der Wagt, and Gerard Loosschilder SKIM	An Empirical Comparison of Approaches to Analyzing Menu-Based Choice Modeling Data
	We empirically compare two approaches to analyzing menu-based choice modeling data on their predictive validity: a single logit-based model that predicts the performance of combinatorial alternatives using Sampling of Alternatives versus a serial cross effects model. We apply the approaches to an ad-hoc dataset to investigate similarities and differences.
Rohit Pandey Columbia University and John Wagner Nielsen	Building Expandable Volume Consumption onto a Share Only MNL Model
	We describe our experiments with various attempts at extending a share only respondent level model using CBC/HB to an expandable consumption model, culminating in a new, successful model. This took the best features of other available models and gave logical results, also comparing well with actual data for two studies.
Break
Session 8
Scott Ferguson and Callaway Turner North Carolina State University Joseph Donndelinger and Mark Beltramo General Motors Research and Development	Using Respondent-Level Utilities to Create Targeted Initial Populations for Genetic Product Searches
	This paper demonstrates how respondent-level utilities can be used within a genetic search to create an initial population better suited for complex product searches. To maximize applicability, we also demonstrate how the genetic search can be extended to handle multiple objectives and enforce basic constraints in the solution structure.
Kevin Lattery Maritz Research and Bryan Orme Sawtooth Software	Can We Improve CBC Questionnaires with Strategically-Placed Level Overlap and Appropriate ‘Screening Rule’ Questions?
	We test different approaches to using respondent preferences (a priori vs. on-the-fly) to create CBC questionnaires with strategically-placed level overlap. We also investigate the elicitation and use of “unacceptable” levels. Our work builds on previous efforts at Sawtooth Software Conferences by Chrzan, Zepp, White (2010), Lattery (2009), and Johnson, Orme (2007).
Friday March 23, 2012
Session 9
Jay Weiner Ipsos Media-CT and Marcos Sanches Ipsos Reid	Being Creative in the Design: Performance of Hierarchical Bayes with Sparse Information Matrix
	We present a Discrete Choice case study where the nature of the business problem forces us to be creative on generating the design. We explore the design generation process and compare Hierarchical Bayes with simple Multinomial Logit to show that the latter can be a viable option in some cases.
Brian Griner and Ian McKinnon Kantar Health	Leveraging the Upper Level Models in HB for Integrated Modeling of Multiple Stakeholders and Decision Processes in Complex Market Environments
	This presentation will focus on the use of the upper-level models in HB to link choices of different stakeholder groups in complex market settings. The presentation will review the conceptual framework for the integrated approach using the upper level models as well as an application to a pharmaceutical case study and suggestions on research design and analysis in practical applications.
Mike Egner and Robby Hart Ipsos Open Thinking Exchange	Modifying Bayesian Networks for Key Drivers Analysis: An Overview of Practical Improvements
	By accounting for the relationships between drivers, Bayesian Networks are one of only a few methods that produce accurate measures of driver impact. However, they are vulnerable to several issues that threaten their ability to consistently perform well. We discuss these problems, and propose specific modifications to mitigate them.
Break
Session 10
John Howell and Greg Allenby Ohio State	Local Monopolies Arising from Fixed Costs
	Many products have a two-part price where consumers are asked to make an up-front investment in a product and then purchase add-on modules for that product. This two-part pricing distorts behavior leading to significant usage of the purchased products and greater resistance to competing products. We demonstrate how to apply economic models to estimate this sticky behavior and predict purchase volumes.
Kurt A. Pflughoeft Maritz Research and Felix Flory evolve24	A Comparison of Auto-Recognition Techniques for Topics and Tones
	An explosion of consumer-generated media coupled with advances in semantic technologies has provided new opportunities for text analytics. In this research, several auto-recognition and toning methods are used to identify comment categories and sentiments. The results from these techniques and from human coders are compared.
Conference Adjourned

Download Brochure