Training Descriptions

Introduction to Conjoint Methodologies (CBC/ACA/CVA)
Bryan Orme, Sawtooth Software, Inc.
In this session, we’ll introduce conjoint analysis, a powerful technique for quantifying how people make tradeoffs among products/services defined on multiple features. Motivation and essential theory will be presented in a non-technical, intuitive fashion. We'll discuss traditional (card-sort) conjoint, Adaptive Conjoint Analysis (ACA), and Choice-Based Conjoint. The pros and cons of each technique will be highlighted, including sample size issues. We’ll also provide a brief introduction to market simulations and outline the proper interpretation of conjoint data.


Stated and Derived Importance Measurements & Forecasting / Monte Carlo simulation
Keith Chrzan, Maritz Research
This session covers two issues that research practitioners face very frequently - attribute importance measurement and volumetric forecasting. Though they may be less sexy than conjoint analysis and segmentation, they are an important, and under-researched aspect of a researcher's job.

The first portion covers different types of stated and derived importance measurements and shows the results of two empirical comparisons. One of these comparisons involves new and traditional stated importance measures and finds that some stated methods perform dramatically better than others. Some perform so well, in fact, that the question arises whether they might outperform derived importance measures. The second comparison assesses the ability of stated and derived importance methods to make out-of-sample predictions. Together these empirical tests identify "best practices" in both stated and derived importance measurement that are better than most current practices.

The second portion covers volumetric forecasting. After reviewing various forecasting methods, we describe an additional Monte Carlo method that can be used to improve any of the forecasting methods. Participants will learn how to build Monte Carlo forecasts in Excel, and will walk through the construction in a case study.

This session covers two 2 hour tutorials.

Selling Conjoint : conjoint sometimes is perceived a specialist method that requires deep understanding of both the researcher and the client. That need not be! This tutorial covers a brief overview of the methods, and their benefits and pitfalls related to conjoint projects. Also, presentation techniques and comparisons with alternative methodologies are discussed. A basic understanding of conjoint is advisable.

Build Your Own : the technique where we ask respondents to compose their ideal product (e.g. Dell computers), given budget constraints, is called for in specific research. This workshop will give you a peek into how this method can be implemented and analyzed and eventually turned into a simulator. Basic understanding of survey authoring and analysis is advisable.

Methodologies like conjoint and MaxDiff enable the researcher to identify the value of different features of a product or service. Thereby, these methodologies bring in the customer perspective when designing or pricing a (new) product or service. Based on conjoint results a mobile phone manufacturer can, for example, assess the perceived value of an MP3 player on a new phone. Moreover, this value can be compared to the value of an FM radio on the phone. With this information product development or pricing decision is no longer only based on ‘gut feeling’ or extrapolated insights from the past.

But methodologies like these do not make clear why customers value particular features in a service or product. Especially when dealing with new-to-the-world products or when a study should provide insights for marketing communication, more information is needed.

The Means-End Chain theory provides us with a qualitative methodology, laddering, that takes the measurement of feature value and importance to a next level. Laddering helps to directly link product or feature preferences to the perceived customer benefits that make the feature valuable. Moreover, it identifies and links the deeper customer motives and values that ultimately drive behavior to the product or feature. Laddering is not an alternative to conjoint or MaxDiff in a market research study but can be applied in addition to these techniques. Whereas laddering is traditionally regarded a qualitative technique, there is also a quantitative derivative called SOAP (Structured Open Association Patterns).

The tutorial will consist of interactive exercises and will provide an overview of the Means-End Chain theory as well as the practical steps in designing and analyzing a laddering or SOAP module. Cases from various industries (e.g. Telecom, Food, Automotive) will be presented.

• Dealing with prohibitions: design testing, collapsing attributes, and alternative-specific designs.
• The four "random" design strategies. Characteristics of good designs, and selecting the best design for your study aims.
• Suggestions for analysis: Counts through HB.
• Interactions, worry about them or not? A practical approach to investigating and modeling interactions.
• The thorny issues: number of levels effect, constant-sum vs. discrete choice, partial-profile vs. full-profile
• A peek at the next level: customized CBC layouts (the flexibility and power of alternative-based discrete choice designs), adaptive CBC.

This course heavily leverages the CBC/Web system, but attendees will not be using the software in class. The presentation is designed as lecture, with time for classroom discussion and questions.

Advanced Scripting with JavaScript and Perl
Frank Berkers, SKIM & Lennaert Goris, SKIM
This training explores how JavaScript and Perl (in the form of Unverified Perl) can enhance your questionnaire. In this world of Web 2.0, interactive web-sites are a new standard. While a questionnaire does require a certain flow from page to page, the standard questions can benefit from a more 'interactive' look. This tutorial teaches you the basics to go beyond the standard HTML elements and create pages that adopt themselves to the respondent taking the survey.

Some exposure to HTML and scripting questionnaires will help benefiting from the workshop.

All attendees are invited to bring a laptop with CiW already installed. An SSI Web demo can be downloaded from either http://www.skimgroup.com/software/downloads or http://www.sawtoothsoftware.com/downloads.

Introduction to MaxDiff and Method of Paired Comparisons Scaling
Bryan Orme , Sawtooth Software, Inc.
Researchers are commonly asked to measure the preference or importance of multiple items (brands, attributes, positioning statements, etc.). The standard (and easiest) ways of doing this (such as 10-point or Likert scales) have significant problems, including lack of discrimination and scale use bias. Maximum difference scaling (also known as best/worst scaling) as well as the method of paired comparisons (MPC) are well-suited for overcoming these issues. They are particularly useful for cross-cultural research, where comparisons need to be drawn between heterogeneous samples.

This class assumes no prior knowledge of the techniques. And, MaxDiff tends to be easier to understand, manage, and present to clients than conjoint analysis.

CiW Questionnaire building
Chris King , Sawtooth Software, Inc. & John Howell , Sawtooth Software, Inc.
This tutorial/workshop is for those who already have a basic understanding of how CiW works. We expect that you have used the product for several projects. We expect that you are familiar with all the standard question types, beginning skip logic, and simple list building. The tutorial is designed to expand your knowledge of grid questions, CBC (conjoint) questions, list building, skip logic, global settings, passwords and quota control. Finally, we will introduce advanced topics including foreign languages, CAPI, linking with other questionnaires, skins to change questionnaire appearance, unverified Perl, and the use of JavaScript.

All attendees are invited to bring a laptop with CiW already installed.

Market Segmentation
Norbert Wirth, MarketTools

Targeted at experienced market researchers on supplier and client side, this tutorial provides a 360 view on market segmentation. The topics covered include managerial as well as technical issues such as:

• Development of clear objectives
• Early involvement of key stakeholders
• Definition of the segmentation base
• Data transformation
• Various up to date segmentation techniques
• Approaches to evaluate segments regarding their technical quality
• Criteria for segment actionability and economic evaluation from a business perspective
• Data warehouse tagging
• Implementation of segmentation across an organization.

Some special attention will be paid to the implications of conducting international market segmentation projects. This tutorial will provide comprehensive guidance to conduct and manage complex segmentation studies with a deep understanding of their analytical, organizational and business dimensions.

Introduction to Market Simulations and Building an Excel Simulator
Chris King , Sawtooth Software, Inc. & Lennaert Goris, SKIM

In this tutorial, attendees will use Sawtooth Software’s tool for market simulation (SMRT) and use Excel to build a market simulator. We’ll assume attendees already have some experience with conjoint methods. However, we will review basic part-worth utility and importance interpretation. The focus of the course will be on the motivation and mechanics for conducting market simulations.

We’ll introduce and explain the mechanics of First Choice, Logit Rule (Share of Preference), and Randomized First Choice simulation models. We will discuss base case formulation, sensitivity/elasticity analysis, segmentation analysis, and scale factor (exponent) tuning. Class members are invited to bring a laptop with SMRT and Excel installed. A temporary student lab version of the SMRT software will be given to attendees (for non-commercial use only) prior to the tutorial. Those with a professional version of SMRT may use their own installation.

In the second part of the course, we will replicate much of the SMRT simulator using Excel ( First Choice and Logit Rule only). A simulator built using Excel allows increased flexibility, better report/graphing capabilities and extensibility.

(Excel is a registered trademark and product of Microsoft Corporation.)

Statistical Techniques for Marketing Researchers: From the fundamental to the leading edge
Jon Pinnell, MarketVision Research

The tutorial will provide an overview of many of the common statistical techniques used in marketing research and will provide an introduction to less common, but more powerful techniques. The methods discussed will include techniques to:

Understand relationship between variables:

The common techniques used to understand relationships between variables include: correlation, linear regression, logistic regression and discriminant analysis. These techniques will be reviewed with focus on interpretation and application. Three less commonly used techniques, path analysis, partial least squares and structural equation modeling, will be introduced and examples will illustrate how they can add additional value to standard analyses.

Understand differences between individuals:

Cluster analysis has been a common tool for many years. The difference approaches to cluster analysis will be reviewed, along with their strengths and weaknesses for marketing research applications. Three additional topics will be discussed including archetypal analysis, which like cluster analysis is an interdependence technique. We will then discuss two dependence approaches that account for differences between individuals: latent class modeling and Bayesian methods.

Visualize data:

Different methods of visualizing data will be reviewed, including multidimensional scaling, canonical discriminant analysis, correspondence analysis and partial least squares. Recent advances in data visualization will be introduced.

Applications will be shown related to product design, market opportunity assessment, segmentation, customer satisfaction measurement, data mining, and image positioning.

Beyond the basics with CBC/HB and HB-Reg
John Howell, Sawtooth Software, Inc.
This tutorial is designed for intermediate hierarchical Bayes users. It will cover the advanced settings in CBC/HB, including the attribute settings and coding methods, interactions and how they are used, prior variance and prior degrees of freedom, demographic filters, constraints, and convergence. Attendees should have a basic understating of how to operate CBC/HB. The tutorial will also briefly cover HB-Reg and how to best structure the data for use in the program.

Laptop computers are not required for this class, but attendees not familiar with the software are encouraged to download the CBC/HB and HB-Reg demos. The demos can be downloaded from either http://www.skimgroup.com/software/downloads or http://www.sawtoothsoftware.com/downloads.