2017 Advanced Research Techniques (ART) Forum

Hyatt at Olive 8
1635 8th Avenue, Seattle, WA 98101
6/25/2017 8:00 AM - 6/28/2017 5:00 PM
Register before 5/26/2017 12:00 AM for early registration fee

Data scientists, marketing professionals and academics attending the Advanced Research Techniques (ART) Forum find a network of colleagues who are serious about developing tools that can solve the next generation of problems in the field of marketing. 

New hands-on sessions, focusing on repeatable research, will make sure you leave with practical takeaways--and the right code--for applying sophisticated methodologies and quantitative techniques to support strategic marketing decisions.

At the ART Forum, you’ll find a community not only dedicated to advancing their research skills, but also ready to help support each other through the process. This in an intimate event where you’ll have access to speakers and other smart minds throughout the entire event.

When you attend the ART Forum, you’ll leave with reproducible research from each session. It’s our way of helping you make an impact as soon as you get back to the office.

Click here to learn more and get a free downable sample of reproducible research.

2017 Schedule At A Glance​

2017 Program Committee

Chris Chapman, PhD - Chair

Senior Quantitative Experience Researcher, Google

Marc Dotson
Assistant Professor of Marketing, Brigham Young University

Elea Feit
Assistant Professor of Marketing, Drexel University

Joseph Retzer
CRO, ACT-Market Research Solutions

Melinda Thielbar
Senior Research Statistician Developer, JMP Software

Robert Welborn
Assistant Vice President, Enterprise Data & Analytics Office, USAA

Robert Zeithammer
Assistant Professor, University of California, Los Angeles​​


Who Attends This Conference

Since its inception in 1990, the conference has focused on how sophisticated methodologies and quantitative techniques can support strategic and tactical marketing decisions. The attendees are experienced research practitioners who use advanced methods, and marketing academics interested in industry problems.

AMA Cancellation Policies​

Sunday, June 25, 2017

4-Hour Tutorials (optional, for an additional fee)
TUTORIALS - 8:00 AM - 12:00 PM

A.  Becoming an Expert in Conjoint Analysis
Bryan Orme, President, Sawtooth Software
Keith Chrzan, Senior VP, Sawtooth Software

Join Bryan Orme and Keith Chrzan for a four-hour tutorial based on their new book, “Becoming an Expert in Conjoint Analysis” (to be released during 2017).  The instructors will share concepts and power tricks that have taken them decades to learn.  Attending this course and carefully reading the associated book will accelerate you toward expert status in conjoint/choice methods.  
This course will take a software-agnostic approach.  We’ll focus on theory, practical tips, and know-how.  The vast majority of what we’ll cover can be accomplished using a variety of commercial and open-source software tools.  
The course outline will closely follow the chapter organization of the book:
  • How people choose
  • Reducing hypothetical bias
  • Advanced CBC designs
  • Adaptive designs and customization
  • The None alternative
  • Holdout choice tasks
  • Dealing with many attributes and levels
  • Sample size decisions
  • Coding conjoint data for estimation
  • Part-worth utility estimation
  • Allocation and volumetric choice experiments
  • Statistical testing
  • Willingness to pay
  • Building choice simulators
  • Optimization using choice simulators
  • Situational choice experiments
  • BW Case II
  • Introduction to MBC
B.  Hierarchical Bayes Methods for Marketing 
Jeff Dotson, Associate Professor, Brigham Young University
Elea McDonnell Feit, Assistant Professor, Drexel University

Hierarchical Bayes (HB) statistical methods have become common in marketing and many of the cutting-edge models presented at ART forum are built using HB methods. If you’ve always wanted to know why HB is so important in marketing, what is “under the hood” of popular HB packages, or how to start estimating HB models on your own, this tutorial is for you. Our goal is to prepare participants to be smart and sophisticated users of HB methods, including widely-used models like HB choice models. While our eye is toward the use of these methods in practice, we will provide the solid grounding in the theory of Bayesian inference and Markov Chain Monte Carlo (MCMC) estimation that is needed to use these methods with confidence. Basic modeling ideas will be presented in Excel, with R and the MCMCpack package used to illustrate more sophisticated models. Participants are encouraged to bring their laptop with Excel, R and RStudio installed and follow along.

C.  Introduction to R 
Marc Dotson, Assistant Professor of Marketing, Brigham Young University

Learn the basics of data analysis using R. No previous statistical programming experience is required. Please bring your own laptop with the latest version of R (cran.r-project.org) and RStudio (rstudio.com) installed.

D.  An Introduction to Probability Models for Marketing Research
Michael Braun, Associate Professor of Marketing, Cox School of Business, Southern Methodist University
Arun Gopalakrishnan, Assistant Professor of Marketing, Olin Business School, Washington University in St Louis

Central to a complete understanding of today’s “leading-edge” market research techniques is a sound intuitive appreciation of the basic foundations upon which these sophisticated tools are built. For example, both hierarchical Bayes models and latent class models build on simple probability modeling concepts (e.g., zero-order choice process, Poisson counts, conditional expectations, and exponential interpurchase times) — yet how many researchers are comfortable at precisely defining these concepts or explaining the motivation for using them? This tutorial aims to fill in these gaps by bringing practitioners fully up to speed on the basic methods that may underlie many of their current or future research activities. Our two broad objectives are (1) to review the basic terminology and logic associated with the area of probability models as applied to marketing research problems, and (2) to develop participants’ skills through a set of case studies that demonstrate the model building process in detail. We will illustrate all of the steps required to develop a probability model, estimate its parameters, and interpret the results. Careful and extensive use is made of the Solver tool in Microsoft Excel, which makes it possible to construct all of these models within a familiar spreadsheet environment. By the end of the tutorial, participants should be quite comfortable with all of the aforementioned principles and models and the managerial issues that surround them.

2-Hour Tutorials (optional, for an additional fee)
TUTORIALS 1:00 - 3:00 PM

I. Machine Learning for Humans (Marketers)
Sanjog Misra, Professor of Marketing, University of Chicago Booth School There has been a lot of buzz of machine learning (ML) and the value it brings to data analytics. In this tutorial we will introduce ourselves to some of the fundamental elements of the ML toolkit as it pertains to marketing. The topics we will cover include:
  1. Fundamentals (Bias Variance Tradeoff, Cross Validation,…)
  2. Regularization (Lasso, ElasticNet, gamlr)
  3. Trees and Forests
  4. Ensembles and Super Learners
  5. Deep Learning
  6. Inference (or the lack thereof)
In particular, we will focus on applications of the methods to marketing problems by adapting and extending them as needed. The tutorial will make heavy use of the R programming language and code will be provided for hands on exercises. 
4-Hour Tutorials (optional, for an additional fee)
TUTORIALS 1:00 - 5:00 PM

F. Using Stan to Estimate Hierarchical Bayes Models 
Elea McDonnell Feit, Assistant Professor, Drexel University
Kevin Van Horn, Senior Data Science Engineer, Adobe

Stan (http://mc-stan.org/) is a new open-source tool for estimating complex statistical models using MCMC. Similar to JAGS and BUGS, Stan allows you to specify complex models using a modeling language. Once the model is specified, Stan automatically generates a routine to sample from the posterior, allowing the user to focus on the model and posterior estimates, rather than on details of the MCMC routine. In this hands-on tutorial, we will show how to use Stan to fit popular models in marketing including the hierarchical choice models typically used for conjoint, a nested logit model for the no-choice option and data fusion models. Users will be encouraged to run code on their own laptops during the workshop. Stan can be accessed through R, Python or Matlab and several other statistical programming tools; our focus will be on using the RStan interface. Users who are already familiar with Bayesian inference and basic R syntax will get the most out of this tutorial. Instructions for installing Stan and example R code will be provided. Users are encouraged to install Stan before the tutorial.

G. Web-Based Conjoint Simulators using RStudio’s Shiny 
Jake Lee, President, Quantum Strategy, Inc.

Bring your laptop. This hands on tutorial will show you how to develop a basic conjoint simulator with an html interface and with R in the background to do the calculations. If you’ve ever thought excel simulators are slow or clunky and you are not quite comfortable with C++, Shiny is worth a look. This tutorial will briefly cover what is possible and then provide you with the background and tools to build a simple conjoint (MNL) simulator. Simulators can be shared as an R file or hosted securely on the cloud.
  • Experience with R is helpful, but not required.
  • Experience with the R package Shiny is not required.
H.  Probability Models for Customer Base Analysis 
Michael Braun, Associate Professor of Marketing, Southern Methodist University

Arun Gopalakrishnan, Assistant Professor of Marketing, Olin Business School, Washington University in St Louis​

Customer-base analysis seeks to use information on the history of customer purchase patterns to identify which individuals are most likely to be active (or inactive) customers and to predict future purchasing patterns by those customers listed in the firm’s transaction database. Any researcher hoping to make statements about “customer lifetime value” must deal with these issues, but unfortunately the set of commonly available tools is not well-suited for the task. This tutorial builds upon the basic “platform” provided in our introductory tutorial to provide a set of techniques and models tailored to address these situations properly. We focus on developing the models entirely in Excel and provide attendees with the relevant spreadsheets and notes on how to implement the models “from scratch”. Our goal is to provide the attendee with tools that can be applied immediately (maybe with some slight modifications) at his/her place of work.

The structure of the tutorial is as follows:
  • Introduction to the idea of customer-base analysis
  • Overview of the concept of Customer Lifetime Value (CLV) and the presentation of a general framework for its calculation
  • Brief review of the probability modeling basics required for model building (e.g., review of binomial, geometric, Poisson, exponential, gamma, and beta distributions; discussion of common mixtures such as the NBD, betageometric, and beta-binomial)
  • Presentation of probability models that can be used to answer various managerial questions including the calculation of CLV. (The empirical examples come from settings as diverse as e-tailing, the charity sector and media subscriptions)
  • Generalizations of the specific models presented in this tutorial making links to the broader modeling literature

I.  Details to follow

5:30 - 6:00 PM  First Time Attendee Orientation
6:00 - 7:30 PM  Welcome Reception with Poster Session and Exhibits


Monday, June 26, 2017 

8:50 - 9:00 AM  Welcome
9:00 - 9:30 AM  Predicting Bundle Preference Using Configuration Data
I-Hsuan (Shaine) Chiu, Ph.D. Candidate, Department of Marketing, University of Iowa
Gary J. Russell, Henry B. Tippie Research Professor of Marketing, University of Iowa
Thomas S. Gruca, Henry B. Tippie Research Professor of Marketing, University of Iowa

Researchers and practitioners often use a configuration data approach (“build-your-own-bundle”) to study bundle utility structures. This procedure results in a large choice set that makes calibration of bundle choice model computationally challenging. Current proposed approaches simplify the estimation task based on behavioral assumptions but ignore the interdependence among products. We propose a modeling approach that deals with the challenges of configuration data and provides clear insights into customer preference segmentation. We show that the configuration data can be represented by a Multivariate Logistic (MVL) choice model, given certain assumptions about the individual choice process and preference heterogeneity. We also develop a procedure for determining optimal bundle design that takes into account the preference structure and price sensitivity. We provide an empirical illustration of the proposed modeling process that illustrates both the strengths and weaknesses of configuration data. 

9:30 - 10:00 AM  Monetizing Ratings Data for Product Research
Nino Hardt, Assistant Professor of Marketing, Ohio State University
Alex Varbanov, Principal Scientist, The Procter & Gamble Company
Greg M. Allenby, Professor of Marketing, Ohio State University

Features involving the taste, smell, touch, and sight of products, as well as attributes such as safety and confidence, are not easily measured in product research without respondents actually experiencing them. Moreover, product researchers often evaluate a large number of these attributes (e.g., >50) in applied studies, making standard valuation techniques such as conjoint analysis difficult to implement. Product researchers instead rely on ratings data to assess features for which the respondent has had actual experience. We demonstrate a method of monetizing rating data to standardize product evaluations among respondents that improves purchase predictions by about 20% relative to existing methods of scale adjustment, requires only a few additional survey items and is easy to estimate.

10:00 - 10:15 AM  Discussion

10:15 - 10:45 AM  BREAK
10:45 - 11:15 AM  Sampling in the River Twitter: A DIY Tool for Social Media Tracking via the REST API
Jack Horne, Owner Jack Horne dot Net
Jane Tang, SVP Advanced Analytics, Maru/Matchbox

We describe a low cost and reliable sampling solution for social media monitoring for Twitter, using Python scripting and automation via virtual Amazon Web Services.   We illustrate the use of this DIY tool, incorporating Twitter buzz data in brand tracking via Kalman Filter.  

11:15 - 11:45 AM  Deep Architectures for Consumer Heterogeneity
Sanjog Misra, Professor of Marketing, University of Chicago Booth School

Ubiquitous data comes with challenges but also with opportunities. One key problem is the modeling of consumer heterogeneity in the realm of big data. In this project I outline a proposal to revisit the issue of modeling consumer heterogeneity by combining traditional models of consumer behavior with targeted deep learning on key parameters of interest. The approach is theoretically simple, significantly more flexible and practically easier to implement than current approaches to dealing with heterogeneity. I demonstrate this approach in the context of two real case studies, one involving linear models and the other with discrete choice models.

11:45 AM – 12:00 PM  Discussion​

12:00 - 1:00 PM  LUNCH
1:00 - 1:30 PM  Perceptual Choice Experiments: Enhancing CBC to Get from Which to Why
Bryan Orme, President, Sawtooth Software

Typical CBC market simulators predict choice likelihoods for multiple product concepts within competitive market scenarios; but they provide no insights into perceptions—the why’s behind choice. We show how to augment CBC questionnaires to add predicted agreement on perceptual/diagnostic dimensions to simulators, by product, as an interactive heat-map.  The experimental design follows standard DOE theory. The modeling may be done with any software for MNL.  The simulator and heatmap are easily built in Excel.

1:30 - 2:00 PM  Toward Reconciling Stated and Revealed Preferences
Greg Allenby, Professor of Marketing and Statistics, Ohio State University

Conjoint studies are widely used in practice to predict demand for products and product configurations that may not currently exist.   We investigate the consistency between stated (conjoint) and revealed (transaction data) preferences using a unique conjoint dataset from individuals for whom transaction data is also available.  Using a volumetric demand model, we find consistency between stated and relative preferences for brands and product attributes in the frozen pizza category, while estimates of satiation, the budgetary allotment and preference for the outside good are not as well aligned.  We explore the use of screening rules to better reconcile differences in the datasets, and find that estimates from these models are more divergent than those from simpler models.  Implications for the measurement of consumer demand and future research opportunities are discussed. 

2:00 - 2:15 PM  Discussion

2:15 - 3:00 PM  BREAK

3:00 - 3:30 PM  Have You Met Stan Yet? (Demo)
Elea McDonnell Feit, Assistant Professor, Drexel University and Senior Fellow, Wharton Customer Analytics Initiative
Kevin Van Horn, Senior Data Science Engineer, Adobe

Stan (http://mc-stan.org/) is a new open-source tool for estimating complex statistical models using MCMC and is championed by a number of contributors including Andrew Gelman and Bob Carpenter. Similar to JAGS and BUGS, Stan allows users to specify their own models using a relatively simple modeling language. Once the model is specified, Stan automatically generates a routine to sample from the posterior, making is possible for both novices and experts to quickly formulate and test new models. In this demonstration, we will show how to use Stan to fit the hierarchical choice models typically used for conjoint along with an example of a custom-built nested logit model for conjoint with a no-choice option.

Bayesian Price Sensitivity Meter (PSM) with Excel (Demo)
Aku-Ville Lehtimäki, Doctoral Candidate, Department of Information and Service Economy, Aalto University School of Business
Outi Somervuori, Ph.D., Department of Information and Service Economy, Aalto University School of Business

Price Sensitivity Meter (PSM) is a widely used yet simple pricing technique. The respondents are asked what would be their idea of the price on (four) different ordinal levels (from too cheap to too expensive). However, in its current form the technique used only as presenting descriptive statistics. We show that PSM can be enhanced using statistical inference and Bayesian paradigm. Furthermore, we show that after the improvement the tool can be still implemented on any spreadsheet software (e.g. Microsoft Excel).

3:30 - 3:45 PM  kNN Logit Ensembles for CBC Studies (Demo)
Dmitry Belyakov, Research & Analytics Team Leader, Ipsos Comcon

This presentation demonstrates how an ensemble of diverse solutions can be created using a method similar to kNN. The algorithm proposed defines similar respondents, combines them in various combinations and then runs logit estimation on the group data. This approach allows us to obtain both stable and diverse estimates.

3:45 – 4:00 PM  Demo Q&A

4:00 - 4:30 PM  Charles Coolidge Parlin Marketing Research Award
The Charles Coolidge Parlin marketing research award was established in 1945 by the Philadelphia Chapter of the AMA and The Wharton School in association with the Curtis Publishing Company to honor persons who have made outstanding contributions to the field of marketing research.  Established as a memorial to Charles Coolidge Parlin, who is recognized as a founder of marketing research, The Parlin Award is today a preeminent national honor.  To be awarded the Charles Coolidge Parlin Marketing Research Award, distinguished academics and practitioners must have demonstrated outstanding leadership and sustained impact on advancing the evolving profession of marketing research over an extended period of time. 

4:30 - 5:00 PM  Discussion

5:00 - 6:00 PM  Networking Reception with Poster Session and Exhibits
Tuesday, June 27, 2017

8:50 - 9:00 AM  Day Two Kickoff
9:00 - 9:30 AM  When Words Sweat: Identifying Signals for Loan Default in the Text of Loan Applications
Michal Herzenstein, Associate Professor of Marketing, University of Delaware
Oded Netzer, Associate Professor of Marketing, Columbia University
Alain Lemaire, Doctoral Candidate in Marketing, Columbia University

We present empirical evidence that borrowers, consciously or not, leave traces of their intentions, circumstances, and personality traits in the text they write when applying for a loan. This textual information has a substantial and significant ability to predict whether borrowers will pay back the loan over and beyond the financial and demographic variables commonly used in models predicting default. We use text-mining and machine-learning tools to automatically process and analyze the raw text in over 18,000 loan requests from Prosper.com, an online crowdfunding platform. We find that loan requests written by defaulting borrowers are more likely to include words related to their family, mentions of god, short-term focused words, the borrower’s financial and general hardship, and pleading lenders for help. We further observe that defaulting loan requests are often written in a manner consistent with the writing style of extroverts and liars. 

9:30 - 10:00 AM  Digital Response to Traditional Advertisement: Toward a Deeper Understanding of TV Ads and Post-ad Search Spikes
Rex Y. Du, Marvin Hurley Professor of Marketing, University of Houston
Linli Xu, Assistant Professor of Marketing, University of Minnesota
Kenneth C. Wilbur, University of California, San Diego

Advertisements in traditional media such as TV lead some multitasking viewers to take immediate, measurable actions online. We analyze minute-by-minute TV ad insertion and online search data for daily fantasy sports and pick-up truck brands. TV ads with small audiences can produce detectable post-ad search spikes for the advertised brand, with 75% of incremental searches occurring within two minutes. TV ads also generate post-ad searches for competitor brands. Search spikes vary with ad content: they are larger after brand-focused ads than after price-focused ads, and after less-informative ads than after more-informative ads. Search spikes also vary with contextual media factors (e.g., TV network, daypart, program genre), but their effects differ across brands. Taken as a whole, our findings suggest that marketers should consider post-ad search spikes in conjunction with other metrics when evaluating and purchasing TV ads.

10:00 - 10:15 AM  Discussion
10:15 - 10:45 AM  BREAK

10:45 - 11:15 AM  Advances in Card Sorting Data Collection and Analyses (Demo)
Simon Blanchard, Assistant Professor of Marketing, Georgetown University

Card sorting tasks provide a flexible research methodology to study how consumers see products/brands as going together. In this methods presentation, recent developments in ways to administer sorting tasks will be discussed. Then, SumPi, a software package for the analysis of sorting task data, will be demonstrated.

Preference Based Conjoint: Can it Predict Dozens of Products? (Demo)
Jeroen Hardon, Director Methodology and Innovation EU, SKIM
Marco Hoogerbrugge, Research Director, SKIM

Conjoint analysis is often used for very complex markets, with a large number of attributes and products in the market. Ideally we would replicate the complexity of the market that exists in reality as good as we can in the design of the conjoint survey. Although we already proved in the past that having more reality makes the predictions better, this is not always feasible. Most particularly, when there are quite a few attributes, we cannot show too many concepts in a single task as this would lead to information overload for the respondent. The key question in this presentation is to check what extent the discrepancy between the number of concepts in a task (3, or 4) versus the number of products in the conjoint simulation model (20, or 50) may influence the results of the study, by comparing CBC, ACBC and PBC (preference based conjoint).

11:15 - 11:30 AM  Web-Based Conjoint Simulators (Demo)
Jake Lee, President, Quantum Strategy, Inc

Interactive data analysis tools have made major leaps in the last few years. In particular, a package called Shiny made by RStudio is opening up new opportunities for conjoint simulators. This presentation will showcase what you can do with a shiny simulator and highlight some of the major challenges to getting started.    

11:30 – 11:45 AM  Demo Q&A

12:00 - 1:00 PM  LUNCH

1:00 - 1:30 PM  Optimizing Marketing Campaigns Using SQL Server and R

Sheri Gilley, Senior Software Engineer, Microsoft Corporation
Xinwei Xue, Principal Data Scientist Manager, Microsoft Corporation
Carl Saroufim, Data Scientist, Microsoft Corporation
Bharath Sankaranarayan, Lead Product Manager, Microsoft Corporation

This session will provide a walkthrough of how you can optimize lead conversions.  Marketing campaigns are not only about what you say, but also when you say it.  Effective campaigns driven by advanced analytics systematically test and learn delivery timing to optimize open rates, click through rates and conversion rates.  This solution demonstrates how you can leverage the power of R combined with SQL server to perform predictive analytics.  The solution was developed by the Microsoft R Server Product Team indicated above.​

1:30 - 2:00 PM  Using Dynamic Factor Analysis to Explain and Forecast Multivariate Response Variables: An Application to Survey and Social Media Metrics
Kirk Ward: Research Methods and Offer Innovation, Kantar/TNS
Ziad Elmously: Marketing Science Center, Kantar TNS

Time-series data collected via surveys and on social media venues often comprise numerous overlapping and noisy metrics.  Dynamic Factor Analysis (DFA) provides a viable framework for consolidating time-series data and mitigating their temporal volatility.  Applying DFA to a simulated dataset and Kantar tracker surveys, we extract a small number of common trends that streamline and measure the covariation among a large number of time series and their associations with explanatory variables.  The factor loadings quantify the contribution of the common trends to the individual time series and divide them into a smaller number of groups.  We discuss potential extensions such as spatial smoothing in order to capture geographic dependencies and peculiarities.  In summary, we demonstrate the application of DFA as a method to shorten data collection instruments, temper the volatility of data collected across time as well as space, capture metrics’ geographic nuances, and forecast a multitude of time series.

2:00 - 2:15 PM  Discussion

2:15 - 3:00 PM  BREAK

3:00 – 3:30 PM  Details to follow 
3:30 – 3:45 PM  Q & A

3:45 - 4:00 PM  BREAK

4:00 - 4:30 PM  Details to follow​

4:30 - 5:00 PM  Roundtables

5:30 PM  Optional Dine-arounds

Wednesday, June 28, 2017

8:50 - 9:00 AM  Day Three Kickoff

9:00 - 9:30 AM  Iterative Multilevel Empirical Bayes (IMEB): An Efficient Flexible and Robust Solution for Large Scale Conjoint
Kevin Lattery, Vice President Methodology and Innovation, SKIM

The typical approach to conjoint analysis uses Hierarchical Bayes with Gibbs Sampling to integrate over an upper level multivariate normal.  This works well with small to medium size data, but with many parameters may be inadequate or impractical.  We describe an approach that fixes the upper level prior based on the data (hence Empirical Bayes) and cross-validation of a scale parameter.  This fixed prior eliminates MCMC iterations, making it much faster than standard HB.  The Empirical Bayes approach also avoids issues of inadequate convergence and improper scaling that may occur with MCMC.  We also describe some of the additional flexibility of this approach, including respondent level customization and the ability to easily estimate utilities for additional respondents.  
9:30 - 10:00 AM  Modeling Consumer Choices with Revealed Preference Data: a Big Data Challenge
Ewa Nowakowska, Senior Director, Data Science, SUPERCRUNCH by GfK
Jakub Glinka, Senior Data Scientist, SUPERCRUNCH by GfK

Ever increasing amounts of data are being collected on product choices made by consumers in the market place. This data is not only larger in volume but is also different in nature from stated preference data traditionally leveraged in market research. This paper examines German TV market data to highlight modeling challenges posed by revealed preference data as well as to show how they may be addressed. The authors present a scalable implementation of nested logit, combined with BASS innovation diffusion model, to explain choices consumers make in the technical products market.

10:00 - 10:15 AM  Discussion

10:15 - 10:30 AM  BREAK

10:30 - 11:00 AM  Details to follow

11:00 - 11:15 AM  Best Paper - Adjourn

Sunday, June 25 – Tuesday, June 27, 2017

Poster Sessions

Bayesian Price Sensitivity Meter (PSM) with Excel
Aku-Ville Lehtimäki, Doctoral Candidate, Department of Information and Service Economy, Aalto University School of Business
Outi Somervuori, Ph.D., Department of Information and Service Economy, Aalto University School of Business

Price Sensitivity Meter (PSM) is a widely used yet simple pricing technique. The respondents are asked what would be their idea of the price on (four) different ordinal levels (from too cheap to too expensive). However, in its current form the technique used only as presenting descriptive statistics. We show that PSM can be enhanced using statistical inference and Bayesian paradigm. Furthermore, we show that after the improvement the tool can be still implemented on any spreadsheet software (e.g. Microsoft Excel).

Computational Intelligence in Portfolio Optimization
Peter Kurz, Head of Research and Development, KANTAR TNS

The work compare particle swarm optimizer, ant colony optimization, genetic algorithm simulated annealing and a multi-verse optimizer concerning computational performance and finding the global optima in the context of portfolio optimization based on discrete choice experiments. 
The presented results are based on simulated and  real empirical datasets conducted  from KANTAR TNS in Germany. All datasets represent a nested attribute structure. The number of possible concepts (search space) varies between 4.2*104 and 4.57*1017. The results show that the algorithms are very equal in computational speed and mostly find the optimal solutions in case of our simulated data sets. The optimization runs on our empirical data sets show a different picture, the optimal portfolios are relatively equal, but in most cases it seems that the optimization only results in  local optima. Finally we present some first suggestions which optimizer can successfully be used under which conditions. 

Direct Estimation of Key Drivers from a Fitted Bayesian Network

Ben Cortese, Ph.D., Statistician – Knowledge Systems and Research (KS&R)

Key driver analysis (KDA) is a technique to identify a subset of attributes, known as key drivers, which have a strong impact on a target attribute.  There are a wide array of techniques to estimate attribute level driver scores, but those most commonly used are unable to provide information about the interactions between drivers.  Instead, we propose a new algorithm, Bayesian network key driver analysis (BNKDA), to calculate driver scores directly from a fitted Bayesian network.  This technique provides both the directed acyclic graph (DAG) and corresponding driver scores to tell a cohesive story about key drivers to a target attribute.  The algorithm is compared to two widely adopted driver analysis methods – Kruskal’s relative importance and partial least squares path modeling (PLSPM) – through simulation studies.

Endorsing Risk in Advertising
Minghai YE, Dr., Prof., Director of Department of Business Administration, School of Economics and Management, Tongji University
Yipeng SHA, Doctoral Candidate for Marketing, School of Economics and Management, Tongji University
Wei LI, Doctoral Candidate for Marketing, School of Economics and Management, Tongji University

The use of celebrity endorsers is prevalent in advertising, but it is not without risk. Negative information about a celebrity has negative impact on consumer evaluations of endorsed brands. This study plans to collect data of negative celebrity information (including illegal events like drug-taking and unmoral events like derailment or fighting) during 2011 to 2016 in China and compare consequences from social attendance and marketing activities. The research uses big data technique to collect data and get a new dataset to examine endorsing risk in advertising. The research is just underway and welcome any suggestions, comments, recommendation and insights.

Patient Typologies as Sequences: An Optimal Matching Analysis of Disease and Treatment Trajectory
John Hartman, Head- Predictive Analytics, Phoenix Marketing International
Sandy Balkin, Senior Director- Global Insights and Analytics, Sanofi

We employ an optimal matching approach based on sequence analysis to derive dissimilarity distance metric between each pair of patient sequences. We use these “distances between sequences” as the input to generate a typology of diabetic patients. Once the patient typologies have been distilled, the patient dataset is leavened with physician level data to provide the inputs for a multi-level model of therapeutic agent choice. The overall analysis allows us to locate physician-specific agent preferences in patient care.  

Already registered on site? Please log in first in the upper right corner of the site before proceeding.

Select Attendee Type:




Select Event Tutorials:

Sunday - 8:00 AM - 12:00 PM
Non-Members : $345.00
Members : $320.00
Non-Members : $345.00
Members : $320.00
Non-Members : $345.00
Members : $320.00
Non-Members : $345.00
Members : $320.00
Sunday - 1:00 PM - 5:00 PM
Non-Members : $345.00
Members : $320.00
Non-Members : $345.00
Members : $320.00
Non-Members : $345.00
Members : $320.00
Sunday - 1:00 PM - 2:55 PM
Non-Members : $172.00
Members : $160.00
My Selected Tutorials:
Non-Member $
Member $
Base Price For The Event

Conference Location

Hyatt at Olive 8
1635 8th Avenue
Seattle, WA 98101
Telephone: +1.206.695.1234
Reservations: +1.888.421.1442

The luxury downtown Seattle hotel features upscale amenities and chic accommodations with beautiful views of the city. Located near the Seattle Seahawks stadium, the spaces exude a Pacific Northwest spirit and offer guests an eco-friendly environment for relaxing, dining, and exploring. Hyatt at Olive 8 is the new standard for green luxury hotels in Seattle.
The stunning hotel is walking distance to Seattle’s most famous attractions, shopping and restaurants. A few blocks outside the magnificent glass doors are historic Pike Place Market and the Puget Sound.

Room Rates:
$259.00 king bed, single/double occupancy per night
$274.00 two beds, single/double occupancy per night
Rates are subject to applicable state and local taxes, currently 15.6% and $2.00 tourism assessment fee (subject to change without notice).

Rates are based on availability.
Please reserve your overnight accommodations early as there are a limited number of AMA special rates available.

Reserve a Room Now

Or call reservations at +1.888.421.1442 and ask for the American Marketing Association room block.

The reservation cut-off date is Thursday, May 25, 2017.
The AMA has negotiated special group rates for the ART Forum. You must make your hotel reservations by May 25, 2017 and properly identify yourself (American Marketing Association) to qualify for the special group rate, based on availability. After May 25, 2017, the rooms will be released for sale to the general public, and all reservations will be accepted on a space available basis at the hotel's available rate - not at the group rate. A credit card or deposit in the amount of (1) one night’s room and tax for each confirmed room will be due at time of booking. Cancellations made after 3:00 pm PST the day prior to arrival and no-shows will forfeit the initial deposit. If you fail to arrive on your confirmed arrival date, your entire reservation will be cancelled and your credit card will be charged for one night’s room and tax.

Check in time is 4:00 pm, and check out is at 12:00 pm. Anyone arriving earlier than 4:00 pm will be checked in as soon as a room(s) becomes available. Guests checking out early may be assessed an early departure fee. Upon check in, guests will be asked to verify their departure date. At that time, scheduled departure dates may be altered.

Hotel Features, Services and Accommodations
Please visit Hyatt at Olive 8 for a full list of services and features.

Area Attractions and Events
Please go to Visit Seattle for additional city information.

Conference Attire
Conference attire is business casual. Meeting rooms tend to be cool so you may wish to bring a sweater/jacket.

The AMA is committed to providing equal access to our meetings for all attendees. If you are an attendee with a disability and require program accommodations, please contact the AMA Meeting Services Department, and a member of our staff will ensure that appropriate access arrangements are made. If you have specific disability related needs for your hotel sleeping room, please be sure to communicate those directly to the hotel when you make your reservation. In an effort to provide the highest quality of service to all attendees, we require that details of all access requests be communicated to our office at least 14 days in advance of the beginning of the meeting.

AMA Travel Program

American Marketing Association has partnered with Delta to provide our attendees a 2% - 10% discount for the ART Forum in Seattle, WA (SEA). The Delta Meeting Network discount is based on the booking class of your ticket.

Reserve your ticket now​ 
or call Delta Meeting Network Reservations at +1.800.328.1111.  When booking online, select Book A Trip, click on Advanced Search and enter the meeting event code: NMPS2 in the box provided on the Find Flights page.

Reservations may also be made by calling Delta Meeting reservations at +1.800.328.1111 Monday through Friday 7:00 am - 7:00 pm CDT.
*Please note that a Direct Ticketing Charge will apply for booking by phone through the reservation number above.

Tower Travel Management
AMA's travel coordinator, Tower Travel Management, is available to assist with reservations. Call +1.800.542.9700 within the U.S. and Canada. Reservation lines are open Monday through Friday 7:00 am - 7:00 pm CDT or, you may contact them via email at association@towertravel.com.

Tower Travel will proactively research airfares on ALL airline carriers to ensure that the lowest available fares are offered to all attendees. They are dedicated to providing superior customer service and hassle-free travel arrangements. Please note that fees, restrictions and cancellation penalties will apply.

From Sea-Tac Airport:
Follow signs I-5 Northbound. Continue on I-5 North for 14 miles. Exit from the left lane at Seneca Street. Right on Sixth Avenue. Right on Pike Street. Left on Eighth Avenue. Continue forward through Pine Street. Hyatt at Olive 8’s entrance is the second driveway on the left.

From I-5 North Bound:
Exit from the left lane at Seneca Street. Right on Sixth Avenue. Right on Pike Street. Left on Eighth Avenue. Continue forward through Pine Street. The entrance to our downtown Seattle WA hotel is the second driveway on the left.

From I-5 South Bound:
Take the Union Street Exit. Continue forward onto Union Street. Turn right on Seventh Avenue. Right on Pike Street. Left on Eighth Avenue. Continue forward through Pine Street. The Hyatt at Olive 8 entrance is the second driveway on the left.

From I-90 West Bound:
Merge onto I-5 North via Exit #2c toward Madison Street/Convention Center/Vancouver, BC. Take Madison Street/Convention Center exit. Follow exit forward onto Seventh Avenue. Right on Madison Street. Left on Eighth Avenue. Continue forward through Pine Street. The entrance to our Seattle WA downtown hotel is the second driveway on the left.

Taxis are readily available at Sea-Tac International Airport. The trip to the hotel is approximately $55.00. The return trip to the airport from downtown Seattle is a flat rate of $40.00.

Shuttle Service
Airport Express provides shared shuttle service to and from Sea-Tac Airport for $18.00 per person, one-way. Advanced reservations are required for trips to the airport, and recommended for transfers from the airport (but not required). Please visit Airport Express to make reservations or call +1.425.981.7000.

Public Transportation
Link Light Rail

Parking at the Hotel
Valet parking is available at the hotel for a discounted rate of $50.00 per night (for AMA attendees), including in and out privileges. Rates are subject to change without notice. 

​Sponsor & Exhibition Opportunities: 

To learn more about AMA Sponsorship opportunities, please contact Andrew Nelmes at anelmes@ama.org or call 312.542.9088.​

Customized partnerships are also available!

Become a Member
Access our innovative members-only resources and tools to further your marketing practice.

 Upcoming Webcasts


 Why Attend


"The ART Fourm has helped me gain proficiency in hierarchical Bayes methods, segmentation, and probability models.  I apply these methods in every project I conduct!"
Past ART Forum Attendee

"The ART Forum helped me develop from a textbook statistician to a creative market researcher."
Past ART Forum Attendee

 "The ART Forum provides a great opportunity to network with marketing research practitioners that are working on interesting problems."
Past ART Forum Attendee


 Join the AMA

​The AMA connects you to a world of resources that deliver results, and help you succeed today and into the future. Join the AM, and put the power of AMA membership to work for you.