Pacific Southwest Region 9 UTC

PSR Co-Hosts Discussion and Criticism of Choice Modelling

Monday, August 13, 2018 - 8:02pm

By Lin Zhu, Master of Transportation Engineering, USC

On Monday, July 23th, 2018, at USC’s Verna and Peter Dauterive Hall, students, faculty, staff, and practitioners  had the pleasure of listening to the presentation of Stephane Hess, a Professor of Choice Modelling at the Institute for Transport Studies and Director of the Choice Modelling Centre at the University of Leeds, present, “From Behaviour to Data to Models and Back Again: Where Next for Understanding and Predicting Human Decision-Making?”  This event was part of the METRANS Pacific Southwest Region University Transportation Center (PSR) 2018-2019 Research Seminar Series and was cohosted by the Leonard D. Schaeffer Center for Health Policy and Economics.

“I’ve been active in choice modelling [increasingly] over the years” said Hess, “and that increase is really correlated with increased computational power - we can now estimate models in a few minutes.” Hess added that at the same time as this increase our traditional thinking is being challenged by ideas from behavioral economics and mathematical psychology and that important work remains to be done on improving the quality of our modelling forecasts.

“After early work in psychology in the 1920s and major econometric developments in the 1960s and 1970s, the mathematical modelling of human decision-making has evolved substantially – most notably in recent decades,”  Hess shared.  “Novel model structures have been put forward, alternative data sources have been explored, and the methods have gained a foothold in new application areas, notably health and environmental economics. At the same time, our traditional thinking is being challenged by ideas from behavioral economics and mathematical psychology, and important work remains to be done on improving the quality of our forecasts. Developments in one area of application also regularly fail to make the transition into other fields, and the impact of data quality on results is often not discussed.”

“I draw on my experience of working in both theoretical and applied choice modelling across different fields,” he added, “and my research presents a critical view of the reliance on cross-sectional (and often hypothetical) choice data, makes the case for moving beyond treating individual decisions in isolation, and discusses the potential benefits and pitfalls of moving away from models grounded in economic theory.”

Professor Hess put forward three criticisms of choice modelling. First, its reliance on hypothetical data. He explained that choice modelling can work very well for goods that users have experience buying, but for unfamiliar contexts we cannot obtain reliable valuations from hypothetical data.

Professor Stephane Hess addressing PSR attendees

Second, choice modelling relies on cross-sectional data. Hess shared a model which shows a negative correlation between age and travel - if one is older they generally travel less- and negative correlation between cost and travel - if travelling is cheaper it is generally performed more frequently. Yet, these trends do not reveal individual preference or actions. As the vast majority of data sets and models are estimates on cross-sectional data the findings attributed to differences across people are not necessarily relevant for a single individual over time.

Lastly, Hess shared that there is an inherent issue with treating choices in isolation. In reality, the choices that we make are very interrelated. Often, choice modelling asks participants to state their preferences directly; however, an analysis of choice modelling over the last 15 years shows the practice has become too reliant on stated preference data.

In contrast to stated preference data, choice modelling can utilize revealed preference data. In this model, the buying preferences of subjects is revealed through their purchases as opposed to being directly listed or stated by the subject.

Overall, the opportunity to listen to a knowledgeable professor on this intriguing, and at times complex matter, was well received by the students and transportation experts in attendance, and I personally thank both our presenter and our hosts for making this opportunity available.

Stephane Hess is Professor of Choice Modelling in the Institute for Transport Studies and Director of the Choice Modelling Centre at the University of Leeds. He is also Honorary Professor in Choice Modelling at the University of Sydney and has a director position at RSG, a leading North American consultancy company. His area of work is the analysis of human decision-making using advanced discrete choice models, and he is active in the fields of transport, health and environmental economics. Hess has made contributions to the state of the art in the specification, estimation and interpretation of such models. With over 100 peer reviewed journal papers on the topic, his contributions have been recognized for example by the 2014 Outstanding Young Member of the Transportation Research Board (TRB) award for exceptional achievements in transportation research, policy, or practice, the 2010 Fred Burggraf award handed out by the Transportation Research Board, and the 2005 Eric Pas award for the best PhD thesis in the area of travel behavior modelling. He is also the founding editor in chief of the Journal of Choice Modelling and the founder and steering committee chair of the International Choice Modelling Conference.

The Leonard D. Schaeffer Center for Health Policy & Economics, one of the nation’s premier policy centers, measurably improves value in health through evidence-based policy solutions, research and educational excellence, and private- and public-sector engagement.  The Center was established in 2009 at the University of Southern California through a generous gift from Leonard and Pamela Schaeffer.  This gift reflects Mr. Schaeffer’s lifelong commitment to solving healthcare issues and transforming the healthcare system. Learn more about the Schaeffer Center at https://healthpolicy.usc.edu/

 

 

 

About the Author:

Lin Zhu is a first year Graduate Transportation Engineering student at USC. She ipreviously worked for Airport Transportation Management Department in China and currently the Opportunities Editor for METRANS. She is hoping to have a career in Intelligent Transportation System and can be reached at zhu113@usc.edu