Yesterday a PhD student posted a series of questions regarding the analysis of the choice of acquiring and using a car on the mailing list of the UTSG (universities’ transport studies group). He had spoken with experts in marketing and psychology who claimed (a) that this choice is both rational (travel utility – price, time, etc) and emotive (symbolism of a brand for an individual and their self image), and (b) that transport analysis — and more specifically analysis using discrete choice models — only considered the rational side of things, ignoring the emotive. With his post he wanted to canvass the views of transport academics on these matters, and asked three questions:
- Is (b) above true and discrete choice modelling deficient in the way it approaches matters such as ‘symbolism of a brand’ or ‘aversion to a mode’?
- If it isn’t, how does discrete choice modelling account for such things?
- Has any work actually quantified the % split between the ‘rational’ and the ’emotive’ when choosing to acquire and/or use a car. Could this percentage vary by culture?
These are interesting questions and I couldn’t resist responding. This is (and edited version of) what I wrote:
I am afraid that the experts you have spoken to are not aware of recent developments in discrete choice modelling in transport studies. There is a growing number of empirical studies using discrete choice models to account for what you term ’emotive’ factors (more on this below). These studies tend to adopt one of the following approaches:
- Include measures of people’s attitudes, beliefs and values directly into the utility function — a good example of this line of work is provided by the work of Pat Mokhtarian at UC Davis. For an example of a study of vehicle choice, see her co-authored paper ‘What type of vehicle do people drive?‘ published in 2004 in Transportation Research A.
- Use a latent class model whereby people are endogenously classified into groups with similar preferences/attitudes as part of the choice model. I don’t have an example of this approach in the context of vehicle type choice at hand, but this approach is increasingly used (see e.g. Walker and Li’s 2007 paper ‘Latent lifestyle preferences and household location decisions‘ in Journal of Geographical Systems) and is directly applicable to decisions about what sort of vehicle people drive.
However, there is also a problem with the way the three questions are framed. There is a long history of dualistically opposing the ‘rational’ and the ’emotional’ in mainstream western thought, but evidence from a wide variety of sources suggests that this opposition is in itself very problematic: it is not possible to separate the two. In the neurosciences, for instance, it is now increasingly accepted that emotions are indispensable to good, or at least effective, decision-making — the work of Antonio Damasio is a key source here. In fact, one could argue that the opposition of the rational and the emotional is a somewhat curious particularity of the vagaries of mainstream western philosophy, and certainly not shared by all the mainstream philosophers of the past centuries. David Hume and some of the utilitarians, for instance, already acknowledged that the rational and the emotional were entangled in complex ways in (good) decision-making. These arguments imply that your third question is very problematic from a philosophical point of view — the question cannot be addressed in a meaningful way if one also wants to incorporate non-Western countries/cultures.
At the same time, there is much to say in favour of the criticisms articulated by the experts from marketing and psychology, but the key here is to distinguish between the ‘rational’ and the ‘instrumental’. In mainstream discrete choice modelling of travel behaviour, decisions regarding vehicle ownership, and so on, there has long been a tendency to privilege the instrumental aspects of behaviour — e.g. how can I get to X the quickest, at the lowest monetary costs and with the highest level of comfort? This one-sided focus on the instrumental has a long and complex genealogy and derives in part from theoretical considerations but also reflects data availability issues and the difficulty of constructing reliable measurements of the more-than-instrumental factors that mediate people’s decision-making.
The ‘expressive’ dimensions of behaviour and decisions about vehicle ownership have long been side-lined in transport modelling, and until fairly recently there has been little attention for the links between transport and identity (and symbolism) and for the links between transport/movement, sensory experience, affects and feelings. The most powerful work on those expressive dimensions can nonetheless be found outside the body of research drawing on discrete choice models; two strands of work would be particularly useful to consider:
- Research informed by thinking from behavioural psychology — see e.g. the work of Linda Steg (among others her 2005 paper ‘Car use: lust and must‘ in Transportation Research Part A)
- Research belonging to what John Urry and Mimi Sheller have called the ‘new mobilities paradigm’ — the best place to start here would be the work of Peter Merriman (his 2009 paper ‘Automobility and the geographies of the car‘ in Geography Compass provides an excellent introduction to this line of thinking)
Considering the more-than-instrumental is important in choice modelling; models that only consider the instrumental dimensions of individuals’ decision-making — e.g. travel time, travel cost, comfort, reliability — seem to systematically over-predict the extent to which people will change their behaviour or choices compared to what happens in the real world. The more-than-instrumental must be given much more attention if we want to move towards more sustainable transport systems.