How we choose: Applying 'decision science' to transportation behaviors

Can scientists understand human behavior enough to figure out what drives the choices you make? In fact, it's called "decision science," and it's something that Anna Spurlock, a behavioral economist with Lawrence Berkeley National Laboratory (Berkeley Lab), specializes in.

Spurlock spearheads the WholeTraveler Transportation Behavior Study, a three-year project that has attempted to analyze why and when some people adopt certain technologies-such as electric vehicles, ride-sharing, ride-hailing (like Uber and Lyft), and online shopping-while others don't.

The study is part of the SMART (Systems and Modeling for Accelerated Research in Transportation) Mobility consortium, which is a multiyear consortium of several national labs developed to further understand the energy implications and opportunities of advanced mobility technologies and services. The SMART Mobility Consortium consists of five pillars of research: Connected and Automated Vehicles, Mobility Decision Science, Multi-Modal Freight, Urban Science, and Advanced Fueling Infrastructure, and is funded by the Department of Energy's (DOE's) Vehicle Technologies Office (VTO) Energy Efficient Mobility Systems (EEMS) Program.

The WholeTraveler study started with an online survey in 2018-over 1,000 San Francisco Bay Area residents responded. The survey included questions around car ownership, commute locations, demographics, personality traits, and a life-history calendar that looked at travel behaviors related to major life stages and events between the ages of 20 and 50. The survey results provided Berkeley Lab researchers with a treasure trove of data and was a major cornerstone of the Mobility Decision Science pillar of SMART Mobility.

I study how people make decisions about energy-related topics, such as energy-efficient appliances or products, time-of-use utility pricing programs, or transportation. How do they trade off decisions on energy or cost or different factors? And what are the implications of those decisions? It's taking the domains that matter for the energy "pie," and of those domains, figuring out what's driven by people's behavior and how can we understand the behavior that underlies it.

Some of what I do might relate to concepts that come from psychology, but when talking about behavior, a lot is data-driven. For a lot of the work I do, we have some data that either directly observes people's choices or things they've done, or sees the implications of it. We use machine-learning techniques to derive and reveal patterns, and also use statistical and econometric analysis to test hypotheses.

When we started assessing the literature, we found there was very limited data that tracked people over a long period of time. Longitudinal surveys are very expensive to do. But with our life-history calendar, we were able to get at that. We could ask about their choices at the shorter-term time frame, like the day to day; at the medium-term time frame, like choices about what vehicle they own or whether they own a vehicle; and at the longer-term time frame, like where they live and whether they have kids, and understand how they're interrelated.

There's a lot of need for better understanding the dynamic relationship between longer-term decisions and life transitions that may affect transportation choices. What life events trigger changes in transportation behaviors, and for whom? How permanent or flexible are those changes? What types of solutions could result in a more energy-efficient transportation system if we can understand what's behind certain behaviors in this holistic sense, as well as the barriers to other types of options?

We also found there was very little survey data on some of the emerging transportation technologies and services, such as connected and automated vehicles, e-commerce and delivery, ride-hailing, and ride-sharing. The overarching purpose of SMART Mobility is to understand the system-level implications of these technologies and services and how they'll change people's behavior-and how that will have implications on the transportation system. We wanted to cover all of the key emerging transportation innovations and their relationship to multiple facets of transportation behaviors, which is why we called it the WholeTraveler study.

There were a couple things. For one, the results we're getting from the life-history calendar data is something we're getting a lot of interest in from the transportation research community. This is an area that was understudied. We're getting insights into how key life events-such as finishing school, partnering up, having children-related to your transportation choices.

For example, we found that for people who had their children in a middle range of about 26 to 32 years old, having a child is associated with a statistically significant increase in the probability they'd be regularly driving. But if they had their children young-under 26-they were less likely to be regularly driving. And if they had their children older, having a child had no impact on the rate of driving. When we dug into the underlying patterns we found that-and this isn't a huge surprise-as people age there's a tendency towards stronger car dependence. Those having children for the first time over 32 were already pretty car-dependent, whereas for those having their first child between 26 and 32, the child triggered a faster transition to more frequent driving. For those having their children young, on the other hand, they were less likely to be working full-time as a result of having children, and so were less likely to be regularly driving.

And related to that, we found that once people hit a certain level of car dependence, the habit is very persistent. This was already somewhat known, but we showed it in a new way. So when thinking about it from a policy perspective, when you're projecting people's behavior, some of the strengths of these patterns of persistence can be important for modeling those patterns appropriately.

There are all these consultant reports out there with a starry-eyed view of these transportation innovations. Some say things like, the rate of car ownership by 2030 will be cut in half, or 95% of people by x year will be relying on services such as ride-hailing and ride-sharing.

But I see the patterns that relate to things like children and the strength of trends toward car dependence, and it makes me skeptical that those types of projections could be realistic. There are real barriers for some people to shedding or not relying on a personally owned vehicle, depending on their life context and related constraints.

We do have more work to do, though. We'd like to expand the survey beyond the Bay Area, and integrate the life-history calendar data into a land-use and transportation simulation model so we can better understand the extent of the energy impacts of these types of life transitions.

Provided byLawrence Berkeley National Laboratory