There is no such thing as unbiased research – is there anything we can do about that?

If you are reading this editorial, you are likely to be a highly educated, high-income individual living in a city, who reads English fluently, has easy access to the internet and a mobile phone, studies or works full-time and has a more-than-passing interest in transportation. You experience the world through your body – abled or disabled, gendered, with skin and hair associated with a particular genetic background. A lifetime of experiences has shaped your familiarity with the transport system. Perhaps you view this transport experience through a defining lens – personal experience with road trauma, an interest to particular mode of transport, or a physical disability. Or perhaps you believe that your experience in the transport system is a generic, “default” experience, more or less similar to your friends, colleagues and fellow human beings.

When researchers show up to work, we like to think that we leave that personal experience behind. We work hard to collect and analyse data in an unbiased way so that we can draw objective conclusions about the transport system and the people who use it. And yet research involving humans can never be truly objective. Every decision is touched by our subjective experience with the world, or our research positionality: which research questions capture our interest, which papers we read and cite, which methods we consider to be valid, which assumptions we make about those methods and how we interpret what we find (Nihan & Debbie, 2022).

The tendency to start from our own subjective experience of the world is a mental shortcut, or heuristic,¹ called egocentric anchoring (Epley et al., 2004). Heuristics help us process the overwhelming amount of information we take in and the decisions we make every day. When we try to estimate something about the outside world, we often start from our own experience before adjusting based on other information. Yet that “adjustment” is usually still biased by our own experience. For example, a recent study of transport professionals in the US found that their experience of the transport network was much more urban and multi-modal than the average American. Moreover, this experience influenced their estimates of how other Americans use the transport system (Ralph & Delbosc, 2017).

Another common heuristic is confirmation bias, whereby we unconsciously seek out or pay greater attention to information that confirms pre-existing beliefs or hypotheses, and conversely we pay less attention to information that conflicts with our beliefs (Nickerson, 1998). Although the scientific method is designed expressly to overcome this bias, it is an insidious heuristic and has been demonstrated to colour judicial reasoning, policy justification and medical diagnoses (Nickerson, 1998).

These biases are part of the human condition, and drawing from personal experience can be a great strength in generating new hypotheses to explore. Yet they can also work to inadvertently shift the research field toward particular narratives (Lowe, 2021). I posit that these biases come together around new and emerging topics that excite the imaginations of transport researchers, such as the changing travel behaviour of younger generations, the potential role of autonomous vehicles or the future impacts of COVID-19 on travel behaviour. Because these topics are future-focussed, they are particularly vulnerable to egocentric anchoring and confirmation bias, partially reflecting “our” vision of the future (Glenn, 2016).

Take for example the interest in modelling the adoption and use of autonomous vehicles. The “three revolutions’ scenario (sometimes referred to as a “heaven” scenario) paints a future where traditional automobility is replaced by a fleet of shared, electric and autonomous vehicles (Sperling, 2018). This scenario is held up as an end-goal of this technology, perhaps in part because it feels achievable and ideal for the people testing these scenarios. A recent study of the potential environmental benefits of shared, autonomous electric vehicles used the following justification for assuming that in the future, 70% of autonomous vehicles would be shared:

Ridesharing and autonomous vehicles are naturally complementary in many respects, and the two technologies will likely mutually enhance one another. Furthermore, Stoiber et al. (2019) conducted a stated choice experiment with Swiss households and obtained results supporting the assumption that autonomous vehicles are more likely to be used in a pooled mode than through private ownership. Therefore, in this study we consider a future transport sector with substantial adoption of SAVs rather than consider ridesharing or autonomous vehicles individually. (Jones & Leibowicz, 2019, p. 282)

The Stoiber et al. (2019) study cited above drew from Swiss adults who had enough free time and technological savvy to complete a 30-minute online stated preference survey, about a future scenario where all vehicles were fully autonomous and the only options are individual autonomous car ownership, autonomous-taxi membership or an automated shuttle to the train system. The headline finding was that 61% of respondents would opt for an autonomous-taxi or autonomous shuttle + train, and the authors noted that “different social groups behave differently regarding sharing or pooling” (Stoiber et al., 2019, p. 280). Yet this carefully-qualified finding is used to justify testing the impacts of the “heaven” scenario of a shared, autonomous and electric vehicle future.

Other studies have looked into which segments of society are more likely to adopt shared, autonomous vehicles. They generally find that highly educated, urban-dwelling, technology-savvy young adult males are more likely to be willing to adopt shared autonomous vehicles (Narayanan et al., 2020). Perhaps I am drawing on my own confirmation bias, but this author wonders how many researchers on this topic are themselves urban-dwelling, technology-savvy young adult males? How many consider the feasibility of shared autonomous fleets for people in rural areas, people without smartphones, parents of multiple children in car seats, or even dog owners? This question is not just a matter of accurately predicting market share. If the fundamental inequity of a market-driven Three Revolution scenario is not at the forefront of the dialogue, the tech-savvy and advantaged will enjoy heaven while historically disadvantaged communities will continue to be left behind in hell (Cohen et al., 2017; Creger et al., 2019).

Overcoming bias

No matter how objective we believe we are, it is impossible to completely overcome the heuristics and biases that are hard-wired into the human brain. Yet becoming aware of these biases, within ourselves and others, is a first step toward counteracting them. We might reflect more consciously on our own experience with the transport system and how that may be shaping the research we conduct. We might take greater care in seeking out studies, stories and examples that challenge our perspective and more carefully represent opposing views when we present our work.

This practice may be particularly fruitful if you work with communities and decision-makers outside of academia, where we might remind ourselves that opposing views are also coloured by lived experience and confirmation bias. For example, local businesses often oppose reductions to nearby parking because they fear losing revenue, yet they consistently over-estimate how many of their customers drive and under-estimate how many walk, cycle or take public transport to their shops or restaurants (Volker & Handy, 2021; Yen et al., 2020). Perhaps they are drawing from their own experience when making these estimates, as a survey of retailers in New Zealand found that 75-96% of retailers used a car to get to their shop, compared to 66% of their customers (Fleming (Allatt) et al., 2013). In another example, Jarret Walker refers to “elite projection”, the tendency of high-level decision-makers to believe that transport solutions that are convenient for them (be that high-speed rail, hyperloops or subsidised electric vehicles) are good for society as a whole (Walker, 2017). In these examples, helping decision-makers step outside their own experience may be a necessary step in shifting perceptions.

More effective than “thinking” our way out of bias is to actively elicit the perspectives of people who are different to ourselves. How many of your students, colleagues, co-authors or research clients are women, gender-diverse, LGBTQI+, disabled, carers of young children or elderly parents, follow a non-majority religion or are part of an ethnic minority? How many grew up in a rural area, in a “blue-collar” family, in a different country? We all look at the world through a lens of our own experience. No single team, and certainly no single person, can possibly represent the full diversity of society. But the greater the diversity around the table, the better we are equipped to identify problems and propose actionable solutions that benefit a broader representation of society.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Entire fields of social psychology are devoted to identifying and understanding heuristics; for a comprehensive introduction I recommend the book by Kahneman (2011).