Attending scientific conferences is considered to be an essential part of academic activity that contributes to one’s performance and professional development (Sanders et al. 2022). At the same time, there are rising concerns about the environmental externalities generated by long-distance travel, and air travel in particular (Chiambaretto et al. 2021). Recent studies have quantified the carbon footprint associated with a (sustainability science) scientific conference (Neugebauer et al. 2020), conducted a qualitative research of (sustainability) scientists’ dissonance between pro-environmental attitudes and flying behaviour (Schrems and Upham 2020), and discussed alternatives for conference travel emissions (Klöwer et al. 2020).
We conduct a stated-preference choice experiment as part of a survey concerning long-distance travel for attending scientific conferences and estimate choice models describing the decision to travel to a conference and the decision to do so by either train or plane. In this context, we address the questions: how concerned are academics about the environmental impacts of long-distance travel associated with attending conferences? what are the determinants of travel mode choice, and what role do carbon emissions play therein?
To elicit academics’ preferences towards conference travel we developed a stated-preference choice experiment in which respondents were asked to consider travelling to an international conference at the core of one’s interest and having to make a choice between travelling by train or plane or opt out from travelling. For train and plane the following attributes were included: travel time, travel cost and the amount of emitted CO2. Each respondent was confronted with 12 choice situations with fixed access, airport and egress times (see Figure 1). Next, respondents were asked to rate 14 attitudinal statements related to environmental awareness about the impacts of flying, flying behaviour, flight shame, the attitude towards online conferences and carbon offsetting on a 5-point Likert scale.
We obtain the trade-offs between travel time, travel costs and CO2 emissions by estimating a Multinomial Logit (MNL) model using Pandas Biogeme (Bierlaire 2020). When someone chose the opt-out option, the following question asked the respondent to also choose between plane and train only, given the same attributes. This ‘forced choice’ is accommodated in the MNL model through availability conditions. Note that our MNL results likely underestimate the standard errors given the panel nature of our data.
The survey was completed in May-June 2022 by a convenience sample of 104 academics (studying primarily transport topics) residing in Europe. The dataset is available at Cats et al. (2022).
Figure 2 presents the answers to the questions about respondents’ environmental attitudes and behaviour. The statements are in decreasing agreement (the green bars). Remarkably, the statements with highest agreement concern the increasingly frequent discussions of environmental issues in respondents’ social environments. Half of the respondents (strongly) agrees with the statement that one is flying less for environmental reasons, but only 7% indicates to have stopped flying altogether. In addition, a small majority (52%) (strongly) agrees with the statement that flying stirs one’s conscience and makes one feel guilty about it. Despite these attitudinal positions, the statements describing actions – e.g., about regularly carbon-offsetting the flights – score relatively low. Finally, only limited substitution of physical conferences by online ones might be possible.
Across all choice tasks academics choose to travel to conferences by train roughly twice as often (64% of the responses) as by plane (35%). Only in roughly 1% of the situations did our respondents decide to skip the conference visit altogether. Table 1 shows the results of the MNL model. The large alternative specific constants reflect the strong preference to travel to the conference as compared to skipping the trip. Travel time by plane is valued much more negatively than travel by train. This finding may, to some extent, be confounded with the selected attribute ranges: in-vehicle travel times by train were roughly five times longer than by plane. The travel cost parameter is relatively low, with implied value of travel time of 521 €/h and 109 €/h for plane and train, respectively. This reflects the fact that employers pay for conference travel, which was also highlighted in the survey introduction. Lastly, the parameter for CO2 emissions indicates that academics are willing to pay € 5 to reduce 1 kg of CO2.
To provide insight into the relative importance of each attribute, Figure 3 presents the range of impact of each attribute on the probability of choosing train. Departing from the probability of an ‘average train’ option (of 66%), it shows the impact of each attribute within the range used in the survey (e.g., the probability that train is chosen, given minimum to maximum plane travel time, ranges from 52% to 78%). Since this is based on an MNL model, the relative impact on the probabilities to choose the plane or to skip the trip are the same. As expected, travel times have the largest impact on the mode choice. However, the impact of CO2 emissions is also considerable; the probability that train is chosen given high (173 kg) and low (87 kg) CO2 emissions by plane ranges between 57% and 75%, respectively.
We find that academics are embedded in social and professional environments that are conscious of the environmental impacts of flying and that many feel guilty about flying. Still, these attitudes do not readily translate into behaviour (e.g. carbon offsetting, preferring online conferences). In addition, while CO2 emissions play a role in travel mode decisions, their role is not as decisive as the one of travel time, which is the most important attribute. Note however that there is heterogeneity in our sample in terms of environmental actions, as seen, for example, by roughly 10% of respondents that have stopped flying altogether (see Figure 2). Overall, we conclude that the so-called attitude-behaviour gap (Juvan and Dolnicar 2014), can also be observed among a segment of highly aware academics. From a policy viewpoint, this means that measures that rely on changing attitudes and/or raising awareness (e.g. through information campaigns) will have little effect on flying practices. Instead, more strict sustainable conference travel policies are likely necessary, such as travel quotas or a carbon budget.