By Nikolai Bode, University of Bristol
Suppose a fire breaks out in a high-occupancy building, making it necessary to evacuate many people. On the one hand, this requires infrastructure and management that facilitate the efficient and safe movement of people to a place of safety, which requires in turn clear signage, sufficiently wide exits and enough exit routes, to name a few examples. On the other hand, evacuations can only be successful if building occupants start to move toward safety without delay when instructed to do so by alarms or safety personnel, without stopping to gather their belongings, for instance. As a result, the required time for evacuations is often considered in terms of two components, the evacuation time and the pre-evacuation time, sometimes also referred to as the pre-movement time.1
Through experiments with volunteers, mathematical models and computer simulations, research has helped to develop guidelines for the safe movement of crowds.2 However, the research-based evidence is much thinner for pre-evacuation delays. That is unfortunate, because observations in real emergencies have shown that the behavior of individuals in the pre-evacuation phase can cause substantial delays in evacuation times, sometimes longer than the time it takes individuals to move to safety.1, 3-8 A good example comes from unannounced fire drills in UK department stores, where the pre-evacuation time contributed 30–50% of the total evacuation time for individuals.6 In addition, anecdotes abound about restaurant visitors wanting to finish their meals, people not wanting to abandon their cars in tunnel fires, and evacuees searching for wallets or clothing.
Part of the reason that research has so far failed to explain pre-evacuation delays comprehensively is that they depend heavily on the context and have a wide variety of causes.9 Examples include difficulties in hearing fire alarms; attempts to obtain more information; and other deliberate preparatory actions, such as collecting personal belongings, putting on warm clothes or waiting for others. How and when these aspects affect evacuation times depends on a variety of factors, including warning systems, occupant characteristics, training of any staff involved, time of day and type of building.1, 3-8
Despite this complexity, research is increasingly trying to quantify the contribution of key factors to pre-evacuation delays. Arguably one of the most widely reported behaviours is individuals collecting objects, such as keys, wallets, electronic items or handbags before they attempt to move to safety.1,3,7,10–11 Except for one study that revealed a positive correlation between the degree to which university students are attached to their belongings and their pre-evacuation times in fire drills,10 there is limited experimental research on what might cause individuals to take this risk. That is why we decided to investigate how risk-taking by collecting objects before evacuating is influenced by three factors: evacuees’ knowledge of a building, the behaviour of other evacuees and how evacuees are attached to the objects they can collect (potential gain versus loss).12
What makes this study stand out from more-conventional fire drills is that it was conducted in a virtual environment. Participants were asked to complete a few simple tasks by controlling the movements of one person in a computer simulation of a crowd evacuation from a building. This setup is comparable to playing a simple computer game.
Participants could collect up to 10 abstracted objects (“coins”) before attempting to evacuate within a time limit. By adjusting how much of the building was visible and how the computer-controlled evacuees moved, and by showing participants’ messages, we simulated changes in the three factors under investigation (reduced knowledge of building layout, crowd evacuation behavior and loss-aversion). This setup allowed us to conduct the study with visitors to the Science Museum in London; we managed to recruit more than 1,200 participants. Figure 1 provides a screenshot of the experiment and readers can try the experiment at www.evacgame.eu.
Figure 1. Computer-based evacuation experiment at Science Museum, London, UK.
We first confirmed that collecting more coins was risky, because it affected evacuation success. Thus, the number of coins participants collected could be used as a proxy for the level of risk they took. Interestingly, we found that many participants showed extremes of risk-taking behavior by either not collecting any coins (26.5%) or collecting all coins (18.1%).
While the movement of the simulated evacuees and the information participants had about the building affected the adoption of these extreme strategies, only the suggestion that participants could recover lost objects had a robust effect on the average level of risk-taking, regardless of extreme strategies. We found no effect of gender, but older participants were, on average, more risk-averse.
These findings confirm what we already assumed — the role of loss aversion — but they also raise additional questions. For example, the prevalence of extreme behaviors requires further explanation and, in contrast to our findings, previous research suggests that there are gender differences in risk-taking.13 Possible explanations could be linked to the experimental setup.
On the one hand, virtual experiments allow complete control over the experimental setting, are typically time-and cost-effective, and make it possible to study stressful scenarios that cannot be simulated in other experiments for ethical and safety reasons.14-15 On the other hand, research is keenly that the validity of virtual experiments for real-world contexts is far from clear: Behavior in virtual environments may not be the same as real-world behavior.14-15 For example, some participants may take the experiment seriously, remember safety instructions they have heard before and not collect any coins. Other participants may treat the experiment as a computer game and try to complete the challenge of collecting all coins in time. Younger participants may be more used to playing computer games, which may help to explain age-related findings. Furthermore, the simple layout of the building used in our experiment could explain why familiarity with the building does not affect pre-evacuation delays, in contrast to what has been suggested previously.1
Considering these caveats begs the question of what our study can be used for. An article in issue 12, article 6, of this magazine discussed the uses of virtual environments in safety training and our virtual experiment makes it possible for people to experience evacuation scenarios interactively and to make mistakes they can learn from in a safe environment.
There is an alternative use for studies such as this one. All experiments are, to a larger or smaller extent, abstractions of reality and therefore face questions about their validity. Our study showcases the potential of using simple virtual experiments to safely, quickly and cheaply scope behavioral processes causing pre-evacuation time delays in crowd evacuations. Conducting experiments online, potentially using mobile devices, could further increase cost and time savings.16 Employing this technology for preliminary scoping studies on evacuee behavior could help identify key aspects that should be explored in more-realistic (and more-expensive) evacuation drills and thus provide starting points for further research.
Studies such as this one can be used for research in understanding human behavior in evacuations.
1Proulx, G. (2002) Movement of people: the evacuation timing. In DiNenno, P.J. (ed.) SFPE handbook of fire protection engineering, 3rd ed. 343–366. Quincy, IL: National Fire Protection Association.
2Schadschneider, A., Klingsch ,W., Kluepfel, H., Kretz, T., Rogsch, C., and Seyfried, A. (2009) Evacuation dynamics: empirical results, modeling and applications. In Meyers RA (ed.). Encyclopedia of complexity and systems science, 3,142–3,176. Heidelberg, Germany: Springer.
3Zhao, C.M., Lo, S.M., Zhang, S.P., and Liu, M. (2009) A post-fire survey on the pre-evacuation human behavior. Fire Technology 45:71–95.
4Kobes, M., Helsloot, I., De Vries, B., and Post, J.G. (2010) Building safety and human behaviour in fire: a literature review. Fire Safety Journal 45:1–11.
5Johnson, C.W. (2005) Lessons from the evacuation of the World Trade Center, Sept. 11, 2001, for the future development of computer simulations. Cognitive Technology Work 7:214–240.
6Shields, T.J., and Boyce, K.E. (2000) A study of evacuation from large retail stores. Fire Safety Journal 35:25–49.
7Nilsson, D., and Johansson, A. (2009) Social influence during the initial phase of a fire evacuation—analysis of evacuation experiments in a cinema theatre. Fire Safety Journal 44:71–79.
8Liu, M., Lo, S.M. (2011) The quantitative investigation on people’s pre-evacuation behavior under fire. Automotive Construction 20:620–628.
9Lovreglio, R., Ronchi, E., and Nilsson, D. (2016b) An evacuation decision model based on risk perception, social influence and behavioural uncertainty. Simulation Modelling Practice Theory 66:226–242.
10D’Orazio, M., and Bernardini, G. (2014) An experimental study on the correlation between ‘‘attachment to belongings’’ ‘‘pre-movement’’ time. In Weidmann, U., Kirsch, U., and Schreckenberg, M. (eds.) Pedestrian and evacuation dynamics. 2012 [RET-C2] Berlin, Germany: Springer.
11D’Orazio, M., Longhi, S., Olivetti, P., and Bernardini, G. (2015) Design and experimental evaluation of an interactive system for pre-movement time reduction in case of fire. Automotive Construction 52:16–28.
12Bode, N.W., and Codling, E.A. (2018). Exploring Determinants of Pre-movement Delays in a Virtual Crowd Evacuation Experiment. Fire Technology 1–21.
13Granié, M.A. (2009) Effects of gender, sex-stereotype conformity, age and internalization on risk-taking among adolescent pedestrians. Safety Science 47:1,277–1,283.
14Kinateder, M., Ronchi, E., Nilsson, D., Kobes, M., Müller, M., Pauli, P., and Mühlberger, A. (2014) Virtual reality for fire evacuation research. Ganzha, M., Maciaszek, L., and Paprycki, M. (eds). 2014 Federated conference on computer science and information systems (FedCSIS) 313–321. IEEE.
15Bode, N.W.F., Wagoum, A.U.K., and Codling, E.A. (2015) Information use by humans during dynamic route choice in virtual crowd evacuations. Royal Society Open Science 2:140410
16Brown, H.R., Zeidman, P., Smittenaar, P., Adams, R.A., McNab, F., Rutledge, R.B., and Dolan, R.J. (2014) Crowdsourcing for cognitive science — the utility of smartphones. PLoS ONE 9:e100662.