Nudging Civilians to Evacuate War Zones

Nudging Civilians to Evacuate War Zones

Photo: / palinchak
15 December 2022

As of November 2022, more than one-third of Ukrainians left their homes because of Russia’s full-scale invasion, the UN reports. However, despite heavy shellings, unbearable living conditions and direct threat to life, many people decided to stay in their homes. Are there affordable and efficient instruments that could help motivate people to leave dangerous zones and therefore to save their lives? Behavioral economics suggests nudging.

Nudging is generally soft interventions relying on choice architecture or information provision to steer behavior in a desired direction. In recent years, there has been an exponential growth in the popularity of nudges and behavioral insight teams (sometimes referred to as nudge units) to the point that many western governments invested in their creation (United Kingdom, Australia, etc.). They applied nudges to numerous settings ranging from COVID-19 to climate change with different degrees of success. The reason why this instrument is so widespread, both among public and private institutions, is it has often been found to have a substantial impact and to be cost-effective. To the best of our knowledge, nudges have not been used before in the context of text-alerts from danger zones. However, they seem promising, and we decided to test it in an experimental setting.

Experiment design

We took a sample of 2000 respondents, randomly split it into ten approximately equal groups, and suggested each group one of ten different text-alert messages with a call to evacuate. Those ten messages varied in two dimensions: first, we provided five various frames (Control group, Gain of Life, Loss of Life, Deteriorating Living Conditions, Military Effectiveness). Along the second dimension, we included information about an evacuation plan provided by the authorities (Provided Plan). After having seen the message, participants were asked to evaluate on a scale from 0 to 10 how effective they believed a given evacuation message would be at convincing fellow residents of their city or village to evacuate. 

For the control message, we used a real-life message aired by local authorities in the evacuation of civilians during this war. For the design of other messages, we were guided by previous findings in behavioural economics which suggest that losses loom larger than gains (Gain of Life vs Loss of Life). We also relied on the evidence indicating that making the consequences of actions more salient influences behaviour (Deteriorating Living Conditions vs Military Effectiveness).

What should be told so that people listen?

We find that the Loss of Life message is the one with the highest level of approval. Nonetheless, the response to this message is not statistically higher than to the rest of the messages along the frame dimension, suggesting that the framing itself does not seem to matter that much. We assume that this is because people are very well aware of the risks, so they don’t need external “motivation” to evacuate. 

However, what we find does make a significant difference is whether subjects were provided with a plan by the authorities or not (instructions on transport available, aid provided in a new place, a list of personal belongings to be taken). The inclusion of the plan leads to an increase of at least 5% and possibly up to 10% in the perceived effectiveness of the messages. This is particularly important given how strongly related are intention to evacuate and actual evacuation behaviour. The magnitude of the effect might not seem large but given that the scale of the problem involves hundreds of thousands of people, even a single digit improvement potentially represents thousands of saved lives. 

Digging deeper into the effect, we also found that women (who often determine the evacuation decisions in households) react much more strongly to the provision of an evacuation plan than men (who hardly react at all). Finally, those who have a personal plan in place evaluate all messages to be more effective, that is there is no difference in their evaluation perception whether or not a plan is provided by the authorities. If anything, the effect is slightly negative but not statistically significant. In other words, having a personal evacuation plan and receiving information about the authorities’ plan are (imperfect) substitutes.

Policy implications

Our findings could be implied immediately in Ukraine and other war conflicts all over the world. First, we show that providing a concrete evacuation plan is fundamental if governmental organizations or NGOs want to convince civilians to evacuate from a war zone. The framing of the message itself has little impact. This can also be viewed as reassuring as it also means that the authorities can hardly make substantial mistakes when it comes to the wording of their message. This is possibly due to the fact that those facing the threats of war are not really naive about the potential dangers of sheltering in place. Hence, the approach should rather focus on reducing uncertainties for those who would like to leave but do not feel they have the means or opportunities. The most important take-away is to provide a concrete plan of evacuation (i.e., a course of action). We are also aware of the fact that the amount of information the authorities are able to provide might be reduced due to the threat of a potential attack on the evacuation area.

Additionally, our results suggest that the personal evacuation plans and the authorities’ plans are substitutes, and that existence of a personal plan is largely determined by access to means of evacuation (i.e., a personal vehicle) and the resources to sustain oneself during evacuation (spare income/cash). Therefore, in order to maximize the number of evacuees we recommend a two-step approach. The first step is an early warning that the situation in the settlement is becoming increasingly dangerous. This first message should instruct people to make their own personal evacuation plan if possible (along with guidance on what such a plan should entail). The main objective is to get civilians to be ready to leave their houses. This message should also mention that those who do not have independent means will be evacuated by the authorities. If the danger persists and evacuation becomes inevitable, the second step would be to instruct everyone who can evacuate independently to do so (ideally while the outward journey is still relatively safe). If the government (or an NGO) has an evacuation plan, this second wave of messages should prioritize informing individuals on how to participate in the authority’s organized plan. Overall, we argue that more effort should be put into testing the effectiveness of the messages and understanding the determinants of evacuation behaviour from war zones to target the population more precisely. There are around 50 other armed state-based conflicts worldwide at this very moment, and understanding how to increase civilian evacuation is clearly important for all of them. 

The research design

We surveyed 2006 respondents in 10 war-affected oblasts in Ukraine (in the East, South, North and Center) in July 2022. The sample contains both subjects that evacuated and those that decided not to evacuate (respectively 37% and 63%). Some of the evacuees have already returned home. The survey was conducted by Gradus Research Company. 


Martinez, Seung-Keun and Pompeo, Monika and Sheremeta, Roman M. and Vakhitov, Volodymyr and Weber, Matthias and Zaika, Nataliia, Nudging Civilian Evacuation During War: Evidence from Ukraine (November 30, 2022). Available at SSRN.

The research was conducted within the initiative Behavioral Science for Ukraine by a group of authors: 

  • Seung-Keun Martinez, University of Nottingham
  • Monika Pompeo, New York University (NYU) – New York University Abu Dhabi
  • Roman M. Sheremeta, Case Western Reserve University
  • Volodymyr Vakhitov, American University Kyiv
  • Matthias Weber, University of St. Gallen – School of Finance; Swiss Finance Institute
  • Nataliia Zaika, American University Kyiv


The authors do not work for, consult to, own shares in or receive funding from any company or organization that would benefit from this article, and have no relevant affiliations