Have you received an Emergency Mobile Alert before? Or sent one, if you’re at a warning agency? It’s a short warning message that arrives on your mobile phone via cell broadcast - these are also called Wireless Emergency Alerts. Here in New Zealand, they are usually used only for potentially life-threatening events. Many of us received them before and during the COVID-19 pandemic lockdowns. I remember going into the Ministry of Health to help write example messages during the first lockdown, and it was quite eerie knowing how important it was to get the messaging right.
A lot of effort goes into the preparation of these messages. Agencies who issue them, including emergency management and emergency services, must quickly prepare an informative and accurate message under pressure, which is sent out to all phones that are within the range of selected cell broadcast towers.
The information that is included, and even the order that it appears in, can influence people’s responses to the warning. In this article I describe:
What should you put in a short warning message?
The key elements of a warning message are as follows:
The source of the message, in the case of an Emergency Mobile Alert, is the agency issuing the warning. In other contexts, sources of warnings can be family and friends, neighbours, social media, TV or radio, and so on – anyone from whom you receive the warning. It is important to avoid using acronyms in source names as much as possible, unless you know for sure that people relate it to your agency. Including an authoritative and credible source of the warning in the title can help increase behavioural compliance [1, 2]. If the issuing source is not perceived as credible or authoritative, then some work needs to be done to increase this over the long term. Initiatives such as ongoing engagement with the community, listening to and addressing their needs, taking steps to increase perceptions of accuracy of the warnings and the knowledge of the source [3], can enhance awareness and improve credibility [2].
Hazard information needs to be specific and detailed, to provide people with a rationale for taking action. Descriptions can include the characteristics of the hazard, such as it’s severity, magnitude/height, and speed. The geographical area it will cover is also important [2]. Having prior knowledge about the hazard from education campaigns helps trigger people’s visualization of what the hazard is, and how it might threaten them. If there is room in the message, explain what the more unfamiliar terms mean. For example, ‘storm surge’ could be accompanied by ‘(inland flooding of the sea)’. A ‘volcanic eruption’ could be supported (if deemed appropriate by volcanologists, as it depends on the volcano) by ‘An eruption may create dangerous local hazards like hot flying rocks, dense clouds of ash and steam, and lava flows. Widespread ashfall may also occur’.
Describing the impacts of the hazards on the recipient if they do not act can be helpful in increasing people’s risk perceptions, and potentially their actions [4]. This could include that the hazard is life-threatening, power outages could occur, coming into contact with flood water can cause injuries and sickness, and so on.
Guidance information is crucial in a warning – there needs to be no doubt in a person’s mind about what they are meant to do next to keep themselves and their families as safe as possible. The actions need to be achievable, affordable, and easy to do. The receiver of the warning also needs to believe that taking the action will be effective in reducing their risk. Having clear location information about where to evacuate to is very important (people often don’t know what ‘high ground’ means for tsunami evacuation). Links to more detailed guidance information can help get around message length constraints, and can often point to existing website information (such as getting your house ready for a cyclone).
Location information is a clear description of where the people are who will be affected by the hazard, and who needs to take action [5]. Location names need to be understandable by all people that the message relates to, including tourists and others who don’t know landmarks in the area.
Time relates to:
The time of issuance of the warning (this is often included by default by the message title)
The time the receivers should start to take action, and/or when to have responded by (using a specific time rather than saying e.g., ‘in 30 minutes’) [2]
The duration of the response (e.g. how long to evacuate for)
The time of the next update, if known.
It is debatable whether to include the time that the hazard is due to arrive. This may muddy the waters with time to have responded by and cause a delay in action – it likely depends on the context and whether evacuations are being called. Time should be in 12-hour format (e.g., 6 pm) as some people can’t easily understand 24-hour format (e.g., 18:00) [6].
A link to more information helps reduce delays while people seek more information. If technology allows it, an active hyperlink to a website can be helpful, or a short and easy-to-remember-and-spell URL if it can’t be clicked. The website will need to be able to handle getting thousands, if not millions, of hits at once without crashing. Referring to a radio or TV station is another option, however this can slow down response if an immediate action is required, and those channels would need to provide more detailed information than the mobile alert.
Finally, if there is room in the message, then a short update on what the authorities are doing about the situation may be helpful. This can increase confidence in the authorities, but its usefulness may depend on context and type of hazard.
What order should these elements go in?
The order of these elements can influence people’s behavioural response to the message and differs slightly depending on message length. The research that investigated this didn’t include ‘impacts’ or ‘links’, so I haven’t added them below, but I suggest putting ‘impacts’ after ‘hazard’, and a ‘link’ at the end. The optimal order the research found was as follows:
Messages of 90 and 140 characters: source, guidance, hazard, location, time [2]
Messages up to 280 characters: source, hazard, location, time, guidance [8]
Messages of 1,380 characters: source, hazard, guidance, location, time [2].
However, these results relate to risk perceptions and intended actions, rather than actual responses during an event. So rather than focusing too much on trying to get the message in this order, it is more important to make sure that the message contains all the elements, and that you consider the characteristics of a good warning message, which is up next.
What characteristics should a warning message have?
If you include all the elements described above, then the warning message is complete and more likely to be useful. This is the first characteristic required, as shown in the image below. If the message is an update, then say so.
The warning needs to be specific enough that people personalise it – i.e., they know whether it relates to them. This can be achieved by mentioning location names that have meaning [9]. In a tsunami warning, for example, instead of describing a warning area as spanning from one ‘Cape’ to another, which people likely don’t know, use city and town names and landmarks that have been proven to be well known. This makes the warning more accessible to tourists as well, who aren’t familiar with the local geography.
Once technology allows maps to be included in short warning messages, this may also help people understand whether it relates to their location or not. This is already possible for warning messages that are on a website, or X (Twitter), etc., and links in Emergency Mobile Alerts could take people to a website with a map on it. Be explicit in the directness of the message. Saying that people ‘in this area’ are to act leads to people not believing it relates to them – specific locations should be included instead [2].
A warning message needs to be clear, or easily understood. Seems obvious, right? But warnings often end up filled with acronyms, jargon, and no context, meaning that someone coming across it for the first time doesn’t understand what it is talking about, and importantly, what they are meant to be doing about it to keep themselves safe.
The message also needs to be accurate – double check that the content is factually correct (e.g. location names), not misleading at all, and there are no spelling or grammatical errors. Any errors, and subsequent corrections, can jeopardise trust in the source of the warning. Have a fresh pair of eyes look over it before sending it off, if possible.
Guidance information needs to be perceived to be effective at reducing the risk, affordable, and easy to do.
The information needs to be internally and externally consistent. This means that the message shouldn’t be confusing and say both ‘do not travel’ and ‘go to higher ground’, for example. Information across several channels, such as different agencies, should give the same type of messaging and advice to reinforce what people should do.
Finally, the warning message needs to be issued in a timely manner. I’m a big fan of enabling people to take actions to reduce their risk in advance of a hazard occurring, using forecasts. So instead of waiting for confirmation that a highly dangerous event is actually occurring in an area, a warning message should go out beforehand. Otherwise, it could well be too late for people to act, with access routes blocked off, or it’s too dangerous to go outside to prepare for the event.
What formatting aspects should be considered in a short warning message?
The focus of this article is on Emergency Mobile Alert type of messages. Technology constraints on character limits often change, but in an example of a limit of 1395 characters, some research [9] has suggested that longer messages may be more effective than shorter messages in prompting people to respond. There is some debate on this though - recent research [10] from the US found not much difference in compliance between 90-character and 360-character messages, and it likely depends on the type of hazard. Longer messages mean there is enough information that it minimizes the time people spend ‘milling’ or searching for more information. Having paragraph breaks in the message can help break up the ‘wall of text’, making it easier to read.
Careful use of ALL CAPS can help communicate the urgency of the warning. However, having an entire message in capital letters makes it harder to read and people feel like they’re being shouted at, so use it sparingly.
I hope that at some point soon, technology will enable us to provide Emergency Mobile Alerts in multiple languages to ensure that everyone understands them, according to user preference or perhaps digital footprints. Using icons and emojis may be effective in helping get the main point across too when language is a barrier, although I would like more research done to investigate whether this impacts the credibility of the message.
Keen for more?
Here are a few resources you can look at for more detailed information:
Potter, S. (2018). Recommendations for New Zealand agencies in writing effective short warning messages. GNS Science Report 2018/02. https://shop.gns.cri.nz/sr_2018-002-pdf/
Potter., S. (2021). Why some people don’t respond to warnings: writing effective short warning messages. Australian Journal of Emergency Management. 36:1. https://knowledge.aidr.org.au/resources/ajem-january-2021-why-some-people-don-t-respond-to-warnings-writing-effective-short-warning-messages/
Sutton, J., Olson, M.K., Waugh, N.A (2023). The Warning Lexicon: A multiphased study to identify, design, and develop content for warning messages. Natural Hazards Review, 25 (1). https://ascelibrary.org/doi/10.1061/NHREFO.NHENG-1900
Additionally, I recommend checking out the Warn Room website, which has been set up by Dr Jeannette Sutton in the US https://www.thewarnroom.com/contents-guide
References
There are many other useful references on warnings out there – these are just a select few I've cited above.
[1] Wogalter, M.S., Kalsher, M.J., Rashid, R. (1999). Effect of signal word and source attribution on judgments of warning credibility and compliance likelihood. International Journal of Industrial Ergonomics, Volume 24, Issue 2, p 185-192. https://doi.org/10.1016/S0169-8141(98)00025-0.
[2] Bean, H., Liu, B.F., Madden, S., Mileti, D.S., Sutton, J., Wood, M. (2014). Comprehensive testing of imminent threat public messages for mobile devices. College Park (MD): National Consortium for the Study of Terrorism and Responses to Terrorism. 192 p. https://www.dhs.gov/sites/default/files/publications/Comprehensive%20Testing%20of%20Imminent%20Threat%20Public%20Messages%20for%20Mobile%20Devices.pdf
[3] Peters, R.G., Covello, V.T., McCallum, D.B. (1997). The determinants of trust and credibility in environmental risk communication: an empirical study. Risk analysis. 17 (1): 43-54. https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1539-6924.1997.tb00842.x
[4] Potter, S., (2018). Recommendations for New Zealand agencies in writing effective short warning messages. GNS Science Report 2018/02. https://shop.gns.cri.nz/sr_2018-002-pdf/
[5] Mayhorn, C.B., and McLaughlin, A.C. (2014). Warning the world of extreme events: a global perspective on risk communication for natural and technological disaster. Safety Science. 61:43-50. https://www.sciencedirect.com/science/article/abs/pii/S0925753512001014
[6] Potter, S. H. (2018). Intended responses to a tsunami evacuation message using Emergency Mobile Alerts in New Zealand (GNS Science Report 2018/14, Issue. https://shop.gns.cri.nz/sr_2018-014/
[7] Potter., S., (2021). Why some people don’t respond to warnings: writing effective short warning messages. Australian Journal of Emergency Management. 36:1. https://knowledge.aidr.org.au/resources/ajem-january-2021-why-some-people-don-t-respond-to-warnings-writing-effective-short-warning-messages/
[8] Wood, M., Bean, H., Liu, B.F., Boyd, M. (2015). Comprehensive testing of imminent threat public messages for mobile devices: updated findings. College Park (MD): National Consortium for the Study of Terrorism and Responses to Terrorism. https://www.dhs.gov/sites/default/files/publications/WEA%20-%20Comprehensive%20Testing%20of%20Imminent%20Threat%20Public%20Messages%20for%20Mobile%20Devices%20Updated%20Findings.pdf
[9] Sutton, J., Wood, M., Huntsman, D., Waugh, N., Crouch, S. (2023). Communicating hazard location through text-and-map in earthquake early warnings: A mixed methods study. Natural Hazards Review, 24 (4). https://doi.org/10.1061/NHREFO.NHENG-1723
[10] Carlson, E., Bean, H., Ratcliff, Cl, Pokharel, M., Barbour J. (2024). Do 36-character Wireless Emergency Alert messages work better than 90-character messages? Testing the risk communication consensus. Journal of contingencies and crisis management, 32 (2).
Comments