FEATURE3 March 2014

Lightning never strikes twice...

Crawford Hollingworth explores how cognitive biases affect our response to floods and storms.


The weather in the UK this winter has been the wettest since records began, and it has attracted a great deal of media and government attention. Homes have been flooded, families evacuated. High winds (up to 140mph at times) and even a few tornadoes have torn down power lines; flights and trains were cancelled, and infrastructure was damaged. 

While there have been some policy errors in managing the problems, the situation has also served to illustrate just how irrational and complex our behaviour and beliefs can be. Our responses to extreme weather and natural disasters are subject to a host of cognitive biases and heuristics which play on our minds – often causing more troubles than they solve.

In the first part of this article, we’ll look at three of these biases – availability bias, optimism bias and gambler’s fallacy – and we’ll examine how they drive our thinking. In the second part, we go on to analyse some initiatives which harness insights from the behavioural sciences to improve our response to what the weather throws at us.

“I can imagine this happening so it must be likely” 

Availability bias is a mental shortcut we use to assess the likelihood of something happening. When we don’t know the exact probability of an event we try to estimate it by drawing on other similar examples or experiences which come to mind. Most people’s memories are far from perfect and the examples we draw on are often those which surface easily, and these are usually the most vivid, recently seen or experienced events. 

Bringing to mind examples of extreme weather events and natural disasters, or imagining them happening, is a perfect example of the way in which availability bias affects us. In periods of calm, sunny or predictable weather, we sometimes struggle to think of instances of extreme floods, storms, high winds, blizzards etc – and consequently think they are less likely than they really are. Such underestimation can land us, quite literally, in deep water. When the weather is good and we have no recent memory of things being any different, we might buy homes built on flood plains or on the seafront, believing our chances of being flooded to be slim to none.

The reverse is also true. Now, in the midst of one of the wettest and stormiest winters on record, we have no problem imagining how bad the weather can get and we’ll probably overestimate the likelihood of apocalyptic-like events. The media helps contribute to our awareness by dramatising events, or using live broadcasts of spectacular weather moments. This kind of coverage encourages us to accentuate the intensity and the ubiquity of it all. The media has also created a new lexicon for extreme weather conditions: think Snowmaggedon, Stormageddon, Snowpocalypse, Polar Vortex, Superstorm, Weather Bomb etc. These terms help cement our acceptance of the likelihood of these things happening. Right now, the first question anyone looking to buy a house will be asking is, “Does it ever flood around here?” – and houses on hilltops will be in high demand.

Availability bias can lead us to behave very erratically when taking out insurance policies. People often arrange insurance to protect against an event which has just happened because they now have a vivid and recent memory of it occurring. Yet, as time passes and the memory fades, people often let their policies lapse, sometimes even believing that it’s less likely that such an event will happen again, simply because it has not happened for a while. A US study to track the proportion of flood insurance renewals found that only 74% of policies were still active after one year, and just 36% still in place after five.

The same heuristic also explains why we are more likely to believe in climate change during extreme bouts of weather. A study conducted by Eric Johnson and Ye Li at Columbia Business School surveyed  1,200 people in the US and Australia in three different studies to determine their opinions about global warming. All respondents were asked whether the temperature on the day of the study was warmer or cooler than usual. Respondents who thought the day was warmer than usual were more concerned about global warming than respondents who thought the day was colder than usual.  

That people base their attitudes to global warming on what’s happening within their local weather system helps explain why public belief in global warming can fluctuate. 

“Always look on the bright side of life”

Many of us are overly hopeful and optimistic, believing bad things will never happen to us despite what the statistics say. Tali Sharot and her colleagues at UCL have conducted many research studies into optimism bias and estimate that around 80% of us are affected by it, even if we don’t realise it. Moreover, even when we are informed of the true likelihood of an event happening to us, we still only marginally adjust our expectations. 

Similarly, we tend to believe that the chances of being affected by a flood or a storm are less than they really are. Qualitative research in the aftermath of a tornado outbreak in the US in 2008 showed that one of the main causes of fatalities was due to optimism bias. More than 50 people died and 350 were injured when tornado after tornado swept across nine Southern US states. When the National Weather Service reviewed why people had not responded to the tornado warnings they found that a major cause was people’s irrational optimism, and their belief that bad things only happen to other people. For example, one Kentucky resident said, “They [tornadoes] always seem to hit down the road.” And a Tennessee family said they heard on the radio that the tornado was in a town just upstream of where they lived “…but we didn’t think it was going to be here”.

We might also overestimate the potential for the development of new technologies to help us guard against the weather, either in predictive power or in infrastructure to protect our houses and land from natural disasters. Yet, while weather forecasting and environmental technology have improved substantially in the last 100 years, it is unlikely to ever be foolproof.

“We must be due some good weather now?”

After weeks of rain and wind, you might be thinking the bad weather can’t possibly last much longer and that we must be ‘due’ some nice, dry, sunny weather soon. Yet this is an example of another cognitive bias – gambler’s fallacy.

Gambler’s fallacy is the misconception that if something has already occurred once, or more frequently than normal, then it is less likely to occur again in the future. A typical example is getting a streak of heads when flipping a coin. People misunderstand randomness and think that it should be the turn of tails to ‘even it out’. The bias is also called the Monte Carlo fallacy, after a 1913 feat in a casino when a roulette wheel landed on black 26 times in a row. Thinking that ‘red must be due’ most gamblers bet against black…

Howard Kunreuther, professor of psychology at the University of Pennsylvania, is an expert on insurance and behavioural economics. He has studied our poor attempts at estimating the probability of natural disasters and reckons that the language used to communicate risks and the likelihood of a flood or disaster can often be misleading to the average person. “Scientists often talk about a ‘100-year return flood ‘, but many individuals do not understand what that means. Some who have suffered a flood believe they will not have another flood for 100 years.”

Leveraging cognitive biases to spur us into action

Although cognitive biases can lead to irrational behaviour, we can also draw on our susceptibility to those same biases to spur us into action. Below we describe three examples of organisations using behavioural science to prompt and steer action effectively.

1. Tornado warnings 

In the US, 163 people were killed in the 2011 tornado which hit Joplin, Missouri. It was a death toll that could have been avoided. Many of the dead could have been saved had they heeded the tornado warnings and taken adequate shelter and protection. But people didn’t respond to the warnings and take appropriate action. 

According to behavioural scientists, part of the problem was the dry, dull warnings – such as “You should activate your tornado action plan and take protective action now” – which failed to conjure up an image of the dire consequences of a full-blown tornado. Availability bias encouraged people to think a real threat to life was unlikely because they could not imagine it happening. People also tended to seek confirmation from other sources before seeking shelter, such as hearing sirens or even needing to see the tornado approaching with their own eyes before taking action. Ken Harding, a weather service official in Kansas City said: “We’d like to think that as soon as we say there is a tornado warning, everyone would run to the basement. That’s not how it is. They will channel flip, look out the window or call neighbours. A lot of times people don’t react until they see it.”

To combat this, the US National Weather Service (NWS) trialled a new style of tornado warning in 2012 in the states of Kansas and Missouri. Their ‘Impact Based Warnings’ make the warning more emotive and salient, conjuring up images of destruction, reminding people of the real danger to life, and so making the likelihood of a tornado seem higher. They are now using compelling descriptive messages, such as: “You could be killed if not underground or in a tornado shelter. Many well-built homes and businesses will be completely swept from their foundations.”

Mike Hudson, the chief operating officer at the NWS Central Region, says: “It paints a mental picture for people in the path of the storm.”  Such was the initial success of the two-state pilot, it was expanded to 14 states across central US in 2013. Of course, there may be a longer-term danger that the new-style warnings could create a ‘cry wolf’ effect if the threatened tornadoes turn out to be less destructive than anticipated. People may stop taking the warnings seriously and revert to their old ways. We shall see…

2. Flood warnings

The UK Environment Agency and the Met Office have also been trying to make flood warnings more salient in order to get people to take action to protect themselves from flooding. After much research and public consultation, the Environment Agency changed the flood warning codes in 2010 to make them simpler and easier to understand. They focused on improving the icons and the language used. The highest alert, ‘Severe Flood Warning’ comes with an accompanying warning of ‘Danger to Life’. This draws on availability bias by making the risk more real and frightening with the intention of prompting people to take action. 

 3. Be prepared

In the Australian Capital Territory (ACT), the region around Canberra, a new initiative has been implemented to get residents to take action and make sure they are prepared for extreme weather events such as fires, storms, earthquakes and floods. Called ACT First, it has been developed by Green Cross Australia in a partnership with the ACT government and support from the Australia National University. It website harnesses a number of behavioural science concepts, including:

  • Social norms: the website states what proportion of residents have taken  steps to be ready for a natural disaster. For example, ‘84% of responders have learned how to turn off their mains’ or ‘73% of responders keep a list of emergency contacts’. We have a natural impulse to do what others do, so these messages are likely to make an impact. They also showcase a range of real case studies, including video interviews and stories of how residents of ACT are getting prepared, with the aim of making other residents feel they should also take some action. We tend to relate well to personal stories.
  • Emotive response to imagery: Embedded in the social norms statements are photos sent in by responders, showing that they are ready for anything. For example there is a photograph of some young children proudly holding up their family’s list of emergency contacts and a photo from another responder showing her turning off her mains supply. These images help get our attention and generate an emotive response which can motivate us to take action. Perhaps seeing a photograph of children makes us think of our own children and whether we would be ready to protect them in an emergency.
  • Tracking residents’ actual rather than intended behaviour: The ACT First initiative tracks real actions that residents have taken. It does not poll people, or ask their opinion regarding intended behaviour. This helps to convert attitudes and intentions into real behaviour change, as well as preparing people for extreme weather events.
  • Commitment bias: The website prompts residents to make their own emergency plans, chunking down the steps to make it simple to do – which helps them to think through each of the components. By thinking through the detail and recording their plans, residents are likely to feel more committed to them. When we have made plans – particularly when we have announced them publicly – we are more likely to see them through.

Crawford Hollingworth is co-founder of The Behavioural Architects