In a crisis, promote reliability (Part II)

This is Part II of a three-part series about effective communication during crises like our current pandemic of Covid-19. In Part I argued To avoid panic, promote efficacy. This post looks at how we can craft reliable communications during times of crisis. A third post will take a longer view, looking at anticipating inequity in how we respond to crises.

I’m calling this model TAPPER.



And always (3.) Anticipate Inequity.

Don’t need the explanation?
Download a one-page tip sheet here.

I was initially concerned this post might be a bit technical, relevant only to experts and insiders. But reliability has become a major political issue over the past week of the online conversation about Covid-19 and Australia’s response to the outbreak. People have been making and circulating ‘home-brew’ graphs, many from engineers and IT guys. Without any modelling skills at all, and without any information on Australia’s outbreak other than infection numbers, they visualise doubling time and little else.

A small crew of smart friends and I have spent the whole week trying to slap this stuff down on twitter. At a time when we are seeing stock-outs of medicines, packaged food and cleaning supplies, stoking panic with talk of exponential spread is like crying fire in a crowded theatre.

(I was delighted to see this post on Friday, Flatten the Curve of Armchair Epidemiology, with suggestions for helping curb the recent outbreak of a concerning pathogen, DKE-19.)

In my last post I talked about framing messages to prevent panic. To be clear, I mean panic in its ordinary sense, rather than a clinical state of panic. There is a lot of that about, right now.

It’s a bad idea to establish a direct line from your pixels, through your retinas, to your amygdala. I’ll write more about maintaining mental health during a crisis in another post. In the meantime, take a break, take a breath, or take a seat.

Managing panic in official communications

A lot of panic has been channeled into anger with the government for not taking more drastic control measures, like closing down schools. Others, mainly silverbacks with relevance deprivation syndrome, have demanded GPS ankle bracelets for people with confirmed infection, and nationwide lockdowns.

Bill Bowtell in particular has been claiming the Australian government is relying on herd immunity to curb the epidemic. This is a flat-out lie, but it’s a touchpoint for community anger. As I predicted in Part I, panic and anger make it harder to process rational messages, such as the Chief Medical Officer (CMO) Brendan Murphy explaining we don’t currently need school closures because children so far don’t appear to play a major role in chains of transmission.

Panic also means audience members can tend to dismiss ‘emotionally dissonant’ messages. If I’m really anxious, and the PM and CMO are saying ‘keep calm, it’s under control,’ I’m likely to dismiss that message, mistrust those spokespeople, and get even more angry that ‘we’re not doing enough.’ This may explain why the first government campaign was not well received.

The PM and CMO have been doing press conferences every day or so. These have ranked well against the recommendations I covered in Part I — detailing specific actions that we can personally and collectively take, demonstrating adherence to expert medical advice, emphasising unity among the ‘national cabinet’ (including premiers and chief ministers), and, usually during follow-up questions, carefully explaining the rationale for these actions.

For the most part, the press conferences are doing the right things, and during a rapidly evolving situation they are a more effective communication strategy than campaigns, which have a long lead-time and can quickly go out of date.

But if the government can’t manage panic, then all its careful work and good advice will go by the wayside.

And this is where I have concerns. The Prime Minister keeps saying we shouldn’t panic, that twitter ‘isn’t real,’ we should only listen to the advice on the Australian government website, and that our Australian-ness will see us through this crisis.

Now, this is a country that had a race panic about Asian people and restaurants as vectors for coronavirus, so I question the premise, but it’s not intended to be taken at face value. He’s gaslighting anyone who has concerns with his approach, in the best (or only) way he knows how, which is dog-whistling to Aussie ‘battlers’ that those concerns are un-Australian. As Billy Leask has observed, in these moments our PM is doing ‘don’t panic’ communication rather than effective risk communication.

The epidemic is fundamentally political, and I am referring to participatory democracy, not two-party politics. We all have a personal stake in the governance of the epidemic as a health issue. It also affects the economy, it involves questions of income and racial inequity (on which, more in Part III), it concerns access to healthcare, it impacts on international relations—and so on.

Preventing panic is not grounds for dismissing political participation.

However… neither is partisan politics an excuse for stoking panic. This past week, it has felt like there is a mobile army of retirees on twitter, lurching from issue to issue, mobilised by their blinding hatred for the PM (‘Crime Minister’) and screaming with absolute certainty that what we need is school closures RIGHT NOW.

And, I mean, I get it. Everyone needs a hobby, and a club, and to feel relevant. But while this issue is political, it isn’t partisan. There are five Labor premiers and four LNP in the ‘national cabinet.’ Most of the actions Scomo has taken can be traced to the published advice of the AHPPC, which consists of the Chief Health Officers of every state and territory and the CMO.

The suggestion that we should all ignore everything except official government advice tees me up for this piece, looking at reliable communication in a crisis.

Suddenly, reliability is a political imperative.

We live in an information ecology

The days in which everyone tuned in to hear Walter Cronkite deliver the truth are decades past. We all receive information from many directions at once. There are many demands on our attention and we have to make choices — who to attend to, what to focus on, and ultimately, what to believe. In research into in communication and culture, we talk about the ‘information ecology’ (much supply) and ‘attention economy’ (very demand).

I have prepared an infographic.

Actually this was media researcher Natalie Henry’s work—thanks Nat!

Even if people followed the PM’s advice that ‘twitter isn’t real,’ they’re still listening to radio, watching TV, reading the papers, commenting on FB posts.

Simply being a medical expert, or a scientific researcher, or a government minister, does not guarantee that people will find you credible.

As Alexis Madrigal put it, ‘Even if you avoid the conspiracy theories, tweeting through a global emergency is messy, context-free, and disorienting.’

The information ecology isn’t a bad thing, necessarily—it means we get to see knowledge in motion and hear different perspectives. But it places a cognitive burden on readers to work out what’s reliable.

Trust markers make it easier for readers to assess the reliability of online communication.

A lot of DIY attempts at public education leave out trust markers altogether. This makes it hard to assess whether they are trustworthy. Many readers will take these amateur efforts seriously. But the ecology includes other people. Trust markers give other participants in the conversation the ability to say, ‘this advice is more trustworthy than your meme, Uncle Bill.’

Examples of trust markers

The Nielsen Norman Group has identified four broad ways of communicating trustworthiness in online communication. That’s a respected consultancy founded by web usability expert Jacob Nielsen, and Don Norman, a key figure in human computer interface research.

The four domains are:

  1. Upfront disclosure
  2. Comprehensive, correct, and current (for fast-moving crises, I’m adapting this domain to careful, current, and clear in scope)
  3. Connected to the rest of the conversation
  4. Design qualities

I hope everyone has read this great piece by epidemiologist Dr Kat Snow at the University of Melbourne, explaining what we know and what we can do about the outbreak. On twitter I circulated this image, which highlights the different ‘trust markers’ that are present in Dr Snow’s blog post. Below, I draw out some key lessons from my analysis of this piece and my experience in practice.

Click this image to view the full page.

1. Upfront disclosure

It should be clear: who made this, what are their credentials, when was it made, where does it apply, and why (for what purpose)?

In the piece by Dr Snow, the authorship is clear. We can see her name, training, location, role and institution. We can assess whether her experience and expertise are relevant to the crisis and her argument. (If a person’s bio is all about their media experience, that’s not good.)

In a rapidly changing situation, the publication date is a crucial detail for assessing reliability. Even carefully developed messages will become out-of-date in future. Likewise, say where you’re talking about. Dr Snow writes:

Note: This blog has been written on the 13th of March 2020. The situation is changing rapidly and some of the information below may be out of date in the coming weeks or months.

Even if you are making a graph or meme, it can easily circulate online and be lifted out of its context, so include these details in the image.

2. Careful, current and clear in scope

As discussed in Part I, to discourage panic, the tone and style should promote personal, response and collective efficacy. Here’s a recent example.

It can be difficult to comprehensively sum up a rapidly changing situation, but we can be clear about what is currently known and unknown.

In Dr Snow’s piece, the tone and style are clear and measured, avoiding alarmist language (e.g., ‘the outbreak… is very intense in Italy’). Her statements use ‘if,’ ’could,’ ‘may,’ ‘can,’ rather than ‘will,’ and this signals that they are predictions rather than guaranteed to happen.

The piece is narrative in style. This happened, so we are recommending this action, and it may have that effect. (Past, present, future.) This can make it easy for people to understand the piece.

However, very vivid narratives—such as those we are hearing from doctors in Italy—can feed into a powerful bias in our calculation of risks. As the Nobel Prize winning psychologist Daniel Kahneman has demonstrated, we overestimate the likelihood of events we can easily and vividly imagine.

It is important to ground these narratives with details of the context. With Italy, it is hard to do this, because their local epidemic is so poorly understood.

Lastly, Dr Snow acknowledges that coronavirus risks, needs, and experiences are diverse. It is very hard to assess the reliability of recommendations that are presented as universal—without upfront disclosure and a clear scope. Such recommendations often embed assumptions from their country of origin, such as expectations about health care access.

3. Connected to the rest of the conversation

Governments wish people would read only their own websites. But those sites give very general advice, to avoid overwhelming visitors with information. As I mentioned in Part I, people need high-level advice translated into concrete, personal, everyday practices.

Fearing liability, official sites do their best to avoid linking to any other sites for advice over which they have no quality control. They say, ‘for more information call our information line,’ as if those lines aren’t already clogged with hyper-specific requests for information and reassurance.

Connecting your communication with the rest of the conversation gives people pathways to find additional information and to ‘triangulate’ your own advice.

There is nothing less trustworthy than online communication that does not cite its sources and show its workings enough to understand its key claims. This is another reason why you should never trust graphs if there are no details on how they were produced.

In Dr Snow’s piece, she provides live-linked sources for major claims. When describing recommendations, she identifies which body says what. This can avoid confusion for readers who may be hearing different things in different countries — including, potentially, sub-par advice.

Dr Snow includes one graph that is relevant to the argument and appropriate to the health literacy of her intended audience, the general public.

Some other pieces have been choking with a glut of graphs from all different sources, timeframes, and based on contradictory statistical assumptions. Their purpose often seems to be making the author seem really science-y.

Exhibit A is Silicon Valley entrepreneur and viral content specialist Tomas Pueyo, whose home-brew analysis has over 35m visits.

4. Design quality

This heart of this factor may not be what you’re expecting. People often think ‘good design’ means an appealing visual appearance. However, good design means two things:

  1. its production takes the intended (and possible) users’ needs into account
  2. the format is fit-for-purpose — users can achieve its goals

Production includes consultation, research, communication strategy, message development, writing, proof-reading, sense-checking with users… every part of the online communication process. Adopting the audience perspective is what distinguishes communication from information-giving, which, as we saw in Part I, does not change anyone’s behaviour on its own.

Dr Snow anticipates the health literacy of the intended audience and supports the sense-making goal of health promotion, tapping into their existing knowledge and using clear metaphors to contextualise new information.

The more we can slow down the spread of the virus, the lower this peak demand will be – like turning off your air conditioning on a hot day to spare the electricity grid.

The piece defines key concepts and provides the rationale for the strategies it describes and the actions it recommends people take. Dr Snow assumes that readers want to understand — after all ‘sense-making’ just means ‘making sense of…’ She does not expect users to mindlessly follow directions from scientific experts, doctors, or the government.

Instead of stoking alarm, the piece motivates action by emphasising the gravity of the situation and the efficacy of the actions we can all take to manage it.

These types of measures are hugely disruptive and expensive. But this is a very serious situation, and this is what we need to do to slow the virus down. If we do that, we can protect ourselves, each other, and our health system until we have a vaccine.

In Part III, I am going to build on these reflections. I’ll talk about how, in designing both online communications and government responses to crises, we need to anticipate the diversity of the community, their different needs and risks, and the social processes that put some groups at greater ‘risk of risks’ (Frohlich & Potvin, 2008).


In my first post, I covered tips for communicating without causing panic — and then we saw a week of concerted efforts to do just that. They highlight the importance of developing reliable online communications that can out-compete these shonky messages. That happens over the medium term, however. Unfortunately, there is nothing we can do to stop it as it happens.

In this post, I connected the reliability of communication with the politics of pandemic preparedness. Reliability is at a premium if we want to have an informed public debate about the control measures we can take in a pandemic.

This perspective highlights that health promotion is not just education, not just giving people clear and accurate information — it’s fundamentally about increasing people’s control over their own health, at a personal level and a community level, via personal behaviour and political participation. It’s encapsulated in the founding document, the Ottawa Charter for Health Promotion (WHO Europe, 1986).

My next post will pose a thorny question. Very often, we respond in the short term with mass communications that reach people with good health literacy. We leave the ‘hard to reach groups’ for later.

This is how health inequities emerge — we take longer to reach people who take longer to access help, and we never seem to anticipate inequities and ‘respond before they emerge.’

If you would like to support the work I am doing, a small donation would be gratefully accepted — but sharing this post would be equally awesome! Donations will be used to support this blog and reliable, independent health promotion initiatives like