Best practices for risk communication from the academic literature

Introduction

The EA community has seen a proliferation of organizations and individuals focusing on existential risk with exciting and ambitious projects. This post is the first of a series of posts hoping to provide insight into the specifics of communicating about risk. This work is funded by a grant from the Long Term Future Fund. In this post, I’ll be outlining the best practices laid out in the academic literature on risk communication to provide a rough guide to the risk communication process.

A note about sources and methods: A list of sources is provided at the end, and specific figures or claims will be cited when necessary; but it’s worth noting that overall, most of the steps or tips provided in the post are not cited to specific texts because they were present virtually across the board. This indicates that there is a broad consensus about these components in contemporary risk communication studies.

The National Academy of Sciences has defined risk communication as:

…an interactive process of exchange of information and opinion among individuals, groups, and institutions. It involves multiple messages about the nature of risk and other messages, not strictly about risk, that express concerns, opinions, or reactions to risk messages or to legal and institutional arrangements for risk management, (p. 21; italics in original). To put it simply, risk communication is the act of communicating information about risks to stakeholders, often with the goal of engendering some kind of action to prevent or mitigate the risk in question. In a more technical sense, risk communication is an evidence-based endeavor to deliver this kind of information in the most effective and ethical way possible. Extant research on risk communication generally examines risk communication processes from the regulatory perspective, focusing on how the government and affiliated organizations who are tasked with protecting citizens from risk. Another dominant arm of the literature focuses of health communication and how physicians, public health organizations, or other experts and healthcare providers can inform and empower patients to make well informed and appropriate decisions for their own care.

We can point to four primary functions of risk communication:

  1. The enlightenment function: fostering understanding of risks.

  2. The behavioral change function: fostering behavioral change to reduce or mitigate risk.

  3. The trust function: promoting the credibility of institutions that handle or regulate risks.

  4. The participative function: providing procedures for dialogue, conflict resolution, and democratic planning for risk management and regulation.

Any instance of risk communication will not necessarily fulfill all these functions.

Why information transfer isn’t enough

Risk communication also shares the same challenges of other types of public communication, namely:

  1. Imprecision: communication is mired in the filters and biases of both sender and receiver.

  2. Irreversibleness: Once a message is out there, it’s out there. Miscommunication or misapprehension of a message requires extra work to rectify, and it’s near impossible to remove any incorrect messages from the public record.

  3. Contextuality: the context within which a message is produced, delivered, and received changes the framing, significance and substance of the information it communicates.

It also carries its own specific challenges, stemming from the issues it addresses and the nature of risk itself. While the social sciences and humanities are able to distinguish a variety of approaches to risk, for the purposes of this guide I’d like to draw attention to a much rougher differentiation with direct implications for practical use: a technocratic approach versus a socially situated one. A technocratic approach to risk deal with risk and hazards as a primarily scientific question of probability, and centers primarily on information transfer as an objective and rational process. Socially situated approaches to risk consider both hazards and their potential outcomes within their full social, political, and economic context.

In everyday life, risk can essentially be understood as an experience of uncertainty and lack of control, rooted in some kind of existential anxiety. Thinking about and acting on risk is closely related to emotions rather than being a purely intellectual and rational experience. More importantly, a thing being defined and recognized a risk involves several primarily social assumptions regarding what is at stake and what potential tradeoffs are considered acceptable.

In most cases, averting risks requires some form of change. This can be to individual behaviors (changes in diet, stopping smoking, choosing sustainable transportation) or broader reaching systemic change (banning the purchase of certain products, changing aspects of the political economic system to collectivize certain types of risk). This can also extend to a challenge to deeply held beliefs and worldviews on the legal and social regulation of everyday life – for example, while banning the consumption of hard drugs is generally socially acceptable, a proposal for banning the use of alcohol would be considered and overreach and generate a great amount of debate about the degree of acceptable individual risk bearing and public responsibility.

These deeper political and social dimensions of risk debates may seem superfluous for those involved in the study of the actual risk at hand, particularly when the risk is deeply important or even existential, but must be recognized for healthy risk communication. This is not only a matter of ethical conduct and respect – in many cases, risk communicators are speaking from a position of expertise or legal/​political power, and the asymmetrical relationship between these positions and the public, to some degree, creates a duty of care and consideration for the affected population – but also a practical one: because of the emotionally fraught and adversarial nature of most risk debates, most people will require more than a simple information transfer. Recognizing their perspective and the other factors that go into their understanding and estimation of risks will allow risk communicators to better understand what information is and isn’t salient and how to make sure that their messages are understood and acted upon.

Another way to understand this is to consider the different components of debates around risk. Renn (p. 82) differentiates three levels of risk debates, each with its own communicative and evaluative needs:

LevelIssue of ConflictCommunication NeedsEvaluation Criteria
1Technical expertiseInformation transfer

Access to audience

Comprehensibility

Attention to public concerns

Acknowledgement of framing problems

2Experience, trustworthinessDialogue

Match between stakeholders and public expectations and agency performance

Openness to public demands

Regular consultations

Commonly agreed procedures for crisis situations

3

Vales

Worldviews

Dialogue

Mediation

Fair representation of all affected parties

Addressing the concerns of all parties in the debate

Involvement of major stakeholders

Transparent and inclusive form of decision-making

While all three of these levels won’t be present and salient to the same degree in all risk debates or for all actors, it allows risk communicators to better understand what exactly they’re communicating and how their message is likely to be processed. While the science of risk can be reduced to the statistics and science, risk related practices will come into contact and be filtered through all three of these levels, and taking an exclusively technocratic approach will very likely lead to ineffective messaging and undesirable outcomes.

It’s also important to understand that, to put it plainly, most people don’t really understand risk the way a scientist or statistician can. Even setting aside statistical or scientific illiteracy, there are psychological and social factors that can lead to misperceptions and misunderstandings about risks. Covello (2010) outlines several of these factors:

  • Availability: The frequency of events that are easily “available” the people – events that are memorable or easily accessible – are often overestimated, leading to overestimation of probability.

  • Conformity: People tend to behave in certain ways or believe certain things because they’re what most people around them do or believe.

  • Overconfidence in ability to avoid harm.

  • Confirmation bias: New information that contradicts one’s present beliefs about a risk or behavior tends to get filtered out, while information that confirms those beliefs tend to get filtered in.

  • Aversion to uncertainty: People will tend to demand exact information, which is often unavailable in regard to risks.

  • Reluctance to change strongly held beliefs.

Bratic Arkin (1989) provides a non-exhaustive list of additional and specific challenges in risk communication:

  • Risk is an intangible concept.

  • The public responds to easy solutions.

  • The public calls for absolute answers.

  • The public reacts unfavorably to fear.

  • The public wants personal control.

  • The public doubts the verity of science.

  • The public has other priorities.

  • Individuals are not personally susceptible: people have a strong tendency to underestimate their personal risk.

  • The public holds contradictory beliefs.

  • The public lacks a future orientation: lower income groups in particular have difficulty envisaging and relating to the future.

  • The public personalizes new information: individuals need to be able to translate population or society level information (like morbidity rates) into personal risk, which is often challenging.

  • The public does not understand science.

Additionally, risk communication frequently occurs in adversarial contexts and centers around issues that are emotionally fraught and prone to outrage. In this context, it’s important to approach risk communication from an informed, intentional, and ethical perspective.

Best practices in risk communication

As with most things, risk communication efforts will go better if you have a plan. For risk communication, Zummo Forney and Sadar propose that a plan should include the following information:

  • Purpose and objectives

  • Stakeholders

  • Communication team

  • Communication risks and mitigation

  • Key messages

  • Communication tools and activities

  • Action timeline and calendar

  • Estimated costs

  • Performance evaluation

Purpose and objectives: know your goals

Have clear and well-defined goals for your risk communication. These goals should be:

  • Aligned with your organization’s policies and mission statements.

  • Actionable.

  • Appropriate for your objective.

  • Realistic.

  • Measurable.

A well-defined goal will be the articulation of the desired outcome of your project and will guide you in choosing your target audience and constructing your messages. Too broad or nebulous goals, like “inform people about AI risk” will lead to risk communication messages and practices that have no clear target audience and no specific actionable takeaways, meaning that a lot of potential momentum and impact will be lost, and that the overall effectiveness of your project will be significantly diminished. Instead, a goal such as “Inform people about AI risks to motivate the choice of certain career paths related to AI risk reduction” allows you to tailor your messages both to a desired outcome and a specific target groups—which can later be chosen on the basis of which groups will be most likely to choose a career path based on new information—which in turn allows for the selection of appropriate tone, channels, and content to actually affect the change you want to see.

Stakeholders: Know your audience

An otherwise robust message delivered to an irrelevant or incapable audience will have no impact. Once your goals have been set, it’s important to determine who the proper target audience for your message is. Different audiences will have different goals, needs, and abilities, and tailoring risk communication messages to relevant groups allows the message to be better understood and actionable.

Identify the impacted or capable groups.

Usually, there will be a wide array of potential audiences for risk communication messages about a given issue. Policymakers, government officials, effected populations, and private organizations and people making decisions about the issue (such as the developers of a technology that poses risk) are all potential audiences. Although non-exhaustive, Palmund’s list of actors and their roles in relation to climate change is an example of how different actors engage with a certain risk:

ActorRoles in conflicts over global climate change
Risk bearers

Communities in low-lying areas prone to flooding.

People in areas subject to drought.

Humanity at large

Risk bearers’ advocates

Activists in environmental organizations.

Some scientists.

Some politicians.

Risk generators

Stakeholders in industry, agriculture and transport systems emitting large quantities of green-house gases.

Consumers not choosing energy efficient housing and transport means.

Risk researchersScientists.
Risk arbitersGovernments and governmental agencies.
Risk informersProducers, journalists, writers etc. in media (newspapers, television, books, films etc.)

Partially reproduced from Palmlund p. 199

Each group listed here has a different role to play in relation to the risk at hand, has different levels of preexisting knowledge and literacy, and has different incentives and desires in relation to the risk and any action taken (or not taken) to reduce or mitigate it. As such, each group will require a different type of communication carrying different information.

For maximum effectiveness, every instance of risk communication should have a well defined primary target audience. The choice of audience will depend on several considerations, including but not limited to:

  • The kind of work the organization or person doing the communicating is involved in.

  • The effected population: who is at risk and to what extent. It’s possible that there are segments within the broader effected group that will be more effected than others.

  • Level of existing knowledge. Effected or capable groups, or segments of these groups, may already have access to the relevant information, making certain types of risk communication redundant or irrelevant.

  • Who is able to take necessary action: when working to communicate risks, it’s obviously important to target the audience that is able to take the intended actions. Some risks like climate change are relevant to virtually all people, but not all people have the same capability to take action to prevent or mitigate the risk or harms related to it, some risks like workplace exposure to harmful materials will only impact a relatively small group of people, while some risks such as economic harms from corporate corruption will only be actionable by a small group of people, like regulatory legislators. Broadly speaking, these capable actors will either be

    • the group that will be impacted by the risk and can either avoid it (for example, by not using a potentially harmful product)

    • those who can take action to mitigate risk,

      • by being in a decision-making position to take systemic or institutional action or

      • by being able to take personal action to mitigate risk to the self or community (for example, take steps for community resilience against disasters by forming community search and rescue groups) or participate in a broader action (for example, regular recycling to decrease environmental impact).

Ultimately, the choice of audience is closely related to the goals of the organization and their theory of change.

Understand your audience

Beyond simple identification, it’s important to understand the particular characteristics of the audience being spoken to. This involves:

  1. Understanding their concerns about the risk and its context: This will allow for an appropriate framing of the risk. It’s easy to alienate an audience when discussing a politically or economically controversial risk if there’s an impression that the risk is overblown or suggested precautions would be against their best interest. It also allows for the calibration of proposed actions to be more realistic and attainable. For example, speaking to inhabitants of a town where the primary economic activity is mining about the environmental and health risks of their work can run the risk of alienating the audience if the message overlooks any economic difficulties that may be caused by changing the standard operation of the mining industry, or if the message is framed in such a way to make the miners and their dependents feel under attack by sanctimonious outsiders telling them they’re destroying the environment. This is particularly important if the target audience is policy professionals. Policy making and legislation is an activity that is embedded in a complex web of consideration and relationships, and effective advocacy requires the recognition of this context and how it influences realistic policy goals and outcomes.

  2. Understanding their level of literacy and technical competence: Much of risk communication involves messages about probabilistic and at least somewhat technical risks. The presentation of the relevant information needs to be in an appropriate format and at an appropriate level, or risks being misunderstood or simply ignored by the intended recipients. Zipkin et. al. (Zipkin et al., 2014) draw attention to the “collective statistical illiteracy” as a barrier to risk communication in the healthcare setting that must be recognized and addressed. While it’s important to calibrate the presentation of technical information for a lay audience, it’s equally as important to consider the legibility and relevance of such information when speaking to expert or professional audiences as well. Policymakers and legislators will require a certain standard of evidence and potentially additional information when considering the necessity, appropriateness, and feasibility of proposed action. While it’s more reasonable to expect that they will be able to parse technical information, these offices are generally working under strict budgetary and time constraints, meaning that the use of formats like policy briefs with appendices with additional detailed information makes it more likely that they will be able to consider the provided information.

  3. Understand what the audience can actually do: Having expectations that are too high will predictably lead to failure. An example of this can be found in public health communications: a perfect public health communication campaign encouraging vaccination cannot reasonably lead to behavioral change if the audience cannot afford to get vaccinated. Not only will this not lead to the intended consequences, it can have negative outcomes if the framing of messages were done in such a way as to moralize the intended actions and cast non-compliance in a negative light by shifting blame and responsibility from capable and responsible actors to those who are not able to take meaningful actions for change.

Listen to your audience

While risk and science communication has long been understood as a one way street where an expert educates others about a certain issue, contemporary literature is in broad agreement that

There is a significant body of research that focuses on risk communication as an interactive process between various stakeholders. Virtually all of this literature conceptualizes or at least focalizes risk communication as a regulatory activity, undertaken by the government or affiliated agencies, and as such will not be covered in detail in this write up. Broadly speaking, it argues the necessity of the involvement of all stakeholders, but particularly the public, in the risk communication and decision making process, and the merits of treating the process as a dialogic and recursive undertaking rather than a linear, one way monologue. While much of this is not applicable for the organizations this review is targeted at, one important insight is the necessity of listening to the audience risk communication messages are targeted at. This is both an important part of a feedback loop that will lead to effective communication and change, by allowing communicators to understand how their message is being received and what potential barriers to uptake and compliance are, and an important part of ethical risk

Communication team: be credible

A major factor determining the efficacy of risk communication is the credibility of the message, which in turn is strongly related to the credibility of the organization or person delivering it. So fostering trust and credibility is an important part of effective risk communication.

There are a lot of factors that go into the perceived credibility of an actor. Some of these will be out of your control: some relatively immutable aspects of personal or organizational identity, such as political leanings, funding affiliations or past actions, can cast doubt over an otherwise robust message. If an organization delivering messages about lung cancer risks is funded through a grant provided by a tobacco company, people will naturally be more skeptical about the intentions of the organization and the message itself. If an organization has been wrong before, there will be more suspicion about them actually being right this time around. However, there are many things that can help build trust and credibility. Covello (2010, p. 89) defines six components of trust:

ComponentDescription
Perceived competenceDegree of technical expertise in meeting institutional mandate
ObjectivityLack of biases in information and performance as perceived by others.
FairnessAcknowledgement and adequate representation of all relevant points of view
ConsistencyPredictability of arguments and behavior based on past experience and previous communication efforts
SincerityHonesty and openness
FaithPerception of “good will” in performance and communication

Building off of these components and the literature, a list of assorted tips to help build and keep trust and credibility:

  • Be clear about what you do and don’t know, and what can and can’t be known at the moment.

  • Listen to your audience and make it clear that their needs and concerns are being considered.

  • Follow through on commitments and statements.

  • Walk the walk. The actions of the organization should line up with their messaging. There’s a reason companies like BP are met with a certain degree of disdain when speaking about how individuals must act to avert climate change.

  • Be consistent. While details and the level of information provided can change at different times or when speaking to different audiences, ensure consistency in key messages, and clearly address and explain any necessary inconsistencies.

  • Act with integrity. The halo effect is at play. Corruption, malfeasance, inappropriate behaviors etc. can all act to discredit the otherwise correct and valuable message of an actor, both in the public eye and in professional sectors.

  • Be humble. Overstatements and grandiosity are unlikely to lead to a well received messaged, and being confidently wrong about something will do more damage than being wrong with the necessary caveats and disclaimers.

  • The channels and mediums used for communication can also lend to or detract from credibility, depending on the target audience. Some audiences will consider information disseminated on platforms like TikTok with a degree of disdain, while others will write off information in the New York Times. Knowing how and where your target audience receives trusted information is important for establishing and maintaining credibility.

  • If you were wrong about something, apologize with sincerity.

  • Recognize and address “uncomfortable” facts. If you’re communicating on a topic that is contentious or in an adversarial discourse, which you almost certainly will be, any facts that you omit or gloss over will most likely be used against you. More importantly, it’s ethically important to recognize any facts or information that may work against your message, even when aiming to persuade.

  • Acknowledge the perceptions of stakeholders. Even if they are not in line with factual reality, it’s important to treat the people you’re communicating to with dignity and respect, and being sensitive to their worldview will buy you a lot of credit.

  • Acknowledge uncertainty. Most, if not all, risk involves a degree of uncertainty. Disregarding this will not do any good.

  • Be mindful of comparisons. Many risks involve a degree of outrage, and comparing risks that carry a high degree of outrage to things that do not will most likely not be taken well. Perceptions of flippancy or disregard are likely to undermine trust.

  • Be mindful of your tone. Most people do not appreciate being lectured to and will have a hard time building a trustful relationship if they feel they are being patronized, looked down on, or not taken seriously.

  • Use trusted voices.

  • Establish relevant credentials, and avoid hiding behind irrelevant ones. A professor of psychology is not necessarily qualified to speak about the dangers of petrochemicals, and using their academic credentials to establish authority is both likely to backfire and is arguably unethical.

  • Cite your sources.

  • When it makes sense to do so, explain the reasoning and process behind the risk estimation and proposals to mitigate. This may not be appropriate to put on a one page flyer, but having a website laying out more detail in accessible and clear language will help those willing to do the work get a stronger understanding and provide you with more credibility for your transparency and accessibility.

  • It can be helpful to have a designated person carrying out communication tasks. This presents a consistent face for your organization and message, which can help build trust with audiences – an identifiable person is easier to build a trust relationship with than a faceless organization.

Key messages, tools and activities: constructing your message

Develop key messages

Determine the key points your message should be getting across and formulate key messages around these points. Key messages should essentially be the gist of your communication platform: if someone only has one or two minutes to hear you out, this is what you tell them. It’s helpful to think about both what you think your audience must know and what they’re likely to want to know when determining these points. Key messages will allow you to prioritize and appropriately frame information, ensure consistency, continuity and accuracy across all instances of communication, and minimize confusion and error. Based on previous works, Zummo Forney and Sadar argue that effective key messages should be concise, simple, strategic, relevant to the audience, compelling, memorable, real, and tailored to the audience.

Tailor your message

Tailoring of the message can be thought about in three ways.

  1. Tailor your message to your risk.

    1. Be timely.

    2. Avoid extraneous information while providing as complete a picture as possible.

    3. Take the necessary and appropriate tone. Flippant messages about serious risks or extremely grave messages about trivial risks will most likely not be met with the desired response.

  2. Tailor your message to the needs of your target audience.

    1. Provide the information that they need to know.

    2. Provide the information that they want to know.

    3. Acknowledge and deal with the concerns and perceptions of the target audience.

    4. Respect emotional and social needs alongside informational ones.

  3. Tailor your message to the capabilities of your audience.

    1. Make sure any desired or proposed actions are ones that your audience can take.

    2. Understand the level of literacy of the audience and present information accordingly.

Lundgren and McMakin (p.101) provide insight into how information about audiences can be used to tailor messages:

Information learnedHow to tailor the message
Audience unawareUse graphic method—high color, compelling visuals, and theme.
Audience apathetic (or feels like victims)

Open risk assessment and management process to stakeholder participation.

Show where past interactions have made a difference.

Provide choices.

Audience well informedBuild on past information.
Audience hostile

Acknowledge concerns and feelings.

Identify common ground.

Open risk assessment and management process to stakeholder participation.

Audience highly educatedUse more sophisticated language and structure.
Audience not highly educated

Use less sophisticated language and structure.

Make structure highly visible, not subtle.

Who the audience trustsUse that person to present risk information.
Where the audience feels comfortableHold meetings in that location.
The method by which the audience gets most of its informationUse that method to convey your message.
Who makes up the audienceEnsure that the message reaches each member.
How the audience wants to be involved in risk assessment or managementIf at all possible, given time, funding, and organizational constraints, involve the audience in the way they want to be involved.
Misconceptions of risk or process

Acknowledge misconceptions.

Provide facts to fill gaps in knowledge and correct false impressions.

Audience concernsAcknowledge concerns and provide relevant facts.

Presenting data

  • Use the same denominator, especially when comparing probabilities.

  • When using verbal expressions of probability (likely, possible, probable, etc.) be mindful that these can be understood differently by different people, even if they refer to defined probability intervals within your field. It’s generally useful to pair these expressions with numerical expressions of probability and vice versa to ensure everyone is on the same page. Using additional visual representations when possible can be particularly useful.

  • Frequencies over percentages: Systematic reviews such as Zipkin et. al (2014) have found that when speaking about risk and prevalence, using frequencies over percentages, such as formulating the prevalence of a certain type of cancer as “1 in 5 people” rather than “20% of people” is more effective for lay people.

  • Data visualization: Visual displays of data, probability, and risk factors in the form of charts or graphs can do a lot of work in overcoming statistical illiteracy. It’s also important to note that different types of visualizations provide different levels of comprehension. While some types of visualization (ie. box and whisker plots) are obviously difficult to understand for those without statistical training, simpler representations like bar and pie charts can also provide different levels of comprehension and miscomprehension depending on the specifics of what is being communicated and how the visuals are displayed. Other factors can also influence how these figures are read. For example, being placed close together on the page or visual similarity can be understood as implying comparison where there is none. It isn’t easy or really feasible to provide a broad overview of which visualization techniques work best for every case because it can vary across types of risk (health risks, environmental risks, etc.), the statistical literacy of the target group, the nature of the data (comparative, probabilistic, etc.), and context and familiarity. This is a case where testing is particularly useful to understand how the figures and statistics in your risk messages are being understood or misunderstood, instead of assuming something like a pie chart or a flowchart of possible outcomes will be easily and fully understood by your target audience. In the meantime, Eppler & Aeschimann (Eppler & Aeschimann, 2009) provide a useful and brief guide to using different types of visualization methods not just for statistical information but also for the risk process overall, and Zipkin et al. (2014) and Yang (Yang, 2020) provide brief but broad overviews of published research on the efficacy of specific visualization techniques which can provide a good jumping off point for your specific goals and subject matter.

  • While initial presentations of data, particularly when speaking to lay audiences, should not be bogged down in excessive detail and explanation to avoid information overload or confusion, providing access to reader friendly information about how data is collected, processed, understood, and contextualized to reach certain conclusions can be very valuable to bolster the credibility of the message and allow interested and affected parties to become more involved and better informed about the issue.

Framing matters

The way your message is framed can change how it is understood. One example from health communication research is the finding that, when speaking about potentially harmful interventions like invasive surgery, when physicians framed the surgery in terms of potential gains rather than potential losses, patients reported lesser perceptions of harm and were more willing to accept potentially harmful interventions.

Framing is unavoidable in communication, meaning that even if your goal is to provide “objective” information with no intention of mobilization or behavior change, it’s necessary to pay attention to how your statements frame the issue. Minute details like the grammatical structure of sentences (can vs. cannot statements, for example), the order in which harms and benefits are presented, and the general tone of the message and its delivery will all have some impact on how your message is received.

To combat any unintended framing effects, and to temper any intended effects to ensure ethical and non-manipulative risk communication, once again, testing can be very useful. Seeing how a focus group reacts to and processes your message can help recognize where certain framing effects are coming into play and allow you to modify your message or add in additional information to counteract those framing effects.

Using video

Video can be a very powerful tool for risk communication, particularly when it appeals to emotions. Dramatized or narrative videos can provide better understanding and engagement over lecture-like videos that are primarily didactic. Incorporation of some form of interactivity can bring additional benefits in terms of increasing comprehension, salience, and increasing the longevity of imparted information. However, poorly made videos can alienate audiences, particularly younger people who are accustomed to the format.

Be mindful of wording

When wording your message, there are four main points that can lead to confusion or misunderstanding:

  1. Ambiguity: Words that have multiple meanings, for synonyms that are used interchangeably can lead to ambiguous messages.

  2. Vagueness: Some words can mean different things to different people, or not be clear enough in general and leave messages to vague to be acted upon or well understood.

  3. Underspecificity: Words or phrases that are too broad or general can lead to differing interpretations of the message.

  4. Context dependence: Certain words can carry different significations when used in different context or when placed next to other words.

Test your message

There is a lot of information about how certain phrases or graphs are understood or misunderstood, or about how certain groups prefer to receive information. These present a good jumping off point for your specific needs, but cannot – because they’re generally context dependent – and should not be where the evidence in your evidence-based communication ends and begins. If you consider risk communication to be an important part of your work, you should test your messages as much as possible. In my personal opinion, this is perhaps the most important take away from extant risk communication literature. While it’s easier to generalize for specific subdomains – like how constituents prefer their elected officials or regulatory bodies to deliver information, or how cancer related risks are best communicated, or which types of risk communication can lead to higher uptake of potentially risky medical interventions – the more or less unique social, political, and economic dimensions of every particular risk means that the insights into one of these issues will not necessarily translate well to a different one. A patient’s understanding of how “likely” a surgery is to have complications will not be the same as someone’s understanding of how “likely” AI-caused catastrophes are when reading about it in a headline on Twitter. Language that comes across as respectful to a group of professionals or policymakers can cause feeling of frustration or hopelessness in a population with an average 8th grade level of literacy because they can’t understand what you’re saying.

So, whenever possible, test your messages with members of your target audience. This will allow you to:

  • Assess comprehension and figure out what parts of your message causes misunderstandings.

  • Assess retention and recall of information to understand what is sticking with people and what isn’t.

  • Identify which points may cause strong emotional reactions and backlash.

  • Identify any barriers to communication – cultural, linguistic, literacy related etc.

  • Identify any sensitive points that require added care or attention.

  • Assess the relevance of the information.

  • Identify any additional information that the target audience might want to have that isn’t included in the message.

Communication risks and mitigation: evaluate and realign

Once you’ve tested your message, you have the opportunity to realign your messaging in line with what you’ve learned. The testing process can show you what isn’t working and allow you to fix it, but not all issues will emerge during testing, and not all issues that do emerge will be easily fixable. However, it is possible to create a roadmap for the actual deployment of your message based on feedback. For example, if there are indications that the target audience doesn’t understand your risk estimates, you can provide additional information or present the data in different ways; or compensate for any cultural gaps in your messaging that are causing the audience to see your message as irrelevant.

It’s also fruitful to consider risk scenarios outside of what is indicated in your testing. Red teaming, and particularly red teaming by communication professionals or lay people who don’t have the same insight into the issue as you do, can be a useful tool here. Reading your own communication in the most negative, bad faith interpretation, for example, allows you to prepare for the objections of any adversarial parties outside of the high stress environment of actual confrontation or debate.

On the practical side of things, having logistical contingency plans in place for the delivery of your message is also a good thing.

Working with the media

In many cases, people will be hearing your message as mediated by the media. Even if your message is delivered directly to the affected groups, like in doctor patient interactions or town halls about local crises, there is a very high chance that some segments of your target audience or entirely different groups will hear about it through the news or other forms of media.

News media in particular can also a be a very strong tool for risk communication because of its reach, and for the legitimacy it lends less well known causes and organizations. Dealing with the media, however, presents its own set of considerations.

As a high level note, if the scope and budget of your project allows it, this may be a good area to delegate to professionals – they will have more contacts and a stronger sense of the media landscape, which reporters are better equipped and qualified to cover your issue, and will have more savvy in terms of specific wording and presentation for the media. On a logistical level, it also makes sense to have one designated person to deal with the media, even if they aren’t only assigned to that task or a PR or media professional. Having a specified contact person minimizes confusion and redundancy.

The media is its own specific context where certain incentives and disincentives will shape how your message is filtered and mediated. News media will tend to focus on issues that are highly emotionally charged or controversial. This can push their frames and focus towards conflict, failures, negligence, scandals, and risks and threats to children in particular. The narrative structure preferred in media in conjunction with the incentive to focus on emotional stories prioritizes stories that can be understood as narratives with villains, victims, and heroes. This is not the case for all media and all reporters, but has been reportedly pointed to as a consideration in several pieces on how to work with the media in delivering risk messages.

Covello provides a list of tips for working with the media, which is further expanded upon in the appendices of the same chapter. Some items that stand out as particularly useful are:

  • Prepare informational materials for journalists in advance of any interaction and allow them enough time to digest and understand these materials and ask any necessary follow up questions.

  • If possible, prepare a variety of materials that present different levels of detail. Depending on their workload and turnaround time, some reporters may not have time to read more than a one- or two-page press release, while others will benefit from extended explanations, background information, and FAQs.

  • Have designated people with sufficient seniority and communication skills interact with members of the press.

  • Provide regular updates and access to experts.

Emotion, affect and facts: the ethics of risk communication

Using persuasion

In most cases, risk communication aims to effect some kind of change. This incentivizes using persuasive rhetorical tactics in communication. This isn’t necessarily unethical: persuasion is a valid goal in risk communication, and in many cases those who are doing risk communication feel that persuading their audience is an ethical mandate in and of itself, because they are trying to avert or minimize risk. However, it is easy to slide into unethical territory, particularly when making judgement calls about what information is necessary or irrelevant. Persuading and informing people are goals that should be balanced throughout the process. Presenting information selectively or omitting conflicting information or leaning on manipulative framing to persuade your audience crosses ethical lines.

Dealing with uncertainty

Risk is inherently related to dealing with uncertainty. Talking about risk is by its nature talking about uncertainty, but the uncertainty about risks occurring is not the only kind of uncertainty involved in the process. A lot of risk issues are not just uncertain in regard to the risk of the event occurring, but also in regards to the science behind the risk itself. Uncertainties in this vein – confidence intervals, so to speak – should be made clear to the audience.

Emotional appeals

Appealing to and dealing with emotions and affect in an important part of risk communication. But again, there is a line that must be treaded carefully. The elicited emotional reaction must be in line with the issue at hand. Eliciting unwarranted emotional reactions in the vein of moral panics or scare tactics is unethical and manipulative. Another issue is related to proposed pathways. Emotional appeals that are out of sync with actionable steps can lead to panic, hysteria, and apathy, which can in turn lead to inaction or counterproductive action. It is also likely to cause severe emotional distress.

Appendix: The Mental Models approach to risk communication: a step by step guide

A popular approach to risk communication is the Mental Models approach, detailed in Morgan et. al.’s 2001 book Risk Communication: A Mental Models Approach. This approach centers the existing beliefs and understandings of audiences in the communication process and proposes a five-step process for designing communication interventions.

  1. Create an Expert Model: Review current scientific knowledge about the risk that is being adressed, summarizing it from a practical standpoint for external review. Use an influence diagram to represent and interpret expert knowledge from various disciplines. This model is then reviewed by technical experts to ensure balance and credibility.

  2. Conduct Mental Models Interviews (with the target audience): Carry out open-ended interviews to elicit people’s beliefs about the hazard in their own words. The interview protocol, shaped by the influence diagram, ensures coverage of relevant topics and clarity of respondents’ intentions. Analyze responses to compare these mental models with the expert model.

  3. Conduct Structured Initial Interviews: Develop a confirmatory questionnaire based on beliefs from the open-ended interviews and the expert model. Administer it to a larger, representative sample to estimate the prevalence of these beliefs within the population.

  4. Draft Risk Communication: Use interview and questionnaire results, along with decision analysis, to identify incorrect beliefs and knowledge gaps. Draft a communication piece and have it reviewed by experts to ensure accuracy.

  5. Evaluate Communication: Test and refine the communication with individuals from the target population through read-aloud interviews, focus groups, questionnaires, or problem-solving tasks. Repeat until the communication is clearly understood.

This is a very research-focused iterative approach. By focusing on how people think and what they already know, the mental models approach helps to bridge the gap between expert knowledge and public understanding, leading to more informed decision-making and better management of risks. This type of approach can be particularly useful for cause areas like AI risks where understanding of and meaning given to the risk of hand can be wildly divergent between experts and other audiences, as well as withing audience segments. Morgan et. al. elaborate on each step of the process in great detail in their book.

Appendix: A new model for policy communications by Aidan Muller

In a two part series of posts on Cast From Clay, Aidan Muller puts forth a new model for policy communications.

The first posts argues that think tanks (and other policy actors) need to move beyond traditional report-based communications. It emphasizes shifting towards storytelling to engage policymakers, who are often too busy to read lengthy reports. The article suggests adopting a new communication model inspired by marketing tactics, focusing on concise, impactful content that builds a narrative to guide policymakers from awareness to action.

The second post expands on how to implement this new communication model. It likens the process to a “Policy Adoption Funnel,” using short-form content like tweets and videos to capture attention, and more detailed content later to solidify support. It advocates for “transmedia storytelling,” where each piece of content is a self-contained part of a larger narrative, enhancing engagement across multiple platforms.

This model has great potential for existential risk communications because a significant amount of work in this space involves policy actors and recommendations. I suggest any actors working in these spaces read the posts here (Part I) and here (Part II). While there isn’t much elaboration offered in the posts in regards to the details of concrete steps and practices, the guiding principles should be able to open up a useful avenue.

Sources:

Note: Due to extreme overlap in content between many of the sources, I have not individually cited each source where relevant within the text in the interest of readability.

Arkin, E. B. (1989). Translation of Risk Information for the Public: Message Development. In V. T. Covello, D. B. McCallum, & M. T. Pavlova (Eds.), Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations (pp. 127–135). Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8_19

Baram, M. (1989). Risk Communication: Moving from Theory to Law to Practice. In V. T. Covello, D. B. McCallum, & M. T. Pavlova (Eds.), Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations (pp. 37–43). Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8_5

Bowen, S. A. (2010). Ethical Responsibility and Guidelines for Managing Issues of Risk and Risk Communication. In Handbook of Risk and Crisis Communication. Routledge.

Bruin, L. A. F.-M., Wändi Bruine de. (2013). The ‘Mental Models’ methodology for developing communications: Adaptations for informing public risk management decisions about emerging technologies. In Effective Risk Communication. Routledge.

Callaghan, J. D. (1989). Reaching Target Audiences with Risk Information. In V. T. Covello, D. B. McCallum, & M. T. Pavlova (Eds.), Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations (pp. 137–142). Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8_20

Campbell-Árvai, J. Á., Victoria. (2013). Risk communication: Insights from the decision sciences. In Effective Risk Communication. Routledge.

Candlin, C. N., Crichton, J., & Firkins, A. S. (2016). Crucial Sites and Research Orientations: Exploring the Communication of Risk. In J. Crichton, C. N. Candlin, & A. S. Firkins (Eds.), Communicating Risk (pp. 1–14). Palgrave Macmillan UK. https://​​doi.org/​​10.1057/​​9781137478788_1

Corvellec, Å. B., Hervé. (2013). A relational theory of risk: Lessons for risk communication. In Effective Risk Communication. Routledge.

Covello, V. T. (1991). Risk comparisons and risk communication: Issues and problems in comparing health and environmental risks. In R. E. Kasperson & P. J. M. Stallen (Eds.), Communicating Risks to the Public: International Perspectives (pp. 79–124). Springer Netherlands. https://​​doi.org/​​10.1007/​​978-94-009-1952-5_6

Covello, V. T. (2010). Strategies for Overcoming Challenges to Effective Risk Communication. In Handbook of Risk and Crisis Communication. Routledge.

Covello, V. T., McCallum, D. B., & Pavlova, M. (1989a). Principles and Guidelines for Improving Risk Communication. In V. T. Covello, D. B. McCallum, & M. T. Pavlova (Eds.), Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations (pp. 3–16). Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8_1

Covello, V. T., McCallum, D. B., & Pavlova, M. T. (Eds.). (1989b). Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations. Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8

Covello, V. T., Sandman, P. M., & Slovic, P. (1988). Risk communication, risk statistics, and risk comparisons: A manual for plant managers. Chemical Manufacturers Association Washington, DC.

Das, S., Mare, S., & Camp, L. J. (2020). Smart Storytelling: Video and Text Risk Communication to Increase MFA Acceptability. 2020 IEEE 6th International Conference on Collaboration and Internet Computing (CIC), 153–160. https://​​doi.org/​​10.1109/​​CIC50333.2020.00027

Dieckmann, R. G., Nate. (2013). Communicating about uncertainty in multistakeholder groups. In Effective Risk Communication. Routledge.

Downs, J. S. (2013). Video interventions for risk communication and decision-making. In Effective Risk Communication. Routledge.

Ellis, S. M., Jaye. (2010). The Precautionary Principle and Risk Communication. In Handbook of Risk and Crisis Communication. Routledge.

Eppler, M. J., & Aeschimann, M. (2009). A systematic framework for risk visualization in risk management and communication. Risk Management, 11(2), 67–89. https://​​doi.org/​​10.1057/​​rm.2009.4

Fahlquist, S. R., Jessica Nihlén. (2013). Risk communication and moral emotions. In Effective Risk Communication. Routledge.

Galloway, D. M., Christopher. (2010). Warming Warnings: Global Challenges of Risk and Crisis Communication. In Handbook of Risk and Crisis Communication. Routledge.

Garg, V., Camp, L. J., Connelly, K., & Lorenzen-Huber, L. (2012). Risk Communication Design: Video vs. Text. In S. Fischer-Hübner & M. Wright (Eds.), Privacy Enhancing Technologies (pp. 279–298). Springer. https://​​doi.org/​​10.1007/​​978-3-642-31680-7_15

Hanna, P., Vanclay, F., & Arts, J. (2016). The Communication and Management of Social Risks and Their Relevance to Corporate-Community Relationships. In J. Crichton, C. N. Candlin, & A. S. Firkins (Eds.), Communicating Risk (pp. 171–188). Palgrave Macmillan UK. https://​​doi.org/​​10.1057/​​9781137478788_11

Hart, P. S. (2013). Boomerang effects in risk communication. In Effective Risk Communication. Routledge.

Henwood, K. L., & Pidgeon, N. (2016). Interpretive Environmental Risk Research: Affect, Discourses and Change. In J. Crichton, C. N. Candlin, & A. S. Firkins (Eds.), Communicating Risk (pp. 155–170). Palgrave Macmillan UK. https://​​doi.org/​​10.1057/​​9781137478788_10

Kasperson, R. E., & Palmlund, I. (1989). Evaluating Risk Communication. In V. T. Covello, D. B. McCallum, & M. T. Pavlova (Eds.), Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations (pp. 143–158). Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8_21

Lichtenberg, J., & MacLean, D. (1991). The role of the media in risk communication. In R. E. Kasperson & P. J. M. Stallen (Eds.), Communicating Risks to the Public: International Perspectives (pp. 157–173). Springer Netherlands. https://​​doi.org/​​10.1007/​​978-94-009-1952-5_9

Lundgren, R. E., & McMakin, A. H. (2018). Risk Communication: A Handbook for Communicating Environmental, Safety, and Health Risks. John Wiley & Sons.

Morgan, M. G., Fischhoff, B., Bostrom, A., & Atman, C. J. (2002). Risk Communication: A Mental Models Approach. Cambridge University Press.

Muller, A. (2018a, August 20). A new model for policy communications (Part I). Cast from Clay. https://​​castfromclay.co.uk/​​insights/​​a-new-model-for-think-tank-communications-part-i/​​

Muller, A. (2018b, October 1). A new model for policy communications (Part II). Cast from Clay. https://​​castfromclay.co.uk/​​insights/​​a-new-model-for-think-tank-communications-part-ii/​​

National Research Council. (1989). Improving Risk Communication. National Academies Press. https://​​doi.org/​​10.17226/​​1189

Palmlund, I. (2010). Risk and Social Dramaturgy. In Handbook of Risk and Crisis Communication. Routledge.

Renn, O. (1991). Risk communication and the social amplification of risk. In R. E. Kasperson & P. J. M. Stallen (Eds.), Communicating Risks to the Public: International Perspectives (pp. 287–324). Springer Netherlands. https://​​doi.org/​​10.1007/​​978-94-009-1952-5_14

Renn, O. (2010). Risk Communication: Insights and Requirements for Designing Successful Communication Programs on Health and Environmental Hazards. In Handbook of Risk and Crisis Communication. Routledge.

Renn, O., & Levine, D. (1991). Credibility and trust in risk communication. In R. E. Kasperson & P. J. M. Stallen (Eds.), Communicating Risks to the Public: International Perspectives (pp. 175–217). Springer Netherlands. https://​​doi.org/​​10.1007/​​978-94-009-1952-5_10

Roeser, S., Hillerbrand, R., Sandin, P., & Peterson, M. (2013). Introduction to Risk Theory. In S. Roeser, R. Hillerbrand, P. Sandin, & M. Peterson (Eds.), Essentials of Risk Theory (pp. 1–27). Springer Netherlands. https://​​doi.org/​​10.1007/​​978-94-007-5455-3_1

Sandman, P. M. (1989). Hazard versus Outrage in the Public Perception of Risk. In V. T. Covello, D. B. McCallum, & M. T. Pavlova (Eds.), Effective Risk Communication: The Role and Responsibility of Government and Nongovernment Organizations (pp. 45–49). Springer US. https://​​doi.org/​​10.1007/​​978-1-4613-1569-8_6

Schütz, P. M. W., Franziska U. Boerner, Holger. (2013). Communicating inconclusive scientific evidence. In Effective Risk Communication. Routledge.

Spitzberg, P. A. A., Brian H. (2010). Myths and Maxims of Risk and Crisis Communication. In Handbook of Risk and Crisis Communication. Routledge.

Van Eijndhoven, J. (1991). Risk communication: The need for a broader perspective. In R. E. Kasperson & P. J. M. Stallen (Eds.), Communicating Risks to the Public: International Perspectives (pp. 393–412). Springer Netherlands. https://​​doi.org/​​10.1007/​​978-94-009-1952-5_18

Visschers, V. H. M., Meertens, R. M., Passchier, W. W. F., & De Vries, N. N. K. (2009). Probability Information in Risk Communication: A Review of the Research Literature. Risk Analysis, 29(2), 267–287. https://​​doi.org/​​10.1111/​​j.1539-6924.2008.01137.x

Wilson, A. Z., Robyn S. (2013). Construing risk: Implications for risk communication. In Effective Risk Communication. Routledge.

Yang, F. (2020). Data Visualization for Health and Risk Communication. In The Handbook of Applied Communication Research (pp. 213–232). John Wiley & Sons, Ltd. https://​​doi.org/​​10.1002/​​9781119399926.ch13

Zipkin, D. A., Umscheid, C. A., Keating, N. L., Allen, E., Aung, K., Beyth, R., Kaatz, S., Mann, D. M., Sussman, J. B., Korenstein, D., Schardt, C., Nagi, A., Sloane, R., & Feldstein, D. A. (2014). Evidence-based risk communication: A systematic review. Annals of Internal Medicine, 161(4), 270–280. https://​​doi.org/​​10.7326/​​M14-0295

Zummo Forney, S., & Sadar, A. J. (2021). Environmental Risk Communication: Principles and Practices for Industry (2nd ed.). CRC Press. https://​​doi.org/​​10.1201/​​9781003083443