How to Communicate Uncertainty Without Losing Credibility

Abel Gustafson & Matthew Goldberg
March 2024

Uncertainties are inescapable, even in good decisions. A good medical treatment is beneficial for most patients—but never for all of them. A good investment is a wise bet but never a guarantee. A good scientific study could very likely be overturned in the future by new, better evidence. The world is an ocean of uncertainty, and we are forced to make decisions while swimming in it.

Because some degree of uncertainty is unavoidable in almost everything, organizations are faced with a decision: how should we handle uncertainties when communicating to our stakeholders? Should we project an air of confidence, or should we openly share all our caveats and unknowns?

Many leaders fear that openly discussing uncertainties will have harmful effects such as injecting doubt, distrust, and perceptions of incompetence. But a counter-argument is that being maximally transparent will actually increase trust—and that showing a strong grasp of all the uncertainties is actually a sign of greater competence. In this article, we summarize our research showing the specific types of situations in which communicating uncertainty tends to have those negative effects, as well as the specific situations in which the effects instead tend to be positive.

Not All Uncertainties Are Equal

Two reasons that communicating uncertainties can have either positive or negative effects are that 1) there are very different types of uncertainty, and 2) there are very different things you can communicate uncertainty about. We’ll tackle the first here, and the second in the next section.

There are typically four types of uncertainties that people express: known unknowns, unknown unknowns, quantified error, and conflict. Here, and in the table above, is a quick rundown of these four types. Known unknowns are what you know you don’t know. These are often things like unanswered questions, out-of-sample inferences, or even simple hedging (“maybe,” “could”). Second, unknown unknowns are what you don’t know you don’t know. This is the inescapable possibility that future evidence might change your current knowledge and the current evidence, even if you don’t currently know how or in what way. Third, a possibility range is a specified distribution or likelihood of potential outcomes—such as a probability of 0 < P < 100, a confidence interval around an estimate, or a forecast with a margin of error. Last, conflict is when there are contradictions within the evidence or disagreement among experts (e.g., some for, some against).

In a recent series of studies, our cofounder discovered that those negative effects of communicating uncertainty—decreasing trust, increasing skepticism, stalling action—are happening almost exclusively when the uncertainty is about that fourth type: conflict. That is, emphasizing controversy among experts or showing contradictions in the evidence (some for, some against) is the primary cause of all those negative effects we tend to be worried about. 

In contrast, across dozens of controlled experiments in peer-reviewed studies, the other three types of uncertainty usually do not have negative effects. Instead, they either have no effect at all, or sometimes they even have positive effects such as strengthening perceptions of trustworthiness, honesty, competence, and expertise. This shows that many types of uncertainty—specifically known unknowns, unknown unknowns, and possibility ranges—tend to be harmless to share, and potentially even beneficial.

Uncertainty About What?

Beyond the specific type of uncertainty, a second important factor that determines people’s responses is the topic or context. Generally, people’s tolerance for uncertainty varies widely depending on the object of uncertainty. For example, you would not be distressed if there is a 70% chance of rain, but you would be distressed if there is a 70% chance that your fiancé will say “I do.”

Some scientific studies have found this is even true with uncertainty stemming from conflict—typically the most damaging of uncertainties. One of these experiments (Jensen & Hurley, 2012) showed participants news articles with quotes from disagreeing experts with opposing perspectives on an issue. Those news articles varied in what issue was being discussed: either toxic sludge contamination or an effort to revitalize the wolf population. So, one group of participants saw conflict among experts about toxic sludge, while the other group saw conflict among experts about wolf repopulation. The study found that seeing disagreement about wolf repopulation did not reduce credibility perceptions of scientists, but seeing disagreement about toxic sludge did. 

Another experiment (Gustafson & Rice, 2019) found that portraying disagreement among scientists about topics like climate change and GMO foods had negative effects on perceived credibility of scientists. But there were no negative effects of the same scientific disagreement when it was on the topic of the harmful musculoskeletal effects of operating heavy machinery—which is a relatively obscure and irrelevant topic for most people. And in all three of these topics, there were no negative effects of known unknowns, unknown unknowns, or possibility ranges.

A good explanation for these differences in effects across topics is that people will be much more likely to demand certainty (and be very uncomfortable without it) when the stakes are high, the risks are serious, and the issue feels personal or immediate. Conversely, people are often quite comfortable with some uncertainty—and many people even expect it—when the situation’s potential consequences are not serious, immediate, or personal.

So What

We can feel confident in communicating uncertainty to our stakeholders when two criteria are satisfied: the uncertainty is not due to disagreement or contradiction, and the topic does not have significant personal stakes. In such situations, being transparent about uncertainties—such as gaps in knowledge, unanswered questions, humility about current findings, probabilities, and margin of error—can all have positive effects that strengthen credibility.

And certainly, we can expect negative effects when the uncertainty is about conflict and does have significant consequences that are personally relevant. In these situations, people crave certainty so they can have a clear view of the problem and a clear path toward solutions.

Outcomes become less predictable when only one criterion is satisfied, such as when there is a serious threat but the uncertainties are in the form of known unknowns or possibility ranges. For example, when a hurricane approaches land, people who live in its path will understandably want clear, confident information about the danger. It is undoubtedly very helpful for scientists and government officials to all present a consistent, unified set of information and recommendations (i.e., no conflict).

But simply due to the unpredictable nature of hurricanes, it is vital to still communicate the many possible paths of the hurricane (often visualized as the “cone of uncertainty”; left). Much of the public may appreciate this use of a “possibility range,” because it illustrates what areas might be hit, and therefore who should prepare or evacuate. Even though most of that cone will not experience a direct hit, communicating that uncertainty range is still very helpful to the entire group.

In sum, uncertainties are inevitable, and as leaders and strategists we often face decisions about if and how to share them with key stakeholders and audiences. The best current research shows that we should not avoid sharing uncertainties altogether, because some specific types in specific situations can have important positive effects. But of course, this advice could always be changed  by future research. There’s uncertainty about uncertainty, too.

cite this

Gustafson, A. & Goldberg, M.H. (2024). How to Communicate Uncertainties Without Losing Credibility. XandY. New Haven, CT. Retrieved from: https://www.xandyanalytics.com/how-to-communicate-uncertainties