Prepared by Ilham Akhsanu Ridlo for SciCOPH (Science Communication for Planetary Health) Workshop Faculty of Public Health, Surabaya Indonesia 7-8 November 2024

Versi bahasa Indonesia disini Mengomunikasikan Ketidakpastian

Introduction

Uncertainty is an intrinsic characteristic of scientific inquiry, reflecting its iterative nature in refining our understanding while acknowledging the limitations of absolute certainty. Effective communication of uncertainty ensures that audiences grasp the nuances of scientific conclusions without losing trust. This guide provides a structured approach to communicating uncertainty clearly and credibly. By embracing and transparently communicating uncertainty, we can strengthen public trust in science and support informed decision-making.

Understanding Types of Uncertainty

Uncertainty in scientific communication can be broadly classified into two main types: aleatory and epistemic uncertainty. Aleatory uncertainty refers to inherent randomness, while epistemic uncertainty arises from limitations in knowledge. Each type requires a distinct communicative approach, and understanding the difference helps convey uncertainty effectively.

1. Aleatory Uncertainty: Aleatory uncertainty refers to the inherent randomness or stochastic variability present within a system or process. This form of uncertainty often arises from unpredictable factors and natural variations. Examples include fluctuations in meteorological conditions, measurement variability, or biases due to sampling. Quantitative approaches are often employed to communicate aleatory uncertainty effectively, such as using confidence intervals, probability distributions, or ranges. These numerical representations convey the potential variability and help audiences understand the stochastic nature of the phenomenon. Visual representations, including violin plots and error bars, further enhance comprehension, especially for audiences less familiar with statistical terminology.
2. Epistemic Uncertainty: Epistemic uncertainty originates from limitations in our knowledge or understanding. It encompasses uncertainties arising from incomplete data, model imperfections, and differing expert judgments. This type of uncertainty is more challenging to quantify, as it is inherently related to gaps in knowledge. Nevertheless, transparency regarding epistemic uncertainty is crucial for fostering public trust. For example, when experts hold differing views on a model's predictions, communicating these disagreements openly allows audiences to recognize the boundaries of scientific knowledge. Such openness encourages the public to view science as an evolving process that adapts as new evidence emerges.

The work by van der Bles et al. (2019) offers a structured approach to addressing epistemic uncertainty, categorizing uncertainty into three key objects: facts, numbers, and science. Moreover, they distinguish between direct and indirect uncertainty—direct uncertainty pertains to specific events or values, whereas indirect uncertainty relates to the quality of evidence and the reliability of underlying data. Grasping these distinctions helps communicators effectively tailor their messages to convey uncertainty, thereby avoiding ambiguity and enhancing the audience's understanding.

Techniques for Communicating Uncertainty

Effectively communicating uncertainty requires the use of various strategies that cater to the audience's needs and the specific context. Below, we present practical techniques that can facilitate this communication:

• Quantify When Possible: When feasible, communicate uncertainty with numerical values such as means, standard deviations, or confidence intervals. Quantitative representations lend concreteness to the message, but even approximate ranges can be more informative than vague descriptions. For example, instead of stating, "There is a possibility of rain tomorrow," specifying, "There is a 40-60% chance of rain tomorrow," provides greater precision, helping audiences to understand the extent of the uncertainty.

• Visual Aids: Visual aids are invaluable tools for communicating uncertainty, particularly for those who may struggle with numerical literacy. Graphical elements such as error bars, violin plots, and probability distributions can effectively illustrate variability in data. By translating complex statistical information into visual formats, graphs, and charts make uncertainty more comprehensible.

  Examples of Visual Aids:

  1. Violin Plot: This violin plot depicts the distribution of temperature measurements over a year for two cities. This type of visualization communicates aleatory uncertainty by concisely illustrating both the median and density of the data points.

     

     This violin plot illustrates the temperature distribution over a year for two cities. Here's an explanation:

     • The plot compares annual temperature patterns between City 1 and City 2.
     • Each "violin" shape represents the temperature distribution. Wider sections indicate more frequent temperatures, while narrower sections show less common ones.
     • City 1 (yellow) displays a wider temperature range, suggesting greater climate variability.
     • City 2 (orange) shows a narrower distribution, implying a more consistent temperature range.
     • The black bar in the center of each violin represents the interquartile range, with the white dot indicating the median temperature.
     • City 2 appears to have a higher median temperature and more frequent high temperatures than City 1.

     This visualization effectively communicates aleatory uncertainty, showcasing central tendencies and the full range of temperature variations for each city throughout the year.

  2. Error Bars in Bar Chart: A bar chart displaying average annual rainfall across three regions, with error bars indicating possible measurement variability. This visualization provides a quick understanding of the potential range of values.

  3. Probability Distribution Graph: A probability distribution graph predicting future sea level rise under a given climate scenario. This bell-shaped curve conveys the range of outcomes succinctly.

  4. Confidence Interval on Line Graph: A line graph showing projected global temperature increases over the next 50 years, with shaded areas representing the 95% confidence interval. This visualization effectively conveys uncertainty without overwhelming the audience.

  5. Comparative Bar Charts: Side-by-side bar charts comparing hospitalization rates for vaccinated and unvaccinated populations. Annotations that include uncertainty ranges help convey data variability in a straightforward way.

  Visual aids simplify complex concepts and make statistical information more accessible. They are particularly effective in enhancing comprehension for audiences that may be familiar with numerical data at varying levels.

• Use Clear Language: Avoiding jargon and overly technical terminology is vital in communicating uncertainty to non-specialist audiences. Instead, use accessible terms such as "likely," "probable," or "unlikely," but be mindful of potential variations in interpretation. For instance, the term "very likely" may suggest a high probability to scientists but could be interpreted differently by the general public. Accompanying these qualitative descriptors with numerical equivalents or concrete examples can bridge comprehension gaps.

• Explain Context and Implications: Providing context is crucial for understanding uncertainty. When presenting uncertain outcomes, place them within a broader context to ensure the audience understands their significance. For example, in discussing climate change projections, clarify that while uncertainties exist regarding the magnitude or timing of temperature increases, the general trend remains consistent. By emphasizing what is certain alongside what is uncertain, audiences are less likely to feel overwhelmed and more likely to grasp the overall message.

Building Trust Through Transparency

Establishing trust is a fundamental aspect of effective uncertainty communication. Transparency about knowledge limitations is central to building credibility and fostering trust between scientists and the public.

• Acknowledge Limits: Acknowledging the limitations of current knowledge helps build trust by demonstrating a commitment to openness. Sharing both the confident and less certain aspects of scientific knowledge presents a complete picture, fostering trust. For example, acknowledging unknowns about the long-term effects of a new treatment allows the audience to make more informed decisions. Transparency also mitigates the risk of backlash if subsequent findings diverge from initial claims. Over time, this approach contributes to building long-term credibility and solidifies the relationship between scientists and the public.
• Avoid False Precision: Overly precise communication can lead to misconceptions and damage credibility if new evidence contradicts earlier assertions. Instead, communicate estimates with appropriate ranges to avoid implying false accuracy. For example, rather than stating, "Sea levels will rise by 30 cm by 2050," it is more responsible to provide a range, such as "Sea levels are projected to rise between 20 and 40 cm by 2050," thus reflecting the inherent uncertainty and avoiding the impression of absolute certainty. Providing realistic estimates helps maintain credibility and reinforces trust, thereby fostering a stronger, more resilient relationship with the audience.

Case Study: Climate Change Communication