Hurricane Matthew strikes South Carolina (9 October 2016). NullSchool screenshot

Rewiring science-policy with visual information design

Co-authored with Marian Dörk, Jason Dykes, Greg McInerny, Denise Young*

DNA is imprinted in our collective consciousness as a twisting double helix not a string of base-pair letters. Compelling graphics are central to communicating scientific ideas and we are in the midst of a new visualisation revolution.

New technologies, ever more data and faster internet speeds are inspiring data analysts and visualizers to break free from the static plane of a chart. A new culture of design is feeding technological development as creativity challenges what can be delivered digitally. Interactive, multi-dimensional and multi-scale visual tools are opening up new ways to interpret and share information with global audiences.

Yet dry text remains the norm in science policy. Written reports, assessments and executive summaries continue to be churned out by organisations like the Intergovernmental Panel on Climate Change (IPCC), United Nations organisations, the World Economic Forum and the World Bank. Audiences are switching off — or worse, there is no audience. When the OECD’s Better Life Index embraced online and visualization technologies it reached over five million users between 2011 and 2015. A recent survey of policy stakeholders in developing countries indicated that databases and online resources were considered more useful than print publications. Policy briefs fared particularly poorly as an effective communications tool.

Scientists cannot afford to have their messages ignored. We argue researchers, and policymakers, adopt an evidence-based approach to knowledge and information design. This requires a fusion of expertise across science, policy, design, informatics and communications. And should build on scientific insights on, for example, how the brain processes visual information. Moreover, given the impact of visualisation, an essential component of societally relevant knowledge production of the type undertaken by Future Earth for example, is co-design and co-production with information designers — both academics and practitioners.

Shaping thinking

The media know readers want information delivered with clarity and brevity and invest heavily in data-analytic capacity. Data journalists work with graphic artists and programmers to make information tell a story, from tracking the spread of ebola to drone attacks.

Rich, engaging, rapidly interpretable graphics can reshape world views.

This message is hardly new. In 1786, the Scottish inventor of bar charts and line graphs William Playfair wrote, “Men of great rank, or active business, can only pay attention to general outlines…with the assistance of these charts, such information will be got, without the fatigue and trouble of studying the particulars of which it is composed.” In 1858, Florence Nightingale’s “rose” diagram paved the way for the nursing profession by exposing to the British government that preventable diseases like typhus killed ten times more troops than battle wounds. Bill Gates said a pie chart in a newspaper breaking down the major causes of death among children persuaded him to accelerate his philanthropic efforts in global health.

Effective tools and expertise to improve clarity and cut the distance between ignorance and understanding can’t be brushed aside.

Scientists are increasingly employing complex visuals in their professional publications. Astrophysicists spellbind with fly-through computer simulations of colliding galaxies. Recently, CALTECH theoretical physicist Kip Thorne teamed up with staff at special effects company Double Negative who worked on the Hollywood sci-fi movie Interstellar to show how a black hole warps the perception of distant stars to an observer close to the black hole. In 2014, geographers, anthropologists and visualisers demonstrated the cultural pull of urban centres such as Rome, London and Los Angeles over 2000 years by tracing the movements of 150,000 notable people through birth and death records.

Visualisation of light bending around a black hole. Kip Thorne et al Classical and Quantum Gravity, 2015. IOP Publishing

And the reach of subjects such as as meteorology are being reinvigorated through sophisticated online visual tools. Weather forecasts have been revolutionized by depicting wind fields, precipitation, temperature and other meteorological conditions. Such displays are often animated and increasingly scalable from local to global scales (Earth.nullschool, Ventusky).

Yet the bulk of climate policy remains wedded to traditional formats. While the IPCC employs graphic designers, its reports rely on conventional plots and charts negotiated through an intergovernmental process involving hundreds of climate scientists and bureaucrats. This process is destined to run into communications problems: IPCC summaries for policymakers attract breathtakingly low readability scores (Nature Climate Change Linguistic analysis of IPCC summaries for policymakers and associated coverage). Graphics should be developed with those who need to interpret them and in light of established best practice. Expert review of visual encodings, design, aesthetics and interaction is rarely used. Yet, the visualisations are central to the communications of the report. The investment in visualisation does not match the great societal significance of these images.

The grand challenge for international bodies and frameworks such as the IPCC, the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES) and the UN Sustainable Development Goals (SDGs) is to create fit-for-purpose systems to gather, synthesize and contextualize complex information from disparate sources and allow users to interact with them in ways that are meaningful, personal, and impactful. An expert review of IPCC’s fifth assessment report, conducted by academics and professionals in information design and data visualisation should be undertaken immediately, before the next report gets underway.

High impact? Climate scientists suggest this graph depicting four future carbon dioxide emissions pathways and their respective impact on global average temperature from the IPCC’s Fifth Assessment Report (Working Group I, 2013) is an example of effective communication to policymakers.

Design for impact

Visualisation researchers generate evidence about how to attain clarity, reach and impact. They explore questions such as how people respond to different colours; how to reduce bias in graphics; and how to communicate uncertainty effectively.

Well-designed data graphics can engage an audience, make complex data comprehensible, and help readers remember patterns in the data. However, some research findings on the effect of design decisions on levels of engagement, understanding, and recollection are rather surprising. For example, several studies suggest that visual embellishment and difficulty can actually increase the viewer’s engagement with a visualization and the comprehension of the represented data [Bateman, Hullman]. Such findings counter the principles of clean and minimalistic design as promoted by the influential statistician Edward Tufte, but stress that the experience and design of data visualization are complex. While there is robust evidence at the perceptual level, we need more research at the experiential level to understand how viewers can be supported to engage with visualization and make sense of the data.

Data display can obscure as well as reveal. Emphasising some facts and omitting others is a powerful way to unearth patterns in a dataset but may also advocate certain perspectives. For example, when charting the deficit in a government budget balance a focus on expenses favours budget cuts while an emphasis on tax revenues might suggest new sources of income.

Uncertainty is especially difficult to convey. To address the challenges of conveying uncertainty, the IPCC adopted simple language conventions for describing in text the likelihood of an event (certain, likely, unlikely), the amount of evidence (low, medium, high) and agreement amongst evidence (limited, medium, robust). But even within a single working group chapter or report summary uncertainty may be presented in diverse ways. Readers must then acclimatise themselves to the nuances of statistical terminology and the visual grammar within each figure. Graphical consistency and clear graphical conventions based upon empirically informed methods would help (MacEachern, Gschwandtner) .

Good information design is particularly important in areas of high societal concern such as climate science or stem-cell research, where uncertainty is significant, bias is a challenge and research is unfolding under the public spotlight. Visualisation can help cut through the confusion but, because of its persuasive power, it may also collide with special interests seeking to cast doubts and manipulate evidence.

In the public debate surrounding climate change, the “Hockey Stick” graph showing a recent uptick in the 1000-year northern hemisphere temperature record became a flashpoint. Its power as a communication tool drove skeptics to extraordinary lengths to discredit it, but to no avail.

More recently, skeptics have moved to focus on the perceived ‘hiatus’ in global temperature rise since 1998, which is based on a selective view of the data. Choosing 1998 — when a strong El Niño drove temperatures to a record high — as the start of a tracking period is biased. Besides, meaningful climate trends can only be discerned over several decades. Once in the media, misconceptions generated through memorable graphics are hard to exorcise, though the recent dramatic surge in global temperatures has quietened many of these voices.

New experimental platforms such as Global Forest Watch and the Global Carbon Atlas offer interactive monitoring information for governments, businesses, and NGOs and communities. While these initiatives still require a thorough evaluation to judge their success, such innovations suggest a growing appetite from scientists, funders and citizens to explore better interactive visualisation tools. Science needs to embrace a broad array of expertise to ensure impact, particularly relating to design, narrative and user experience. Indeed, user experience, or UX in the trade, is a foundation stone of companies like Google, Facebook and Apple who are in the business of creating products people want to use because of their usability, value and appeal. Many organisations find their brands are increasingly dependent upon the user experience — the worth of a brand may be determined by a simple heuristic: did the user retrieve useful information, which they can themselves use, efficiently. This attention to the user beyond academic colleagues is often entirely absent in science-policy communication.

What needs to change

The data visualisation community spans both academia and practice, narrowing the gap between scientific and public knowledge. This community is mathematically literate, technologically adept and design savvy. It bridges the physical sciences, neuroscience and psychology, computer science, communications and design. In an increasingly visual world, they are peerless translators of scientific knowledge.

Yet collaboration between the two remains the exception rather than the rule.

How should science engage with this visualization revolution? How can we encourage collaboration among these communities such that the visualization community is not just playing a role as translators of science, buttaking part in co-design and co-production of knowledge around some of humanity’s biggest challenges on global sustainability?

First, graphical expertise must be understood, valued and promoted in the scientific community beyond service provision and towards deeper intellectual engagement. The cartographer Mark Monmonier lists four “paramount communication skills of the educated person”: literacy, numeracy, articulacy and graphicacy (visual literacy). Undergraduate students should study graphical techniques as they learn about numbers. Doctoral students should receive training in visualization and research scientists should be supported and updated as information design research and technologies evolve.

Second, research institutions should create spaces where visualizers can come together with researchers to co-design and co-produce knowledge and new approaches to communicating science. Ideally this should include helping researchers to take advantage of the democratization of available visualization tools and technologies.

Third, these “hack” spaces should support user-centred and participatory approaches to design. Successful information design requires detailed understanding of users with producers and consumers engaged in the design process. Programmers, designers, user-interface experts, journalists and users should operate on a level playing field. This approach to co-production of knowledge will challenge outmoded power hierarchies in science where communicators provide a translation service after publication in peer-reviewed literature.

Finally, we need to nurture data labs to act as incubators to conceptualise, prototype, test and release novel analytic and visualization platforms around societal challenges, like Ed Hawkins’ recent spiral global temperature graphs. This can happen best in new, neutral spaces. For example MIT’s Media lab was set up in 1980 to explore how computers, publishing and broadcasting might converge. Similar places are now needed to bring together the right groups to develop solutions around themes such as urbanisation, food security, disease control, sanitation, energy and climate.

As Bill Gates has said, “If you show people the problems and you show them the solutions they will be moved to act.”


AUTHORS*:

Owen Gaffney, Future Earth, Royal Swedish Academy of Sciences, Stockholm, SE 104 05, Stockholm Resilience Centre, Stockholm University, Sweden.

Greg J McInerny Centre for Interdisciplinary Methodologies, Department of Social Sciences, University of Warwick, Coventry, United Kingdom

Denise Young, International Council For Science, 5 rue Auguste Vacquerie, 75116 Paris, France

Marian Dörk, Institute for Urban Futures, Potsdam University of Applied Sciences, 60 06 08, 14406 Germany

Jason Dykes, giCentre, Computer Science, City University London, EC1V OHB

*All authors contributed equally to this article.