AN OPEN LETTER TO THE IPBES ON THE VISUAL DISPLAY OF INFORMATION

PDF Version – Open Letter to the IPBES

An output of the Tansley Working Group on ‘Information Visualisation for Science and Policy’. Contact: Greg McInerny – gregory.mcinerny@cs.ox.ac.uk

About the ‘ecoViz’ Tansley working group: This NERC funded working group, sponsored by Future Earth, has held two short workshops (25-26/11/2013 and 18-20/03/2015) to explore a multi-disciplinary view of information visualisation in science-policy interfaces. The group involved a diverse set of expertise spanning scientists to designers, and from social scientists to communications experts, computer scientists and software developers. Our expertise spans academia (universities including Oxford, City (London), NYU, Imperial, UCL, York, Cambridge, UEA, Potsdam, Institute of Zoology, Dundee, Napier, Amsterdam), governmental agencies (e.g. CEFAS, CEH, ONS), NGOs (e.g. ICSU, IGBP, SRC, FututreEarth, IPCC, UNEP-WCMC, UNDP), journals (BMJ), as well as SMEs (Periscopic, CartoDB, Vizzuality) and freelancers (e.g. http://truth-and-beauty.net/, http://pauladematos.co.uk/).

 For more information please get in contact & see the working group website https://ecoviz.wordpress.com/

We urge the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES, http://www.ipbes.net/) to place a far greater importance of the visual display of information. The IPBES goals require the best possible communications to support understanding and engagement across diverse stakeholders and end-users audiences.

Graphics and visualisations should then be purposefully developed to support the wide variety of IPBES goals and deliverables (e.g. assessments, capacity building activities, policy support tools and  communication strategies) in terms of the diverse role of ‘visuals’ play in supporting IPBES users and participants – from research communities and policy makers, to citizens and the media (McInerny et al. 2014).

Many graphical representations of the same data and information are possible and the impact and utility of these variations will differ. Just like models and statistics, graphics are a compromise amongst a host of trade-offs. The most objective, rigorous, engaging and effective graphics and visualisations will only be developed by addressing the challenges of their production and by engaging appropriate expertise (McInerny et al. 2014).

In addition to improving the efficiency or effectiveness of visualisations and graphics, IPBES may consider how a new visual approach can improve economy in its processes, its outreach and its impact, whilst soliciting new forms of engagement and feedback.

We suggest that the first priority could be to address graphics and visualisations within the “Guide on the production and integration of assessments from and across all scales”. Establishing this guidance will be a critical step for IPBES in facilitating support for authors in the development of high quality assessments and reports, and enriching strategies for communications and stakeholder engagement.

Specific guidelines have been developed for similarly pervasive topics such as ‘uncertainty’ (Mastrandrea et al. 2010, IPBES guidelines). However, guides for graphics and visualisations have not previously been developed for the IPCC (Intergovernmental Platform for Climate Change) or MEA (Millennium Ecosystem Assessment), despite playing a central role in those processes and reports. Graphics and visualisations are mentioned in IPCC communication strategies (http://bit.ly/1KWrZL2), and are a focus for the ‘future work’ (http://go.nature.com/MCYYHn, http://go.nature.com/KL5f34).

However, without real guidelines and support approaches are more likely to be improvised rather than guided by an informed approach based on expertise and experience. The IPBES is presented with a unique and important opportunity to seize upon visualisation expertise, techniques and technologies, and take control of its graphics and visualisations

We note that graphics and visualisation satisfy all the criteria of a ‘priority capacity building need’ (chapter 13, IPBES 2015). For instance, graphics and visualization can be integrated into all deliverables (criteria a); can enable the work of the platform (criteria b); can build on existing initiatives and institutions (criteria c); and have a critical role in stimulating awareness of the platform (criteria d).  In addition, graphics and visualisation support and cross-cut between multiple “capacity need categories” (chapter 13, IPBES 2015), such as Developing the capacity to carry out and use assessments (category 2), Improving accessibility to data, information and knowledge (category 4) and Developing enhanced and meaningful multi-stakeholder engagement (category 5).

Our initial comments (derived from two workshops, see details above) represent the first step towards making graphics and visualisations work for IPBES. In the following we provide:

  • specific comments on the existing guidelines for assessments
  • initial proposals to the secretariat.
  • example guidelines for authors.
  • selected key resources.

SPECIFIC COMMENTS ON THE CURRENT GUIDELINES FOR ASSESSMENTS

Currently the guidelines for producing charts, graphs and visualisations are limited, e.g.:

  • Producing visual aids such as charts, graphs and pictures to easily communicate messages within the text. Use of these supporting visuals may also increase the chance of greater media coverage (UNEP, 2007)” [IPBES/3/INF/4, lines 1-3, p136]
  • Uncertainty can also be presented in graphical form using radar plots or snowflake charts that signify increasing confidence as it increases in size. See Moss & Schneider (2000) for further discussions on graphical approaches to communicating uncertainty.” [IPBES/3/INF/4, lines 3-5, p79]

These guidelines do not encompass the breadth of knowledge and best practices for the production and use of graphics and visualisations. For instance:

  • Without appropriate attention and sensitivity to design, not all ‘visual aids’ are effective in communicating a message or finding, nor do they necessarily increase the chance of engagement or greater media coverage. Different designs of a graphic can alter its efficiency for different tasks, and affect other traits such as perceived reputation and memorability. Moreover, graphics can communicate messages that text cannot do so easily, and should not be considered solely as a device to support text. Instead, they may be used to lead discourse.
  • Radar plots and snowflake charts are one of very many means to depict uncertainty. The radar plot may not necessarily have any benefits over other visual encodings, except where it saves page space or injects diversity into the types of charts and graphs used in a report. Moreover, radar plots can be prone to a variety of perceptual biases associated [E.g. Fuchs et al. 2014]. More broadly, the addition of uncertainty to a graphic involves increasing its complexity and authors may need some understanding of how users will perceive that extra information [e.g. Correll et al. 2014], when the inclusion of uncertainty may require a different design for a graphic, and how attuned that approach will be to the audiences knowledge and skills.

These comments are only a small example of the topics that the ‘guidance to assessments’ could expand on. A significant body of scientific literature and expertise exists that can inform this process and should be consulted and involved in developing content on graphics and visualization to appropriately support authors and IPBES. Some initial guidance to authors is given below.

INITIAL PROPOSALS FOR THE SECRETARIAT

We put forward an initial set of considerations that could inform the approaches taken by IPBES in relation to graphics and visualisations. This guidance derives from discussions at our two initial workshops (see above, https://ecoviz.wordpress.com/) and the paper by McInerny et al. (2014).

  • Set a new standard for the visual display of information within science-policy. In addition to supporting the core goals of the IPBES and its deliverables, this new standard could also influence the development of policy support tools that IPBES consumes and also influence existing and future science-policy interfaces.
  • Develop a chapter on the production of visuals and graphics to the ‘guidance for assessments’. The topic of graphics and visualisation cross-cuts many activities and documents, supporting the case for a chapter focused on this topic. A specific chapter would avoid fragmenting or repeating information across multiple documents and chapters, whilst allowing an appropriate selection of expertise to develop the material.
  • Link this guidance on graphics and visualisation to content across the IPBES deliverables. Many IPBES deliverables will benefit from this fresh approach to graphics and visualisation, especially given its crucial role in joining deliverables such as (but not exclusively): Deliverable 4(b): Development of an information and data management plan and Deliverable 4(d): Set of communication, outreach and engagement strategies, products and processes. Graphics and visualisation should be included during the scoping processes.
  • Establish appropriate processes for reviewing visualisations. Reviews can provide crucial feedback throughout the whole production process. Direct invitations for review comments on graphics and visualisations should be possible at each review stage, and be solicited in ways that engage relevant communities. Reviews may review a template designed for this task.
  • Consider what evaluation procedures would assist the IPBES goals. Road-testing, evaluation and assessment of impact of graphics on stakeholders will ensure IPBES receives appropriate feedback on these activities and products..
  • Develop capacity and processes to appropriately engage and support graphics and visualisation expertise. Specialists related to graphics and visualisation specialists may play in a number of roles within IPBES; from contributing authors, reviewers, task force members, consultants and within technical support units. IPBES should develop the ability to engage expertise from different domains of visualisation, graphic design, information design and cartography for appropriate tasks. Relevant expertise may be found across academic, business and freelancer settings and may also be found in small pockets of existing science-policy interfaces, such as within the IPCC technical support unit of WG2.
  • Develop strategy and processes for engaging graphics and visualisation expertise. This strategy and understanding of how graphics and visualisation can work with IPBES will ensure that IPBES can act with authority on all the issues involved. For example, by understanding the requirements and timelines involved in enlisting expertise, and by understanding the differences in outcomes associated with strategies of outsourcing ‘design’ to external expertise versus embedding design expertise, or relying on the ad-hoc practices of non-specialists.
  • Convene a workshop and expert group/taskforce to further develop IPBES capacity for graphics and visualisations. A workshop should be convened to develop and engage an expert group or task force, or to engage assessment authors to address the issues in this document, and to ensure that the IPBES maximises its capacity for handling graphics and visualisation. As assessments are underway this should be a priority in order to support the ongoing work of IPBES authors.
  • Consider how training may support IPBES participants and a design ethos. Scientists rarely receive training in visualisation and graphics. Training should support individuals in developing the appropriate skills and knowledge. Training could also cover topics such as co-design, participatory design, and user-centred design which may require expertise in UX design, facilitation, ethnography, design and the social sciences.
  • Recognise that, like science, the domains of graphics and visualisation are not entirely resolved subjects and that the context of science-policy is new. As the IPBES strikes new ground in science-policy it will also explore a host of new issues in the use and impact of graphics and visualisation. Opportunities to learn from the IPBES experience may be taken to encourage research in this area.

EXAMPLE GUIDANCE TO AUTHORS 

Just as with the composition of text within IPBES reports and outputs, graphics should use appropriate framing, visual grammar and terminology to ensure that the standards of the IPBES are met. A range of visual products can be produced (e.g. visual displays of information may include graphs and charts, infographics, visualisations, maps and interactive graphics and require different expertise) and may require different expertise and approaches to their development. We put forward a number of key considerations for lead authors and contributing authors. These guidelines originate from discussion sessions at our workshops (see above, https://ecoviz.wordpress.com/) and are a first step towards more comprehensive guidance. Some key resources are provided.

  • Establishing the objectives and audiences for graphics and visualisations should be the first consideration in their production. For instance, graphics developed for assessments and their SPMs (summary for policy-makers) may need to be rethought and revised for the policy-maker audience  rather than (re-)produced only as analytical/diagnostic graphics that may be more traditionally suited to scientific audiences. Developing an open rationale for graphics (who the audience is, what the objectives are) will encourage the most productive review processes and helpful inputs from experts.
  • A wide range of stakeholders and users may be engaged with graphics and vary in their scientific, visual and statistical literacy, numeracy and understanding of uncertainty. The target audiences should be considered carefully when selecting the narrative of a graphic and the techniques used for depicting information. For instance, whilst scientists may be familiar with techniques to depict uncertainty or statistical differences, these may not always be understood by a broader audience. In some cases, the production of a figure for non-specialists may require greater editorial consideration and input from communications experts.
  • Lead authors and co-chairs should oversee the use of all images in chapters and assessments. This will be a key step in achieving consistency and quality within and between assessments and across all IPBES outputs.
  • Graphics specialists should be considered in technical support units and as contributing authors. The IPCC WG2 recognised some of these issues and successfully incorporated graphics expertise into its Technical Support Unit. In addition, we suggest that graphics and visualisation specialists could contribute as contributing authors.
  • Different graphics suit different data characteristics. It should not be assumed that one type of graphic will highlight the same messages or characteristics for all data sources. There are predictable perceptual and cognitive biases associated with the use of particular types of charts that depend on the type and characteristics of data being displayed. For instance, different graphs may differ in the effectiveness of displaying data with different properties (e.g. strength and direction of correlation) and more fundamentally, the effectiveness of particular visual variables (such as position, length and area, colour, transparency and shape) depends on different data types (Bertin 1967; Mackinlay1986; Maceachren1995; Harrison et al. 2014). 
  • When charts and graphs are reused from the scientific literature their suitability must be considered. Graphics from the primary literature may be subject to a wide variety of individualistic design choices and may not have been designed for this particular use. Graphics may require a redesign to suit the new context of use, and to fit within conventions used within a report (e.g. audience, framing) and the methods of production.
  • Ensure that the encoding of information in visualisations and graphics has a level of consistency (e.g. scales, colours). Consistency reduces the burden on users of the assessments. Inconsistency relies on users learning the grammar of each individual chart, graph or map, and increases the potential for confusion where similar information are depicted differently and different information are depicted similarly. Where visual encodings have multiple meanings (e.g. error bars) care is needed to ensure the specific use is communicated.
  • Visualisations and graphs should be designed as an independent unit such that users do not have to locate and refer to the text to understand its contents. To be independent of the text, a graphic may require greater consideration of the framing of the narrative; clearer explanations of visual encodings; or more details on source materials, data types and/or methods involved in deriving that information.
  • Producing visualisations in an independent format will facilitate and encourage their use and re-use. In addition to the above guidance, it is important to consider that visuals may be reused by the media (e.g. newspapers, tv, blogs), the public (e.g. social media), other assessments (e.g. cross-referencing) and throughout the IPBES itself. Enabling these different kinds of reuse presents a variety of opportunities to evaluate the different visual strategies and examine the reach/impact of IPBES.
  • Data should be made available for graphics. Making data available in advance of or during report writing will encourage openly designed graphics. Linking graphics to data resources will also encourage the most transparent interpretation and effective re-use of graphics after reports are published. Examples of policies best practice may be found in the scientific literature (e.g. Nature, PLoS) and organisations such as the Office for National Statistics, UK (ONS) and the OECD (Organisation for Economic Co-operation and Development) where visualisations are linked to data platforms using digital identifiers. As part of an open data strategy, these steps will encourage traceable reuse of IPBES outputs (e.g. the media).
  • Effective graphics are underpinned by clear references to scales, indication of geographic projections and baseline measures, definition of terms and clearly established context. Clarity in all these facets of a graphic will reduce the potential for misinterpretation and produce salient graphics. In cases where the data is ‘complex’ or ‘complicated’ it is important to ensure that the design suits both the data, the purpose of the graphic and the audience.
  • Authors should consider if a particular design for a graphic allows uncertainty to be depicted efficiently. Adding uncertainty to a graphic may alter its effectiveness and authors should be open to reconsidering how effective a graphic is with this extra layer of information. Visual encodings (such as blurring, more complicated glyphs and error bars) may affect the interpretation of all information in a graphic, and not just the uncertainty (McInerny et al. 2014). Care should be taken where visual encodings indicate uncertainty but obscure the actual range of the data. Supplementary annotations or alternative framing of the graphic may assist in developing a successful graphic for complex topics.
  • Consider whether a map is the best depiction of geospatial information. Geo-spatial data is likely to feature heavily in the assessments. However, maps may not always be the most effective method for displaying that data and information. Authors should consider if readers can draw the same conclusions as experts when viewing maps side by side, or are able to assess areas in maps (e.g. areas associated with projections at different time steps, or for different scenarios) when the map is composed of regions of complex shapes. In many cases, supplementary charts and graphs may be needed to support maps.
  • Authors should also consider what knowledge the interpretation of a map or graphic requires. The narrative surrounding the use of a map may require detailed knowledge of other factors; such as co-variation in the changes of conditions within projections or methodological nuances such as the difference between relative likelihoods versus actual probabilities.
  • Different graphics may be produced more effectively in different technologies. A wide variety of software are available for producing graphics and visualisations. Technologies differ in how easily and effectively different kinds of graphics can be produced. Selecting appropriate technologies should consider how easily graphics can be modified or updated following review.
  • The scope of expertise required should include user experience design. Graphics will be hosted in a wide range of media (from reports, to interactive graphics, web portals and decision support systems) and expertise in user experience (UX-design) should be involved to ensure that these outputs and end-uses are fit for purpose.

SELECTED KEY RESOURCES 

This list is not meant to be exhaustive. Instead it highlights some key resources that can provide context to the issues of graphics and visualisations, and also assist in the production of ‘visuals’.

REVIEW AND ACADEMIC PAPERS

WEB RESOURCES

BLOG POSTS

BOOKS

References

  • Bertin J. (1967) La sémiologie graphique. Paris: Gauthier-Villars.
  • Correll, M., & Gleicher, M. (2014). Error Bars Considered Harmful: Exploring Alternate Encodings for Mean and Error. IEEE TVCG (Proc. InfoVis). 20(12), 2142–2151.
  • Fuchs, J., Isenberg, P., Bezerianos, A., Fischer, F., and Bertini, E. (2014) The Influence of Contour on Similarity Perception of Star Glyphs. IEEE TVCG, 20(12):2251–2260.
  • Harrison, L. Yang, F. Franconeri, S. & Chang, R. (2014). Ranking visualizations of correlation using weber’s law. IEEE TVCG (Proc. InfoVis). 20(12): 1943-1952.
  • IPBES, 2015. Guide on the production and integration of assessments from and across all scales (IPBES/3/INF/4).
  • Maceachren, A. M. (1995) How maps work, New York, NY, USA, The Guilford Press.
  • Mackinlay, J. (1986) Automating the Design of Graphical Presentations of Relational Information, ACM Transactions on Graphics, Vol. 5 ,No.2 pp 110 – 141.
  • Mastrandrea, M.D. et al. (2010) Guidance Note for Lead Authors of the IPCC Fifth Assessment Report on Consistent Treatment of Uncertainties, IPCC.
  • McInerny, GJ. Chen, M. Freeman, R. Gavaghan, D. Meyer, M. Rowland, F. Spiegelhalter, D. Stefaner, M. Tessarolo, G. & Hortal, J. (2014) Information visualization in science and policy – engaging users & communicating bias. Trends in Ecology & Evolution. 29. 148-157. DOI : 10.1016/j.tree.2014.01.003
Advertisements