Visualising health doesn’t have to be abstract. People tend to remember infographics that feature recognisable objects (Borkin et al, 2013; Stones & Gent, 2016 forthcoming). For instance Zigmund-Fisher et al. (2014) found that risk recall was significantly more accurate in icon arrays when more anthropomorphic icons were used, (such as restroom icons, outlines of heads, and detailed photographs) than with other abstract icon types such as blocks, ovals or simplified faces.
By using ‘real’ images we can also make the data more real for most of our audience (though there is still unease about using them with low-literate audiences).
Refs: Borkin, M. A., Vo, A. A., Bylinskii, Z., Isola, P., Sunkavalli, S., Oliva, A., & Pfister, H. (2013). What makes a visualization memorable?. Visualization and Computer Graphics, IEEE Transactions on, 19(12), 2306-2315.
Zikmund-Fisher, B. J., Witteman, H. O., Dickson, M., Fuhrel-Forbis, A., Kahn, V. C., Exe, N. L., … & Fagerlin, A. (2014). Blocks, Ovals, or People? Icon Type Affects Risk Perceptions and Recall of Pictographs. Medical Decision Making, 34(4), 443-453.
What place does health data have on our streets? We live with a myriad of numbers and statistics reported every day though they are easy to miss. One of the deliverables of this infographics project was a series of street-based infographics (or “infoventions”) around Bradford and Leeds in Yorkshire, UK. These were focused around the issues of health and transport in relation to air quality. The purpose of the work was to alert viewers to: alarmingly high air pollution measurements (by covering an actual air quality monitoring station on one of the busiest roads in bradford), deaths-related to air pollution in Leeds (by placing work underneath a car) and the relationship between diesel cars produced before Sep 2015 and the amount of harmful nitrogen oxides they are legally allowed to emit.
By making numbers either site-specific or tied (quite literally) to the objects they relate to, they represent a reality unachievable on the page. They also invite us to look closer by initially raising a ‘visual difficulty’ (see Hullman‘s work).
Hullman states that visual difficulties can induce engagement “by manipulating novelty, tailoring and personalization, challenge and game-play, and aesthetic appeal.” In these infoventions, engagement is designed to occur via novel encounters, tailoring to the environment and aesthetic appeal.
As a basis for the work, mock ups of a whole series of street infographics were shown to
members of the general public and they were favourably received by 90% of the 85 people interviewed. What emerged from interviews was the need to not ‘preach to the converted’ and to ensure work is both timely and transitory. Designs were deemed more effective for attention when placed in appealing but unexpected environments.
When data was shown in mocked-up street infographic form or standard drawn form it was, in 3 out of 4 design pairings, much more likely to be recalled in street infographic form and so we believe the infovention is an effective method for making numbers related to health both more noticable, real, memorable and tailored for communities.
Public Health data is usually concerned with population-level data. This data is usually visualised via a ‘neutral’ objective tone. For instance, icon arrays used for showing proportions consist of identical circles or icons of various consistent shapes. The idea of neutrality in design is problematic – see Robin Kinross’ 1984 article ‘The Rhetoric of Neutrality‘ – and as designers we should be on the look out to challenge the norm and question how infographics communicate on more subliminal levels.
Visualising the health of a population is particularly challenging as it’s constantly shifting. A statistic at any given moment is out-of-date the next and can only ever be based on approximations. It may well be time to acknowledge this within the visual language that we choose to use. Using less precise visual languages, found in unpredictable materials may, in some sense be more precise as a whole. 80 holes drilled in a board result in the same holes on one side but a different ‘story’ on the other, randomly dictated by the reaction of the material. Whilst 80 people may have the same condition they are not the same people and may be dealing with the condition in 80 different ways. What room is there for capturing what I term here ‘more precise imprecision’? I raise this as a question here for us all to consider though there are several interesting places to look for precedents, particularly regarding visualising uncertainty of the future…
Public Health datasets are very rich indeed. You only need look at some of the powerful tools available in the UK such as http://www.tobaccoprofiles.info. Here you can view each city/area and view the prevalence of smoking and smoking-related disease found there. At www.tobaccoprofiles.info the presentation of data is, however, not overly compelling, though its colour coded tables are very legible and transparent for those in local authorities to extract data from. It strikes me though that these are true stories worth sharing outside local authorities and by presenting the data differently (shown above) there are ways of highlighting key messages and getting people in local areas talking.
One of the findings of the testing with the general public in West Yorkshire (we spoke to people in Leeds, Halifax and Wakefield in February 2015) was that they generally appreciated a visual element that reflected the subject of the data presented. This only needed to be quite subtle, particularly when the subject matter could be potentially very negative. In the example above the visual cue is sensitively done using broad shapes to suggest pregnancy and smoking. The colour red used here acts as a warning that there is something that needs our attention.
The people we spoke to also generally appreciated ‘something different’ that didn’t look like a graph or have too many visual elements in it that needed working out. In the example above the visual design has been influenced by the flowing designs of David McCandless. There are, of course, some sacrifices to be made when moving away from more standard formats such as bar graphs. One sacrifice here is complete accuracy in the placement of the elements (in this example, vertical distance is not mathematically indicative of performance). However, absolute values are very clearly displayed, so the viewer can see, for instance that Leeds and Calderdale share the same values even though one is above the other. Where do aesthetics give way to mathematical accuracy when designing for the public? It’s still a grey area. What is of absolute importance is that the key story, that Yorkshire and the Humber is performing poorly nationally, comes across as part of a journey through the visualisation. The end of the ‘smoke’ in this image reveals the story, helped by an annotation layer than picks out ‘the national average’.
What also came across very clearly in the public testing was that people recall and pay attention to data that affects them directly. The first link in this chain of engagement is to make the data more reader-friendly so the viewer gives the data story a chance. The local nature of this compelling data should help make the story relevant.
Local public health data belongs to the public in those local areas. In the case above, Barnsley people need to know there’s a problem, with 1 in 4 women still smoking when pregnant. We need to find better ways to engage people with the data – extracting data from http://www.tobaccoprofiles.info and experimenting with its form is just the start…
What makes an infographic more memorable? We’re all busy designing them or commissioning them, but do we really know what people remember about them? What makes the graphic and its message stick in our minds?
One of our projects recently involved testing short term free recall of health infographics. Whilst we’re still analysing all the data it’s interesting to share a few thoughts so far.
We interviewed 90 members of the general public. We spoke to a wide range of people from all walks of life, aged 18-79. We showed them various single message infographics, captured what they remembered about them and we also asked them why they think they remembered them.
According to the participant comments, what seems to make infographics (and more importantly their messages) ‘stick’ is down to (in order):
1) A personal connection with the data
3) Visual appearance
Content that relates to us personally and tells us something new, according to participants comments, seems more important for recall than the visual qualities of the design. This is a challenge for public health data which focuses on the health story of the population rather than an individual. How then do we get the individual into this story?
One idea that we’re testing is the use of local spaces, to bring the data closer to the people and it make it more relatable. We’ve been doing that through mock ups but will be implementing some of the ideas in reality too after more development and testing.
The top two images here are of a street infographic mock up and the same data presented more conventionally. When we showed these two designs to people (as part of two different sets of 12 infographics of mixed styles) they were more likely to remember the street infographic than the plain version and, most importantly, remember what its message was. 36% of the group that saw the embellished version remembered it whereas only 2% of the group that saw the plain version could recall its message.
Given the fact that on average people didn’t have a high recall rate (on average people remembered only a third of the infographics they’d viewed) 36% is actually quite high so in fact the visual presentation of this image DID make a difference to recall. Perhaps it did that by forging a more personal connection with the data through the depiction of a familiar space or perhaps the ‘newness’ of the presentation technique helped recall. It’s early to make such a claim but it does suggest that what makes health infographics sticky is a complex area which we need to spend more time thinking about.
These findings and more will be published in a forthcoming journal paper though a fuller set of results will be made available on this site too in the form of an extended abstract.
Do you have any studies/case studies on recall and infographics you’d like featured here? Let us know…
In 1924 Harold T Larsen wrote some words that still very much ring true today:
“Are long, monotonous columns of figures statistics if they do not stimulate interest, thought and action? In public health work, perhaps more than in any other field, the matter of presenting figures so that they compel the attention of casually interested people makes the answer to that question worthy of careful consideration”.
How many communicators and designers of public health data wrestle with that question 90 years on? I suspect a lot of us will hold up our hands!
Larsen goes on to say how important the individual is in public health, concluding that “any method of presenting facts that will excite his interest and increase his knowledge of preventive medicine is of value. The use of clear but unusual graphical representations which will arrest his attention and make him think are both justifiable and worthy of use”
Whilst we know much more today about what people understand in terms of graph design for health (see http://www.vizhealth.org for some well tested guidelines), we still don’t know much about what grabs people’s attention or even ‘excites’ them. Can health statistics even be exciting for instance? These are still important questions and I think it’s important to credit Larsen with some foresight. He was making these statements in the United States around the same time that Otto Neurath in Vienna was working on his famous Isotype systems with the aim to make data more compelling and easy to understand. Larsen’s work, in terms of academia, appeared to have little impact (according to Google Scholar it has only been cited 4 times). Whilst we’re probably all aware of the ‘famous’ public health visualisations of the past from John Snow or Florence Nightingale, Larsen’s voice is almost silent today.
His paper is not a particularly easy read and reflects the time it was written. It has none of the ‘snappiness’ of today’s popular and shared content. Instead the paper requires slow reading and a quiet appreciation of the carefully hand rendered visuals.
He presents some intriguing formats such as the spire graph shown below.
He praises this unusual format since it draws attention to its top, guiding the eye to the peak. Though rare to make an appearance these days (and not found in Harris’s ‘Information Graphics: A Comprehensive Illustrated Reference’) it does have a resemblance to the ‘monster’ imaginative made by Nigel Holmes. The chart at the top of this page, given a splash of colour, would sit comfortably within Taschen’s ‘Information Graphics’ weighty tome.
Larsen certainly had some views which cause concern. His views on colour, for instance, are misguided – “Colours in graphics embrace a fallacy not widely recognised”. This is not an empirical paper (how many existed in 1924?) but it’s one of the few papers from a by-gone era that treat the public as being worthy of graphics that engage. He also highlights some of the debates that we continue to discuss such as the ‘controversy’ over use of the third dimension in graph design.
I’d definitely recommend it as a read and I’d class it as a hidden gem in the history of public health graphics.
Larsen, H. T. (1924). GRAPHS IN PUBLIC HEALTH REPORTS. American Journal of Public Health, 14(7), 585-591.
Can we take advantage of the power of infographics to highlight issues with our research methods? Here’s an update to the ‘Evidence Base‘ page today that includes an infographic of sample sizes (samples range from 5 to 3536) and sample types (students or the general public) taken from 31 academic papers published from 1985 to 2014. This is very much a work in progress and the graphic will expand over time so please check back.