The authority of a graph

With so many graphics software applications now available, it’s never been easier to turn complex data sets into interactive charts, maps, and infographics. However, using contemporary visuals to represent and understand big data often raises the question, “That looks nice… but, what exactly is it meant to show?” network Source: http://commons.wikimedia.org/wiki/File:Social_Network_Analysis_Visualization.png?uselang=en-gb

Performing an internet search on ‘data visualisation’, you will undoubtedly come across the statement: “studies show the brain processes images 60,000 times faster than text”. The actual number is unsubstantiated, and has been debunked extensively on the web, but it is a fact that the brain interprets written words as physical objects, and reading is a cognitive skill that requires further neuronal processing than a picture entails. In the learning analytics arena, communicating progress data to students as a dashboard of charts allows for ease of transmission, but there is a risk that graphs could be misinterpreted by some and carry disproportionate meaning. This suggests that we shouldn’t release “dashboards” without due support to help students interpret and make sense of their data.

The unintentional impact of a pretty picture

The persuasive nature of graphs is illustrated humorously in two articles I recall, both published in respected journals. In the first, we see convincing graphs showing a significant correlation between the stork population and numbers of out-of-hospital births (New Evidence for the Theory of the Stork, [1]) and in the second paper, a graph reveals the remarkable correlation between countries’ per capita chocolate consumption the number of Nobel prize winners (Chocolate Consumption, Cognitive Function, and Nobel Laureates, [2]). The data for these trends are genuine but they are nonsense correlations by which the authors demonstrate the fallibility of scientific research. I’m not suggesting that academics will mislead students with irrelevant data, rather that the visual nature of a graph can appear to carry more authority than if the data were presented as text, a list or table. Indeed, it’s been shown that people are more likely to be persuaded by graphical representations of data compared to the same information in text format (e.g. Blank Slates or Closed Minds?, [3]). With that awareness, we must ensure supportive lines of communication exist between tutor and student to avoid a negative graphic of progress resulting in disengagement with the learning environment. Equally, we must ensure that students are not reduced to passive consumers of this information, but are given guidance on options for action to help them improve their learning outcomes.

So, what type of visualisation works best? I confess, I have become somewhat jaded with the fast-fact infographic, distilling ‘key’ information into a colourful cartoon format (one can’t help but wonder if some of the facts have been added simply to fill a space and/or utilise the pretty imagery available) and, although it clearly has a place in many contexts, I don’t think it is a format suitable for disseminating learning analytics data to individual students. Graphs can convey a significant amount of information rapidly and explicitly, but we must be sure to label axes and provide an informative legend to avoid any ambiguity. From scatter plots, bar graphs, pie charts, through networks, heat maps, bubble charts, and steam graphs, we have a lot of choice. Whilst the primary goal is to communicate information, we can aim to select graphic formats that are both aesthetically appealing and intuitively insightful. The added bonus of today’s visualisation software is the interactive element, providing students with a live, visual window on their progress, in comparison to their peers. I like the idea… but will my students? I shall update you in a future blog post!

[1] Höfer T, Przyrembel H, Verleger S (2004) New evidence for the theory of the stork. Paediatr. Perinat. Epidemiol. 18(1):88-92.

[2] Messerli FH (2012) Chocolate consumption, cognitive function, and Nobel laureates. N. Engl. J. Med. 367(16):1562-1564.

[3] Brendan Nyhan B, Reifler J (2013) Blank slates or closed minds? The role of information deficits and identity threat in the prevalence of misperceptions. http://www.dartmouth.edu/~nyhan/

Paula Smith, IS-TEL Secondee