Regarding 'Chart Junk'

Graphical elegance is often found in simplicity of design and complexity of data. 

The visual/functional spectrum is something that's grown and grown as an interest for me, particularly after trawling through the blogs and comments on Stephen Few's blog (the ones on McCandless are incredibly interesting, I found this recent one extremely insightful) that there was a clear divide among people who 'practice' data visualisation (let's not even get into the 'is data viz a science or an art' debate) - Should your chart/visualisation serve the purpose to inform, or merely attract interest? Infographics and KPI Dashboards are two examples at either end of this spectrum.

It's very easy to take the numerous amount of literature, blog posts and advice from experts as gospel - you can enforce a bar chart on most data types, and everyone seems to swear that pie charts are evil (well, almost everyone - Matt Francis has kicked off a website to debunk data viz myths) - but falling down deeper and deeper into this rabbit hole has lead me to question things a lot more. Why are thing referred to as chart junk? And most importantly, what IS chart junk?

Chartjunk | Refers to all visual elements in charts and graphs that are not necessary to comprehend the information represented on the graph, or that distract the viewer from this information (Wikipedia, inspired heavily by Tufte)

I feel like there's a bit more to the term chartjunk than this. Tufte mainly talks about the visual elements - which refers to 'bad' use of colour, shading, borders etc. The 'unnecessary use of ink' is seen to be distracting, and adds nothing - how much 'data ink' can you delete before the visualization stops making sense?

The above chart was found here on Bloomberg Businessweek, where they explored some Tufte-isms - Including needless colour coding, needless visual elements such as gridlines, 3D effects and labeling, and simply anything distracting from the visualization as a whole.

Essentially, anything non-informative or elements of the chart which aren't adding anything to the visualization qualifies - added dimensions, questionable use of colour, visual format elements like grid-lines not adding any more information for the end user. While reading up for this post, it was fascinating to read Tufte on this subject - the man is *so* passionate! Take these choice parts (from here);

"The interior decoration of graphics generates a lot of ink that does not tell the viewer anything new... Like weeds, many varieties of chartjunk flourish"

But for me, chartjunk can also refer to bad chart decisions - and this is actually something I've been guilty of during 'trials' of charts and dashboards.

Let's start with the dashboard; with 'effective dashboard design', depending on the message that you're trying to convey using the visualization, each part of the dashboard should contribute something - otherwise, it can count as 'chartjunk'.

- How much value is that bar chart adding?

- Is that trend line showing the same data differently?

- Why are there packed bubbles used here?

The principle, however, remains the same - How much can you erase/delete before the objective or purpose of the dashboard (or visualisation) becomes redundant? Robert Kosara has a really interesting blog post on chartjunk (here) which links to Stephen Few (here) - The crux being that visual elements and use of ink aren't bad 'period'.. But more that they should be used with care, and the end user in mind.

For the way that 3D charts/shapes, shading etc work - fine. But for the use of colour, there's a whole different spectrum to consider; but with relation to chartjunk, and taking it at the face value of 'usage of data ink' - I tend to agree with Few and Kosara... It depends, therefore 'relative' chartjunk has it's place.

The key part of gridlines and strong colours is the impact; by this, it refers to weight of a line chart, or the gradient used for a bar chart or the strength of the colour of the gridlines. Depending on the end user, gridlines are  useful for allowing the eye to follow, and shading highlights certain areas - and extreme shading (also known as highlighting..) brings out the strongest elements of a visualization, the parts which carry the message or that the end user should be drawn to.

Visual elements can help guide the user, and ensuring that there's enough guidance, given that context in visualization is key - a good question to ask yourself is, where does this viz sit in the wider scheme of things? How does it support or add to the analysis/question posed?

I think the term 'chartjunk' has evolved and developed, but the ever poignant statement of 'it depends...' remains over it's head. As always, I would love to hear your thoughts on this topic!