Effective Data Visualisation Examples for Researchers

The main goal of data visualisation is to communicate complex statistical information as simply and memorably as possible using graphics. Improve how you present data with 5 examples of effective data visualisation based on research from psychology, information technology and data journalism.

Last year, I presented on “Data Visualisation and Research Communication” at several research institutes. I’m still getting started with data visualisation, so I want to document the research I did for these presentations for other people who are just starting out in the field.

Let’s get started by asking the obvious question. What is an effective data visualisation?

The primary goal of data visualisation is to communicate complex statistical information as simply and memorably as possible using graphics. If you lose sight of this aim, the effectiveness of your visualisation will suffer.

“The purpose of visualization is insight, not pictures.”

Ben Shneiderman, Professor of Computer Science at University of Maryland

Researchers may not know how to use design techniques to improve the structure of their visualisations. Many also are more comfortable focusing on their methodology rather than the significance of the end results, even when talking to non-scientific stakeholders.

Designers like myself often come at things from a visuals-first perspective, which sometimes doesn’t pay enough attention to the actual data. Unless they’re very specialised, designers probably don’t understand statistics enough to be comfortable suggesting alternative visualisations and stick to safer topics like adjusting colour, font-size and line-weights.

Ideally, both analysts/data scientists and visual designers can collaborate to focus on their strengths. Both sides need to understand how research into human perception and memory, along with established best practices in computer science, business intelligence and data journalism can help them create better data visualisations.

To try and bridge this gap, I want to focus on five key strengths of data visualisation in communicating research outcomes. These are not the only things to take into consideration, but they’re a good grounding in what makes for effective data visualisation before we start talking about strategy.

  1. Your brain recognises visual shapes far faster than text or numbers.
  2. You can understand complex data more efficiently as a visual pattern
  3. You can present significant results clearly
  4. Visuals allow your writing to provide a deeper analysis
  5. Relevant, informative visuals are critical for your outreach efforts

1. Your brain recognises visual shapes far faster than text or numbers

There’s an oft-cited figure that people can process images up to 60,000 times faster than text, but there seems to be no source for this exaggerated claim beyond a marketing campaign by 3M!

A more reliable source is a 2014 neuroscience survey conducted by researchers at MIT, which discovered that test subjects could still recall key details of images after being exposed to them for as little as 13 milliseconds. Importantly, certain types of visuals are easier to remember – faces and people are far more likely to be recalled than landscapes, for instance.

Researchers observed similar results in testing how memorable different visualisations are in a joint Harvard/MIT study. Participants saw a series of visualisations for one second each. Recognisable images or symbols (especially human-centric ones) were far more likely to be remembered accurately than more abstract visuals.

Interestingly enough, the same study found that when it comes to charts, less common (and more complex) chart types such as treemaps are far more memorable than the more common types such as bar charts. Researchers hypothesised that they were different enough to stand out, but the branch/flow structure was familiar enough to be easily understood.

Visualisation of the 'black budget' in US intelligence spending by Martin Grandjean
Visualisation of the ‘black budget’ in US intelligence spending by Martin Grandjean

 

2. You can understand complex data more efficiently as a visual pattern

An early example of data journalism is John Snow’s cholera map (1854), which linked the source of cholera to a pump which was later discovered to have become contaminated by leakage from a septic pit.

This map of pumps, cholera cases and streets was published by Dr Snow after the outbreak to illustrate and support his theory of contamination through human waste in a visual summary, without having to leaf through every patient’s individual case history.

John Snow's cholera map, sourced from the University of Delaware
John Snow’s cholera map shows cholera cases by their address on a street map of Soho, London in 1854. Original file from the archives of the University of Delaware.

Even simple data sets such as sales tables can benefit from visualisation for comparing trends or looking for patterns. That’s not to say that tabular data doesn’t have its place – it depends on the strategy of presenting the data in the first place. Think about giving a presentation – it’s far more useful for participants to have a printed hand-out of your data in table form, while your presentation focuses on visualising key points.

Research by Joline Morrison & Doug Vogel on presentations, published in Information & Management, 1998, found that visual aides help improve comprehension and retention of information.

Don’t dismiss the insights you can get from data visualisation when exploring data either. As datasets grow in complexity toward ‘Big Data’ involving terabytes of data, visual interfaces are becoming an essential tool for even data scientists and analysts to observe trends and test hypotheses. There’s probably no better example of this in action than Hans Rosling’s pioneering work with interactive visualisations at Gapminder. Seriously, check them out if you’re not already familiar with what they do. 

Imagine the tools Dr Snow would need to track patterns in a similar scale outbreak in modern London. In fact, epidemiologists now have global insight into disease alerts through online applications such as Health Map.

3. You can present significant results more clearly

We’ve already talked about how visuals can be understood and recalled faster than text, and how visual patterns are easier to read and can help with comprehending information.

For even better comprehension, Tom Davenport, Senior Advisor to Deloitte Analytics highlights the importance of focussing on telling a story with your data to help engage decision makers. One of the interesting points he makes is that among recruiters of business analysts, better communication skills is one of the top requirements, not methodological prowess. 

Andrew Kaplan from LinkedIn shared three examples of effective data-driven storytelling, one of which, from Bloomberg, used data visualisation to address the question “What’s really warming the world?”.

Visualisation of public data on climate change by Eric Roston and Blacki Migliozzi, Bloomberg Business 2015
What’s really warming the world?” Visualisation of public data on climate change by Eric Roston and Blacki Migliozzi, Bloomberg Business 2015

 

This chart is an excellent example of how to use simple animation and focused, rather than kitchen-sink graphs to answer specific questions and show trends clearly.

Stories are very powerful tools for communication, but they can seduce you to disregard inconvenient facts for a more appealing narrative. For more on this, I recommend reading this article by Jonathan Gottschall on Theranos and the Dark Side of Storytelling.

However objective you intend to be, I would argue that researchers have a responsibility to remove ambiguity and room for manipulation by emphasising context, relevance and key points. It’s probably impossible to eliminate the risk of findings being taken out of context or misrepresented, but by focussing on the things that are significant, you can make it much harder.

In short, stories are perfect for engaging with stakeholders and keepings things snappy and exciting by not focussing on methods and table after table of data, just so long as you don’t cross over into misrepresenting the significance of your data.

4. Visuals allow your text to provide a deeper analysis

A good visualisation provides a high-level overview of a particular query and should be self-explanatory. The accompanying article can delve deeper into context and analysis of the topic, according to Alan Smith, Data Visualisation Editor at the Financial Times.

In general, it’s a good idea to restate key points while avoiding too much outright duplication of information which can frustrate readers. After all, one of the most common complaints about technical books is the amount of redundant information included!

Visualisation by Ella Koeze / FiveThirtyEight.com of changes in pickups by Uber / yellow c
Visualisation by Ella Koeze / FiveThirtyEight.com of changes in pickups by Uber / yellow cabs in NYC.

In an analysis of the competition between Uber and traditional taxis in New York City on FiveThirtyEight, the authors use data-based visualisations such as the example above throughout the text. Although there is a certain amount of cross-over, the copy gives additional context that isn’t possible to include in the graph (Uber’s incentive programme, for instance). The visuals work to clearly show what is happening, while the text supports this by asking the whys – why is this happening, why does it matter, and so on.

If you work with qualitative and quantitative data, then the focus of the text can shift toward including representative samples of the former, for instance. A general rule of thumb is the visualisation should deal with your results, while the text deals with the significance of your results.

Another useful place to use visualisations is at the end of a section to serve as a quick recap of the most important points, bringing together ideas that were still quite abstract to the viewer in into an ‘aha!’ moment where they suddenly see how things relate to one another.

5. Relevant visuals are critical for your outreach efforts

High-quality visual content is one of the best performing types of content for social networks, and data visualisations can help convey complex stories in an immediate format that’s actually useful to people! 

Data visualisations have been identified as one of the hottest trends in ‘content marketing’ by Alexandra Samuel in the Harvard Business Review because it allows users to access quantitative data that’s vital to them in a digestible format. 

Now you can’t slap any old chart on to social media and expect magical results. Resources do need to be invested into making sure the chart is the right size and format for the social network it’s going on, that everything is legible, and that you’re targeting the right audience with your post.

In this example from the Financial Times, you can see just how popular a topical data visualisation can be!

Newspapers such as the Financial Times, New York Times and Washington Post have invested heavily in data visualisation, and data journalism is leading the application of academic theories towards better communication, so we’ll be revisiting more examples from them.

Wrapping up and further reading

You should have a good idea about the value of data visualisation now.

For even more information on this, I recommend starting with Data Visualization for Human Perception by Stephen Few, on the Interaction Design Foundation website.

Other useful resources include the Chart Doctor blog on the Financial Times, Storytelling with Data by Cole Nussbaumer Knaflic, and the blog of Alberto Cairo, The Functional Art.

If you’re already excited about the possibilities, it’s probably tempting to jump straight into creating new visualisations. However, if you don’t have a plan, you might end up spending a lot of time and other resources on visuals that don’t have any real value.

In the next post in this series, I’ll dig into visual strategy a lot more, so please stay tuned. After that in Part 3, I’m going to dig even deeper into the research into what makes data visualisations useful, with specific tips and techniques to get consistently better, more informative and memorable visualisations. 

If you want to receive updates on new posts automatically, subscribe to have posts sent to you by email, or follow me on Twitter or LinkedIn.

Want more impact from your ideas?

Sign up for monthly updates on how to present and promote your ideas and expertise.

Janto McMullin

Want more impact from your ideas?

Sign up for monthly updates on how to present and promote your ideas and expertise.

Leave a comment