Datavisualisering i samfunnet
[Content warning: suicide] The section “Challenges with presenting and visualising data and discoveries in news articles” is based upon a news report covering suicide statistics. Feel free to scroll past that section.
Datavisualisering i samfunnet (data visualisation in society) was a free, half-day conference, arranged to conclude INDVIL, a 3.5 year research project administered by the University of Agder on trends and consequences of increased usage of data visualisation for communication (my paraphrasing).
Data visualisation is an area of software development that has piqued my interest in recent years. I bought a copy D3.js in action two years ago (still not found time to finish it) and do my best to follow talks from conferences that are published online such as those from OpenVis 2018.
The day started with lunch and a small datavis exhibition with multiple interesting stands, the media was of course represented with Bergens Tidende and Verdens Gang showing some of their visualisations such as reporting on maintenance status of Norway’s bridges and train delays. A company called Ferdio had a stand showing off their Data Vis Project library of visualisation inspiration. There were a couple of stands relating to visualisation literacy, Datatrotters were there with their board game for teaching visualisation literacy (I placed an order online, got the shipping notification today) along with Tallgrafikk a website with the goal of teaching data visualisation literacy. Lastly the more fun statistrikk.no was there showing off work of representing data visualisations as knitting patterns.
Presentation of the book
The first session started with an introduction the INDVIL research project quickly leading on to an introduction to the book that will contain easy to digest articles about the topics that have been researched. The book will be published for free by Amsterdam University Press early in 2020.
The book is described as an important step towards the creation of a publication on the topic of philosophy in data visualisation, which reminds me that I have the highly recommended A Philosophy of Software Design sitting on my shelf waiting to be read.
Then the researchers that were present were invited up on stage to give a lightning talk on a part of the book that they contributed to. Lots of interesting topics came up that I wasn’t able to make notes on during the presentation, but it was enough to inspire that I should get my hands on the book when it comes out.
The second section of the afternoon’s program was split into four parallel sessions, it was incredibly difficult trying to choose here as all sounded interesting but in the end I landed on the “Design and esthetics in data visualisation” track that felt like one of the more practical sessions. I think this day could have easily been a whole day conference with a maximum of two tracks.
Why we should start designing visualisations from our own life
This presentation by Sara Brinch, one of the researchers, was proposing that it can be useful to use visualisation techniques on your own life in order to both better understand yourself and other peoples visualisations. Examples of sources for data could include a diary or your mobile phone.
For those with less technical skills Jill Simpson’s A Day of OCD was provided as an example and the book OBSERVE, COLLECT, DRAW! by Giorgia Lupi was recommended as it is a guided diary for collecting information graphically.
This talk inspired many thoughts about what I could do in my own life. Recently I commented on a slack discussion that I liked the annual plan visualisation from the website of a consultancy that a friend recently started at, I’m now wondering if something similar could be repurposed to document family or life activities that will happen throughout the year.
Another idea we’ve been discussing at home would be to have a world map so that we could discuss with the children when one or the other parent is off travelling to a conference or similar. This could be augmented with physical pins representing who and with information of when. Maybe I’ll develop this thought over the Christmas holidays.
Why the look of simple elements matters
This was a very interesting talk on how we communicate information through the styling of lines. Verena Lechner has researched how we communicate information through the use of lines whether they are continuous or broken up (dotted, dashed, etc), what colour they are, the thickness. Continuing into techniques that are enabled in digital mediums such as opacity and widening lines as mechanisms that can be used to communicate about the uncertainty of the data points.
References used here were the xkcd on earth’s temperature, which makes use of three different line styles (dotted, solid, dashed). The length of the line tells a story in itself, then the three diverging options for the future communicate alternative outcomes with the dashed lines indicating uncertainty.
Sharpiegate was also used to illustrate the great significance simple lines can have.
All in all, a great presentation from someone who appears to know her subject in excrucuating detail. It’s always fascinating to hear from people who have very deep knowledge of a topic the rest of us don’t actively think of on a day to day basis. Looking forward to reading this section of the book.
Challenges with presenting and visualising data and discoveries in news articles
Next up was Mari Grafsrønningen from NRK. This talk opened up by playing the audio of a distraught mother looking for answers after her son had committed siucide while under observation. For some reason, as someone with at least one friend who has been “in the system” and a relatively new parent who hopes never to experience something simar, this caught me completely by off-guard, the difficult topics this talk relates to could have been mentioned clearly in the schedule for the day. I’m not sure that a content warning would have changed my choice of track for the day, but I would have been more mentally prepared.
With that out of the way, that incident was the trigger for a huge investigation and the talk goes into the challenges of extracting insight from unstructured data, condensing large amounts of data in a manner that can be digested and presenting very sensitive data while maintaining confidentiality and credibility. The result of this work was the suicide in psychiatry article.
Generally an interesting talk but I’d be very interested in a follow-up article in a year’s time to see if things have changed during the two years since the original article, but I guess that’s not going to be as headline-grabbing.
What helps us to design data visualizations, and what doesn’t
Birger Morgenstjerne from Ferdio presented 10 of the tips from Ferdio’s Information Designer’s Notebook
- Have a goal
- Story is king
- Invest in sketching
- If you can’t explain it simply, you don’t understand it well enough
- Test early, test often
- Collect, sort, analyze, visualize
- Put numbers in context
- Don’t let the tools decide your design
- Use common data visualisations for common understanding
- Data is imperfect
Not sure anything I write here will summarize any better than going and reading through the notebook yourself. Through this notebook and the Data Vis Project mentioned earlier Ferdio have published some great resources for people working with datavis.
This part of the day concluded with Q&A panel discussion with those that held the talks. From the audience questions there was a clear fascination for the topic of lines, which speaks to how well the topic was presented.
State of the Art 2019
I had to head home at this point in the day to collect kids from kindergarten but I caught up with Andy Kirk’s closing keynote from the live stream.
There were multiple subjects covered in the talk that I found interesting, firstly the fact that we’re now visualising more and more qualitative data which needs new visualisation techniques to communicate the fuzziness inherent in qualitative analysis. The difficulties of emulation vs. plagiarism when it comes to new visualisation techniques. The changes that the office of national statistics in the UK have made to guidelines in presenting conclusions that can be extracted from the data instead of just presenting data and expecting people to see any significance in the data. Sonification as a novel technique for interpreting data. Lastly, I noted that the challenges of communicating uncertainty in data was a recurrence of a common theme throughout the day.
My greatest take-away from the day was learning about the issue of visualisation literacy, we tend to think of graphs and diagrams as methods of making data easier to digest than reading tables of data, but there is still a huge potential barrier to entry and pitfalls to look out for when someone creates a visualisation with the intention to mislead, with the classic truncated y-axis, for example.
I’m no graphical designer but this day was hugely inspirational, looking forward to reading the book when it comes out and I hope I can find a way to get more involved in datavis on future projects. It’s quite a shame that this was a one off event coupled to the end of the research project, it definitely seems like there’s room for a datavis specific event in Oslo’s conference calendar.