#btconf Düsseldorf, Germany 08 - 09 Nov 2021

Matthias Stahl

Matthias is a trained biochemist and bioinformatician. After dismantling mechanisms of multi-resistant bacteria and a dive into the big data secrets of childhood leukemia, he realized: No breakthrough is possible without compelling data visualizations. Luckily, at the same time, he discovered his passion for the combination of data & statistics with art & design.

In 2020 he founded his data visualization studio higsch, where he specialized on creative, interactive data visualizations. After a year of exciting projects across topics in politics, medicine and human rights, he decided to join the weekly news magazine DER SPIEGEL. Today he is leading their Graphics & Interactive department.

Want to watch this video on YouTube directly? This way, please.

Looking like a Lost Sheep – the Story of the Lonely Chart

Data visualization consists of lines and bars and we love the simplicity of scatter plots. But data visualization is much more than mere shapes. Visualizations are emotional, they are shouting ‘hey look at me’ and eventually let the observers immerse in the data.

Oh sorry, this is just one of my bold dreams.

Out in the world we often see charts and graphs that just try to tortuously mirror the data they were fed with. Sometimes they do not communicate with their readers at all or do not even catch the readers’ attention.

But how can we change this, and which role can data art play in this dance between visual and reader?

Let me take you on a journey through the world of data visualization and how it could look in future.



Matthias Stahl: Hi, there. I am Matthias Stahl and, first of all, I am really sorry that I cannot be in Dusseldorf today. At least I am streaming. I am based in Hamburg here, so not that far away, and I am hoping to feel a little bit the vibes of this wonderful location in Dusseldorf where you are.

I’m following up the conference on Twitter and via live stream all the time. And it’s really cool what Marc has put together here. Thank you a lot, Marc.

So, today, I want to take you on a journey through the world of data visualization. I want to tell you the story of the lonely chart or why we need more data art. So, I love data visualization. I really live and breathe data visualization. But to be honest, it took some time to realize that.

I first started out as a biochemist, so I really started biochemistry back in the days in Munich where we started in the research group, bacteria. Bacteria can be the reason for multiple diseases that we might have not for virus diseases. That’s not a but, but for other diseases. [Laughter]

And the problem with bacteria is that they get really fast resistant against antibiotics. And sometimes, especially in hospitals, you can get a bacterial infection that cannot be treated anymore with any of the antibiotics out there that we get from most pharma companies.

This is bacteria that are multi-resistant. And we, in our research group, tried to find alternative treatment strategies to fight these multi-resistant bacteria, and that was quite cool. And one of our key methods that we used -- attention -- was mass spectrometry. And it’s not really important to know what it is, but it’s just important to know that it produces a lot of data.

It produces like one gigabyte of data per hour, and we need to analyze the data. And it’s not like that you could just open up Excel. No, you have to write usually your own analysis software. And so, I switched over to the field of bioinformatics because I always had a little bit of a passion for programming. And so, I programmed our own analysis software, and that was quite cool.

One of the cool things with programming--with bioinformatics, which is the data science of biology--is that you need to do a lot of data visualization. In fact, it’s the key part of data science, in my opinion, is data visualization because then you can make the breakthroughs. Then you transform information into knowledge.

I was so fascinated about that that I even did a Ph.D. I wrote a dissertation and did research work on bioinformatics versus biochemistry and all those kinds of things. But you know what? I had the most fun with writing that dissertation when I did the figures. It was really cool to select the colors, to select the right shapes, to select the right charts, and to make really unique data vis for that dissertation.

Anyway, I thought, well, it’s nice that you love your job but, well, let’s go on. Let’s learn bioinformatics, the solid, the right way, and I moved to Stockholm, which is a hotspot in Europe for bioinformatics. There, in Stockholm, I programmed a lot. I programmed the whole day, so I was not in the lab anymore, but I got the data from the lab workers and we worked on childhood leukemia, and we tried to find better therapies for these children just by writing code and doing data vis.

This was really fascinating to me but, after two years in Sweden, I also realized, well, making the figures is still the fun part and it’s still my real passion about my work. And I also realized that while making figures for research projects is one thing but, of course, you could apply data visualization for so many other topics as well. In politics, it’s important. It’s now important for epidemiology if you look at all the corona pandemic charts that newspapers have, for example.

And so, after all, I decided to quit academia and to start my own company, Higsch, which is a data visualization studio, because I just saw, well, taking the math and the data that I knew from my bioinformatics job, but combining that with thoughtful design, with cool tools, and also the part that you communicate with your audience so that you have to listen, “Hey, what do they want, and how can I tell them a story?” that was really, really exciting to me.

And so, this figure from Ali Tobin (a friend of mine and data vis colleague) really shows that passion of what data vis really is, and I think that’s quite cool. And so, yeah, I worked for a year in data visualization as a freelancer. I had many projects, and I will show you some of them later on.

Yeah, after a year, I eventually got contacted by Der Spiegel, and they asked me to do the same thing but just for their newspaper, which is one of the largest in Germany, so I couldn’t decline that offer. Now I’m working at Der Spiegel and doing data vis there. Yeah, that’s a really cool thing.

Yeah, we heard it, I think, yesterday, quite a lot of times: “Follow your passion. Do the job that you love.” I think, so far, I’m quite happy that I had the chance to follow my passion.

This could end like that, for example. So, when I worked for Der Spiegel, we have a print magazine, but we also have a website. Sometimes, we are very happy that our data vis, even if it’s just small things, ends up on the cover page like that one in the summer, which was really cool.

Yeah, so now let’s dive a little bit deeper into the topic and, for today, I brought you a dataset, which is the under-five mortality rates of children by UNICEF. This data set is quite popular because it’s able to tell so many different stories. Yeah, I just want to explain to you how it is built up.

You can see on the top left-hand side there are the countries, like Seychelles, Mauritius, Libya, and so on. There are 200+ countries in the dataset. For each of the countries, we get the timeline of child mortality rates starting in the ‘50s.

For Seychelles, you can see here the first figure is 116.85. This means that from a thousand born children, 116 died under the age of five years old. This is a very high child mortality rate but quite common in the ‘50s, actually.

You can also see that there are missing values because not all the countries report their data steadily or started reporting later, and so on. Now, if I would say to you, “Hey, can you just visualize that?” because I don’t want to look at the very large table, but can you just put together a quick visualization that explains to me the whole table? Then maybe you open up Excel and you choose the standard default plot, a line chart, and we get something like this where just all the countries are a single line, so every country one line. We can see that child mortality rates, overall, are declining from the ‘50s up until the most recent date, which is in 2018 here.

There are spikes in between, but we cannot really access this data, so we do not really know which country is it. The color choice, yeah, there are some repeating colors, I would say, and it looks nice but it’s just telling one story, which is child mortality rates are going down.

Yeah, this is a good thing, but this is something that usually you do not surprise your readership, your audience with that fact because I think everyone roughly knows, well, child mortality rates are decreasing because we have better medicines. We have better healthcare systems. We have better hygiene standards. After all, that’s true for all of the world.

So, this is something I often call a lonely chart because we have a lot of these charts out there on websites, in newspapers even, and nobody is really looking at them because they are just not attracting the readers’ interest. And so, they just stay lonely and nobody wants to look at them.

I think we have to do a step back and just ask ourselves what is data visualization then if that is not a good data visualization. What is it? I really like the analogy that data visualization is like a radio show. Why? Because you as the one that’s building up the data visualization, you are the show master and you want to communicate information during your radio show.

But on the other hand, you don’t see the audience, so you can talk and you can inform, but you do not get direct feedback of your audience the same like a radio show. But at the same time, you also want to inform and maybe you also want to entertain. If you succeed in doing both, that’s really cool because when you entertain, informing is easier because your audience is more open to it and so on.

Data visualization is really like a radio show. If we do now a little physicalization experience and look at one of these radio devices, then of course the radio device is the medium that the radio show needs. We also need the sender, of course. That would be like the show master. And we need the receiver, which are the listeners to your radio show.

If I transform that now to data visualization, the sender is of course you because you are creating the data visualization. The receiver is your audience, whoever that will be, so we’ll talk about that in a second. And your medium, this can be various things in data vis.

It could be just paper. Think of our news magazine, for example. Think of a report or something. it could be a browser, which is maybe the most popular form right now where you have the largest audience at the moment. Yeah, these are static data visualizations, but these are also interactive data visualizations.

It could also be an audio device because there is a thing like data sonification where you try to turn data into audio signals, into (I wouldn’t say) noise, [laughter] but into nice tones. And there are many more, so there is also data physicalization where you really use bodies, tools, or something like that to represent data, so you have a various amount of different media that you could use.

But you have to keep in mind that talking about your medium, you are definitely not in control of the channel selector. At least not in direct control. You do not know if your audience will look at your data visualization or if they just turn the page and say, “Well, it looks not interesting. Next one.”

And you are not in control of the volume button. I think that’s really important to know. You don’t know how strongly the data vis affects your audience. But you are in indirect control of these two buttons, and we’ll talk about that; how we can be in indirect control of the channel selector and the volume button.

This chart here really may be the channel of that data visualization is selected, and someone is shortly looking at it, but then the volume is really turned down because it’s not interesting. While the colors look somewhat interesting, but then you cannot read anything and you just see something that you already knew.

Would you listen to that radio show? Well, if not, how can we produce then effective data visualization and have the name it already says? In the center of your data visualization, you have the data, so you have to look at the data. What is in there? Are there any patterns?

Then around the data set, you build a story. Well, there’s something else around. But in fact, you build stories (in plural) because there might be multiple stories in your data set.

And this thing that I drew now very faintly around, we’ll talk about that in a second.

But let’s just have a look why the story is so important. I always love to bring up the Twitter post that goes around like every half-year. It’s unfortunately very hard to find the original source, but the data really is like unsorted Lego bricks. Then, as a data scientist, you might know that already. You saw that you arrange it. You wrangle through data. You tidy up your data set.

In fact, you present it visually. That would be more or less the lonely chart because, well, you can read it. This would be like a bar chart right now, and we can count how many different Lego bricks there are. But this is not really attracting the interest of your audience.

You need to explain it with a story. This is nicely shown here by building up that house, but this is, of course, just an analogy. You need to build a story around.

And of course, it could be a thousand stories because data visualization, unlike a text passage, is able to really tell multiple stories at once and, depending on your audience, your audience might see different stories in the data vis, or see them in a different order, and that’s quite interesting.

Going back to that model, we have the data. We build multiple story levels around. Then there’s something around, which I want to talk in the last part because that’s also very important. Stay tuned for that.

What does that really mean? I thought it might be best to explain this really abstract model, data story, and also the whole idea of what data visualization by showing you my process. When I want to do a data visualization, I have four distinct phases.

Phase number one is assessing. That means I ask questions about the data set and, at the same time, I try to define my audience because, you remember, during the radio show, I have no chance to see my audience. But I can at least assume things about my audience.

Asking questions about the data set, this could be: What does the data show? - very basic. Then a little bit advanced: Are there any patterns that are interesting? Are there any limits, maybe? Are there things that people think that they just can’t explain, but actually the data set cannot explain it?

And, at the same time, defining the audience, so who could it be? What do they already know, or what do they care about, or are they able to read statistical charts? Things like that are quite interesting.

Yeah, a really cool talk yesterday about the sketching. Really, this is something you could use sketching really well because you just take a piece of paper and you define your audience by sketching. You can draw a little bit, but you could also just write down your ideas who the audience might be.

What does that mean regarding our child mortality data set? Well, we can ask questions and see, while there is of course the child mortality rates and we already know they are going down. But if we advance a little bit, we could see that some countries perform better than other countries. That might be an interesting story later on.

We could also see that not all the countries have declining rates. There are really countries that had rising child mortality in the end. We haven’t seen that in the initial chart. But when you scroll through the data set, you can see it.

Child mortality, of course, depends on continent. This might be also something which might be an interesting story to tell. For example, how does Europe behave in comparison to Africa or Asia or even South America, which is really not known, not commonly known?

Talking about the audience, this is something I always do in parallel. I look at the data. Then I look at the audience. Then I switch back sometimes. Are they used to statistics?

Just as I said before, I would say if I assumed that we have a Der Spiegel readership, they are used to statistics. They are also usually Internet users, so they’re using our app and our website a lot, so they know how interactive graphics work.

We have a very wide age range, and we also want to tell our story to the older ones, but we also want to tell our story, but maybe a different story, to the younger ones. That’s usually a large challenge that we face.

That would be the first phase, so assessing. After that, I go to actual planning. For the planning, I, first of all, define concepts.

I define for myself -- also by sketching, by the way -- a message and the structure. It could also be multiple messages, of course, but it’s usually cool to have one center message where you build your structure of the data vis around. That means the message here could be, “Well, child mortality is declining worldwide, but with different rates and not everywhere.” That would be a centerpiece of message that we could put in our data visualization.

Talking about the structure, it could ease the message transporting or to convey the message by just showing the child mortality range sorted by continent, sort them by country, and maybe also sort them by time. These three variables might be quite important.

Of course, you try to not make everything at once, but here I decided now to build an interactive graphic, so we could play with different user interactions, maybe. But just for the basic structure, that is good to know that sorting by continent, country, and time might make sense to convey the message above.

After that, after I define the concept, I go and usually find inspiration. There are multiple ways to achieve that. I always recommend go for a walk in your neighborhood. Go out to nature and just look what there is in terms of colors, in terms of shapes. Or you find, also, inspiration when you go to websites like Pinterest.

At Pinterest, you can also create a so-called mood board, which is just a white canvas where you collect different artworks, different patterns, different colors, different other data visualizations that you think that these transport the mood of your data vis. This could look like that, so here is a general mood board where I collect radial stuff where I think, well, how could data visualization look like when you want to show radial things.

I show you this because I also had the impression that radial data visualization might be quite cool for the child mortalities because there is this little paradox that you have data, which is really sad, which is showing that thousands or millions of kids were and are dying, but on the other hand you bring in the circle, which stands for perfectness. This is always a really good combination to put in a little bit of tension into your data vis, which is quite good to attract the interest of the audience.

You find inspiration and you decide a little bit what is the mood of my data visualization. Then it’s time to sketch the first graphic ideas. You really take a piece of paper and just say, “Okay, well, if this is a circle data visualization, I’ll draw a circle. Then I have the continents, maybe. Maybe it might be a good idea to focus not on the whole range of data, but just showing the last 20 years.” Things like that, you can sketch.

As you can see here in that sketch that I did for the data visualization of the child mortality rates, I really cannot draw. [Laughter] It looks ugly, I know. But a sketch is usually just for you or for presentations at Beyond Tellerrand, of course. If you understand your sketch, that’s fairly enough.

I’ll show you another sketch, which is maybe a bit clearer. It’s a closeup where you have, in the center, we have the different countries arranged. Each angle is a different country, and then they are sorted by continent. Then the radius is starting from 1998 and showing, in circles, how large, how high the child mortality rate was, and how it developed until 2018.

That was just the idea. There is no color involved yet. This is something that you can pick from the mood board later. But this is basically how sketching works.

But you have to remember here one important rule that I would recommend you 100% to follow. Don’t think of any tools here. Just sketch with pen and paper. Don’t use any tools that restrict you in your creativity because when you think, “Okay, I have to build that visualization later in JavaScript with the help of D3, maybe, or in Illustrator or,” I don’t know “in Figma,” and so on.

If you have that in mind, you restrict your own creativity because you say, “Well, I don’t know how to implement that with D3, so let’s take another representation.” I think that’s really bad because you have to just--

Yeah, you just need to bring your creativity to the paper so that you can develop the ideas. After that, you can ask yourself, “Okay. How do I build that?”

That’s exactly my third phase, the building process. Then you can choose a tool, and you build the data visualization and iterate over the design.

Now, I know a lot of people who are saying, “Well, I want to have the design before I build that,” and I think that’s possible in like 1% of the cases. In most of the cases, it’s not possible. The simple reason is your data set is so large so that you cannot imagine (while sketching) how it really looks like in the end. So, you have to code or to build with a tool first and then you see what’s happening.

Of course, there are little helpers, so you could start with a tool which has a low barrier like raw graphs or Datawrapper where it’s very easy to just produce a simple chart, and you can see if it works. But then when it comes to the custom data vis like the one you sketched, maybe, then you have to build it with an advanced tool like Tableau or with JavaScript indirectly or Illustrator. Then you really have to design by coding, by iterating over the building process again and again.

Then the fourth phase is just publishing. I say “just.” Well, just is a little bit simplified. Publishing means that you collect feedback after all, so you throw your data vis out there. Maybe you publish it in the newspaper. Maybe you just put it on Twitter, maybe on your personal Web page. Maybe you give it to a client.

Then it’s really important to just listen to them and see, well, does it work? Do the stories I had in mind really reach the audience, or do they see completely different stories? That’s really important because usually, you do not stop doing data vis by publishing. You go to the next project and again and again. And so, you collect experience by collecting the feedback.

These are the four phases: Assess, so ask questions about the data. Define the audience. Then plan first, which is about the whole concept, the whole message. Build it and publish it.

In my case, the data visualization about child mortality looked like that in the end. This is an interactive graphic. I want to take you just to my browser here. You can see that when you hover over these circles, we can see highlights. But let me, first of all, describe the structure a little bit.

We can see the countries here around, and they are ordered by continent. Then we get the three years, 2018. Let me just enlarge that a little bit that you can see the labeling. We can see here 1998, 2008, 2018, and the size of the circles.

There is also a legend here up there represents the mortality rate. Then you can see that, for example, here for Africa, this is where I’m hovering.

Okay. I need to go back to the original zoom. Here where I’m hovering, this is Africa. They are sorted by child mortality rates in 1998.

One story now could be that we can really see how the different countries in Africa performed now over 20 years until 2018. We can, for example, see that in 2018, which is the outer circle, the data is not ordered anymore, so the countries really performed differently when it comes to the development of the child mortality rates.

For example, there is one country, Rwanda, who performed really good who has really low child mortality rate but started with a very high one. Whereas other countries directly next to them performed not as good as Rwanda.

So, there are multiple stories. There’s also this outer ring. I would encourage you, by going to my website, higsch.com, to just have a look on your own and try how visualization works. Yeah.

This is how one could visualize child mortality rates. You’ve seen that I started out with the data and then I built the story around, so I focus on the continents, so I showed that there are countries that have rising mortality rates and declining mortality rates. It would be the outer circle. You’ll see that when you try it out, and so on. I’m really curious to know which stories you discover when you go to that data visualization.

But this data visualization has a little problem. It does not convey emotions, and that would be the outer layer of the whole data vis production story. The emotions, emotions are really, really important, and I’ll show you why.

In the data visualization that I just showed you (and I already highlighted Sierra Leon, for example), in the center, we see the full data set. We can see that the child mortality rate dropped to 105. But can you really imagine what that means? You can you really imagine how many children died?

it's really hard because you also wouldn’t care if it’s 105 or if it’s 95 or 150. There is no feeling to it, and that’s the problem with this specific data visualization. Although the design might be cool. It’s circular, and I don’t know what. It’s built with JavaScript and Svelte, by the way.

But it does not convey the emotion, and this is important. Why? Because of psychic numbing. The more who die, the less we care. So, if you have one person who is dying and you know the individual story, you really are sad, and you are thinking of what is going on there.

But if it’s two persons, three persons, or (if you think of the large conflicts in the world) thousands, hundreds of thousands of people dying, we do not really care about because we have no psychological connection. We have no emotional connection to these statistics.

The statistics is important, but the story behind is also important. We can see that in our real life, every day. For example, you’d go to the New York Times website, and you have a look at the newly reported deaths by day in the U.S.

This is a chart or a screenshot that I made a couple of days. This is really a live data, more or less. We can see that in October, there are still a thousand people dying every day in the United States. Did you know that? I didn’t know that. I assume that many Americans also do not know that.

They are seeing this since months, since one and a half years or so, and psychic numbing. We do not care about these thousands of deaths because we have no emotional connection.

On the other hand, when I bring up an image (also from the New York Times) -- this is an article about Bergamo and the high death rates there in Italy at the beginning of the pandemic -- this picture is conveying so much emotion even though it’s not showing a single dead body. But it’s taken at a cemetery. We all know how it is there. We all know how emotions on a cemetery are, so our experience is now connected with the image and the information which it is conveying.

At the same time, in the foreground, we have a person that felt sick, which is tested for COVID here. This picture is showing a lot of emotion, whereas this chart is almost showing no emotion. Although, this is about thousands of dead people, and this is just about a person in the foreground seen, yeah, feeling sick, and a remembrance ceremony. It’s not even a funeral. It’s a remembrance ceremony in the background. This is something we need to bring into our data visualizations.

We need to bring in emotion to our data visualizations because while it’s cool to see the statistics, so while cool in a sense -- it’s a sad topic, but cool in a sense that we are able to put together these statistics, and these statistics are used to make politics and so on. But without the emotions, we cannot really relate to that really sad topic.

We need the emotions, but it’s quite hard to bring emotions into data visualization. There’s a little trick. We could use data art. We could really use data art, and data art is something which is completely different from generative art.

I guess many of you know what generative art is because you are probably all on Twitter and, usually, our timelines get flooded by generative art, which is art that is programmed but based on randomness, not on data. Data art is really based on data, and then you try to be artistic. You try to bring art design together with the statistics. This can help a lot to bring emotions to our charts. How could it look like?

Earlier this year, I did a project with Journalism is Not a Crime. This is an organization which is located in London. This is run by Iranian journalists that are pressured in their countries, so they are maybe--

There are a lot of Iranian journalists that are in prison because of what they write and say. There are a lot of journalists in exile. There are also journalists who have been killed in custody. They have a very comprehensive list about these journalists, and they also had a little representation of these numbers on the website, like you can see it here.

But the problem was that these numbers, of course, are not conveying any emotion. And so, I got the task to build the data visualization, which attracts readers to them, which really makes readers think about these people and their fates.

I went through my process, and I just want to highlight the inspirational part because I thought, “Well, all these people that are journalists in Iran, they only want to make their country better,” and I’m convinced that they are really proud of their country. Maybe not of the government, but of their country as such and of the people living there.

When I think of Iran or the Persian area, I always think of these special Persian architecture where the architects used the differently colored tiles to build these lovely geometric shapes on buildings. I really love that, and I think this is something which people there are really also proud of. And it’s beautiful, and it has this stark contrast to the actual topic, which is dead journalists, which is Persian journalists, and so on.

I thought that might be really cool to bring together. In fact, for the interactive data visualization that I designed, I used that tile pattern. I built these rosettes where each of the tiles really is one of the journalists.

To show you that a little bit better, let’s switch to my browser again. This one is Journalism is Not a Crime. We can just go to the data visualization, which is called Visualizing Censorship in Iran. Then we get these differently colored rosettes here that are completely inspired by Persian architecture.

Here now you can see by hovering that each of these tiles really is a journalist. For some of them, we have images, so then you directly get a connection, “Okay, this is really a person behind the tile.” We get some short biography and some more information about it.

But at the same time, the users are able to play with the statistics. For example, I could (here with these buttons) change the organization. Right now, I can see, okay, there are journalists. But there are also social media activists in the data set. The coloring is also showing categories in death cases, the investigating institutions. So, for example, here there are a lot of social media activists in purple that are pressured the IRGC Intelligence.

Yeah, now you can, as a reader, just play with the colors. For example, I could just switch the color to represent gender. Now here I see that there are male and female journalists, male and female social media activists.

I could also look at the status. For example, the yellow ones here or the brown ones here, they are killed. Most of them are released from prison, so they went through this whole process already. And, at the same time, if I move back to gender, I can also switch the organization of these rosettes, so the arrangement. For example, I could say, “Well, go by gender.”

Now I have color and gender at the same time. Now I can do the organization by the investigating institution, for example. Then I see here is the Ministry of Intelligence. Here it’s unknown, unfortunately. Here is intelligence and so on.

I could also, of course, do it by occupation again. That was the initial view. I could also go back to the whole view on all of the journalists.

This is good from one perspective from the statistics perspective so that you can see, well, it’s a lot of people in there. I cannot count. Now it’s just more than 800. It’s a lot of people there. But at the same time, there are individual stories hidden behind each of these tiles.

If your audience wants to, your audience can also dive a little bit deeper into these personal stories. Here we have no picture, but we can read now all of the biography.

I can also go back. Let me take one with the picture. Of course, these data sets are not complete for all of the cases because it’s quite hard to build up these databases. You can see all the information about this person and really read the story, which is behind, and dive one level deeper into the whole story.

This is how data visualization should work. It should talk about the statistics because that’s also one part of the truth. But it should also talk about the individual stories that are behind the statistics. This is just one possibility how you could do that.

It could also be that your audience does not want to play. That’s completely possible. Yeah. The question is, what do you do then? Now with that Iranian data visualization about the journalists, really your audience needs to play around, and you need to motivate your audience. If you defined your audience, if you are sure about your audience that they will do it, it’s absolutely fine to do it.

But if you have an audience like we often have on our Der Spiegel website, we do not have an audience that wants to play, usually, because we are a news website and people scroll, scroll, scroll, and see what’s interesting. Click into it. Read the important bits and parts. That’s it. They usually don’t want to play.

But then you can switch to other techniques. There is, for example, the scrollytelling technique that I want to show you next. For that, let me just switch back to my browser. That’s a Spiegel article that we just did a couple of weeks back when the Nobel Laureates of this year, in the beginning of October, were announced. We wanted to show all different Nobel Laureates, yeah, that we had over the last, like, 100+ years.

This again was done by showing the individual stories -- a little bit at least -- plus the statistics. It’s always a nice thing when you combine these two by scrollytelling. For example, here I tried to catch the audience by showing the Nobel Prize winners from this year sorted by category. There is, for example, Benjamin List, who is a German guy, for example. Maybe you’ve heard the news early in October. We just explain here -- and I’m scrolling now -- by these little scrolling boxes.

We tell -- yeah, we tell the story, and we’re just introducing, “Hey, here are these circles. Colors are the prize category.” Yeah, each of these circles is one person.

Then we highlight some of them. Tell the story about that. Tell the story about them. Then we move them into the context of all the Nobel Laureates that we had since 1901. Then we tell the story of the statistics. By keeping the little photos here, you can still see, okay, each of these circles is now a person.

Similar to the Iranian visualization, we can now just say, “Okay, how many prizes were there for the different categories?” We can highlight people, so these laureates had multiple prizes. Then we can start reorganizing the whole visualization again and again. The only thing the user has to do is just scrolling through it.

We can also play with charts here, so this is just a histogram where we showed the age distribution of all the laureates. I just want to mention here that animation is quite important for such visualization which just connects the different slides to each other. Here we show the age distribution by prize category, and then we again go to the individual story and highlight some of these stories: Who were the youngest? Who were the oldest?

Then we show by country, by birthplace, and so on until, in the end, we have the overall view with all the Germans again. This is a nice way how to engage your audience with an emotional data vis because you then connect the individual story with the statistics. I think this is something we need to do more. I think if we do more of that in data visualization, we will see a lot less lonely charts.

The data, the story, the emotions, that’s the receipt for attractive and effective data visualization.

For the story, it’s important to assess, plan, build, and then, of course, also to publish. But to build the story, it’s the assessing, the planning, and the building.

For the emotions, have a look at data art, what is possible with data art.

When you have that in mind, you are able to produce really good data visualizations that are not lonely charts, that are just effective, that use data art, that use the power of emotion to attract the interest of your audience and to convey your message to your audience.

Thank you very much.