German Elections – The Data Vis Explanation


Gosh, the German election system. I still remember learning about it in school. Or rather: I don’t. I was really not interested in politics back then. Of course, I blame my teacher. So every four years I stare at this sheet of paper in the voting booth, trying really hard to remember what it means when I put my cross next to a party’s name.

So what does it mean? What are Germans like me voting for exactly when they’re voting on Sunday? This blog post and an essay spreadsheet I recently created exist solely for the purpose of finding that out. Let’s start (also, you might want to read that blog post on your desktop computer. Sorry.):

We can’t vote for our chancellor directly. Instead, we are deciding on the approx. 680 people who will move into parliament, and who then will vote for the chancellor and tons of other stuff in the next four years. To decide who these people are, I’ll make two votes on Sunday: The First Vote and the Second Vote.

1 Who?

The First Vote is for a person, who is often affiliated with a party and wants to represent my election district in parliament. Germany is currently divided into 299 election districts, which all have roughly the same number of citizens: 280,000 people:


These election districts only become important when they are elections. Most people have no clue about the size of their election districts. And because they only take inhabitants into account, this size can be hugely different. Hamburg with its 1.8 million inhabitants is divided in six election districts. Mecklenburg-Vorpommern, another state in the north, has also six election districts. Although the state of Mecklenburg-Vorpommern is thirty times as big as Hamburg when it comes to size, they both have roughly the same number of inhabitants – and therefore the same number of election districts.


The First Vote in the 299 election districts works after a “The Winner takes it all”-system: Only the candidate who gets the most votes in their district will move into the parliament. That means that the parliament is already half full just with the 299 winners from the election districts.

For instance, in the election district “Hamburg Nord”, a candidate for the Christian Democratic party (CDU) won 39.7% of the votes at the last general election in 2013. The 2nd place went to a candidate from the centre-left social-democratic party SPD, with 34.8% of the votes. The CDU guy won and got a seat in parliament for sure. The SPD guy went home and cried (or maybe he didn’t. I don’t have any information on that, really.)

That’s how it looked like in 2013 in all election districts. The election districts are coloured according to the party of the person who won this district.


It’s pretty black. Black is the color of the CDU, the party of Angela Merkel. When you look at this map, you’re probably close to thinking: “So the CDU won like…what, 80% of the seats? That’s awful! The poor other parties!”

But don’t pity too early! There is hope: The election districts and the First Vote don’t actually determine how much percent of the parliament seats are taken by which party. That’s the job of the Second Vote. Only with the Second Vote we voters decide what fraction of the seats each party gets. The Second Vote is the reason why the Green party won only one of the 299 election district in 2013, but ended up having 13% of the seats in parliament – and why the CDU didn’t win 80% of the seats, but 40%.

2 How many seats? Also, who else?

The Second Vote is for a party; not for a specific person like the First Vote. Or, to be exact: The Second Vote is for a list of candidates from all parties in my state. In each state in which a party wants to be voted for, the party sends a list of candidates to the election administrator of this state and says: “Please put our party on the ballot in this state.” When I vote for a party with my Second Vote, I actually vote for this state list of the party.

So how do we get from these votes to the seats? Well, in each state, each party can earn a specific amount of seats. The number of seats per state is determined by the number of inhabitants, similar to the election districts. The German election administrator just takes the 598 seats that exist in Parliament and allocates them to the states: Hamburg gets 13 seats, Berlin gets 24 seats, etc.


The more voters in a state vote for a specific party, the more state seats this party gets. The Christian Democratic party won 32% of the votes in Hamburg in 2013, so it gets 32% of the 13 seats for this state (which is…um…5 people 1 ).


But who are these 5 people? Here, it all comes together. Remember the First Vote? And the guy who won an election district in Hamburg for the CDU? This guy gets one of these 5 seats. That’s the interesting thing: All candidates who win their election districts thanks to the First Vote, are part of the seats which a party wins thanks to the Second Votes.


Sooo…that means that the CDU has only 4 seats in Hamburg left to fill with their people. And here, the list that the CDU gave the election administrator earlier makes its appearance. The 4 remaining seats are filled with the first four candidates that are on the CDU-Hamburg-list 2. The fifth person (and sixth person, etc.) is out. Meaning, when a party wants to make sure that the party superstar gets a seat in parliament, they put her at the top of the list – then, if the party gets only one seat in that state, at least the superstar is in.


3 All the seats, in all the states

Here’s a quick summary of what you’ve just learned: Germans have two votes. The First Vote answers the question who will get a parliament seat to represent my election district. The Second Vote answers two questions: How many people of each party will be in parliament for my state? And: Besides the people who got elected directly with the First Vote, who else will get into parliament for each party?

Let’s zoom out. That’s how all the won seats looked like for every single state at the last election:


Here are the most obvious things this overview can show us about the German election in 2013 (these things were true-ish before and will also be true-ish this election):

1 More than any party, the CDU/CSU got almost all their votes in form of election district votes. Only 25% of their seats come from state lists. (In comparison: The Greens had to fill 98.4% of their seats with state list candidates; since only one guy won an election district for them.)
2 The Left party is far stronger in former East Germany than in the former West German states. That’s mostly because it has their roots there: The Left party developed out of an old East German party.
3 The SPD, the Left and the Greens won far more election districts in cities than in rural areas. These three parties won 70% of the election districts in the three city states Hamburg, Berlin and Bremen; but they won only 18% of the election districts in the rest of the states.

That’s it! Do I feel better about voting on Sunday now? Heck yeah. Here is the ballot again, this time translated into German. If you still don’t have a clue about what it all means, try this very nice visual explainer by the Bloomberg peeps. And please send me an email so I can try to explain it better: Thanks!



  1. Here is where it gets a bit tricky. Actually, 32% of 13 seats are 4.2 seats, which is nowhere near 5 seats. But like most countries, Germany has an election threshold. That means that parties don’t get into parliament when they don’t get 5% of all german-wide Second Votes. In Hamburg, people voted a lot for these smaller parties that didn’t end up getting into parliament. In fact, only 86% of the people voted for parties that did end up getting into parliament. That means, we calculate the 32% of this 86% “big-party-voters” – which gives us 37.2% “effective” votes for the CDU-party. And 37.2% of 13 seats are 4.8 seats…which get rounded up to 5 seats. 

  2. Sigh. I know. I totally left out that very inconvenient but probable case that a party has more direct candidates in a state than it got seats thanks to the Second Vote in that state. These extra seats are called Overhang Mandates and are a pain. If you like pain, you’re welcome to learn everything about them in this spreadsheet essay I built. 

My Why and How of Blogging


My blog used to have blog posts; back then, in the days, when I was still writing posts more often then every three months (yes, my last post is from May). I will pretend it’s a summer break. To celebrate the slow decline of this once lively blog, I didn’t hesitate when Designfeast asked me to tell them why and how I created such a lively blog. They had questions. I had answers. Here is the original post on Designfeast, and here is a copy:

1 Why did you create a web site of regular entries?

A Curiosity. I want to think about things. Questions like: Why is that? Should it be like this? What happens when? Also, I want to think about many different things and don’t know how they’re all connected yet. That’s why I’m writing blog posts and not a book…yet…

B Clearer thoughts. Writing helps me to structure my own ideas and to think through arguments. Does my super genius opinion about how stuff should work actually make sense? Every time I need to translate my blurry thoughts into concrete words, they get their first reality test.

C Feedback. Almost every time I push something in the world, I get something back. People build on my ideas, coming from their own perspectives. Or they argue against my ideas out of reasons I haven’t thought about. Both is beautiful and makes me grateful: It makes me humble to look at comments to my articles. Writing my blog definitely made me held my beliefs less tightly and keeps reminding me of the complexity of this world.

D Giving back and pushing forward. The world gives me a lot of ideas, so I want to return the favor. I do believe in pushing a field (data vis/data journalism in my case) forward together; in figuring things out together; in constructive arguments and collaborations. It’s beautiful to see that happening over many years and across many countries; seeing people come and go and get excited and change the mindset of people in the field, baby step by baby step, with every blog post they write and every talk they give. I want to be one of these people.


2 What web-based solution did you select and why?

Back in the days, I’ve used Tumblr and I enjoyed its convenience. But now I use Jekyll, because convenience is overrated…or at least not as important for me as the following three reasons:
A It’s simple. It doesn’t have tons of features I don’t need.
B I’m in full control of all the files that build the site, and can actually understand what’s going on.
C Jekyll forces me to write in Markdown and to use Github, and I wanted to get to know both technologies better.

3 What is your definition of a good blog and what are three good blogs that you frequently visit?

A good blog inspires me on a periodical basis. I don’t think that I can give a more detailed definition, since what happens within the limits of it can be quite different:

Tim Urban’s Wait buy why is definitely at the top of the list. This blog educates me deeply about things, it builds arguments beautifully, makes me change my mind and supports the concepts it explains visually. And boy, I loooove visual concept explanations.

Nathan Yau’s FlowingData and Andy Kirk’s Visualising Data need to be named as blogs that keep me and thousands of other data vis enthusiasts informed about what’s happening in the scene. I especially enjoy the posts that offer an opinion about the quality of a data vis work, instead of just stating that it exists.

Brandon Stanton’s Humans of New York, because it opens my eyes for the situation outside of my privileged filter bubble and lets me understand how people got to where they are right now.

4 How do you create content for your blog?

The pipeline from “quick idea” to “tweet-able blog post” is long and wonderful and distressing. The perfect path would look like that:


1 Ideas. Often, ideas come out of conversations that I keep thinking about, out of an own pain point or out of a question.

2 Research. Once I’m curious about things, I do research about it. The time for that ranges from a minute to a week of googling, reading books and scientific papers. Indeed, I discovered my love for reading papers! Once you want to answer a very specific question, searching for the most precise answer is tons of fun. Like my little brother, I just keep asking “But why? How?” until my curiosity is satisfied.

3 Structuring the post. When I’m happy with the information I gathered, I try to communicate it in the best possible structure. I use a tool called “Workflowy” for that. It lets me move around my thoughts until I find a flow that doesn’t make me want to cry anymore. To understand causes and effect better, I often visualise them with pen and paper in the process.

4 Writing. Once I have the structure and decided what I want to communicate, the task of actually writing the post becomes far less daunting. I write in markdown, using an editor like iA Writer, Sublime or Atom.

5 Add images to the post. I’m a visual thinker, and images will result out of the process of writing a post anyway. Often, I include these illustrations into my articles after refining them with Adobe Illustrator and a drawing tablet.

6 Publishing. After writing the article, I publish it to my blog. I read it again there, find tons of mistakes, fix it, republish it, formulate a tweet, proofread it, take a deep breath, send the tweet, read the article again, find more mistakes, fix it, republish it an x-th time, and then distract myself from looking at my Twitter notifications with food or so.


5 How do you stay organized and motivated to contribute to your blog?

One answer is: I don’t. I aim for one article a month, but if that doesn’t happen, then that’s ok with me. Another answer is: I give talks at conferences! I like to submit talk proposals about things I’m curious — but have no idea about. Then I have a deadline to do the research. Posting a transcript of my talk is the easy part at the end.

6 For those aspiring to make a web site composed of regular thoughts and/or images, what is your advice?

Get a second blog.

“Whaaat,” I hear you gasping, “I can’t even fill one”.

I hear your gasps. But think about it: Why is it that you can’t fill your blog? Maybe because you think your thoughts are not worth to be published. Or maybe because you posted that one crazy good article two months ago and you can’t think of anything better. So get a second blog. A trash-blog. For the bad articles. Just start writing them. Several of my articles were born on my secret trash-blog. Works really well for me.

7 What is your quest in blogging?

To teach and to learn.

Why You Don't Believe In Facts, And How To Fix It


This is the transcript of a talk I gave at Re:publica in Berlin at the 9th of May 2017. You can find all the slides in that insanely fast GIF up there as PDFs on Github. You can also find a video recording of this talk on YouTube.

Hi! This is me, and every single one of you:


We’re standing on the ground of hard facts. (“Der Boden der Tatsachen”, as Germans would say.) Well, ok, there are no hard facts; let me rephrase that: We’re standing on the ground of concepts with different levels of certainty; concepts entangled by confidence intervals. For example, it is a kind-of-certain fact that approximately 97% of climate experts agree that humans are causing global warming.


When we’re accepting a “fact”, we start to believe in it. Eg we can believe that global warming is real. Beliefs are important: They let us marry people, let us found companies and research projects, and let us go to church every Sunday. And beliefs are the basis of every ethical discussion: We need to agree on what’s happening before we can discuss what we should do.


But it seems like in more and more cases, we don’t have a discussion of what we should do – but what is true in the first place. What is really happening? What is truth? What is the reality? What are the facts? It almost seems like some people stand on the ground of a completely different reality; sharing completely different facts. But of course, that’s not the case. The reality is still the same, pretty much.


What I’m currently doing with my life, is visualising the facts. I’m visualising data, often from the Census or Scientific Studies. And then I’m communicating it to the public and say “Here, look, these are the facts”; let’s have a discussion on these grounds. So when people don’t take them seriously, I have a problem. I’m not doing my job well.


Today I want to explore how I can do my job better. Am I just preaching to the choir with my data vis? Or is it possible to convince somebody with facts, numbers, science, reason?

To do so, I will explain first where false beliefs actually come from. Why don’t we all believe in the same? Then I want to explain why it’s so hard to get rid of false beliefs once they’re planted inside of you. And then we’ll use this knowledge to explore what we can do to believe more certain things – we in this room, and I as a data vis person.

A. Where false beliefs come from.

Let’s start with talking about the reasons for false beliefs. Of course there’s only one:


Fake News is something a lot of people have talked about. (One of the best things, in my opinion, can be found on First Draft News.) I actually think that sharing fabricated content and misinformation on social media is only a minor cause of having false beliefs, but a major symptom.


One of the actual big reasons is tribalization. We’re social animals living in different tribes (our family, our friend groups, our favourite online forum, our Twitter feed). And for social animals, social support is more important than believing the truth about something that’s far, far away like climate change.

In the first instance, beliefs can become a marker of identity. We belief what our family and our friends told us. When we share articles on Facebook, we want to demonstrate who we truly are. We want to define ourselves. “Defining” literally means “to limit, to bound”: We draw boundaries around our identity, also to attract like-minded people.


Showing who we are is especially important to demonstrate whose team we are on. We want to show the other tribe members (friends, forum, party, church) that we are good members of the tribe.


We could see that especially well in and after the election. When confronted with the photo of Obama’s inauguration from 2009 and Trump’s inauguration from 2017 and asked which photo had more people, lots of Trump supporters gave the wrong answer. They said that Trump’s inauguration hosted more people. Not because their perception was flawed. But because it was a way to show their support for Trump. It’s not truth vs lie, it’s their team vs the other team.


But showing off your beliefs with information doesn’t just gain us the trust of a team. It’s also a test to see who WE can trust. These people who unfriended us after we posted a long essay about how much we hate Trump? Yeah, it’s good they’re gone. We’ll definitely not drop by their birthday party this year.


Sadly, that happens more and more in the two-party-system in the US. It’s one thing that less and less government people agree with members of the other party:


It gets disturbing when you hear that 58% of Republicans and 55% of Democrats had a very unfavourable attitude towards the opposite party in 2016; up from 21%/17% in 1994. Less Democrats marry Republicans and the other way round. People of the two parties don’t trust each other as much as they did only a few years ago.

Interesting enough, for forming beliefs for the sake of tribalization, it doesn’t matter if we understand what we believe. They are too far away from us. Ask yourself: How does climate change work…exactly? Could you explain it? I know I couldn’t. But people from my tribes and I believe in it, and we know someone is from another tribe if she or he doesn’t.

The third reason for forming false beliefs I want to talk about are fallacies. This could be a talk on its own, so I will just mention one example: Generalising from anecdotal evidence. Which means seeing one instance and applying that to the whole concept. Saying “Man, it’s so cold and rainy for May. Global warming, yeah right” for example:

image image

B. Why it is so hard to get rid of them.

So once we got a false belief, we just keep it until somebody comes and says “Dude, that’s not right”, and then we get rid of it – right? Unfortunately not. We defend many beliefs like crazy.

Since we get so many beliefs from our tribes, questioning the belief means questioning the tribe. Beliefs are connected to the person who gave it to us, the situation we’re in, the memories, our upbringing etc. If a belief turns out to be untrue, then all of this is in doubt.


But there’s no need to get rid of the beliefs anyway. One quick Google search tells us that there is tons of evidence for our belief, and our tribe confirms us on a regular basis. And if one online forum tries to convince us of the opposite, we can just leave and can go to another one. The internet is truly a paradise for finding confirmation.

image image

Once we have a belief, we perceive information differently. Information that confirms our beliefs we agree easily with; we question it less and remember it better. The opposite of this Confirmation Bias is Motivated Skepticism: When we get confronted with information that oppose our beliefs, we apply more skepticism than normal. Everything seems untrustworthy suddenly: The author, the sources, the medium.


An example for that can be found in the following article, which explains why you should vaccinate your kids. As somebody who agrees with the premise, I also don’t question the sources: “International studies”, it says. Sure. And there’s this one study that basically explains that you have less respiratory problems when you don’t have the flu? Sounds about right:


Ok, fine. I’ll admit, I changed the original article a bit. The actual article can be found here and speaks against vaccines.

And suddenly every second word makes me angry and suspicious at the same time. “International studies”? Yeah right. I know which kind of studies THESE ones are. Certainly not peer-reviewed, I bet. And this journal called “Clinical Infectious Diseases”, they mention? I’m sure it has a super bad image in the community. I wouldn’t trust one article published in there. And Well, I’ll certainly unfriend everyone immediately who mentions this site on Facebook.


Confirmation Bias and Motivated Skepticism don’t just make us perceive information differently, but also strengthen our beliefs – in both directions. Consuming lots of confirming information (more investment) nourishes my belief:


It’s supposed to work also the other way round: If you show me information opposing my belief, my belief will become stronger. You might have heard of it as the Backfire Effect, but it’s not well proven scientifically. I dare you to find a good paper and I’ll change this blog post.

Close-mindedness is the last reason I’ll mention today. Some of us are it more, some are it less: We are different people, and some of us accept opposing information better than others.


I think it’s important to keep that in mind, to not give up on people (“Whyyyyy the heck do you not accept the truth?”).

C. How to believe more true things.

So after hearing about the obstacles that keep us from believing more true things…how can we still attempt to do so? The most obvious idea is to convince someone, brute force: “Hey, you believe x, but that’s not true. You should believe y.”


Often, that’s neither the most successful nor the most kind attempt. Let’s maybe not do that.


Instead, we can try to make them see the truth themselves. We can ask them questions (like Socrates back then): Why do you believe what you believe? How did you come to that conclusion? What would be the consequence of your thinking?

And we can point out the difference between the facts and their beliefs in a subtle way.


The closest data visualisation I’ve seen to this approach is Bloomberg’s “What’s Really Warming the World?”. It leads through different factors that “Climate Deniers” see as reasons for climate change: Is it sun activity? Deforestation? Ozone Pollution? No, it really is greenhouse gases.


The next idea, actively avoiding tribalization might be the most powerful approach against false beliefs, especially when we believe that tribalization is one of the main reasons in the first place. The easiest way to avoid tribalization is to find people with different opinions, and then to actively listen to what they have to say. Why do they believe the things they believe in? What are their rational reasons?


The German ZEIT Online is attempting conversations like that. Right now they have a call for people who want to meet different-minded people in their neighbourhood, on one afternoon in June.

During such conversations we can try to build understanding and empathy, which might let us discover some hidden truths and emotional reasons:


The third way to make us believe more true things is to change our attitude; towards ourselves and towards others. This attitude shift can happen in different ways.

We need to accept that there are millions of facts we don’t know, that not knowing things is normal (not devastating); that changing beliefs, therefore, is necessary to grow. Which means that having beliefs as marker of identity is not the most marvellous idea. Our beliefs come and go to help us act in this world; but we are not defined by them:


A great idea to realise this attitude is to get people in a truth-seeking mode. The excellent paper “Science Curiosity and Political Information Processing” by Kahan et al states that “individuals can use their reason for two ends—to form beliefs that evince who they are, and to form beliefs that are consistent with the best available scientific evidence.”

The paper show that people who are more intelligent are more likely to have extreme beliefs (true and false), which seems to be bad news. But it also shows that people who are more curious and driven by scientific thinking tend to believe more true things:


So we should make people more curious! Quizzes and games are a great way to do so; and of course I need to mention “You Draw It” from the New York Times Graphics team here. What a great coincidence that I wrote about curiosity in a blog post last month, where I explained the beauty of this very graphic.

I can also think of three more “changes in attitude” that are closely related to data vis: Showing complexity (including uncertainty), and planting doubt. The latter means explaining that the world is not as simply measured as we sometimes assume it is.

But let’s focus on showing complexity. I would assume that most data visualisations out there are bar charts. Which brings two problems with it: First, every kind of data looks the same; and second, it simplifies extremely.

And hey, that’s good. Simplifying is our job, among many things. But more and more often I appreciate approaches like the one from Accurat, who show the complexity of the data out there better than any bar charts. The form gives a more honest view of the world. Our beliefs are already simplified enough: Forms that remind us that the facts are not that simple could become more and more important.


The last approach to believe more true things is to remind ourselves of our skin in the game. Having skin in the game means to be directly influenced by the outcomes of something; in this case our beliefs. We have skin in the game when we think that it doesn’t rain, but it does – because we’ll get wet. We get immediate feedback.


But we don’t have skin in the game with most tribe-defining beliefs like climate change, pizza gate, voter fraud. We are not effected by the outcome:


Remember that awesome “Science Curiosity and Political Information Processing”-paper? It states that “Farmers … have been observed to use information on climate change to form identity-congruent beliefs when they are behaving as citizens but to form truth-convergent ones when they are engaging in the task of farming, where they have an end—succeeding as farmers—that can be satisfied only with that form of information processing”. Which basically means: When you ask a Republican farmer if he believes in climate change, he will probably say no – until this belief gets important for his goals to be a better farmer. Reminding ourselves and others of the skin in the game and the personal ways they can benefit from information will make them listen.

D. What does it all mean?

Wow, you’ve made it! That was one long article. Here are all the points we’ve encountered while you were reading this article; and believe me, each category could be filled with far more reasons, causes and examples:


But what should we do now, after having all these information? Is it a good option to not believe in anything anymore?

I’d say no. I disagree with people who say we need to build beliefs carefully. Beliefs are important for acting in this world – and I’d rather want that we act than not act. In my opinion, we should have grand beliefs. But these beliefs show be flexible like a gummy bear: image

Strong opinions, loosely held. That’s a phrase I really like and that I’d like to apply to beliefs as well: Strong beliefs, loosely held. We want big beliefs to act. But when we encounter new facts, we want them to update easily.


Thank you! If you have thoughts, new categorisations and critique, please email me: ( or look me up me on Twitter (@lisacrost).


Embracing Curiosity


Curiosity. Some people have it a lot, some have it a little. All the data (vis) people I know have tons of it. “I wonder if there’s data out there for it..”, “We could also try to find out…”, “Maybe, if we can prove with data that…” are common phrases you can hear wherever data people gather.

That’s why more than two years ago, I declared curiosity as one of the four core traits of data vis designer:


But although I love the feeling of curiosity while working with data myself, I’ve never considered the curiosity of the data vis reader. That is, until I encountered Tim Harford’s article “The Problem With Facts” a few weeks ago. He cites a paper that studied the effect of curiosity on reasoning:


Robin Kwong transformed this idea in his article “How do we create a need for news?”:


Help people to explore the world? Help them to learn about it? Not giving final answers to people, but making them ask questions? That seems like a goal worth pursuing for journalism.

But I’m conflicted.

On the one hand: Yes, curiosity is amazing. Why? Often, curiosity is directed towards something very specific. It comes with a concrete goal: Finding out what happens when you jump in that puddle. Finding out if your best friend has a new boyfriend. Finding out if there’s a regional pattern with unemployment rates in your country.

Having these intrinsic motivations and acting to get there is one of the best feelings in the world. It’s what people talk about when they talk about “Flow”. You forget the time. You feel productive. Your work feels important and meaningful. There’s just you and the goal. And everything you do is to reduce the obstacles between you and the goal. And when you reach the goal, a mini-Christmas is happening.


I want readers to get that feeling when they interact with my data visualisations. I want them to feel curious about the data. I want them to ask questions, and to find answers. I want them to feel empowered, and to really grasp the data.

On the other hand, I want to make sure that they get the most important statements in my data vis. I want to save them the time to figure out these points on their own. They shouldn’t learn how to use a completely new data vis tool, just to get to an underwhelming point after half a day of digging. My job is to spend all that time with the data for them, so that they don’t need to spend that time themselves. I want to deliver answers to them, fast.

A compromise is needed. I think I need to say good bye to the idea that the most efficient data communication is always the best way to go. For people to care about my data, I need to make them curious, first – to the right degree:


There is a sweet spot between my current “get the reader the information the fastest, even if that means it’s a bar chart”-mindset and the too-open field of wonder in which the reader will quickly loose interest; especially if he doesn’t know if it’s worth it.

That sweet spot has been hit by many people in the past. Explorable Explanations have been doing that really, really well. And in the news graphic realm, the New York Times did an amazing job with their “You Draw It” graphic:


It’s a quizzing tool that plants a question in the reader, before showing her the information. The reader needs to build a hypothesis first. Then she can check if her belief was right or wrong. It’s science in a mini-format.

But it’s science that’s rewarded. The reader trusts the NYT that it will be worth it to work for the answer; that the question the reader asks will lead to an interesting answer. Not every data is worthy of such a treatment. In 2011, Kaiser Fung introduced the metric “return on effort” for evaluating graphics. He asked: How much will the reader be awarded for her effort to read the graphic? The reader will be annoyed to work for her insight, if the insight is disappointing. We need to adjust our work to take that into account:


When we first make a reader curious and then deliver the answer, we can have a greater impact than when we just deliver the answer. The question is: What are good ways to make people curious? What are ways to plant questions in people’s head? How can we make people curious about the world, even the ones who don’t score high on curiosity naturally?

And to think one step ahead: Can we, instead of giving people answers, or first plant questions and then answers – can we give people the questions without the answers? Questions we don’t have any answers for yet? How can we empower people, and make them interested in trying to solve the yet-to-be-solved problems?

Curiosity is like a superpower of people. When curiosity is fired up, people seek out information actively, and suck it up like a sponge. Let’s make sure we understand and use that power.

Please help me to answer these questions. Write me an email ( or find me on Twitter (@lisacrost).

Why Do We Visualise Data?


This is the transcript of a talk I gave at INCH in Munich at the 10th of March 2017. You can find all the slides in that insanely fast GIF up there and as PDFs on Github.

Finding a definition for data vis is ok-ish easy: Data vis represents data with visual elements to communicate information. Today I want to focus on a part of the data vis definition that is a bit overlooked: The part that tries to answer my favourite question of them all: “Why?”

So yes, why. Why the heck are we actually visualising data? I want to answer that question in five parts:

1 Data vis goals: the overview
2 Different industries have different goals.
3 Different data visualizer have different goals.
4 Goals influence each other.
5 What are your goals?

Data vis goals: the overview

In this first part, I will present my mental model of possible data vis goals; sorted into three categories. Feel free to skim that part! The first category, Attention & Beauty, includes goals that relate to the quick first impression of a data vis project. After the first initial impression, we move to goals that relate to the decoding of data: Understanding. And the last category, Implication, contains the sense-making and long-term impressions of a data vis piece.


(These three categories are nothing new. Francesco Franchi said that data visualisations can “Inspire & Entertain, Inform, Encourage” and Andy Kirk stated that the user of a data vis needs to “perceive, interpret/decode, comprehend”. That’s both awfully close to my categories of “Attention & Beauty, Understanding, Implication”.)


Attention & Beauty: 1. Get them to read: Data vis is still considered to be new and shiny and is often used to attract some eyeballs. Like the ads on Times Square, some designs out there try really, really hard to get your attention. An example is the typical colourful long-format infodesign blogpost. But every magazine will tell you that infographics work wonderfully to get a reader’s interest after pages of long text. That’s related to the 2nd goal:

Attention & Beauty: 2. Attract different learning types: Different people absorb information differently. It’s like kids who choose different toys on a playground. Infographics are amazing for visual learners. An example is Vox’ interview with Hillary Clinton. Here, the reader can get one and the same bit of information via audio (in the video), via the transcript or via the graphics.

Attention & Beauty: 3. Go with the hype: Have I mentioned that data vis is still considered to be new and shiny? And people love shiny stuff. Even if there is no rational need for data vis, many people feel that they need it for their business/article/publication. Just because everybody else has it. Sometimes, that’s not a bad thing. “Going with the hype” also includes copying successful and fancy visualisation formats. Remember NYT’s Snowfall from 2012? Every publication copied it; and made digital storytelling better in the process.

Attention & Beauty: 4. Create beauty: We can’t deny it. Data vis can look beautiful, at least according to a famous info design studio and a famous data vis subreddit. Good examples can be found in the tweets from @accidental__aRt: We have no idea what data is displayed, but it’s still interesting to look at. Not because of the content, but because of the form.


Understanding: 1. Understand the data: Before we can communicate something to the outside, we need to understand it ourselves. When I sit in front of R Studio I feel like in a lab, forming hypotheses and testing them. Visualisations help data scientists to make sense of the data; and they help quantified-self enthusiasts to make sense of themselves.

Understanding: 2. Explain: Once we’ve understood the data, we can teach it. A schoolroom is a good metaphor for that process; and data journalism is a good example. If you want to find good explainers, just look at every Washington Post graphic for the US election 2016. Man, they did some great work.

Understanding: 3. Explore: But sometimes, we can’t or don’t want to go as far as explaining the data. Sometimes we want our users and readers to explore the data for themselves. In this case, we build tools for them that make it as easy as possible to get the information they’re looking for; like libraries do (you know, the ones with books; not your Javascript ones). Dashboards are the by far most prominent examples in this category.


Implication: 1. Prove: In some cases, making people understand something is not enough. You want to convince them of something (hopefully The Truth™), like a lawyer want to convince the judge of the innocence of her client. Scientists are giving proofs with data and data vis all the time; and I’m always happy when I see journalists using the scientific toolbox. An example is the FT article that shows the correlations between Brexit votes and demographics.

Implication: 2. Correct world views: We are not rational. Although it would often be more useful to act based on The Truth™, we act as a consequence of blurry feelings and beliefs. To counteract them with data could change the wrong and harmful world views of (some)(open-minded) people. A great example is Max Roser’s Our World In Data.

Implication: 3. Evoke feelings: Data vis is probably not the preferred medium to evoke feelings (video, photo, illustration; heck, even text is better at that), but you might find yourself in the position where you’re looking to get an emotional response about your data from the user/reader. Data vis projects that achieve that are “U.S. Gun Deaths” by Periscopic or “Fallen” by Neil Halloran.

Implication: 4. Evoke actions: Once you got your reader to feel something, you often don’t want to stop there. You want them to do something: donate, participate, etc. To be honest, I can’t think of any data visualisation that tries to convince me to do something specific. Let me know if you got an example!

Implication: 5. Go meta: Data defines our lives in many ways. Data vis can show that and can make us think about how data is collected, stored or analyst or used for decisions. Data art can achieve that in its extremes; but “normal” data vis can show a taste of that. Giorgia Lupi, co-founder of the studio Accurat has been an advocate for that for years. In 2015, she stated that Accurat’s visualisations want to “embrace complexity” and “maintain the nuances of the data”.

Different industries have different goals.

After going through all the goals, we will now have a closer look at the data vis goals of certain industries.

Let’s start with business slideshows: They have very little emphasis on goals of the first category, Attention & Beauty: Since the data is often crucial to the user, she will actively try to understand it and doesn’t need to get convinced to do so. image

The opposite is true for Data Art: Good data art looks appealing. But it also has a huge focus on sense-making in a deeper way (“Implication”). image

Bad” Data art often offers the same (kind of) beauty, but without the implication. It serves as decorative illustration in this role. It’s still worth existing, e.g. for cover designs of business reports. image

And then there is Data Journalism. Its biggest focus in on understanding. It needs to attract readers and as we’ve seen, wants to change people’s mind; but most often, that’s not its main priorities. image

Different data visualizer have different goals.

We all get into Data vis because of slightly different and also because of slightly similar reasons. Personally, my main motivations to do data vis are to create something beautiful, to make myself and other people understand something, to offer an overview and to change people’s beliefs. That is very close to the goals of data vis for journalism, and I do feel very home in that field.


Why is it important that personal goals align with field goals? Because goals define priorities and how much time you spend with which task. If your job is to make data beautiful, but you don’t care about beauty at all, you’re in the wrong job.

Another way to think about goals and priorities is to ask: What do you sacrifice for what? If your main priority is beauty, you will sacrifice understanding for a fancy new visualisation form. For me, understanding is more important than beauty. I’m somebody who says: “Ok, let’s show the boring bar chart instead of the multi-dimensional slope chart, since nobody seems to understand the latter one.” But beauty AND understanding are definitely more worth than just understanding; and I’m willing to put a lot of time and effort into beauty.

Goals influence each other.

Goals/outcomes are interconnected. Some influence each other positively, some influence each other negatively. For example, if your goal is to make your data visualisation beautiful, you’re more likely to have evoked emotions as an outcome than with a boring bar chart. Evoking emotions is a side-effect of beauty.

But HOW goals influence each other is up to discussion. Different data visualizer have different opinions about how the goals of the three categories are connected. As an example, I want to talk a bit more about the relationship between beauty and understanding.

Moritz Stefaner, the “truth and beauty operator”, sees a tight connection between truth (understanding) and beauty in Alberto Cairo’s book The Functional Art: “Truth and beauty have always been two sides of the same coin. […] For Buckminster Fuller, beauty was an indicator of truth.”

Francesco Franchi argues the opposite in his book: “While creating an infographic, one moves within a spectrum: on one hand, there is aesthetic, and on the other, the information.”


Who’s right? Are understanding and beauty two extremes that couldn’t be further away; or extremely closely related?

Both are right. Yes, bringing more attention to the style can lead to more understanding. Just remember Edward Tufte’s Data-Ink-Ratio.


But that’s only true to a certain point, in my experience. If the focus on beauty is too high, the effect inverts: Understanding goes down. That’s why data art (with a high focus on Beauty) often has a low focus on understanding; and that’s why data vis with understanding as its highest priority can only be beautiful to a certain degree.


This example shows how goals can positively and negatively influence each other. But it also shows how opinions about goals can differ. I wonder what you all think about the relationship between beauty, understanding and implication. Let me know!

What are your goals?


I want to make you think about your goals when you visualise data. What are your personal priorities? What are the priorities of your industry or company you work at? Where do you stand between beauty, understanding and implication? And where do you want to stay?

Please write me an email with your thoughts, goals, new categorisations and critique ( or look me up me on Twitter (@lisacrost).