As data editor for the Guardian US, Mona Chalabi contextualizes big numbers with her signature illustrated — and often provocative — data visualizations. But her end goal isn’t just a shareable image; it’s to make sure readers understand the big picture. In this talk, Chalabi shares her approach to storytelling with data, including:
- What common mistakes to avoid when presenting data
- Why data visualization can never be totally subjective
- How repetition and surprise can be wielded to emphasize important information
So my job is the data editor at the Guardian US. And it’s a job that involves basically finding numbers in all kinds of different places, kind of buried in the appendix of a pdf or in an academic piece, and then translating those numbers into either written pieces or charts and graphs. And no surprises, I’m going to be talking about the charts and graphs today. And more specifically, I want to talk to you about a problem that we in journalism face. And it’s a problem where we have bounces. Now, some of you might not know…we’re just illustrating it now. And some of you might know it, but a lot of you in the room have been bounces at some point in your lives. You may have even bounced this morning on your way to work. And basically, what bouncing means is it’s arriving at a webpage on our site and thinking ‘no’, and quickly leaving.
It means you didn’t find the information that you came for and that’s really worrying us as journalists. It means that we messed up. Now, each organization kind of measures bounce a little bit differently, but at the Guardian we consider you a bouncer if you were on the page for 10 seconds or less. So for all of the types that I’m going to show you today, I’m going to show you them for 10 seconds or less and see if, hopefully you managed to understand the information that was within them. But before I do that, I want to talk to you about the reasons why people bounce when it comes to charts. Like basically what constitutes a bad chart.
And I can only speak for myself here. I am going to bounce if you show me a chart with no labels. I’m going to bounce if you show me a chart with too many labels. I’m going to bounce if you show me a chart with the wrong labels. Don’t say a third and then show me something other than a third. I’m going to bounce if your labels are too small. And I’m going to bounce if you…I’m gonna bounce…I’m just gonna bounce. No explanation needed. So what is the solution to all of this? Well, before we kind of talk about the solutions, I want to talk about the classic chart types that people like me use. These are kind of like my bread and butter. They’re the things that I come back to over and over again to kind of create these visuals. But I think there’s a real problem with some of these kind of classic depictions. For one thing, I think that charts like this kind of give this false veneer of like perfect scientific objectivity.
They make it seem like the data is so pure and precise and that’s not the truth of data. I often talk about polling just because it’s obviously been a huge deal in this country, like the last election and stuff. And to illustrate one of the problems of polling, I’ll quickly poll the room and ask anyone here who is cheating on their current partner to raised their hand. Yup. You get a similar issue when you ask about how many people have third nipples. And whether or not they plan on voting for Donald Trump, which is like a third nipple, I guess in some ways. It’s a private affair and it’s something that people don’t necessarily want to share. But by the time that the data gets embedded in a visualization like this, people have forgotten all of that kind of ambiguity and that messiness.
There’s another problem with these charts, which is that if you were to remove the labels, they can be about anything whatsoever. Right? So they’re also not memorable because they don’t connect the subject matter with the visualization itself. And my work strives to kind of change that. So, I try to connect the subject matter with the kind of depiction or the visualization itself. So I’ll walk you through a couple of them. And very upfront, I’m trying to write about really, really serious and important stuff here. So like after Hurricane Katrina, if you’re trying to show rainfall levels, instead of just sort of like a classic bar chart, why not use for scale a president who is six foot two? And if you’re talking about an economy that’s diving, that’s kind of in free fall, why not show a diver?
And if you’re talking about hangovers, you can show someone vomiting. Yeah, it spikes Saturday and Sunday. Also, I just find the search time really hilarious. There is no cure. But yeah. And if you’re going to be talking about the KKK, then show the KKK. And like, I think images like this are kind of uncomfortable and I kind of want them to make people feel a little bit uncomfortable. So, another problem I think with a lot of visualizations is that they demand too much of the reader. So I don’t know if any of you guys have done this where you’ve kind of like landed on a page that’s an interactive and you just have no idea where to click first. It’s basically asking you guys to find the story instead of doing our job as journalists, which is giving you a story.
So this is an example. This is on NYC data and it’s all of the dog names in the city. It’s like a really fascinating data set, right? But to interact with it, you kind of have to hover on the circles or else you’re kind of like entering a name. Obviously you’re entering your own name or entering the name of an ex to be like there are this many dogs with your name. And so I kind of faced a choice with visualizations like this, which is to condense the information and to tell a story. And my story can either be the weirdest, most niche names, or I can choose the most, like the biggest dog names in New York. And this piece always really, really surprises me as well, because I thought it would be so niche and the only people that would be interested in it were dog owners, but actually like loads of people that were commenting underneath it are people who know people called Bella and Max that were like, “Ha, ha ha. You have a dog’s name.”
Finally, just one more visualization that I made because I think it’s a rare example of an acceptable pie chart. Hopefully it’s memorable. Okay. So, I want to talk a little bit about the ways— you know, I’ve talked about this stuff before—about ways of communicating uncertainty and marrying together subjects and the visualization. What I want to talk about today is something which is a little bit different, which is another tool that we have at our disposal in this age of kind of short attention times when people are bouncing. And that is sequence, sequence. Can you guess what the third thing is? Oops! I went too far. Surprise. Because the point of the surprise thing is to actually hold you there in a way that’s really, really surprising.
So the point of sequencing, you can do it in a couple of different ways. One way that you can sequence the information to kind of hold people there is to introduce the subject before you show the visualization. So, introducing the subject, introducing the data, and obviously this only works on an iPhone. This scale is uncomfortable.
Another way that you can kind of sequence information is by showing one axis at a time. So, this next chart, which introduces the inflammation piecemeal was actually a chart that was designed for the visually impaired. I did a conference like this and a blind woman came up to me at the end and was like, “What you’re doing for people like me?” And I was really uncomfortable because the answer was honestly nothing. So I tried to design this chart, which is again about testicles, which is a running theme I guess. So yeah.
This is an audio chart. During puberty, boys testicles get bigger and, as they get bigger, their voices deepen. Tom will now demonstrate.
Hi, my name is Tom. This is how I sound with testicles that are one milliliter. This is how I sounded like the testicles that are five milliliters. This is how I sound with testicles that are 10 milliliters. This is how I sound with testicles that are 15 milliliters. This is how I sound with testicles that are 20 milliliters. This is how I sound with testicles that are 25 milliliters. This is how I sound with testicles that are 30 milliliters.
So, another way that you can introduce sequencing is by breaking down different historical periods. Completely different example here. And you can also sequence by showing one chunk of the information at a time. It doesn’t necessarily need to be historical chunks, but it can still be different chunks. So another example. This came from a study which is still forthcoming, so I’m not actually supposed to show it, but I want to. I still think there is space for interactivity in these kinds of visualizations, right? So I still want readers to kind of explore it, but I want to guide their exploration so they’re not overwhelmed. So this next visualization was designed for Instagram where obviously you can have a gallery of multiple images. And the idea is that readers will swipe through. That’s just one year’s worth. And this isn’t just for moving images. You can also do this with static images. So in the next example, I also feel like it’s sequencing because what I’m trying to do is kind of sequence the order in which your eye processes the chart. So I’m trying to make sure that you kind of start at the top and work at the bottom. Now I know that not every single person in the room is an online data journalist, so I want to talk about some of the ways that this can be applied in your world too. So this isn’t just digital. I recently got a commission to have kind of two pages in a magazine, a women’s magazine, to kind of do whatever I wanted with it. And again, I wanted to kind of explore this issue of equal pay on the magazine. And I was thinking about how when it comes to print, people do have slightly different attention times, right? Because they have a physical object in front of them. And I was trying to think about examples where I had physical objects. And it kind of made me slow down and process the information a little bit differently. And two examples came to mind. One was board games and the other one was, you know those like magazines? I don’t know how many of you guys read them, like the really trashy magazines for girl in the ‘90s when it was like, you know, this thing.
And it was like, “Is he into you?” And you’d follow it. “No, he’s not.” And so we wanted to kind of channel a similar thing about a very different subject matter. And so, my final visualization on the two pages looked something like this. Again, it isn’t out yet. And the idea is that you start at the top. You have a dollar. And then you see how that dollar changes depending on whether you’re a man or a woman, whether you’re married or unmarried, and depending on your race or ethnicity. And it means that people can engage with the data for themselves. The thing I think is quite powerful about this is it allows you to see multiple different possibilities and that’s the truth of the data. Right?
Averages today, in a diverse society, kind of don’t mean very much. People want to understand how the data relates to their gender, their age group, kind of where they live in the country. And again, this can kind of be sequenced. So this next example is about, again, desegregating the averages.
And finally. I also want to say that this isn’t just for data. One of the things that’s quite exciting for me about sequencing is that when you took a really, really long time to make something and you don’t want someone to just look at it for two seconds, you can also slow them down by showing them your process, by sequencing your art basically. And this is something that I made for Instagram stories that people could kind of just tap through.
And then the very last thing that I want to say is that this stuff really, really matters. So I’m working with data and working with the communication of information. And bounces matter because if they leave the page with either the wrong information or no information whatsoever, I’ve failed at my job. One of the very first charts that I showed you actually, that infographic that had too many labels on, was supposed to be an infographic that helped people understand whether or not to get a knee replacement. If that information has failed, then you have seriously, seriously let people down. And I hope that sequencing can kind of be a way out for all of us. Thank you very much for listening.