Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #72: Maarten Lambrechts
178 Plays8 years ago
Hi everyone, welcome back to the PolicyViz Podcast! This week, I’m pleased to chat with data journalist, designer, and (new) visualization consultant, Maarten Lambrechts. Maarten has built a ton of great projects and analysis, so we dive into his process, his...
The post Episode #72: Maarten Lambrechts appeared first on PolicyViz.
Recommended
Transcript
Introduction and Sponsorship
00:00:00
Speaker
This episode of the PolicyViz podcast is brought to you by JMP, Statistical Discovery Software from SAS. JMP, spelled J-M-P, is an easy to use tool that connects powerful analytics with interactive graphics. The drag and drop interface of JMP enables quick exploration of data to identify patterns, interactions, and outliers.
00:00:19
Speaker
JUMP has a scripting language for reproducibility and interfacing with R. Click on this episode's sponsored link to receive a free info kit that includes an interview with DataVis experts Kaiser Fung and Alberto Cairo. In the interview, they discuss information gathering, analysis, and communicating results.
Guest Introduction: Martin Lambrechts
00:00:49
Speaker
Welcome back to the PolicyViz Podcast. I'm your host, John Schwabisch. On this week's episode, we're going to go a little over the Atlantic. I'm happy to have on the show Martin Lambrechts from Belgium. Martin, I was close, I think. I'm pronouncing your last name, but maybe not right there. Yes, actually very close. Martin, thanks for coming on the show. How are you doing?
00:01:10
Speaker
I'm doing fine. Thanks for having me. I'm really glad you could come on the show and chat with me. We have a lot of great projects of yours that I want to talk about. But first off is congratulations on new website, new journey, I guess, into the freelancing sphere, data journalist, data designer, and visualization consultants.
Career Journey: From Bioengineering to Data Journalism
00:01:32
Speaker
I think you have your work cut out for you. Before we get into maybe some of your projects, do you want to talk a little bit about your background and
00:01:39
Speaker
what your goals are, what your dreams are for the new venture you're undertaking? Okay, so I graduated some 13 years ago as a bioengineer and I also have worked some years as a bioengineer and an agricultural economist.
00:01:59
Speaker
But then I stumbled more or less into journalism. And for the past two years, I was working at Mediafin. It's a Belgian company and publisher of two newspapers, The Ted in Dutch and Le Coe in French. And I worked there as a data journalist and a visual journalist making charts and maps interactive and static.
00:02:22
Speaker
And a few weeks ago, I switched to Freelancer, so I'm now for hire. I teach, I give workshops in data visualization and data journalism, and I also design visualizations.
00:02:40
Speaker
Yeah. So the whole range of things that people need. So if you are in Belgium or Europe or anywhere in the world and you need someone to help you with your data visualization needs, Martin is a great guy to go to. And so we're going to talk about some of his projects so you can get a flavor of the type of things that he works on.
Interactive Projects and Statistical Uncertainty
00:02:58
Speaker
I want to start with one of the really neat projects you did. And I think you launched this at Malofie conference last March, your rock and pole.
00:03:08
Speaker
project is sort of an interactive data story on explaining uncertainty in poll results. Do you want to talk a little bit about how you got into that and why that was interesting to you, then a bit about how you built it and what you're trying to do with it? Yeah, so I was a bit annoyed about how polling results were reported here in Belgium, but I guess the situation is similar in other countries.
00:03:31
Speaker
So reporting focused on small differences in poll results and I want to show the public but also my colleague journalists that focusing on small differences is not a great idea because in polling results you have uncertainty and small differences can be the result of randomness.
00:03:51
Speaker
And I wanted to explain it visually and without using any statistical formulas because I knew that a lot of people would just switch off with the first formula. And so I built it interactively and I also made it a bit like a story. So I explained these things step by step and people can interact. They can generate their own polls and they can see how polling more people influences the polling results.
00:04:21
Speaker
And in the end, they can generate eight polls at the same time. And then they can see the differences between those polling results. And they can see that randomness actually can play a big part in polling results. So it was my attempt to explain one aspect of polling and the uncertainty that goes with that. But yeah, visually and interactively. So without using any statistical formulas.
00:04:50
Speaker
Do you think that people don't understand uncertainty because we don't we being whoever we is capital we don't do a good job of explaining what uncertainty means that we use terms of margins and fair and people just don't understand it or
00:05:07
Speaker
or we present this 52% of people support this candidate, and we provide it as this solid, hard, specific number, and maybe we should be a little bit softer about how it's presented. Where do you see the pain points, I guess, of presenting uncertainty to news readers or media readers or people who may not just be familiar with the statistics and the underlying uncertainty?
00:05:36
Speaker
Yeah, I think uncertainty is something really hard to understand and even I struggle with it sometimes. The thing we need to do is make it visual. There are some techniques to show uncertainty. You can first of all visualize the interval, the confidence interval. You can play with animation. So there are a lot of techniques and there's also research about how best to represent uncertainty visually.
00:06:06
Speaker
So first step we need to take I think as visualizers is when there is uncertainty also show it in our graphics.
00:06:14
Speaker
Because then people, but also the journalists who are reporting on the story, they will notice that there's something there and when they see overlapping intervals, they might be a bit more hesitant in drawing big conclusions from the data. I think we need to just show uncertainty when there is uncertainty visually.
00:06:37
Speaker
And then people will start to be more susceptible to uncertainty and maybe even understand it better. Yeah. I mean, there are lots of reasons I think why uncertainty is not well understood. And from the sort of media perspective, it's a lot easier to give the point estimate of something. This is 46%, as opposed to saying, well, it's 46%, but plus or minus x and y.
00:07:04
Speaker
how do people get around or content producers really get around that where you want to give something quick and short and get to the point but also you need to convey this uncertainty you've mentioned making it visual but I can imagine a lot of times where you know you're writing a report or a journal article or a news article I mean how do we as people who are working with data and visualizing data how do we try to educate people I think a little bit more on what uncertainty means what probability means and those sorts of things
00:07:35
Speaker
Well, I think we just have to have the journalistic reflex that if there's something we don't understand, we need to ask kids to someone who does understand. So when the reporter wants to write a story about polling results, for example, he wants to be sure that there's a story there. He just needs to talk to the people who conducted the poll or other people who are more statistically savvy. So they just need to ask for advice. Yeah. I think.
00:08:04
Speaker
Yeah, the people conducting the polls also have a big role to play in the whole system of polling, in my view, because often they don't do a very good job on reporting the results to the reporters. The confidence intervals and margin of errors that are put in the small print at the end of the documents, for example. Yeah, for journalists to understand what is in the results is actually
00:08:31
Speaker
really hard because the explanation uses the jargon, the statistical jargon. So people conducting the polls also have a responsibility in explaining what all the results mean.
00:08:44
Speaker
Yeah, that's a great point.
Trends in Visualizations: Interactive vs Static Graphics
00:08:46
Speaker
The rock and pole project you did was an interactive visualization tool, but I know you also have some strong feelings maybe we should say about the relationship between interactive visualizations and static graphics.
00:09:01
Speaker
You were at Malofie last year. Archie say, give a great talk on why maybe the New York Times is sort of moving a little bit away from interactivity. Can you talk a little bit about your feelings on maybe moving away from interactivity towards, towards static graphs? Yeah, I think Archie did a good job there in, in just showing that also the big newsrooms who used to do these very complex interactives.
00:09:28
Speaker
And they're stepping away from that. And, um, this was actually good news for smaller newsrooms because, um, producing a static graphics is a lot easier and, and needs less time, um, and then less people to do it. Yeah. It's, it was actually good news, but stepping away completely for interactivity, um, I think isn't the solution because, um, for some data sets, it's actually useful to have some, some interactivity, for example.
00:09:57
Speaker
If you want readers to discover the stories in the data that are most relevant to them, you have to give them some search forms, for example, or some other controls to find their own stories.
00:10:11
Speaker
And so I think we can step away completely from interactivity because interactivity also can be really powerful. Without interactivity, my rock and pole wouldn't be so comprehensible and would be more boring, I think. We like seeing moving things on the screen, so I think
00:10:38
Speaker
Um, yeah, we should be keep doing interactives. Um, but as Archie said, um, that's, it's, it's really hard to make it work on all screens. Well, I see some websites who are doing really great interactives at the moment. Um, but they're not in journalism. They, they're more like, um, arty projects or, uh, experiments.
00:11:03
Speaker
And they work really well on desktop. And when you open them on a small screen, they just say, I'm sorry for this, you need a bigger screen. And I think there's some value in that because yeah, if you want great visual and interactive things, you need pixels. It's a bit of pity that news is consumed so much on smaller screens. When I started working in journalism,
00:11:29
Speaker
It was all desktop. And obviously we really have to face this reality, but we can't throw away the interactivity completely just because people are reading news on smaller screens. There's a lot of value of putting more complex and visual interactive things on bigger screens. So yeah, you have to balance the two, I think. Yeah. What I find interesting about interactivity is that
00:11:53
Speaker
In some ways, it's more complex to create because you need to know how to use those tools or those programming languages to create the interactivity where everyone can make a chart in Excel and that's fine. Things like Tableau and high charts and things, maybe they're not that complicated, but it requires some other level of skill or knowledge. But on the other hand,
00:12:16
Speaker
In some ways it's lazier. I feel like a lot of the interactive pieces are lazier because the creator just says, well, here's an interactive piece. You can just go dive into it and go figure it out on your own. Whereas in the static pieces, we have to think about the annotation and the notation and the text to help the user understand. Do you sort of see that conflict? Do you think that people just put things, make things interactive because they have the tool in front of them and they can, and it's just an easy way to do it?
00:12:44
Speaker
Well, I don't think it's lazy and I don't really see a conflict. If you make things static, then you have to think really well about the message you want to convey. And that's a great thing because it lets you focus. But for me, the greatest pieces are the ones where you guide your reader through a data set, maybe in a
00:13:09
Speaker
a sequence of static charts or maps. And then in the end, you give the control to your reader and you say, well, here's all the data, go look for yourself and try to find your own stories or the stories more relevant to you. And I think those are the most powerful because you explain first what is in there and what are according to you are the stories in there.
00:13:35
Speaker
And then you just give away all the data and you give the reader some controls to find the wrong story. And I think those are the pieces I enjoy the most. When you start a new project or when you're given data, do you immediately start down in your mind? Do you start down a static path or do you start down an interactive path when you're thinking about the final project? I think that develops along the way. But in most cases, you have an idea of what you want to achieve.
00:14:06
Speaker
So I think the decision to build an interactive alongside some static visuals or just interactive on itself, I think this decision is made pretty early in the whole process.
00:14:27
Speaker
Because, yeah, you need different things that the data needs to be structured in different ways. For example, if you're building static things, you can mostly calculate some aggregates, for example, and build some charts.
00:14:45
Speaker
And if you want to go interactive, then you start most of the times with the raw data, the non-aggregated data, because you want to show all you have. So I think this decision is made pretty early in the whole process.
Tools and Techniques for Data Visualization
00:15:03
Speaker
Can you talk a little bit about your development and creation process and the sort of suite of tools that you use?
00:15:09
Speaker
Well, what I do when workload permits it, and those is mostly during summer and winters and during holidays, is learning new tools or extend my knowledge of tools I already use. And last summer, I took the occasion to learn some R and also ggplot2, the R package for visualization. And I noticed that
00:15:37
Speaker
When you know those tools, you know how to work with them, you don't need a lot of Excel anymore because, yeah, it has some advantages. Using R, for example, everything you do is recorded. You build scripts, and when new data comes in, you can just run the script against it, and then you have in the end a new visualization with the fresh data.
00:16:05
Speaker
In Excel, a lot of the times you just have to start from zero and repeat all the steps you did before. So I use R to make sketches. What I have done also is produce PDFs within R, which I sent to the graphics people, the people who make graphics for the newspaper.
00:16:27
Speaker
So I make like the structure of visualization and they finish it off for the newspaper. And then for interactives, I use D3, like most of the online interactive visualizations. So
00:16:49
Speaker
Yeah, I can really recommend people to learn some art because it's really powerful. And if you invest some time in it, you'll notice that it will save you a lot of time later on. And it can seem like a big mountain, but I can guarantee it really pays off. Yeah. You mentioned when you were at the newspaper working with the design staff.
00:17:15
Speaker
How do you get together? How do you work in teams? Or maybe the better question is, what's your view of the skill sets that are needed in a team to have this entire workflow of writing and data visualization and maybe data analysis, statistics, and design? If you could build your perfect team, do you have this core skill sets in mind already?
00:17:44
Speaker
Well, I think you have to have someone who's comfortable working with data. And then you also need someone who knows how, well, what visualization fits to what data, for example.
00:18:00
Speaker
And then you need someone who can finish the design. So someone who knows design, who knows typography. If you're in a print environment, you need to have someone who understands the logic of print because it's really different from online. Yeah, and then obviously a reporter. And I think it's also really important for reporters to know a bit about visualization, to know what's
00:18:29
Speaker
chart types are out there and what chart types could fit their story and also how to mix charts with text.
Skills for Effective Visualization
00:18:37
Speaker
Because a lot of the times a chart is just something to illustrate a written text and I think those are missed opportunities. You really need text and visuals to work together. Is there a skill that you think is missing from... Let's start with the data visualization.
00:19:00
Speaker
People who sort of you know regularly make visualizations and do data visualization is there a skill set that They're really missing like if I if someone asked me, you know, what skill set are researchers missing from communicating their work, you know, one of the first things I would say would be you know, they're not thinking visually and so they're not thinking about how to communicate to a sort of broader audience is there a skill set that you think the data visuals that you know people who are who are
00:19:25
Speaker
You know primarily making visualizations. Is there a skill set that they are? Back the letter term lack of a better term. Maybe they're lacking Well Really interesting that you asked that because just today I was looking at a cartogram of the Netherlands which is not my my own country and I was noticing that it was very hard to interpret and
00:19:53
Speaker
because I wasn't familiar enough with the normal shape of the country of the Netherlands. So I couldn't really estimate the deviations that were shown in the cartogram. So in the cart, the polygons of the provinces in this case are distorted according to some data value.
00:20:19
Speaker
But I'm not from the Netherlands, so I couldn't really understand the map. And what we often forget is that people who are not looking at graphics day in and day out, like me or like other people who make those graphics, that it's often not so easy for lay people to get into graphics they're not familiar with.
00:20:48
Speaker
So we really need to pay attention to that. We have to add enough annotation so people can understand everything in the graph really well and the bar to get into it is lowered. But on the other hand, I also think that we need to experiment. And especially in media, I think we have a role to play in educating our readers on
00:21:17
Speaker
less common graphics because sometimes a story can benefit from a graphic that's not like your normal bar chart or line chart. It's more like an exotic graphic. And I think we just have to, in those cases, go for it and publish those weird graphics because they can illustrate the story sometimes much better than the day-to-day bar charts and line charts.
00:21:47
Speaker
But yeah, we need to be conscious that not everybody is looking at charts and maps all day long. That is certainly true. Do you have a particularly favorite non-standard, shall we say, chart type these days?
00:22:08
Speaker
Well it still surprises me how few slow graphs I see in media especially because they're so efficient and also very elegant and in a print environment in a newspaper they fit really well because they can fit into one column of text for example or maybe two and a lot of times when reporters want to show the differences between two
00:22:37
Speaker
time periods. They use bar charts.
00:22:41
Speaker
And a slope chart is just so simple and beautiful and effective, and we're still not using it.
Underused Visualization Methods: Slope Charts
00:22:48
Speaker
So that keeps surprising me. Yeah, I think that's right. And I think it's one of those, like you said, one of those graph types that's sort of new to most people and requires some explanation of not only the content, but also how to read the chart. So it sort of adds that layer of complexity, I guess, to a lot of readers.
00:23:11
Speaker
But in essence, it's just a line chart with two points in time. Martin, this has been really interesting.
Conclusion and Future Projects
00:23:22
Speaker
I'm really excited about the work you've been doing 2016 and certainly excited to see what comes out of your new efforts here in 2017. So thanks for coming on the show and good luck this year.
00:23:33
Speaker
Well, thank you very much, and thanks for inviting me. It was a real honor. It was great. And thanks to everyone for tuning into this week's episode. I hope you've enjoyed it, and please check out Martin's work on his website and the various projects that I will link to on the show page. So please also let me know what you think of the show, guests that you'd like to hear from, comments, suggestions, either on the website at policyvis.com or on Twitter. So until next time, this has been the Policy Vis Podcast. Thanks so much for listening.
00:24:11
Speaker
This episode of the PolicyViz podcast is brought to you by JMP, Statistical Discovery Software from SAS. JMP, spelled J-M-P, is an easy to use tool that connects powerful analytics with interactive graphics. The drag and drop interface of JMP enables quick exploration of data to identify patterns, interactions, and outliers.
00:24:30
Speaker
JUMP has a scripting language for reproducibility and interfacing with R. Click on this episode's sponsored link to receive a free info kit that includes an interview with DataVis experts Kaiser Fung and Alberto Cairo. In the interview, they discuss information gathering, analysis, and communicating results.