From the course: Data Visualization: A Lesson and Listen Series

Listen: Jen Christiansen

From the course: Data Visualization: A Lesson and Listen Series

Start my 1-month free trial

Listen: Jen Christiansen

“

(upbeat music) - I'm thrilled to have Jen Christiansen, Senior Graphics Editor at Scientific American here with me today to talk about science visualization. Jen studied scientific illustration in school and has been with Scientific American on and off for about twenty-five years, with a stint at National Geographic in between. And she's a student of the field. She writes about the history of science visualizations at Scientific American and thinks about how we can get better at communicating science centric topics visually. Welcome and thank you so much for joining me today. - Thanks so much for the invitation. It's an honor to join you. - So I want to start off just asking this question. As a student of the field, maybe you can start us off with a very brief definition of science visualization, at least within the pages of Scientific American. What do you include in that definition and what are a few maybe examples that you might include to help illustrate some of the categories? - Sure. To my mind, the term scientific visualization kind of refers to the full gamut, from representative illustrations like dinosaur reconstructions, to schematics and diagrams to data visualizations. Others might have a narrower definition, one that's more focused on kind of classic data visualization like charts and graphs, but in the world of science, research and publishing, I think that everything that we illustrate is rooted in data collection at some point of the process. So like dinosaur bone length measurements for those reconstructions, and documented experiments that help us understand things like how a virus works, to also just straight up charting of numbers. But there's also like photography and other forms of data collection that allow for things like black hole images from the event horizon telescope. Although the term scientific visualization can be applied to so many things and varies a bit from discipline to discipline, it really comes down to me in my day-to-day work as a visual science journalist. It kind of comes down to translating quantitative information related to science research into imagery. - That's great. Yeah, it's such a broad thing. It's beyond data visualization. It is sort of this, I refer to it as like a sister topic or a sub topic because it crosses so many different domains. You said that you studied scientific illustration and you've spent your entire career really in this field. What would your advice be to someone on the outside? Maybe not from the science background who might need to work with scientific data, and don't think about necessarily illustrations for this. Maybe more on the charts and graphs side potentially. So let's say as an example, they're working with epidemiological data from COVID-19 say, what are some words of wisdom you might share to keep someone from maybe getting into trouble as they approach a project like that? - Yeah, I recommend starting with some basic research on the topic at hand. Try to learn kind of the key bits of jargon, basically the specialized terms of vocabulary. With things like mortality and public health data, that might mean doing some basic Google searches on things like Crude Death Rate and Adjusted Death Rate. But you also need to remember that that basic research doesn't make you an expert. So talk to the people who are the experts. If you have questions, reach out to the corresponding author on a scientific paper that you're interested in, or in the case of public data portals, do a little research on the general topic and try to find a scientist that specializes in that area of research. It's also important to remember that on the science bit, most of the data you'll be working with has been analyzed and peer reviewed in journal articles, but that it doesn't mean it's flawless. So you must remain kind of... You need to retain a critical eye. Read the method sections of papers and data portals, and become familiar with the questions that guided data collection, as well as the subjects of the study itself. So this is particularly relevant with regards to health studies, as the results might be rooted in a homogenous sampling of people. And that's a detail that should inform how you frame the data. Again, though, with all data visualizations, be cognizant of data collection methodologies. Again, especially true with global health data in which data collection methods are rarely consistent over space and time. So seek advice from content experts when trying to determine which authoritative source is considered the best option for the subject or a regional focus of your particular story. And then upon plotting out the data in your preliminary explorations, question any and all surprising patterns. For example, does the prevalence of a disease rapidly change? Check the documentation to make sure that the shift isn't due to a change in data collection methods. And then check with a content expert before jumping to conclusions. Although, sometimes the right decision is not to visualize the data at all. So Amanda Makulec has a great post on this topic that's titled, "10 considerations before you create another chart "about COVID-19." And one of the tips she includes there is a reminder to be cautious about making generalized predictions or comparisons based on regionally specific data. - Yeah. Bring up a great point. That article is a really good article. It was written on the publication for the Data Visualization Society, if I remember. And that's actually the advice I gave in the lesson that precedes this interview. Essentially the two things you just said. Talk to an expert, and yes, be very, very, very careful 'cause you're probably not an expert in that subject. It's interesting that you mentioned the idea of looking into methods because, if you've studied science and you understand statistics, you can do that and do that very well and really understand the methodology. For someone who's outside the field, even that can be somewhat challenging. Any advice you can share in terms of how a non-expert, even someone who doesn't have a scientific or a statistical background, might think about things like that. - Sure. I think, reading through the methods as they're documented, even if you don't understand every bit of it, it kind of helps prepare you for a conversation with that expert. So I often just kind of go very slowly through. If it's a scientific paper I go through and I highlight the sections that I think are important to the filter that I want to look at this data through, or things that I don't understand that might help guide the question when I do get the expert on the phone or through email. It's mostly just to kind of try to learn as much as you can about the information. So you can highlight what you need to learn more about and how you can kind of make the best use of that expert's time. - Yeah, that makes sense. And I think people who do what we do, we're theoretically very data literate at a minimum. And like you said, we can detect a strange pattern shift, something which we can then ask the expert about. So, that's great. I know that you studied geology, as well as studio art in school. Not biology, not chemistry, not physics, some of the other subjects that you cover. So how important though was that general scientific education for you, enabling you to do the work that you do? And do you work with freelancers who have like maybe sometimes zero scientific knowledge? If so, how does that tend to work? - Sure. Yeah. So some geologists would argue that geology covers all of the other disciplines, (Jen laughs) - Fair enough. - But I was only... I only studied it as an undergraduate, so I didn't kind of get quite as deep into physics and chemistry as some others may have. But my studies in geology really helped me, start to understand how the practice of science works. And perhaps most importantly, how to read a scientific paper. And in terms of working with freelancers. So when dealing with highly processed abstract and specialized data, like with genetics and astronomy, I find it useful to work with freelance data visualizers that are domain experts or have a science background in that area. There are so many really specialized conventions in those topic areas, both in terms of the vocabulary and how the data is structured. But if communications lines are kind of open with a scientist that collected the data, and if the topic isn't too esoteric, it can be kind of fabulous to see what a freelance data visualizer that isn't a specialist in that field can do with the data. They tend to be less beholden to kind of standard visualization practices of their particular scientific field. And they often bring a really kind of fresh and fun solution to the table. But in those cases, I breathe a lot easier knowing that a content expert will be reviewing things for accuracy. - Yeah. That's the balance, right? Bringing that fresh perspective, but also retaining the adherence to accuracy, scientific realities. It's a tough thing to accomplish sometimes. What are some of the unique challenges to science visualization specifically as opposed to more broadly, information visualization outside of the scientific fields? - Yeah, so I think that scientific visualization is rooted in a tradition of using data visualization mostly as a tool for analysis. So many charts and graphs that I see from scientists are pretty complicated to decipher. But once you learn how to read that chart, you're probably set up pretty well to be able to read other charts in that same discipline. But they aren't optimized necessarily to communicate results to people who aren't already fluent in the chart language of that discipline. So they're kind of... They're optimized for specialists looking for patterns that they're already primed to understand. And I think that sci viz, may rest on that a little bit too much. Scientists and others that work with scientific data need to remember that, they need to make design decisions based on each context and each intended audience. So a visualization form that you use in the lab for analysis might not be the best form to communicate results to your peers in a paper. And yet a different form might be better suited for a poster presentation. And still another version might be best for a press release where you're trying to reach a broader audience. - Yeah. It's like the curse of knowledge, right? So the scientists do what they do. They're experts in what they're experts in. And they're used to speaking to people like themselves and getting them to take a step back is always tough. So it's our job is to be the translator sometimes it seems. So, Scientific American where you work had its 175th anniversary this year, the longest continuously published magazine in America. That's amazing. I had no idea that was the case until this year. And I really loved the project that Moritz Stefaner created looking at trends over time, based on the words published in the magazine, different topics and themes as they evolved over time. I know that your website, says that you think and write about science history of visualization. So I wanted to just to ask you to share maybe some interesting examples or stories of, projects, visualizations, things you've seen in the archives over that rich history that might be informative for our audience. - Sure. Well, one of my favorite examples is kind of from the more recent history, the 1970s. But hands down, my favorite data visualization from Scientific American's archive is probably better known to the folks watching this video as the cover art for the band Joy Division's debut studio album titled "Unknown Pleasures". That album cover was designed by Peter Saville in 1979. But the graphic appeared in 1971 as a full page image in Scientific American in an article on pulsars. So pulsars are a category of rapidly rotating neutron stars. It turns out that the original chart was created by Harold D. Craft and published in his PhD dissertation at Cornell University in 1970. But I just love that a chart created by a graduate student for a really niche area of astrophysics, became a pop culture icon, with a bit of an assist from Scientific American which appears to have been a stepping stone when you trace that path of publication. - Yeah, that's a great example. I can just imagine. I wish there was like real time Google Trends, 'cause you know that there's like a spike in searches for Joy Division cover art right now. But do you know any of the backstory behind that? Like why that was chosen to be the cover for that album? Like what the connection was? - Yes. Apparently the band brought Peter Saville the book, "The Cambridge Encyclopedia of Astronomy". And it had been printed in that book along with several of the other illustrations from that same Scientific American article. Yeah. So the band members pointed it out and then Peter Saville of course did his magic with making lots of design decisions that made it the iconic cover that it is, no title, no other indicators. So he definitely did some serious design work on that, but it looked like it was through the encyclopedia. - Interesting. Yeah, that's great. It is nice to see data visualization cross into the public domain, public consciousness like flatten the curve data. I mean, like to an amazing degree more recently. Looking at 105 years of history, 175, rather, years of history, thinking about more recent history in data visualization and information design, any insights that you have on trends, innovations, interesting developments that might be also helpful for our audience. For instance, I wrote an article for the Data Visualization Society publication also on "The Past, Present and Future of Scrollytelling", which of course is this mechanism that is used all the time in data storytelling these days. I'm wondering what you think the next big thing might be in data visualization specifically based on, what you're seeing in terms of trends and experience over your... Even in the Scientific American pages specifically. - Well, I think that one of the next big things 'cause been kind of percolating for a little while now but it's been gaining momentum recently, although I suppose that's how it often works with these trends. Specifically, the idea that data is messy and it's an artifact of our own biases. I think that often, especially in the context of science, we look at data as cold, hard indisputable facts. I mean, I think it's accepted within science savvy audiences that additional studies and research will add more information and help shift the shape of that larger dataset. And it may result in shifting interpretations over time. I mean, that's pretty much how the, practice of science itself works. But I fear that we too often create and interpret discrete charts with a sense that they represent indisputable and absolute truth kind of full stop. But it only really represents the outputs of a particular line of questioning or a specific model or a specific experimental setup. So how does this translate into a tangible data visualization trend? Well, I think we'll see more and more designers grappling with how to portray things like uncertainty and error ranges, particularly as they relate to climate models and public health projections. And this is certainly already happening, following the lead of perception and visualization researchers including, Jessica Hullman, Matthew Kay and Lace Padilla. But I also think we'll see designers grappling with how to portray both what the data includes and what it excludes. For example, data broken down by gender, often just includes two categories, male and female. But how do folks that identify as neither of those options in that binary view, like how are they represented in the data? So for people interested in thinking on that more, I recommend checking out the project "Missing Data" by Mimi Mimi Onuoha. There's also writings on "Data Humanism" by Giorgia Lupi and Jer Thorp. The organization Data for Black Lives and its Executive Director Yeshimabeit Milner and especially the book which is called "Data Feminism" by Catherine D'Ignazio and Lauren Klein. - That's great. Yeah, I've seen a lot of talks and articles about uncertainty visualization over the last couple of years. Alberto Cairo did a great piece that showed up in the New York Times about, the hurricane visualizations and what does the cone of death really mean? And which it's good to see that even, getting to the general public in that way. We're running out of time. I just want to thank you very much for being here and speaking with me today and for, of course, the amazing work that you do at Scientific American. Thank you so much again for sharing your thoughts and wisdom with us. I really appreciate it. - Thanks for the conversation. I really enjoyed it.

Contents