Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #101: Diana Yoo
171 Plays7 years ago
Diana Yoo has over ten years of experience as an information graphic designer and as an art director/design director. As the Art and Creative Director at the Pew Research Center, Diana lead a multidisciplinary team to develop digital projects of all...
The post Episode #101: Diana Yoo appeared first on PolicyViz.
Recommended
Transcript
Introduction and Guest Welcome
00:00:11
Speaker
Welcome back to the Policy This Podcast. I'm your host, John Schwabisch. I'm here with Diana Yu, the art director of the Pew Research Center here in DC. Diana, welcome to the show. Thanks, John. Yeah, I'm glad we finally got to get together because we met in person for the first time at Tapestry a few months ago. So now we actually get to sit down and chat.
Role of Art Director at Pew Research Center
00:00:31
Speaker
And it sounds like you guys are doing some incredible work here at Pew. I want to start by getting a sense of your background and what it is you do here as art director.
00:00:40
Speaker
Yeah, sure. Well, I've been here for nearly eight years. And as art director, I'm sort of responsible for just the general creative direction of the center. And I created sort of our visual look and feel that is translated in our reports on our website, well as just our overall brand.
00:01:04
Speaker
which extends to our logo, et cetera, on all of our collateral. And I have a team of about five designers currently who create most of our work on the website. And in our reports, our researchers do as well many of the graphics in our reports.
Collaboration Between Researchers and Designers
00:01:23
Speaker
So can you talk a little bit about the, how does the workflow work here? So when you have, you know, a huge team of researchers going through the six of you essentially. So how do you funnel that down?
00:01:35
Speaker
Well, it's interesting because actually the designers tend to spend about 80% of their time working on the static graphics that go into our reports. And the thing is that with only six of us, there's only so much we can do and we produce a lot of reports, hundreds of graphics in each one in some cases. And so our researchers
00:02:01
Speaker
use templates that we created to do the bulk of those graphics and our designers focus on the more complex ones. And the templates are in Excel? Templates are in Excel and Word. And we've made them as consistent as we can. We've given the researchers guidelines and best practices to use. And in general, it works really well. It's amazing how far we've come just in the last five years since we implemented them. And they're able to do
00:02:31
Speaker
Again, I think at least 75% of the graphics on their own and they look really great. Then the designers come in, clean things up here and there, and then for the graphics that are more involved or maps or those things the designers do themselves.
Adapting Graphics for Diverse Audiences
00:02:49
Speaker
So let's talk about audiences then in terms of how Q communicates to different audiences because I saw a graph about a week ago or two now and it was from the UN, essentially it's UN or UNESCO maybe, UNICEF. UNICEF, right. And so it had all these different countries and it had three years for each country and instead of doing like a little line chart or paired bar charts or what you sort of might,
00:03:17
Speaker
do as a default and excel right it had a bar for each country for one year and then a circle for another year and on a line for another year all sort of like on top of each other and i found it really hard to read even though the title was really good.
00:03:32
Speaker
The title sort of met the audience, but I sort of got into a little bit of a Twitter discussion with people about, you know, I don't really like this graph. I think it can be made clear. And someone said, well, maybe the people who read this report are familiar enough with the graph that it's okay. And my response to that was, well, maybe they just don't know. They just don't know what it would be if it, you know, what it would look like if it was better. So I would guess that UNICEF is doing the same sort of thing that you're doing.
00:04:00
Speaker
trying to communicate to different audiences sort of like here's a graph for the lay reader or the maybe the less sophisticated reader and then but here are all the details even packed within the graph right so so when you're thinking about communicating to say the public and then to the sophisticated researcher are you trying to walk that line as well I think we are I think we do tend to make sure that our graphics first and foremost are readable to
00:04:29
Speaker
general audience or an informed audience. But we also try to make sure that if it's possible to label everything, you know, we are very strict about ensuring that our axes are labeled properly, that we're clear about the framing of the graphic
00:04:50
Speaker
I think we do a pretty good job of that in our static graphics. It's something that just in general we think about with our research reports, making sure that they are readable to a general audience or to an informed audience, but include detailed tables or other things for those that are interested in learning more or taking our data and doing something else
Balancing Readability and Detail
00:05:15
Speaker
with it. That's something that is
00:05:17
Speaker
Obviously, many people may be listening to this right now, may have used our data in the past, and we try to make that accessible. The reports often are not places that we will have all that detail. It will be in a detailed table, or it will publish our full data set. Those are the things that are really useful to that.
00:05:37
Speaker
That little niche. Exactly. So do you get pushback from the researchers here about creating a graphic that may try to be reaching a wider audience? And the researchers are like, no, no, no, I understand this graph. And therefore, people like me will understand this graph. But maybe not the wider world. Do you have that tension, too? Oh, yeah. There's definitely creative tension, I think, in a good way.
00:06:07
Speaker
where, especially when we're discussing things like economic concepts that have a very wonky audience, for example, we may have sort of a staged approach in the report. So the overview, you know, that's the part that most people will read, or at least read the first page or two of. And that's going to be geared more toward general takeaways and big findings.
00:06:33
Speaker
As you get further and further into the report, you're going to find some more detailed tables, some more detailed complex charts that we're expecting. Really only that niche audience is going to be interested in reading. And those more complex charts, are they repeating content that was delivered earlier? Maybe repeating and then adding additional pieces to it? Yes. So are you thinking when you're writing, when you're creating or designing a report like this, are you sort of thinking,
Structuring Reports for Varied Interests
00:07:03
Speaker
I don't know what the right analogy or metaphors maybe like the martini glasses sort of the good one, right? It's like the broad overview and then diving in and so are you how are you thinking about physically like designing the report type? So does it have an executive summary? And then so can you just walk us through maybe how you sort of design each of these reports as you busy trying to meet each of these different audiences? Sure. Absolutely.
00:07:24
Speaker
It's hard to say design because the researchers do so much of this. The general report structure is an overview, which we've also called an executive summary in the past. And it has a very specific approach. And it's not necessarily bullet points, but it really goes through as briefly as possible the main findings. And then usually reports are structured sort of by
00:07:51
Speaker
topic area or something, some logical structure within it to direct people who are interested in different aspects of the work. So, for example, if we're talking about the middle class, there may be, you know, different aspects of the middle class that researchers are interested in who are reading this report and that will be sort of organized by chapter.
00:08:12
Speaker
We may also do that with larger, like religion reports, for example. We have global religion reports where it'll be broken down by Christianity, Islam, on and on. Right. Now, because you have these multiple audiences, have you done any testing with these different audiences?
Insights from User Testing on Graphics
00:08:31
Speaker
Because one of the comments I got about this UNICEF graph and this Twitter thing was,
00:08:35
Speaker
Well, maybe their readers are comfortable with it and that's why it's that way. My instinct is that they're probably not actually asking the readers whether it's a good graph. They just have been doing it this way for a long time and just continue to do it. Have you sat down with users and readers and said, you know, what do you like about this graph? Are you getting content out of this graph?
00:08:56
Speaker
Yes, we have done user testing. A lot of it is with our digital work, so interactives, and we focused a lot on getting our special features sort of to work really well on the web. And so much of our research is associated with that. But we have talked to users about specific graphics.
00:09:17
Speaker
And I think we found some interesting things and we don't have any really huge conclusions to come to. But for one, we've got a little bit of some mini research in that we did a science knowledge quiz a couple of years ago. And one of the questions in the quiz was we presented the audience with a chart. It was a scatter.
00:09:58
Speaker
I think this comes with data literacy. Yeah, or graphic literacy. Yeah. I mean, they're all related, but yeah. This is the one where it's like 60 something percent, know how to read a scatterplot. Exactly. Yeah, that's an interesting one. Good. Okay, cool. Okay, cool. Do you need to... Yeah, so we'll cut it later, but yeah, let me just count down and just continue. Five, four, three,
00:10:12
Speaker
graph. And that's it again. Sorry.
00:10:25
Speaker
So on the podcast a few months back, you were talking to Kim Reese and she actually mentioned how she felt like the data visualization community and really academia and the research community as well have a tendency to underestimate the public's data and graphic literacy.
Public Understanding of Data Visualization
00:10:43
Speaker
And of course, this is something that Alberto Cairo has talked about as well. And in his book, he used one of our charts, which was from a science knowledge
00:10:54
Speaker
survey that we did. And in the survey, we presented respondents with a scatter plot that showed sugar consumption on the x-axis and number of decayed teeth per person on the y-axis. You can imagine this. And each dot was a country.
00:11:17
Speaker
So they were asked then to choose from four responses to identify which statement best described the data in the chart. And overall, 63% of American adults gave the correct answer, which was the more sugar people eat, the more likely they are to get cavities. Which, you know, again, 63% were able to read this chart, which is pretty great. And, you know, 79% of college-educated respondents got it right.
00:11:43
Speaker
and 84% with a postgraduate degree got it right. I think it's great. I do think that maybe this task was made a little easier by the fact that we had these response options, that people were able, if they couldn't quite figure it out. They got to choose from some conclusion, right? Exactly. I suspect we would have gotten a different result if we had
00:12:09
Speaker
just asked people to interpret the chart without any answer options. This sort of relates to some of the results of user testing that we've found in
00:12:25
Speaker
with our charts and usually associated with an interactive feature. But the main thing is that we found that even college educated users, which is sort of our main baseline, they experience a bit of a learning curve when they're presented with unusual
00:12:41
Speaker
uh, chart types or unfamiliar chart types. And so before, sorry, before you go on, so unusual chart types would be like anything out of the bar column line pie chart group. Okay. So, so we, so a scatter plot would be, we would call an unusual chart type for this. I think so. Okay. All right. Great. So, uh, one of the things that we've seen is that, uh,
00:13:08
Speaker
Something our UX specialist refers to as satisficing. I think I've heard it before and elsewhere. But people have a tendency to kind of invent a good enough interpretation of charts. And they're just not always careful about consulting headlines, titles.
00:13:24
Speaker
legends, all the things that we put there specifically to make them easier to read. We record some of our user interviews and I've heard some where a user has been asked what they think a chart means and they'll look for a moment and they'll go, okay, yeah, yeah, okay, I see. This chart means that XYZ is the case and that makes sense. Great.
00:13:51
Speaker
The trouble is XYZ is totally wrong. And they've already moved on. They've already moved on. They'll say they had a great experience. Oh yeah, this was, you know, I had no problem understanding what's going on.
00:14:03
Speaker
But they got it wrong. And these aren't charts. You're not just showing them the scare plot region. It has a title and it has all the labels and all the annotation, right? Absolutely. It's exactly as we plan to publish it or as close to week as we can get. So is that just terrifying? A little. Yeah.
00:14:26
Speaker
So because I think we always think that if we have active titles, if we have good annotation, that people will be able to not just understand the graph, but they'll understand the content. And it sounds like at least some of the people who are looking at the graph, they didn't get that.
Importance of Clear Graph Titles and Context
00:14:45
Speaker
So what do we do about that? I don't know.
00:14:51
Speaker
I was writing some notes before this and that's basically where I thought of this. I don't know what to do about this. But you guys do have very active titles, right? But that's one thing that I've noticed about Pew. And actually, the UNICEF example is actually a weird example of this case, right? Because the title of that chart is pretty active, even though the visual itself is, I would argue, not that great.
00:15:18
Speaker
So I guess the first question is, is that a philosophy that you guys have here that our titles are going to be really active? And then when you think about that, and when you're making them active, are you also thinking about, are you trying to be as sort of objective as you can with the title? Or are you sort of leading people down a road to like, here's the causal story we're trying to tell. Right.
00:15:48
Speaker
I think using the word objective is tricky because, of course, we're... I think neutral is a better word. So you can say, here's... Let me go back to the scatter plot, right? Here's a scatter plot of tooth decay and sugar consumption.
00:16:05
Speaker
Alternatively, that title could have said something like, the more sugar you eat, the more likely your teeth are going to rot. So those are two different titles. So I guess are you on the more active side or the more neutral side? I think we do. Part of our style guide is to be with the more active side. And I think of it more as any chart needs to have
00:16:32
Speaker
a point that it's making. Of course, charts, especially with a lot of data in them, may be making many points. There may be multiple things you can take from them, but usually you have to focus on one main takeaway. That's what the headline should give you a hint of. Then we use the subhead in the chart to provide some more context. For example, I was mentioning earlier that
00:16:59
Speaker
the framing of a chart is really important that people understand, you know, what is the basis for this? Are we talking about all adults? Are we talking about just internet users? And the subhead allows us to essentially write the rest of the graph as a sentence. So it'll say percent of American adults who say abortion should be, and then the labels on the chart would say
00:17:25
Speaker
legal in all cases, illegal in some cases, illegal in all cases, things like that. So that a person, again, who is paying attention and reading it can easily figure out kind of what the chart is saying without just having to interpret the labels. But you're also doing what I think is really clever about that is you are in some ways forcing the reader to go into the chart.
00:17:49
Speaker
Because you could just say, you know, more than half of Americans believe X, right? And that would be true. But the way you frame the title forces someone to actually go into the read the data labels, right? Yeah, at least the sub title. Yeah, takes you into it. And I think
00:18:09
Speaker
With something like what I just described, it's fairly straightforward. There's probably not going to be a lot of confusion, even if we don't have that detailed subhead. But it helps avoid confusion when you have more complex things happening in the chart. For example, it would be very easy to confuse
00:18:28
Speaker
a chart that is either showing the percentage of Hispanics who identify as Catholic versus the percentage of Catholics who are Hispanic. Because you'd have a Hispanic label, you'd have a Catholic label, depending on how those are oriented, it'd be very easy to confuse those two. So again, I always say we need to write it so that it reads as a sentence as much as possible, that the data in the graph completes the sentence.
00:18:57
Speaker
people don't read it that makes it very hard but you know we've done everything we possibly can to we've given them a hint as to what they're going to see and then we walk them through and use the data and the chart to complete that.
00:19:13
Speaker
So do you think of these charts as applicable to a PDF report the same as on the website as an HTML graph, the same as what you might put on Twitter?
Consistency in Messaging Across Platforms
00:19:27
Speaker
Because you've made this active sort of succinct headline, do they work on all these different platforms or are you thinking about
00:19:32
Speaker
But we have this thing works really nicely here in the front part of the report. We're going to make it different for the back part of the report. We're also going to make it different for the website and for Twitter. Like there's so many ways to reach people about like, are you thinking about all these different ways or have you gotten like that core philosophy more or less worse for all the different channels? I think we've tried to make that core philosophy work for all the channels. There are cases where we find we need to
00:20:01
Speaker
adjust things, especially for, for example, the featured image for a report. The format for that is very specific to what Facebook will show or what Twitter will show. And so we format our graphics accordingly. But in terms of the messaging, it's the same in all those cases.
00:20:23
Speaker
Of course, we may have tables within the report that are not things that we're necessarily planning to put out on social media, but as much as possible, try to keep them consistent. I think it does work pretty well. For workflow reasons, it's important too because we just don't have the manpower to be able to make a special version for every single one. We essentially create the PDF first.
00:20:52
Speaker
and then work backward for the online version, and then... And that does ultimately, yeah. And then everything goes on, essentially, exactly. And I'm sure that's probably what many other organizations are working with as well, especially if they're creating something like this UNICEF report, for example, is a PDF only, I believe. Yeah, yeah. And the thing about that UNICEF report is that the graph would not work on Twitter. Like, it's too big, it's too dense, there's too much on it.
00:21:23
Speaker
However, one could imagine taking five of the 30 countries or something and making that smaller graph. But like you said, you're not going to do that for every report. You have a report with 35 figures. You're not going to start doing that for all the different figures. It's too time consuming.
00:21:38
Speaker
But of course, if people come get a conclusion out of a graph and think that they actually understood it, but they didn't, then I'm really not sure. I'm really not. It's like shaking the whole foundation of everything you do. Yeah, I know. I was afraid to mention that. It is something we found, and it's hard to say exactly. It's hard to extrapolate how much the user testing scenario is
00:22:01
Speaker
very specific. They've been presented with something. They know that they're supposed to be testing it and looking for something. Usually there's something that the person who's asking for the testing is looking for. It's hard to say whether that's how people in the wild, I guess, experience things.
00:22:24
Speaker
It seems that definitely there are cases in user testing where when people read a chart correctly, for example, and it doesn't compute, it doesn't match with something that they already think is the case, their tendency is to go, oh, there must be a mistake here. Because this doesn't match my expectations. Yep, doesn't fit my expectations, or I don't believe this to be true.
00:22:52
Speaker
So there must be a publication error here. And so that's something that we encounter, I think, on social media all the time.
Challenges in Chart Interpretation
00:23:01
Speaker
People will engage with us and say exactly that. So I can only conclude that there are some things that this user testing that are real world scenarios. So when you do hear that feedback, this doesn't really accord with my expectations.
00:23:19
Speaker
Does that tend to be unlike the simpler chart types? I wonder if people see a bar chart and they're familiar with that and so they may take a little more time with it because interpreting the graph is so much easier than say a scatter plot which is a little more unusual and maybe they just don't take enough time with it or it's not as familiar with it.
00:23:43
Speaker
and they read it as if it agrees with their preconceived notions, and so then they just do that and move on, whereas a bar charge is so much more familiar that if it looks much different than what they expected, they're willing to contact you guys and say, wait, this isn't right. That's a good point. I don't know that we've explored that specifically, or we haven't done enough user testing in this way to be able to compare
00:24:13
Speaker
the reactions to a simpler chart versus a more complex chart, but it's definitely something that I think we should do because it's important. We have an informed audience, but it's important to know what they expect and
00:24:28
Speaker
and what their graphic literacy level is. Yeah, absolutely. I mean, it's great that you're doing the user testing because I don't think a lot of places do the user testing or enough user testing. So I think it's great that you're doing that. So that's fantastic. Well, Diana, thanks so much for coming on the show. This has been a lot of fun and educational, perhaps shocking.
00:24:51
Speaker
Certainly educational. So thanks so much for coming on the show. Yeah, absolutely. And thanks to everyone for tuning into this week's episode. If you have any comments or questions, please send me a note on the comment section below or on Twitter, or you can find Diane on Twitter or the rest of the Pew Research staff. I'm sure they'd love to hear you yell at them about what you didn't expect in their graphs. So thanks again for tuning in. This has been the policy of this podcast. Thanks so much for listening.