00:00:00
00:00:01
What is Data Visualization? From the Expert Behind PolicyViz
3.4k Plays1 month ago
In this episode Autumn and Dr. Jonathan Schwabish discuss the importance of strategic thinking in data visualization and the key elements of good data. He emphasizes the need to understand the data and how it was collected, as well as the importance of starting bar charts at zero. He also highlights common mistakes in data visualization, such as distorting or lying with visuals, and the potential impact of data visualization on policy decisions. Looking to the future, he discusses the role of AI in data visualization, the integration of AI into visualization tools, and the potential of augmented reality and virtual reality in data visualization. Jon Schwabish discusses the different data visualization tools he uses, including Excel, R, Tableau, Datawrapper, and Flourish. He emphasizes the importance of choosing the right tool for the specific use case and audience. He also highlights the need for policymakers and individuals to be trained in interpreting and using data visualizations effectively. Schwabish discusses the ethical considerations in data visualization, such as using inclusive language and considering accessibility.
Keywords: data visualization, strategic thinking, good data, common mistakes, impact on policy decisions, AI, augmented reality, virtual reality, data visualization tools, Excel, R, Tableau, Datawrapper, Flourish, policymakers, data interpretation, ethical considerations, inclusive language, accessibility
Subscribe to Breaking Math wherever you get your podcasts.
Become a patron of Breaking Math for as little as a buck a month
Follow Jon Schwabish on Twitter and on YouTube. Also go give PolicyViz Podcast a follow
Follow Breaking Math on Twitter, Instagram, LinkedIn, Website, YouTube, TikTok
Follow Autumn on Twitter and Instagram
Follow Gabe on Twitter.
Become a guest here
email: [email protected]
Transcript
00:00:00
Speaker
Welcome to Breaking Math, the podcast where we not only prove theorems, but also look at regressions, policies, and procedures in the world of data visualization. I'm your host, Autumn Feneff. Today we're joined by Dr. Jonathan Schwabisch.
00:00:15
Speaker
He's an economist, writer, teacher, and data communications expert. He is considered a leading voice of clarity and accessibility in how analysts, researchers, and scholars communicate their findings. Across four books, he has provided comprehensive guide to creating, communicating, and distributing data-rich content.
00:00:37
Speaker
Better presentations coaches people through preparing, designing, and delivering data communication products. Elevate the Debate helps people develop a strategic plan to communicating their own work across multiple platforms and channels.
00:00:54
Speaker
Better Data Visualizations details essential strategies to create more effective data visualizations. His most recent book, Data Visualizations in Excel, helps readers create better graphs and charts in the Excel software tool. So stay tuned to listen to what good data looks like on this week's episode of Breaking Math.
00:01:25
Speaker
Hi, John, how are you doing today? I'm well, Autumn. Thanks for having me. Thank you for coming on the show. Now tell the guests a little bit about what you do behind the scenes at Policy Phys and also your career as the data analyst and economist.
00:01:43
Speaker
Sure yeah I mean it's a long it's a it's a long path so we'll keep it we'll keep it short uh because we got a lot to talk about of course yes yeah I'm trained as an economist um I did my my schoolwork here in the states at the University of Wisconsin and Syracuse University um I spent about a decade working for the congressional budget office which is uh uh based in the budget arm of the U.S. Congress and while they're kind of towards the end of my tenure I started to get I think a little frustrated with how I think the media wasn't really using a lot of the analyses and reports and things that we were producing and even like members of Congress weren't using it. It was just kind of frustrating. And so I started to think about, you know, what is it like, how can we get our stuff to stand out? And this was like in a pre social media days, right? So you had sort of more, you know, kind of legacy media, you had, you had sort of dedicated channels.
00:02:40
Speaker
And um it turns out there's this whole world of data visualization of people actually thinking carefully about how to communicate your data, which is something that like I don't know, I was never exposed to in any school, like certainly graduate school. There's never any talk of like, how do you present? How do you write? How do you speak? How do you make good graphs? Like it's just not part of it. And so I um got, I dove headfirst into this, into this world, um met some great people and then just started of making things and and pretty quickly realized that like, if you just spend a little bit of time,
00:03:16
Speaker
um Thinking about how your reader or user or or audience them whoever it is, how they are going to use it and what they need, you're just going to make better stuff. And so I'm not a designer, I'm not a developer a web developer, um'm you know, again, I'm going to come to select.
00:03:33
Speaker
ah But all you have to really do to start is just to think about what people need. Like maybe they don't even need a graph. Maybe they just need a table or maybe you've been showing them tables, but they need a graph. Like that's the sort of thing is just like strategic thinking.
00:03:49
Speaker
um And so I was there, I worked at CEO for a long time, and then I moved to the Urban Institute, which is a nonprofit. I'm based in Virginia, but the Institute's in DC, so right across the river. And um yeah, I still do economic research, mostly on nutrition policy and and disability policy. And then I started my own side, little hustle, as the world is these days. PolicyViz, where um I do similar sorts of work,
00:04:16
Speaker
Um, you know, consulting with folks, building tools, um, you know, uh, writing all all the sorts of good stuff, hand man podcasts. I mean, all the stuff on them, you you know how it is. Like yeah always, always. It's par for the course. That's right. That's right. One thing helps the other helps the next thing. and that's That's work in in the 2020s is just a constant hamster wheel. So yes um yeah, I mean, I come to this world, again, not through a design eye, not through a kind of computer ah science eye, but through the, I guess I would say the data, the research eye, and recognizing, I think, that a lot of folks who also come on that ah same journey, that same path,
00:04:59
Speaker
don't think about the other pieces of the puzzle. I think more so now ah than in the past. clearly I mean, the the tools that we can talk about later, the tools are better, the browsers are faster. The media organizations have accelerated how important data and data visualization is in their workflow. So there's a lot of reasons why people are more attuned to it than they were when I first started. um But it is, I think, at its core,
00:05:27
Speaker
If you're thinking about how do I get my research, my data, my analysis, my math out there so people can understand it, like the visual, we know visuals work, and you don't really have to, again, get an MFA in design, just think about what is your, you know, that key person, that key group, that key stakeholder, what do they want to see, what's going to help them do their job better, what's going to help them find insights. And once you sort of identify that, the rest is is really just gravy on top.
00:05:56
Speaker
Definitely. Now I've done data, whether it's been in my courses or, you know, just in working, right? So everybody's a scientist. Everyone talks about good data. What exactly is good data? Ooh, that's a really good question. and I guess it's going to be, I think in in some ways it's going to be specific to the sector or the the field or the science, right? But I would say in general, what is good data? I mean, I think good data is, has, now we're not going to get into all the statistics here because we'll we'll be here all day. and It'll be super fun, we'll ah ob be all right.
00:06:40
Speaker
um We like numbers here. We do like numbers, right. So what so what is what what constitutes good data? I think first is its the sample selection is neutral and objective. um That doesn't mean it doesn't have outliers or it you know necessarily follows a normal distribution, um but the sample itself is is not biased. um um you know We understand who did the survey and how the survey was conducted um and how the data were collected. I really do think that's an underappreciated part of working with data is understanding how things are actually collected. I mean, we know
00:07:22
Speaker
I mean we know right now I'm not again I'm telling all the things I'm not so I like I'm not a survey methodologist but we know that people answer surveys differently if it's pen and paper versus on the phone versus on their cell phone as a you know type again versus on their desktop we know those are different where you know people answer questions differently depending on the order of the survey, depending on what the options are and individual questions. So really understanding the data that you're working with and how they're collected, I think that's really the key. And you know, it's a kind of a tricky question. I mean, we're we're like end of August here recording. And just a few days ago, the Bureau of Labor Statistics came out with like massive
00:08:07
Speaker
um ah revisions to the unemployment rate data. Like, massive revisions, right? And so, if you said to me, yeah, the, bla do you know, do you think the Bureau of Labor Statistics has good data? I would say, yeah, it's like, it's the best, right? Like, they are the best. And then they have these massive revisions. And I don't think that means that they're bad. It just means, like, if I use those data three weeks ago, I'd have to redo my analysis now because the data have not seen so I think, I guess, part of this is depending again on what you're doing and what your sector is and what your goal is data is kind of a living breathing thing and you have these changes and updates and edits and, and it's it's hard to
00:08:51
Speaker
You know, keep track of and i mean it's hard it's hard work. I mean, I guess i'll I'll end it there for folks who are, you know, maybe who are just getting started in the data world working with data. And it doesn't matter if you're, you know, a school teacher looking at the survey results of parents or, you know, you're downloading data from the Census Bureau. It's all hard because all these questions apply.
00:09:13
Speaker
And I think if you're getting into this journey of of being a data person, I think really understanding what the data are or where they come from, how they were collected is really the first step. Sure. Now, what are some of the most common mistakes that people make when creating data visualizations?
00:09:32
Speaker
So so i I struggle here, because it's a really it's a really interesting question, but I think my, like a lot of people's knee-jerk reaction to that to that question is here are some specific rules, right? You need to follow this rule and not follow this rule. but But data visualization is interesting because it's a little bit of a combination of art and science. And so there is a science of it, particularly around how you use data. There's also art. So how do you put rules around art? So I think that's like,
00:10:00
Speaker
You know, I would just I would just caveat my answer with that. So I would say instead of I will give you one rule. This is the only rule that I believe in and that OK, well, let me let me plus was a like should pie charts sum to 100%?
00:10:14
Speaker
Yes, yeah they should sound to 100%. Is that a rule or is that common sense? I don't know. Like, I kind of just feel like that's common sense. But if you want to call it a rule, like, okay, the only rule that I really believe in is that bar charts, horizontal or vertical should start at zero, the axis should always start at zero. And that's because the way we read or understand a bar chart is by the length of the bar. And so ah the thought experiment that I like to do is create a bar chart of just two bars and imagine the increments on your y-axis are each, say, an inch.
00:10:47
Speaker
and calculate the height of the bars when the bar started zero. And now start to graph it something other than zero, have the same inch height on each increment and recalculate. What you'll find is that you have this overemphasis in the height solution bars. So just a little bit of math can get you a long way here.
00:11:04
Speaker
So what, you know, what else makes a good graph? I mean, for me, I look for a few things in, in, in visualizations are still like the phase that took my years off the life when they like warning signs. I guess I call these warning signs for people like trying to destroy July things like percent change when maybe the levels are meaningful, right? If you say, oh, this thing grew by 300%. It's like, okay, yeah, it went from one to four. Like, is that meaningful for one person to four people? I guess maybe that's not meaningful, right? It depends on your population size and the data, right? Right. Exactly. Exactly. So it it totally depends. But like, uh, you know, those, that's the sort of thing that like, especially if it's not a topic I'm particularly well, especially if I am familiar with it, but, but also if I'm not familiar with it, like,
00:11:58
Speaker
Okay, are they showing me the percent change because this, you know, we're starting with a small baseline and maybe that the growth doesn't really matter. So that's one, um log scales, not that I think log scales are bad, it's just, you know,
00:12:12
Speaker
A lot of people have a hard time interpreting log sales and so you don't see them a lot but that's but that's one. um Another one that I love and I have so many examples of these ah is when people overlay regression lines on scatter plots.
00:12:28
Speaker
Because it's so easy. Like, Autumn, I feel like it i feel like you and I can could talk about this world. Absolutely. Right? Like, it's so easy to be like, oh, here's some polynomial. And now you've got this weird line that makes zero sense. Or the opposite, where it's just a linear one, right?
00:12:45
Speaker
Or they just figure out what's the line of best fit and it is just not going according to the data. Yeah. I mean, there's all like, that's the thing about regression models, right? Is you can, you can play around with the model.
00:13:01
Speaker
And then, I mean, you know, a dishonest person can get a model to fit and they can say, oh, I ran a regression and lots of people do like, Ooh, that's pretty fancy. But you're just what does it mean? Right. What does it mean? Yeah. Right. You're just the US kind of line. Yeah.
00:13:16
Speaker
And then I guess the other one, this is sort of just a general one, um but comes back to the bar chart point, is anytime someone has made an arbitrary decision in their graph, this is it sort of just a general one. An arbitrary decision to me is the one to look out for. So the bar chart not starting at zero is a great example. If you start your bar chart at something other than zero, you've made some arbitrary decision or where to start. it You started at 34% or $699.
00:13:44
Speaker
That's arbitrary. And so why did the person do that? Right. A really good example is ah there's this website. And I don't actually know if he keeps it updated anymore, but it's this this guy's named Tyler Vegan. It's V-I-G-E-N dot com. and It's called Spergus correlations. You're nodding. So you know this site. This site is great. Yes. And all he does is create these dual axis line charts, so one line to one axis and a line to another axis, and just plays with the ranges and scales of the axis. And if these lines look like they're lying on top of each other, when, you know, they're seemingly totally separate trends, but yeah again, you like you these arbitrary decisions about what range you're going to use, what min and max, what increments you're going to use, you can make these kind of look the way you want them to look. And so those are the things that are, that are like, that I look out for when I'm well i'm looking at people's visualizations.
00:14:38
Speaker
especially when you're doing policies and procedures and you need to make those decisions right then and there. Right, exactly. So do you have another example that can talk about data visualization and how that has a significant impact on policy decisions?
00:14:58
Speaker
Yeah, I have a couple. I mean, it is a little tricky, right? Because you want to be able to say this kind of like to our regression point, right? Like you want to be able to say, here's a graph that affected this thing that affected this outcome. And life just isn't that that simple. So impact in that way is hard to measure. But I will tell you the story. i I made an infographic, which is like, you know, infographics back in the day. And we'll let this big tower things or combinations of graphs and text and pictures and all this stuff. And I made several of those and and and discovered, I think pretty quickly that people just won't even reading these tower infographics. There's just so much to scroll through. It just wasn't just wasn't natural. So I made an infographic.
00:15:43
Speaker
that um accompanied a 150, 200 page report for at CBO. It's just eight and a half by 11. The whole point was to print it. CBO, members of Congress liked paper. So like, let's, I mean, back to the very beginning of this conversation, right? Let's make things that meet people where they are, right? If you're in a print town, print stuff out. If you're in a, you know, an area where there's not broadband, your big old dashboard ain't going to work. Sorry. So like meet people where they are. ah Sure.
00:16:17
Speaker
So I made this ah made this one page infographic and we delivered it to you know to members on the hill and a few months later the director was was testifying and the ranking number of the of the of the committee like held up that printed infographic with likes his scribbles all over it yeah exactly it was like pumping my fist I was like this is amazing because I did yeah you're like okay so like does that mean You know, there's this particular outcome. No, but you have provided the information to the person so that they they can use it. Look like and I am by no means saying members of Congress are dumb or can't read, although I could maybe make that mistake that I'd like.
00:17:04
Speaker
they're busy people. yes They're busy people, right? They've got a lot of stuff on their plate. And so they're not going to read 150 page report because they have to read 12 of them a day. Right. So like you're thrown so much information yeah and policy and I would say even sometimes in academia, when you're teaching, you're making quick executive decisions, even as a CEO or CEO all of a company, right? Yeah. Sometimes you have 501 things. What is the simplest form of data or policy or procedure that's going to come across and just be like, here, look at this picture. Right. And that's what clicks. That's what sticks in your head. Sometimes it's not all of that fine text.
00:17:51
Speaker
I mean, if you're yeah if you're the CEO of a company, you don't need 150 page report to dive into the data. That's what your employees are for, right? At the end of the day, they're they're bringing you the recommendation or the answer or or whatever it is, right? So like, here are the five numbers you need to know. And again, it's not a knock on anyone's intelligence. It's just the way the world is that we have lots of competing demands and depending on your role, again, you might be the the budget analyst for organization XYZ and your job is to read the 150 page, but maybe the president of your organization, they're also not gonna read it. And you know that's okay, we've all got lots of things. So um yeah, so this this you know this one pager,
00:18:40
Speaker
Again, it wasn't like, oh, he has this one pager and therefore we got this outcome, but it was clear we need this one pager and he read it and he used it and he took notes on it. And that's, that's a fist pumping win right there, right? That's, yeah that's what we're, that's what we're trying to do. We're trying to get information that analysis data into the hands of people who can use it. And, and we do that the best we can and in a clear, objective, and useful way. And then we hope that they don't want to make the right decision.
00:19:10
Speaker
Definitely. Now, with that, where do you see the role of data visualization evolving in the next five years, maybe even the next decade? Yeah, I mean, I think the first answer has to be how is, well, it's going to happen with AI, right? I think that's the first clear question, clear answer.
00:19:33
Speaker
um You know, a lot of data visualization work is done through code, done in programming languages, R, JavaScript, Python, Svelte, several others.
00:19:45
Speaker
on you know, AI makes the coding task easier. It does. You know, you just, you're right. I mean, just like, hey, write me a code that does this thing. And you know, I use it for a lot of my, actually stuff when I do like more advanced stuff in Excel, I use it to write VBA code and like, it's generally right. Like, I mean, you know, so I think that's the, that's the first place, certainly on the coding, please. I think we're already kind of almost there.
00:20:12
Speaker
Then I think the next question is, how does AI get embedded or integrated into the data visualization tools? This is already happening, you know, Microsoft has their co-pilot tool, Tableau has something similar. um But how does it get integrated into the tools where you can say, you know, hey, here's my data, show me six options or six variations.
00:20:35
Speaker
Or, okay, great, ah show me that bar chart you just created using a style that, you know, Autumn would use, right? Like, I mean, that's that's the right that's the capability it's gonna have, right? Or, you know, okay, make that bar chart in the and the form that, you know, the Economist or the Washington Post would would make it. um You know, I think,
00:20:56
Speaker
That's sort of almost the low hanging fruit in some ways. I think a lot of tools kind of already do that. I think the next stage is how is AI gonna enable, how is AI gonna help us see more and better insights into the data, right? Especially if it's complex and with a lot of different trends, it could show you a lot of patterns that you're not picking up yourself.
00:21:26
Speaker
Right, and and I also worry, i mean we know there's bias built into these tools already, and I worry a little bit about You know, if you have, you know, again, the tool doesn't know the quality of the data. So it's just going to use the data as if this is right. So there's that caveat. But I think it's also, if you're asking the tool to get, or I guess be informed by other sources or the resources, what is it pulling from? Like, what is it looking at? um What other literatures is it reviewing? You know, if you were to say, you know,
00:21:59
Speaker
I want to i want to now analyze these 12 variables. What are some recommended models? right What is it pulling from? like What is the the research basis that it's that it's reviewing? right Is that bias towards you know um We can make it simple, like US s research versus UK versus Asian you know research in Asian countries. um Is it just looking at certain fields? you know like There's a lot of questions, I think, there. um and And I'm curious to see, I think the other thing that I'm curious to see how things will evolve in the terms of the future of data is is how our um Our technology and not just our ai technology, but like our physical like our phones in particular, like, how are our phones going to change in the next several years. That will either facilitate or not facilitate our use of data in different ways. And by that, I mean.
00:22:53
Speaker
um We saw kind of in like the early 2000s into like the, you know, early 2010s, like this real huge increase in people making interactive data visualizations on every book, every board or every line chart was interactive. You could click, you could swipe, you could do all this stuff. And then start folks started to realize like, okay, everybody's on their phone.
00:23:15
Speaker
No one's like filtering and searching. So we don't need, no one's clicking. So we don't need like all this fanciness. Let's go back to static graphs. And then the phones got really big and then they got sort of smaller again. And so where, where are we going to end up in the next few years?
00:23:31
Speaker
Right? Like in terms of size, speed, different platforms. I mean, you know, and then, and then I guess the last thing I would say is, you know, where is augmented reality, virtual reality going to fall into this? Um, I've seen some really interesting use cases. Um, and there's a lot of research going on around the world and especially in AR, um, to kind of give like this, I don't know what, I don't know what, I don't know what these folks would call it, but to me it's like, it's almost like.
00:24:00
Speaker
virtual tactile visualization so I can actually like. Take the board, like it's a minority report, right? Where I could take the board, I could zoom in, right? In the in the sort of the air, right? So is that- It's more interactive. Yeah. And it's not mouse-based. And there's some, there's also some advantages there from an accessibility perspective, right? If you're unable to use a mouse for lots of different reasons, where I could break my arm tomorrow and not be able to use my mouse, um but I might be able to use my hand to zoom in and zoom out and pan in school.
00:24:32
Speaker
Yeah, I mean, I think there's some, there's some interesting use cases there. And so, you know, there's a lot of interesting things that are going to happen. um And we'll just, you know, I'm not scared of the terminators yet. So I think, you know, we're, it'll be interesting to see. I think something that you haven't really thought about also is with large amounts of data is going to be the security.
00:24:57
Speaker
that you are going to have to think about, are these going to be, you know, we just saw Microsoft have a whole collapse with their system. Right.
00:25:09
Speaker
for all the transportation for airplanes globally. So what does that mean ah for even Google, right? yeah So our largest platforms are Microsoft, Google, and there's one other. okay And most most government things are on Microsoft. Other systems are, and even if you're talking about crypto, it's AWS.
00:25:35
Speaker
e Right, right. So that that really brings in some questions of where's the security going to lie? Is it going to be in these systems or are they going to be smaller towers that are like stored for companies when you're doing all this data, right? And I didn't I didn't back up from that, right? Because if you're worried that your data is at risk,
00:26:01
Speaker
are you as willing to answer a survey? There are federal laws on the books that say what the Census Bureau can and can't do to protect people's privacy because they know if they say, oh yeah, we're just going to share all of your information, you're less likely to answer that survey. So if our data is at risk through not the Census Bureau's fault, but through how they are stored and shared and and that sort of thing. Are people just going to need more reluctant to answer surveys? And that's going to hurt our data quality. So, there's yeah, there's a lot of... And and and i focus on the I keep mentioning the Census Bureau, I know, because um that's just you know where where my world is. But you know if you're going to do the school teacher one, if you were if you're in a school teacher and you're plucking data on only do if you're collecting data on kids,
00:26:57
Speaker
and your data are stored in the cloud. And that data is lost or hacked or revealed. Like, you know, there's a there's a lot of of um considerations here about data privacy that medical systems. Yeah, yeah, it's yeah. And it's hard and it's hard work. I mean, this is hard stuff to do and to do it well. Of course. Now, with that, what are some of your favorite tools or software for creating data visualizations?
00:27:26
Speaker
Ooh, okay, so that's a great question. um ah um go ah I'll tell you my the first one I use, and a lot of people are gonna cringe, um but I use Excel for a lot of my work. I was going to say, I knew that you were that was the first thing that was going to come out of your mouth. i'm like I almost wanted to have a flashcard. Well, I mean, you know, a lot of people are like, oh, Excel's the worst. Excel makes bad databases. Like, first off, Excel doesn't make anything. Like, we make this stuff. Yeah. Like, needs cell control. Excel has a couple auto functions, which are not fantastic all the time. Yeah, I mean, look, it's good for what it's good at. If you need to make
00:28:08
Speaker
a bar chart for your report excels great at that. um I would never use it to like make some sort of interactive data visualization for a website but that's not what it's for. So I think there's when people talk about tools they get I think a little bit too much in the like you know what's like objectively the best and I think it's again just kind of a use case in an audience but i I'll say i have ah I have a few that I do use so I use Excel a lot um a lot of the time.
00:28:37
Speaker
um I use ah the R programming language, um yeah for particularly for maps. um I will say I'm not a great Rcoder, um but that again is what's great about knowing lots of people who do lots of great work. You can always lean on your friends, so that's that's great.
00:28:53
Speaker
um ah So I use, so the library and in R is ggplot. So I use ggplot for maps and then some other graph types that are relatively easy to grid in R. I do use Tableau for some dashboarding work. um And then I'm sure people are like, why not Power BI? um Power BI, for those of you who know, the Microsoft version, I think it's it's fine, but it's not on the Mac. And I'm ah i'm primarily on my Mac, so...
00:29:20
Speaker
Sorry, like Apple and Microsoft get over it so we can get the car behind the map and more people can use it. And then there's two, at least two other tools that I use fairly regularly. They're both based in the browser. So bottom of your point about data security, these are tools you would not use to secure data.
00:29:41
Speaker
because it's saved into the cloud. But and there are two tools, Data Wrapper with a W and then Flourish, um the Flourish Data of Visualization Tool. Both are really nice tools. Both are based in the browser. Both have pretty broad um ah libraries of of graphic types. And what's really nice about them is that you can create a graph you can grab an embed code, drop it on their site pretty quickly. That way it's responsive to different sizes. you know it has the They both have interactivity with them, but like, okay, if you know if it's gonna resolve the response the responsiveness issue that like a JPEG or a PNG is not, then I'm kind of willing to have a little bit more clickability on it, even though I may not need it. so
00:30:30
Speaker
Yeah, I think that's like five tools that I use. I'm not, again, I'm not like a web designer. of The tools that that are datavis our data visualization team use at at Urban,
00:30:42
Speaker
ah for a lot of our online immersive storytelling, storytelling things ah is JavaScript, books particularly D3. And then um also Svelte is the other one that they're using with a lot of things like Node throw in there, Node.js throw in there and a bunch of other things that I do not code in. So, um but yeah, there's a lot of data, there's tools out there, but those are the ones those are the ones that I pretty much rely on.
00:31:06
Speaker
Those are the ones that you get a little more interactive on the website. So. Yeah. So if you go, like if anyone goes to like the New York times or Washington, but or really any like major news organization and there's some sort of interactive visualization or like we scroll through and something animates and move around, it is almost certainly built in D free. Yeah. Um, which again is JavaScript. Um, almost, almost certainly. Um, so.
00:31:32
Speaker
Uh, which again, like, you know, if you can code in that language, uh, more power to you. And again, like, so, so I think here's like the flip side, right? Of the, of the Excel makes bad database Excel's been what it's good for. You wouldn't use D three to do the analysis, right? Cause that's not what it's good for, right? You wouldn't do your analysis on almost and then you, that's a data visualization pool. Same thing. Um.
00:31:53
Speaker
ah here's and another good example Flourish and Data Wrapper, both tools I really like, they are not spreadsheet tools like Excel. So you can't like load your data into either of those tools and say, okay, I want to multiply this column by four. You you have to go back to your your data tools, which could be Excel or R or whatever tool you use, and then bring the data back in, right? Those are data visualization tools. In Excel, and it's the reason, one of the reasons why i've I've always used it is you could do your analysis and your visualization right there in the same
00:32:27
Speaker
Uh, uh, same environment, um, and. Uh, I know lots of people who are like, well, that's what ours for and you should do it in our and programming is. More stable and more reproducible and and I don't and I and I do when I totally agree with that. Uh, the only caveat I guess I would add is a lot of the folks that I work with. Um.
00:32:48
Speaker
uh, tend to be like small nonprofit organizations or advocacy organizations. And like they have a data person and the data person is the data person cause they like showed some affinity of working with like Excel or Google sheets. Right. And so that's what you' happens with a lot of cases. Yeah, yeah, exactly. And so that person's not going to like go learn R or they have to do their day job. Um, and you know, they're not going to be able to hire someone to do that. So everybody has Excel, you know, pretty much everybody knows how to make the basics. And so my goal in, in helping folks with Excel is just to show them how to push the boundaries a little bit. Cause the, the rules in the matrix, then in the major in Excel or like the matrix, some could be bent and some can be broken. And so.
00:33:32
Speaker
um that's how i sort of but That's how I sort of approach the the tools world, yeah. not Sure, I totally get that. Out of curiosity, how do you think policymakers can be better trained to interpret data and use data visualizations? I mean, I guess, but are policymakers any different than anybody else?
00:33:57
Speaker
Okay. How can anybody, anybody? Yeah. I mean, when I tell folks, because a lot of the times when I teach, my teach data is one of the things that I lean on is showing people lots of different graph types, because we all know line bars, pie charts, bar charts, because we learn those in, I don't know, whatever third, fourth, fifth grade. It's not like human beings have the bar chart DNA strand, right? Like we have to learn these. So, um,
00:34:26
Speaker
I mean, I think about, you know, my undergraduate work, my undergraduate work. Again, no one taught me how to make good graphs or how to make good presentations or how to write well. I think part of what we need to do in the education, especially in sort of like the advanced degrees or or and and there is to ah teach people how to be good communicators.
00:34:55
Speaker
right? It's not just the eat of his life again, how to be a good writer, like how how to be a good storyteller. How do you Yeah, like all these and it matters for every job you're going to go into like, yeah i you years years ago ago I have a ah friend still who who' still CBO, um and she manages multiple people. And she's like, it shocks me how these, you know, early career folks come into meetings with me, and they're unable to like, have like a meeting. And I think like,
00:35:21
Speaker
being able to communicate in that way. And that's not a presentation in front of an audience of 5000 people or 100 people. It's a one on one presentation. And that's a skill. Yeah, I mean, my, ah I have ah my oldest, my oldest ah kid is in as a senior in high school. And, ah you know, one of the things and you know, she has lots of friends who have already gone to college. And one of the things that we We've been talking to her and lots of others, kids as they go into schools. like and We hear this from a lot of admissions officers at schools. Don't be afraid to go talk to your professors. like you You should go to office hours. like Even if you don't have a burning question or a problem, because talking to professionals, talking to your professor, talking to senior people is a skill. um and I'm sure there's lots of educators who listen to this, like talking no students talking, it's a skill you have to learn.
00:36:13
Speaker
It is. I find that to be crucial for networking, especially i so some some of the classes that I taught were 150 freshmen in a course, that's best right? but Biomedical engineering, 150 freshmen, yeah and fresh out of high school, never interacting with somebody. ah You get these professors that come in,
00:36:43
Speaker
older, I'll say older, 50s, 60s, very professional and seem kind of harsh and unapproachable until you get the op and their office hours, they're sitting cross-legged on the desk saying, Hey, what's up? um And you don't get that interaction until what, eight weeks in the semester? right or Or even if they are more formal, like yeah you're going to graduate and go into the workforce and you're going to have a boss who walks to the, what some boss is going to walk to the office of no with no shoes and socks and another boss who is going to wear three-piece suits every day. See, and I've had professors i've had no professors with no shoes on.
00:37:26
Speaker
Yeah, yeah, i'm ah that I'm not a fan of, I'll just I'll just say that right. I don't want to see your feet. That's I'm just putting that out there. right Yeah. um But I think, you know, you asked about policymakers, but I think it's really anybody. And I think especially if you're you know young, early career, particularly college students,
00:37:44
Speaker
um you know, or or even, you know, high school students, like go talk to your teachers like that. And that skill, you never perfect that skill, right? You just become a better communicator. You're able to have conversation, be able to guide conversation, learn how to, you know, have facial, you know, ah you know, look in people's eyes. I mean, I think the technology we have, which is great, you know, we have these supercomputers in our pockets, but, you know, we sort of lost, I think, a little bit of that face-to-face communication skills. And um those are still necessary. And essential. Yeah, they are. They are essential. And so, you know yes, I would love for us to...
00:38:29
Speaker
I would love, for example, for calculus to be ah way less emphasized in high school and for statistics and probability to be way more emphasized because you know calculus is for the engineers and for the economists, but for most.
00:38:45
Speaker
people, most kids are not going to need calculus, but everybody needs to understand how probability works. Everybody, you know, you know, we're in the midst of a presidential election here, right? Everybody needs to understand that when you see 42% and 41%, that those are statistically the same thing, right? And, and, you know, calculus doesn't really help you do that. um So I would love, you know, I would love to add data visit into that, because I think that makes for a more educated, more data literate society.
00:39:15
Speaker
um But I think to your point, it's just about being a better communicator and that takes practice. And I don't think you can start early enough about learning how to communicate with people. Now, what ethical considerations do you take into account when creating visualizations for policy purposes, for procedural practices, or just to an audience?
00:39:42
Speaker
Yeah, I mean, this is an area that I've been thinking a lot about in the last few years, particularly starting with the racial justice protests in Spring of 2020. And there are a variety of ways that I think about this, and and I think they're sort of like The simple to the more complex and the simple, not that it's easy to do, but sort of like maybe the more straightforward is the best, better term. It's like, what words do you use to describe people in communities? Like, do you just describe yeah people by their skin color or do you say, you know, Hispanic?
00:40:17
Speaker
Employees or Asian students, right? As opposed to describing people like as their color. And I think yeah like an an interesting one that. We've seen this change over time is like moving towards people first language. So we use like people with disabilities rather than disabled person, right?
00:40:35
Speaker
Yes, but there's nuance and subtlety and evolution in language that makes like coming up with like a rule really hard because there are lots of ah groups and people who prefer the identity first language like disabled person right autistic person. um I have an example in my notes somewhere of um There's an organization that works on behalf of autistic people and they surveyed their numbers and like 80 something percent of their members said that they prefer identity first language rather than peer person first language. So these are like really hard questions. um But the overarching
00:41:14
Speaker
I think thread through um this body of work that I've been doing at the Urban Institute for the last three years or so, which we call our Do No Upon project. So you have about six, seven reports on this. um the The sort of thread that pulls through everything is to just have empathy for people, which I think is just like a good rule of thumb, just yes living. um But for Data and DataViz,
00:41:37
Speaker
is would you like here's the thing that I just I just tell everybody like would you feel offended if you were represented in that way in the chart the graph the diagram right if you were described in that way in those words if the color in the graph was assigned to you as a person how would you feel if you were represented in that icon of that person how would that make you feel right and I think that's a very I was going to say simple, but it's not really that simple because we all have our own experiences. um So it'd be hard for me as a, you know, as a white man to sort of, you know, know how, you know, a black woman would feel and seeing a graph in this way. But you, I think, you know, that's what empathy is. And you try to just put yourself in someone else's shoes um and think, you know, the thing that you can do is say, okay,
00:42:28
Speaker
is this icon, this graph, this word, let's make it my identity. Is that how I want to be described or shown? And that's sort of like the beginning. And then you kind of move into more depth from there. So how are you going to order the bars in your bar chart or the entries in your table? Are they going to put white and male in the top row or the first bar, which is how Census Bureau products are. They put white is always option one in the race question on race. So but we don't have to follow that ordering. that's just That's just based on structures and institutions we've had in this country for 200 years. So can we maybe think about, well, maybe I should order it based on the
00:43:14
Speaker
the magnitude will sort by the bar plate or sort of alphabetically or, you know, another one that's not, you know, I mean, every survey that the Census Bureau does on on gender. um They're all binary, by the way, it's always male and female and it's always coded male is number one and females number two, which is not alphabetical. I mean, I guess it's it's binary. So I guess it's not, but it's not alphabetical in the way we think about it. yeah It's not alphabetical and it's not.
00:43:40
Speaker
um It's not ordered by population size. So it's just saying, well, men go in the first position. And, you know, I think that's how we can start to think about being more equitable in and inclusive and in the work. And I'll just say here, like, I do get people who push back on some of these ideas that, you know, it's just, oh, it's just woke and it's just, you know, liberal, blah, blah, blah, blah, blah.
00:44:03
Speaker
and And I'm happy to take an ethical moral stand on these issues, but I'm also happy to take kind of like a capitalist stand, right? If I'm selling a product,
00:44:15
Speaker
Um, I want more people to buy that product. And if people feel respected and they feel valued and they feel like you are being empathetic to their lives, their experience, they're more like, I mean, I don't think you need science to know this, right? They're more likely to buy your stuff if they feel like you can value them. And, um, so, you know, I'm happy to fight on both those hills, but, um,
00:44:41
Speaker
Uh, yeah, I guess I would just kind of like wrap up. I like, I mean, like the empathy piece is what kind of drives all this. And that, and that gets you a good chunk of the way there. It's just speaking. How would I feel if I was shown in this way, you know, this icon of this.
00:44:56
Speaker
but like an icon of a woman in a dress with a baby pink color like is that right how we want to represent y'all is that everybody yeah right yeah is that everybody is that how we want to represent people do we want to talk about the history of pink that goes with it it was a man's color originally okay yes you know this book yes Yes, so the secret lives of color. It's fabulous. Fabulous book. And I look I have I can't believe you just mentioned it. I have my bookmark on the chapter on pink, because she talks about how it used to be flipped. Try like the mid like post World War Two, it was flipped.
00:45:36
Speaker
It's the artist in me, okay? That's the secret behind the scenes. yeah I am also an artist, not just a mathematician. But that stuff that stuff is like these hidden stories, right? They're just fascinating. Right? Yeah. Yeah. Now check out the next one is the color green. Okay.
00:45:54
Speaker
All right. Uh, I don't know how far they go in. I don't remember how far they went into the book. I was flipping through it, but green is everything. Yep. Royalty, but also now, uh, green is a trustworthy color for businesses and brands. Right. So if you take it from like, uh, not just having purple for, um, a lot of Europe.
00:46:22
Speaker
as royalty. A lot of the immigrants, even into the 90s to 2000s, especially from Portugal, Italy, Spain, they would drive green cars when they came to America.
00:46:37
Speaker
e Because wow in America, they were building their wealth, but especially having an immigrant family. You were building your wealth, and that's that's some of the unspoken language of color. It's also interesting, like just the cultural differences in color.
00:46:56
Speaker
yeah Right? Like in in Western civilization where like red is stop, bad, you know, negative, whereas in Eastern cultures, it's the exact opposite. I mean, I, I've always, you know, my work is almost entirely US focused, US domestic focused. And I've always Like felt bad for people who work at like some international organization where you have to make like this chart has to work for your US audience and your European audience and your Chinese audience. Like that has got to be so difficult to think about, you know, your colors really matter and, and and are are just interpreted in very different ways. And that right that's just a huge, huge challenge. Not to mention.
00:47:37
Speaker
back to the equity piece on like accessibility, right? Like accessibility is also an equity issue. And so ah the native is world, I wouldn't or you would and have argued lots of places that the field is a little, I think a little over overly focused on color vision deficiency, like, and and yeah for about three but lots of types of of disabilities and impairments that impact people's ability to understand information or to use computers or to use definitely or whatever it is. Right. um I did, oh, I've been doing this project at work. It's a data physicalization project. So it's like, it's making, you know, making graphs with physical things. So it's like, here's a ah map of the United States in post-it notes, right? So each state is a post-it note. And here's a disc, little wooden disc, like put the disc on the state that you were born in, something like that. So we make these fun little graphs. And as part of some some more research into that, I spoke to a bunch of folks who do museum installations.
00:48:38
Speaker
And just thinking about the the variety of challenges you need to consider, right? How are kids, you know, you're going to build ah a table, right? In the Museum of Natural History, how are kids going to be able to see it? How is someone in a wheelchair going to be able to see it? um Is the room going to be light or dark? Like, I mean, there's just so many variations to think about. Definitely. It's really just kind of a fascinating sign, especially in the area.
00:49:04
Speaker
Now, uh, behind the scenes of what you just, you do on a daily basis, you also have policy viz podcast. Yeah. Tell us about it because I know that a lot of folks here, we do have a large audience. Um,
00:49:23
Speaker
We do different things, more engineering side, but also a lot of these people are in policy. They're in government and engineering and yeah also higher ed. So the stuff that we do is this also the things that we can learn from you.
00:49:39
Speaker
Yeah, so yeah, so so yeah, so thanks. Yeah, but I mean, my show is really talking to folks in and around the data visualization field for focusing on how do I communicate my data better? How do I communicate my data to have impact?
00:49:57
Speaker
um So lots of ah authors come on the show. um ah the One of the guests I had at the end of of my last season was Georgia Lupe, who's ah really well known for her sort of bespoke visualizations. She had a really, she ah she had this incredible story in the New York Times about long COVID um that was kind of looked like um like ah um paint brushes, like different yeah different swatches of paint. It was just just beautiful work. um Oftentimes i'll I'll try and get people on who um can talk about tools, like specific leopardy clothes and how to build tools. on I'm going to talk to someone from Envivo ah in the fall, right, which is the qualitative data data tool. I've talked to people who use R from Posit, the Posit
00:50:45
Speaker
a company to talk about how to use R. With doing data visualization and working with people in policy and communication, and I'll even throw in educators, science, communicators. What is the biggest thing that you want people to take away from this podcast, from this episode, and also learning about more data in general. Okay. And also listening to your podcast. Yeah, well, obviously, that's obvious, obvious. Yeah, I think and this is this is nice. This is is kind of where we started, right? So it's like, for a I think the thing to start with, is who is your, ah who is your audience? What do you want them to do? What are the insights you want for the final discoveries you want them to make? How do you power? How is your work and help them?
00:51:41
Speaker
learn better, do their job better, whatever you want them to do. And if you could just think really hard about your audience and and a lot of this is just as an evolution, it's a journey. you You put stuff out, you see what works, what doesn't work, you see what your audience needs, doesn't need, what they use, what they don't use. But I think if you start there, a lot of the other stuff sort of falls into place and that's, um and a lot of this is also just um learning and growing and trying and changing. i I get asked a lot to teach like data is you know 201 like a 201 class and I'm like the 201 class is just doing it like you just gotta you just gotta get in there and do it if I'm not going to teach you a tool how to code in this language or how to build a dashboard in this thing if we're just talking about
00:52:27
Speaker
poor principles of data this there is no actual one tool on it's just doing it and you got to get out there and you got to try it and some of it's going to work some of it's going to knock going work But I promise people that if they think really hard about who they're trying to reach.
00:52:44
Speaker
And even for for for your educators, right? I mean, all the educators who are listening to this, you know, who have taught at different grade levels know that sixth graders, ninth graders, and twelfth graders learn in very different ways, right? And so yeah they know that instinctively. And so we, you know, the rest of us have to learn that, that that it is about knowing who your audience is. and And the rest of it, I think, just kind of falls into place.
00:53:10
Speaker
Wonderful. Thank you, John, so much for coming on this episode of Breaking Math. Thanks, Autumn. That was a lot of fun. It was a pleasure having you.