#207: Data Visualization in a Low-Attention World with Philip Bump

As analysts, we conduct analysis on behalf of the business to (hopefully) provide them with clear and objective information to help with making decisions. We use visualizations of data and, when we’re really hitting our stride, we even tell data stories. So, how does that compare to mainstream journalism and the stories they tell, especially when there is data that can be visualized in support of the story or the analysis? There could be no better guest than Philip Bump, long-time columnist for The Washington Post, author of the How to Read This Chart weekly newsletter, and author of a soon-to-be-published book about the baby boom generation!

People, Platforms, and Posts Referenced in the Show

Photo by Keighla Exum on Unsplash

Episode Transcript

[music]

0:00:06.2 Announcer: Welcome to the Analytics Power Hour. Analytics topics covered conversationally and sometimes with explicit language. Here are your hosts, Moe, Michael and Tim.

0:00:20.5 Michael Helbling: Hey, everyone, it’s the Analytics Power Hour and this is episode 207. You know, it’s said that the human mind is more designed to fit things than it is for the truth. And in analytics, we often find ourselves challenged by the obvious in some ways, what should be obviously apparent is somehow morphed into a completely separate thing that someone at the table just believes to be true. Until we grasp how the mind works, this weird little truth is no end of concerning. So what can be done? There’s just so much data, not just in business, but in the world generally, and all of us have to be careful. We don’t just loop up the facts and information that kind of agree with our priors. Hey, Tim, you are a data and analytics guy. You ever have to change your priors based on new information?

0:01:11.7 Tim Wilson: Nope.

0:01:16.9 MH: Hey, perfect. So proving it out. And Moe, you live in Australia. So I assume it’s the opposite in Australia because, you know, the other side of the world?

0:01:24.2 Moe Kiss: Everything’s upside down.

0:01:25.3 MH: So okay, so the opposite, the mind works in the opposite way there. So that’s pretty awesome. They’re very data driven in Australia, which is awesome. I’m Michael, but we needed a guest. So someone who could kind of help us delve into this topic a little bit more. Philip Bump is a national columnist for the Washington Post. Prior to that, he also led politics coverage for the Atlantic Wire. His writing focuses on the data behind polls and current political rhetoric. He has a very popular weekly newsletter called How to Read This Chart. And his new book, The Aftermath, is available for pre-order now and will be coming out January 17th. Welcome to the show, Philip.

0:02:05.0 Philip Bump: Thank you very much. Happy to be here.

0:02:08.2 MH: Now, being a fellow Northeast Ohio person, did you fall into the Browns camp or are you a Steelers fan? Because you’re sort of in that gray area that might go either way.

0:02:17.1 PB: Yeah, so well, I’m originally from upstate New York. So you could also, I could have been a Bills fan. I went to high school in…

0:02:21.9 MH: Well, that would be acceptable.

0:02:23.7 PB: I went to high school in Warren, which is just outside Youngstown, which is basically about halfway between Cleveland and Pittsburgh for those not familiar. But I actually also went to Ohio State. And so my football loyalties are as an Ohio State Buckeye, I actually have very strong feelings about the NFL and I’m generally anti, which I’m not going to get into now because that’s one of the least popular things I can say. But so my football loyalties are with the great state of Ohio.

0:02:49.8 MH: But you, wait, you’re in DC right now, which has like the most like highly regarded NFL franchise. Oh, no, you’re trying to not.

0:02:57.4 PB: I’m in New York. I could also play in Jets, also the Giants. I could also play in the Pats. I can go anywhere.

0:03:03.0 MH: Okay, go Giants this year and you’re only an hour away from Philadelphia. So there you go.

0:03:08.0 PB: I would not play in the…

0:03:12.8 MH: This year, I mean, there’s…

0:03:13.2 PB: Bandwagon available.

0:03:15.2 MH: All right, we’re not here to talk about football, American football. We’re here to talk a little bit about how people consume data, data visualization, things like that. So honestly, one of the first questions I’d love to just jump into is, how did your career intersect with this topic and how did it sort of become the jumping off point in what you do today?

0:03:36.9 PB: No, it’s fair. It is a long, convoluted and unusual path. I have always been good at math, which is probably the starting point for a lot of people who have similar conversations. But I ended up, really one of the former things I did is when I was at Ohio State, I had a job in what was at the time called a computer lab, which no longer exists, which is where kids who didn’t have computers would go and they’d use a computer. And I worked there from midnight to 8am. This is right when HTML was emerging, when the web was emerging. I taught myself HTML. I learned how to do basic, you know, markup language coding and then eventually actual coding, coding. And so I created a blog and I had a blog, you know, about 20 years ago that I had made myself. When I found what I was doing is I was commenting a lot on what was happening in the news, but I was also doing a lot of sort of data visualization along with that. I ended up working at Adobe as a designer for a while, the makers of Photoshop, Illustrator, so on and so forth.

0:04:26.8 PB: So I have a design background, but all of my career stuff tends to orient around numbers. So I had this blog, I started doing writing on the side and ended up doing it professionally. And so, you know, I had this sort of mixed background where I was of my own volition doing analysis of the news sort of just for fun because I thought it was great. And then I ended up getting a job doing that for The Washington Post, which seems to have worked out.

0:04:50.2 TW: So can I ask, when it comes to, you know, large media like The Washington Post, I historically was under the impression that they had the data visualization team or group that would be kind of the contributors. But in your case, it seems like, one, you do a lot of writing that is just analysis that doesn’t necessarily have data visualization. You sometimes do writing that has visualization and then you have like your whole weekly newsletter that is all about visualization. Is that the norm or do you fall in kind of a weird middle tier where you do both long-form writing or medium-form writing, I guess, and the actual visualization analysis work?

0:05:37.6 PB: It’s a good question. And it is indeed unusual. We do have a great visualizations graphics team at The Washington Post who do, you know, when you see big, splashy, you know, beautiful, well-designed, impeccably designed graphics at The Post that that’s invariably who did it. But I feel like there is a need for numeric analysis within the context of news articles more often than the peers, right? And, you know, I just feel like when you’re writing about a poll to not actually show that visually is doing the reader a disservice. The Washington Post is a newspaper. And so as such, it was later in adopting a sensibility about how multimedia can be included in news articles than other outlets might have been, you know, particularly digital first outlets. But I come from that background. And so for me, it was very natural to incorporate this stuff into the work that I was doing. I knew how to do it. I, you know, can make a graph very, very quickly. And I can have it say what I want to say. And, you know, honestly, a lot of times stories follow from the data visualization rather than the data visualization necessarily driving the story.

0:06:46.0 PB: I’ll look at the data and say, “Oh, that’s interesting,” like, I did this story before the midterm elections that was looking at the extent to which Fox News was talking about crime. And there was a story that did very well and got a lot of attention, but solely based on my doing the data analysis, then visualizing it and being able to see, “Oh, look at this spike break here.” That then became the story. And that’s not something you can do if you’re saying to the graphics department, “hey, I’d love to have you crunch these numbers and see what you come up with.” And so, yeah, it is, it’s an unusual space. I think it’s a space that more people need to occupy. And I think it’s a space that for people who do this stuff natively, there’s a lot of opportunities in media, I think over the longer term, if you are able to bring to it a sensibility of how data can be visualized and how that can inform what it is you’re trying to convey more broadly.

0:07:32.1 MK: So tell me though, like, I feel in my world that I work in, I’m often trying to convince people, even data people about the importance of how you visualize certain graphics. And I don’t know, I’d like to think that there’s a gray zone, but I feel like there isn’t, that there’s people who get it and people who don’t get it. And I don’t know if that’s the right summary of an understanding of data viz. But in my experience, I think that’s what it feels like. In your industry in media, like, what have you found? Have you had to convince people that this is important? Or is because it comes like you have such a good background in it, and you know, it’s important. It’s just blended into your work, and you haven’t had to fight for it. Or do you feel like you do still have to kind of make a case?

0:08:21.6 PB: No, I don’t think so. I mean, what I do is unusual at The Post. And I mean, there is when I go to write an article, we have an upload tool that was, you know, the Washington Post created basically its own back end a couple of years ago. And so part of that there’s a back end tool where you can upload images. And one of the image categories is for columnist created graphics. And then it says in parentheses, for example, bump. So it’s like, it is unusual enough that in our back end, it is identified as being something that I do. So yeah, no, but I mean, I was doing this before I came to The Post, right? I went from the Atlantic wire to The Post, in part because this is what I did. This is what I brought. So, you know, The Post hired me knowing that this was part of my skill set. And so I don’t think they were surprised. There definitely were some conversations at the beginning of my career at The Washington Post in which I was I was gently and then less gently encouraged to make sure I was following the brand guidelines, and so on and so forth.

0:09:17.5 PB: But you know, I mean, the graphics team, as I said, is great. And, you know, I mean, honestly, I think that if I were to come to them, with every request I have, or want to include a graph in an article, they very quickly want to see me, you know, drawing a quarter. So that makes me happy having to do it myself.

0:09:33.0 TW: What are your tools? I mean, in the, in how to read this chart, you sometimes kind of allude to some of the especially on kind of the the geo type tools, but do you have a go to tool set? Because in how to read this chart, you often you’re like, well, this is another free tool that you can do this cool thing with you have a set of tools that you’re going to or does it? I mean, you must.

0:09:56.3 PB: Yeah, no, yeah, I do. I do. I have a, I do like to share some tools that I use. And we haven’t talked about those. But the core set of tools I use are two extremely expensive ones that are really not ones that most data analysts would use, which are Microsoft Excel and Adobe old school.

0:10:13.0 PB: But you know, so it is like, I do a lot of data now. I mean, like, look, I’m old school, like, I still do data analysis using pie code pearl and like, in order to do big assessments of, of, you know, text data, I use Excel, I’m just really good at Excel. And so it’s faster for me to use Excel to like, parse it to parse data. I use Illustrator because I worked at Adobe and I used Illustrator for decades. And so like, I’m just very really good at Illustrator. So I can make it, you know, but then there are online tools to like this thing that I talked about before in the newsletter, rawgraphs.io, which allows you to take tabular data and very quickly put it into different formats. I use that a lot. It’s really good. And it’s just fast. And you know, you don’t have to dig into or do anything along those lines, which is not my core skill set anyway. Is there, are there undoubtedly better tools to do what I do? Yes. Am I very fast at doing them using the tools that I have? Also yes.

0:11:02.1 MK: Oh, Moe, like insert Moe googling rawgraphs.io. I’m like, I feel like I’m gonna have to spend a couple hours digging around. It looks very cool.

0:11:11.9 PB: It’s great. It’s a great tool.

0:11:14.0 TW: Well, and it is and maybe this is a little back to where I think Moe kind of a slightly different angle from where Moe was. It’s interesting, coming from an analytics in the business world where we say, Oh, we crunch the data. And then we create and maybe experiment with different visualizations. And then we, what we, I’m going to speak broadly for the whole industry, drop it into a slide deck, and then try to tell a story. Whereas I’m kind of intrigued by the more journalistic approach, I think if I’m stating this fairly, so I’d love to get your thoughts that you’re maybe looking at the visualization, but then you’re literally writing a narrative, like actually figuring out what’s, what’s the lead? What’s the sequence, putting that rigor around it, and then I think the visualizations, depending on what that narrative is, drives what visualizations show up where, which seems super… I mean, it goes back to you kind of having a knack and a talent for it, but it feels like it’s a different mindset that would serve a lot of analysts well, to try to get into that space.

0:12:32.2 PB: Yeah, there are, I would say there, there are three types of stories that I do. And that sounds like I’m about to reveal something very fascinating, but it’s not actually what’s going to follow. The three types of stories I do are ones that don’t use data visualizations, ones that use them to supplement the story, and ones with center around data visualization, right? Which, yeah, okay, no kidding. Like, wow, great, thanks.

0:12:58.5 PB: But that’s important, right? So I do a lot of stories that are just like, you know, here’s what’s happening in the world. And here is, you know, one way in which you might look at it. Then there are stories I do where I look at patterns that I see in politics that then, you know, here’s a, here’s some poll results that show how this trend has formed over time, and so on, and so forth. And then there are ones that I did, like the aforementioned story I did about crime and Fox News’ coverage that are really centered around the visualization. And even if it doesn’t necessarily seem to the reader that it’s centered on visualization, it may nonetheless be, of course, there are times when I’m like, “Hey, look at this cool chart,” I just do a cool chart, right?

0:13:34.4 PB: Like, you know, those are, I guess, sort of a fourth category, subset of the third category. But, but yeah, I mean, it is, it is always and this is a little bit particular within the media industry, this is a bit of a hackneyed way of expressing it. But it’s true that I always have a question that I’m setting out the answer. And then sometimes the answer to that question is made more comprehensible through these adaptive visualizations. Sometimes it is isn’t.

0:14:00.3 TW: Is there any, generally speaking, does it tend, of those three groups, is there any sort of pattern when it comes to what gets I mean, visualizations feel like they share more easily than a headline, not that not that people can’t get incensed on Twitter about, you know, just the headline of your piece and not a visualization, but from an engagement or a comprehension, and I’ll say engagement, not just like and retweeting, but actually seeming to engage with the content, is there, do you sense a variation across those three types?

0:14:39.8 PB: I tend to be or I try to be assiduous about not actually looking at metrics. Because again, I think no, like, seriously, I think it is, you know, I think it’s more useful for me to stay true to what it is I think I’m interested in, because that served me well, than it is to be trying to chase what I think people are interested in.

0:15:00.8 PB: So it’s hard for me to say I have a vague sense of how things do because, you know, we have like, here’s what’s most popular list, and I can see what shows up on there or not. But I do, I will say I’m more tuned to it on Twitter where you can’t escape the metric. And so I will say that when I have something that is bolstered by an interesting chart, it does tend to do better on Twitter, but of course, Twitter also rewards visual stimulus in a way that other outlets, you know, even The Post itself doesn’t necessarily. So that’s a very long way of saying, I can’t answer that specifically, but it certainly is the case that there are times in which having a chart, even if it is a tertiary aspect of the story itself, does help it get more traction online.

0:15:41.9 MK: So I have to share a small anecdote, because it was possibly one of the most frustrating email chains I’ve ever been on. We have a good friend of the show, Emily Oster, who is brilliant, and has a newsletter as well. She’s fantastic. And she’s one of those people that like, lays out the data very clearly and walks you through kind of like her assessment of what it’s saying. And I had a friend who then replied to me basically being like, Oh, my gosh, can you believe this? And it was the complete opposite of what the article said. Like, I mean, it was the polar opposite. And I actually read it being like, how? How can someone have completely misinterpreted this from someone who’s a very clear, articulate writer? Do you find yourself in that situation? And do you get frustrated by it? Or like, because there must be times that you put out content that gets misinterpreted or potentially even misused intentionally. How do you handle that once your work goes into the wild?

0:16:41.0 TW: And in his Twitter, you just rationally, you just rationally explain that they’re incorrect. And everybody says, Thank you very much for clarifying and that was rain.

0:16:49.8 PB: That’s the plus side of social media. Yeah, I mean, look, I write about a lot of things that are very… So the things that I find frustrating actually aren’t data visualization things. They are instead things like debunks of obvious nonsense, where you nonetheless have people who crop up and, you know, insist that the thing you’ve just debunked is infact accurate, which it obviously isn’t. But I have an anecdote for this, which is that in 2009, I spent six months on jury duty in a high profile case in New York City. And at the end of that time, you know, after weeks and weeks and weeks of sitting in a jury and, you know, going back to the jury room with fellow jurors, I had a very strong sense of what had happened and what hadn’t, and expected everyone in the room to be on the same page as me, which is not how juries work. But beyond just that, even after 11 of us had reached a consensus, there was still one person who didn’t agree. And it became very apparent very quickly that, that this person was responding in not, responding emotionally, not rationally.

0:17:54.6 PB: And so we had to come up with sort of an emotional way of convincing her of what was rationally obvious. And it was really informative, because it was, you know, it was, you know, I was this nerdy kid. And, you know, I guess I was a kid back then. But I was this nerdy guy. And I had this expectation that people would respond in the same way. And then I was disabused of that. And I had to figure something else out. And so it is not surprising to me when that happens. It happens with a great deal of regularity, not really necessarily always about data visualizations, although there have certainly in times in which people have, it’s far more common for people to complain about a visualization than to misinterpret one. But, but yeah, I mean, this is, this is the nature of presenting an argument to the world is that people will miss the argument. And then you either can choose to engage with them or not.

0:18:47.3 TW: I guess that’s, I mean, that may be another kind of a difference between the world of media is you’re saying you’re pursuing what you personally have curiosity, or maybe a hypothesis or you’re kind of it sounds like in a position where you’re in the, you’re kind of in the service of your own curiosity with as long as you head in and with true, fine intentions, more often than not, that will produce a product that is of value to the readership. Whereas I think in an analyst a lot of time they’re in service of the curiosity or agenda of their business, the stakeholders who are asking the questions. And I think maybe that sometimes as an analyst, we start feeling frustration because really we’re in the service of somebody else’s curiosity and hypotheses. And we’re then overlaying our own. I mean, the grass is maybe always greener on the other side. And I’m like, your job sounds awesome.

0:20:00.2 MK: I would say the difference though, is that the business are often making a decision based on what we presented, which like you want the business to make the best decision and you’ve spent all this time, you know, working on something to make a recommendation. Whereas like in this case, like in your case, sorry, Phillip, it might be more about like something that someone’s interested in or something like that versus they’re actually making a decision about something.

0:20:26.9 PB: Yeah, yeah. I mean, look, I can’t speak to your world. I will say that you guys probably get fewer death threats than I do. So that was sort of the flip side of it.

0:20:34.1 MK: Definitely.

0:20:34.5 PB: So you know, look, I mean, you look, what I’m engaged in is absolutely an exploration of what I think is important and how, you know, what I think people need to understand about the world, but it deals with a very contentious set of issues. And, you know, opens you up to feedback that’s not positive. So you know, I say that mostly as a response to yeah, look, it’s not, it’s not sunny necessarily, as a general rule, but you’re right that it does afford me the ability to at least, you know, the mandate of journalism is ideally speaking truth to power that those in positions of power, be they in politics or in media or in, you know, the economy that you say to them, actually, this is wrong. What you’re doing is wrong. When it needs to be said, that’s important and rewarding. And I do get to follow my gut on that. But you know, there is obviously a downside. I know that doesn’t answer your question. But I can’t speak to your world. I can only speak to mine. So I’m speaking to it.

0:21:38.3 PB: And I don’t say that because I want people to feel piteous of me. I’ve known this a long time. I’m used to it. I’m just saying that I just want to be very clear that when the way you framed it, which is I would love if my job were like that, but it’s just, you know, especially in this moment, it’s all just often not.

0:21:57.0 TW: Yeah. And that’s a… I mean, I do.

0:22:00.6 PB: That was a total drag, wasn’t it?

0:22:00.7 TW: No, well, I mean, I’m a little bit of a, I mean, the listener certainly Michael and Moe know. I mean, I’m a little bit of a political junkie. So the, I see the journalism, I see that as being a you know, democracy dies in darkness as like a pretty noble, noble pursuit. So have you met Bob Woodward? No.

0:22:24.8 PB: I have.

0:22:25.6 TW: You have? Oh, man. Okay.

0:22:27.3 PB: He shows up a lot more in my emails when people are like, what happened to The Post that used to be Bob Woodward and Carl Bernstein and blah, blah, blah. I get that…

0:22:35.0 TW: Yeah. Bob Woodward never made a chart.

0:22:37.8 PB: Exactly. He wouldn’t use a pie graph for God’s sake.

0:22:42.8 TW: Yeah, I guess, but in some sense, your stakeholders are the American public, right? They’re just not coming to you with, I don’t know. This is, I do not think I’d put enough thought into the distinction between the roles heading into this. And so now, therefore, I’m thinking out loud, which is never good.

0:23:01.3 MK: Dangerous. Dangerous.

0:23:03.5 PB: You know, and look, if what this thing gets to that, you want to cut this whole section out, that’s fine. Like, I, I…

0:23:08.7 TW: No…

0:23:09.0 PB: Legitimately, it’s fine. Yeah.

0:23:11.3 MH: I think there’s an interesting dichotomy because in the world of business analytics, in essence, we’re supposed to be doing something somewhat similar in a lot of cases where we are also using data to, but it’s not framed as speak truth to power, but in a certain sense, speak truth to power or speak truth to business leaders or other leaders, which is the data shows that this is the thing that’s happening with clarity or the narrative that’s emerging from the data. And I think that’s, it’s kind of an interesting juxtaposition because in the world of business, well, yeah, we probably don’t get death threats.

0:23:52.9 MH: We do get yelled at and we do get fired and we do get dismissed. And a lot of times you’ll find pretty unhappy analysts who are sort of like in jobs where, you know, with the phrase quiet quitting is out there now. So we can use that one, like where they are just they’re undercover because nobody listens to them anyway. And it’s, it takes a lot of bravery, I think. And I think maybe that’s the common thread is if you’re going to be a great analyst, whether you’re going to do that sort of in the public sphere or within a company, it takes a lot of courage, I think, to continue to pursue showing people the truth, right? Or as best as you can. And I think maybe that’s the through line that we could kind of draw there because it’s certainly the risks or the profile is different, but it, it’s the function is, I think can be somewhat thought of as similar.

0:24:44.2 TW: Nope. I want Philip to be up on a pedestal and I want to be down.

0:24:47.5 MH: Oh, I absolutely, it’s a whole different thing, but I think there’s some, we can all claim some nobility, Tim, which is what I’m trying to give people.

0:24:58.2 PB: Look, look, you know, the, the, and I say this, so I spent some time, I was in AmeriCorps for a while and there’s a great Martin Luther King quote, which is, you know, that, that I was paraphrasing here that there’s nobility in all work, right? You know, and he uses the right, you know, if you’re going to be a street sweeper, be the best street sweeper you can be, right? And you’re right that these are both occupations and by both, I mean, you know, everything that sort of encompassing here that focus on representing the world as it is. Right. And I think that is a more noble occupation than say Tucker Carlson, whose job is not to represent the world as it is. Right? You know, like I think that there is absolutely nobility, no matter what it is and no matter what, that comes with saying, here’s what I’ve looked at and here’s what this says, and this is what you should know. That I think is a through line. And yes, the repercussions are very different. And, you know, I don’t want to get fired either, right?

0:26:00.2 PB: Like, you know, I can’t, I have to do a good job from that standpoint as well. So, you know, I think there, there are equivalences and I do, you know, I raised that before simply to say that, to point out that there is this added level of frustration that you get from doing the work that I do, then hopefully the work that you guys and your listeners do that no one should have. You know, my, my goal, I guess what I’m saying too, is my goal is not to have all of you guys get death threats…

0:26:31.7 MH: Yeah.

0:26:32.5 PB: And I can chart that too.

0:26:36.3 MH: But I think two things, right. It’s sort of like we all need to sort of follow that example. Like we see you setting in the public sphere, which is keep showing up, keep trying to do that thing. It’s awfully easy to sort of sometimes let it be like, all right, I won’t bring it up because I know we won’t get a good reception about this. Because so-and-so believes this and we can’t, we will never get headway on this topic. So I won’t even bring it up in the meeting, right?

0:27:01.8 MH: That kind of stuff happens in the business world all the time where people are just like, yeah, that’s not, it’s not a battle worth fighting. And I don’t know that we all need to sort of be, you know, chaotic in our approach to this, but you know, you have to kind of pick your spots, but also this idea of like, yeah, there is something to this beyond just sort of like, I crunch numbers at the service of whatever corporation or organization I work for, as opposed to like, I’m trying to bring and shine a light of objectivity of the data actually has some clarity and some things to say about what’s actually happening that may or may not agree with what you think of as the answer. It’s not a, by accident, the way I intro’d the show, which is all of humanity kind of like, we want to fit things into what we already think. And that’s the way our brains work, which is kind of fascinating, you know, in its own sense, but it’s sort of why we have to train people how to become good at this. It’s interesting. Do you see, because obviously you interact with so many people in the world, do you see sort of an increase of data fluency?

0:28:10.8 MH: I don’t like to say literacy, but like data fluency across sort of like, do you feel like it’s increasing or is it about the same? Like you’ve been doing this for a long time. Like what are your perceptions of the public, I guess?

0:28:24.2 PB: That’s a great question.

0:28:24.3 PB: I honestly, I’ve never really considered it. And that is, has some irony coming from a guy who started a newsletter that was introducing that, right? I think it’s sort of in necessarily the case that people are more aware of how to read basic data visualizations than they used to be simply because it must necessarily be the case. And I haven’t seen this measured, but it must necessarily be the case. People can count them more often than they used right? And you know, the year 1990, if you open the newspaper and there may have been a couple of charts in there, that, that might’ve been your exposure to it and a school book, something along those lines, which isn’t necessarily people it’s, you know, television news. Now, you know, you’re sort of inundated with them, particularly, you know, the tool set to make them is obviously broadly available. Now, of course, you have a lot more interactive things, which encourage exploration, which didn’t exist back then. But I think it must necessarily be the case. But I am a human being and the things that stick out at me are at the times when people don’t get it because, you know, we tend to look at the negatives more often than we ought.

0:29:25.7 PB: And so, yeah, I mean, it definitely is the case that there are things that I think are intuitive, like scatter plots. I think scatter plots are incredibly intuitive and I use them all the time. So I think they’re very useful. Just look at them and go, “What the hell is that thing?”

0:29:38.3 PB: And it’s like, “Look, man, there’s two axes.” It’s not complicated. You know, this is part of why I started the newsletter because I want people to feel comfortable. And so the newsletter is super light and it sounds like a plug for the newsletter, which wasn’t meant to be, but might as well. But, you know, it’s meant to be super light and easy to access and get people to just look at graphs that they wouldn’t otherwise see and say, “Hey, that’s actually kind of cool looking. What is it actually saying? And I still don’t really get it. And I get a lot of emails each week.” It’s like, “I still don’t really get it, but it’s fun to look at. And I thought it was a good story. And that’s good. You know, I think that’s fine.

0:30:04.7 PB: And I think that I have, you know, lots of people who willingly sign up for a thing that literally is like, “Hey, it’s, you know,” I wouldn’t use this analogy because I don’t think it’s apt, but in the 1990s, the X for Dummies books were super popular. And I was always just like, why the hell would you buy a book called X for Dummies? Like who stands at the checkout and has that in their hands? It’s like, “Yeah, I’m an idiot. Look at me right?” But I think that the way that we do this with how to read this part, it’s just more of like, you know, it’s more like expanding your awareness and expanding your knowledge. And I think, you know, even if I get an email from people that like, “Hey, I didn’t quite get it.” I think the fact that they signed up for it is great. And I think beneficial.

0:30:48.4 MK: Now that you’ve mentioned scatter plots, which you do like, and I also struggle with why some people don’t understand them because they seem really intuitive to me. And you can, I’m going to give you like a multiple option, choose your own adventure. Are there any things with data viz that you always do or that you never do? Like a couple of tips that you’re like, I definitely find myself regularly doing this, or I absolutely never do this.

0:31:13.8 PB: I tend to do a lot of sort of experimentation and I do data visualizations that don’t work. Right? Like sometimes I’ll do a visualization and I know it’s sort of pushing the boundary and I know it’s more complicated than it needs to be. And I just kind of like it. And I’m just kind of throwing it out there and God bless the Washington Post for giving me the space to do that. At no point in time has anyone ever come to me and said, this was a terrible graph. You shouldn’t have done it. So one thing that I do do that I encourage people to do is just try different stuff and try different visualizations. And some of the things I’ve had that have done the best have been things that were different than what I normally do. Like I remember when Ebola was emerging in 2014, I did a visualization at The Post that had, I believe, a visual representation of all 300 at the time million Americans. And then three of them were colored red to indicate the number of Ebola cases. And it was just this huge, like it was a massive pain on web browsers to load the thing.

0:32:10.4 PB: You know, like people clicked it and people actually wrote it and were like, hey, actually your JavaScript code here is, could be streamlined a little bit and be more efficient. So, you know, like the writers helped me code it a little better. But, you know, things like that, I just, it was an exploration of trying to figure out a way to convince information in a way that was useful that I think worked. Are there things I don’t do? Nothing that I can think of. I mean, honestly, you know. I’m all for more data visualization, however it works and whatever it looks like. I’m hard pressed to think of something that I would say no to.

0:32:46.7 MH: So you would say yes to a 3D exploding pie chart just for the record?

0:32:50.5 PB: Yeah, yeah, honestly, yeah.

0:32:54.1 MH: Oh, wow.

0:32:56.8 PB: No, for real because it’s like, you know, sometimes what, you know, if that works and it’s visually interesting, sure, who cares? Like, honestly, I mean, it’s harder to make an illustrator. You gotta use this weird devil control and stuff like that.

0:33:07.6 TW: It’s hard to make an R too, because you have to use the polar coordinate and the help even says don’t do this. But so, I mean, I do want to go…

0:33:15.3 PB: But I mean literally, who cares? I mean, here’s the other thing. I’ve never taken a class in design in my life. And so I have this sort of, you know, I don’t have any sort of, I’m not fancy or fussy about it. And I’m not implying that you are. I’m not calling you a pretentious pie chatter.

0:33:39.4 MH: Well, we do have a term for Tim. He’s the quintessential analyst. So, yeah, we are pretentious on this podcast. Thank you very much.

0:33:48.5 TW: No, I mean, the fact that, I mean, your first initial reaction of trying different things, like, I do feel like that is, it drives me nuts. So maybe I do sound a little pretentious when people are like, did I represent the data accurately? Yes. Good. I’m done. Move on. And I think that’s one of the reasons that I like even the basic, one of the premises of how to read this chart is, let’s look at something and try it a couple of different ways to see if it actually becomes more useful. And I think that takes a lot of discipline to get in the practice of saying, just because I represented it accurately, and it makes sense to me, I’ve got the curse of knowledge. I’ve been buried in the data. I know exactly what it’s representing. So I love that part. But on the scatter plot front, which I’m also big fan of scatter plots, but also have seen, like, if you take the old ice cream and drownings over time and you plot that as a scatter plot, you’re going to see this really strong inverse correlation. So kind of shifting to the question of the old correlation versus causation, which I feel like on the business side, people are starting to get a better understanding of some of the pitfalls of seeing a correlation and not like understanding things like confounding factors and factoring those in.

0:35:20.7 TW: Like when you’re, how much do you need to get into your analysis being causally valid with the numbers versus causally valid based on your analysis of what’s going on?

0:35:35.8 PB: So I just use a simple rule of thumb, which is that correlation always equals causation. Then I don’t worry about it.

0:35:40.0 TW: That’s good. Okay.

0:35:42.1 PB: That’s a joke. Yeah, I mean, I try and be very, very conscientious. The Washington Post is an organization that unlike a lot of organizations is necessarily predicated on conveying as accurate a set of information as you can. And so that means when there is a person who commits a murder and is standing there holding the gun and shouting, I did the murder until he is convicted. We say that he is the alleged murderer because we don’t know until we have a certain set of facts that have been established in a certain process and gone through. That’s is just what we do for our media organization. And so I have occasions in which I think causality is blindingly obvious. And I don’t say, and I still am very careful about saying, of course, this may not be linked because that’s what we do.

0:36:31.1 PB: This is an alleged causation, I guess you could say. And so I think I probably am more conscientious about that than I would normally be out in the regular world simply by virtue of working for The Post. And so as such, I am not often burdened by the question of, can I draw this line because I am assiduous of making sure I don’t unless it’s absolutely brilliantly blindingly clearly the case. That’s all.

0:37:01.8 TW: As I think about it, as the flip side, actually, there are times where there are other outlets or people who are making causal statements and you wind up tackling it as a, maybe let’s dig in where this might not be. I mean, you’ve done a decent amount of late on the crime front and there is, on the one hand, there is just like crime data sucks in the US for a myriad reasons. But I guess, do you sometimes find yourself doing that when you’re taking somebody who is declaring causality and you’re actually saying, one, there actually may not even be correlation here. So let me pick that apart. But also, even if there is, there’s not necessarily a causal link?

0:37:43.9 PB: I mean, I have spent probably 45% of my life since November 2020, making sure that people understand that presumed causality involving data and numbers as relates to votes being cast is not correlation with existence of voter fraud. Like the past two years have been a bull market on people who want to make sure that the world understands that correlation isn’t always correlation and much less causation when it comes to these claims that have been made. So yeah, absolutely. I mean, it’s just like I could have made a career out of being a guy who can do numbers analysis in the post-2020 world just because it’s a nonstop flood of nonsense that needs to be addressed.

0:38:37.7 TW: So have there been any examples in this? I’m also kind of alluded to, I’m kind of fascinated with sort of the journalism and journalistic standards and some of the things that will blow up where a journalist made a mistake or a journalist misrepresented something. Like have you had cases where you had a data go out that you did inadvertently misrepresent whether it was visually or not? And are you willing, if so, are you willing to share how that came about and what happened?

0:39:17.2 PB: Oh yeah, it’s a nightmare, right? I mean, like every time you put something out, it took me a long time to build the confidence that each time I published a story, I wasn’t like, “Oh God, what if I f’d up a number,” right? It just takes a while. You learn to trust yourself, but absolutely. My favorite example, just because it’s so goofy, is one time I did this story and this is why I love my job at The Post because they let me do stories like this. But one time I did a story, it was probably June of 2018, my guess, maybe 2017, and I did a story that was literally like, “What if when Donald Trump was going down the escalator in Trump Tower, the escalator were infinite and he never stopped moving down the escalator? Where would he be in space and time?” That was the story.

0:39:58.3 PB: And so I figured out like the orientation of the escalator relative to north-south. I figured out the rate of speed. I actually went to Trump Tower and rode the escalator to see how fast it went. And so I actually extrapolate it outwards and I showed… I illustrated in Illustrator, this big escalator extending out into space and how high it would basically be above the Earth’s surface. And I published the story and it was a lot of fun. It was just basically math and nerdy and goofy. And then this math teacher writes in and he’s like, “Love this story. It’s really funny, but you messed up your trick and you used… ” I was doing like SOHCAHTOA and I did like the CAH instead of the TOA or something like that.

0:40:39.3 PB: And he was right. Or it was something even less than that. Maybe even just the Pythagorean theorem. But I had shown a triangle and I’d gotten the sides wrong and I had to switch the sides. And so it ended up repositioning him in space and a whole different place. And I had to add this, append this whole thing on the bottom. And it was just the funniest correction I’ve ever had to do. ‘Cause I was totally flat wrong with what I ended up with. And the stakes were so low that it was like totally un-painful. So that’s my favorite story, is a massive correction I had to do, but the stakes were so low that it was fun to actually do the correction. But so yeah. Yeah.

0:41:11.6 TW: But is there… I mean, I assumed is the Washington, is there… There’s gotta be, is there a formal workflow when you actually need to issue a correction? Is that in the Washington Post corrections database for all time at this point?

0:41:25.5 PB: There actually isn’t a corrections database or there wasn’t at the time. Once we got the new CMS, there is now a correction tool. I’m not sure if that’s logged somewhere. So that one escaped that, but corrections take different forms. Sometimes it’s… I accidentally take the… I misspelled someone’s last name, for example, and that, we’ll often get a, “This name was written or misspelled.” Sometimes it’ll just be like, there’s a typo. I said 2100 instead of 2010 and Z just tweaked that. So it depends on what the correction is, but no, you have… If you want to search seek that one out, you’ll have to search for yourself. But I think if you do a search for, if Donald Trump had never gotten off the escalator in Trump tower, there probably aren’t a lot of hits.

0:42:04.4 TW: Well, but I am… He was going down the escalator. So did you take him through the earth and then it’s kind of where he came out? To me, what would be fascinating is where did he emerge on the other side? ‘Cause he was coming at an angle, right?

0:42:17.8 PB: That’s correct. He ended up in that. He ended up… Basically went through the earth. Yes. Over the course of this time. And then the escalator presumably kept going in that exact same direction. So basically emerged out of the earth, like when you were a kid and you’d dig down to China, it was like that sort of thing. And it was… He ended up, if I remember correctly, somewhere just inside earth’s atmosphere above like some Island in the South Pacific.

0:42:41.3 MK: Nice place to be.

0:42:41.7 PB: But he didn’t. He actually got off and gave a speech.

0:42:43.8 MH: This is very useful for when the XKCD starts taking submissions for their next version of what if.

0:42:50.1 TW: Yeah.

0:42:52.8 MK: One quick question I do have, you mentioned that you worked at Adobe and you kind of always been interested in this space. Is there anyone that’s inspired you or that whose work that you follow that you look to in the industry? Where… Is everything self-taught? I feel like data viz, there are so many people that are involved in this, that like you kind of get bits from different people, but that’s been my experience. So I’m curious to understand how you’ve kind of like developed and who’s… If there are people that have inspired you.

0:43:23.4 PB: I’m gonna give what at first will seem like a very normie answer, which is our good friend, Edward Tufts, but I’m gonna go in a different direction with it, which is when I was at Adobe, they paid for me to go to one of the seminars, which I… 80% of your audience, I’m sure has gone through a similar experience. So you go and at the time, this was in the early 2000s in Silicon Valley. And I went to the seminar and sat through the morning session. It was very fascinating. I got all the stack of books, which was three at the time. And then there was lunch. And then after lunch, he went into this thing about how to make your charts for PowerPoint good. And I was just like, “I’m not staying for that.” And I left. And it’s only years later that I was like, “That is so brilliant.” ‘Cause that’s how he gets all the businesses to pay for people to go to the seminar. Right, man? Like it’s brilliant. And so that was also part of what I was thinking with how to read this chart was like, I’m gonna sucker people in and give them like… What I want them to do is I want them to like learn about charts and obviously of course, click through the Washington Post and eventually subscribe to Washington Post. But I’m gonna give them something of value to, which is understanding these things and at times learn how to actually make these things. I’m gonna do the same thing.

0:44:31.7 PB: I’m going to be conscientious about the value I’m giving to people that’s inescapable so that I can then get them to also click through to the Washington Post and subscribe to the Washington Post. And I just thought like from a standpoint of how you actually use your skills to engage people in a way that also benefits you. Brilliant. And he does nice charts, I’ll add that.

0:44:52.6 MH: We should have thought of something like that for this podcast.

0:45:00.8 TW: Well, but there’s a part where it’s kind of similar in that I feel like because you have enough of an audience that you have that it feels like you have people will share, “I found this thing.” Some of the times when you find like the, this was a chart that was found in a magazine that went out of print in like 1958 and you shared it. So you get that bit of like things coming in that you get to put your eye on and say, “Is this good, bad, cool? Did it work? Did it not?”

0:45:29.1 PB: Yeah, it’s nice. There’s been a sort of a community that’s evolved around it. It turns out there are a lot of retired professors who really love to talk about data visuals and I’m all for it. It’s great. I get all these emails. “Hey, check this thing out.” And it’s just it’s fun. It’s a lot of fun. And my job isn’t always fun. So it’s nice to have this sort of fun stuff.

0:45:48.9 MH: Have you thought about setting up a Discord server for all the retired? I’m just kidding. That’s a joke.

0:45:53.0 PB: Honestly, I have thought about how I can expand it, but I also have a lot of fun.

0:45:56.6 MH: No, I mean, like retired professors are probably not going to use Discord. Yeah.

0:46:01.1 PB: Hey man, let’s put it in. You never know.

0:46:03.9 MH: Maybe. Maybe you should…

0:46:06.0 PB: Who would have thought that if you did a show about a newsletter or about how to read charts, people like, “Hey, let’s check that out.” I guess you guys probably would have assumed I would.

0:46:14.3 MH: Well, no. ‘Cause we said, “Let’s talk about analytics on a podcast, and I’m like, who’s the analyst in the… ”

[laughter]

0:46:20.4 PB: There you go.

0:46:21.8 MH: Here we are. Okay. But we do have to start to wrap up. One thing we love to do is go around the horn and share a last call, something we think might be of interest to our audience. It can come from anywhere. You’re our guest, Philip. Do you have a last call you’d like to share?

0:46:34.0 PB: Yeah. I’m gonna be totally self-serving in lines with what we just talked about. I have this book that’s coming out in January and the pitch I’m gonna make is that it has 130 distinct charts in it. There are basically a chart for every three pages in the book. Really, really. I think obviously fascinating to look at the baby boom, what happens after the baby boom, how power shifts in the United States, but there’s a ton of data visualization, including some stuff that I cannot wait. That’s going to end up in the newsletter as well. Just some really, really great visualizations that I can’t wait for the book to come out because I’m really excited about them. But look, you have me on, I have this thing to plug. Sorry guys, I got swamped.

0:47:10.2 TW: No, good.

0:47:12.2 MH: Perfect. We call that synergy.

0:47:13.7 PB: Yeah.

0:47:15.7 MH: Or in the words of 30 Rock, vertical integration. Okay. What about you, Moe? What’s your last call?

0:47:21.6 MK: I’m actually gonna take a leaf out of Philip’s book and be a little bit self-serving. I have got a completely different last call, which is actually a request from our listeners. So what I wanna understand is how you like to absorb information about what’s going on in the industry. So I wanna know whether you mainly listen to podcasts, you look at blogs, read newsletters, watch YouTube. Do you do training? Do you stay up to date on social media? I’d love you to tweet me or message me on measure Slack. I’m just trying to figure out like, “What are all the cool kids doing now?” Hopefully everyone’s not on BeReal. But yeah, I just… I wanna check in and see how people are staying up to date on what’s going on in the industry.

0:48:04.5 MH: Oh no. I hear on analytics power, our TikTok account coming.

0:48:11.3 TW: Oh, dear. I can’t even finish the tweet more than once every two weeks.

0:48:17.7 MH: Okay. I’m not gonna say anything else because I don’t want to get our listeners thinking it down any certain lines. Tim, what about you? What’s your last call?

0:48:23.9 TW: So I’m gonna do a quick follow-up from my last episode of last call that was the Google Analytics alternatives guide that Jason Packer wrote. We now have, if you haven’t already gone out and gotten it, we actually have a… You can get it through Gumroad as an ebook, PDF, or you can get it through Amazon. This doesn’t help with the Amazon. If you get it through Gumroad, we actually have an analytics power hour discount code for a 15% discount. But here’s the best part. The discount code is rock flag and discount. So I just wanted to plug that book again. It’s like the couple of people I know who’ve gotten in and have looked at it have said, “Yeah, it was great.” But my main last call, past guest Ben Stansel, which I could tie to Philip. Go read some of Philip’s writing ’cause it is hilarious.

0:49:16.2 TW: How to read this chart is got… I’m becoming part of the little different hooks. I’ve read them enough to know the parts that are kind of recurring verbal hilarity. Ben Stansel’s footnotes are the same thing, but he wrote a post on his substack called Data’s Invisible Hand, which is kind of a thought experiment of what if the data teams… What if the stakeholders had a fixed set of tokens and they basically had to bid and compete for the data teams time and it’s…

0:49:49.2 MK: Oh, I actually wanna try this.

0:49:50.6 TW: He goes deep into all the like, “What about this? How would this work? How would this actually work? And how would you compensate the data theme?” So it is completely a thought experiment. Probably not… Some of the commenters are like, “We pretty much do that.” And I’m pretty sure they do not do what he has outlined, but it is a fascinating kind of thought experiment. So that is worth a read. Data’s Invisible Hand. What about you, Michael?

0:50:17.4 MH: Well, I read an article recently. So in analytics, we often find ourselves hand in hand with sort of change management in some way, shape, or form just because we’re analyzing things. Let’s change things. So I read this article that has this model for change management that I’d never come across before. And I really like it. I think it’s called Lippitt-Knoster Model for Managing Complex Change and I thought that was pretty fascinating.

0:50:45.6 TW: It rolls right off the time. I’m shocked that it hasn’t just made it into the zeitgeist.

0:50:51.6 MH: Well, people don’t know about this. I always assume when I don’t know about it, it must be something like the people learn like in MBA school or something, but whatever the case, I found it very fascinating and it was actually a really good layout. So we’ll link that in the show notes and you can take a look if you are doing organizational or change management. That is something probably worth your time to understand that model. There’s others too, but it seemed like a good one.

0:51:15.2 MH: All right. Well, we would love to hear from you. If you aren’t already subscribing to Phillip’s Newsletter, How to Read This Chart, I’d say go ahead and start there, but we’d love to hear from you. And the best ways to do that are through the Measure Slack group or on Twitter or on our LinkedIn page. So please reach out, let us know how you’re doing and what other topics we should cover. So yeah, we’d love to hear from you. And of course, no show would be complete without a huge thank you to our producer, Josh Crowhurst. Thank you, Josh, for everything you do. And then once again, Philip, thank you so much for coming on the show. What a pleasure to have you fellow Ohioan. Thank you very much.

0:51:55.8 TW: OH.

0:52:00.7 PB: Oh, I have to throw… I’m a New Yorker from Western New York.

0:52:01.6 MH: Yeah. Okay. New Yorker.

0:52:04.2 PB: But if you rep my home town…

0:52:06.1 TW: But if we give you an OH, you’ll respond with a…

0:52:09.7 PB: Well, I’ll say hi.

0:52:09.8 TW: Okay, that’s good.

0:52:10.9 MH: Okay, there you go…

[overlapping conversation]

0:52:13.1 MH: Yeah. Yeah. Yeah.

0:52:15.3 PB: Like, come on.

0:52:17.6 MH: Tim, yeah. Perfect. All right. Anyways, thanks for coming on the show. We really appreciate having you. We really enjoy your writing though… So I’m sorry for all the other things you get in the mail, but at least, you know you’ve got a few fans out there.

0:52:30.5 PB: I appreciate it.

0:52:33.0 MH: And anyways, but yeah, we really appreciate it. And I’m sure I can say for both of my co-hosts, Moe and Tim, no matter what your data visualizations are looking like right now, remember, keep analyzing.

0:52:48.9 Announcer: Thanks for listening. Let’s keep the conversation going with your comments, suggestions and questions on Twitter at @analyticshour, on the web at analyticshour.io, our LinkedIn group and the measured chat Slack group. Music for the podcast by Josh Crowhurst.

0:53:07.0 Charles Barkley: So smart guys want to fit in so they made up a term called analytics. Analytics don’t work.

0:53:12.1 Tom Hammerschmidt: Analytics. Oh my God. What the fuck does that even mean?

0:53:21.4 MH: So basically, this way will go is I’ll kick it off. I might ask you if you’re a Browns or Steelers fan. Given that it read one of your recent… I think it was one of the recent newsletters where you showed where you grew up and I was like,” Oh, I didn’t grow up too far from there.”

0:53:36.8 PB: Where’d you grow up?

0:53:38.4 MH: I grew up in Lakewood near Cleveland.

0:53:39.7 PB: Okay, got it. Sure. Oh, yeah. Okay. Yeah. I’m happy to… North East Ohio.

0:53:46.4 MH: North East. Yeah. Yeah.

0:53:46.6 TW: All right. I’m in Columbus married to an Arawakan girl, but I’m let them introduce themselves.

0:53:51.8 MH: Okay. Quick if you have many other topics you can be like connected to our guests with. I’m just describing the longest. Okay, let’s go.

0:54:05.5 TW: Rock flag and scatter plots die in darkness…

[music]

Leave a Reply



This site uses Akismet to reduce spam. Learn how your comment data is processed.

Have an Idea for an Upcoming Episode?

Recent Episodes

#211: What To Do Next?

#211: What To Do Next?

https://media.blubrry.com/the_digital_analytics_power/traffic.libsyn.com/analyticshour/APH_-_Episode_211_-_What_To_Do_Next.mp3Podcast: Download | EmbedSubscribe: Google Podcasts | RSSTweetShareShareEmail0 Shares