Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #56: Andy Kirk
138 Plays8 years ago
Welcome back to The PolicyViz Podcast! I hope you had a great, safe summer. To kick off the fall slate of episodes, I’m very excited to welcome Andy Kirk to the show. Andy–as I’m sure you know–runs the popular website Visualisingdata.com,...
The post Episode #56: Andy Kirk appeared first on PolicyViz.
Recommended
Transcript
Introduction to Podcast and Guests
00:00:00
Speaker
This episode of the PolicyViz podcast is brought to you by Tableau Software. Tableau helps people see and understand their data. Tableau 10 is the latest version of the company's rapid-fire, easy-to-use visual analytics software. It includes a completely refreshed design, mobile enhancements, new options for preparing, integrating, and connecting to data, and a host of new enterprise capabilities. To learn more, visit tableau.com.
00:00:39
Speaker
Welcome back to the PolicyViz podcast. I'm your host, John Schwabish. Happy fall, everyone. I hope everyone had a great, safe, and happy summer. Back to another great slate of guests and episodes on the PolicyViz podcast. And to kick it off, I'm very happy to have Andy Kirk from Visualizing Data and author of the new book, Data Visualization, a Handbook for Data-Driven Design. Andy, welcome to the show. Good to see you.
00:01:06
Speaker
Thank you, John. Good to be back. It's been a few months, but nice to be back again on this award-winning Pulsiviz podcast.
00:01:13
Speaker
Thank you, sir. How are you? How was your
Andy Kirk's Journey to Writing His Book
00:01:15
Speaker
summer? Very good. Thank you. Nice and restful. Yes, well, I walk around with a book and hope that people will come and listen to me do some readings known again. But yes, quite restful after a two-year period writing the book. As you know yourself, that kind of come down from the kind of intensity does require a little bit of a breather.
00:01:37
Speaker
batteries fully recharged and good to go. Good, great. Well, I want to talk about the book and a number of things that you go through in the book, but why don't we start by having you talk a little bit about yourself for folks who may not be interested in the one and only Andy Kirk.
00:01:52
Speaker
Sure, yeah. So I'm a UK-based data visualization specialist, a freelancer based in Yorkshire in the north of England. And I kind of get involved in quite a few different things. My public first, I guess, is as a visualization website editor, blogger, visualizingdata.com, all the S's. But also, obviously,
00:02:14
Speaker
I'm an author, but I do a lot of training courses around the world and also teach at Mica at Baltimore alongside yourself and also Imperial College in London. And I do research work and quite a few kind of speaking gigs. So all sorts of different things involved in data visualization because it is fundamentally a passion. So I like to get involved in as many different things as possible.
00:02:40
Speaker
So lots of good stuff and of course, your latest, greatest project, labor of love, I'm going to guess, labor with a you because of that. Your handbook for data-driven design, it's great. I'm loving it. I really enjoyed going through it. I want to talk generally about some things in it and then dig into just a couple of details that you wrote about.
Purpose and Target Audience of the Book
00:03:03
Speaker
Can you talk generally about the book? Who's it for? What do you hope people are going to do with it? There's a great part in the very beginning of the book about how you hope people are going to use it. I want to hear from the man himself. So what were you thinking when you were writing this book? Who are you hoping was going to use this and how are they going to use it? What were you thinking?
00:03:21
Speaker
Yeah. So basically, winding back a little bit, I have written a book before this, and it's a book that was okay at best. And I won't go into all the experiences that came of that. But really, this was, to a certain degree, unfinished business. But it was also an opportunity to write a book at the point at which I actually felt, I guess, mature enough and developed enough in this subject to truly talk about
00:03:44
Speaker
my own convictions about what I believe is right and wrong, good and bad, and the way to do visualization. So the essence of the book is that visualization is a game of decisions. And to make good decisions, you need to be aware of all the options that you've got, and then be aware of the influencing factors that shape what choices you can make, what skills have you got, and how does that restrict or open up opportunities around chart types or interactivity.
Structure and Workflow in Data Visualization
00:04:12
Speaker
what are the formats that you have into aspire towards and what does that impact in terms of the choices that you have to make for a given context. So the entirety of the book is structured around this workflow process, which will maybe come into more detail shortly, but hopefully what this gives people is a very practical
00:04:32
Speaker
pragmatic way of making sense of what is potentially quite chaotic sense of creative and analytical choices that in and of themselves are not complicated. They're just quite numerous in number and varied at the point at which they need to be dealt with. So the book was written for primarily social scientists, researchers,
00:04:56
Speaker
and I guess specifically undergraduate, postgraduate and kind of postdoc level readers. In effect, that kind of means it's everything other than those involved in the life sciences. But as I'm sure you'll be able to kind of share yourself, it's very difficult I find to write a book that excludes potential interested visualization people because it is such a neutral and portable activity that
00:05:23
Speaker
everyone really has got an entry point that hopefully they can find something relevant to this piece of work. But primarily, as we always talk about and as we always preach about, design for your audience, the audience intended for this was the kind of social science type subject areas.
00:05:40
Speaker
So why the social science folks? Is that because that's what you used to do or is that just like there's just the array of data that you can use in the social sciences or or like the biological sciences sort of you know a different type like why did you angle down and I love that angle because that's my angle but yeah why did you head down that avenue?
00:05:59
Speaker
Well, the primary point for that is that the publisher, Sage, are focused on doing books for that audience. But secondly, likewise, it is also my background, working in the police force and the universities and different organizations in the past. I've always been on that side of things, so it's the audience that I feel most confident talking to, really. Yeah, so I forgot to pick up on the point you mentioned about. In the opening section of the book,
00:06:27
Speaker
I do kind of request people to use the book, physically use the book, fold it, write on it, scroll on it. Don't tear it, perhaps. Let's draw the line there. But I want this book to be used, to be kind of, you know, bent backwards so people can read out a passage. And I want people to have markers and little adhesive tags sticking out because there's been an interesting thing that they've found. There are many beautiful books out there.
00:06:53
Speaker
that are almost too beautiful. It's like glossy, coffee-tail books and you almost feel like you need to wear white gloves to open them up and kind of go over each page. Let me say, I'm delighted with how it looks in its print format. The quality of it is really nicely done by the publishers, but it needs to be something that is practical and is indeed used. I want to, maybe down the line, I'll be asking people to send me photographs of a very kind of worn and used book because that will, to me,
00:07:19
Speaker
that would be a major success that people have actually found it useful that they've gone back to it and they've referred to it. Right. You talk a lot about process in the book and I think unlike a lot of other books on data visualization, you spend a lot of time talking about the process of
00:07:36
Speaker
developing a visualization, creating the visualization, working through the data, developing a production process.
Creating Visualizations: Kirk's Four-Stage Process
00:07:44
Speaker
Can you describe that production process a little bit to get us started? I want to dive in a little bit and I have some follow-up questions. Yeah, sure. I have a four-stage process based around how I've always found it the most logical way to organize my thinking.
00:07:59
Speaker
And very briefly, the first stage is what I term formulating your brief, which is very much rooted in the language of graphic design or project management, but it's establishing the requirements. Why are you even here making a visualization? So why are you doing it? What's the trigger curiosity? What's the trigger opportunity that you're trying to pursue? What do you intend to achieve with it? What would be a successful accomplishment from this work?
00:08:25
Speaker
Around that, what are the factors that will shape what you can and can't do, what you should or shouldn't do? Practical things like timescales, access to software, who is the audience that you have in mind, all those kind of classic things that don't take a lot of time to deal with, but they have such a huge bearing on your choices later on, so get that done first.
00:08:45
Speaker
The second stage is what I describe as working with data, which might have actually arrived before you even start the brief stage. But in terms of organizing the book, I have to create a linear sequence, obviously. So working with data is four steps. It's getting the data, it's examining the data, getting a sense of its physicality, its shape, its size and condition, transforming it, so modifying it, creating your calculations, cleaning things up, reshaping it, getting rid of stuff you don't need.
00:09:13
Speaker
And then lastly, exploring it to familiarise yourself with the qualities and the insights of the data to educate yourself before you then start to commit to what you might do with it. And the third stage, editorial thinking is indeed the stage of commitment.
00:09:29
Speaker
defining what you will and you won't show to people, determining what you will and won't include in that display of data and what things you might seek to emphasize. So editorial thinking, as we know, there's no sense of pure objectivity in visualisation. So it's about making decisions about what to do and what not to do, which then leads on to the fourth stage, which is the design stage.
00:09:51
Speaker
working through the kind of design solution, all your options, rationalising which of those you actually should choose, which chart type should you use, what interactivity could you use if it is indeed a digital solution, what annotation, labelling, captions, introductions, and then the colour choices and finally the composition choices, where things will go, how big things will be.
00:10:14
Speaker
So that process I've always found is the easiest way I find in practice, but also in teaching to kind of organize my thinking, organize my critical thinking, hopefully to get the most effective solution in the most efficient way. And I think efficiency has to be a very important factor in all this.
00:10:32
Speaker
Right and you mentioned in the book having a design brief so when you are. Creating a visualization are you you have a physical. Document that you're working out of to follow the four steps that you sort of outlined like how to what what is your what is the Andy Kirk. Individual process to walk yourself through those four steps.
00:10:50
Speaker
It's absolutely that. I have a relatively loose template that has all the prompts at least. It's not too much box ticking or filling this text box for this section. It gives me the prompts because if I'm doing stuff for myself,
00:11:08
Speaker
I need to be reminded of the things I need to think about and when. If I'm working with clients, I need to be able to manage the conversation. This is the time to think about this and this is the time to think about that. So I do have a loose, I wouldn't say template approach, but kind of an aid memoir approach, which helps to organize my own thinking, whether it's alone with collaborators or with clients.
00:11:32
Speaker
And when you are thinking about taking this process for other organizations to apply it to their own work, do you have an organization in your head or is this applies universally no matter if you're a design firm, if you're a news organization, in your research organization, or do you have a specific sort of workflow in your head when you're working in this way?
Adaptability and Universality of Visualization Process
00:11:56
Speaker
It's a good question, and I think, let's say from my blink of perspective, I believe there is a universality to it, but it's specifically for those occasions whereby you're doing a one-off project that might not be repeated, or you are doing an initial development of something that will be repeated, but you need to kind of get it right first time round to create the infrastructure so that the second time round, let's say every month, every quarter,
00:12:26
Speaker
you're almost just dropping the new data and all the thinking has been done in that first upfront development. Obviously, when you are faced with a repeated task, you don't always need to go through the process of thinking about why we're doing this and who's the audience because that's been done already. It does really focus more on those one-off projects or those initial developments
00:12:50
Speaker
And again, within those four stages that I just briefly outlined, there are all sorts of stages of iteration.
00:12:58
Speaker
in a book, you have to write it in a linear fashion. But there are points where you go backwards and forwards, that you stay in a loop for a little while. There are probably stages where you do them out of order in terms of, I'm sorry, out of sequence based on the order that I presented stuff. So it is a framework. It's not a prescribed thou wilt do this all the time. And hopefully those people have got the kind of mindset to take on board a framework and not
00:13:25
Speaker
just to purely rely on a very kind of instructive stage one, stage two, stage three type approach. They'll be the ones who can most flexibly adapt this. And it does need to be adapted. It needs to kind of fit into your own natural rhythm about how you work and how you think about this challenge. You know, we've all got different mindsets. So it is not to enforce it, but just to give you a sense of a framework that hopefully will be immediately applicable to you or at least will be adaptable to you.
00:13:51
Speaker
But it's also, I think, I mean, the way I read it was that it's also applicable to teams. I mean, you make a point of saying, you know, know your skill sets. Maybe you can't do all this, but, you know, there are other people on your team or you need to hire someone. So, you know, I kind of feel like it applies. It's a process that's not just for an individual, but for the, you know, for groups, right? Well, that's right. And I mean, you know, if you look at, let's say, newspapers,
00:14:13
Speaker
Most projects that you see there perhaps have two or three names against it. So you are looking, let's say at least for the most ambitious stuff, you are looking at involving two or three pair of hands. And so you do need to organize resources and you do need to organize task dependencies. And again, it's not to make it overly systematic or overly project managed, but
00:14:35
Speaker
We don't do these things without deadlines. So we can't afford to neglect the sense of organized approach, efficiency approach. And that's absolutely paramount when you are working beyond just your own individual pursuit. It's something that, you know, you do see that team environment.
00:14:54
Speaker
Yeah. The other thing that's interesting about your book as opposed to many that are out in the market is that you also spend a lot of time talking about data. Different types of data, how to work with different types of data and sort of the link between the visualization and the data.
00:15:11
Speaker
As opposed to, here's a bar chart, you actually talked about the data first and then sort of get into it. So you have process, you have data and then the visualization. And I'm curious, what do you think people are most likely to do wrong when they work with data? There's a few things that come to mind. I mean, I think the first one that people do wrong is, can it go through the motions?
00:15:33
Speaker
Let's say, I mean, the classic example would be something like, I have spatial data, I must map it. It's about trying to, I mean, I talk about this, I think, in the book, and I talk about it in trading courses in particular.
00:15:45
Speaker
It's about having that kind of intimate acquaintance with your data going beyond the surface. It's not just how many things have you got and what type are those things. It's about what is this thing about? What's the phenomena? What's the meaning of the data? Is it about people? Is it about activities? Is it about transactions? Because all those kind of semantic dimensions that aren't necessarily captured or visible at the surface are cues or indeed clues to how we might best visualize that subject.
00:16:15
Speaker
So I think you've got to always have that sense of I'm going to take this data set on its own merits and I'm going to really be open minded about what it's telling me and what it's not telling me. I think the second thing is to have the commitment and the discipline and perhaps even the courage to make a call.
00:16:32
Speaker
to say that, okay, even in the relatively simplest datasets, the relatively simplest tables of data, there's actually lots and lots of ways that you could analytically pursue that and therefore lots of different ways you could visualize it.
00:16:46
Speaker
And I think sometimes people are, and this is perhaps a cultural thing or an organizational cultural thing, people are afraid to not include something. People are afraid that that one person who wants that one single statistic that no one else is interested in needs to be served. And practically that's not always possible. And so it's having that commitment, which is that in effect, which is that third stage of editorial thinking to say, just because we've got all this stuff, it doesn't mean so we have to use it. Let's make a judgment about what we are seeing
00:17:15
Speaker
is the most relevant, the most representative, the most interesting thing to emerge. And let's make a call about what not to include. And let's have conversations with people who know the domain and ask them. So it's really about getting under the skin
Common Mistakes and Recognizing Data Gaps
00:17:28
Speaker
of data, I think. And that's what people hopefully will get from that chapter in particular, that it's not just a mechanical thing of flicking through a spreadsheet and doing some basic descriptive statistics. It's about thinking carefully about what all this data means
00:17:42
Speaker
Yeah. I want to make sure I mention, because you were talking about making sure you pay attention to things that may not be there. So I want to make sure that I highlight your openViz talk from a couple of years ago. We talked about the zeros in the data and the missings in the data. The design of nothing. Yeah. Because it was a great talk and talked about that exact point. So I'll make sure I put that on the site because it was a really interesting talk and something people should be thinking about, obviously.
00:18:07
Speaker
And that's it. And I think I've got a quote in the book from Joe Thorpe, which is about embracing the gaps, about asking questions about what the gaps mean and not being put off by those, but actually maybe embrace them and say, okay, this is the story. It's not the presence of data. The absence of data is the story about a lack of transparency, lack of availability, lack of infrastructure to capture data about that entity or that phenomena. So yeah, always embrace the nothings.
00:18:36
Speaker
The gospel according to Andy is graced to nothingness. So just keep that in mind everyone. Okay. The Seinfeld said there's always something in nothing. That's right. That's right. Okay. So we talked about what do people most likely do wrong with data? So what do people – what do you think people are most likely to do wrong when they create a visualization? Not – we don't need to talk about the 3D and the terrible stuff. But what's the sort of the core things that people tend to just miss when they're creating?
00:19:05
Speaker
I reckon I could play bollock down two or three, maybe even two, but the first one is kind of being led by ideas or being led by a desire to do cool things. I'm always hesitant when I say the word cool attached to anything in visualization because it usually means that there's an imbalance in the focus, but this is how it should go. Data should lead to the design, not the other way around. We need to be respectful of what the data is telling us and respond accordingly.
00:19:32
Speaker
You might have a desire, and I often find this in projects I work on. I'd love to work on a Sankey diagram task today. But the data might not be something that lends itself to a Sankey, or I've got this little desperation to do some small multiple area charts at some point, but I've not had the data that lends itself to doing that thing. So ideas are important to capture, kind of remove and extract from your head at the start. Do not be led by them. To have them as a kind of a little bit of background inspiration or influence, but
00:20:02
Speaker
You know, let data talk and then respond accordingly. So having all your options is one thing, but knowing what the right choice is, is the real outcome. I think the second thing is simply the wrong chart to show the thing you're trying to show. I sense that's getting better, actually. I think the literacy of what each chart does in its role and what it doesn't do in its restrictions, I think that's getting better.
00:20:24
Speaker
across the board but yeah picking the wrong chart to do the wrong thing I think so often a very well-intended and you know just a simply an innocent mistake to make but I think that's perhaps the other common mistake.
00:20:35
Speaker
I don't want to cause a fight or anything, but I want to see what you think here. So people in the field tend to get sort of riled up, up in arms about certain visualization sins, right? So zero baselines, pie charts, word clouds, whatever would have you. Is there one of those types of issues that you take a strict line on that you have your, you know, you're really militant about? Or are you, well, it depends. There's gray in everything.
00:21:01
Speaker
Yeah, I think I'm a lover, not a hater. I think there are people out there and we're not going to cause fights by naming names. There are people out there who do, for good reason, see things in a very black and white world. And I think it can be because people are inherently creating and causing those kind of gray scales.
00:21:24
Speaker
Of those things that you listed, and there are others, of course, I don't think there are any that I say don't do. I mean, I include pie charts because pie charts have a role to play, and the evidence of people like Robert Kazara, and I can't remember whose collaborator it was, Steve Harrells, possibly. Well, on the pie chart stuff, I think it was Drew Scow. It was Drew Scow, of course it was. Yeah, sorry. They've shown evidence that actually there are benefits, and there are
00:21:50
Speaker
increased evidence of the ease of readability of a pie chart. So what I do in the, perhaps, the flagship section of the book, which is a chart type gallery, I include the pie chart in there, but I include with that the caveats of when you probably shouldn't use it. When you've exceeded, let's say, three categories, it might be better off using a bar chart, for example. I include the word cloud.
00:22:12
Speaker
And yes, there'll be people who are preparing an angry mob to come round to my house and say, how dare you? But there's a role to play the work, Cloud, in my mind, for yourself. When you're analyzing and exploring data, you're getting a sense, you're getting a gist of what you've got.
00:22:28
Speaker
in a passage of text. Personally, I would never use it to display and communicate to others, but it is still a chart type ally when it comes to that stage of exploration. So most things have, even if it's a very small caveat, they do have those moments when they can become useful. They might not be the most useful, but they can be part of an artillery of tactics. I think the one thing that I am absolutely clear about, truth is an obligation.
00:22:56
Speaker
So one of the principles I talk about in the book is about trustworthiness, which is about your desire to try and get as much trust earned from your audience as possible. But that's assuming that you are being truthful. And so never lie, don't make things up. Don't make wrong claims. But trust is a difficult thing to gain, but truth must always be an obligation.
00:23:18
Speaker
More words to live by, everyone. Absolutely.
Future Tools and Resources for Visualization
00:23:23
Speaker
Okay, so I want to close things up by looking ahead to what you've got going on. Of course, the book is out doing great. You have the website and the blog and your monthly roundups. You also have your existing research projects, but are there other projects that you have that you're working on that you're coming out with? What are we looking forward to from visualizing data over the last few months of 2016?
00:23:48
Speaker
Yeah, so I think the first focus after the book has been out has been actually to try and... Sleep, sleep a little bit. Well, sleep, sleep, and then try and earn a living again, having been totally off the... So kind of consumed by the book writing. But the next kind of project of any sense of magnitude, and certainly in a public environment, is something that I've been wanting to do a long time now.
00:24:12
Speaker
to the questions that I get asked most at my training workshops and even just on kind of emails and discussions on the blog which is which tools allow me to make which charts and which charts kind of make in which tools. And so what I'm looking to develop
00:24:30
Speaker
and it should be out very, very soon, is some kind of interactive, browsable matrix that will list all the chart types that kind of come from the book, but also I'll probably expand on those as well. And then lists either all the tools, but maybe perhaps some of the major tools. And it invites people around the world to contribute what they are aware of being possible, either standard off the shelf chart,
00:24:58
Speaker
or as a kind of a hack, workaround, plugin, achievable chart, so that we can get a sense for those people who are trying to navigate around the plethora of tools that are out there. If they want to make a certain chart, which are the tools that will equip them with that possibility?
00:25:15
Speaker
what I'll probably do is looking to give people a sense of an example image and maybe a link to a tutorial which explains how to make that chart if it's more than just an off-the-shelf prospect. So I think this is, you know, this feels like an important thing not just for
00:25:32
Speaker
Beginners, but even for myself, you know as we all know there are so many tools out there and getting a sense of what they all offer Even if it's just a narrow view of what chart types they offer I think will help everyone get a sense of where to where to look to kind of broaden their skill set, right?
00:25:48
Speaker
I mean, I think for a lot of people, just to know what tools are out there, I think is super useful. And then to be able to apply and know what they can do with it. Andy, always great to talk to you. Thanks so much for coming on the show. My pleasure. Really enjoying the book. For those who haven't picked it up yet, they should.
Conclusion and Promotions
00:26:04
Speaker
It's Data Visualization, a handbook for data-driven design by Andy Kirk, published by Sage. And of course, you can check out all of Andy's great stuff on visualizingdata.com, visualizing, of course, with
00:26:15
Speaker
All the S's. All the S's. All the S's all the time. Andy, thanks again for coming on. My pleasure, John. Thank you. Thanks, everyone, for tuning in to this episode of the Policy of This podcast. We're now back for the fall with a whole new set of guests coming up over the next few months. So thanks again for listening. And until next time, this has been the Policy of This podcast.
00:26:49
Speaker
This episode of the PolicyViz podcast is brought to you by Tableau Software. Tableau helps people see and understand their data. Tableau 10 is the latest version of the company's rapid fire, easy to use visual analytics software. It includes a completely refreshed design, mobile enhancements, new options for preparing, integrating, and connecting to data, and a host of new enterprise capabilities. To learn more, visit tableau.com.