Being a Data Viz Consultant | Cara Thompson | Data Science Hangout

Hey there, welcome to the Posit Data Science Hangout. I'm Libby Herron, and this is a recording of our weekly community call that happens every Thursday at 12 p.m. U.S. Eastern Time. If you are not joining us live, you miss out on the amazing chat that's going on. So find the link in the description where you can add our call to your calendar and come hang out with the most supportive, friendly, and funny data community you'll ever experience.

Can't wait to see you there. I am so excited to introduce our featured leader today, Cara Thompson, Data Visualization Consultant at Building Stories with Data. Cara, welcome. How are you doing? Thank you so much. Thank you so much for having me. I would love you to introduce yourself, tell us who you are, what you do, and something that you do for fun.

Okay, so I am Cara Thompson. I live in Edinburgh in Scotland, and I am a Data Visualization Consultant. So it's my joy in life to help data teams make the most of Data Viz to communicate the outcomes of their research really effectively. And I think I found over the years that my niche has seemed to define itself as helping teams of specialists communicate with a non-specialist audience. And it's just, yeah, it's great. I enjoy doing that a lot. You can probably guess some of the things that I do for fun from the stuff that's behind me. I'm a musician. Mike and I have talked about this a few times, and I'm convinced that my musical education plays a strong part in how I approach data visualization.

I run as well. I enjoy running. I find it's a great, yeah, it's a good headspace generator. I'm a mum of two young girls, and so having, you know, half an hour to an hour where nobody is asking me for anything is really wonderful.

Being interdisciplinary

I remember asking you, do you have any sort of data viz books to recommend? And you said, no, I don't. I don't worry about data viz specific stuff. I don't worry about data viz specific books or data viz specific conferences because I am interdisciplinary and that's what makes me special. So can you talk a little bit more about that?

Yeah, sure. I mean, I have to say I do have a lot of respect for people who write data viz books, and I think there is a lot to be learned from them. But I think a huge part of what makes me approach data viz the way I do is the background that I have. So I grew up in a family where we were speaking two languages. So I speak French and English. I speak French with my daughters. And so there's an element of learning that people process the world in slightly different ways through different languages, which I think has been useful to me. I enjoy playing music. I enjoy writing music. If anyone in my group of friends is getting married or having a big party and they've got some musical friends, I'll try and write parts for everyone so that it all works well together.

Mum is a brilliant artist. She paints. And so she used to take us around art galleries growing up. And there's a lot to be said for just being curious about how other people use shapes, colour, movement, and how across all these things we encode information, whether it's language, music, visual arts. We're all encoding it and then decoding it. And it's just something that I have always found fascinating. And so I took the multidisciplinary thing through to my degree. I did a degree which was a joint degree in music and psychology, which I really enjoyed because when I got bored of writing my psychology essays, I would write a piece of music and then I would get stuck on that and I would go back to writing the essay and back and forth, which was great. It also taught me a lot about managing multiple deadlines, which I didn't always succeed at doing as a student.

So I think I did that as an undergrad. Then I did psycholinguistics for my masters and then in the PhD revisited how the brain reacts to strange things in music and language, how we process patterns, what we can learn from that as to how we're wired to be excited by new stuff. And yeah, I just love hanging out with people who are interested in different things to me because I think we've always got a lot to learn from each other.

You mentioned to me museum displays that you really like the way that you see things done. And you also said, like, how do comic books do this? How do we present things visually? Can you talk a little bit about some of your favorite inspirations that are comic books?

Yeah, of course. I mean, so comic books are a really big thing in France and, you know, they are throughout the world, but French kids will tend to grow up reading Asterix at some point and Tintin. You know, there's a lot to be learned from Tintin. It's actually historically quite useful. There's some stuff that you have to, you know, reimagine a little bit nowadays. But I think, you know, growing up with that notion of telling stories through pictures has been quite fun. So how do you know, you know, if there's a speech bubble on its own in a little box, how do you know who that belongs to? How do you infer what has happened between image one and image two? How is movement communicated? How do you see the different angles that people are looking at? And then it's really fun seeing different cultures. So like Japanese comics would be quite different from European comics in terms of how shapes and movements and stuff are translated.

And there's quite a lot of similarity between the two, but also really interesting differences to look at. And then you've got the thing about, you know, the more specific a person's face is, the less likely you are to identify with them. So if you make a more generic face, people are more likely to identify with the character. And if you think about, you know, the IKEA flat pack furniture instructions, very generic people that are drawn in there. And I guess that's another interdisciplinary thing, you know, growing up building flat pack furniture. That was me and dad. But, you know, growing up with him, both of us used to like, really like that problem solving. Like we had, where did this piece go? How did they, why did they draw it this way around? And I think for a while as a kid, I thought maybe, you know, drawing IKEA instructions books could be, you know, one of my future jobs. It's just been really interesting thinking about that notion of translating something complex into something easy has been a thread all the way through.

Career path to data visualization

I would love to ask you a little bit about your background because you mentioned your master's and your PhD, love that that was super interdisciplinary. What did you do after the PhD that got you to where you are right now?

Yeah, so after the PhD, so my PhD was in how the brain reacts to weird things in music and language. There wasn't a huge amount of funding for pursuing that at postdoc level and then going into academia, although that had been my intention. So I applied for a bunch of postdocs and didn't really get anywhere with that. But where I did end up was teaching research methods to students who were enrolled on a psychology module within a music department. And so that was really fun because these were people who hadn't come from a research background at all, but they were doing music psychology stuff. If you think about music therapy or the potential community involvements in music programs, they wanted to do some research that could help to emphasize the impact of what they were doing. And so my task was to give them what they needed to do that, to do it well, to write it up.

And during that time, a job came up at the Royal College of Surgeons in Edinburgh. So in order to be a surgeon in the UK, you have to pass one of the Royal College of Surgeons exams in order to become a consultant. And so these exams are really, really important. They're career defining for people. And an exam, like a research design study, is a measurement tool. So the task was figure out how precise the measurement tool is, figure out how we could make it more precise, help the surgeons understand the imprecision, the implications of that. And what do we do with candidates who achieve exactly the pass mark? Are they having a good day and actually they're not really safe to be let loose on people? Or are they really competent and they're having a bad day? And these were all things that required the surgeons to grasp statistical concepts in terms of the relationship between the items in the exam, probabilities, false positives, all this kind of stuff.

And surgeons are not used to not understanding stuff because they've always been top of the class. It's something that they found quite disconcerting. And so I think what I really enjoyed with them was meeting them with explanations that they could really grasp really well enough to make informed decisions. And data visualization became a key component of that. So rather than bamboozling them with statistical concepts, I would show them the data and we could see, right, here's the pattern. This is what it looks like. These are the implications. What do we want to do about it? And so it was just really, really interesting to be in that world of, you know, communicating education data to specialists in a completely different field and helping them feel confident that they were making sensible decisions for the exam.

So I did that for about a decade and during that time got more and more excited about data visualization. And then I went on maternity leave with my first daughter and discovered the Tidy Tuesday challenge. It was COVID. I needed something to keep my brain going with the coding side of things. And my maternity leave got extended because of the lockdowns. So I thought, right, I'll get involved in this. This looks like fun. And it was great. And it really, you know, it really transformed things for me. And that's where the learning out loud in R came from, although I think I've always been keen on learning out loud. But doing it in the R community, people were so welcoming. It was a real confidence boost at a time when I desperately needed it.

But doing it in the R community, people were so welcoming. It was a real confidence boost at a time when I desperately needed it.

And so I'm just so grateful that that challenge was there at a time when I needed it. And I'm grateful that it's still there, that it continues. And it's a source of inspiration for so many. So yeah, seeing what other people were doing, like, how nice did they do that? And then you find the code, you're like, okay, okay, that makes sense. I can, I can, I can do something. And so then you build on it and figure out how it works and build your own stuff that people go, how did she do that? And then you go, well, here you go. Here's the code. And seeing other people building on the stuff that I've done is just really, really rewarding. It's a magical feeling.

Data viz in specialist vs. mainstream contexts

So the question goes something like, can you talk a little bit about the differences between the kind of infographics that you might see in the media, like BBC website or Financial Times, and the kind of what I'm calling death by data viz that we get in pharma industry or for publications or for the kind of regulated industries? Because it seems that there's very little in common. I mean, I would love for some of those publication visualizations to be closer to your best practices and those kinds of more easy to read, easy to understand concepts. But can you just kind of, why would people do those kind of really nasty data visualizations?

Yeah. I think there's an element of you do what you know, right? So if your field has always produced this specific type of visualization using this tool that everybody uses, then you just go with it. If the aim is to communicate just within your field, then that's going to be the path of least resistance, A, in terms of time investment, B, in terms of getting other people who are familiar with that to understand it. But I think there is a real case to be made that people, I think probably since the pandemic and more data was being shared during that time publicly, people have got more used to seeing visualizations, people are more willing to engage with them outside of the specialist fields. And so if we want that to be done well, then we maybe do need to start engaging a little bit more with formats that have been more successful in more mainstream media.

I think there's, yeah, it's probably a question of priorities as well. You know, I feel I had to fight quite hard against imposter syndrome when I would go to conferences and see all these people doing really cool stuff with the data, and I was like, I just make it look pretty. But actually making it look pretty, it's not just about making it pretty, is it? It's about making it more functional, making it more accessible, making it easy for people to understand. So yeah, I think it's probably priorities, it's time, and it's that people see it and think, oh gosh, I don't have the skill to do that, whereas in actual fact, you know, people can build up the skills to do these kind of things.

Choosing tools and managing overwhelm

Hi, everyone. Thank you, Cara, for being here. My question is kind of coming from a place where I feel overwhelmed often by sort of the number of data tools and packages available for any given task or project, so I'm wondering if you have thoughts on, like, how do you decide what to use and for what purpose, and then how do you balance between sort of deepening a focused skill set versus expanding your toolkit by learning new things, new packages and tools?

Yeah, that's a great question. The overwhelm is real, I think, for everyone, so don't feel bad about that. That's probably a good thing, you know, to realize in a sense how small we are in a huge field. You know, it's good. It keeps us wanting to learn. I would say the thing that I encourage people to do when I'm giving data viz workshops is start from where you are, so think about the types of data that you are currently interacting with, and think about the types of stories that you're telling with the data that you're interacting with, and focus on a small number of graphs that you think are a really good fit for that combination of things, so the interaction between the type of data and the type of story. I always encourage people to start with pen and paper rather than starting with tools, so if you were to draw your data story on the back of a napkin to explain it to a friend of yours over coffee, what would that look like? And then from there, figure out what tools can get you close to that.

There are some great resources out there. There's the data visualization catalog, which is tool agnostic, so there are a whole lot of different data viz types on there. What I love about the catalog is that for each data visualization type, there are caveats, like you might want to use a pie chart, but here are the reasons why this might not be the best plan for your data, and the same with all sorts of graph types. I picked on the pie chart because everyone loves to pick on the pie chart. They do have their place, but yeah, it's really useful for having that conversation with people about why different types of graphs might be suited to different data sets or different stories.

A lot of the time it's linked to tutorials with different tools. If you're looking specifically for coding stuff, then data to viz, I think Jan Holtz developed that, is a really good tool where you'll find tutorials for doing all that. But I would say start where you are, start with pen and paper, and get really good at a small number of things. Then the skills that you've learnt on those small number of things will be transferable to different visualization types, and then you can grow from there. But yeah, don't worry. There's always going to be stuff that we don't know. I feel the same. I had a client the other day ask for a specific type of plot, and I was like, oh, I might just have to Google that very quickly to figure out what it is. So yeah, it's perfectly normal.

Specializing as a consultant

I think in general, I might have understood it wrong, but my understanding is that you focus mostly, if not exclusively, on data visualization. And I wondered if you ever find it limiting or there's many different projects that keep it fresh, let's say. And then I was just curious what you would say to somebody who would like to specialize so much.

That's great. No, that is absolutely the case. I do specialize in data visualization. And to me, this has been something that I wrestled with when I first started out as a consultant, because you can code. So I can do this. I can do this. I can do this. I could build these kind of pipelines. I could do this kind of stuff. And so the question, I think, to ask yourself if you're thinking of doing this kind of stuff is, what is the thing that I do slightly differently? So what is the thing where I could really hone a craft and be world class at this particular skill?

There is a temptation to try and be all things to all people and be as helpful as you can. And initially, you think, if I diversify what I do, then I can get more clients because I can do all sorts of different things. Whereas the natural fact, if you do one thing really, really well, then you gain a reputation for being very good at that thing. And the more you do it, the better you get at it. And the more you can talk about projects that you've done that are similar, the more you attract the same kind of clients. So there's a lot to be said for niching, for thinking quite carefully about the skills that you want to be showcasing.

And yeah, I guess when you're starting out, you probably want to push a bunch of different doors and see which one's open easier than others. But for me, it's definitely been data visualization. And I like the fact that I get to interact with a lot of different people. So I'm doing some stuff at the moment for the British Heart Foundation, and it's been really interesting interacting with their team and seeing how they're using data. And I did some stuff for a mental health advocacy charity in Scotland, which was really great, just seeing the data that they'd collected and how they were using that to try and inform policy decisions. I've done some stuff for research groups who were comparing how trustworthy videos are perceived to be, depending on whether they were AI generated or human generated.

I've done stuff equipping research teams to go and build their own visualizations, which I really enjoy as well. So I think there's always enough to keep me excited about what can be done with data viz. And it's the combination of, you know, I'll do it for you, and I get to learn something about what you're doing. And that's really exciting. Or I build tools or training kits for you. And then you get to go and build more stuff than I could ever dream of building. And that's a really exciting thought as well. So I think, yeah, to answer your question, I don't think it gets stale. I don't think I'm going to run out of projects because this is just a really interesting skill to be adding in to what teams are already doing.

And I also love that I don't have to be the expert on the subject matter. So I get to trust their expertise. I have enough statistics in my background that I can spot something that's not maybe the way that it should be. But I love the fact that I get to make other people look good. And there are plenty of people out there to help with that. So yeah, it's good fun.

Finding clients and working out loud

Jared says, with regards to being a consultant, how do you find clients? Or do clients find you as a result of what we have continued to call working out loud? And then the anonymous question was, how did you find your niche? So maybe these two can go together.

Yeah, that sounds good. So I think finding my niche, it was a process. But it was actually quite a fun one. And part of it is when people come to you and go, I really like what you've done here. Could you do it for us? And that kind of makes you see, OK, this is a useful skill that people are excited in, in which I stood out. Part of it was asking clients directly why they approached me. And that helped me refine the language that I use to talk about what I do, which was a really interesting one as well. The whole thing of having the scientific rigor, but also the creative side and marrying those two things was something that came back a few times in what clients were telling me when I asked them that question.

In terms of finding clients, a lot of it is about learning, working out loud and becoming recognized within my group of peers as somebody who knows what they're doing, who's also flexible in the way that they approach things. And so initially, a lot of my work came from people like yourselves. We would hang out online. We would do Tidy Tuesday stuff. And then we'd go, I really want to bring that to my team, but I don't quite have the skill set that's needed to do this bit of it. Could you help us with that? And that was really great. A lot of it is word of mouth and recommendations from other people. That's not a long term strategy for running a business. So you do have to be a little bit proactive about publicizing what you do. But yeah, a lot of the time, if I'm honest, clients do tend to find me rather than the other way around. But I'm probably more aware now of the potential in what I share online for it to be reached by different audiences. So my peers who get some useful skills from it and also potential clients who see it and go, actually, we could do with that as well.

Yeah, no, don't worry about it. Go for it. I think it's funny, because I thought this was going to come up because we talked about it briefly, but I was just thinking about it earlier today. So I don't have a LinkedIn post schedule. I don't have a list of topics that I'm going to talk about. I don't do any analytics on what time of day I'm posting, and the impact that that has on my reach, and all that kind of stuff. I'm not trying to game the algorithm. It genuinely is, oh, wow, this thing. I didn't know this thing. Part of it is I'm talking about it so that other people can learn about it. Part of it is that then I can add it to my blog so that the next time I get to this thing that I don't know how to do, I can find the solution that I found two years ago. But it genuinely is coming from a place of, wow, I'm excited about this thing that I didn't know about.

Or like yesterday, I spent a day in frustration trying to hunt down this bug. I've just switched from Windows to Mac, and that's been a whole journey in and of itself, but it's been quite fun. And there was an error message that was coming up, and I was getting stuck on it. And so all the way through, I was like, once I find this, I'm going to tell people what the solution is. And then no one else will have to fight with this thing in the way that I just did.

So, yeah, I would say if you don't be, particularly in our communities, I think there is a genuine desire to be helpful to each other. And so coming in with a question that's a genuine question, not one of those, like, oh, if you ask me a question, then you cannot get access to my newsletter kind of question. That's not a question. But a question where you genuinely want to learn something from the community, then ask it. And that will create interactions. It will open up conversations. You will be more integrated in people's networks, and you'll understand things better. So, yeah, do stuff, make mistakes online. I've made mistakes online. And actually, sometimes you learn a lot from those as well. You post this thing like, oh, this is great. And someone goes, actually, that's not how it works. OK. And then you learn from that.

Trends in data visualization

I'm wondering if there are trends in data viz right now that you feel really excited about or that you think is really cool, or also trends in data viz that you dislike. I'll just give one example of something I've seen a lot lately, which is there's a lot of generative data visualization out there that basically uses LLMs to you ask a question of your data, and it magics up a data visualization. And I don't know whether to be suspicious of that. I mean, it feels kind of sketchy. But yeah, what are trends that you feel are exciting and maybe some things that you dislike and feel are bad?

Yeah, good question. I think one of the things that I'm finding quite exciting is just people coming up with new ideas and trying to see how we can help people connect to the stories behind the data. So I think as we've got more and more used to seeing graphs, people are getting more and more used to that. And seeing large-scale data, I'm quite excited by the small-scale data stuff, and people using data to emphasize personal stories. I think Georgia Luppi has got a brilliant manifesto. I think it's Data Humanism Manifesto, where she talks about trying to help people connect to the reality behind the numbers and how we do that well. So that's probably the opposite end of the spectrum from let's throw all the data in the world at this thing and see what comes out the other end. But I think to me that's really important, and it kind of ties in nicely with the way that I tend to think about the people that are behind the data stories that we tell.

In terms of stuff I'm less excited about, yeah, I have to say the AI hype is something that causes me to ruthlessly unfollow people on LinkedIn, because it just clogs up the feed with all these recycled things. And some of it is great, I think there are some really exciting tools out there that can help people generate new ideas, but I would say don't delegate your specialism. So I feel like, so I play multiple musical instruments, I play the electric guitar very badly, and I deliberately bought it a couple of years ago to teach myself to fail hilariously at stuff, because normally musical instruments, I'm quite good at just picking them up and going and see where we land. But I feel like, you know, learning to use a new tool can sometimes feel like learning to use a new musical instrument. If you know what you're aiming for, so you know what you want it to sound like, you know what you want it to look like in the end, then you can pick up the tool much more easily. And I think that, you know, to me the strength in AI and in using that is to help us get to where we want to get to, rather than replacing the creative process.

And I think that, you know, to me the strength in AI and in using that is to help us get to where we want to get to, rather than replacing the creative process.

I'm probably slightly biased coming from a family of creatives, but seeing the damage that's being done there as well is not something that I'm excited about. So yeah, I think there's lots to be said for new tools, for using AI to explore them, but also for remembering the people behind the data, and also remembering the, you know, the ecological impact of all the queries that are running, which is not not negligible as well.

AI and the data science transition

With AI just becoming such a big topic these days, I want to know what parts of your job you see being automated by AI? And if AI has right now or even if you see it in the future impacting your business, and if so, in what ways?

Yeah, good question. And the honest answer is I haven't actually explored AI all that much. And so I've used it for write me a step-by-step instruction for how to do this thing on this device that I don't know. I've used it for get me the right regex string that I need. Although there is something so satisfying about getting that right yourself from the get-go. It's one of those, oh, did I do it? Yes, it worked. So I haven't really, I don't really use it at all regularly. Just every now and then for troubleshooting, but then it doesn't always get to the right answer anyway.

I think sometimes there is a tendency to think that AI can do everybody else's job, but not your own. So I've seen a lot of people talking about how it can be useful for cleaning up data. I've worked with Crystal. I've loved working with Crystal. And I know that I would trust Crystal way more than I would trust an algorithm to do that kind of stuff right. So I think there are specialisms that we don't always appreciate in other people's work that we think can be delegated, which I think is a bit of a dangerous way to be thinking.

To me, there's stuff about checking the accessibility of color palettes, for example. There are a few AI tools that are coming up in that field that are really quite exciting, or you can say tweak this and bring this in, and that's good. But then you still have to test it. You still have to check that it's doing the right thing. You have to check that it's not compromised on the creative vision that the client had. So I think using it for get me from this place to the next step is something that I can see it doing really well. I think maybe in terms of business day-to-day running, there's maybe a little bit more where it could be useful than necessarily in the super specialized skills.

Where I really do not like it being used is people who use it to churn out a whole bunch of content online and just, oh, do we need that? Does the world need that? Does the planet need us to use its resources in that way? I'm not convinced. But, yeah, I think using it wisely, no problem at all, but being aware of the need for specialists still would, I think, keep us on the right track.

Common pitfalls in data viz

What are some common pitfalls in DataVis and how can you avoid them?

Yeah, I mean, there are pitfalls. There are so many of them. One of the pitfalls is to just do what you've always done and stay comfortable within that. I think that comes back to Mike's question earlier. Sometimes it's best to ask for forgiveness than ask for permission when it comes to trying out new ideas in DataVis. I think a common one is to just throw too much information at people and to have it all be at the same kind of level of perceived hierarchy. So if you're putting a lot of text into a visualization, people are going to spend a lot of time reading that. If you pare it back a little bit, that's helpful. If you organize it visually, so you add some text hierarchy into it, making sure that people process it in an order that's logical, then that will help as well.

Often as data people, we come at things with like, oh, look at all this data, look at all these patterns, how can I show them all? When in actual fact, the question is, what do my users need? What is the first thing that should stand out from the visualization and what could I put behind the curtain, either with some interactivity or with extra graphs in it? I think that to me is a big pitfall that I see a lot. It's that desire to show everything that you've done in a graph, which isn't super helpful for the people.

Convincing people to care about data viz

Have you ever had to convince people to care about data visualization?

Yes, yes, I think, yes, I do. I tend to speak at conferences which aren't data-vis focused because the people who go to the data-vis focused conferences already care about visualization, and it's the people who go to the other conferences that sometimes I feel I need to speak to more. So yes, there is definitely the group of people who say, well, this is just making it pretty. I don't need that. Or do we even need it at all? One of the things that I do at the start of a workshop that I give regularly is set up this really silly scenario about needing to go to the palmer penguins' islands and needing to see two different species of penguins, but I am terrified of penguins with long beaks. Help me plan my trip. And then I give them the standard academic write-up of how many penguins are on each island and what the mean beak length is on each island, all that kind of stuff. And that's not super helpful. Then we move to a table, which isn't very helpful. Then we move to a table that's been stylized, which is a little bit more helpful. Then we move to an ugly graph of all the data, and then we move to a graph where I've spent some time organizing it visually so that it makes sense, and suddenly the story stands out. And I think a lot of the time we think it's a lot of extra effort for us to do it, but the effort that's reduced from the users is really worth it.

So I think sometimes, yeah, if people are looking to argue that DataVis isn't worth the time, then spending a little bit of time and then showing them the impact that it can have is a better way to do it than just going through conversationally.

Parting career advice

Good question. I think earn a reputation for being a good person to work with, and that is a holistic good. Okay, so you can be super skilled at this thing, but put people off with your approach, or you can be slightly less skilled and bring people in and encourage them and be a good team player. And I know which of those two I would prefer to work with. So I think that to me is a big piece of career advice. Get good at your skill, focus in on something that you think you could be world-class at, hone that craft, but don't neglect relationships, they are so important. Yeah, be a good community member with your skill.

Get good at your skill, focus in on something that you think you could be world-class at, hone that craft, but don't neglect relationships, they are so important.

Thanks so much, everybody. I'll hang out for just a second more so people can say goodbye and save that chat. And I'll say, see you later. Next week, we have Sajay Suresh, Senior Director of Data and Applied Science at Microsoft. So, it's going to be really, is such a fun person to talk to. And I look forward to seeing you next week. Bye, everybody. Bye, everybody. Nice to see you.