Storytelling through Data Visualization
This episode is all about creating meaning from data, and making it easy for your audience to understand by making it visual. We can apply the same principles we use to design dashboards to presentations - whether you’re communicating design concepts (and the data-driven decisions behind your design) or user research findings. I had such a great time talking with my guest Thomas Watkins about how to show meaning using data, I know the term ‘data visualization’ sounds a little overwhelming to some people in UX, if you aren’t designing dashboards. But it doesn’t need to be. I think we did a pretty good job of making data visualization accessible for any level of data interest in the episode.
Everyone in UX should be using data - whether you’re in design or research or PM or developer or marketing - and using data in a visual way to communicate important information to the person reading it. If you create presentations, for your design or for your research, if you have any opportunity to use data to explain your design or research recommendations… this episode has great information for you. Whether you are data-curious or data-shy or you lean in the data geek direction like me. I learned some great guidelines about making data visual in more effective ways, and I’ll bet you will too, even if your eyes glaze over at terms like “magnitude comparison” or “scatter plot.” I love how Thomas talks about explaining the context of the numbers, and not just ‘decorating’ numbers with meaningless donut graphs, one of my pet peeves!
Thomas Watkins is the founder of 3 Leaf consulting, a design collective that combines psychology and design principles to create usable products and services. Thomas is a thought leader, speaker, and industry practitioner in Houston TX. The scope of his work has included interface design for mobile, SaaS system architecture, usability research, and data visualization.
“By Separating the signal from the noise”
Understand which aspects you want to communicate to you audience
Communicate the context of the data
Don’t use visuals as “decoration”
Explain your graph of choice , don’t just assume everyone in your audience understands it because it has labels .
Highlight why your audience should care - what does the data mean
Thomas Watkins is a thought leader, speaker and industry practitioner located in Houston TX. He is a life-long learner who has a passion for bringing greater clarity to the world.
Thomas has made it his career’s focus to combine technology with design psychology in order to drive business success. He specializes in helping his business partners bring their own brilliant ideas to life, by translating complexity into simplicity. The scope of his work has included interface design for mobile, SaaS system architecture, usability research, and data visualization.
LINKS
https://www.3leaf.consulting/
https://www.instagram.com/3leafmethod
https://www.linkedin.com/in/watkinsthomas/
Show Link - Graph Selection Matrix
https://www.perceptualedge.com/articles/misc/Graph_Selection_Matrix.pdf
TRANSCRIPTION
Leigh Arredondo 00:01
You X cake is all about developing the layers you need to be more effective in your work and to be happy and fulfilled in your career. I'm your host Lee Alan arrow Dano, and I'm a UX leader and leadership coach. Hi, Thomas, thank you so much for joining me on UX cake. Today, I'm really excited to talk to you about visualizing data for more powerful presentations and kind of like data visual visualization in general. Where are you talking to us from today?
Thomas Watkins 00:35
I am in Houston, where I'm based and I met the Met, we work one of the WeWork offices.
Leigh Arredondo 00:42
Nice. Yeah. Well tell us a little bit about what you do in the world of user centered design and research.
Thomas Watkins 00:49
Yeah, sure. So I'm a UX practitioner. And I lead a team of UX folks, we tackle problems where people are trying to innovate in business. And so I work with a lot of startups, I work with a lot of companies who are trying to do something new, like a startup. And we ended up learning a lot of kind of like interesting, unique things in that space. And in addition to that, we kind of we do data visualization, as well, I am very, very into data visualization. And I've kind of been on a personal journey and or a mission for the past, I don't know, maybe 10 years or so to bring data visualization more into the consciousness of UX practitioners, because I think that's where it belongs. I think it kind of belongs with UX people. But for some reason, if you go to like data visualization conferences, and you kind of see what's going on in that space, it's mainly dominated by kind of technologists, and a little bit data scientists, but mostly business people who just like making fancy graphics. And so I think there's a lot of, there's a lot of opportunity for growth in that area. And that's something I'm very passionate about.
Leigh Arredondo 02:22
Yeah, I love that you love data. I think data is so important. And so that's why I wanted to, you know, talk about this. I know a lot of folks, particularly in design less so in research, who are a little bit maybe scared of data, maybe they just don't understand it, maybe didn't learn about it in, you know, in their schooling. And I think for designers, it's equally important that we don't have to be data scientists, we don't have to, you know, be like deep into data. But we do need to have data to inform our decisions. And we need to speak about how it informs our decisions and design decisions and research findings. So I'd love to hear a little more about your, your passion for data visualization. And how did you become so interested in it?
Thomas Watkins 03:14
Yeah, so the, I'm actually fortunate enough to have. So I have a big background in research. And I spent years toiling in a graduate school basement dealing with data and designing experiments, running participants, taking the data off of a floppy disk and loading it into a statistical program. And doing that day in day out for years. And with that, I had a developed a really strong comfort level with data where it wasn't scary for me. So I got used to thinking about data in terms of columns and rows, in terms of independent variables and dependent variables, in terms of descriptive statistics and things that you need to do to data in order to analyze it. And so when I got out of school, and I started working in UX, I did kind of regular UX like everybody else. But then I started getting into data visualization, I had to work on something that caused me to want to read Stephen Few's work. And then I went down a rabbit hole of getting deeply, deeply obsessed with data visualization, it kind of had a little bit of an advantage over maybe the average designer who might not have that kind of background. So when I was learning, data visualization, and what you're supposed to do with different types of graphs and displays and what process you're supposed to go through with the data. For me, it was quick to pick up. So I think it was a combination for me of both having the interest having a little bit of that background being in the UX space and a little bit on that. And then so also for myself, when I did started doing consulting, started designing a lot of dashboards. And so when you're dealing with dashboards, of course, you're dealing with a scenario where you have a target user and that target user needs to receive information as a part of their daily decision making process, and you need to optimize it for that. And so as a professional getting lots of practice designing lots of different dashboards, that put me in a position to kind of, you know, gain an understanding of this stuff. So by the time I started kind of teaching this in workshops, and trying to evangelize, this is something that I think should be one of those things in our toolkit, not that every UX person has to be an expert on it, but having it in our toolkit to where it's something you could be an expert on, kind of like accessibility, right? All designers should know something about accessibility, but not everybody is going to be an accessibility buff, right? And so it's a little bit similar with data visualization is its own big space, everybody can get like a little bit. So they're not being basically good with their data visualization design. But then there's a deeper path for those who are super interested in it a nice addition to UX teams, right? So if you have a UX team would say, you know, six people, it's kind of nice to have maybe one person on the team who's, who's kind of comfortable with that kind of problem.
Leigh Arredondo 06:06
Yeah. And actually, that was you kind of hit on what one of the questions that I had as a follow up for you about y'know, how much is enough knowledge? You also mentioned a book or an author, Stephen few. Yes. Can you give us the name of the book?
Thomas Watkins 06:24
Absolutely. So I think Stephen Few would be the kind of most important and accessible and very thorough body of work. And I think I would start with probably Information Dashboard Design by Steven Few. And then he has a number of other books that are that are good that get into different areas and different slices of the data visualization problem. Show Me The Numbers. Now You See It are a few others. it's a perspective that the purpose of data visualization is communication, okay, and what kind of communication? Well, you're communicating with a person's perceptive and cognitive systems. So you want to craft the visual attributes of your data visualization, so that they best speak to our way of perceiving. And so the ultimate goal is, I want a person to be able to look at this data display and be able to gain insights very easily, and not just gain insights. But particularly what's important about this data, right? different data sets have different types of things that are important about them, you know, so for example, for, you might be looking at data for stocks, and the most important thing might be trending direction, and volatility, right. Or you might be comparing the performance of multiple different entities, most important thing to grasp from it is a quantity comparison .So depending on the type of thing that the person is supposed to get from it, we can actually craft the data display to optimize for someone's ability to see that to separate the signal from the noise in a particular data set. And that is the perspective that we want to approach as data visualization is. And as UX people who are adopting that as a skill set, we want to learn a little bit about about how to do that, how do people perceive things? What is the particular use case with the persona and what they need to glean from the data? And ultimately, what are the best practices of how to bring all that together? And we can do a lot of great stuff in this space doing that. Yeah, I think maybe it helps to even bring an example of research. So people can who are listening can sort of visualize what you're talking about. So let's say there's a, there's a data set that you have from user research, and the number of people who's rated something, you know, a concept, let's say on a scale of one to five, or one to 10, and you're trying to communicate something about something meaningful about that data. So just using that as an example, or if you have a, you know, another example that's better.
Leigh Arredondo 09:40
But yeah, just kind of talk talk to me about that a little bit. How to think about it, I guess, if you're, if you're sort of if you want to say, I want to make this really clear this, what I'm trying to get at that, you know, most people were kind of neutral. You could look at the numbers, right and say, oh, yeah, there's, you know, 30% of the people said it was great. But then let's look at what's what's this the rest of the 70%?
Thomas Watkins 10:06
Correct? Yeah, so um, so a lot of times, what we're trying to do is compare quantities of things. And so when we look at how do we most accurately judge the the size of a certain thing, were very accurate if we're talking about like the position of dots, or the length of lines. And if you do, and when I teach data visualization by workshops, I start off with an exercise where people have to make judgments about different objects and things and we judge the accuracy. And they're able to see for themselves that we're better at some things than others. One thing that we're bad at is judging the volume of two dimensional areas. And we're even worse, if it gets three dimensional, we tend to dramatically underestimate. And there's, there's actually equations that can predict how bad we are at this. So that's one thing, if we want people to judge quantities, you know, maybe, for example, a bar graph or a dot plot that to be confused with the scatterplot, a dot plot where the position of the data is like the end of the bar graph, you know, that is more effective. And also, the other issue is the vocabulary of our data visualization. So people run to pie graphs a lot, because it's a very well known type of graph. But pie graphs are specifically designed for parts a hole into information. And that's not always what you're interested in. So if I have a dataset that has, let's say, 100 different categories, and I say, Okay, well, we're trying to compare the size of each one, put it in a pie graph, well, first of all, there's going to be too many slice slices, pie graphs don't scale, well, you know, maybe take the top 10 or something like that, plot those on bar graphs, and then you're able to compare quantities. You know, there's other problems that pie graphs, you're comparing two dimensional area, you're comparing your mapping angles, because of the slices on two quantitative judgments. And we're also bad at that. So it takes a bunch of things that we're bad at and puts it into one graph. So you know, each type of graph is good at the different things, scatter plots are good for showing how correlated two items are. And that's what they're in vintage for. And that's kind of what you're able to do with it. So it's a little bit about that it's a little bit about from a design pattern perspective, no different from regular UX, where we want to be aware of existing design patterns, lists versus dropdowns, versus, you know, other types of things. It's the same thing in data visualization, there's different things tools you have at your disposal. And you know, you shouldn't pick it based upon what type of graph is your favorite type of graph, as you you'll often find yourself in that kind of situation with maybe an executive who likes a certain type of graph that says, you know, I like scatter plots, or I like bubble charts. And, but we have to be there to kind of inform the process and say, This is what this graph is good at. And it's not good at this other things. So we want to be able to do that. Did you want to do like a specific example? Like a specific?
Leigh Arredondo 13:04
I know, it's hard, because so many of us are visual thinkers. We're using complete words. Yeah, so please tell me in words, what that would look like. But actually, yeah, could use kind of describe? Well, I have two questions. Right. My first question actually, is, how does one go about figuring out what the right graph for visualization is? And then maybe we can just kind of talk through an example that maybe someone would use because I'd love to talk about this in a way where people could use this in their presentations. We'll get to that too, a little bit more. But you know, how to visualize user research data, for example. So first question, I kind of molded those together. First question being, how does one find out or figure out what is the right way to represent this visually? And then maybe if you could try to describe visually what it might look
Thomas Watkins 14:04
like? Yeah, so I would advocate with all of UX. And with data visualization as well start with the persona, considering the persona, user centered design, who is the recipient of this data? Some things that you might consider there is, what are their skill sets, proclivities, and other things like that? Then the scenario, what decisions do they need to make based upon this data display?
Leigh Arredondo 14:33
Yes, so that's a very, very good point there because like, sometimes you have to present this information to program managers or product managers, or product owners. And that's very different than presenting to the sales and marketing team, and very different than presenting to executives,
Thomas Watkins 14:53
right. Totally. Yeah. And then if we're talking about the President's presentation, we have the benefit of at least realizing what the main point about this dataset is that we want to communicate and designing exactly for that, rather than the more complex situation of designing a dashboard, or a situational awareness display where we have to anticipate what the data is might show and designed for that, which is, which is more complicated. But then so if we're talking about presentations, I think that's a good start, you know, who is the audience? Like you said, if it's, if it's sales, folks, if it's executives, there's different aspects of the data. That's interesting. So we want to understand the data first, as the folks working with the data and look at it and say, Okay, what am I trying to communicate, and you want to communicate it in a way that's honest, yet gets the point across, right. So you don't want to do dishonest things with your graphs, like exaggerate things or make them you know, not to scale. To make your point. You want it, you want it to be honest. And but you want it to be enlightening. So one of the one of the trends that I've seen a lot for, I don't know, maybe the past 10 years or so, I call it the donut stamp. And it's the, it's a donut graph with just a number in the beginning in the middle of it. And you'll see dashboards and are very common, right where you see like a number, and this is going to label. And it's kind of this perspective that the data consists of maybe an interesting number that's there to be decorated. And that's a very common approach to data visualization, of just saying like a number, and then I'm going to decorate it. Well, putting together your presentation, instead of decorating your number, surround it, not with a decoration, but with context. So what context might be interesting. So if we say that our our annual revenue was $50 million, compared to what is it compared to previous years? Is that the interesting thing that we want to communicate? Is it compared to other big players in the industry? You know, are we making some kind of a competitive or strategic statement with it? What is it that we're trying to communicate, so that would be kind of my one major piece of advice that you'll go very far just trying to do that is plotting the data along with context, that makes for a much more interesting and useful data visualization,
Leigh Arredondo 17:26
I can totally see that. Because, and I see this a lot with research findings, especially the team is so wrapped, you know, is so deep in it, and they really, they come to conclusions. And they they give these numbers like, Okay, but what does that mean? Like, that seems significant is significant enough for you to raise it to the level of, you know, an executive summary, don't make the the audience have to figure out what that means?
Thomas Watkins 18:01
Totally. So there's so many things to speak on there. Because putting on your researcher hat is a little bit what you're doing in that case. So like, I'm thinking in terms of like academic standards around presenting data, you want to set the proper stage, kind of like what you're alluding to? Why should people care? What numbers are kind of expected? I think setting the stage around the research in general, we need to do that even before getting to the data. And so that's kind of what is the research question. You know, why matters? And then going into, okay, you know, this is what we did, and all of that stuff, and then presenting the data in a way that it's just the results are clear. So if it's, maybe it's let's say it's results from a Likert scale, and the strongly disagree to strongly agree for, for listeners who may not be familiar with the term I use, you want to show a lot of times in those cases, the summary of how people answered it, here's a couple of things to think about. How do we summarize the data? Because that's really all we're trying to do with data visualization, we're trying to visually summarize it. But even before that, with data thinking, right, thinking of statisticians, not just the design thinking, but the data thinking, how do we give people an idea of what's going on with the data, you have measures of central tendency, and you have measures of distribution? I think these are sometimes, you know, overlooked, we think about mean, median, and mode with measures of central tendency, what did the data tend to do? And then you have your kind of indicators of how to how was this data distributed? So, range, standard deviation and things like that. And, you know, if you if you plot the district in a histogram, you can see the distribution. So imagine that for each question. item on your Likert scale, you have, you know, the bar graphs, and vertical bar graphs stuck together so that it looks like a distribution with the strongly disagree to strongly agree, all plotted together. And then now you can see the shape of the data how people answered and draw like a line vertically, where the mean is usually you're going to be using mean for that so that people can see. So now you're not just saying like, well, the average person said, Oh, they agree with that statement. Well, you could see agree, but you get the richer context of how do people generally answer and then this is, you know, where people land. And now people can see, like some of the data that might not have been reported? Like, is it bimodal? Or is it like heavily skewed and, and things like that. So presenting the results, setting the audience up to where they're going to be able to, they're ready for the results right there ready to perceive it. And then, and then marching through the results is the way to do it. Here's another tip. Another thing that we don't do often enough, folks, in general don't do it often enough, is they just don't give the proper orientation around how to read the graph, they throw the graph up, the audience is kind of stuck sitting there staring at the screen and kind of looking at it and kind of trying to interpret it, throw up the graph and say what it is, even if it feels obvious, it's a good way to orient, people tell people the x axis is showing the responses. The y axis is showing how many people responded to each one, boom, now you've added so much more clarity and taking taken so many question marks out of people's heads and got it to where they can know now they're where you are where you're they're interpreting the data,
Leigh Arredondo 21:46
right? Even if it's labeled, like you're talking, even if it's legs, you're talking through it just like actually, that's right to what what it is you're
Thomas Watkins 21:55
That's right, that's right. As soon as you throw up the graph, say like, oh, this is the such and such graph, the title of the graph, you might say what the graph is. And this is a time series showing number of sales across three departments over the past two quarters. And you know, it Okay, time series, now they know kind of a little bit about it. And then the y axis is showing number of scales. Okay, so now everybody's know everybody's on the same page. And that goes a long way to print on presenting data.
Leigh Arredondo 22:25
One of the things that you kind of alluded to, which is like what I love about data and looking at kind of the anomalies in data, so you mentioned like a Likert scale, and you've got showing the median, I think, is what you said, and then you sort of like hurriedly said, you know, and then you know, there might be interesting skews and data or you know, some other anomalies, which actually is, is what a lot of people miss, right. And so I just wanted to make sure that we talk about looking for the interesting story and data, and then telling that interesting story in data in a visual way. Excellent point, I guess. Yeah. And I'll just hand it to you. But I think sometimes it's really, you would get us a very different story, if you just did the mean. And you didn't show Oh, everybody either either loved it, or hated it. You know, it's sort of like, it's actually,
Thomas Watkins 23:24
that's right. Yeah. Yeah, no one was in the moment, and everyone was on the sides. Yeah, that's an excellent point. And yeah, it's this is just a really good sub topic of it. Because it also speaks to be one of the advantages to being able to present the data versus the complexity is around designing a data display where the data is going to be different and vary each time, you can get familiar with the data. And you can find out what you think is interesting about it, and highlight it, highlight it in the actual data, display, circle things, point to things and show people why they should care. Now, you know, hopefully, you're not doing it in a deceptive way. And hopefully, folks in the audience are informed enough to be able to see see the data also for themselves. But in terms of communicating insights, what you're doing is you're you're giving folks the point that you want them to get the same way they would receive it if you just wrote a report or something like that, but they're also seeing the context of the real data. And so that's kind of one way of handling that. So that might be like in a line graph where there's a big spike at some certain point and people can see it, but then you might, you know, kind of highlight what that is for changes in trends, you know, and other things like that. But yeah, presenting it as a good opportunity to be able to do that.
Leigh Arredondo 24:46
Yeah. You mentioned earlier about instead of decorating numbers, I loved how you put that instead of decorate numbers, showing the context. Do you have any Can you give us any more sort of things? examples of how someone might do that visually.
Thomas Watkins 25:02
Yeah, so there's the donut stamp. That's a one big example, I think the biggest example I can think of is, right now is benchmarking showing a set of magnitudes or something like that. And you're saying like, let's say you're showing the, you know, our sales for this year, or, I don't know, a number of downloads or something like this, you might not even have the data that to compare it to, but you can oftentimes find it. Sometimes you can generate the data, if you have that your organization, if you have access to the folks who, you know, maybe you have a data science department or something like that, ask people if they have they collect data on anything, and sometimes you can find stuff. Sometimes you can go to those, like those Gartner media reports and things like that and find, you know, industry averages for things like that. Find data on it. Now, this gets a little bit tricky, because you'd have to make sure that you understand enough about the data that you're doing apples to apples comparisons, sometimes it's you want to think first about what matters. And, you know, why should it matter? You know, we hear oftentimes, like, you know, in the news, you'll hear like big numbers, this will add, you know, such and such a billion dollars to the deficit go oh, my gosh, that sounds like a bit. Is that a big number? I don't know, sounds? I have no, I have no idea with it. That's big. But you see that happen all the time, where the magnitude or the you know, there was a 700% increase in such as compared to what is doesn't normally very like that is it is held or more wild than before. That's part of the integrity of this discussing and communicating data is getting in the habit of giving context. So if the integrity plus the clarity, and helping people really be able to interpret things in the proper way.
Leigh Arredondo 26:54
Yeah, I love that. You also mentioned going out and finding data, like secondary sources of data, there's so much information out there that it can take some digging, if you don't have access to like a page source of industry standards, but there are paid sources of industry standards out there as well. So yeah, going out and finding secondary data that you could use as a benchmark is is can be super helpful, like, Okay, so we're seeing this sort of behavioral data, right, in our on our website, you know, is that good or bad? Percentages can be really deceiving. When you're talking about something like any kind of behavioral information on a website, it could be like we have 3% conversion, is that good?
Thomas Watkins 27:49
That's right. And then, and then here's, here's another one, sometimes you have to calculate the data based on something that's already being collected. And I'll give you an example. With a lot of startups when they're bootstrapping and just trying to get the code built, they'll collect data that's collected in like a logging sense, but not Dilek. Set up for BI. So here's an example. Let's take your example of like website, conversions, what you might see is like you might be gathering, how many customers ever have bought something. And what they might have on the back end is just an ever accumulating number, like each new purchase just adds to the total. And then that's, that's the only metric that they have. And then, and then getting the other, you know, it might be like, surprisingly complex to build the back end to be able to collect that. And there's a lot of things like that. And in the startup space, you'd be surprised like how many just like it's just logging in evergrowing accumulation? Well, you can calculate something more useful from that, even though the Data Science Department says they don't have those numbers, you can take that number and then divide it by, you know, how many folks per time period converted on our on our service. And then you're able to now have metrics that are more meaningful, you can say that, you know, 70 people per day, bought something on the website, you know, this year, on average versus last year, when it's 30 people per day, now you're able to get something meaningful from it. So then, that's where I, you know, I talk a lot about data thinking, where you have to think, design thinking not only for the data, insights and how to put it together and how to design the graphs and so forth. But you also might have to engage in data thinking so you, you have some massive data that's being collected, what are the things that you can do to the data to get the numbers that you want? Another example of data thinking might be computing ratios, because a lot of times what might matter is a ratio we deal with ratios all the time, but we don't necessarily think about it like when you drive down the street, your miles per hour, that's that's a ratio. Right? And it's, it's meaningful, and that's how you get arrayed. So sometimes that's another interesting thing you can do with the data is, is say like, Okay, well this thing by itself kind of matters, but not much without context is other thing matters. But if I do sales per such and such one, now I have a ratio, or how many conversions compared to marketing dollars spent on a campaign. Now we have this meaningful ratio that built in has the relationship between two different things that I care about, and automatically tells me the relationship between those two factors. So that's a lot of data thinking, that kind of plays a role in being able to do data communication, effectively,
Leigh Arredondo 30:40
data thinking, is just the thought that came into my head is That sounds scary for someone who, you know, maybe doesn't think data is cool. Yeah. I am curious, like for the folks who take your workshop, like, what tends to be one of the, like, biggest, is there anything that people find surprising that they thought was going to be kind of like, ooh, scary? And and it turns out, like, what makes that shift? Yeah,
Thomas Watkins 31:13
I think a lot of it is realizing how ineffective the kind of impulsive first response of how would I visualize this data often can be and how it's not that complicated to pick something that's like, way more way more effective. You know, one of the things that we've got to think about when we visualize data, is we're not making diagrams, we're plotting points. So like, a lot of times, you might be in a business meeting, someone goes up to the board, and they say, like, well, here's how I see us versus the competitors. And then they'll start, they'll draw quadrant, and then they'll put dots that represent the different players in that. And then, so that's fine for a diagram for you just drawing something because you're kind of showing the relationship. But if you're thinking about data visualization, you've got to prioritize, how does the quantity of the different data points yet encoded visually in a way that the meaningful thing comes out. So if I'm comparing the magnitude of different things, I want to graph that maximizes that so bar graphs are very good or bad. If I want to show, if I want to show the shape of something, I might choose a line graph, because that's something that I pre attentively can get a gestalt of how is this thing shaped. And so if I'm trying to show trends or something like that, then you know, a line graph will often be very effective for that. Or if I'm trying to show the relationship between two variables, a correlation, you know, scatterplot, might be the most effective depending on on exactly what you're doing. But what you can see, instant, instantly perceived features of the graph are apparent right away. And now it's just a matter of just being properly oriented so that your, you know, your perception is directed in the right direction. But I think I think once folks kind of get used to it, it's kind of fun. But it but if I had to boil it down to kind of takeaways that I think if you're not such a big data person, some things that you can think about, I would say, measures of central tendency, so giving people an idea of what is the data generally doing. So mean, median, and mode, usually mean, aka average, median is used for datasets where there are extreme examples that will skew the data. So there's no longer representative. That's why median often used for things like income, because you have a few number, a small number of celebrities and super, super rich people who would would drag the line too far over. So you use median, because it's a robust against extremes, like things like that giving people an idea of what the general data generally do measure of central tendency, and then a measure of distribution. How is this so you can either display it in some kind of a distribution, standard deviation is a little bit sophisticated for maybe the average audience that may not be that into the academics of it, but I think most of you will kind of get in range is a simple one, but not as descriptive. But I think getting this area is called descriptive statistics for the listeners. And if you just get the basic descriptive statistics, that's usually fairly good for giving people a good impression of it, and then and then plotting it in a way that that that it can be understood.
Leigh Arredondo 34:41
Yeah. So for someone who is interested in having more effective displays of data, how could they learn without like, diving into I gotta say tough too tough these books. Yeah, I don't know. Much Sorry about the other author that you mentioned, but like, How could someone kind of figure out okay, I have some data, what is the right way to display this? You know, like, based on the principles that you've been talking about the principles of perception,
Thomas Watkins 35:17
one thing they can do, they can sign up for my workshop, which will be USPA. International 2022 in San Diego, but if you if you don't have time for that, and if that's not in your schedule, yeah, there are there are some recent, The tricky thing is finding things that are really, that you can trust. And by trust, I mean, coming from a perspective. That is That is correct. I'm looking up right now, Stephen view, did kind of like this one page, cheat sheet. That kind of gets you pretty far. And I'm trying to look up the name of it, because they use it a lot. Yeah,
Leigh Arredondo 35:55
we can, for sure. put links in the show notes. Yes.
Thomas Watkins 35:59
And so I've found there's, there's some, there's some bad cheat sheets that I do not recommend. But there's there's some good ones. Yeah, there's one that's called a conversation starter that floated around for years. It's called Chart suggestions. A thought starter? It looks interesting. That's,
Leigh Arredondo 36:18
yeah, that's the bad one. Okay, so don't use Don't
Thomas Watkins 36:20
Don't, don't use that one. What you want to use is if you want a cheat sheet, there is graph selection matrix. My Stephen few perceptual edge, there's a bigger problem, I think it's the preference, political aspect, the
Leigh Arredondo 36:37
preference of the person seeing it, you mean, no, the
Thomas Watkins 36:41
the preference of the the folks who work above you often, who will have their own favorite types of graphs, favorite types of displays, they want things to be flashy, and colorful. And there's this misconception that if you don't make your data display, flashy and colorful, and interesting enough that people won't be engaged with it. And that's an incorrect assumption, the easy example I like to give to people is, let's say that you woke up one morning and you logged into your bank account. And then you saw that your bank account had a million dollars in it, you'd be very excited, you don't need the text for 1 million to be flashing or moving or colorful, in order for that to be. Because it's it's something that it's data that's already very meaningful to you. And so you don't need the declaration. And so it's really the engagement is really about finding things that matter to people presenting that, and highlighting the things about the data that will be that will be engaging to the end. And I promise you every single time no matter what the recipients, no matter what their job role is their job title, whether it's blue collar work, white collar, no matter what kind of, if you show somebody a dashboard, or a display, that shows metrics that are important to their day to day work, they will interpret it and if there's something interesting going on there, that's that would be meaningful and meaningful in real life, and you're plotting it on a data display, it will be interesting to them. Because that's how people that's how people perceive it. And so I think a lot of that is really one of the biggest hurdles is working with executives, a lot of it is going to be an educational process. And so doing some testing along with it, saying, Hey, we showed this to the actual users, it performed well, with people being able to use it with people being engaged, and having some of that stuff to kind of back it up will help get over those hurdles, if you have, you know, open minded people in your organization. But that is very often a little bit of a battle that needs to be to be fought.
Leigh Arredondo 38:44
Yeah, that seems like a really pretty important takeaway, which is find out what's actually most meaningful to the audience that you're showing the data to. And then it will be and and, and then takeaway number two is make sure that you are having the
Unknown Speaker 39:06
context. That's right. That's right. That's right, that makes it meaningful.
Thomas Watkins 39:12
What does it mean? Where people are able to get it and get insights and creates a lot of interesting conversations around the data as well.
Leigh Arredondo 39:20
I love that conversations around the data. So in summary, you gave us some resources, and I'll have a link to that graph selection matrix for like cheat sheet. That sounds awesome. You mentioned that you've got a workshop coming up. Tell us just a bit about how people can find out about what you're what you've got going on because you do workshops occasionally.
Thomas Watkins 39:45
Yeah, absolutely. You can do relief dot consulting, and that's a relief. Relief. Yeah. The number three and then the word leaf consulting. Find me on LinkedIn. Thomas Watkins. Yeah, in the the workshop that I mentioned. It's kind of designed for business professionals and UX designers and covering what are some of the approaches we want to make for effective graphs for presentations, for dashboards? And for data discovery platforms? What should we be thinking about? And what techniques can we use? And we do we practice examples where we redesign poorly constructed graphs, practice, redesigning them, discuss them as a group, and practice taking data and materializing that into data visualizations and discussing the pros and cons of different approaches, and getting it to where we get an increasing comfort with the practice of data visualization.
Leigh Arredondo 40:42
That sounds like such a great opportunity for anybody who can take that workshop. I think you mentioned when we talked before, that you also do workshops sometimes in your area,
Thomas Watkins 40:53
right? Correct. Yeah. So I talk a lot about design psychology, and the importance of when we're designing products, when we're solving problems, thinking about things in a very user centric, you know, human centered design, where we're taking into account and understanding of how our people will construct it, and generally, and then how our is your user, specifically? And considering all those kinds of factors when we're designing things, designing products that needs to meet people's needs, it needs to resonate with users.
Leigh Arredondo 41:27
And that's in Houston. Yes. Are you do these talks?
Thomas Watkins 41:31
Mostly I travel around, but yes, I'm situated in
Leigh Arredondo 41:33
Houston. Thank you so much, Thomas. Thank you, and I really enjoyed our conversation. And thank you so much for sharing with the audience. Absolutely. Hey, if you enjoyed this slice of UX cake, please rate it and subscribe. tell others what you liked about it. It really helps us spread the word and get this free content to more people. You can follow UX cake on LinkedIn, Facebook and Instagram, and get all the episodes and show notes at UX kake.co. Thank you for listening and sharing the UX cake