Are you curious about how to tailor your data visualizations to either explore new insights or clearly explain your findings to different audiences?
How can you effectively use data visualizations in clinical trial reports, interactive dashboards, and more to achieve your specific goals?
In this episode, I’m diving into the fascinating world of data visualization. I explore a crucial yet often misunderstood distinction: exploration versus explanation in data visualizations.
By the end of this episode, you’ll grasp the 10 key differences between these two approaches, helping you tailor your visualizations to meet specific goals. Whether you’re designing a clinical trial report for regulators or creating interactive dashboards for data analysis, understanding these differences is essential.
Join me as I unpack how to effectively communicate your data’s story or discover new insights through tailored visual tools.
Episode Highlights:
- Purpose
- Audience
- Design
- Interaction
- Task
- Usage Time
- Improvement
- External Impact
- Software
- Creation Process
By understanding and applying these key concepts, you can significantly enhance your ability to communicate data effectively and discover valuable insights.
If you found this episode insightful, please share it with your friends and colleagues. Let’s spread the knowledge and elevate the standards of data visualization in our professional community.
Transform Your Career at The Effective Statistician Conference 2024!
Don’t miss this conference, a transformative event focused on “Transforming Healthcare.” Join us on November 7-8 and 11-12, 2024, to gain insights from leading experts, engage in interactive workshops, and expand your professional network. Secure your seat now and be part of a community driving innovation in statistics and healthcare! Register today and take advantage of our early bird pricing.
Register now!Never miss an episode!
Join thousends of your peers and subscribe to get our latest updates by email!
Get the
Learn on demand
Click on the button to see our Teachble Inc. cources.
Featured courses
Click on the button to see our Teachble Inc. cources.
Transcript
Exploration Or Explanation: 10 Key Differences For Data Visualisations
[00:00:00] Alexander: Welcome to a new episode of the effective statistician. And today we will talk about data visualization, which is [00:00:10] a really, really great topic. As I We talked about this in lots of episodes already on the podcast. By the way, [00:00:20] we have these nice new playlists on different topics. And if you’d want to learn more about them, just check out the [00:00:30] effective statistician dot com where you will find them very, very easily.
[00:00:35] Alexander: So today I want to talk about two very, [00:00:40] very different things in terms of data visualization that people very often confuse or that they are not [00:00:50] clear about it. And that leads to lots of confusion, misunderstandings, and these kind of different things. And these two [00:01:00] concepts are explain versus explore.
[00:01:04] Alexander: So data visualizations can help you to. [00:01:10] Explain data. So that means that you already have a story there. You already have an understanding of what you want to say about the [00:01:20] message and, you know, the audience and the circumstances. And then you can tailor your data visualization to this specific [00:01:30] audience for these specific purposes.
[00:01:32] Alexander: So. For example, this could be a figure that you include in your clinical [00:01:40] trial report so that regulators can understand it and you get basically approval based on [00:01:50] these figures. Another example. Another thing could be you have a poster and you have a figure on the poster and [00:02:00] there it is. It needs to be self explanatory.
[00:02:03] Alexander: It needs to be pretty fast to understand because people won’t take a lot of time looking at the poster. [00:02:10] And you want to probably have it for a specific audience that usually attends this conference and the audience you are interested in. [00:02:20] A completely different thing are exploratory data visualizations.
[00:02:28] Alexander: Exploratory data [00:02:30] visualizations serve you to understand, well, explore the data rather than to explain the data. [00:02:40] For exploratory data visualizations, these, for example, are tools like Spotfire or Shiny apps. [00:02:50] These kind of things that help you to dive into the data, to see patterns, to check data, to actually find the [00:03:00] stories that you then want to communicate about or drive certain actions from.
[00:03:07] Alexander: So, Whenever I [00:03:10] talk about data visualizations to people, there’s very, very often either focus just on one or the other, [00:03:20] and that leads to a lot of problems. Now, I want to go through a couple of different examples. Major [00:03:30] differences between these explaining and exploring type of data visualizations. So first is in terms of [00:03:40] how you design this.
[00:03:42] Alexander: So if you have a tool set explores data, then very, very often you will [00:03:50] have something like a, a standardized way of looking into the data and the tools. stays the same, and you exchange the data. [00:04:00] So for example, you might want to have something that looks into the safety data of a specific study. And then [00:04:10] you look into this again and again over the time of the study, the more data you get in, or maybe it’s [00:04:20] across a complete compound with multiple studies and you get more and more data.
[00:04:25] Alexander: And you look into this with the same tool again and again. [00:04:30] Maybe you can even do that across different compounds or studies and have then more or less the same tool. [00:04:40] So you have a standardized design and you change the data with [00:04:50] customized, high quality. Planning data visualizations, that is very different.
[00:04:58] Alexander: There, [00:05:00] you customize your data visualization to specific Data to a specific audience, to a specific purpose, [00:05:10] to a specific message. So, a really good data visualization will be different whether you put it in a [00:05:20] report, or whether you put it on a poster, or use it as slides that you present and talk to. It will look different whether [00:05:30] you show it to your CEO, or whether you show it to your internal colleagues.
[00:05:35] Alexander: Or to treating physicians or to key opinion leaders. It [00:05:40] will always kind of adjust to the different audiences. So you have a lot of customizations here. [00:05:50] Because in this customization lies the efficacy of this data visualization. Yeah, whether it [00:06:00] achieves its goal or whether it doesn’t achieve the goal.
[00:06:06] Alexander: Goals are also very, very different. [00:06:10] And we will talk about this in a minute. The next point that is very different is the design. [00:06:20] These exploratory data visualizations usually have a very, very technical look out. Because, you know, you [00:06:30] yourself or your team or colleagues will work with that. So it might actually have.
[00:06:38] Alexander: The actual [00:06:40] variable names in there. So the, you know, what you get from the atom data sets, things like that, so that you can easily [00:06:50] identify and go back to the programs. You will never have that in a data visualization that you show at a conference. [00:07:00] You also very often have it like just from a design, from a look and feel, it will look very, very technical because you will probably not spend [00:07:10] a lot of time in looking exactly what kind of figures colors you have.
[00:07:16] Alexander: Kind of, maybe you have many more colors because it [00:07:20] needs to adapt to kind of different settings of the variables and different settings of the of the data. In a [00:07:30] customized, high quality, explaining data visualization, there you will have something that looks [00:07:40] very attractive, super professional, like it was done by really good designers.
[00:07:49] Alexander: It [00:07:50] will have very, very clearly discussed and identified colors, whether that is for a certain branding and these kind [00:08:00] of things. It Will just look much more beautiful and sometimes in this beauty. There’s even [00:08:10] some Function in it. Yeah, it needs to be Attractive people may want to have a look at it because it looks really really [00:08:20] nice So it attracts people to stop at a booth and actually look at it Or just not, you know, [00:08:30] really look at the poster and spend more time or not just scroll over it in your mobile phone, [00:08:40] but actually stop scrolling and having a look.
[00:08:43] Alexander: All these kinds of different things can be triggers for creating something that is [00:08:50] attractive and very, very appealing. Of course, another difference, and you have probably noticed this already, is [00:09:00] very often the audience for these explanatory and exploratory data visualizations. Whereas, when we talk about [00:09:10] exploratory data visualizations, these are usually scientists, you, your colleagues, your team, yeah, experts that [00:09:20] work with the data.
[00:09:24] Alexander: When we think about explaining, these are usually very, [00:09:30] very different audiences very often less scientific or very, or less knowledgeable about the data. These can be [00:09:40] upper management. External audiences but also including regulators and, and payers. Yeah. [00:09:50] So people that are not directly touching the data itself.
[00:09:57] Alexander: Another difference is [00:10:00] how you interact here with the different data visualizations. So if you have, exploratory tool, then of course you [00:10:10] want to interact with it quite a lot. You will have all kind of different features to work with the data and spend a lot of time with the data. [00:10:20] This is also true This is also possible for explanatory data visualizations, but this is usually much more [00:10:30] limited.
[00:10:32] Alexander: Where, and that is the next difference, we have for tasks, yeah for exploratory data [00:10:40] visualizations, we talk about filtering, searching, selecting, combining, all kind of different things, maybe even analyzing, you know, [00:10:50] summarizing. These kind of things. So, what is this? This is what it means with exploratory data visualization.
[00:10:55] Alexander: With explanatory data visualizations we often have [00:11:00] no interaction, just to view it. Maybe there is some scrolling or a hover over functionality. But it is [00:11:10] usually very very limited. Tell a very, very specific story, and then it [00:11:20] doesn’t make a lot of sense usually to have a lot of interactions with the data visualization.
[00:11:28] Alexander: The next [00:11:30] thing that is related to this task and the audience is how many times people actually work with this data visualization. [00:11:40] If you think about an exploratory data visualization, people work minutes, hours, hours. I was maybe even days with the same data [00:11:50] visualization with explaining data visualizations.
[00:11:55] Alexander: It’s more kind of one minute and less. Yeah. [00:12:00] So how long will people look at a figure in a manuscript or report? Yeah. Maybe a little bit longer, but [00:12:10] usually you measure that in seconds and not in minutes or hours. Another interesting aspect. Is how [00:12:20] you improve these kind of data visualizations over time with the exploratory data visualization, because you [00:12:30] usually use it again and again and again and again.
[00:12:34] Alexander: You can actually improve it and because you use it yourself or people [00:12:40] that you know use it. You can build in feedback loops and customize it, add to it delete certain aspects [00:12:50] that are not helpful. All these kind of different things. Whereas Set is much harder with explanatory data visualizations.
[00:12:59] Alexander: [00:13:00] Because in these settings you will. You will often not exactly know who is looking into this.
[00:13:07] Alexander: Another big difference, [00:13:10] and that is something that people that focus only on the exploratory side very often get wrong, is [00:13:20] the external impact. The external impact for these exploratory data visualizations is [00:13:30] usually close to zero. None. Nada. None. The external impact of these [00:13:40] explanatory data visualizations used at advisory boards, used in conferences and publications and promotional material, all [00:13:50] these kind of other areas can be huge.
[00:13:54] Alexander: It works on a very, very different level. It reaches far more people as we talked [00:14:00] about, and it can make a huge impact in terms of how maybe your compound, your study [00:14:10] separates from others. I’ve seen that. For myself, where we created a very, very specific, very innovative [00:14:20] way of showing how patients improve over time and actually including individual patient level data.
[00:14:29] Alexander: And that [00:14:30] was so impressive that people really loved it and first time really understood the variability between patients and within [00:14:40] patients over time. They first had a really understanding of how different levels [00:14:50] of response related to each other, how baseline data was related to follow up data, how, whether all patients more or less behaved the [00:15:00] same or whether there were very, very big differences or subgroups of patients.
[00:15:05] Alexander: All that was with this new [00:15:10] data visualization, and we got lots of very, very good feedback from clients. And I think the best feedback was that [00:15:20] the competition actually copied it. So they obviously saw the value as well. Another topic that I want [00:15:30] to mention is usually there’s a difference in terms of What software you use to produce it.
[00:15:38] Alexander: So if you think [00:15:40] about exploratory data visualizations, you will also think about interactive dashboards [00:15:50] like ShinyTools or Spotfire or other software. Of course, there are certain things you can do with Spotfire. With both [00:16:00] software, yeah, but usually, nowadays, you will look into something like RShiny or Spotfire.
[00:16:06] Alexander: For the explanatory [00:16:10] things, yeah, you know what is probably the tool that is most often used? It’s Excel. [00:16:20] Because most often these data visualizations are actually produced by people outside of stats. Statisticians will [00:16:30] probably more go to SARS or especially to R with its really, really nice data visualization capabilities.
[00:16:39] Alexander: [00:16:40] Jump is another tool that maybe more people in other areas work to explore the data. [00:16:50] Finally, a big separation between these two things. is how you create these data [00:17:00] visualizations. The exploratory data visualizations there you are your own audience or your [00:17:10] colleagues and you will create And design it having your own needs or those of your [00:17:20] colleagues in mind, and you don’t need any further inputs there.
[00:17:27] Alexander: This is very different to [00:17:30] explanatory data visualization, where you are not the audience. You’re not part of the audience. In order to [00:17:40] understand what the audience really needs, you actually need to talk to them. Or at least a sample of them. So you need to co [00:17:50] create the data visualization together with the audience.
[00:17:53] Alexander: You need to get feedback about it. This great data visualization that I talked [00:18:00] about We had lots of checks about with the audience should the scatterplot which was animated move up or should it move [00:18:10] down what kind of colors should be in there, how we design it, where we order it where, which kind of [00:18:20] Data we prune out because it’s not so necessarily, these are more kind of these extreme cases, lots of, lots of different things [00:18:30] we talked about with the audience to make it fit for the audience.
[00:18:36] Alexander: And so. You need to co create, you need to [00:18:40] spend a lot of time talking with the audience and listening to the audience and showing them different versions of it so [00:18:50] that they can give you feedback. Co creation is really, really important for any really good high quality data [00:19:00] visualization. Of course, you also have co creation for the exploratory things.
[00:19:06] Alexander: Well, you do it with yourself, so you co create with [00:19:10] yourself or your colleagues. But there, kind of, the feedback loop is a very, very different one. So, in summary, [00:19:20] there are many different aspects that There’s a huge difference between explanatory and exploratory design, [00:19:30] standardized, technical, customized, the different audiences, the interaction, the different tasks, the time you interact [00:19:40] with the data visualization or you consume the data visualization, how you improve it over time, how you create it.
[00:19:48] Alexander: The external [00:19:50] impact is vastly different. Understand where you can use one or the other. [00:20:00] You will need both and you need to understand where you can use both. In the next episode, I will go through a couple of different [00:20:10] use cases that will help you to better understand Where you can all use exploratory data visualization, and I’m pretty [00:20:20] sure there will be some in it that you haven’t not thought about, and where you can create more value for your organization.
[00:20:29] Alexander: By [00:20:30] the way, we have a new video coming soon. And then we have a data visualization course. Especially about data visualization. So if you’re interested in this topic, then check out the show [00:20:40] notes and have a look into our data visualization course.
Join The Effective Statistician LinkedIn group
This group was set up to help each other to become more effective statisticians. We’ll run challenges in this group, e.g. around writing abstracts for conferences or other projects. I’ll also post into this group further content.
I want to help the community of statisticians, data scientists, programmers and other quantitative scientists to be more influential, innovative, and effective. I believe that as a community we can help our research, our regulatory and payer systems, and ultimately physicians and patients take better decisions based on better evidence.
I work to achieve a future in which everyone can access the right evidence in the right format at the right time to make sound decisions.
When my kids are sick, I want to have good evidence to discuss with the physician about the different therapy choices.
When my mother is sick, I want her to understand the evidence and being able to understand it.
When I get sick, I want to find evidence that I can trust and that helps me to have meaningful discussions with my healthcare professionals.
I want to live in a world, where the media reports correctly about medical evidence and in which society distinguishes between fake evidence and real evidence.
Let’s work together to achieve this.