Anyone can create a data visualization but not everyone can create a data visualization that is effectively communicating the data – and this is what we should aim for. Done wrong they leave the audience confused.
In this episode, I’ve gathered up six problems around creating data visualisations, so you can improve the quality of your visualisations.
Stay tuned while I talk about these 6 problems:
- Just relying on a simple table without data visualisation
- Not taking context (message, audience, circumstances, channel) into account
- Lack of testing with the target audience
- Having an unclear message to start the design
- Incorporating every comment
- Relying heavily on templates
Listen to this episode now and share the link with your friends!
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
6 Problems around creating data visualisations
Alexander: You’re listening to the Effective Statistician Podcast, the weekly podcast with Alexander Schact and Benjamin Piske designed to help you reach your potential, lead great science and serve patients without becoming overwhelmed by work. Today I’m talking about six common problems with data visualizations and of course how you can overcome them.
So now some music. As you know, I’m a really, really big fan of data visualizations because they help people to understand data much faster. And I’m speaking here more about the figures that actually tell a story, not so much about data visualization. That helps to explore data where you want to find patterns into data.
So listen to this episode. These are the things that really I see very often across the [00:01:00] board. By the way, there is also really good PSI, special interest group about data visualization. Check that out on the PSI homepage at psiweb.org so you can learn all about the different things around PSI, what PSI does.
You doesn’t need to become a PSI member to join this special interest group. But of course, there’s a lot of great things that are happening around PSI anyway. By the way, PSI is also helping me with this podcast. So head over to psiweb.org to learn more about PSI activities and maybe become a PSI member today.
So what are the big problems that I see all the time in terms of data visualization? The first is, there’s no data visualization at all. People [00:02:00] just rely on a table, and the most common reason I hear for this is I don’t have time for this. Now, what I hear when someone says I don’t have time is actually this is not a priority for me.
Now, priority depends very much on reward, expectations, and things like that. If your supervisors, if your more senior management doesn’t take that as something important, then that’s not rewarded. Very often, people always also will think, well as a decision, it’s not my job. That’s a job by done, by medical writing or someone else.
But really they can do the [00:03:00] usual things like maybe a bar chart or maybe even a forest plot, but all the things that rely on patient level data, it will be really hard for any medical writer. And also, do you really wanna leave it to the medical writers to tell the story around your data? Control
that process, work proactively? It’s also very often that it’s no time because the processes are very hard. Yeah so if you try to follow the same processes that you have for tables for data visualizations, of course it’s hard. A table you can rather easily pre-specify and make very crisp.
Okay. This is a standard table and we put the numbers in there. That doesn’t work for. Very good high quality data visualizations. For [00:04:00] good data visualizations, you will always need to adapt a couple of different things. Of course, you can have, let’s say your standard ones laid out those that get somewhere in an appendix and things like that.
But for some things that you wanna present to Wall Street to upper management at a conference or in a publication, you need to have something that looks really good and that is always based on lots of iterations and improvements. So don’t use the same processes that you have for
tables for data visualizations, you need to have a much higher frequency, much higher speed of getting feedback. It needs to be co-created with the audience in mind. The next problem that I very often see as that people don’t take the context into account. So this is problem number [00:05:00] two, whenever I run my data visualization workshops.
And if you go to this, then you in the future then you will probably not fall into the trap. But about 95% of the statisticians fall into the strips that I have at the beginning of my workshops. So I give data table to the participants and then ask them to visualize that. And if they need any further information, they can just ask.
Rarely someone asks about things like, who is the audience? How will it be presented? What is a key story here? What is the most important part of the data that we actually want to talk about? And these are really important things. Figure that goes into a dossier, [00:06:00] a report a paper where people can look at it.
If they want for hours, that can be much more complex, detailed sophisticated. Figure that is shown on a slide for 30 seconds at a conference needs to be simple, super simple. You can’t have kind of three lines of form of footnotes on that. It needs to be crisp and clean. Also imagine you present that in some kind of interactive way on a smartphone, on a desktop or at a conference on a touch screen.
You need to take that into account. You can’t just take the same thing and copy it over. That doesn’t work. But I see it all the time. Something is produced for a clinical study report and send copies and paste it over into present. And [00:07:00] that only leads to someone else picking it up and correcting it manually.
Look into the material set is used for promotions. Is that exactly what is in the figures? In the, or in the tables, in the manuscripts, in the CSR, mostly not. Because, they are experts, the marketing people, the sales people, they want to have that really crisp and clean. They produce usually very good quality graphics.
You can learn a lot from them, but of course, if you give it directly to them, then you can also ensure. It’s done right from a statistic point of view, and you get much bigger quality and you can also provide with patient level [00:08:00] figure sets, they actually can’t work on. Question or problem number three is lack of testing with the target audience.
I’ve once had a interview with Alberto Cairo, and if you scroll back in, in the podcast he’s a famous data visualization guru and the podcast with him is one of the most downloaded episodes, actually. Even, he sometimes creates data visualizations that he then needs to. Start from scratch again after he has shown them to the audience.
But of course, you only get that inside that it doesn’t work well with the audience if you test it. And then you will see all the different things that the audience expects. Let’s say maybe in this specific [00:09:00] audience, say, have a specific association about what certain colors mean. Yeah. Because they’re associated with certain disease symptoms or with maybe certain brands that they are aware about that you may not think about, maybe they’re familiar with.
Certain types of graphs or they have familiarity with certain types of concepts. You may even learn whether kind of your line should go up or down. All these kind of different things. You can also test. Whether you can easily explain the graphics or whether they are self exploratory, explanatory.
So these kind of things are really important. You’ll get that when you test with the audience. Next problem that, I very often see is number four. You start with an unclear [00:10:00] message.
This unclear message leads to a lot of confusion. Let’s say you get asked to present the efficacy data for endpoint X for study Y in a figure. Now what about this endpoint is so important? What do you really wanna show? Do you wanna show how it, how fast it develops at the beginning? How big the difference to placebo or the active comparator are?
Is placebo actually relevant or is only the active comparator relevant? Are there certain time points that you wanna bring home? Are there certain parts of the endpoints that you wanna emphasize, do you wanna look into if it’s a combined endpoint into the components as well? If it’s
patients reported outcome into the different items of it. Do you wanna show[00:11:00] patient level data so that you can see how consistently the effective, all these kind of different things Yeah. Can be included in very generic ask for providing a figure. And if you are not clear about this, then you will get conflicting feedback when you test it.
Some will say, well get rid of this. Others will say at that the next one will say, no, we need a bar chart instead of a line chart. And another person will say, but what about the SAP groups? If you are not clear on what exactly you wanna show, what exactly the title is. Yeah, having a telling title is always a great start.
And also helps you to get clear within the team on what you really want to show. Then, you only then. You [00:12:00] can deliver a table set or a figure. A figure that really crisply shows this message.
Next problem number five, trying to incorporate every comment. You will not get to a really good outcome if you try to please everybody. It is impossible, and also you don’t need to please everybody you need to please those that are representative of the audience. And honestly, mostly these are non statisticians.
So if the other fellow statistician say, oh, you need to add detail X, Y, and that maybe that is not the most helpful comment because more detail is usually not [00:13:00] what’s the audience needs. Think about what are the most important Comments, what are the most important remarks? What are the most important feedback that helps you to get things across?
Don’t take into account everything that’s impossible. Try to come up with some things that best helps you to convince a message and that best resonates with your target audience. Problem number six, using templates too much. Yes, I know everybody wants to standardize things. Everybody wants to make things faster.
Everybody wants to make things easier and so on. But great explanatory data visualization. Don’t follow templates. Yes, if you want to just put them into your CSR and the appendix, you can have templates and. [00:14:00] groups should use a certain figure 200 times, but the ones that you put in your manuscript, the ones that are very front and center and post us and presentations, you can’t just go with the template.
It chiro, I think, once said, if you can see with which software it was created, Then is probably not done well because then you just used some kind of template from a software and didn’t think consciously about what exactly you want to show how exactly the figure should work. In my workshops, I start with pen and paper, and one of the reasons why I do this, Then people think very intentionally about all these [00:15:00] little things that they want to do and say, don’t fall into the trap of using a template that might not be helpful for them.
Maybe you wanted a specific form in there. Maybe you want to have specific colors in there. Maybe you want to have. No grid lines or you wanna have grid lines or whatsoever, all these different things you will only think about if you are consciously deciding all these different things. Yes, templates can make it faster, but usually don’t lead to better results.
So be really careful with using standard templates. Where looking into other graphs is really helpful, is getting inspiration. How could you do it differently? And there’s a lot of free contents [00:16:00] that I have about data visualization on the effective statistic homepage. So check that out. Look into our library, what we have there.
Also look into the visualization special interest group on their homepage. They have also a lot. One last thing. Gary and myself have also created a course that is called Winning With Words and Crafts, the Effective Ation Presentation and Data Visualization Master Clause. That also speaks about how you can improve your data visualizations.
So that’s another thing you can check out. And from time to time, I’m organizing these workshops that I referred to in the episode today a couple of times. If you are on my email list, then you will get announcements regarding this as well. Also, follow me on LinkedIn and best you process this little bell I ings, [00:17:00] is that you get all my announcements as well.
So check that out. Sign in for the newsletter on the effective statistic homepage so that you don’t miss these kind of opportu opportunities. Thanks so much. That was a, hopefully a really good episode for you of not making these six mistakes, which were no visualization at all. The second thing, not taking the context in terms of who are you presenting to and which environment are you presenting, and what is the key message that you want.
Third problem was lack of testing with the target audience. Fourth problem was having an unclear message to start the design of fifth, trying to incorporate every comment and last overusing templates. So vi. [00:18:00] You’ll be far better than most of the other state decisions in terms of data visualizations.
The show was created in association with psi. Thanks to re enter team at vvs who help us to show in the background. And thank you for listening. Reach your potential lead grade science and serve patients. Just be an effective statistician.
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.