Episode #254: Maarten Boers

Summary

In this week’s episode, we delve into the pivotal role of visual clarity in scientific research. Join me and Professor of Clinical Epidemiology Maarten Boers as we discuss his new book, Data Visualization for Biomedical Scientists. If you are at all interested in being a better science communicator—and especially if you are interested in publishing your work in academic journals—this episode is for you! We talk about how Maarten’s book extends beyond the world of biomedical science into good table design, small multiples, and how academic publishing needs to get its act in order.

Topics Discussed

• The Necessity of Clear Experimental Procedures: We highlight the significance of understanding every step within an experiment. Our discussion unpacks the ways in which clear, precise procedures facilitate reproducibility and validation of scientific work.
• Deciphering Scientific Terminology: Maarten’s book emphasizes the importance of demystifying complex scientific jargon. We examine strategies for breaking down terminology barriers for both specialist and general audiences.
• Graphical Excellence in Research Communication: We focus on the power of well-titled, labeled, and annotated graphs in conveying research and analysis.
• Impactful Captions and Visual Storytelling: Captions are more than mere descriptions—they’re a gateway to engaging the reader. We explore how to craft active captions that not only inform but also captivate and retain the reader’s attention.
• Challenges in Academic Publishing: We confront the practical challenges researchers often encounter with journals, their design (or lack thereof), and other publishing pitfalls. We talk about how to effectively intervene when production staff mishandle figures and how to work within the constraints of journal page limits.
• Ensuring Accuracy in the Publication Process: Our conversation also touches on the responsibilities of researchers to ensure their findings are presented accurately and effectively, even in the final stages of publication.

Resources

Guest Bio

University Medical Centers (1997–2023), and rheumatologist (recently retired as a staff rheumatologist in the Reade Institute). He is director of Epiconsult BV, that hosts his consulting activities. He holds a PhD in Medicine from the University of Leiden, Netherlands and an MSc in Clinical Epidemiology from McMaster University, Hamilton, ON, Canada. Dr. Boers’ main research interests have focused on treatment of rheumatoid arthritis and on outcome measures in clinical research including the early intervention and combination therapy in rheumatoid arthritis (principal investigator of the COBRA trial and follow up studies) and the rehabilitation of glucocorticoids as essential treatment (principal investigator of the GLORIA trial published in 2022). He co-founded the OMERACT (Outcome Measures in Rheumatology) consensus initiative in 1992 (www.omeract.org). He has (co)authored more than 500 full papers on topics both in and outside of rheumatology.

Related Episodes

Sponsor: Tube Buddy

TubeBuddy is the main tool that I use to manage the backend of my YouTube channel. It has been a huge lifesaver when it comes to time management and properly optimizing my channel. They have tons of free tools you can check out right here!

Different Ways to Listen to the Show!

New Ways to Support the Show!

With more than 250 guests and 10 seasons of episodes, the PolicyViz Podcast is one of the longest-running data visualization podcasts around. You can support the show by downloading and listening, following the work of my guests, and sharing the show with your networks. If you’re interested in financially supporting the show, you can sign up for my Patreon platform, make a one-time payment via PayPal, or shop one of the show’s sponsors.

Transcript

00:12 – 00:17
Welcome back to the PolicyViz Podcast. I’m your host, Jon Schwabish. Thanks for tuning in

00:18 – 00:23
to the show. On this week’s episode of the podcast, I welcome Maarten Boers, scientist who’s

00:23 – 00:29
written a great new book on visualizing data, something I really am interested in, obviously,

00:29 – 00:34
and I’m sure you are too having listening to this podcast. Now Why should you read Maarten’s

00:34 – 00:38
book and listen to this podcast in addition to all the other great books that you’re reading?

00:38 – 00:44
I think there are 2 big things to consider with this book. 1st, He spends a lot of time talking

00:44 – 00:48
about table design, and we’re gonna talk about it in the interview. It’s something that doesn’t

00:48 – 00:53
really get a lot of attention in the data visualization library. So from my perspective, it’s

00:53 – 00:59
really Maarten’s new book, my book, and Steven Fuse’s book are really the few books that talk

00:59 – 01:02
about table design. So that’s one thing. If you’re interested in better tables, this is the

01:02 – 01:07
episode for you. The other thing that we talk about a lot is small multiples boers what Maarten

01:07 – 01:12
calls matrix graphs. So we talk about good styles and designs and ideas and concepts about creating

01:13 – 01:18
better small multiple charts, Which if you don’t know are exactly what they sound like, a combination

01:18 – 01:24
of smaller multiple charts. And the last thing that we talked about that is really interesting

01:24 – 01:30
for me, From my perspective as someone who’s, worked in publishing in the academic field and

01:30 – 01:35
in academic journals is how the academic publishing doesn’t focus on data visualization and

01:35 – 01:41
how that’s been a detriment to the ability to communicate, research, and scientific findings

01:41 – 01:46
more effectively. So there’s a lot of information and guidance packed into this interview and

01:46 – 01:52
into Maarten’s book. So I hope you’ll enjoy this week’s episode of the podcast, my interview

01:52 – 01:59
with Maarten Boers, starts right now. Hi, Maarten. Good. Well, afternoon, your time. Morning,

01:59 – 02:01
my time. How are you? Happy New Year. Good to see you.

02:02 – 02:07
Hi. Good to see you too. Thank you for inviting me to do this, interview. I’m really excited.

02:07 – 02:12
This is, terrific. I’m excited to chat with you. I’ve got your book, Data visualization for

02:12 – 02:18
biomedical scientists. We’ll talk about how it’s not just for biomedical scientists. And there

02:18 – 02:24
are some unique pieces in your book that you don’t find in many other data visualization boers.

02:24 – 02:29
So so I’m excited to chat with you. So, Maybe we can start with, your background, and then we

02:29 – 02:34
just jump right in to talk about the book. The first question I have about the book is, What

02:34 – 02:40
are the unique challenges with visualizing data that that biomedical scientists face? And then

02:40 – 02:44
I think we can because the book is definitely broader than than just that, but I think We can

02:44 – 02:48
start there. So, so kind of 2 questions for you. Just just kick the whole thing off.

02:48 – 02:56
Yeah. Well, I I mean, To be really honest, I don’t think there are unique challenges to biomedical

02:57 – 03:06
science. The reason I decided to focus on what is my own field is that there’s lots of general

03:06 – 03:14
data visualization books around, and Some of them also address scientific papers and and and

03:14 – 03:24
presentations. And in the process of writing my book, I had some feedback from some, very notable

03:24 – 03:31
people in the field, And and one of them said, you know, why write another book on data visualization?

03:31 – 03:36
I mean, look at my book. That’s that’s all people need.

03:36 – 03:38
Right now. Yeah. Right. Yeah. Yeah.

03:38 – 03:43
And I was I was I was writing this book because Even though that book was great, and there are

03:43 – 03:49
many great books out there, and I, you know, I reference them all in my book. It’s not like

03:50 – 03:56
I I suddenly invented this field, but, I was struggling with what I could not find in those

03:56 – 04:02
books that were day to day issues in the papers I was trying to write or mostly I was trying

04:02 – 04:10
to supervise in in my in my my fellows and finding out that that There was no guidance at all,

04:10 – 04:16
and everybody did it wrong, including senior scientists. Nothing was being Corrected from the

04:16 – 04:22
journals. If you look at the journal templates, you know, 99% are simply horrible and could

04:22 – 04:28
be easily improved. So I started, you know, Writing about that and writing to journal editors

04:28 – 04:34
and and when I was on editorial board, spending time to to improve papers and writing guidelines,

04:34 – 04:40
and, You know, that sort of expanded into something, then I decided, you know, COVID came along.

04:41 – 04:47
Mhmm. Let’s let’s write a book. So that’s Basically, how it happened. But I think and and and

04:47 – 04:52
biomedical science is my field, so I know what’s wrong Jon my field. And if I were to say, okay,

04:53 – 04:59
and by the way, you know, economics or another field, they have poor graphs as well. Mhmm. Then

04:59 – 05:04
I didn’t really have to go and study those journals to make sure that I wasn’t making a fool

05:04 – 05:08
of myself, and people would fall all over me. Oh, you haven’t seen me done this and that. I

05:08 – 05:15
know my field. Right. So I thought, you know, if I stick to that, nobody can blame me for going

05:15 – 05:18
outside of my, comfort zone. So so

05:18 – 05:22
but let’s talk about that a little bit because, like, I will tell you in the economics field,

05:22 – 05:27
like, the graphs are also fairly terrible in most journals. And so you you spend there’s, like,

05:27 – 05:31
this this section right at the end that’s kinda tucked away, but it was kinda like my favorite

05:31 – 05:38
part because it it talks about this peer review process that nobody talks about. And and I guess,

05:38 – 05:46
I’m not sure what my question is other than why do you think Peer reviewed journals haven’t,

05:47 – 05:55
tapped into what we know is data visualization is a popular means to communicate research and data.

05:56 – 06:07
Because nobody feels enough of an expert to comment on it. And, so it’s it’s from from the junior

06:08 – 06:13
to all the way to the senior level, that nobody knows how to do it.

06:14 – 06:14
Yeah.

06:14 – 06:22
And so so, You know, I I was interested in in data visualization right from the very beginning

06:22 – 06:29
of my scientific career before, So I always spent a lot of time, you know, doing stuff, but

06:29 – 06:32
nobody of my supervisors would ever comment on that part.

06:32 – 06:33
Right.

06:33 – 06:40
And if you made a table, Well, you just made a table, and nobody said, gee, why don’t you you

06:40 – 06:44
know, why is the order of your categories like this? Or why have you ordered your numbers in

06:44 – 06:50
this way? And why is this table so terrible to read? It was sort of a a natural phenomenon that

06:50 – 06:53
tables and graphs would magically appear.

06:53 – 06:54
Yeah.

06:54 – 06:59
And then right from the first version, they would go all the way into the published paper. And,

06:59 – 07:06
unfortunately, That is either still the case, or journals have caught up and put in some guidelines

07:06 – 07:08
that make tables and graphs worse.

07:09 – 07:16
Right. They built these templates that are not built by I mean, I’m sure well intentioned, but

07:16 – 07:19
not people who may have expertise and how to do these well.

07:19 – 07:25
Or they and and, I mean, the most recent version of this is that the the bigger journals have

07:25 – 07:32
put little software machines in place Mhmm. Where, once you are past the acceptance stage, you

07:32 – 07:37
go into the proof. You get your proof, But the proof is no longer a PDF, but it’s actually a

07:37 – 07:43
web page where you can, you know, correct your paper. And the web page has a little table machine

07:43 – 07:48
in there It does not allow you to format the table properly.

07:48 – 07:48
Right.

07:49 – 07:51
Or it makes it so difficult that everybody gives up.

07:52 – 07:57
Right. There’s a revenue stream problem in academic publishing, right? Like my first, the first

07:57 – 08:02
paper I wrote, I remember the first paper that I will coauthored, back in the day, the journal

08:03 – 08:08
helped make all the the graphics. You know, they they had all the editors. I mean, there was

08:08 – 08:11
a team, and now You’re on your own. It’s like, you

08:11 – 08:14
know Exactly. Yeah. Exactly.

08:14 – 08:20
And do you do you think I mean, this is looking ahead, but do you think that Maybe some of these

08:20 – 08:25
artificial intelligence tools and as the, you know, data vis tools maybe start developing it

08:25 – 08:29
better, that will be helpful, or do you think we’re just gonna be spinning our wheels and

08:30 – 08:36
Well, yeah, well, if you say AI, AI needs to be trained. And if AI is gonna be trained with

08:36 – 08:43
the current body of literature, they will Engrained is the the horrible practice that we have now. Right.

08:43 – 08:44
Yeah. That’s right. That’s right.

08:44 – 08:51
I hadn’t even thought of that, but you’ve given me a new nightmare to contemplate. That’s that’s

08:51 – 09:01
not gonna really help. I mean Yeah. The software has improved Immensely. So there’s much better

09:01 – 09:07
tools that you can use today than there were when I started out. I mean, when I started out,

09:07 – 09:10
we didn’t have computers. I mean, I’m that old.

09:10 – 09:11
Punch cards.

09:12 – 09:15
And, actually, I started before we had punch cards.

09:15 – 09:16
Before punch cards. Okay.

09:16 – 09:26
So so so my first, scientific article was drawn by a scientific artist Oh, wow. Who who had

09:26 – 09:28
the, you know, the like architects have?

09:28 – 09:29
Yeah. Yeah. The big drafting.

09:29 – 09:35
And then they they would have paper, and then they would have sets of of dotted and dashed lines

09:35 – 09:41
on on paper, and they they would put it on the line, and then they would scratch the ink onto

09:41 – 09:48
the well, that was that was that was pure art. Yeah. Yeah. And I I remember dreaming of a computer

09:48 – 09:54
program when when When the first Mac started when when the the the Mac first Mac came on the

09:54 – 09:57
market, I dreamt of a program that would do things.

09:58 – 09:59
We’ll do it. Yeah.

09:59 – 10:04
And when that program appeared, it isn’t around anymore, so I can safely name it without sponsoring

10:04 – 10:12
anybody, Cricket Graph. Oh, wow. When that program came along, I bought my first Mac because

10:12 – 10:17
I had dreamed of that program. You know, you click on an axis, And it opens, and you can change

10:17 – 10:18
the scale. That sort

10:18 – 10:19
of thing.

10:19 – 10:24
Well, that’s, you know, routine now. It wasn’t it wasn’t around. So Right. I I’ve seen it all

10:24 – 10:31
come, and, I mean, now there’s such good programs, and and some of the programs are invented

10:31 – 10:38
by people who know about graphs. So the default is already half or three quarters good, which

10:38 – 10:45
means the errors, If you just keep on the default, you already have a pretty good Right. Starting

10:45 – 10:48
point. And many people don’t get past the default, as you know.

10:48 – 10:49
Yeah. Right.

10:49 – 10:56
When Microsoft changed their standard font, everybody uses that font because nobody ever changes

10:56 – 11:01
a font. Right. And it’s with graphs and with tables. It’s it’s the same way. You have one template.

11:01 – 11:03
Oh, that’s the way to do it. Okay. Fine.

11:03 – 11:10
Right. Right. So so you’ve mentioned a couple times tables. In the first chapter after the introduction,

11:10 – 11:17
the first chapter in your book is on table design. And By my count, there are only 3 data visualization

11:17 – 11:22
books that spend any significant time talking about tables. Your book, my book, and Steven Fuse

11:22 – 11:24
book. And those are the only 3 that I’m aware of

11:24 – 11:25
that’s happening in

11:25 – 11:34
the same time. And I’m curious, I guess my curiosity really stemmed, why you decided to do tables

11:34 – 11:35
at the very beginning of the book?

11:37 – 11:39
Well, I was gonna do a book on graphs.

11:39 – 11:40
Uh-huh. Right.

11:40 – 11:48
And then I I thought yeah. But I need to do tables first. Yeah. Because tables in a way, are

11:48 – 11:56
simpler. Mhmm. Although, making a good table is pretty difficult, but tables are simpler because

11:56 – 12:00
you don’t have all these, you know, directions you can go in.

12:00 – 12:01
Yes.

12:01 – 12:04
And they are sort of the basic staple of of science.

12:05 – 12:05
Yeah. Is

12:05 – 12:12
it that’s not graphs. It’s tables. Yeah. And I think nobody writes about them. So I was highly

12:12 – 12:18
Fired when I I didn’t have your book then. I just had Steven Few’s book, and I thought, he is

12:18 – 12:23
a guy who writes about table, and it’s it’s It’s really I mean, his table chapter is really very good.

12:23 – 12:24
Yeah. So,

12:26 – 12:32
like with all things, you know, do I need to add to that? Yes. Because Steven does not address

12:32 – 12:38
the issues I have in my field with tables and and extensively described in the chapter where

12:38 – 12:43
you have, And I’m I’m moving away from that, but that’s what you see very commonly. You have

12:43 – 12:50
multiple numbers within one cells, mhmm, within one cell that are, you know, unclearly separated.

12:50 – 12:56
So you have these number blurbs that almost become words inside a cell, And then on the next

12:56 – 13:01
line, you have the same blur but slightly differently. And, you know, within 4 lines, your table

13:01 – 13:02
is a total mess. So

13:02 – 13:02
Yeah.

13:02 – 13:10
Yeah. There were there were quite a A number of things that Steven Few didn’t have to focus

13:10 – 13:17
on because his tables We’re basically about business and sales and and housing and whatever.

13:17 – 13:22
So it’s Right. Sort of one category, which is either dollars or bricks So whatever.

13:23 – 13:23
Yes.

13:23 – 13:29
And we have all these multi item tables with different things. So Yeah. There’s there’s highly

13:29 – 13:34
specific things that I was struggling with Before I wrote the chapter and I had sort of found

13:34 – 13:39
some solutions boers, which were universally rejected by journals if I submitted, they say,

13:39 – 13:42
yeah. But this is not standard. Yeah. But just down at the zoo.

13:44 – 13:49
I mean, it’s so interesting. Like, they have their own publishing platform, whatever tools they

13:49 – 13:55
use. Yeah. And it’s like, that’s that’s it. And if it doesn’t fit in that little box, then

13:56 – 14:02
Well, it’s it’s worse than that. I think most of the journals have one template, which, you

14:02 – 14:11
know, looks like a format. Mhmm. But what they do is they just Pour your numbers in the cells

14:12 – 14:17
and then forget about it. There isn’t Right. You know, there’s some good journals that make

14:17 – 14:24
very nice tables, but, Usually, it’s just, oh, you’re submitting a table. Okay. Here’s our matrix.

14:24 – 14:28
Just pour it in. There you go. This is our format. This is how we always do it.

14:28 – 14:29
Right.

14:29 – 14:32
So there’s no there’s no thinking behind it. It’s just production.

14:33 – 14:39
What would your top recommendations be for for people when it comes to making good tables.

14:40 – 14:41
I can name only 1.

14:42 – 14:42
Okay.

14:42 – 14:43
Only only 1?

14:43 – 14:44
Love it.

14:44 – 14:45
Because I have a lot of them.

14:45 – 14:47
Yeah. Yeah. Yeah. Yeah. 1 would be good. Yeah.

14:48 – 14:52
The the I think the first recommendation would be proper alignment.

14:53 – 14:54
Yeah.

14:54 – 14:59
It’s it’s about categories and numbers, and if your numbers are not properly aligned, they’re

14:59 – 15:00
difficult to read.

15:00 – 15:06
Yeah. It is fascinating to me when I see this. I’m sure I mean, you know, science is science.

15:06 – 15:10
So they have like you said, you have multiple things in a cell. You’ve got the, You know, a

15:10 – 15:14
coefficient and a standard error with stars and and parentheses around it. And it is shocking

15:14 – 15:20
to me how the numbers are not aligned, and it is just Just objectively difficult to read. I

15:20 – 15:26
don’t think we need any sort of, like, like, real fundamental study to to prove that point.

15:26 – 15:28
It is just harder to read.

15:29 – 15:36
Exactly. So so and and it’s so basic, and it is it is fairly easy to correct. I mean, there’s

15:36 – 15:41
some issues, you know, shall we do it this way, that way? But it’s not like, oh my god. How

15:41 – 15:47
am I going to solve this problem? It’s just saying, this is how you do it and implementing it.

15:47 – 15:54
I think the the other thing about tables is to communicate to people that, you know, tables

15:54 – 16:02
are actually Boers of like very simple graphs. So you have to think about the ordering of the

16:02 – 16:11
information in your table, Which which which I I term after one of the other giants, telling

16:11 – 16:19
a story. So it’s not like you make a table and you order your categories by alphabet, or by

16:19 – 16:30
the order in your database, mhmm, Or by, you know, random? Yes. You know?

16:30 – 16:33
Whatever your code says the variable name is? Yeah.

16:33 – 16:38
Usually, things you find important are gonna be at the top of the table. Yeah. So don’t let

16:38 – 16:46
me scroll through 18 rows of noise before, you know, we hit on the thing you wanted to show

16:46 – 16:50
me in the first place. Yeah. So those those are it’s more than one. That that would be the 2

16:50 – 16:56
things, and then there’s all the other stuff. Because as you rightly noted in the prep for this

16:56 – 17:01
interview, I think the the tables chapter is actually the longest of the whole book.

17:01 – 17:07
Well, it’s it is interesting because Steven’s Chapter on Jon tables. I don’t remember, but exactly

17:07 – 17:11
where, but it’s somewhere in the middle of the book. The chapter I wrote on tables is at the

17:11 – 17:16
end of my book. And I just I did I found it interesting that it was at the beginning of this

17:16 – 17:21
book, but when I think about at least in economics, and I’m guessing it’s similar in your field.

17:21 – 17:28
If I think about picking up any journal and just going page by page and counting graphs versus

17:28 – 17:35
tables, I’m sure there are more tables, than than graphs in any random journal that I select.

17:35 – 17:41
Yeah. And and and in every paper, there’s a table 1 become it always comes before any graph,

17:41 – 17:41
and

17:41 – 17:47
it hits gives you the baseline data, so it’s sort of really, really basic. Yeah.

17:47 – 17:55
Yeah. The other interesting, piece of the book is you have an entire chapter Dedicated to, you

17:55 – 18:00
know, what you call matrix graphs. Other people call it small multiples, trellises, panel chart.

18:00 – 18:06
I mean, whatever. Right? But So, again, I’m I’m curious about the decision to include to to

18:06 – 18:11
write an entire chapter about it. I think it’s, I I don’t I’m not saying I disagree with it.

18:11 – 18:15
I actually agree with it, that it it deserves its oh, it’s one of the one of my pillars of good

18:15 – 18:21
data visualization is Think small multiples and and see where’d you go. So, again, what was

18:21 – 18:25
your thinking behind a whole chapter dedicated to small multiples? And then we could talk about

18:25 – 18:29
what your, you know, Top recommendation would be when people are are creating it.

18:29 – 18:36
Okay. So so that chapter is actually has 2 parts. The Sort of the basic part, the basic principles

18:36 – 18:45
of small multiples, which covers topics like, you know, trying to harmonize your axis So that

18:45 – 18:51
you can do away with a lot of labeling and and trying to, again, get the order right for the

18:51 – 18:57
message you have in your data, Mhmm. Whether it be you know, we like to compare horizontally

18:57 – 19:01
rather than vertically. So if your main comparison is horizontal, they should be like this,

19:01 – 19:08
all that sort of stuff, which goes for, all of those, what I call matrix drafts. But the real

19:08 – 19:22
story is The second maarten. And the second part is my effort to make A good matrix graph for

19:22 – 19:29
basic scientists. So in in my field, basic scientists are the The people in the labs, or they

19:29 – 19:33
call themselves usually translational because they want people to feel that they are connected

19:33 – 19:40
to the bedside. You know, they do stuff with cells and with with with DNA and with, experimental

19:40 – 19:41
animals and what have you.

19:41 – 19:42
Right.

19:43 – 19:52
And within my field, that is a really close shop Mhmm. Of people who have their own methods

19:52 – 20:01
of doing research. And if you’re not In the in crowd, you usually don’t understand anything

20:01 – 20:06
of what’s going Jon, and that’s because they have their own codes. They have their own abbreviations.

20:06 – 20:11
They have their own way of doing things. They have their own way of doing statistical analysis.

20:11 – 20:17
And I’m, you know, sort of a statistician as well, Doing a lot of things that in the rest of

20:17 – 20:24
the world is found to be quite objectionable and not really very right, like doing, experiments

20:24 – 20:31
with very, very small sample sizes and then doing, parametric statistics To see differences,

20:31 – 20:38
significant differences, and then doing multiple tests, and I can go on and on. So There’s a

20:38 – 20:51
lot of things in in methodology, in, my type of methodology, Which is not the lab, mechanistic

20:52 – 20:59
methodology, but really statistics, Yeah. How to set up your experiment, how to do controls,

20:59 – 21:06
and all that sort of thing, which is, Well, difficult to understand, and and usually not the

21:06 – 21:13
way I would recommend, but the other side of it is they do a lot of, So they have this theory

21:13 – 21:24
of, you know, whatever. Agent a, blocks The production of, some sort of stuff that you need

21:24 – 21:31
or not need from, from the raw compound, And we’re gonna prove that that is the fact and also

21:31 – 21:36
that it has a biologic effect that is relevant. Mhmm. And so the so they have this system where

21:36 – 21:42
they say, okay. Let’s first look in the genes, whether this is happening. And then let’s look

21:42 – 21:48
at the gene expression, and then let’s look at whether the cells who have that gene extraction

21:48 – 21:53
are actually in the place where they should be doing the work or not. And then if we have that,

21:53 – 21:59
then let let’s look at, You know, in whole lab animals, and let’s see whether the agent that

21:59 – 22:06
is needed for that process is actually being blocked or not. So they have A lot of ways to buttress

22:06 – 22:11
their theory, which is really very cool. You know, in in human experiments, we have this drug,

22:11 – 22:16
And we put it in a human and see, you know, what happens to the human, and we have to infer

22:16 – 22:23
all those steps in between. So Their if you want statistical methodology, I think is very poor,

22:23 – 22:32
but their biological methodology It’s very, very advanced, very precise, very elegant. They

22:32 – 22:37
always have, you know, 3 or 4 ways of proving their point. Yeah. We maarten here and there and

22:37 – 22:43
there and there, and that’s why this is true. So really good stuff. Beautiful science. And Yeah.

22:43 – 22:49
I always thought when I saw these Horrible mattresses that they produce because they have, like,

22:49 – 22:55
in an 8 minute presentation, they will have Twenty slides, and all those slides are small multiples

22:55 – 23:01
of at least 16 graphs with unreadable letters with no explanation. And they’ll say, So here

23:01 – 23:07
you see that a was logged by b. And by the way, oops, next slide. B was logged by c. Right.

23:07 – 23:14
You know, another 20 graphs. So it was Oh my god. Oh my god. I I can’t So I thought, you know,

23:14 – 23:20
it may just be that I’m too stupid to understand this stuff, And all the rest of these basic

23:20 – 23:24
science people, they understand, and they’re happy with these slides. It’s just me being stupid.

23:24 – 23:24
Yeah. So

23:24 – 23:29
I just said, okay. I don’t understand this graph, and I had a colleague, you know, I work in

23:29 – 23:34
a in a university bath department with a lot of basic scientists, And there was a very nice

23:34 – 23:41
presentation from one of his PhD students. And I saw from the slides okay. Here’s Someone who’s,

23:41 – 23:46
you know, sort of interested in in in trying to convey what’s going on to me who is not in the

23:46 – 23:51
lab. So that was the starting point. And I said, look. I saw this presentation. It’s nice. I

23:51 – 23:57
understood about 25% of it. Mhmm. I want to understand 100%, and I wanna see your graphs, And

23:57 – 24:04
I wanna see whether I can improve it Mhmm. After understanding so that it’s better.

24:04 – 24:04
Yeah. Is

24:04 – 24:09
it okay? So, well, it was a bit of, you know, to and fro before I had the data and I had the

24:09 – 24:15
graphs and I had so I started building this. I think, Yeah. I wasn’t working full time, and

24:15 – 24:16
I think it cost me 3 months

24:17 – 24:17
Oh, wow.

24:18 – 24:23
To understand the science Oh, yeah. Really on the detail level. And, again, you know, the science

24:23 – 24:29
is not really very difficult. It’s just understanding what’s going on in the nucleus with that,

24:30 – 24:37
cytokine and that stuff What is being activated and repressed? Sure. But you have to know what

24:37 – 24:43
kind of abbreviations they’re using. Are they useful? And then you find out they always do the

24:43 – 24:53
same experiment. It’s Jon generic experiment, and they do that In different settings over and

24:53 – 24:53
over and over again.

24:53 – 24:54
Over and over again. Right?

24:54 – 25:03
It’s like this. Okay. I have a negative control. Mhmm. Yeah. It’s normal saline or or or unstimulated

25:03 – 25:11
cells boers, Wild type baby baby calves or whatever. It’s, you know, this. Okay. Let’s see what

25:11 – 25:12
happens. That is my zero condition.

25:13 – 25:13
Yeah.

25:14 – 25:21
And then I have my positive control, which is, let’s inject this baby calf with something which

25:21 – 25:25
will turn All the white blood cells on. We know it does.

25:25 – 25:26
Right.

25:26 – 25:31
So okay. In our system, let’s see what happens. And we have, you know, All the locusoid counts

25:31 – 25:37
going up or all the loop, interleukins being activated, whatever. So that’s my positive control.

25:37 – 25:44
Okay. Mhmm. Now let’s put in the agent we think is going to do something in this system.

25:44 – 25:45
Right.

25:45 – 25:50
Let’s see what happens in this leukocyte kind of whatever the system is I’m using to measure.

25:50 – 25:56
Okay. We have that. Okay. Now let’s see what we what happens If I put in a blocking agent without

25:56 – 26:02
my stimulus, nothing happens because nothing’s being seen. Now let’s put them in together. Is

26:02 – 26:07
it really blocking? Yeah. Because Yeah. The original is going down by 50%.

26:08 – 26:08
Right.

26:08 – 26:15
Okay. So, this is blocking that. Okay. Good. Next experiment. Let’s try the same thing in another,

26:16 – 26:20
setup with different readouts. Is it happening there too? Is it happening there too? Is it happening

26:20 – 26:25
there too? Is it happening there too? And then they cycle through this because once they’ve

26:25 – 26:29
shown this to be a blocking agent, they’re gonna see, okay. If I block it and now I stimulate

26:29 – 26:34
something else, What will be the effect downstream, and it goes on or not? But the basic experiment

26:34 – 26:39
is always positive control, negative control, Action blocking.

26:39 – 26:39
Right.

26:39 – 26:41
So those are the generic labels.

26:41 – 26:42
Yeah. Yeah.

26:42 – 26:47
So why should would I need to read 16 graphs with hrzbq25

26:51 – 26:51
Right.

26:51 – 26:56
And have to go down to the footnote, which is half page long. And halfway, it says, Stimulates,

26:58 – 27:01
a b c l l five o x

27:01 – 27:01
Yeah. Yeah.

27:02 – 27:06
Which is the readout system that I’m using in this little panel.

27:06 – 27:07
Right. Right.

27:07 – 27:11
Anyway, I’m I’m using a lot of words, but that chapter cost me 3 months to make. And in the

27:11 – 27:17
end, well, you saw you’ve seen it in the book. I hope you said, okay. You know, from the original,

27:17 – 27:22
this is quite an improvement because I sort of can see what they’re doing with little text saying,

27:22 – 27:28
in this experiment, we showed blah blah blah, but not that. And then here, this was stimulated,

27:28 – 27:34
and the labels, The text and the labels are as generic as possible with as little abbreviations

27:34 – 27:41
as necessary, and then boers of walks you down that path of little multiples so that at the

27:41 – 27:43
end, you have the story.

27:44 – 27:45
Yeah. It’s kinda like a little comic book,

27:45 – 27:48
Yeah. It’s a comic book. Exactly. That’s the word.

27:48 – 27:53
But it it it’s interesting. Your your story of that what’s interesting about it is that your

27:53 – 28:03
focus is not so much on, you know, should this axis be this width, and should the grid lines

28:03 – 28:09
be this color, And should the line be such and such? It’s it’s interesting to hear you talk

28:09 – 28:17
about it because the focus of of that story was on Making the labels and the language accessible to people.

28:17 – 28:18
Exactly.

28:18 – 28:22
And and and it is interesting to me because this is one of the things that I talk about with

28:22 – 28:28
people that that it’s it’s about the titling and the labels and the annotation because, you

28:28 – 28:34
know, a graph, Whether a graph is is beautiful or not is a lot in the eye of the beholder. Right?

28:34 – 28:41
You like blue, I like green, whatever. But it’s being able to Tell people what’s going on in

28:41 – 28:46
the graph with the words. And and, I guess, I’ll say one last thing, and then I’ll I’ll let

28:46 – 28:50
you just respond. That I think you’ve heard this too where people say, oh, you should be able

28:50 – 28:57
to look at a graph and get it right away. And I just I don’t think that’s true because you need words.

28:58 – 29:04
No. Well, you know, there are rare instances. I mean, they say, you know, one one picture is

29:04 – 29:12
worth a 1,000 words. Okay. But, the the usually, the graphs that are immediately obvious are

29:12 – 29:19
those Situations where I say, did you really need a graph for this? So, you know That’s right.

29:19 – 29:27
Yeah. If I show you a graph with A mean length of men versus women, and you see a bar with men

29:27 – 29:32
being higher than women. You say, okay. So men are bigger

29:33 – 29:33
Right.

29:33 – 29:38
Larger than women. Then I say, okay. You could’ve said that in one sentence.

29:39 – 29:39
Right? Yeah.

29:39 – 29:45
Did you really know who this part this is what what pharma used to do. They’ve moved on from

29:45 – 29:49
that, I have say, but they used to show you these graphs, you know, before, after Mhmm. Big

29:49 – 29:58
big bars. Oh, right. Oh, yes. Oh, now I see it in a graph. Oh, man. No. I I think and and there’s

29:58 – 30:08
also this other thing of, I think more, traditional scientist scientific views that, you know,

30:08 – 30:14
you should not impose on the reader Boers interpretation of the data. I’m just showing you the

30:14 – 30:21
data, and then you can make your own interpretation. And I’m not gonna bias you by telling you,

30:21 – 30:29
Well, that’s all nonsense. Yeah. You know, in in in current times, people spend maybe 2 or 3

30:29 – 30:34
minutes on an abstract before moving on to the next. And if you want to hit them, you hit them

30:34 – 30:42
between the eyes. So the the graph and the caption, and and everything is telling you that message

30:42 – 30:47
again and again and again. So your caption should not read, Here’s the results of my experiments.

30:47 – 30:52
I mean, that’s not gonna help. It’s gonna say, this clearly shows that a is blocking b.

30:52 – 30:59
Yeah. I find this fascinating and it happens in scientific publishing all the time where people

30:59 – 31:06
say, oh, I can’t tell you what the result is in the graph. But if you go to the text that they

31:06 – 31:11
write, the text will tell you exactly what the argument is. But, like, when you the graph, it’s

31:11 – 31:18
like it’s just purely descriptive. And I I never understood that disconnect between What I write

31:18 – 31:23
in the text versus what I write in and around the graph? Like, why don’t they don’t

31:24 – 31:30
Well, yeah. I mean, the same discussion around What one calls declarative titles of papers.

31:31 – 31:40
So, many journals will not allow you to say, A randomized trial showing drug a is good for patients with b.

31:40 – 31:40
Right.

31:40 – 31:47
No. No. No. No. No. You have to say A randomized trial of the effect of a in patients with b.

31:47 – 31:50
Yeah. But in the text?

31:50 – 31:51
In a text, it’ll tell

31:51 – 31:57
It’ll tell you that first piece. It’s it is it’s just so frustrating to me. I just I just don’t understand it.

31:57 – 32:02
But I I I want to I want to come back to what you said earlier Yeah. Because It it is really

32:02 – 32:13
2 things. I think it’s sort of, those two things, labeled, I think Jon Cleveland. You’re you

32:13 – 32:17
know of John Cleveland? It’s an old book, but he was also one of the William.

32:18 – 32:19
William Bill Cleveland.

32:19 – 32:23
Bill Bill Cleveland. Yeah. Yeah. Yeah. Yeah. Yeah. Together with, with Tuft. Yep. Those were

32:23 – 32:31
the 2, you know, Pioneers in the field, the Giants, that we still worship every day. But he

32:31 – 32:37
I think Bill Cleveland said clear vision and clear understanding. And and those 2 are you need

32:37 – 32:44
them both, and and they’re not completely independent. But, first, you think, okay. What’s the

32:44 – 32:50
main message here? And and how I’m gonna tell this story, that is the basic, and and that would

32:50 – 32:55
would be maybe your storyboard of your multiple panels saying, I wanna show the first, this

32:55 – 33:00
experiment, then that experiment, then that. And that these three together leads to a conclusion

33:00 – 33:05
which I can use in my next set of experiments, etcetera, etcetera. Mhmm. If if that’s not there,

33:05 – 33:11
if the order there is wrong, then you’re immediately lost. Yeah. Then then the clear vision

33:11 – 33:21
comes in with Not only clear labels, but also consistent labels Mhmm. And color coding, And

33:22 – 33:30
cues in the symbols that you use, and the thinness of the versus the thickness of the series

33:30 – 33:39
lines Mhmm. And always, making your experimental condition slightly more prominent than your

33:39 – 33:45
control, but either by color or by hue or what have you. Mhmm. And that is the technique of

33:45 – 33:53
clear vision. It it’s driven by the story, but you want I mean, it’s as simple as this happens

33:53 – 34:00
all the time. So You have this order of, you know, a, then b, then c, and then you have a graph

34:00 – 34:09
where the order is c b a. Like, you know, immediately, your brain is, hey. What was first here?

34:09 – 34:11
Right. Right. Yeah.

34:11 – 34:14
And that’s usually the the programs that do that. Right? They they they have a Yeah.

34:14 – 34:15
There’s some

34:15 – 34:17
PolicyViz give you the wrong order.

34:17 – 34:17
Right.

34:17 – 34:25
So all these things is designed. And and like in tables, Ordering is designed. It’s so it it

34:25 – 34:29
boers of fits together, and you need to go back and say, okay. Now I fiddled around with my

34:29 – 34:34
labels, but is the story still clear? Oh, no. Because here, I used another label be because

34:34 – 34:39
it fit better or because it was more appropriate in this graph. But given the story, perhaps

34:39 – 34:45
I should change it to the other label because then it the consistency is is is maintained. Boers

34:45 – 34:47
I can’t, but then I thought about it.

34:47 – 34:52
Right. I wanted to ask you one last question on on the small multiples. So some of the examples

34:52 – 34:59
in the book have, Say an odd number of graphs. But the way, at least some of the examples are,

34:59 – 35:04
is, you know, you might have, let’s just say, 2 squares next to each other than a rectangle

35:04 – 35:10
below it that kind of spans across. What would you recommend to people if they have small multiples,

35:10 – 35:16
but they have an odd number? They have 1, 2, 3, and they have this blank area. Right? That’s

35:16 – 35:22
kinda left over. What do you what do you recommend people do with that blank area?

35:23 – 35:31
That’s a difficult question. Also, because, usually, you’re not the one making the page layout.

35:31 – 35:32
Yes, that’s true.

35:32 – 35:45
So so, Any form you produce will be handled or mishandled by someone In the production process

35:45 – 35:55
Mhmm. Turning it into something different. I I remember, with The tables article that I wrote

35:55 – 36:03
that they got the layout so wrong that I ended up, you know, printing the proof, Mhmm. Cutting

36:03 – 36:04
it into pieces

36:04 – 36:04
Yeah.

36:05 – 36:09
Putting the layout together, and then making photographs that say, I want it in this order.

36:09 – 36:15
Yeah. Sure. So if you have these If you have these 3 panels, they could either make a big, white

36:15 – 36:22
area boers they could just fill it up with text, in which case It maybe it doesn’t matter, or

36:22 – 36:26
they put the caption there, which if it’s big enough, you know, fits inside, then you have to,

36:26 – 36:32
you know, place your, I would say then then, you know, you have a a row on top if I do it in

36:32 – 36:39
mirror image, and I would put the the third one on the right side, and then I put the caption

36:39 – 36:45
on the left below the first well, you know, that sort of thing. If if they allow you to fiddle

36:45 – 36:52
around, that’s That’s fine. I don’t think you need to fill up a square space with a, you know,

36:52 – 36:59
a stretch graph just because then Yeah. You’ll have the block filled. But it it it it boers

37:00 – 37:08
remind me To say that it’s really important that you think about the limitations of the journal

37:08 – 37:15
page, whether it be electronic or paper, it’s it it isn’t really helpful if you make graphs

37:15 – 37:20
that fill 1 and a half column, where you know it has 2 columns because either they’ll blow it

37:20 – 37:24
up, which would be nice, or they’ll make it smaller.

37:24 – 37:25
Smaller, right.

37:25 – 37:30
Where, and then everything becomes unreadable, So you have to think in advance. Okay. What do

37:30 – 37:35
I want for this figure? Can it be in 1 column? Can it be in 2, or should it be a full page?

37:35 – 37:40
And if so, Should it be rotated, which is usually not good for figures because people read upright,

37:41 – 37:42
etcetera, etcetera. So that kind of Yeah.

37:42 – 37:48
Yeah. Yeah. No. But that’s but that’s a good point. And it is it is true that a lot of journals

37:48 – 37:53
don’t tell you that until The very end. Or they never tell you that, and they just say this

37:53 – 37:55
is how it’s gonna be, and then you’re you’re

37:55 – 37:59
And and most authors are so thrilled that their paper is finally accepted.

37:59 – 37:59
Right.

37:59 – 38:08
They don’t wanna Jon difficult where the reality is the, you know, the science editor is usually

38:08 – 38:11
on your side, and it’s just as exasperated with the production process

38:12 – 38:12
Right.

38:12 – 38:19
As as you are, but is out of the the loop. Right. The moment it goes into production, you’re

38:19 – 38:25
dealing with technicians Yeah. Who may or may not be helpful. Right. Perfectly fine just to

38:25 – 38:30
say, you know, Sorry. This proof is not okay because this and that. And then, I’ve had proofs

38:30 – 38:31
come back 4 times.

38:31 – 38:32
Yep.

38:32 – 38:37
But then, of course, I’m a cranky old man. So If you’re a junior felon,

38:37 – 38:45
that that would be a good deal. Well, on that note, So the book is data visualization for biomedical

38:45 – 38:51
scientists, creating tables and graphs that work. Maarten, where can folks find you if they have

38:51 – 38:56
Questions boers they wanna bring you in to talk, where what’s the best way to to find you?

38:56 – 39:01
Well, I think it best it best were if you Just were to project the the slides I sent you with

39:01 – 39:11
my personal details, also the The link to the book, if they wanna buy it, it’s it’s QR codes.

39:11 – 39:16
And I have to say, you know, I’m actually, Next week, I’m gonna be in in in Canada, in Toronto

39:16 – 39:25
to give a a data course, which is, you know, a half day affair with Some prep work, and that’s

39:25 – 39:30
really a lot of work to do. Yeah. You know? So I I do travel.

39:30 – 39:31
Okay.

39:31 – 39:36
And I I I don’t I don’t organize these courses myself mostly. I just tell people, you know,

39:36 – 39:40
you want me to give that boers? Okay. Then I’ll come over, and then this is what you need to

39:40 – 39:42
do. So I’m in.

39:42 – 39:47
Yep. That’s great. Well, I’ll put links to the book and to your to your classes site and, folks

39:47 – 39:51
can get in touch. Maarten, thanks so much for coming on the show. Really appreciate it. It was

39:51 – 39:52
a great conversation.

39:52 – 39:53
Thanks.

39:55 – 39:57
Thanks for tuning in to this week’s episode

39:59 – 39:59
of the show. I hope you enjoyed that.

39:59 – 40:04
I hope you will check out Maarten’s book. I put links to his website and all of the things that

40:04 – 40:10
we talked about on the show notes at PolicyViz. And you can check out other resources, other

40:10 – 40:15
tutorials, and other things around the world of data and data visualization at PolicyViz. Thanks

40:15 – 40:21
so much for listening. Until next time. This has been the Policy of this podcast. A number of

40:21 – 40:26
people helped bring you the Policy of this Podcast music is provided by the NRIs. Audio editing

40:26 – 40:31
is provided by Ken Skaggs. Design and promotion is created with assistance from Sharon Sotsky

40:31 – 40:35
Remirez, And each episode is transcribed by Jon Transcription Services. If you’d like to help

40:35 – 40:40
support the podcast, please share it and review it on iTunes, Stitcher, Spotify, YouTube, or

40:40 – 40:45
wherever you get your podcasts. The Policy Mizz podcast is ad free and supported by listeners.

40:45 – 40:51
If you’d like to help support the show financially, please visit our Paypal page or our Patreon page atpatreon.com/policyviz.