design of scientific posters ·

Visit the Poster Hospital to see redesigns of real-world posters and learn practical design guidelines for scientific posters and layouts on a large canvas.

design of graphical abstracts ·

Visit the Graphical Abstract Hospital to see redesigns of real-world abstracts and learn practical design guidelines for graphical abstracts and small figures.

Obesity — a Data Story

Rescuing nuanced pattterns from the clutches of a bad graphic

Here, I go through the details of the stories and their design in my poster of “BMI and Obesity Prevalence for 185 countries”.

a legend that teaches you something

As you'll see the interpretive text of the poster is itself like a mini-poster. It shows data vignettes from the poster and provides a brief explanation.

The legend doesn't simply explain what is shown. It highlights interesting observations and aims to teach you something. The aim is for you to get something out of this even if you don't look at the rest of the poster.

what is shown? how is it shown?

From the title, the poster is obviously about BMI and obesity. So, the first thing to do is define the BMI, whose equation `\textrm{BMI} = w(\textrm{kg})^2/h(\textrm{m})` can be written out in full.

Immediately after this, to place the absolute BMI values in context, I show the ranges for underweight, normal, overweight and obese categories.

The next paragraph describes a key feature of the poster — for a given country, BMI values for males and females are shown separately. BMI is shown as a dot and colored by gender: females are encoded by magenta and males by blue (because, why not). Critically, a solid dot indicates that the BMI in that gender is higher. For example, a solid magenta dot indicates that BMI in that country is higher in females.

▲ Introduction to the poster and BMI encoding.

an example of what is being shown — BMI

Right after explaining how BMI is encoded, I show an example of three countries.

First, we look at Japan, where female BMI is lower. The vignette shows both genders' BMI values and their difference.

I then look at Nepal, which has the same trend as Japan (female BMI is lower) but this difference in Nepal is much smaller.

The last country is Indonesia, where the trend is reversed. Here, male BMI is higher.

This choice of countries is deliberate. They're reasonably recognizable and their BMI ranges are similar so the data in the legend is compact. I sensitize you to two patterns: which gender has the higher BMI and how large is the difference between genders?

Notice that the tick marks on the BMI axis are little dots and BMI is encoded by dots. This is not a coincidence.

▲ A vignette of BMI statistics for three countries.

an extension to what is being shown — obesity

The poster also shows obesity statistics. This category has already been explained above (BMI ≥ 30).

Obesity is shown by a bar and the choice of color is the same as for BMI. The color of the bar shows in which gender obesity is more prevalent.

We see that in Japan the range of obesity is very small and males tend to be more obese than females. In Nepal the range is larger and in Indonesia larger still. And in both obesity prevalence in females is higher.

I introduce the OBP acronym (obesity prevalence) to keep text short. It's the only shortcut I allow myself.

Obesity is shown by a bar and the obesity axis is a bar. This is not a coincidence.

▲ A vignette of obesity statistics for the same three countries as above.

combining the BMI and obesity statistics

I now show how the BMI and obesity statistics can be combined and use the same three countries as above to do so.

I introduce the idea that countries can be categorized by the genders in which BMI and obesity is higher. For Japan this is male/male (blue dot, blue bar), for Nepal this is male/female (blue dot, magenta bar) and for Indonesia this is female/female (magenta dot, magenta bar).

Note that the BMI and OBP axes are distinct.

▲ Having explained how BMI and obesity are shown we can now combine the encodings.

countries by leading BMI and obesity

I introduce the four groups (A, B, C, D) that correspond to which genders have a higher BMI and obesity.

▲ Countries are assigned groups based on which genders have higher BMI and higher obesity.

relationship between BMI and obesity

This is a pretty complicated set of plots — it shows a lot but takes you through it at a maneagable pace.

For each country group (A, B, C, D) we see the OBP plotted against BMI for males and females. The average OBP and BMI across all categories for both genders is shown by a dashed line.

I fit the trends with either a line or second-degree polynomial (this is a subjective choice).

Note that the number of countries in a group is shown next to the group letter. Group B has only one country (Syria). I point out countries that might be of interest: Japan (lowest BMI and OBP in group A), Syria (only country in group B), USA (largest BMI in group C) and Nauru (largest OBP in the entire data set).

Note that the symbols used for the countries reflect the BMI encoding. This is a little unconventional, since both BMI and OBP are being shown on these plots. I thought that keeping color here makes a stronger connection to the rest of the poster.

▲ Trends between BMI and obesity for each country group.

lookup table — find your country

The bottom of the poster, which is the least valuable real estate on the canvas, shows the detailed statistics and ranks for all the countries, alphabetically sorted.

I'm a big fan of showing the raw data in a way that can be easily looked up. Sometimes space doesn't allow for this but when it does, it adds another way for which outliers (more so than trends) can be spotted. But even more importantly, it's a safety net for the reader in case they get frustrated because they can't find a specific country.

▲ A lookup table allows you to quickly find your country and provides all statistics at a glance.

BMI and OBP distributions

To fill empty parts of the poster, I provide the distributions of BMI and OBP values for the entire data set and by gender. I also show the distributions of the differences between genders for these quantities.

Right away, you can see that the differences for which female values are larger have a larger range. Both BMI and OBP distributions appear roughly bimodal.

▲ For data lovers: BMI and OBP distributions by gender.

distributions by country group

Teasing the data apart a little more, here I show the distributions of values and gender differences for each country group. Compare the tails of the difference distributions for groups and quantities where male values are larger (blue) vs females (magenta). The latter have much longer tails.

For all the distribution plots I chose to keep the y-axis absolute.

▲ Digging deeper: BMI and OBP distributions by country group and gender.

averages

Overall and gender averages are shown by colored lines that drop from their respective axes.

▲ A closeup of group A countries (BMI and OBP higher in males). Global and gender averages are shown as drop lines from the axis.

Easter eggs

Always leave room for an anecdote.

▲ An interesting about the USA is called out.