Makeover Monday 2020 W3: UK's sugar intake

Makeover Monday's week 3 was a quick and fun exercise for me. It is basically made out of all the usual classics that make for a good visualisation - a bar chart, reference lines, icons and a sweet colour palette (pun intended). It was simple in the sense that nothing ground-breaking happens in it, but the result was one I'm quite satisfied with. It doesn't always happen. Below you'll see a bit of my process to build it.

The original viz:

The original visualisation was posted by the BBC in their article titled "Is it time to treat sugar like smoking?", which is based on a study published by the British Nutrition Foundation, available here.

What works well:

What could be better:

My approach:

1. Understanding the data & my potential audience:

The data provided by the Makeover Monday team was simple: the age groups available in the research's data and the percentage of total calory intake that represents sugar consumption, across the years of the survey.

Although I could have gone with the data I was given, I noticed the survey results available on the British Nutrition Foundation website added a few interesting insights: intake of other nutrients, such as intake details of fibre, vitamins, protein and fat. It also dives into more detail about the sugar intake research, informing the breakdown of the sources of free sugars consumed at each age group.

I then decided that instead of trying to improve a nearly perfect viz, I'd do my own take on the sugar consumption insights of the survey data.

In my mind, this viz was to be seen by the same people that would be potentially reading the BBC article. That is - the general public that would be interested in knowing more about sugar intake.

2. Defining which story I'd like my visualisation to tell:

This week's viz had a very straightforward message to be told: in the UK people consume way more sugar than the recommended quantity - especially teenagers. A huge portion of this consumption comes from beverages, and as we age we tend to try to make healthier choices - but it's nowhere near enough to be considered healthy.

I quite like the idea of adding interactivity to visualisations. In this case, I wanted to find a way to make people curious about their habits and perhaps find out what they could change towards a healthier relationship with sugar - hence, the idea of including the breakdown of the sources by age group and make it somewhat entertaining.

This is what I wanted my viz to say:

3. Getting creative & Sketching:

The beauty of data visualisation is that it's a creative science, whose main purpose is to communicate a message or story to an audience. It has several parallels with writing for this reason: to write a good book, the writer must understand their audience, their genre, the tropes of a good story, and how to take the readers through the message they're trying to convey.

In Data Visualisation's case, this is achieved through the use of visual encodings: position, scale, size, direction, angle, area, volume, shade and colour. If you want a full master class about visual encodings applied to data visualisation, I can't recommend Alberto Cairo's work enough. His work played a huge role in why I decided to follow the path of developing data viz for a living, and everyone interested in the topic should go read all of his books - but this is probably a story for a different post.

In this week's Makeover Monday I decided I could explore more the use of colour in data visualisation - which is always an interesting topic to discuss and that seems to cause shivers whenever it is brought up. It often goes through doubts like:

All these questions are very relevant to be asked, but I often see answers erring to the side of caution - which results in beginners only feeling safe with shades of grey, or the sparse hint of colour if absolutely needed - and how dare you not make that hint of colour a shade detectable by all ranges of colour-blindness! - which kinda kills the creative part of the data viz science.

A good way to start is this article, in which the author explores all the challenges of coming up with a good colour palette for data visualisation.

Another helpful resource is to use Adobe Colour. The way I do it is to find a good reference image related to the theme I want to explore and pick the colour palette used. This way, I have a general colour guideline to follow in my visualisation.

Finally, a rule of thumb I always follow to make any viz: if it still sends the message if it's in all black and white, then you're doing it right. The hue (how light or dark an element is) and the contrast between elements is usually enough to make for striking attention-grabbing visualisation.

Last, if you're developing for accessibility, a tool I find really helpful is the Colblindor Colour Blindness Simulator. You can snip n image of your viz, upload it and play with the whole range of filters for different types of colour blindness and check if your viz still sends the right message across all accessibility points.

With all of this in mind, I first chose the colour theme I wanted to work with for this viz.

Since it's about sugar intake, I thought a "sweet" kind of colour palette would be fun (think macaroons).

These were the palettes I landed on after some tweaking:

Sketching:

I had a fun time with this one because there were a few ways I could go about telling the story I wanted to tell. The sketch I landed on wasn't what my viz ended up being, and that's alright - no good data visualisation happens in the first iteration. It's always a very long process of tweaking here and there - almost like gardening.

The sketch:

In my sketch, I entertained the idea of making a population pyramid kind of chart, splitting the data between male and female. However, once I built it I noticed how it was taking away from my original story point. I ended up confused looking at it, imagining what I was trying to compare between genders. I ditched that idea on the go, but the rest of the framework remained the same - so sketching was still very helpful! Just in an unexpected way. :)

The viz I didn't go with:

So I moved on with the idea of keeping just the aggregated numbers for the main insight.

4. Building it in Tableau:

Tableau is a tool I love, but it has its annoying quirks. One I discovered this week is that Tableau Public doesn't translate correctly all fonts it has available in the desktop version. Only a handful of fonts are supported across platforms. Not only that, but its conversion font of choice is Times New Roman - if you use your sleek Helvetica stye font and it's not supported, it will become our old seriffed friend once you upload it.

Sarah Bartlett (@sarahlovesdata) - Data viz enthusiast, Tableau ambassador, part of the #datafam and overall great person - kindly shared this resource with me, that explains which fonts are supported or not by Tableau Public - for easier reference:

Also, if you want a detailed exploratory view of all your background and font colour combinations in Tableau Public and what they might look like, Danny Bradley (@vizwithdan) shared this workbook available On Tableau Public.

Moving on to my viz, the build was really simple. A bar chart, with emphasis on the age group I wanted to highlight. One simple annotation right beside it.

The interactivity was reserved only to a dashboard highlight action, that made the icons below the main chart change according to the reader's age group selection above.

This was the end result:

5. Testing it with a non-data viz person:

This week I had little time to run after people and ask them what they thought of it, so I had two friends who happened to be nearby while I was working on it. In general, both of them loved it!

One of them didn't get right away the interactivity bit, so I figured adding an extra explanatory text would be helpful.

The other one commented about the recommended intake and how they'd add to the text above it that it refers the to the 5% mentioned in the axis at the bottom - so they don't need to move their eyes around the viz to understand what it's all about. Great catch! It's a very tiny addition that improves the overall quality of the viz greatly.

Feedback, feedback & more feedback

This week saw a huge number of Makeover Monday submissions, and that's great! There were many awesome ones and I think that I submitted mine a little too late (Wednesday night!), so I didn't make it to the #MMVizReview feedback.

The Community Feedback was overall positive and apart from the details raised above, I haven't heard much more about it... I'm assuming silent feedback means it works!

In summary, the points raised were:

Iterate!

You can find the interactive version on Tableau Public, at this link. This is what the latest iteration looks like:

What I've learned:

In great kiwi fashion: sweet as!

-----

You can follow me on Twitter @DataRocksNZ and check my upcoming Makeover Monday submissions.

And while you're at it, check my Tableau Public Profile