In the last 4 weeks of Makeover Monday we’ve had 3 weeks with live events in Sydney, Amsterdam and Paris, as well as a week of collaboration with #VizForSocialGood and the Inter-American Development Bank. And in a week’s time we’re at Tableau Conference on Tour with a big live session on day 1.
Who else feels just a bit out of breath? Yup, me too. So for this week I wanted a simple dataset with plenty of options and I found one that will fit quite nicely.
For week 22 we’re looking at Internet users per 100 people across countries globally over the past years and decades. The original viz below was posted my Knoema.
What I like about it:
- The interactivity is engaging. The map is animated based on the years in the dataset, so I can see how the internet user base grew over time in different countries
- A map is always a good way to get me interested and I like seeing the countries in relation to one another
- The legend is prominently displayed so I can refer to it easily
- The tooltips give me the necessary information for each country
- The timeline at the bottom shows me that there is data from various years, which in turn draws my eyes to the pay button to explore the animation.
What I don’t like about it:
- The colours are not great. They don’t cater for viewers with red-green colour blindness, plus the green and yellow hues are not distinct enough to differentiate
- The legend changes all the time. WTF? Click on the play button and watch the animation.
The legend changes dynamically for every year, so red, yellow and green never mean the same thing across years. That is really bad, because with an animated visualization, I need to have some constant reference points, otherwise my brain can’t interpret the data that flickers across the screen…
- While I like maps and I understand why a map was used to show internet users by country globally, maps aren’t always the best way to present geographical data. On the one hand, small countries can become almost invisible on a global map, which doesn’t do them justice, while large countries appear disproportionally large, simply because of their size, and even if they don’t have much of an impact data-wise.
On the other hand, when we then apply colour to those country shapes, things can get worse, because colour looks different on a small vs a larger area and again this can impact our interpretation of the data.
- I still can’t get over the dynamically changing colour legend…
- On top of changing dynamically, the legend also doesn’t have any explanation as to what those colours mean. A couple of extra words to explain the numbers would help me a lot, especially because the title is quite far removed from the viz and the brief description doesn’t provide any useful context on the colours.
- While the tooltips give me the basic information for each country, there could be so much more in there. Why not give the number as a %? 7.6 per 100 people makes 7.6%, which is much easier to think about.
Why not also provide an average for the region as a benchmark, or a comparison to the prior year or decade for reference purposes?
What I did:
- Despite picking a simple dataset, it took me a while to arrive at my final viz, because I always struggle to pick a story and then formatting just takes forever
- I decided early on to do a slope chart. I haven’t done one before and I thought it would be a nice and elegant way to show the change over time between two key years – 2000 and 2015
- I wanted to reduce the dataset and grouped countries into regions to arrive at a cleaner and simpler chart. While I knew that with population differences etc. I couldn’t really compare at the regional level, I initially used an average across those regions for my chart. Andy rightly challenged that – when he saw my draft he probably thought ‘WTF Eva, didn’t I just tell everyone last week to NOT do this?’
But sometimes you need to work through these things in a couple of iterations before the fog lifts…
- So out went the region idea and all that was left from my grouping exercise was the decision to pick a single region and go back to the country level. This way the chart wouldn’t be too cluttered and I could try to find a story.
- The smallest regions were Middle East and Oceania, so I picked the Middle East because it’s the less obvious choice, but also because the story seemed more interesting to me with a much stronger growth in internet users
- What I ended up with is the chart below, showing what happened across the countries and elaborating on the top and bottom ranked (by number of users per 100 people) nations.
Click to view on Tableau Public