Makeover Monday, 2017 #20

For week 20 Makeover Monday is collaborating with #VizForSocialGood and Inter-American Development Bank to look at youth employment trends in Latin America and the Caribbean. Great data set, great cause and a great opportunity for our data visualisations to make a difference. For some reason I also felt an increased sense of responsibility to understand and accurately represent the data!


Thoughts on the original charts

The original article walks  the reader through the overall headline figures, explaining the various categories and ending with a look at the sectors that young people are employed in. I spent a lot of time trying to understand how the various categories (Ninis, Nininis, unemployed but studying, informally employed, etc) added up to the total number of 15-24 year olds. So much so that in the end this seemed like a good angle to visualise. If I was struggling to make sense of the categories then there was a good chance others were too, and so explaining that graphically would be valuable.


The makeover

Design wise I wanted to bring in key numbers from the original article as headlines, but present and compare the proportions graphically. Pie charts were an option for the graphical component (given a limited number of parts to the whole), but waffle charts seemed to be a better fit for the flow of the visualisation:



The visualisation is also available on Tableau Public, where you can choose a country to drill into.



A quick nod to Andy Kriebel and his very helpful blog post and video on producing waffle charts in Tableau. This was my first attempt to create a waffle chart and Andy’s video was invaluable!



The first challenge was understanding the categories! You’ll note from the final waffle chart that I’m not quite there yet. If you hover over the grey section on the left (in the interactive version) you’ll see that I’ve labelled it “unknown”. I’m guessing that this category has to be those 15-24 year olds who are studying or training and are not seeking work. It’d be great to hear other people’s thoughts on whether this is correct, and whether I’ve accurately represented the categories.

Challenge two was a bit more prosaic. I built the headline components of the dashboard as worksheets in their own right. Each of these headline worksheets had a single text label incorporating the various numbers with text. What I hadn’t remembered until I finished was that I couldn’t apply a filter from one worksheet to another worksheet with a different primary data source. Rats! I had to go back and start these again, pulling in a dummy value from the waffle chart grid data source so that each worksheet had the same primary data source. Was there a better way to do this? If so I’d love to hear about it.

The final challenge was colouring the text in the headlines to avoid the need for a colour legend underneath each waffle charts. I could improve this aspect because the colours develop as the categories are expanded and consequently some colours are technically given two meanings.


Final thoughts

What else might I change? A waffle chart isn’t always as accurate as a pie chart (unless you can show enough squares!) so the eagle eyed will notice some rounding issues – e.g. two squares shaded for the 2.5 million unemployed who are studying out of 100 million young people . It would probably have been better to have more than 100 squares in each waffle to allow for more accuracy. Adding the percentage into the hover over would help here.

Part of me thinks that a concluding paragraph would be useful, but I wasn’t confident adding this with the unanswered question of the unknowns. Nevertheless I’ve learnt a lot about issues in employment for young people in Latin America, used a new chart type and hopefully contributed something valuable to the overall conversation and understanding.