Who votes, who doesn’t

National Elections! Europe had a lot of them this year. We still remember the Dutch election in March, the UK and the French election in June and the German election in September. And these are just the big ones. Liechtenstein, Hungary, Bulgaria, Albania, Kosovo, Norway, Malta, Slovenia, Austria, the Czech Republic and Iceland all had national elections this year, too.

After that packed election year in Europe, I want to take a general look at voter turnout on this continent. The voter turnout tells us what percentage of the citizens (who are allowed to vote) actually turn up to the ballot on election day and vote. The differences are quite stark in European countries:

Let’s go straight to the decisions I made around this chart. We have a lot to untangle:

Chart choices

This map is a tricky one. We’re comparing apples with oranges: The last national elections in European countries happened in different years. Of the six countries with the highest turnout, only one country voted in 2017.

So is it still ok to compare these values? I think so. But I took safety measures: Besides the wording, I chose a choropleth map as my chart type. Why do I think that’s a safety measure? Let me explain. Afterwards, I’m very happy to hear if you think it’s appropriate and/or sufficient:

A few weeks back we realized that choropleth maps are a bit dull because they can only show one value. But there is another reason why they shouldn’t be the unquestioned first choice for all our geographical data: Choropleth maps are imprecise to our eyes. Humans have a harder time to distinguish different colors than e.g. the different lengths in a simple bar chart.

Have a closer look at East Europe. It’s all very red there. If the labels weren’t there, could you guess which country had a higher turnout at their last election: Lithuania or Bulgaria? There’s a 3.5%-point-difference between the turnout of these two countries, but it’s hard to spot. Imagine seeing this data as a bar chart. You wouldn’t have any troubles at all to even perceive small 2%-point-differences.

I see that perceived imprecision as an advantage for our imprecise data. In fact, that imprecision is the main reason why I chose a choropleth map instead of a bar chart. I want my readers to see a general trend: Nordic countries have a higher turnout in general than East-European countries; the turnout in the four countries surrounding Switzerland is way higher than in Switzerland itself. But I don’t want them to remember later that Bulgaria had a higher turnout than Lithuania by exactly 3.5 %-point.

Although I’m happy to take advantage of one of the disadvantages of a choropleth map, there are more of them that bug me. First, choropleth maps make big areas seem important, even if we have no clue about their population density. Sweden looks really dominant on this map, but fewer people live there than in Hungary. Second: We don’t see important values if the area they’re in is too tiny. I bet you’re not able to find the country with the highest turnout in the data. Should I give you some time?

…It’s Malta! Yes, the five pixels at the bottom of the chart. These tiny pixels had a crazy turnout of more than 90% for more than 5 decades now. (Consult this paper to find out the many reasons.) In a ranked bar chart, Malta would be the star. In our choropleth map, nobody will ever notice it.


I’m getting very excited about Christmas🎄. My vacation starts tomorrow already, but I’ll be back with another chart next week. I’m super excited to present to you…a chart about charts!

Comments