The same data can suggest totally different stories just by changing the presentation. Even little visual details (like the ones we will tune below) can heavily impact data interpretation, especially in an information overloaded lifestyle. The python code to repeat the following experiments is available at the end of the article.
We use a dataset about GDP per capita in the world in $, downloaded from The World Bank.
We purposely start from per capita data instead of showing simple GDP because that would be misleading, since it does not consider the number of people living in a country. GDP in itself is a gross metric poorly capturing the real wealth of a nation, it’s worth checking newer alternatives.
Ordering nations by GDP per capita in 2014 leads to the following:
In order to make the plot say “our economy is going really bad” we just exclude countries in which the GDP per capita is less than the nation we want to make look poor (let’s say the UK).
We applied the same bias ourselves in the first plot, by showing only the first 50 countries over ~250. Working on global data is actually a good excercise to avoid media brainwash and remember how lucky and rich we (western people) are.
To further increment the sensation of fear we make the same operation vertically, by cutting out the big part of the plot and make the y axis start just a little bit lower than the “poor” country amount. This simplification is often done in good intention with the aim of making the plot slicker, but partecipates in making the visualization and the following insights biased.
Oh, poor UK!
Plots like this are rarely accompained by contextual descriptive statistics and take often part in a media agenda. It’s smart to think visually, but when you suspect a bias always take a look at the raw data, check the sources and consider the wider context… or ask somebody you trust to do it for you.
It is a great responsibility for professionals in data visualization and journalism to convey the truth without falling in the temptation of making drama (and get more social media attention, and thus more advertisers funding their publication). The reader is responsible to adopt a critical eye when facing a visualization, and pose questions on how the plot was made and on the data it is derived from.
Data literacy is vital for the newcoming citizens of the world.