[This is a guest post by Ben Willers*, about his infographic project “The Love of Guns”]
Debate over the relationship between civilian owned firearms and crime within a given area continues to rage, with members from each side citing studies to support the belief that guns either increase or decrease the level of safety. It is a subject I have had a passing curiosity on for some time, but one that I have never looked into deeply before now. With many reports centered around studies within the US, I was keen to explore the issue on a wider scale, and see how different regions compare. This became a personal project undertaken in my spare time, and the final visualization displays data for civilian firearms per 100 people, and homicide rates per 100,000 people across 155 countries.
The project began in February 2013 when reports began to emerge that Oscar Pistorius had shot and killed Reeva Steenkamp, his girlfriend of three months, at his home in South Africa. After reading the initial news reports I felt compelled to research further, and began by gathering data on intimate femicides (the killing of a woman by her partner) and how rates in South Africa compare to other regions in Africa and around the world. This lead me towards other datasets, two of which I felt were particularly compelling when viewed side by side.
It came as little surprise to me that the US leads the world when it comes to firearm ownership (source), however the sheer number, almost 90 civilian-owned weapons per 100 people was astonishing. Yemen, the second placed country on that list, drops to an ownership rate of 54.8. Likewise, homicide rates per 100,000 people (source) revealed enormous variation between countries, with Honduras topping the list by a considerable margin.
Further study revealed some interesting trends. European countries have some of the highest gun ownership levels in the world, and also some of the fewest homicides. Africa and Latin America meanwhile have some of the fewest firearms, but suffer from some of the highest rates of homicide. Conversely, some other counties like Japan and Singapore feature close to the bottom of both scales, while South Africa features highly on each. These contradictory findings could be considered disastrous for a designer hoping to present a clearly defined story with firm conclusions, however I find the exploration process thoroughly enlightening. Complex issues of this nature are not always as black and white as some would like to believe.
The greatest challenge I face in any project is finding a way to present data which can be freely explored without exposition from myself, and the first step is deciding how much data to show. Include too much and I risk overwhelming the reader at the outset. Show too little, and I am not allowing them to investigate freely. The data I had amassed on this occasion covered 215 countries, and while it is tempting to present a complete view and show everything, editing is usually necessary to help users locate information which is likely to be most interesting to them. In this instance, I excluded any country with a population less than one million, leaving a more manageable, but still respectable 155.
Every dataset presents unique challenges when it comes to visualization, and this project was no exception. Two-thirds of counties featured had less than 10 guns for every 100 people, however this figure climbed dramatically for the top few, increasing to 88.8 for the US. To visualize all countries would require a scale which can accommodate all values. A linear scale would highlight the differences well, but would also result in a lot of unused space on the page, and countries in the lower half would be displayed so small it would be difficult to interpret their values. An alternative method would be to display countries with exceptionally high values simply running off the page, with an authors note explaining their try value. This would solve the issue of unused space, but this method fails to give any visual impression of difference, a critical factor as far as I am concerned. Another option was to use a logarithmic scale, however this idea was quickly dismissed as it would present a distorted this view.
After much deliberation I settled on a linear scale with the x-axis wrapped around a semicircle. This allowed unused space to be cropped at the sides, while providing room to display exceptionally large values towards the top of the page. Homicide rates per 100,000 people were then added on the right, which upon first inspection appears to produce a remarkably symmetrical effect. In reality though the two sides are anything but similar. Regions featured highly on the gun ownership scale often have low homicide rates and vice versa. To immediately highlight this I color-coded countries by region. In the final version these are yellow for the Americas, blue for Europe, white for Oceania, purple for Asia, and brown for Africa. I also tried using tints of colors for subregions, so North America would be distinct from Central America, South America and the Caribbean, but this became exceptionally difficult to read when all other regions were factored in.
In the center I had considered displaying two choropleth maps, each providing a geographical representation of where gun ownership and homicide rates were high and low. In the end the idea was dropped, partly because the maps would need to be quite large to be readable, and also because I felt they appeared disconnected from the rest of the design.
I instead decided to display a series of rings, each one representing a subregion. A dot is placed for each country where it intersects with it’s corresponding ring, and then these dots are connected by an arc. This has a number of benefits, it allows us to locate countries more easily (as long as we know the subregion) and it provides us with the means to compare countries within a subregion, and also subregions with each other. So, if for example you are interested in comparing the US with other North American countries, you locate the North American ring, then follow it round to easily pinpoint Canada and Mexico.
I believe that multi-layered, non-linear graphic representations of data allow us to make more informed judgments and comprehend on a much deeper level. While this method may not satisfy those looking for immediate answers, I find visualizations of this nature far more engaging and satisfying to digest.
*Ben Willers is a freelance graphic designer in London specialising in information and data visualisation. High resolution examples of his work, including “The Love of Guns”, are available to view at benwillers.com. You can also follow him on Twitter (@b_willers).