Big Data BBQ

The UK Guardian news­pa­per recently pub­lished an inter­ac­tive ring dia­gram show­ing gun reg­u­la­tion by US states. It’s an attrac­tive dia­gram but makes so many ele­men­tary mis­takes in chart­ing that it fails to sum­ma­rize the data in a mean­ing­ful way.

Analy­sis: If the graph makes a sin­gle point, it’s that all states reg­u­late whether a weapon is car­ried in pub­lic or not, but the North­east is the most highly reg­u­lated area for guns, while South­east is the least. Beyond that, it’s not clear what the point of the graph is — how does the user learn more by inter­act­ing with the graphic? What are the sub­tle points of under­stand­ing does the user get from the diagram?

Look­ing closer at the graphic, I find a lot of prob­lems with the execution:

• The cat­e­gories are het­ero­ge­neous: carry in pub­lic is one type of reg­u­la­tion vs. what type of gun requires a per­mit vs. data points on what is reg­u­lated (miss­ing firearms, lock­ing devices and points of sales (gun shows, pri­vate sales). For reg­u­la­tion, these seem like a jum­ble of unre­lated issues stuffed into one cat­e­gory. The cat­e­go­riza­tion is the most sloppy part of this graph and does not add to the under­stand­ing of the issues.
• The under­ly­ing data is poorly doc­u­mented, although there is a com­pan­ion post as to how the cre­ators strug­gled to sum­ma­rize the data (ie fit the data to the visu­al­iza­tion they were attempt­ing. There is also a lot miss­ing to anno­tate method­ol­ogy: when the graph is scaled by pop­u­la­tion, is the out­er­most ring pro­por­tional? or entire slice? what’s the rel­e­vance of “pop­u­la­tion”? Wouldn’t it have made more sense to have the slices pro­por­tional to gun own­ers? What is the time frame for the data? Is there a data table and a clear methodology?
• Color choice seems arbi­trary and changes for each data cat­e­gory. Does red mean “unre­stricted” when refer­ring to reg­u­la­tion but “con­cealed carry” permitted?

How the chart could be improved:

• Clearly define what the point of the visu­al­iza­tion is. Is it to show that there are geo­graphic dif­fer­ences in gun laws? Show how gun reg­u­la­tion impacts states with higher pop­u­la­tions? Show that states with fewer gun laws have more/​less vio­lent crimes com­mit­ted with guns?
• Con­sis­tent cat­e­go­riza­tion scheme and color scheme. a two tiered cat­e­gory of gun types (hand guns, rifles, etc) and chan­nels (gun shows, gun shops, and pri­vate sales) with a sim­ple scale (white= unreg­u­lated, light blue = par­tial reg­u­la­tion, dark blue = full regulation)
• Add a data table and a clear descrip­tion of method­ol­ogy used for cat­e­go­riza­tion and how the data fits the categories.

Over­all, I like the idea for this graph to dis­play mul­ti­ple dimen­sions, but think it is too mud­dled to pro­vide any insights.