Telling the Story with Maps
We had an interesting team discussion last week about the "why" of putting data on maps, a subject Chris D touched on in this silo post after a World Bank discussion on the topic, and Ian Schuler blogged about in his Burma election post yesterday. The question on the table was a simple one: why are these maps important for NDI's partners and our democracy work in general?
One of the themes worth mentioning was a reminder of the importance of providing visualization tools, including but not limited to maps, that not only provide the data in a visually engaging way but also tell a story - a key to making the data speak to and persuade the audience. In many cases, it's not enough to simply put political information on maps without context - in fact this could have the opposite effect and lead to misleading inferences or undermine your goals in some situations.
This simple and seemingly obvious concept is one that our team needs to keep in mind, and reinforce to our partners and staff around the world as web-based mapping applications become more popular.
One technique we've been using lately to accomplish this is to mash-up our main election data with supporting data sets to provide better context - often done using additional map layers. Our recent projects in Bahrain and Afghanistan demonstrate the power of this approach. Shaded areas or points on a map can have much more impact when provided in the context of other factors beyond just geography and/or location - such as demographic, economic, political, health, security - or a wide range of other data sets. I highlighted a couple examples from our recent programs below, and Development Seed - one of our tech partners - has done some nice work for the New America Foundation with Pakistan data here and on Pakistan flood response here.
Finding data that provides the context or supports your case can be a challenge, one reason we're strong supporters of the open data movement.
Another useful approach we've been working in to the projects is to use color to differentiate between different categories of mapped information. This provides the relative frequency of each category visually in the context of the overall data set. A great example is the Burma Election Tracker site, launched by our partners the Burma Partnership <deprecated link removed> last weekend, that provides election related violations that are not only represented by location and frequency using circle size - an increasingly familiar approach - but also color-coded by violation type so that you can infer the relative frequency of each type of violation. The site goes a step further by allowing you to drill down to map views and lists of each incident type (more features including provincial data on mouseover and provincial map layers should be live tomorrow, stay tuned).
A promising next step for us might be to use shape to differentiate report types or origin on maps - so different shapes for SMS, web text, video or audio reports for example; or use shapes to represent the group who made the report - trained election observers, human rights monitors or other civic activists, journalists or bloggers, international election monitors, public citizens, etc.
Lots of room to innovate here. I anticipate using these and other new and interesting approaches in several of our upcoming projects as our partners around the world focus more on using maps to tell their stories.
Here are snapshots from the examples mentioned above:
The Burma Partnership is using data collected from a number of courageous human rights and other civic activists during the election period surrounding the May 7 military elections. The group used map layers to highlight the types of electoral violations using circle size and color that highlights the relative frequency of each type of electoral violation reported.
A story emerges of heavy government involvement in the process (the shades of blue), when the regime "abused its authority and resources to lend political advantages for its proxy parties, going so far as to deny government services to those not voting for SPDC backed candidates". While not a representative sample, the data suggests that government meddling in the process was a widely used tactic to manipulate the outcome of the election.
Bahrain Parliament Elections 2010
During the October, 2010 Bahrain parliamentary elections the government promised to address past problems with voter disenfranchisement caused by overcrowded polling centers in populated areas by opening additional "super polling centers" in densely populated areas.
However, as the maps demonstrate visually, the story is quite different. In all cases except one the super centers (green markers) were placed in low density areas, while the high density areas (filled in blue) were excluded.
The story of the last two Afghanistan elections is one of fraud and violence.
The correlation between security and fraud is evident in this 2009 election results map where red circles indicate the frequency of fraudulent polling center results and the map layer underneath represents frequency of recent security incidents- pink and orange being the highest levels and green less violent. Clear correlation between election fraud and security - that was an importnat story around the 2009 presidential elections in Afghanistan.