Thousands of bloggers are taking the time to discuss a single topic today – poverty. As we sit in our cozy homes, go out to eat, watch movies, or simply read the news on a computer, it’s easy to forget that there are millions of people around the world who aren’t so well off. Blog Action Day is an opportunity to remember and to perhaps help out in some way.
Mapping Poverty Rates
I of course took the visualization route. What better way to get the facts than through data? The US Census Bureau provides lots of poverty estimates, so I took their data and mapped it over the last 27 years. I found it alarming to see that some states had a poverty rate over 20%. I clearly live in a cozy bubble. What does your state look like?
Neat map, and a noble cause. But from a visualization perspective, I don’t find it very compelling. The sizes of those disks are hard to read and compare, and there is a lot of overlap, especially along the east coast. It’s also hard to follow the trends because everything moves. This sort of map would have worked if there was a strong overall trend (gapminder-style), but there doesn’t seem to be one.
I think a more effective visualization would have been parallel bar charts, or at least bars on the map. Wouldn’t look as good, but would be far more effective to actually read the data. Or as a map, do a choropleth. I’m not generally a fan of those, but if you pick a few good, distinct values for 5-7 colors, the overall trends would be far more visible (though you would of course get the usual problem with overemphasizing large states).
This is a difficult chart to do well. You can make it pretty, or you can make it effective, but doing both is quite a challenge.
@Robert – I agree with all your points. Maybe a chart showing changing ranks would have been more effective or some form of time series plots. This was one of those late-night experiments that would’ve benefited from some preliminary work in R – but as with most of my own “projects” that I post on FlowingData, I just wanted to learn something new :)
You may be able to eliminate the overlap problem and simplify the presentation by aggregating the data by region. You could also indicate changes in trend direction with color changes in addition to size changes.
How did you handle the dramatic changes in the definition of poverty behind the data over this period?
Pingback: Blog Action Day - Poverty. « Prioritizing the Paranoias
@Dave – I didn’t. I just took what the US Census Bureau gave me. However, the last definition change was in 1981 and the map only goes back to 1980, so it’s not too bad.
great post thanks. poverty is important. Have you done any work on Dead capital?
Pingback: The New Poor - Internet Penetration Rates by Country | Pointy Haired Dilbert - Chandoo.org
This is a good chart. I found another chart that shows what kind of a mess we are in right now. It shows how this economy compares to the economies of the past depressions and recessions. Here is the link http://www.gotoguy.com/?p=455
I’m curious to know how you made this chart. What tool(s) did you use?
@jason – Take a look at my recent code release for the Walmart map:
https://flowingdata.com/2008/10/21/code-for-walmart-growth-visualization-now-available/
The implementation is pretty close to this one.
The folks at Social Explorer have an interactive map showing the % in poverty for 2000 down to the census track level:
http://www.socialexplorer.com/pub/maps/map3.aspx?&g=0
I have a static view of this map at my site: http://www.visualizingeconomics.com/2007/08/11/united-states-poverty-map/
I was looking at my home town in MO and was fascinated at how it stayed in the non-poverty zone. Then I zoomed in and saw that the poverty circle centered on the state capital, so it didn’t illustrate regions within the state.
Still, it was a cool Q&D finger exercise. I enjoyed the Walmart growth map.