This is a graph showing the decline in the number of houses being built in 2009. There are many problems I have found in this graph, and they are all issues that, while very important, can be fixed easily.

First, I think that the only way that this kind of information can be shown is between two graphs. If one is trying to show a decline in the number of homes being built, the reader should at least be shown how many homes were built 10 or 20 years ago to have a platform to compare against. We should be able to tell the decline in housing prices without the little explanation blurb in the corner. 

Second, a map is not the correct visualization for this kind of data because, of course places with a larger number of people such as New Jersey, Texas and California are going to have more homes than less populated states. The scale being used would be more justified if the same kind of data was expected in every state, but that is not the case. It isn’t surprising that Alaska, Hawaii, Maine and Rhode Island have had the biggest decline in housing because they are all not very populated. Additionally, they may have all had a low number of houses being built to begin with, but the reader would not know because there is no comparison made to a previous year. 

The color scheme of this data visualization is also a bit annoying. Why do the different numbers of homes being built need to be shown in different shades of green? Why not have a different color for each range of numbers? Why the author chose to do this baffles me. When I look to see how many homes were built in Michigan, for example, I cannot tell if it is 3,500 to 7,000 or 7,000 to 10,500. The only way I am able to tell is by comparing it to the states around it, which is not how a good data visualization should work.

To fix these problems, the author should show the data using percentages. Instead of using a map, the percentage decrease of homes being built (how many homes are built in that year / last year), can be shown on a scatter plot. The percentages can be on the y axis and the states on the x axis. If this graph would be to tedious to read, each data point could be a different state and be labeled on the graph itself. The author should also give the exact percentage change for the number of homes. The scale currently being used is 14,000+ for any state that is in dark green. The reader has no idea whether Florida had a decrease in 14,000 homes or 50,000. A different color scheme as well would make the graph more appealing and meaningful. The author could even display the information in two graphs, one for 10 years ago and another for the current data and have no colors on the states, just the percentage change in the number of houses being built. Between the two graphs, the reader would be able to make stronger comparisons for themselves. 

While the concept for this graph is a very good one, these minor changes would make it more effective and aesthetically pleasing and easy for anyone to read.

Source:

http://mattsoave.com/blog/information-visualization-one-bad-one-good/