The Minnesota State Demographic center created an interesting contest around visualizing the educational attainments of the state. So I thought that Dex might be useful in producing some visualizations with the data.
After looking at the data, I chose to dig into a visualization of the educational attainment of its citizens divided by Race and Sex.
The analysis produced a number of interesting visuals which I would like to share.
By far, the most succinct and effective display of this information is the Sankey diagram.
Here is a static view of the diagram:
Each color coded group to the left represents 100 percent of that population based upon race and sex. The connections are weighted by the percentage of that population group which has attained the education level represented in the color coded groups to the right. Hovering over a connection will bring up a popup which displays the exact percentage of that population which has achieve the education level to which it is connected.
Hovering over the color code boxes gives the percentage of the population within that group. It is a bit confusing, however, the boxes to the right can exceed 100% because they are comprised of some ratio of the12 groups of 100% represented to the left. Still, it conveys an effective sense of proportion in a single chart rather than traditional pie bar charts which tend to overwhelm.
You can play with a live versions of Sankey diagrams here, each gives a different insight into the data:
- Race and Gender vs Education Level By Percentage
- Race vs Education Level vs Gender By Population
- Race vs Education Level By Population
- Race vs Education Level By Percentage
- Gender vs Education By Population
When we plug population totals into the Sankey, we see a very different view of the data.
Minnesota’s primary racial demographic is white. Here is a live version of this visualization.
The dendogram is a fancy Greek word for Tree. It’s a nice way to show a lot of data at once, but you have to imagine proportion.
This is another good view of overall data connectivity, if not proportion. However, the Dex version of this view still needs some work for datasets containing this much data, as you can see below.
Digging Into The Data
There are a number of other views which are nice for digging into hierarchical data. They can also be used to convey proportion similar to Sankey, however, Dex is not currently supporting this feature yet.
The first of these views is the Starburst, named for it’s sometimes overwhelming burst of colors.
This Starburst consists of concentric circles consisting of Race, Sex, Education Level and Total Population starting inside working our way out. A live version of this view can be found here.
The partition layout is also nice for quickly digging into a population segment of interest.
You can play around with this view here.
The last of these types of views is the indented tree depicted below.
This view isn’t useful for conveying a sense of proportion, however, it is good for quickly digging into key data. You can see a live version of this view here.
Fun But Less Useful
Hive plots are great for showing connectivity without the clutter. However, Dex still needs to work out the bugs in this view, however, I include it here for fun anyhow.
These are can be used to show interconnections and are just plain fun to play with.
Traditional representations are still quite effective for telling the story. The following pie charts give a pretty good picture.
Hispanic and Latino
American Indian and Native Alaskan
As you can see, visualizations such as the Sankey can display the same or even more information than a series of traditional Pie Charts very effectively and quite succinctly.
Thinking back on this effort, I am left with a couple of thoughts. The Minnesota educational system seems to be working well for the general population. This is something of which Minnesota should be proud. However, it’s educational system is not working equally well across all demographics.
The population in Minnesota is dominated by Caucasians, more so than the national average. Could this be a contributing factor in the fact that the educational system seems to be less effective for its minorities; especially Hispanics and Latinos? Should adjustments be made to ensure greater success in the minority communities? Would this do more harm than good given the relative effectiveness of the general educational system? I don’t know, but hope that analysis such as these help smarter folks than myself make better decisions for the greater good.
To me, that is what the Dex project is all about. Pay it forward.