MOOC Week 4: Unemployment in the US 1960-2012

US unemployment data has been visualized over and over and over in the last year because of the election and recession. Not only is the topic important but the US Bureau of Labor makes the data easily available. It was not surprising that Alberto Cairo chose it for our week 4 exercise of the MOOC, but it also meant that the bar was already set high. The data set, coming through the Guardian Data Blog, was comprehensive if not diversified. My experience in the prior weeks led me to conclude that I should not spend too much time exploring the data and focus on executing a neat infographic. This turned out to be the wrong conclusion.

After a quick survey, the data seemed to lack in depth: providing four years of unemployment data immediately begs the question of what happened before those four years, especially since an economic meltdown had happened in the months prior. So I decided that my interactive graph would show 52 years of data (and it is available). I envisioned a basic visualization with multiple graphs that would allow the user to explore all the data and cut it in multiple ways - by state and by date. Here's how it looks. I chose a grey color theme to keep the focus on the data-related colors, but it seems a bit dull although it might not be the colors as much as the fonts and design.

Unemployment US FG

Data linked to population call for a cartogram because it is not based on the territory, contrarily to weather, agriculture and climate for instance. It seemed also natural to offer the option of comparing any set of states so I took my inspiration from the Google Data Explorer. The states are hidden in a drawer on the right and I'm not quite sure that the users would notice this.

Unemployment US FG2

The closest I came from providing a narrative is in creating ready-made periods, on the top right. In the example below, the user has selected the results of the 2012 election and the lines, bars and states are colored according to each party. I'm not sure how it would really work on the map since I would end up with two color shading, making hard to compare levels of unemployment across parties.

Unemployment US FG5

It seemed interesting to see the influence of the political system on the economic performance. It is unclear if presidents or even Congress can have an impact in the short run, but it is something that partisans constantly bring up. I thought that the users would like to know what happened while their party was in power.

Unemployment US FG3

Two variations in the next version. The most obvious is the circles representing jobs created and lost in each state. While circles are not very precise, I find it an interesting way to show four data points at once: losses, gains, differential and overall size. Also, it is unusual and eye-catching. Some users might need a minute to understand it, but it pays off.

The second variation is in the bottom graph where the data is now relative to the average national rate. It is less radically different than I expected. In fact, I'm not quite sure that it adds enough insight to justify the interactivity.

Unemployment US FG4

I went heavy on the interactivity and customization but I mostly missed the story. I did not look for hence I did not find one. My graph comes across as a source of raw data for people who are interested in economic data, but it does not draw in anyone new. Yet another lesson learned from this MOOC.

Week 3: Aid Transparency

The third week's assignment was right up my alley: aid transparency. It is even more disappointing then that I was not able to complete something worthy. The source data comes from the Transparency Index of Publish What You Fund and takes the form of a ranking of aid agencies according to their transparency score. I expected the students to visualize this ranking, making more apparent the comparison between agencies, highlighting their strengths and weaknesses. I decided to try something different by visualizing the indicators themselves. I thought that it could be a nice way of explaining data transparency by detailing how it is measured. Here is my entry.

Aid Transparency Graph FG

The assignment was for an interactive visualization, so the image below shows some of the interactivity that could be prompted by the users, namely a series of definitions and the capacity to select a subset of indicators.

Aid Transparency Graph FG2

Using Adobe InDesign, means drawing every data point and this took way longer than it should, mainly because it does not add so much to the graph to have very precise data. Most people just give a quick look and are mostly interested in the ways in which the data is visualized, more than the result of the visualization. This point was driven home by the multiple hand drawn sketches of fellow students that did not approach accuracy, but that sometimes conveyed clearly enough their concept. The next week, I wouldn't be caught.

Given the call for a narrative, I spent some time finding and writing some analysis. Again, this is not something that I expect anyone to read -- at least, no one has ever commented on the text -- so it does not seem like a good investment of time.

Aid Transparency Graph FG3

In general, I like to use colors to visually group things, but more than once my audience has been more interested to see things grouped by subcategories, so that's what I have done here with the three categories of transparency. See how the colors are grouped. I have to say that it worked better than I had anticipated.

Aid Transparency Graph FG32

This slide shows only the improvement of each indicator. The main message is that all indicators have improved over the last year, although some much more than others.

In the end, I did not get to produce something of the quality I was hoping for. I picked my colors at random, I did not include a legend, I did not push the analysis, etc. But the week was over and another assignment was waiting. The point is not so much to create a perfect infographic, but to learn and this goal was already achieved.