Enough – #IronViz

Enough – #IronViz

If you want to go straight to the viz, click here.  It is also embedded below my ramblings.

The topic for the second IronViz feeder contest was announced on June 10th…Politics.  Initially, I was not too excited.  The election process in the U.S. is ridiculous.  It takes so much time and money and generally polarizes the country.  I wonder if our founding fathers would be shocked to hear Presidential elections cost in the billions now ($2.6B in 2012)?

We probably all have that family member who brings up politics at every gathering.  If you are like me, this is the cue to check on the kids, get a beer, or whatever, as long as it gets me the heck out of there!  I wondered if I could ever get into this topic.  After thinking about it for a while, I decided to have a little fun with this contest.  I started researching politician sex scandals.  Shockingly, I could not find a clean dataset.  There is a big list on Wikipedia, but it is mostly unstructured text.  Every decent article I found used Wikipedia page as their source.  So, I manually cleaned the Wikipedia data to get a name, year, description, party, title, etc., for each record.  The source is undoubtedly missing data and demonstrates a clear media bias in what events were reported.  I doubt politicians were perfect little angels until the early 1970s.

Click to enlarge

This is a very rough draft.  I am not particularly proud of this one, but I spent some time to clean the data and Photoshop the resignation/shame speech images of some of our elected officials, so I thought I’d show it.

Then on June 12th, the unspeakable Pulse nightclub shooting in Orlando occurred.  Shortly afterwards, I decided to explore mass shootings and Congressional inaction. I originally thought I’d submit both dashboards, but in the end I could not get back into the scandals and I didn’t finish it.

I wanted my new topic (shootings/Congress) to tell an emotional story, to present analysis of the data (without too much bias – not sure I achieved that), and to urge people to make an informed decision on election day.  I also wanted to challenge myself.  I am a huge fan of the #MakeoverMonday Challenge.  A few weeks ago, Michael Mixon put together an amazing viz of Global Arms data that inspired my IronViz dashboard.  As a result, my viz would also be mobile friendly.  I very quickly found a complete dataset of mass shootings from Mother Jones.  It only needed minor cleaning so I was off to a great start.

I set my dashboard to the max 4000 pixels tall (unless you hack the XML) by 375 pixels wide.  I used every bit of the canvas.  I decided on a grid pattern for my viz.  I was going to design in 150 x 185 pixel boxes.  This is actually quite a challenge!  I pulled out all the stops and tricks I’ve learned to make this work.  I have sparklines, slope charts, dynamically sized trellis charts, bar charts, highlight tables, custom icons, dual axis hacks, an ASCII unit chart and much more!

I recently went to an Edward Tufte training class, which means I now have sparklines on the brain so this is where I started.  I was really liking the result, but wanted to add some music to help draw out more emotion.  I have done this before on a couple of desktop dashboards.  In the past, I used the embed URL from a Youtube video and hid a webpage behind an image.  There would be icons on the dashboard to launch the video and mute the video via URL actions.  I muted the video by loading google.com.

This is the video I intended on loading behind a floating image.  However, I learned some new things about mobile devices.  I found out you cannot autoplay videos on mobile, at least on iOS, which is what I was testing on.  This makes sense because of data usage concerns.  I was designing for mobile so I was going to have to remove this trickery.  Then, I started playing with showing the video in a very small space, but it really has to be about 150 x 150 pixels to look nice.  Unfortunately, this was not going to work for the music component because a mobile device will go full screen when you launch a video.  The experience I was going for was having the piano and violin playing in the background while you explored the analysis.  So, I ended up ditching the music, but it was not a total loss because I learned more about YouTube URL parameters in the process.  I used those to specify start/end times on the three videos I did embed.

2016-06-18_22-59-00A main component I wanted to add was the 2013 Senate vote on a new assault weapons ban.  I used a little Photoshop magic on an image of the Senate chamber to float at the top of the worksheet.  I cropped the original, added a new layer, drew a selection around the U.S. flag, deleted the rest of the image, and adjusted the layer underneath to black and white.  This really made the flag stand out.

Then, I wanted to give the visual effect of the semi-circle pattern of desks on the senate floor.  I found the 2013 vote data on the NY Times so I had information on each Senator and how they voted.

I knew Adam E. McCann’s edge case could be turned into a use case here.  I used his parametric equation (detailed how-to blog here) to figure out the x/y coordinates in Excel.  There were a few things I had to change. I only needed to include angles between 90 and 270 degrees so I deleted the other rows.  I also did not have the same number of senators in each row so I started the first row of data at 90 degrees and the second row was 90+(180/n-1) and so on.  I used 180 because I only was visualizing half the circle and n represents the number of senators in the row.  This actually came together pretty quickly because I used his how-to blog before on another viz.  The trick to making concentric circles is to multiply the ID field by x/y in a calculation to create new x2/y2 coordinates.  I originally started the IDs at 1-5, but the circles were too close together in the center.  I fixed this spacing issue by making the IDs 4-8.    If you want to dive into the data behind this viz, you can download it here.  The rest is just formatting.  I added a parameter so you can easily find a particular State by adding size emphasis.  I think this use case could be expanded to look at the change in the Senate over time by party or the vote on certain issues.  Since politics isn’t my thing, let me know if you do this!

I did a fair bit of testing on my phone, since I hope this is the primary device people will use to view this dashboard. I think the mobile experience works well because you don’t see the entire viz at once and scrolling helps the story flow.  The video interaction works well too in my opinion.  I added floating text objects throughout to make the scrolling experience more user-friendly.

This was a fun challenge in designing small, tall, and mobile friendly, but the data is sobering.  I tried to be really cognizant of the sensitive nature of the data, to be respectful to the victims and families, and to not glorify the shooters by including their names or description of events.  I hope this comes across in the design and you will be inspired to be heard on voting day.

One final note.  If you are attending the Tableau Conference, you will need an absentee ballot to vote.  You can follow the important note on the bottom right corner of my viz to launch the embedded form at vote.org.  It takes about a minute to fill out and they email you an application.

Thank you for reading!  Please leave any comments or feedback below.

Interactive version below

Leave a Reply