For week 2 we were looking at the most sought after characteristics for a mate; men vs. women. After a bit of experimentation I finally found what I wanted to do, then it was just on the to execution.
Bar Bell Charts… or DNA Charts as some call them.
Across the top I included the question, and each column included the ranking 1 through 6 and the percentage of men versus women. The men were represented by Yellow and the women by Green. Initially I have the bar doing a gradient between the two circles, but more on that later.
To construct the bar bell charts it’s fairly straight forward.
First I added the nationality (used for filtering) the question, a calculated Rank Label, and the ranking to the columns, and then a dual axis for the rows. The nationality and Rank (Number) have the headers hidden.
Next I changed the first percentage to a line and get the gender as the path. Initially, I set the gender as the color as well, which gave me the gradient bar between the two circles. On Twitter, Andy Kriebel suggested to someone else who took inspiration from my viz that they make the bar the color of the higher percentage. I found that idea really interesting, so I went off to figure out how to achieve it. After a bit of experimentation I finally got to use an EXCLUDE LOD!
The calculation, when added to the first percentage (which uses the gender as the path), excludes the gender then isolated the gender percentages and compared them. The result was a True/False value which when added to the marks card as the color allowed me to assign the appropriate color. That little change added quite a bit (IMHO) to the viz as an second indicator as to which gender had the higher response rate per ranking per question.
Next, I needed a way to filter and change the nationalities so I created a second sheet that was simply a list of the nationalities and added that to the dashboard. After adding a hover action to filter the main viz, and a bit of formatting I was all done.