I was on vacation for a few weeks in the UK and was not able to complete #MakeoverMonday. But I’m back now and I was excited to see the topic for this week. This post walks though my process as I completed #MakeoverMonday Week 30.
Critique of original viz: Pros
- A pictogram is friendly and is basically the shape of a barchart, so it is easy for most people to interpret.
- It is sorted to easily see the food products with the biggest and smallest water footprint.
- It is easy to make comparisons between the different food products and see that “bovine meat” (i.e., beef) has the largest water requirements, and veggies have the smallest.
- I could deduce that each water drop was representing 500 liters since the actual amounts were also on the chart. It was good that the actual amounts were listed on the chart.
- It is good that the chart indicates that the numbers represent “global averages.”
- It is a useful subheading that clarifies what is being represented.
- The source organization is listed.
Critique of original vis: Cons
- The chart junk of the scale and icons of food products is unnecessary.
- There is no link to the original report and the report itself is not specifically named.
- There is no indication that this is data coming from a 2010 report. This infographic was created in 2017 and 7 years is a long span of time. This data may not reflect the current reality.
- There is no legend to state that 1 water drop = 500 liters
- The water drops are a bad idea all around. At least use a liter symbol if you are trying to equate this icon to the measure it is representing. Or don’t use icons/a pictogram for this data.
- There is no take-away statement. What’s the “so what?” Take a stand and make some claims, tell me what I’m supposed to be getting from this information. Similarly, there is no call to action (CTA) to do anything with this information. (This is something I personally think a lot of vizzes lack and intend to write a blog post about this soon.)
- Something that could have helped people to understand the data would be to compare animal versus plant products in the chart and distinguish between the two in some way.
- This chart does not help me to understand what 15,000 liters means. Is that a lot? What concrete representation does that equate to?
- I could just turn this into a bar chart with the existing data, clean it up so there is no chart junk, and call it a day.
- I could do a univariate scatterplot with liters on x axis and types of food on y axis.
- In the plot, I could color code the food by animal products or plant products and add annotations noting that bovine meat requires much more water than any other product.
- I could do a bar-in-bar chart comparing each food product to the bovine meat to show how much more water that requires compared to any other food product.
- I could do calculations of how much more water is required (e.g., beef requires 47x more water than veggies) and chart those.
- I could do a proportional shape chart, tree map, or bubble chart where the marks differ in size for the number of liters required and the marks are color coded for animal products versus plant products.
Digging into the data: What do these numbers mean?
I opened the Excel sheet to start to look at the data before I connected to it in Tableau. I was initially confused because I saw that the measurements for total water were in cubic metres/ton. What does that mean? Does that convert to liters/kg? Maybe this is common knowledge for other people, but I am from America and I rarely quantify anything in liters or kilograms. Even though the original viz reported liters/kg, I didn’t want to make assumptions. After a few confirmation searches I did feel more confident that it was accurate to convert cubic metres/ton directly to liters/kg.
Ok, according to the data, to produce 1 kg (2.2 pounds) of beef it requires more than 15,000 liters of water. How do I wrap my brain around what it means to use 15,000 liters of water? I looked up the amount and found this website which had a nice representation of a water tank that held 10,000 liters:
And this as a representation of 5,000 liters:
Damn! That’s a lot of water for less than 2.5 pounds of beef! What is that, 5 hamburgers?!
- “I’ll just show the total water amounts, not the breakdown of blue vs. green vs. grey water. I think including that breakdown will weaken the main message because it is not central to the story I want to tell with the data, which is the difference between plant and animal products.”
- “Should I label each mark? Or just show trends? To distinguish between plant in animal products, should I create a set? Or does it make more sense to create a group?” (I did both and decided to use the groups in my analyses.)
- “What colors should I use in the chart? A red hue for animal products and a green hue for plant products? Note to self: be aware of visual impairments with red-green color deficiencies.”
- “I’ll use circles as my marks for sure, but do I use filled or not filled marks?”
- “And what’s my story anyway?”
At first I just used the food items dimension (grouped by animal and plant) and the total water measure in my charts. I made a tree map, bubble chart, bar chart, and a scatterplot. I didn’t think any of them were telling a story very effectively and I wasn’t satisfied. Then I had a conversation with my husband about what I was working on and he suggested that I compare how many calories are supplied by each type of food. That is, how many calories can we get for the water resources we are spending? What is our ROI?
Yup, that did the trick and told a much better story. My scatterplot showed how many calories each food type supplied and how much water was required to produce those calories.
At this point I also tried adding some reference lines to the chart to create a 2 x 2 matrix and distinguish which items were above and below average in how much water they required and how many calories they supplied. The grains, pulses, and oil crops were the clear winners as they were in the high-calorie/low-water quadrant of the matrix. I also tried the matrix with the median values. I wasn’t that happy with the reference lines though. I also tried to add just a diagonal line from the 0, 0 point. But hacking Tableau to add a diagonal line required that both axes were 0 to 16,000. And I didn’t want that, so I scratched that idea too.
I added annotations to help certain points stand out, changed the colors to teal and orange, and worked on formatting. My viz is not complete until I run through the Tableau best practices for an effective dashboard:
1) Are my instructions to the viewer clear (if I have any)?
2) Did I format the tooltips for comprehension and consistency?
3) Are my colors on point?
4) Does the arrangement of the charts make sense?
5) Are my fonts appropriate and consistent?
I learned these guidelines when I recently completed the Tableau eLearning modules for Desktop I (Beginner) and Desktop II (Intermediate). (There were super helpful modules and I’m going to write a blog post reviewing them in the near future. Thanks Tableau!)
So at this point I showed the viz to my husband and he said, “Well, that’s cool and interesting. But what about the amount of protein that each food supplies?” Yes, good question! So I quickly made that viz too and it definitely supported my story.
The final section I wanted to add was the CTA and the references. I wanted everything to be a hyperlink (e.g., @dreamsofdata to go to my Twitter account; the Source to link to the actual 2010 report; etc.) but I hadn’t added hyperlinks to a dashboard before so I had to look up how to do it. I followed this blog post from InterWorks (thanks InterWorks!) and it was really helpful. At first I thought I only needed to create a Worksheet Action but when my links weren’t working on the dashboard I realized it was a two-step process: 1) Create the Worksheet URL Action and 2) Create a Dashboard URL Action using the worksheet. At least, that is my understanding now of how that works, because once I did that all the links in the dashboard did finally work. I used all tiled containers and that affected the layout at the bottom of the viz. I don’t like how it looks at the bottom, the spacing is all off, but that will have to do for now!
Since it is a long form viz I needed to follow the advice of a blog post from the Data Duo (thanks Data Duo!) and take screen captures of different sections of the viz and then combine them in Photoshop. Note: Combining images in Adobe Illustrator is a fool’s errand. The images lose their resolution when you zoom in if you use Illustrator. Photoshop is the way to go. I found it very easy to place the image files and then export the final image.
Finally, here is the final viz! It is also posted on Tableau Public.
Note: Please be aware that all of this took me WAY longer than the #MakeoverMonday 1 hour allotment! But I’m sure I’ll be faster next time because I learned so much! Thanks Makeover Monday! 🙂