Week 29 took us back into the world of politics with the recent release of salaries in the Trump Administration. Politics is always a touchy subject, yet the Makeover Monday Community overcame this, resulting in many great discussions this week.

LESSON 1: USE COLORS THAT CONVEY MEANING

Political parties always have colors associated with them. It was great to see so many people using the red & blue colors that are associated with American political parties so well this week. This helped make so many vizzes consistent and easy to understand. Limiting yourself to one or two colors can often be challenging. This week’s topic helped demonstrate that it CAN be done. Next time you’re vizzing at work and you think about including color, consider why you’re doing it. Run yourself through a simple list of questions:

1. Do you need color at all?
2. Can you reduce the number of colors?
3. Are you using colors that associate meaning?
4. Can you use color for highlighting things that are important?

Of course there are many more questions you could ask yourself; this is merely a list to help you get started.

LESSON 2: FIND INSIGHTS IN THE DATA

In the favorites below, you’ll nice several times where I highlight the thinking and excellent analytical work that some of the visualisations included. Every week, the Makeover Monday data is ripe for analysis. Consider applying analytical thinking to your work. What is the data REALLY telling you? What can you add to the conversation? What are the unknown unknowns? Thinking like this will help move from being a good data analyst to a great data analyst.

LESSON 3: ACCOUNT FOR ANNUAL SALARY INCREASES TO NORMALIZE THE DATA

I kind of knew this would happen. Heck, NPR even did it on the original viz. You can’t directly compare the Obama and Trump administration salaries directly without first applying a conversion to the Obama salaries to account for annual salary increases. I increased the Obama salaries by 1.8% (the increase in the max allowable salary). However, as Ken Flerlage pointed out, this was probably not the most accurate either.

Looks like the govt approved a 2.1% raise for fed employees in 2017, but the top were limited by WH maximums. 2.1% might be better.

— Ken Flerlage (@flerlagekr) July 17, 2017

The point I’m making is that you need to put on your critical thinking hat and consider adjustments like this where needed. We saw way too many vizzes this week that inaccurately compared the salaries.

LESSON 4: SCALING DIVERGING BAR CHARTS

Lots and lots of people this week created diverging bar charts, which is a good chart choice for this particular data set. However, when using this chart type, it’s absolutely vital that both sides be scaled the same, otherwise the reader might misinterpret the data.

For example, if you create separate charts for Obama and Trump and place them together on a dashboard, you get a view like this:

The problem here is that the scales aren’t the same. Look at the 125K range. Trump is bigger than Obama, but is it really? Nope! Obama has 36 employees in that range whereas Trump has 29. To correct this, create a calculated field that pushes Obama to the left and Trump to the right.

IF [Administration]=’Obama’ THEN -[Number of Records]ELSE [Number of Records]END

Then set a custom number format on the field so they both display as positive values and you get a perfectly scaled view. Simple!

If you want the text in the middle like the first option, then you can either manually set the axis ranges or you can use reference lines. No matter your method, it’s important to look for these scaling issues. Here are a couple examples of diverging bar charts from the Community that were adjusted properly.

Ken Flerlage

Lee Gibbons

WEEKLY FAVORITES

Author: Ludovic Tavernier‏
Link: Tableau Public

What I like:

• Really like the dark and light sides; helps split the viz well
• Using a unit chart helps show the volume well
• Very clean design
• Good use of color
• Including summary stats then the details after
• Great tooltips
• Check out the highlight action; magnificent!

Author: Pablo Gomez
Link: Tableau Public

What I like:

• What a beautiful design! This could be printed as a poster.
• Using a question in the title and making sure everything in the viz relates back to the question
• Really like the sections with dividers in between
• The BANs stand out and tell me what each section is about.
• Including a single diverging bar chart in each section with summary numbers
• Consistent, effective use of color
• The long-form design encourages scrolling.
• Including instructions
• Nice tooltips

Author: Matt Hoover‏
Link: Tableau Public

What I like:

• Easy to understand design; simplicity like this is hard to do!
• The title tells us what the viz is about and the subtle green bar matches the reference line the title is talking about.
• Nice use of reference lines for context
• Jittering the people helps show the volume of people employed
• Good use of color
• Nice tooltips
• Everything was clearly thought out and has a specific purpose.

Author: Sean Hughes‏
Link: Github

What I like:

• Be sure to check out Sean’s blog post. You’ll really like the iterations he went through.
• Nice title and subtitle! I love a subtitle that summarizes the viz in text format.
• Including a reference line for equality
• Effective use of color
• Overall clean design that’s very easy to understand
• Excellent analytical work! This is what being an analyst is all about! This is the kind of work that gets you hired.

Author: Rob Groves
Link: Twitter

What I like:

• Such a beautiful display of the distributions!
• Enjoyed watching him iterate through the design
• Great annotations next to the barbell charts
• Good labeling
• Made the complex look simple; so hard to do!
• Nice title and subtitle
• I really like seeing people use other tools, like as Rob did here with R.
• Overall, this is another outstanding analytical piece.

Author: Robert Crocker
Link: Tableau Public

What I like:

• Robert asked for feedback. Received it. Acted on it. Resulting in a very good viz.
• Took a chance by looking at the salary comparisons in a different way.
• Found a story in the data; sign of a good data analyst.
• Nice labeling where necessary
• Good use of highlighting
• Using a color in the title as the legend
• Asking a question in the title to help explain what the viz is about