MakeoverMonday: The Bermuda Population

Bermuda

I am going SUPER-SIMPLE with MakeoverMonday for the foreseeable future! There are 2 reasons for this. Firstly, I’m finishing my book and also going on family vacation.

More importantly, it’s a response to comments in Andy K’s post about how many people are downloading the Makeover datasets but not publishing vizzes.

This one sums them up:

Ouch.

For people who feel intimidated about posting their work: I am sorry. Our intention with MakeoverMonday is to encourage everyone to share their work, in order to help them elevate themselves, whatever their skill level. I want to see people share the vizzes that DO take only an hour. Or less. This week’s took me 20 minutes. It’s a straightforward area chart with the colour scheme copied from the original.

If someone wants to spend hours on their Makeovers then that’s FANTASTIC. If people want to take the datasets and spend hours crafting amazing pieces of work, then I applaud that. I love seeing the amazing styles and approaches people take.

One look at the MakeoverMonday Pinterest board and it does seem to an arms race towards who can do the most elaborate work. Pam G said she stopped doing MakeoverMonday because “so many posts were all about the cool graphics being used and not so much about the charts themselves.”

So with that, here’s my plan for the next few months: I’m going to keep it simple and focus on the charts themselves. If you want to join me, then please do! I recommend you read Chris Love’s Keep It Simple post for more reinforcement of this idea.

Why did I choose an area chart this week?

The original viz was titled “Bermuda Population Growth“. I like unit charts, but this one made it hard to compare each year being shown. Taking the title as my cue, I wanted to make it easier to see the growth. I decided seeing total population was more important than the actual Female/Male breakdown. Therefore I chose an area chart: it shows the total at the expense of accuracy of the gender breakdown.

I’ll be keeping it simple for the next few weeks. Who else wants to join me?

 

Scribbles and lines: what does iteration look like?

Capital Punishment in the US

What’s the goal of analysing data? It’s about finding insight, and communicating that insight  in an optimal, engaging way.  

How do you reach that goal?

You don’t get there by first deciding which chart types you want and then adding data to them. Saying “I will start with a bar chart” and adding data to it  is the equivalent of trying to create a masterpiece with a paint-by-numbers set.

You do it by starting off with freeform exploration, improvisation, and shifting perspectives. We might try out 100 different views and keep only one. That single output will be powerful, but you’ll know it’s right because all the other views provide small contextual validations that you gained on the journey to the final output.

What does this look like? How about this squiggle?

squiggle narrow

The start is fast, messy and unpredictable. You’re failing fast and moving on. Over time you focus on the insights you’ve found and gradually hone in on the best articulation of the data.

What does the squiggle look like with real data? I got a chance to find out during MakeoverMonday, week 29, when we looked at data on executions in the USA. For a project being run by Tableau’s research team, I captured a screenshot of every view I made in Tableau (this was automated!). My analysis took 90 minutes. In that time, I built 302 different views of the data. The final dashboard is shown at the top of this post.

Here’s what the exploration looked like:

iteration 300ms

You can see all phases of the squiggle in the animation.

  • Rapid exploration at the start. During this phase I discovered the data had a similar shape to Simon Scarr’s Iraq’s Bloody Toll and decided that, based on that insight, I could base my final output on the design of that infographic
  • In the middle phase, there are periods of formatting the time chart interspersed with exploring afresh. I’m not restricted into doing one thing or another. As I play with the data, I get new ideas to go try out. I’m honing in on something, but still exploring.
  • Finally I spend a big chunk of time finessing the main chart. This is the phase when I’m trying to find the absolute best way to represent the data.

I created tables, bubbles, area charts, pie charts, sideways charts, ugly charts, and more. I threw most of them away. What the animation shows is what happens when an analyst is in the flow with their data.

I found this week’s MakeoverMonday fascinating. I have written many times before about exploration and iteration (e.g. here and here) but this GIF is the first time I’ve been able to truly see what exploration looks like. It’s the first time I’ve been able to validate that exploration can look messy from the outside but is actually hugely productive for the person doing it.

Footnotes:

  1. The Squiggle is based on Damien Newman’s Design Squiggle. Newman applied the squiggle to design.
  2. The screen captures from Tableau were captured automatically. Only changes to the views themselves are captured. Titles, annotations, and laying out dashboards aren’t captured.

MakeoverMonday: The Next to die (capital punishment in the US)

 

Capital Punishment in the US

Anyone who follows my blog will immediately recognise my inspiration for this week’s makeover: Simon Scarr’s incredible Iraq’s Bloody Toll infographic from the South China Morning Post in 2011. I’ve written and spoken about this post many times in recent years.

The original source for this week’s makeover (“The Next to Die”) is a great project. It puts a new perspective on this topic, focusing not on the past but on the future. One thing it doesn’t emphasise, which I learnt by examining the dataset, is how the number of executions is dropping across the US.

Given I learnt that the number of executions is dropping, let’s go back to my makeover. I chose red bars and made them face down: evoking a smear of blood. But if I my biggest learning point was that numbers are going down, surely orienting the bars in a normal way (ie up) would make it clearer?

Well, here you go:

Capital Punishment pointing up

Regular readers will, of course, recognise this is what I also did with the Iraq’s Bloody Toll infographic. Here’s the Iraq’s Bloody Toll and my Capital Punishment story told in both ways.

side by side

 

The perils of bubbles

Another day, and another tragic event. The events in Nice horrified us all. One excellent response is to reinforce that, should this turn out to be a terrorist attack, the individual does not represent all Muslims. The number of terrorist-minded Muslims is tiny. We can make that point using data.

This afternoon I saw the following chart tweeted by Ian Bremmer, which makes this point:

The point being made is great, but look at those circles: if you put that ISIS circle inside the Muslim circle, it’s kinda big. So I took the data from that chart, and redrew it in Tableau. Turns out, the circles were the wrong size. Here’s what the size really look like, if you draw the circles so that their AREA represents the value:

Muslim

 

Can you see the circles for Al Qaeda, ISIS, or the Taliban? They’re up there in the top right. Tiny, aren’t they?

The lesson here is that, if you’re going to use circles, size them according to the area. The people behind the original chart, MIIM Design, were trying to make a valid point, but the circle size misrepresented the sizes, which could cause confusion.

Note: I used the same numbers MIIM used in their chart, taking the upper estimate of each category. I have not done research to check the validity of these numbers. My goal is to make a point about circle sizes, not a political point about the size of different religious or terrorist groups.

Update: I replaced the original image with one with a new title. One commenter suggested, rightly, that I might have implied “Muslim” is a terrorist organisation. That absolutely was not my intention or my belief.

 

MakeoverMonday: Orlando shooting

Download the v10 workbook here
Download the v10 workbook here

As shooting after shooting occurs in the USA, this week we’re looking at politicians’ responses to these shootings. Specifically, the first response statements of members of congress following the tragic Orlando shooting a few weeks ago.

I went through quite the process this week.

progress

I started with a side-by-side bar to examine the data. I don’t like side-by-side bars, so changed to single row with two circles. That made it too hard to see the difference so I joined them as a dumbbell. And then I thought that the GAP between the parties was most interesting.

For a long time my makeover was just the gap:

Dashboard 3

… but this was unsatisfying: the gap is interesting, but what does a 40% gap represent? 90%-50% or 40%-0%? Knowing whether 0% or 50% of a party mentioned a topic is more important than just the gap.

And so I tried the pies. I quite like them. I could have just shown the number, or a bar, or a sized circle, but I thought I’d try a pie. The problem with pie and difference bars: you still only show percentages, not the actual values. However, if you DO show numbers via circle size, you still have a problem: there aren’t equal numbers of Democrats and Republicans in Congress so the numbers don’t relate directly.

percent diff

In the circle chart above, looks like way more GOP condemned terrorism than the DEMs. It’s actually a 20% difference. Maybe a bar would work? This is one where you can go round in circles balancing most efficient visual with the most engaging. I went round in circles before running out of time and sticking with the pie. It’s not perfect!

Pie, circle, bar?
Pie, circle, bar?

The original chart this week is unconventional but does convey the point well.

What’s unclear is just how the slices are sized. By area? Radius? Diameter? The choice of semi-circle is abritrary. Despite that, it is easy to see the difference, even if accuracy is sacrificed.

 

MakeoverMonday: What’s the best US state to raise a child in?

 

50 states

This week we’re looking at quality-of-life metrics in the USA. The original article had a series of 5 maps showing how each state is ranked according to various measures.

This is an interesting challenge: how can you compare ranks among multiple metrics for so many states? My normal approach to MakeoverMonday is to make static charts, but this is one dataset where the only way to provide insight is to use interactivity.

Whenever you hover over a bar, or the map, every area of the dashboard highlights to show the rank of that state in each category. For example, it makes it easy to see that while New Hampshire is ranked number 1 for Family and Community, it’s weak in Health:

Best states to raise a child New Hampshire

You can download the workbook here (it’s built in Tableau v10 beta 4)