The Times did a great job making this graphic (this snapshot is just the top half):
A lot of information is packed into a small space. It's easy to compose the story in our heads. For example, Lee Chong Wai, the Malaysian badminton silver medalist, was suspended for doping for a short time during 2015, and he was second twice before the doping incident.
They sorted the athletes according to the recency of the latest suspension. This is very smart as it helps make the chart readable. Other common ordering such as alphabetically by last name, by sport, by age, and by number of medals will result in a bit of a mess.
I'm curious about the athletes who also had doping suspensions but did not win any medals in 2016.
Reader Conor H. sent in this daily medals table at the NBC website:
He commented that the bars are not quite the right lengths. So even though China and Russia both won five total medals that day, the bar for China is slightly shorter.
One issue with the stacked bar chart is that the reader's attention is drawn to the components rather that the whole. However, as is this case, the most important statistic is the total number of medals.
The other day, a chart about the age distribution of Olympic athletes caught my attention. I found the chart on Google but didn't bookmark it and now I couldn't retrieve it. From my mind's eye, the chart looks like this:
This chart has the form of a stacked bar chart but it really isn't. The data embedded in each bar segment aren't proportions; rather, they are counts of athletes along a standardized age scale. For example, the very long bar segment on the right side of the bar for alpine skiing does not indicate a large proportion of athletes in that 30-50 age group; it's the opposite: that part of the distribution is sparse, with an outlier at age 50.
The easiest way to understand this chart is to transform it to histograms.
In a histogram, the counts for different age groups are encoded in the heights of the columns. Instead, encode the counts in a color scale so that taller columns map to darker shades of blue. Then, collapse the columns to the same heights. Each stacked bar chart is really a collapsed histogram.
The stacked bar chart reminds me of boxplots that are loved by statisticians.
In a boxplot, the box contains the middle 50% of the athletes in each sport (this directly maps to the dark blue bar segments from the chart above). Outlier values are plotted individually, which gives a bit more information about the sparsity of certain bar segments, such as the right side of alpine skiing.
The stacked bar chart can be considered a nicer-looking version of the boxplot.
This chart looks simple and harmless but I find it disarming.
I usually love the cheeky titles in the Economist but this title is very destructive to the data visualization. The chart has nothing to do with credit scores. In fact, credit scoring is associated with consumers while countries have credit ratings.
Also, I am not a fan of the Economist way of labeling negative axes. The negative sign situated between 0 and 1 looks like a stray hyphen that the editor missed.
A line chart would have brought out the pattern more sharply:
The pairing of columns in the original chart signals that readers should compare GDP growth to population growth. A good point, since GDP scales with population.
Controlling for population size can be accomplished by the per-capita GDP growth rate.
The last three years are clearly different. By this metric, different in a good way.
This chart creates a problem for the journalist. The article is about the deal to "save" Puerto Rico which some has criticized as colonial. Presumably, the territory has been in dire straits. There are plenty of metrics to illustrate this point but GDP growth is not one of them.
Via Twitter, Bart S (@BartSchuijt) sent me to this TechCrunch article, which contains several uninspiring charts.
The most disturbing one is this:
There is a classic Tufte class here: only five numbers and yet the chart is so confusing. And yes, they reversed the axis. Lower means higher "app abandonment" and higher means lower "app abandonment". The co-existence of the data labels, gridlines, and axis labels increases processing time without adding information.
A simple column chart shows there is almost nothing going on:
I suspect that if they were to break the data down by months and weeks, it would be clear that the fluctuations are meaningless.
The graphical scaffolding, or what Tufte calls the non-data ink, should provide context to help readers understand the data. This is not the case here.
Worse, the context needed to interpret "app abandonment" is sorely missing.
You might argue with me. Isn't it clear from the chart title? And doesn't the subtitle provide the details of how app abandonment is measured? It says "% of users who abandon an app after one use".
That definition is an emperor with no clothes.
The five numbers could not really be percentages of users because every user has many apps. So one may abandon app A after a single use, but one may also have used app B four times, and app C 12 times, etc.
It seems possible that they are counting user-app pairs. This measure is much harder to interpret because every user is represented as many times as he/she has apps. The more apps he/she has, the more times he/she is represented in the data.
And be careful, we are not counting all apps either. For the definition to make sense, we should be counting only apps that are downloaded in the given year. This means that lurking behind the time series is the proportion of "new" apps and how this evolved over time. It is also murky what "new" means. I am aware that many app developers keep forcing users to download upgraded apps - sometimes, I think these are counted when developers publish app download statistics. Obviously, someone who upgrades an app is likely to be an active user. So whether upgrades or later versions of the same app are counted or not is another question.
Finally, what constitutes a "use"?
From a Trifecta perspective, this is a Type DV chart. There are obvious visual flaws but the real issue is the missing context related to how the metric is defined.
My summer course on analytical methods is already at the midway point. I was doing some research on recommendation systems the other day, and came across the following chart:
Ouch. This is from the Park, et. al. (2012) survey of research papers on this subject. It's the 21st century, people. The column chart copies the older-generation Excel design made infamous by Tufte, and since abandoned. Looking more closely, I suspect that the chart was hand-crafted, not made in Excel.
There are several challenges of reading this chart.
The gaps between columns are narrower than the columns. Only in the last two years do the eight categories all count. So a key task is to learn which column stands for which type of application. Having one's eyes flip back and forth between the columns and the legend below the chart is a big hassle. As readers, we tend to learn a short cut, which is to memorize the order of the categories (first column is book, second column is document, etc.). The incorrect width of zero-valued columns thwarts this simple strategy.
The designer creates another obstacle by sorting the categories alphabetically. Shopping and movies are two of the most important applications and that message is buried.
The key to cleaning up this graphic is to bring the visual design closer to the question being addressed. The question of the chart is how interest in various applications has changed over time.
The answer is that applications are getting more diversified (the rise of the Other), and that Documents, Shopping and Movie applications were growing while research on Image, Music, TV Program and Book stagnated during the study period.
Chris Y. asked how to read this BBC Sports graphic via Twitter:
These are managers of British football (i.e. soccer) teams. Listed are some of the worst tenures of some managers. But what do the numbers mean?
The character "V" holds the key. When I first read the chart title, I wonder why managers are opposed to win percentages. Also, the legend at the bottom right confuses me. Did they mean "W" when they printed "V"? "Games W%" seems like a shorthand for winning percentage.
After looking up John Carver's not-so-impressive record, I learned that the left column are total number of matches managed and the right column is the winning percentage expressed as a number between 0 and 100.
I think even the designer got confused by those scales. Witness the little bar charts in the middle:
The two numbers are treated as if they are on the same scale. The left column is assumed to be the number of matches won while the right column is treated as the number of matches lost (or vice versa). Under this interpretation, the bar charts would depict the winning percentages. Let me fix the data:
While these managers have compiled similar losing records on a relative basis, some of them lasted longer than others. The following chart brings out the difference in tenure while keeping the winning percentages: (I have re-sorted the managers.)
When they finally got the sack, they reached the end of the line.