Tag: edward tufte

A Tale of Two Charts

On Tuesday, I’d put up a post about Claudia Goldin having been awarded the Nobel Prize.

The original post did not have a chart at the start, it had a verbal description of a table from a rather-boring-to-read-but-oh-so-important PDF. A friend to whom I had sent the post for review suggested that a chart might do a better job instead. He was right, of course, and so I went ahead and added a chart. Here it is:

Here is the same data, but shown in a different way:

Which chart tells a better story?

Chart design and visualization is a subject that has long fascinated me. How long? 16 years, to be precise. One of my managers, back when I was a corporate employee, had taken a chart I had created, and shared it with the whole team. Which in and of itself was fine – but he had then also recreated my chart, but with his ideas for how to make it better. Placed next to each other, they showed pretty much the same thing – but in the same way that a stick figure and the statue of David are the same thing.

Now, I am nowhere close to being an expert when it comes to charts, but ever since that day, I have been fascinated by how powerful a well designed chart can be. Powerful in terms of it being a clear way to tell a story, it being a simple way to tell a story – and how much detail can be fitted into a single chart.

But for today’s post, let us focus on clarity and simplicity. Which of these two charts tells a simple, clear story? Would it be the one that I had in the post, or the second one?

  1. How do I want to group the data? Categories (men, women and total) on the axis, or time on the axis?
    What story do I want to tell? Do I want to emphasize how much the data has changed over time, or do I want to make the point that women’s participation has always been lower than that of men (and of course therefore the total)?
  2. Should I use bar charts, or should I use a line chart?
    The answer to this question depends on the answer to the first question. If I want to emphasize how much the data has changed over time, it should be a line chart. If I want to emphasize the fact that women’s participation is always lower, I should use a grouped bar chart. But what if the line chart actually makes this point even more clearly?
  3. Which color scheme is better?
    Again, do I want to emphasize the difference between the categories? Or do I wish to emphasize the change over time. In my opinion, the second chart looks cleaner, and tells a simpler story.
  4. Note that the second chart doesn’t have a reference scale, but the first one does (the “x-axis” with the numbers running from 0 to 60).
    And therefore, to me, the second chart is better in this regard. It has less information that needs to be processed by the viewer, and they can reach a conclusion rather more quickly.
  5. Do data labels help or hinder?
    In my opinion, they help. In fact, well designed data labels take away the need for a reference scale, and so can help make the chart look cleaner.
  6. What about chart titles?
    Ideally, each chart should have a title, that answers the TMKK question. I didn’t include titles in either case because it was a blog post and not a presentation – but I know I would have got a metaphorical rap on the knuckles. When you create a chart, also create a useful title is a good rule to follow. So follow what I say, not what I do!

There are lots of good resources to follow online if you want to get better at using data visualization to tell better stories. Stephen Few, Edward Tufte, Jan Schultink and Nancy Duarte are just four names to get you started – but these four will teach you much more than you will need. And if you poke around in the alleys and bylanes of this part of the internet, you will be able to find so much more.

But please, do learn the art of data visualization, it is quite a powerful tool to have in your arsenal. And if you think the first chart is better than the second one, please let me know why – because I think the second chart wins hands down.

An Economist Talks About PowerPoint

Not me, I hasten to add, but Tim Taylor.

He published some days ago a wonderful little blogpost, commemorating the 20th anniversary of an essay called “The Cognitive Style of PowerPoint: Pitching Out Corrupts Within“. It’s not a short essay, at 25 pages, but it is remarkably well written, full of lovely little anecdotes.

For example, did you know that Richard Feynman has a rant about bullet points?

“Then we learned about “bullets”—little black circles in front of phrases that were supposed to summarize things. There was one after another of these little goddamn bullets in our briefing books and on slides”

https://www.inf.ed.ac.uk/teaching/courses/pi/2016_2017/phil/tufte-powerpoint.pdf

Both the essay and the blogpost are full of lovely little anecdotes and points such as these. I especially loved Tim’s concluding paragraph:

I just think we would all be better off with slide presentations that have fewer bullet points, fewer pages jam-packed with words, and fewer detailed numerical tables that can’t be read by anyone more than 30 feet away. Presentations impose costs of time and attention on others. In successful presentations, your attention is attracted, rather than taxed, and the entire time feels well-spent.

https://conversableeconomist.com/2023/06/01/how-powerpoint-and-other-slide-presentations-can-inhibit-thinking/

Having sat through my fair share of presentations in both the corporate world and in the world of academia, I can attest to the fact that in “unsuccessful” presentations, one’s attention is taxed. Mine has been taxed far too often at far too onerous rates The applicability of the Laffer Curve to the real world might remain a matter of debate, but I have empirical evidence about its relevance to sitting through PowerPoint presentations.

Do read both, Tim’s blogpost and Tufte’s essay – and here are my additions to both of their suggestions:

  1. Consider doing away with presentations entirely as often as you can. You can replace it, as Tufte’s anecdote about Gerstner suggests, with a conversation, or you can go Amazon style and have people write brief notes instead. But avoid presentations when possible.
  2. Do. Not. Read. Out. The. Slide.
    I am a person capable of reading what is in front of me. Not everybody in your audience might be able to do so at all times, of course, but working on the assumption that they are, please don’t read out the damn slide.
  3. Answer the “So What?” question. The title of the slide should not just describe what is in the rest of the slide, but it should also answer the question “So What?”. Gokul Rajaram’s LinkedIn post, which Tim links to, speaks about titles in the second bullet point: “The title does most of the heavy lifting, which means it cannot be passive. It must be action oriented. Eg: not “Subscriber retention” but “Subscribers continue to be retained strongly”. even better “Net revenue retention continues to be > 100%”.”
    I’d go a step beyond and say it could be something like “Net Revenue Retention targets continue to be exceeded at >100% Levels”. Or “need to remain at”, or “get even better than” – or whatever needs to be done as a consequence of the data shown in the slide.
  4. A presentation is a complement, not a substitute. It is there to help you do your job better, it is not there to do the job instead of you. Use it as a reference to help you deliver your talk later. Use it as an inspiration for you to take off on whatever point you want to make. Use it to convey a feeling, a thought, or an emotion (and this is why images are better than words), but don’t use it as a way to for you to be lazy on stage. Quite the opposite, actually.
  5. Make sure that there is a double thank you moment at the end of your presentation. And I should be more specific – make sure that you thank the audience for having listened to you, and make sure that they end up thanking you for having delivered the presentation. Not for finally ending it.