What to connect in dot plot chart?

Member-only story

Do You Know What Is the Right Connection in a Dot Plot?

Incremental Improvements #05: Dot Plot

Weronika Gawarska-Tywonek

--

Last week I came across some unusual dot plot that The Economist did. At first, I thought this was a refreshing way of using this chart. But the longer I looked at it, the stranger it seemed. The graph presents the feelings about the removal of US troops from Afghanistan. Its primary purpose is to compare three “Yes/No” questions among three different groups — all Americans, republicans, and democrats.

My problem with this chart is its overcomplication and misleading line usage. In the dot plot, the length of the line shows the difference between two data points — and the longer it is, the bigger the difference.

This chart is quite the opposite, and a shorter line means a more polarized opinion on the subject. But even this is not always true because the chart doesn’t show all answers (have you noticed that the numbers don’t sum up to 100%?), so imagine two situations in which the dots are close to each other: 1) the two dots are close to each other in the middle of the scale, which means that around 50% of people agree and another 50% disagree, so the opinions are polarized; 2) the two dots are close to each other around the 20-percent-point, which means that the majority gave a neutral answer or no answer at all.

You can see that this way of presenting data is very confusing. So how would I change it?

Good data visualization solutions✔️

Elements that work in this chart?

  • Chart selection — dot plot is a good chart selection because it allows comparison between two groups (or even three if you add an overall result) or two timestamps (before and after comparison). I would only change the way that the points are connected.
  • Comparing different questions — using the same scale and showing questions together show a fuller picture. The audience can easily compare the diversity of views among groups (democrats, republicans, and overall) and within different aspects (comparison across the questions).

Bad data visualization choices❌

--

--

No responses yet

Write a response