« Enhanced tables, and supercharged spreadsheets with in-cell tech | Main | Scorched by the heat in Arizona »


Jean-Pierre Tanguay

I find the new version worse since I don't understand it. Anyone else? Looks like Tetris!


Plotting per head of population would also make more sense an increase the disparity, and just use a bar chart.


I have to disagree Jean-Pierre.
I didn't understand the old version, even after the guy explained it.
I only understood it when I saw the new version.

In the old version you can't easily tell that California is over 50% of sales. Even a rubbish pie chart would have done this!


It took me a long time to realize that the outline inside the blue bar was meant to represent the comparative measure.

I share Jean-Pierre's confusion over the tetris-style blocks used to build the comparative regions - can you explain the thought process behind using this approach instead of a simple stack?


I was picturing something more like this, for example:



jlbriggs & several others: Your version is also fine. The point of this post is getting out of the plot-raw-data setting; look at the data, figure out what is an interesting thing to say, and then design your graphic to highlight that point. Your bar chart brings out the message too. Recently, I have been experimenting with ways to achieve a better balance between precision and engagement. My chart wasn't trying to convey precision at the region by region level.
Thanks for bringing up this discussion.


Also, thanks to the reader who pointed out the "moire" typo, corrected. It's ironic that I had gone to Google to check the spelling while writing the post. Since enough people are misspelling it, Google's auto-complete suggests "moivre pattern" is correct. That is one of the downsides of count-based machine learning approaches; they assume common patterns to be true even when they may be inaccurate.


I truly don't see a problem with the original chart. While not the best form it's not really hard to decipher, moire' patterns aside. The numbers serve to eliminate area confusion for the upper two boxes, and the rest of the boxes are very distinct, size wise. California, shaded as a subset of "West", rather than confusing, is easy to see as a callout of a specific territory in the west (note that it has the same pattern in the background)... (Admittedly, I haven't read the article, but that's what I infer from the chart at first glance).


@kaiser - "look at the data, figure out what is an interesting thing to say, and then design your graphic to highlight that point"

This is a great point, and one that should be printed on the wall in any room where people do this work :)

I feel like this particular experiment was not entirely successful though. Visually, I found the remake more confusing than the original, with some of the same problems.

The lack of precision isn't the problem at all - the odd shapes of the regions were very distractingly confusing as I attempted to discern meaning from what apparently had none.

In addition, the transposing of a second, slightly different version of the overall shape on top of the California area could have used some explanation or a stronger visual connection.

I think engagement is a worthy goal, but I think if it means "try to figure this out" then it's going the wrong way.


The comments to this entry are closed.


Link to Principal Analytics Prep

See our curriculum, instructors. Apply.
Kaiser Fung. Business analytics and data visualization expert. Author and Speaker.
Visit my website. Follow my Twitter. See my articles at Daily Beast, 538, HBR.

See my Youtube and Flickr.

Book Blog

Link to junkcharts

Graphics design by Amanda Lee

The Read

Keep in Touch

follow me on Twitter