Tracking tigers
Finding dots

Rip tide

Nyt_autoAs if a rip tide has torn through, this chart drowned the data in the depths of colors, scales and graffiti.

Scales - every chart has its own scale, rendering it impossible to read across charts.

Colors - every brand has its own color.  This feature is redundant since the data labels already serve the purpose of linking the two columns of charts.

Compression - it is impossible to judge the growth or decline of individual companies, especially since only the current market share is provided.

If anyone has access to the data, please send them over so we can remake this chart.  Or just send in your charts and I'll put them up here.

Reference: "Now Playing in Europe: The Future of Detriot", New York Times, Oct 28 2006.


Feed You can follow this conversation by subscribing to the comment feed for this post.

Jon Peltier

I don't agree that this chart is so terrible.

"Scales - every chart has its own scale, rendering it impossible to read across charts."

The charts do share a common scale, which is simply not explicitly shown. Judging from the thickness and label at the end of each chart, the values are consistent with those in other charts. Also there seems to be a horizontal white line every 10%.

"Compression - it is impossible to judge the growth or decline of individual companies, especially since only the current market share is provided."

While it's true that the charts are compressed, you can still see the bumps in each company's share. And while a factor of two difference in a company's share is important to the company, to me the difference between 20% and 40% (i.e., a significant player) should be more pronounced than that between 1% and 2% (a minor participant).

At least it's not a stacked area chart!

Johnny Won

I disagree with the first post, it is a bad graph because its only functional in showing the progression across the US or across Europe. The graph is only useful if you look at Ford or GM's decline.

You say its relieving its not a stacked chart but a stacked chart would really help this graph to show how these firms expanded and contracted at the expense of others. Maybe a stacked chart for only now and a certain year in the past would make things easier to read?

The post is right about one thing, it's almost impossible to discern a comparision versus continents. But maybe we're missing the point, maybe the point is that Europe's skewed auto groups are what US is emerging to look like.


I agree with Peltier that this is not a bad chart, and won't repeat his points.

With each chart separate, but on a common scale, it's easy -- and just as easy -- to compare Ford US with Toyota US or Ford US with Ford Europe.

I'm glad it wasn't a stacked area chart OR two line graphs with about 10 lines each.

Daniel Staal

Personally, I think this chart is just too noisy to be well understood. A selection of pie charts would have lost some historical change data, but shown the point better.

I think I'd go with a pie chart of 'before' and 'after' for both markets, and a line graph with the three-four most interesting companies over the whole period. It would look like more data, but you'd actually be able to comprend it all.

Xan Gregg

I've scraped the data for this graph using a nifty Mac app GraphClick, and it's available in the CSV file listed below. I've also added a quick mock-up with sorted histograms for 1990, 1998 and 2006.

I think the original graph contains a lot of information, which is useful for static exploration. (I don't think anyone pointed out that the color scheme is organized by continent of manufacturer.) However, the text focused on one particular detail, how US market share is becoming more like EU marketshare, which I think is best shown with histograms.

Incidentally, looking at the histograms, you see that EU marketshare distribution is becoming less flat (more like US). So perhaps the two distributions are converging to a middle ground.

Jon Peltier

No offense, but I still prefer the original.

Johnny -

You can see one company's growth at the expense of others easier with the current chart than if they were all stacked together. In a stacked area chart, you have to be trying to compare angles and slopes, and we're just not very good at it.

Daniel -

By the same token, pie charts force you to evalueat angles. If one wedge pushes the others around the pie, it becomes a struggle to compare the same pie wedge on two pies but at different positions around the pie.

Xaan -

I was going to say "Good job", your histograms show another interesting cut at the data (even though I prefer the original). However, I changed my mind when I read the caveat that bars are not comparable across histograms. This loses the year to year correlation, and the changing order of the bars also makes the user work a little harder for comprehension.

The comments to this entry are closed.