« As good as Bolt | Main | Mountain, molehill »


Kevin Henry

These are good points, but I think the biggest flaw with this Times piece is that "Normal" is given a point value, and every year is considered to be either above or below average. It would be much more sensible to define a range of normal values (for example, plus or minus one standard deviation from the average), and consider the number of years that the rainfall fell outside those values.

Graphically, instead of presenting a series of wildly contrasting spikes, those values in the normal range might be rendered in gray, leaving the color only to the extreme values.


I think the biggest problem is something you haven't mentioned at all: While the article itself speaks about recent developments in the last few years, the data stops at 1992! That's 20 years of climate data left out, and crucially, that includes all of the data that is supposed to bring about the conclusion of the text... small wonder the text and the chart do not tell the same story, the chart doesn't know all the words!

(Incidentally, it seems like your aggregated version would be sensitive to the selection of bin edges, have you tried shifting them around and find the "worst" and "best" selection to convey the message?)

David M

My biggest issue with this plot is that ""Normal" is defined as the average rainfall between 1931 and 1990." Why are we comparing data from 20 centuries to an average across 59 years? Why not compare each year to an average of the whole data set, or even to an average per century? This graph is conveying that on average, Mexico is dryer than average. That statement is obviously contradictory. What it should be saying, I think, is that in the last century, Mexico has been abnormally wet compared to the average of the last 19 centuries.


Rettaw: I didn't mention it because I think they worked off of an old study that hasn't been updated. Since it's a 2000-year approximation, I suppose it takes a huge effort to update the data.
Your other point about bin edges is right on as well. What they could have done is to determine dry and wet regimes, instead of using arbitrary 100-year bins.

DavidM: I'd pretty sure there must be a reason for choosing the 1931-1990 period. We'd have to read the original research paper to see the explanation (or not). We can just think the whole time series as an index, and that period as the reference level. The truth is that whatever reference level one picks, someone would disagree.


Very informative post! Thank you so much for sharing it with us.

Chuck Derouen

interesting read. Thanks for the connections. I never would have connected kitchen cabinets to your charts. But well done!


The kitchen looks good!

The comments to this entry are closed.

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