Texas Grows 70% Each Year

Welcome back, everyone, to Cartastrophe: The Blog with First-World Problems (as a reader rightly pointed out recently). Today’s effort comes to us from the folks at the Associated Press:

This choropleth appeared this morning accompanying a story I was reading online about the new population numbers out of the Census Bureau. Most of the map is unremarkable, but the legend is worth noting. According to the title on the legend, the colors indicate population growth, in thousands. But, the actual numbers in the legend are marked as percentages. It is probably unreasonable of me to believe that the population of Texas increased 20,000%, as that would put their current population somewhere above 4 billion people. I believe that these numbers are intended to be percentages, and that the title on the legend is simply incorrect. Perhaps this map was made by altering an existing product, and the author forgot to make some necessary changes.

The more subtle, and much more common, problem with the legend is the arrangement of the numbers. There’s an overlap to the data classes. If a state had 10% growth, does it go in the third class or the fourth? Better, I think, to add a decimal place to these numbers so that the separation is clear: 5.0-9.9, 10.0-14.9, etc. Gaps between classes make it plain which numbers go in which class. Alternately, a more complex solution is a redesign of the legend. It may be possible to visually clarify that the 5-10 class includes all numbers from 5 up until, but not including, 10. Here’s a mockup of something that comes to mind as a potential design solution:

That may or may not be too difficult for the average reader to interpret. It’s off the cuff, so I’m not entirely certain about its merits, but I do believe there are visual solutions to this problem as well as ones which rely on changing the numbers. The latter may be more clear, ultimately.

The colors for the choropleth are largely fine, but I think the various shades of blue are a bit too close to each other to easily match back to the legend. Reducing the classes by one, or by making the darkest blue even darker and stretching the color ramp out would help ease this.

One Nice Thing: I appreciate the author’s use of small boxes of color next to the state names in the northeast. The states get pretty small up there, and figuring out the color of Delaware can be challenging. With this solution, there are always legible swatches of color associated with each state.


7 Responses to “Texas Grows 70% Each Year”

  1. 22nd December, 2010 at 1:06 pm

    That legend labeling problem is frustrating as a cartographer. Decimals and gaps can be messy in an otherwise orderly legend. (And you have to be sure none of the data actually falls in the gaps—can’t write a nice number like 14.9 if one state is actually 14.95.) But clearly showing the ranges can require a legend design that doesn’t easily fit within the constraints of the project. Somebody really needs to invent an “up to but not including” symbol that we can use in place of a dash.

    Perhaps we should stick to interval notation. [5,10), [10,15), and so forth.

    • 2 Daniel Huffman
      22nd December, 2010 at 4:23 pm

      Interval notation might be nice. One of the things I was thinking about here is that we run into problems with using something unfamiliar. But, if we just bit down and started using something that works well and is unfamiliar, eventually we’d create a level of literacy among the map reading populace where this would become familiar.

  2. 3rd January, 2011 at 12:00 pm

    I think your visual solution is very nice. It might be improved with font and arrow symbol changes but it eliminates the implication of precision that comes with decimal numbers in a legend as well as the value gaps.

    For this map I think it can be assumed that no states growth was exactly on the center line of any range so there is probably no ambiguity in determining what color to give each state for the cartographer.For the map reader (ignoring the typo) you can read each range as saying this color represents growth ‘between 5 and 10 percent’ or ‘between 10 and 15 percent’. It’s a range of 5 percent that only returns information that says ‘this states growth was somewhere between 5 and 10 percent’. No confusion or ambiguity actually arises from having repeated values in the ranges since each range statement is likely always correct in this case and delivers the generally the same level of detail as it would if the legend read in more precise decimal values. If a state happens to be directly on the overlap value this argument falls down of course, but I think that’s unlikely.

    • 4 Daniel Huffman
      3rd January, 2011 at 12:10 pm

      Thanks for the kind words, David. You are quite right in that it’s unlikely that a state is on the edge, and I think I have been looking at this problem too much from the side of the data rather than the reader, trying to figure out which number should go into which class, rather than reading each class as a range.

      Your comment reminds me of another solution which I forgot to mention. I encourage my students to put the actual class ranges on their legends, rather than trying to give a continuous sequence of numbers. That is, something like: 1-4, 8-9, 11-13, etc. This makes sure there are clear gaps between each class, and gives extra information: there are no states between 4 and 8, for example, and any state in the second class is between 8 and 9 (rather than between 5 and 10, if we were trying to make all the classes adjoin each other numerically). It adds a bit of extra precision to things.

  3. 5 Curtis
    20th January, 2011 at 11:13 pm

    David was exactly right. I began my comment on the intention of stating what he already did. If the “in thousands” was removed the chart is easily read.

    Their is one other thing. <5% and <0% are data sets that overlap. But this is not really a big deal. most people will interpret this properly as red being below 0% and the lightest blue color being 0%-5%.

    The simplicity of the chart and ease of reading, vastly outweigh the need for perfect nomenclature of the data sets. Stuff like that only bugs obsessive compulsive people. OC's are impossible to please anyhow.

  4. 6 anonymous
    13th August, 2012 at 1:02 am

    One solution might be to use a notation like:

    [0-5[ ; [5-10[

    It has the advantage of being easily understood, yet it looks pretty ugly in a map legend.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: