4

I have always struggled to come up with a great solution to handle choropleth or any other quantitative interval or dataset classifications in a map legend where the numbers are sequential.

For instance: here is the ArcGIS default legend, which is a bit gaudy and what's the real cuttoff? .55000000001 or .5501 or what?:

enter image description here

And I see many people create their legends like this:

enter image description here

Usually with "TIE" written as "50%", which in this case would be wrong on two levels, 1. These election results (like many) have pluralitiy winners,i.e. winning with 49.8% 2. If a county/unit is 55%, which classification is the unit a member of? It can't be both! And this particular dataset does contain 4 tied units.

I tend to use interval notation in my legends, like this:

enter image description here

Set notation { x| 55 < x <= 67} would work as well, but I'm worried that people won't always understand these methods.

Anyway, I'd like to hear your solutions to this to help figure this out.

1

First off you should remove the "TIE". You can probably stay with 6 fields if you want, but it depends on your data's deviations. You want to be careful with having too many fields because the naked eye cannot perceive the color differences. Removing the white "TIE" is going to allow you to make your 1st and 2nd fields more distinguishable. Right now they are too close in color.

Your label will be different than the actual classification. For what you have there you would the actual values to list as:

50.01-54.99
55-66.99
67-100

Your decimals will change based on your data. If you are setting your deviations, set them to whatever your attributes are. For example: if you data is rounded to 2 decimal places, then set your classes to 2 decimal places. If you use one of the preset deviations, I believe you can choose how many to use (or it will automatically do it based on the breaks).

For the values I listed below, your actual labels would read:

50-54
55-66
67-100

You want to avoid using decimal places for this because they really aren't necessary and you want a choropleth map to be as simple as possible. If you want you could do:

50-55
55-67
67-100

That works too. 50-55 would imply all values between 50 and 55 (which matches your break classifications of 50.01-54.99. 50-54 would be implying numbers between 50.01 through 54. I suppose your dilemma is how do most people read "-"? Do they read it as "to" or "through"?

As far as the aesthetic of your legend, that's up to you. It depends on the overall appearance of your map.

  • The "tie" label needs to remain on the map, as four units are statistical ties. With a divergent color scheme, I think the user can perceive the difference between the 3 shades of blue/red. I would agree with your integer classification labels, except the difference between 54% and 55% can be a few thousand people - That scheme would work if I was classifying number of voters – CCantey Apr 22 '16 at 15:52
  • It depends on the size of your map. If you're showing the entire country, it can be difficult to distinguish those colors. Does it have to say tie? Can you call it "50"? It looks strange having a word represent one value, part of some values as a combo of a word and number, and then some as a number. You should not have tie in the values for TIE-55%. That would imply that TIE (which can only be represented by exactly one value) falls into three separate values. If you need to keep tie maybe do: TIE<br> 50-55<br> 55-67<br> 67-100<br> – Adam Kara Apr 22 '16 at 15:56
  • Good point - +1 for your input - i should have included the map, its a pretty large scale, a few counties wide/tall, so the map is quite readable... On the "50", in many units, percentages less than 50 win, so right now, I feel like "tie" has to be the break point... Also that's why I use the interval notation () [] to show that tie is not included - and thus proofing my original theory that it is too difficult to recognize immediately : D – CCantey Apr 22 '16 at 15:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.