Legend Advice (Normalised Data)

Discussion created by sjacksonesriaustralia-com-au-esridist Employee on Dec 26, 2012
I have a polygons layer (postcodes) and a points layer.
Each point can be of any 5 languages.
I have pre-calculated:
- total points in each polgyon (Total)
- Total of each Language in each polygon (TotalArabic, TotalSpanish, etc)
- Normalised Values for each language (TotalArabic/Total, TotalSpanish/Total, etc)

I have served these up as a dynamic map services at 10.1 SP1, and enabled dynamic layers.

I have a dropdown to choose one of the 5 languages on my app.

I am making use of the GenerateRenderer to request a map (5 classes, natural breaks, fixed color ramp).

I am using each of the corresponding fields for the pre-calculated normalised values for language counts (as I do not think the renderer supports normalising on the fly?).

This works as expected, but I would now like to explain to the end-user what the colours indicate (i.e. darker blues indicate higher overall concentration of XX language within this postcode.

The problem is, the normalised values do not mean much to the end-user:

I am looking for some advise from both a UX & development perspective, on the best path forward.

I was toying with the idea of having a static legend that appears when a value in the dropdown has been picked, with no labels, but some kind of graphical indicator that darker blue means more concentrated. 

I was also wondering if you can override the labels in the legend to show something easier to understand like percentage values.
I assume I could perhaps get around this by hard-coding the class break values in the renderer, but then I lose a true natural breaks classfication.

Would love to hear some ideas on how you would go about this.