So, I have three datasets, same geographic extent and same aggregation zones. The first shows population from 2000, the next shows a modeled population projection into 2030, and the third shows the delta between the years. The layout are basic chorpleth maps. While this seems simple, My problem is with the choice of classification method for the data.
It is critical to demonstrate the 'movement' of the population and where we would expect the population to be located in the future. I usually rely on Jenks for my data, but the breakpoints are moving for each map and the classification ranges shifting, so the maps look totally unrelated and you certainly cannot understand the pattern of movement.
To illustrate my point I included a layout that arrays 2 different themes ( I am making a totally different map, but same choropleth idea and aggregation zones), Streetcars Origins and Destinations. You can see that using the same legend/classification across the different datasets shows how shortnening the streetcar line 'concentrates' origins and destinations. 'Movement' is apparent. But this seems wrong from cartographic purist point of view...
So, here come some considerations.
...What is the downside of using a single legend for the 2000 and 2030? If I do use a single legend, do I choose the breakpoints myself and not concern myself with the arbitrary nature and conflicts over data ranges?
...I am normalizing the data to report density of population...2.5 households per acre. I could also say 3 thousand more people or 50% growth from today. How do you folks report change over time data?