I established the categories with the first data set: 2010 Census (race) and 2006-2010 American Community Survey (income). I updated the income categories on Jan 15, 2020
using the data from the 2014-2018 ACS. In the future, I plan to do annual updates to the income categories and will update the race categories every ten years using
the new census data.
Equal Percentile - each category gets the same percent of data. For instance with ten categories, you put 0-10% of the data in the first slot. I used this
for income. Each income category originally included 11.1% (for sequential color scheme) or 10% (differential color scheme) of the data, however I rounded to the nearest
Sliding Percentile - I created this system to handle race because it has a large tail distribution. Ex. there are a lot of census tracts that have under
1% of several racial groups. If I use a percentile system then I will be creating category breaks for things like 0.3% vs 0.5%. So instead of having an equal
percent of the data in each category, I weighted it so that the lowest categories have a larger percent of the data in them. For instance with 9 categories, I put in each category
22, 17, 15, 12, 10, 8, 7, 5, and 4 percent (instead of 11.1% in each). This allows us to show the differences between higher percents of a racial group that
would otherwise be given the same color (ex. 30% vs 60%). I then rounded the categories to have nicer breaks (ex. to the nearest 0.5% or 1%).
Equal Breaks - This method would create categories with equal breaks. For instance for percent black the categories could be 0-20%, 20-40%, 40-60%, 60-80%,
and 80%-100%. I used this for the race plural layer. I didn't use it elsewhere because it does a poor job of handling anything other than an even distribution.
Jenks Natural Breaks - This method minimizes the difference between each item in a category and the rest of items. I didn't use it because it gives too
little emphasis to differences in the tail.
Color schemes come from colorbrewer2.org
Sequential - a continuum. All race layers use this as do two of the income layers.
Differential - emphasizes the break around the median value. Only used for one income layer.