# Child Poverty and Race, USA 2016

This week’s Tidy Tuesday invited us to examine county-level census data from 2016 in the United States (sourced here). I have been poking around in the data and discovered an interesting visual artefact in the following otherwise unsurprising quick plot:

I split the USA’s 3,142 counties (excluding Puerto Rico) into regions, based on the US Census Bureau’s four statistical regions (pdf). I excluded Puerto Rico because its racial breakdown does not fit the exercise here. And I chose to classify counties by White (%) because that seems the best way to capture the percentage of the population from racial minorities. No doubt this is a hugely crude way to go about the exercise.

Anyway, even with my basic familiarity with the United States I am hardly surprised to see child poverty correlate negatively with the proportion of the population that is white. What is surprising I think is the way that child poverty seems higher in the Northeast and especially in the South in more homogeneous counties.

Still, we have to be careful. When we look at the Southern counties what we are seeing here might relate to a combination of rural poverty and an uneven distribution of people by race. Indeed, it is very striking how segregated the United States are.

## Warning: funs() is soft deprecated as of dplyr 0.8.0
##
##   # Before:
##   funs(name = f(.))
##
##   # After:
##   list(name = ~ f(.))
## This warning is displayed once per session.
Table 1: Counties with population composed of 75% or more from one racial group
State Counties Segregated Not Segregated Segregated (%)
West Virginia 55 55 0 100
Kentucky 120 116 4 97
Tennessee 95 84 11 88
Maryland 24 14 10 58
Arkansas 75 43 32 57
Virginia 133 68 65 51
Alabama 67 30 37 45
Florida 67 28 39 42
North Carolina 100 40 60 40
Oklahoma 77 31 46 40
Texas 254 77 177 30
Mississippi 82 20 62 24
Georgia 159 37 122 23
Louisiana 64 12 52 19
South Carolina 46 5 41 11
Delaware 3 0 3 0
District of Columbia 1 0 1 0

But note also that each state (and I imagine at more granular levels than that) has its own population distributions:

So, as always, more study required.

Gist here.