Data on frontline workers is based on research by the Brookings Institution: https://www.brookings.edu/research/to-protect-frontline-workers-during-and-after-covid-19-we-must-define-who-they-are/
Unemployment data is updated weekly by the Louisiana Workforce Commission via their UI Dashboard: https://public.tableau.com/profile/louisiana.workforce.commission.lmi#!/vizhome/LWCUIDashboard/DashboardMain
For many of the maps, the color scales are broken into five groups that are determined by the entire state’s data: the darkest shade covers the top 20 percent of estimates across the state, the bottom 20 percent of estimates will be colored the lightest shade, and so on for the three shades in between.
There are some exceptions to this coloring scheme: 1. A continuous scale is used to color the race/ethnicity and average household size maps, and; 2. The two “index” values are colored as described below.
Two of the mapped indicators are index values created by other researchers. These values are a composite of multiple measures, such as income or health insurance coverage, that research showed related to the outcome of interest (social vulnerability or risk of homelessness). Once calculated, the index value for each tract was given a percentile ranking against every other tract in the state. The Social Vulnerability Index is calculated by the Centers for Disease Control: https://www.atsdr.cdc.gov/placeandhealth/svi/index.html
Data on renters most at risk of homelessness comes from the Emergency Rental Assistance Priority Index, a measure published by the Urban Institute in response to the coronavirus pandemic: https://www.urban.org/features/where-prioritize-emergency-rental-assistance-keep-renters-their-homes
The color scales for these measures are shifted to show more detail between the higher-percentile tracts, which are the more vulnerable in both cases.
Demographic data comes from the American Community Survey’s 2014-2018 5-year estimates. At smaller geographies, such as the tract-level data used in these maps, the data’s error tends to be large enough to make the data unreliable. Reliability of data can be determined by the estimate’s coefficient of variation (CV), or how large an estimate’s error is compared to the estimate itself. The smaller the CV the better, and we call data with a CV less than or equal to 0.15 “reliable.”
To address the higher CVs of this tract-level data, the data is smoothed: tracts’ estimates are averaged with the estimates of adjacent tracts. In cities with populations of 100,000 or greater, tracts are only averaged with their “true neighbors,” which aren’t separated by rivers or other bodies of water. This process is nearly identical to the one described in The Data Center’s 2012 report, “ACS Mapping Methodology”: https://gnocdc.s3.amazonaws.com/reports/GNOCDC_ACS_Mapping_Methodology.pdf
Each tract has a new CV calculated by a similar method. If the new, smoothed CV is still greater than 0.15, the tract is displayed with crosshatching.
The CVs of the average household size data showed this data to already be quite reliable, so this indicator did not undergo the smoothing process.