Census 2020: Challenges for the IE

As Census 2020 approaches, I wanted to take a closer look at the unique set of challenges that the Inland Empire faces in terms of achieving an acurate count. All the data comes from the Census planning database. In preparation for the 2020 Census, the planning database includes a census measure called Low Response Score or LRS. The LRS can be directly interpreted as the “predicted mail non-response rate”.1

Some caveats here:

  1. You are not supposed to aggregate LRS to calculate county-level estimates. See below

The Low Response Score should NOT be aggregated (i.e., summed or averaged) to different geographies than it is calculated. The Low Response Score is based on a linear regression model that utilizes multiple characteristics of a given census tract or block group to predict mail nonreturn rates for that particular geography. Even though census tracts nest within a county, you should not sum or average the Low Response Score across all census tracts in a county to arrive at a county-level Low Response Score. Separate models would need to be developed for other levels of geography (e.g., county, state).2

  1. LRS only predicts non-reponse for MAIL
  2. Subsequently, the current version of LRS does not account for broadband access

Estimated non-response by IE subregion

Here’s a look at the distribution of tract-level non-reponse rates by tract-population. Some subregions have tracts with small populations but large non-reponse rates and others have large tract-populations with relatively small non-reponse rates.

Total estimated non-response

Sono Shah
Ph.D. Candidate