Thread by @StatStas

Quality of the #2020Census data -- a session on race and ethnicity coding.

Census follows the OMB 1997 standards, does not set the standards: 2 categories of ethnicity, 5 categories for race (AIAN, Asian, Black/AA, Native American/Pacific Islander, White) + "some other race"

The race and ethnicity questions help understand how people self-identify, so research into these is necessary to understand how the U.S. population evolves (more multiracial, more diverse than measured in the past)

There were some proposals to start offering "Middle Eastern / North African" (MENA), but they did not make it to the #2020Census.

4 categories of Hispanic:
- Mexican/Mexican Am/Chicano
- Puerto Rican
- Cuban
- Yes, some other Hispanic -- other countries of origin are listed as examples

Race questions: White and Black had "Print" for write-in lines of detailed origin (German white or Nigerian black, say); American Indian / Alaska Native could add enrolled/principal tribes; multiple Asian categories ~10 countries of origin + Other Asian + Other Pacific Islander

Code list: a four-digit classification www2.census.gov/about/cic/Codi…

Added: new White detailed groups, new Black detailed groups, many AIAN/PI groups. About 99% of write-ins were automatically coded; clerk review for the residual write-ins. #2020Census: 350.5M write-ins, c.f. 54.7M in the 2010 Census

Up to six write-in responses, up to 200 characters per write-in line, no prioritization (c.f. two writeins, 30 characters, priority Hispanicity in the 2010 Census)

(I think I wrote in "Beige" as suggested by my American friends who said there is no way I should refer to myself as "White".)

(And the most meaningful "race" question I have seen was from one of the non-gov organization "White" "Black" "Asian" "Hispanic" "Native American" "immigrant". I could nail that without hesitation!)

2020 ethnicity breakdown: 62.1M Hispanic or Latino, 269.4M not Hispanic

Some other race 49.9M surpassed Black 46.9M as the second largest group after Whites.

The race/ethnicity results should be compared with caution between 2010 and 2020 due to these changes in measurements. [The 2020 was much improved from the methods perspective for the snapshot measurement, but of course for trends, the changes are bad -- S.K., not Census]

Q from the floor: the mission of the Census was to count every person once, exactly once, and in the right location. What about counting the race exactly?? [a trick question indeed].

A: we were not surprised by the #2020Census given what we saw in the 2015 content test (and that whole decade of testing). Before, in 2010, we saw good agreement with demographic projections. That's the best we can say.

Q: @dcvanriper asked for clarification regarding how "Cuban + Thai + Filipino" ended up as "Some Other Race + Asian".
A: all responses in the race write-in space were mapped to race [and apparently did not get matched to any major category groups -- S.K.]

Q: while you created a wealth of information on detailed race categories, people on another floor in your building are working hard to obfuscate that.

[wow, people have strong opinions on #differentialprivacy]

Q [opinion really] the improvements in race / ethnicity measurements did not go far enough because of the political appointees in OMB

As a general comment, Zoom organizers should pay exponentially more attention to the chat. Neither the camera nor the mic are working on this computer that I am connecting with, so I am limited to the damn chat @uscensusbureau looking at you.

The 2020 #ACSdata used the same question format and the coding scheme/algorithms as the #2020Census

Connie Citro observed that the vital records do not have the detailed coding used in the Census/ACS. Vital records are used down the line for the population projections that in turn used as controls for #ACSdata that in turn used to weight all gen pop surveys. What's going on?

Documentation on the #2020Census race codes

https://twitter.com/mecline6/status/1442907449623601155?s=20