2020 Census Questions Submitted to Congress

[muon2]

One of the purposes of the Census is to provide complete data, not merely estimates, of the population by block to be used in redistricting. Up through 2000 asking age and race was sufficient to get the voting age population required to test compliance with the Voting Rights Act. Court decisions leading up to the 2010 Census made it clear that for certain minority groups, especially Latinos, citizen voting age population was the correct measure for testing VRA compliance.

However the 2010 Census lacked a citizenship question and that posed real issues for the states and localities trying to draw plans that would comply with the VRA and be sure that they would stand up in court. There was ACS data with citizenship information at the time of the Census release, but the ACS geographies didn't match those of the Census since it was collected prior to the finalization of the 2010 Census boundaries. ACS data with the new geographical boundaries came out many months after the Census data, too late for some states that have early constitutional requirements.

Even if the Census had the manpower to process a round of the ACS while doing the decennial Census (they don't) it isn't available at the block level, but is aggregated by larger geographical units. The ACS is also just an estimate based on sampling so it has statistical errors, just like a political poll does. If that sample is small, as is the case for a 1-year sample, they can't even make estimates at the block group level and the data is too coarse for redistricting.

To get a good ACS sample that can really apply to redistricting at level needed by mappers the 5-year sample would be needed. But that means waiting over 5 years after the geography is finalized (typically 2 years before the Census) to get an ACS data set, so that might occur 2 years after the redistricting data is first released.

The other way would be to greatly increase the number of households that must answer the ACS in the year after the Census. Increasing the number probably means going back to the long form sent to 10% of all households. That again runs into the manpower problem since the Bureau is completely focused on preparing the Census during that year, and the ACS was designed to spread that workload more evenly through the decade. At that point the Census might as well just pick out the questions needed for redistricting under the VRA and include them in the standard Census form.

[muon2]

I'd say there is an argument to be made behind the motives for this. The fact that they threw in the laughable "Voting Rights Act" excuse shows they have something to hide. I'm just curious whether it's a matter of federal funds or setting up a future case where they can be allowed to draw maps by eligible citizen voters, which would give them another advantage in redistricting, or both. To think that Republican actions here are innocent is incredibly naive. They are leveraging their power at the federal level to help them win future elections, just like they have done in the past.

I'm not sure that the VRA part is an "excuse" on the part of the Census Bureau. I know from panel discussions going back to 2009 that this question of citizenship has been one they have wrestled with long before this administration. At one 2010 meeting expert redistricting lawyers were giving their best ideas for workarounds to try to be compliant with results from the prior decade's cases.

As recently as the 2000 Census, 1 out of every 6 households was sent the long form that included the same citizenship question in the 2020 draft. Cases were using VAP in the 1990's cycle, though for Latinos it was recognized that numbers well above 50% were needed (The 7th circuit went along with a threshold just under 60% for IL-4.) Courts weren't requiring it, and the long form needed to be moved out of the decennial Census, so the question was dropped for 2010.

After 2000 courts leaned move heavily on CVAP, and not just VAP,  because then all groups could be treated by the same use of the Gingles test when identifying that a minority made up 50% of a compact area. However, realization of the impact of those decisions on mapmakers came too late to affect the official list of questions. The courts have continued to move more in the direction of CVAP for the VRA, so shouldn't the correct data be available at the time the plan is prepared?

And yes, I get that there are partisans who would think this will help them in future elections. But I don't see how that changes the underlying issue. We should also be able to look back to the 2000 Census and see how the response rate for the citizenship question on the long form compares to the ACS. If there are real differences, then yes I would agree that more work needs to be done to improve education and outreach on the question. But if not, I think those who are fighting this may be just as partisan as those who are motivated to add it in hopes of winning future elections.

[muon2]

And yes, I get that there are partisans who would think this will help them in future elections. But I don't see how that changes the underlying issue. We should also be able to look back to the 2000 Census and see how the response rate for the citizenship question on the long form compares to the ACS. If there are real differences, then yes I would agree that more work needs to be done to improve education and outreach on the question. But if not, I think those who are fighting this may be just as partisan as those who are motivated to add it in hopes of winning future elections.

Does it even matter at this point if the push back is a partisan fear that Republicans are trying to rig the system in their favor even more? I think that is a completely rational response after what Republicans have been doing since Obama's presidency wiped out the Democratic Party. On one side, you have people who just want the census to stay the same as it has for decades with regards to this question, and the other, you have others who want to add a question that will give them data to possibly bolster attempts to redraw districts in a way that shifts power further to the Republican Party. I think if you were a well-informed, strongly Democratic voter yourself, you would probably have the same fears. You might even be wondering, "why can't Republicans stop trying to put their thumb on a scale that already significantly benefits them at almost every level? Why can't one year go by where scheming politicians don't try to rig the system more in their favor?"

Partisan concerns aside, I still have to ask why it is so important to add such a controversial question so close to the census. There doesn't seem to be a lot of support for the Voting Rights Act excuse - a flurry of responses which could be partisan in itself, but I'm not so sure. I haven't really seen anything that supports this being so important that it has to be done. I concede that you provide a reason for it, but is the reason simply existing good enough in this case?

I would only say again that he citizenship question was in there up through the 2000 Census, and was only missing once - in 2010. It's true that the question didn't go to all households, but those long forms were treated just as rigorously as the short forms that went to the other 5/6 of the US. That treatment included follow up by workers in the field, so we have data from 2000 that can tell us how fairly and accurately the question can be collected.

One area where the past should be a guide are in the results from CA in 2000. This was after CA had started limiting access to services for illegal immigrants through prop 187 (1994). Undercounts in minority areas had been a problem for decades and were again in 2000. If reticence to participate increased because of a citizenship question, it should have been present in CA in 2000. I don't recall any particular stories that cited a concern that the long-form citizenship question would impact CA counts that year beyond the problems that were faced in 1990, but if there is a paper out there I would be interested in seeing it.

My point is that the controversy does not seem to be backed by either history or data. It seems purely political to me.