This chart shows how badly the census could still undercount people of color

The Supreme Court just prevented the Trump administration from adding a citizenship question on the census, at least until the Census Bureau gives a sufficient explanation of why the question is required.

Census experts worried that adding such a question to the census would reduce the response rate from undocumented and immigrant communities.

But even without a citizenship question, the 2020 census is still projected to undercount people of color — even more so than the previous two surveys, according to a report from the Urban Institute.

This chart looks at how the 2020 census would perform if it was conducted at 2010 levels, then compares it to the actual projections using the census planning documents. The effects of the citizenship question aren’t included.

This next chart looks at how severe the undercount would be if the 2020 census performed on the lower end of the Census Bureau’s expectations, and it takes into account the potential chilling effect from the citizenship question:

The projected undercount for the 2020 census is caused in part by underfunding. To reduce costs, the Census Bureau introduced a new online survey method, but had to cancel two of the three tests of the new system in 2018 because it didn’t have the funding. It’s vulnerable, transient communities who are the hardest to find and survey, which is why testing of these systems is so important.

In addition, merely the public narrative around the citizenship question could depress the response rate.

The citizenship question may have been added to suppress turnout, by design. Files from deceased Republican strategist Thomas Hofeller show that he studied the effects of a citizenship question and found that it would electorally benefit white Republicans. He subsequently advised the Trump administration to add the question.

Let’s back up for a second: Why all the hubbub about this undercount?

The Constitution requires a census every 10 years because it’s the data that determines how much electoral power each community gets. When a certain group is undercounted, it essentially means they are being apportioned less political power.

And it tends to be vulnerable populations that are undercounted. These are people who are harder to find, more transient, less trustful of government, and less tied to communities. Here’s how much black and Hispanic voters have been undercounted in every census since 1950. Basically, they’ve always been less represented than white people.

There are ways to statistically adjust for these miscounts, using something called “sampling.” But the Supreme Court has said sampling can’t be used to apportion seats in the House of Representatives, though it left room for sampling to be used for state redistricting and apportionment of federal funds to states.

What that means is that we’re left with this imperfect process to determine how much representation everyone gets in government. But when our government doesn’t invest in the process — and actively sabotages it for political gain — it further compromises the legitimacy of what is supposed to be an objective count.