top of page

Dudley Poston on How Sabotaging the Census Hurts Conservatives


Dudley Poston, one of the most renowned demographers in the world, has recently been writing a set of op-eds on Undercount in American Censuses.

This sounds like it would be a super-boring subject suitable only for methodological treatises.

It turns out to be a mega-important political issue that everyone both on the left and the right needs to be paying serious attention to.

It turns out that you can majorly benefit blue states or majorly benefit red states by tinkering with the Census.

The tinkering, however, does NOT work the way you think it would. You have to know what you are doing.

Donald Trump tried to put his thumb on the scales of the Census so he could increase the power of his political base and undercut the power of liberals and minorities.

He shot himself in the foot. His crudely executed cheats ended up giving blue states and lefties a non-trivial advantage.

However, not all wannabe Census Cheats are going to be as ham-fisted about it as Donald Trump was. A newer better generation of cheaters could do real harm to democracy or could end up really damaging whole sections of the country.

You can read Dudley Poston’s interview by clicking the link below – or you can watch the video that comes with the article. I closely follow Dudley Poston’s logic in the arguments I make below.

Rigging the Census is pretty easy to do.

Rigging the Census can be politically advantageous. You can change the count of how many voters live in what places. When redistricting comes, this can put more representatives in the place that you falsely claimed had lots of voters and fewer representatives in places that you falsely claimed had fewer voters.

You can misrepresent the ethnic composition of an area. This can be useful in defending yourself against claims that your self-interested gerrymandering of voting districts is hurting one particular ethnic group or another.

You can misrepresent the number of people in an area that are senior citizens, or are of school age, or who have disabilities. This allows you to give less funding to senior services, school spending or medical assistance in areas you don’t like and more such funding to areas you like better. (Like those with your political supporters.)

It goes without saying that rigging the Census does terrible things to the actual solution of social problems. If government agencies and planners do not have accurate information about what is going on, they are more likely to make honest mistakes about the programs they create and the services they provide. Taxpayers and everyday citizens get hurt. The agencies get blamed. But the real problems stemmed from the provision of inaccurate information to the agencies entrusted with the responsibility of fixing social problems.

It also goes without saying that rigging the Census does terrible things to scientists trying to use that data to understand how society works. The scientists also contribute to the solving of social problems.

However, a partisan politician might also not care about whether the government solves problems or not. That politician might only care about whether his or her party wins the next election. However, if the cheating gets done in the wrong way, it backfires. Government benefits and legislative seats get shifted to the opposition.

Something like this may have happened with the 2020 Census.

Donald Trump wanted a Census that would have high counts for his supporters: suburban and rural whites – and low counts for his opponents – city dwellers, blacks and Hispanics.

The strategy for getting the Census results he wanted was to

  1. Complete the Census Very Quickly.

  2. Not Use Statistical Techniques for Correcting for Hispanic Undercount.

Why would someone who wanted to increase the seeming percentage of suburban and rural whites in the American population go for a fast census with no statistical correction for Hispanic undercount?

Census Speed. Urban areas, black areas and Hispanic areas take longer to count than do white areas, suburban areas and rural areas. The issue is finding all of the housing units that might have people that you can count.

A census taker identifies housing units by “block-listing” where you walk or drive around a block looking for all the housing units. In a suburb, this is usually pretty easy. In suburbia, there are nice neat distinct single-family houses that are readily identifiable. If there are apartment buildings, they are easy to find and it is easy to locate all of the units in each building.

Rural areas are harder to block list but not by that much. People live in farm houses or if they are poor, they live in trailer homes. Generally, there is some sort of driveway that will lead you to the farm house or the cluster of trailer homes. Sometimes, the farm house has trailer homes around it. (An older farmer might a trailer for his adult children, or let a friend move a trailer on to his land.) There is a small number of people who live way way back in the woods. Some are extreme nature lovers; others are doing something they don’t want you to know about. But generally, following driveways will give you a pretty accurate count.

Cities are much harder to blocklist. They have lots of complicated housing that hides the existence of all sorts of people. Poor people live in basements, or in back alleys. Buildings have funny little doors that lead to unexpected housing units. There are courtyards you can’t see from the street. Homeless people are especially tricky to find. They have encampments that are hidden in parks or under bridges or in industrial areas. Cities have institutions such as hospitals, college campuses, and prisons. These populations are hard to allocate between “really living in the institution” versus “belonging somewhere else”. Finding everyone in a city is generally do-able if you devote enough time to it. But it is a longer, slower and more exacting process than finding residents in other settings.

Rushing the count means rushing the urban count. The population count for the cities goes down. If there were a lot of minority members in the city, the minority count goes down too. If cities are more liberal than are suburbs and rural areas – they would appear to have fewer voters and deserve fewer representatives.

Correcting for Hispanic Undercount. Migrants in general and Hispanics, specifically, are often missed by the Census. People who fear immigration authorities are less likely to fill out a Census form or cooperate with a Census taker. People who come from abusive, authoritarian countries are less likely to cooperate with government figures. Tensions between African Americans and police lowers the willingness of Blacks to cooperate with the Census.

This problem gets solved two ways.

The Census often works with trusted figures in migrant, Hispanic and Black communities to have them speak to their communities. The trusted figures explain the importance of getting an accurate count on the Census. Many of these outreach campaigns have been quite effective.

The second method is to correct for the undercount statistically. The Census has developed various techniques for calculating how many people are missed. They do a Census “the normal way”. Then for certain areas, they do a deep scour. They find how many extra people they find on the deep scour – and what their age, sex and ethnic distribution looks like. Knowing how many people they miss in different settings (different regions of the country, urban vs. suburban vs. rural, rich areas vs. poor areas etc.), they can run a correction for each area that allows them to figure out what kinds of people are missing and where they are located. After-tests of these statistical corrections show that most of them are quite effective.

In 2020, it was decided to do the Census with no corrections. The logic was – why increase the number of Hispanics in the final count?

*  *  *

Superficially, it would have seemed that these devices should have been effective. Political power would shift from liberal Hispanic areas to conservative white areas. Government benefits would also have shifted to conservative white areas.

As Dudley Poston points out in his interview, exactly the opposite occurred.

Many conservative white voters live in areas with increasing minority populations. Their concern about the increasing minority population is the very issue that leads them to vote for conservative pro-white candidates in the first place.

So, when you undercount the increasing minority populations, by implication, you also undercount the areas with conservative whites who worry about minority populations. As a result, both government monies and political representation in Congress shift away from the states where the conservative whites live to areas with stable white populations. Texas, where the undercount issues were substantial, lost funding for government programs and lost a chance to get more seats in Congress. Minnesota, where a thorough and complete census was done, gained government funding and congressional representation. There were parallel shifts across the U.S. with blue states benefitting and red states losing.

*  *  *

What’s the lesson that should be drawn from this?

I personally hope that the lesson would be that people should not try to rig censuses. The best census is an honest, well-executed census that uses all of the best technical methodologies available to get the most accurate possible count.

An alternative interpretation might be that the obvious cheats in censuses don’t work. If one wants to rig the census, one needs to use better and more sophisticated techniques.

You, the reader, can make your own call.

But Dudley Poston is telling you the truth here. The truth is a good thing in censuses.

bottom of page