If the Census Bureau truly values accuracy, it should count incarcerated people at home

The Census Bureau’s current method of counting people in prison and jail is prone to errors with sizable consequences. Counting incarcerated people at home can produce more accurate data.

by Aleks Kajstura, October 30, 2024

It is no secret that the Census Bureau counts incarcerated people in the wrong place — a prison or jail cell — instead of in their actual communities. As a result, when states and local governments use census data to draw new political boundaries during their redistricting process, they inadvertently give residents of districts with correctional facilities greater political clout at the expense of everyone else — a problem known as prison gerrymandering.

For two decades, we’ve worked to fix this problem, but it isn’t the only way that the Census Bureau gets it wrong when counting incarcerated people. The Bureau also routinely mistakenly and unknowingly places prisons and jails in the wrong location in its data or dramatically miscounts them, further distorting political representation across the country. This problem exacerbates prison gerrymandering and warrants further scrutiny.

We examined 2020 Census data to better understand the impact of this often overlooked error. We found that it is not a rare mistake; it is perpetuated across the country and across the decades. And the solution is clear: by counting incarcerated people as residents of their homes the Census Bureau will minimize technical inaccuracies while also ending prison gerrymandering.

The Bureau is wrong to count incarcerated people as if they were residents of correctional facilities

Before looking at the impact of the Census Bureau misplacing or miscounting entire prisons and jails, it is worth revisiting its flawed reasoning for counting incarcerated people at those facilities in the first place.

The Bureau says that it determines where to count people using its “usual residence rule” which says it counts them where they live and sleep most of the time. In attempting to justify its decision to count incarcerated people in prisons and jails, it argued, “counting prisoners anywhere other than the facility would be less consistent with the concept of usual residence since the majority of people in prisons live and sleep most of the time at the prison.”

As we’ve noted before, this is simply not true. Short stays in prison are common. Even people incarcerated away from home for a year or longer do not stay in one place; they are usually moved between multiple facilities. While they are being shuffled between facilities, incarcerated people generally maintain a usual residence elsewhere; their pre-incarceration home remains the only actual stable address. There are plenty more reasons — backed up by hard data — that make clear that by counting incarcerated people at prisons, the Census Bureau is choosing the least accurate option.

The Bureau’s flawed way of counting incarcerated people makes its other mistakes even worse

We often think of the census as a gold standard for population data. And while it is, on the whole, fairly accurate, no matter how careful you are, perfection is not possible when attempting to count over 330 million people.

However, when the Bureau counts an entire prison or jail — which can account for hundreds or thousands of people — in the wrong place, it can have a big impact.

The Bureau routinely misplaces correctional facilities

During the 2020 count, the Census Bureau counted at least one prison or jail in every state in the wrong place. Most of the populations counted erroneously were small, but occasionally, the census reported hundreds or thousands of incarcerated people in the wrong place.

Here are just a few examples from a single state, New York.

The Census Bureau didn’t count any people at Adirondack Correction Facility — a place that holds bout 300 incarcerated people. However, roughly the same number of incarcerated people were counted in a state forest about 175 miles away.¹ Our best guess is the Bureau accidentally counted this facility in the wrong place.

Woodbourne Correctional Facility appears to have been counted twice — once in its actual location,² and also in the empty field next door.³

Empty fields are a common place for the Census Bureau to mistakenly report correctional populations. In Wyoming County, N.Y., rather than counting them in the facility where they were confined, the Bureau counted nearly a thousand incarcerated people in a field,⁴ and another group of over 800 was reported just across the street from their facility’s location — also in empty fields. This is a perennial problem with the census. In the 2000 Census, the Bureau reported nearly 1800 incarcerated people — likely from the Wyoming Correctional facility in a random field halfway across the county.⁵

These examples are all just from a single state, and these problems are not confined to New York. The scope of the problem becomes clear when considering that these miscounts and misallocations exist in every state.⁶

Some might be tempted to argue that the Census Bureau might make similar mistakes if it counted incarcerated people in their home communities rather than counting them at the prison or jail. This would almost certainly happen. Despite its best intentions, the Bureau is not perfect and will make mistakes.

The difference is the size and scale of the mistakes. If the Bureau places a prison that holds hundreds of people in the wrong place, it can severely skew the counts. However, if it counts a single or even several incarcerated people in the wrong place, it would have a negligible impact on political representation and the underlying Census data.

The Bureau’s mistakes create incorrect race and ethnicity data

When the Bureau counts even a single facility in the wrong place, it drops thousands of people into an arbitrary location. This can flip the race characteristics of thousands of people, as the Bureau did in reporting 2020 Census populations for the blocks containing Angola Prison in Louisiana.

The prison accounts for a third of the Parish’s census population, so any population data reported for the prison will heavily influence the demographic picture of the parish. The 2020 Census counted over 4,000 Black people incarcerated at Angola as if they were white. At that time, white people actually accounted for less than a quarter of Angola’s population. Not only does such an error further distort the demographic data of the surrounding parish, but it blatantly obscures the disproportionate toll that mass incarceration places on Black communities.

Furthermore, by creating a false picture of the population in a facility, it also makes it harder for states ending prison gerrymandering to comply with voting rights laws that require accurate race and ethnicity data for drawing districts.

Misplacing prisons & jails exacerbates prison gerrymandering and diminishes trust in the Census Bureau

The Census Bureau’s current approach — counting incarcerated people as if they were residents of a correctional facility — creates inaccurate pictures of communities across the country and leads to prison gerrymandering. This problem is made worse by the Bureau’s persistent, multi-decennial errors that place correctional facilities in random other locations. A single error (and it’s never limited to single errors) haphazardly injects hundreds or thousands of people into a random community’s redistricting data.

Incarcerated people and their families lose trust in the Census Bureau when they don’t believe they are counted in the right place or that they are counted at all.

At a time of increasing distrust in the census, the Bureau should ensure that communities view its data as accurate and feel represented in the process and final results.

The most accurate and equitable solution is to allow incarcerated people to self-enumerate.

Officials at the Census Bureau may worry about how they would implement a change to count incarcerated people in their home communities. The good news for them is they already have a process in place that they use for all other people in the U.S.: simply allow incarcerated people to fill out their census forms themselves — or, as the Bureau calls it, “self-enumeration ” — rather than relying on the facility administrators to provide the date for them. In fact, this is a solution that was already part of the Census Bureau’s plan for the 2020 Census, giving more facility types this option than in 2010.⁷

States that currently reallocate incarcerated people back home in their redistricting data must now rely on their internal facility or correctional agency records to access home address information. These records are sometimes incomplete. By allowing incarcerated people to self-enumerate, the Bureau would provide states with more accurate and complete data and alleviate state officials of a significant burden.

Not only would this change allow incarcerated people to provide their addresses directly, it would also improve race and ethnicity reporting. Currently, correctional facilities don’t report race and ethnicity data in accordance with census standards. For example, in 2020, correctional facilities failed to report Hispanic status for over a quarter of the people they reported. Moving to self-enumeration would mean a more complete and accurate picture of the demographics of people who are incarcerated.

Of course, despite this, there would still be some non-responses for various reasons, just as people on the outside fail to participate in the census for various reasons. In these circumstances, facility records could still be used to fill in any blanks left after everyone had the opportunity to self-enumerate, ensuring that the 2030 Census is more accurate than prior decades.

If the Bureau chooses not to allow self-enumeration to collect home address data, the Census Bureau can follow the limited-but-tested processes used by the states and rely solely on the data already in the records collected by correctional agencies. Information missing from the records can either be imputed, as the census already does for race and ethnicity data, or, as some states do, continue to count people at the facility if all other efforts fail.

The states have pioneered a variety of paths to counting incarcerated people at home, and the Census Bureau has many more tools, giving it the ability to go far beyond state efforts and create an even more accurate enumeration. At the very least, the Census Bureau can replicate state efforts to count incarcerated people at home.

If the Census Bureau wants to be accurate, it should try something different: count incarcerated people at home

The Bureau’s current way of counting incarcerated people isn’t working. Not only does it dramatically distort political representation, but persistent errors in this data create redistricting chaos far outside the prison walls.

Counting people at home would produce census data that is, on the whole, a more accurate reflection of the population. There are concerns that counting people at home will lead to some inaccuracies because of missing home address data in facility records or non-response from incarcerated people asked to fill out the census themselves. However, the current methods used by the Census Bureau have already resulted in serious inaccuracies in the data. By attempting to count incarcerated people at home, the Census Bureau would be abandoning a seriously flawed status quo and aiming for a more accurate census.

Counting incarcerated people at home is crucial to ensuring equal representation, particularly for communities disproportionately impacted by mass incarceration. In fact, nearly half the US population is now living in a place that has rejected the Bureau’s redistricting data in order to avoid prison gerrymandering. It is time for the Bureau to count incarcerated people at home in order to create a more accurate 2030 Census.