Peter Wagner, Executive Director
I need your help. For more than a decade, the Prison Policy Initiative has been at the forefront of the movement to expose how mass incarceration undermines our national welfare. With a lot of hard work and generous support from a small network of individual donors, we've won major civil rights victories in local governments, state legislatures and even the Supreme Court. But our long-term viability depends on people like you investing in our work.

Can you stand up for smart and effective justice policy by joining our small network of donors today? You can make a one-time gift, or even become one of our sustaining monthly donors.

I thank you for your investment in our work towards a more just tomorrow.
... (read more) (read less)

“Address Unknown” — Podcast Episode #1

by Peter Wagner, May 20, 2010  


Host: Peter Wagner, Executive Director, Prison Policy Initiative

Guest: Justin Levitt, Counsel, Brennan Center for Justice
May, 2010


Peter Wagner:

Welcome to Issues in Prison-Based Gerrymandering: a podcast about keeping the Census Bureau’s prison count from harming our democracy.

The Census Bureau counts people in prison as if they were actual residents of their prison cells, even though most state laws say that people in prison are residents of their homes.

When prison counts are used to pad legislative districts, the weight of a vote starts to differ. If you live next to a large prison, your vote is worth more than one cast in districts without prisons. Prison-based gerrymandering distorts state legislative districts, and has been known to create county legislative districts that contain more prisoners than voters.

On each episode we’ll talk with different voting rights experts about ways in which state and local governments can change the Census and avoid prison-based gerrymandering.

Our guest today is Justin Levitt from the Brennan Center for Justice, here to talk with us about how states that wish to change where people in prison should be counted should structure their bills.

Justin Levitt:

Yes, thank you very much for inviting me to this chat. I am an attorney, with the Brennan Center for Justice at NYU School of Law. We are part think tank, part law firm. We work on all kinds of issues of fundamental democracy and justice and my own particular bailiwick is election issues and voting rights issues and particular relevant here, redistricting issues, how local, state, federal districts get drawn and whether they’re drawn as districts should be.

Peter Wagner:

And so one issue that my organization the Prison Policy Initiative and then yours the Brennan Center have been working on for much of this last decade is this question of prison-based gerrymandering. And of course we’ve been working together to try to change how the Census Bureau counts people in prison. But we’ve also been talking about various ways that state and local governments can change the census count. And you’ve had an interesting idea about the best way that these reforms, these state or local level reform bills should be structured. And I was hoping that we could talk about that today.

Justin Levitt:

Sure. I mean, this all comes back to, you’re right, the work that we’ve been doing for the better part of the decade, and in particular, the recognition that when people in prison are counted for redistricting purposes in prison, it really sets up a democratic skew, it skews the democratic system. Vast majority of time, people who are imprisoned are moved into that prison from their home communities, they have very little contact with invididual citizens on the outside of the prison, and they’re not in any way meaningfully represented, which is what districting is all about, by the officials who sit where their district is. They’re used instead as filler in order to plump up the size of districts where the prisons are in a way that really deprives every other citizen of the area, whether it’s the town or the county or the state, of adequate representation. And the biggest point of this issue is not whether or not these people should be counted, everyone should be counted, it’s a question of where.

The default rule that the Census uses is really just a default rule, and it says, we’re going to count you where you lay your head. But in the case of prisoners, we actually know much better where they actually live, where their residence is. The vast majority of states say that the legal residence of prisoners is where they lived before they were incarcerated, and that decision on legal residence affects things that you’ve documented extremely well, Peter. Things like, where you can file for marriage or where you can file for divorce, where your kids get childcare, where they’re part of the school system. It affects all sorts of parts of these individuals’ lives. And the uniform answer is, until a person in prison actually makes an affirmative decision to change their residence, their legal residence, where everybody considers their home to be, is where they used to live before they were incarcerated.

When jurisdictions, whether it’s a local, town, or county, or whether it’s statewide, want to correctly represent the population, when they want to correct the skew that the Census Bureau’s default rule puts in place, and count people for representational purposes where they live, then the bill should represent that, or the whatever structure it is, the governmental decision to correct that skew, should represent that fundamental question of where a person should rightly be counted. And in particular, as we’ve talked about designing legislation, as we’ve talked about designing executive action, as we’ve talked about local jurisdictions or statewide jurisdictions fixing this problem, we’d settled on a structure that hinges entirely on this “where” question. It says, essentially, if you can get the information, if you can collect it administratively, or from the individuals in prison, or by whatever means, about where the individual came from before they were incarcerated, then that’s where you ought to count them for the Census purposes, just like all other purposes. And if you can’t get that information, for whatever reason, and some areas may have administrative difficulty in figuring out where any individual resided before they moved to prison, if you can’t get that information, then you tally those individuals like any other individual when you don’t know exactly where they live.

So for example, military personnel who are overseas, we count them in the Census, and we use that information to apportion representation in Congress, Federal representation. So if somebody from North Carolina is stationed overseas, North Carolina doesn’t get dinged for that, that person counts as a North Carolinian, even though they happen to be living for the moment somewhere outside the country. But the Census Bureau doesn’t know, more accurately, where that individual lives within North Carolina. They know they’re in North Carolina, but they don’t have a specific other residence to tally them. And so they count that individual as a North Carolinian, but they don’t assign them to a particular point within North Carolina for purposes of redistricting the state, or for purposes of redistricting local towns or counties. And that’s the same principle that jurisdictions should use when they don’t know what the real correct residence of a prisoner within their care is.

Peter Wagner:

If you wouldn’t mind maybe just stepping through the two ways this could work and say like a state like Oregon, which is very interested in addressing prison-based gerrymandering, but the legislature has ended its session and won’t come back until January of 2011, where the state has… only partial records of people’s home addresses, and then potentially maybe contrast that with a state that has a very good set of electronic records of where incarcerated people come from. If you could maybe just step through how that idea would work.

Justin Levitt:

Sure, and let’s actually start with the good records state because it’s easier. So in the state with really good records about, well, let me back up. In either state, the basic principle and the underlying principle is, just like in other residency rules, that you should count people in prison where their home is, and, for the vast majority of prisoners, that’s going to be where they came from before they were incarcerated. So that’s the basic threshold rule is that you should count people where their residence is, particularly because prisoners probably uniquely in the population aren’t represented by the people who, by the representatives where the prison sits. That is, the political representatives where the prison sits aren’t looking out for the prisoners needs. And so the basic principle is, they should be represented where their home communities are.

For a state that has really good electronic records, Department of Corrections, or Department of Probation, or the court system, about where that prior address was, it’s real easy. You take the numbers as the Census comes in, you collect the information from whatever that centralized place is, and say, we have 500 people in this particular facility, or 1000 people in this particular facility, we know that two of them go to this Census block, and five of them go to that Census block, and eight of them go to this other Census block. And you just adjust the information that comes in from the Census Bureau to reflect the location of the last residents. That is, that person who would be otherwise reflected in the Census block where the prison is, should instead be reflected in the Census block where their last residence was. And that, for states that have real good information on where people in prison were from, that’s pretty easy. It’s essentially just easy math, adding one and subtracting one, or adding two and subtracting two from the Census blocks where the prison is and where the home community is. For a state that has mixed records, that doesn’t have all the records available, you start in the same place.

It’s the same basic principle that people should be counted where their home communities are, and you also start where you do have good records. So if you’ve got records for half the population in prison, then you start with that half, where the records are good, and you do the same sort of adjustment. Figure out which Census blocks the prison is in, figure out which Census block the home community is in, and when the data comes in from the Census, the thing that drives redistricting, you do the simple math adjustment, subtracting form where the Census block is for prison, and adding to the Census block where the home communities are, depending on the population of that particular facility. For the other fifty percent, where you don’t know where they’re from, you sort of do half that math. You know that it’s not right to count them in the particular location where the prison is, because that’s not home, so you subtract those individuals from the Census block count where the prison is. But you don’t know where to assign them, because you don’t have the information for that half of the population about where their prior residence is. So you either sort of hold them in the jurisdiction as a whole—if we’re talking about Oregon, you assign them to Oregon as a whole, but not to any particular Census block–or you average out that number–you say, we don’t know which Census block they’re in, so a tiny little fraction of that person is put in each Census block around the state.

And that’s essentially, like I said, the same way that we count military personnel overseas when we know they belong to a state, but we’re not sure where within the state they really belong. And that latter procedure is also pretty mathematically simple. It’s a straightforward step of figuring out which Census blocks have the prison facilities and doing the same math that you would do for the people who have addresses on record in subtracting those individuals from the Census blocks where they just don’t belong.

Peter Wagner:

And what about ideal district size, speaking of calculations? So where, what’s the impact on how ideal district size should be calculated?

Justin Levitt:

If you’re calculating the idea district size, and this is something you have to do to make sure that the districts are roughly equally populated under the constitution, then how you count the ideal districts depends on which of those two tactics I just mentioned you take. For indidividuals where you know they’re prior address, there’s no real impact on the idea district size because you’re taking them out of one Census block and assigning them directly to another, and it’s a one-to-one correspondence, and so that really doesn’t impact the idea size at all.

For people where you don’t know where they lived before they were incarcerated, if your solution is to assign them to the state as a whole but not to any particular Census block, then you’d calculate the ideal district size without that tally. So you’d calculate the ideal district size based only on the people who get sorted into districts. That makes sense. If you’re looking for an average district size for people who get sorted into districts, then you wouldn’t include the people who are assigned to the state as a whole. If instead, you think it’s mathematically more sensible, and it amounts to the same thing, if you think it’s mathematically more sensible to assign a fractional person to each Census block because you’re spreading them throughout the whole state, then in the same way, your average district size would include all of those fractions. It would be spread among the whole states, so it wouldn’t change the relative mix. If you had a district that had a certain number of people and that meant it was going to be a certain size, then adding the fractional “we don’t know where they live” people uniformly across the state is not going impact that. The district is going to stay the same size. It’s just a different way of representing the average number so that you know whether you’re within roughly equal population, which is what the Constitution asks.

And those two choices, those two ways of dealing with the math, they may sound technical, but they’re actually very simple ways of taking averages when what you’re talking about is the average number of people within a particular district. That’s the basis for the calculation and so choosing from among the people who are assigned to a district is the right way to take the average. And all of this, I should mention, is much fairer, ensures real representation, and is why we do redistricting in the first place, than the status quo, which is to take people who really belong somewhere else for representational purposes, and to put them arbitrarily where they happen to be sleeping even though their residence is miles away.

Peter Wagner:

So that seems like a pretty straightforward fix to, what has become a very old problem in the U.S. Census.

Justin Levitt:

It really is. Now the better fix, and the longer-term fix, is to go out and collect the information on where people’s prior residences are so that every state moves into the “real good records” category. That’s the right solution, is to make sure that you actually count everyone because you know exactly where they came from beforehand, that’s something that states could do on their own, that’s something that local jurisdictions could do on their own, it’s also something that, given enough time to work out the logistical questions, the Census Bureau could do when it goes and collects the information every April 1 every decade. And that would make it a lot easier to put everybody in the 100% category and the “real good records on everyone” category. And so that’s clearly the way to go in the future. But even this halfway step, of dealing with the ones where you know prior residence as best you can, and dealing with those for whom you don’t know prior residence as best you can. Even that halfway step, it’s both conceptually simple and it’s mathematically simple. It’s easy and straightforward to do. And any state that wanted to do it could do it really easily this next redistricting cycle.

Peter Wagner:

I think that’s what our listeners need to hear. Is there anything you want to add?

Justin Levitt:

Not really. There’s, the one other point is that along with all the simplicity you don’t need any sort of specialized technology to do this work. It doesn’t take a fancy system. All it takes is knowing which Census blocks or where, and the Census corresponds with people in each state so that every state’s got a repository for that so that you know, any given street address, what Census block that’s in. That’s also, I think, something that’s possible to find out on the Census’s website, once the information comes out, nice and easy. And a calculator. And really the calculator is only to add in three people here, five people there, subtract five people there, eight people there. It’s not hard math. So none of this, it’s both conceptually simple and mathematically simple and technologically simple. It doesn’t require investment of any kind, and doesn’t actually take that much time.

Peter Wagner:

Well, that sounds like a pretty important solution. So my final question for you is, so, how can other people learn more about what you do and be in touch with you and the Brennan Center?

Justin Levitt:

In addition to all the great resources on your website, Peter,, the Brennan Center has a collection of resources and has, obviously, legal support to lend to this question. The Brennan Center’s website is That’s B-R-E-N-N-A-N, named after the famous Justice Brennan, where we get our heritage. And there are many resources on the extent of this problem, the scope of this problem, some of the legal solutions, and technical solutions we’ve just talked about. There’s more information on all of that online, as well as contact information to get a hold of us if you are speaking with a local legislator or you’re speaking with someone who has the capacity to make these changes, we are more than able to help out, and draft as appropriate, and help figure out, help you figure out how to do the math as appropriate. It’s really pretty simple, but we’re around to walk you through it if you’re looking for more help.

Peter Wagner:

Well that’s really great. Thank you very much, Justin. I’m very grateful to have you on the program and I’m sure all of the people listening will find this very helpful and will be in touch. Thank you.

Justin Levitt:

I appreciate it, Peter. Thank you very much for the invitation, and thanks again for all the work that you’re doing.

Peter Wagner:

Thank you.

2 Responses

  1. Justin Levitt explains prison-based gerrymandering | Prisoners of the Census says, 1 year after publication:

    [...] more with Justin Levitt, see my podcast “Address Unknown” Podcast Episode #1 where we discuss the optimal way to structure a bill to eliminate prison-based [...]

  2. Kentucky redistricting proposal took steps to minimize prison gerrymandering | Prisoners of the Census says, 2 years, 9 months after publication:

    [...] the midst of redistricting to, as an interim solution, consider the incarcerated populations to be “at-large” inhabitants of the state rather than as residents of districts where they lack legal residence. In this way, [...]

Leave a Comment

Please note: Comments are moderated and there may be a delay before your comment appears. There is no need to resubmit your comment.