Skip Navigation

Justin Levitt Speaks on Alternative Voting Systems and Redistricting Consequences of Incarceration

Justin Levitt speaks to the International Municipal Lawyers’ Association in Columbia, SC on alternative voting systems and the count of incarcerated persons in redistricting.

Published: December 11, 2009

  Hot Voting Rights Topics for Muni­cip­al­it­ies:

Pre-litig­a­tion Use of Altern­at­ive Voting Systems and
Redis­trict­ing Consequences of Incar­cer­a­tion

Justin Levitt, Coun­sel

Bren­nan Center for Justice at NYU School of Law 

Decem­ber 10, 2009*

Down­load Power­Point


Thank you for the oppor­tun­ity to occupy a bit of your lunch time.  My name is Justin Levitt, and I am an attor­ney at the Bren­nan Center for Justice at NYU School of Law.  The Bren­nan Center is a non-partisan organ­iz­a­tion that unites schol­ars and advoc­ates in pursuit of a vision of inclus­ive and effect­ive demo­cracy.  The Center’s Demo­cracy Program promotes policy that elim­in­ates barri­ers to full and equal polit­ical parti­cip­a­tion and that fosters respons­ive and respons­ible governance.  

In partic­u­lar, the Bren­nan Center has been at the fore­front of research on redis­trict­ing proced­ures.  We have extens­ively stud­ied redis­trict­ing prac­tices nation­wide, analyzed both success­ful and unsuc­cess­ful attempts at redis­trict­ing reform, and produced mater­i­als to educate legis­lat­ors and the public at large about the bene­fits and consequences of vari­ous redis­trict­ing meth­ods.  We have test­i­fied with respect to proposed redis­trict­ing legis­la­tion, and assisted advoc­ates and elec­ted offi­cials in draft­ing such legis­la­tion.  In addi­tion, we have parti­cip­ated as amici curiae in many of the major cases address­ing the use of redis­trict­ing for partisan gain or at the expense of minor­ity voters.

In this capa­city, we have stud­ied not only the main­stream of redis­trict­ing — the meat and pota­toes of the menu — but also many of the intriguing efforts crop­ping up at the margins.  I would like to talk to you about two of those topics today, to season lunch with a bit of food for thought about altern­at­ives and options that may be useful to you in the upcom­ing redis­trict­ing cycle. 

Altern­at­ive voting systems

The first of those topics is actu­ally an altern­at­ive to redis­trict­ing entirely: a way for local juris­dic­tions to promote minor­ity voting oppor­tun­it­ies and to comply with the require­ments of the Voting Rights Act, without draw­ing district lines. 

At-large districts

At present, many muni­cip­al­it­ies, school boards, and similar govern­ment bodies — perhaps includ­ing those that you repres­ent — conduct elec­tions using an at-large system.  In a basic incarn­a­tion, for example, if a city elects five coun­cil­mem­bers at the same time, each eligible resid­ent would be able to cast one vote for each of five seats, across the city.            

As you know, there are some reas­ons to favor at-large voting.  It can foster elec­ted offi­cials who are able to speak for and are account­able to the polit­ical unit as a whole, and not merely to one geographic subdi­vi­sion.  It can broaden the pool for candid­ates, with no contender deterred from running for office because of resid­ence in the wrong district.  And most notably, it avoids the cent­ral redis­trict­ing prob­lem: accom­mod­at­ing multiple compet­ing interests in draw­ing lines that alloc­ate polit­ical power once per decade, by placing resid­ent voters into prede­ter­mined groups.

However, at-large districts can also suppress minor­ity voices within a juris­dic­tion.  When each resid­ent is able to cast one vote for each of five seats, across the city, cohes­ive major­ity voting blocs reli­ably win every single seat.  Even when cohes­ive minor­it­ies repres­ent 30, 40, or 49.9% of a city’s voting popu­la­tion, the at-large struc­ture described above will reli­ably deliver all five legis­lat­ive seats to the city major­ity.

For racial and language minor­it­ies, the federal Voting Rights Act protects against this sort of vote dilu­tion.  Much of the rest of the day is devoted to the nuances of Voting Rights Act liab­il­ity, and I do not want to repeat what you have already heard or pree­mpt what awaits after lunch.  I’ll emphas­ize only that the Voting Rights Act prohib­its elect­oral struc­tures dilut­ing the voting power of minor­it­ies who might other­wise have an oppor­tun­ity to elect their repres­ent­at­ives of choice.  For muni­cip­al­it­ies with a sizable minor­ity popu­la­tion and polar­ized voting along racial lines, the default means to provide such elect­oral oppor­tun­it­ies within a city’s legis­lat­ive struc­ture, and the default means to comply with the Voting Rights Act, is to divide the muni­cip­al­ity into districts.  With districts drawn specific­ally to provide effect­ive elect­oral oppor­tun­it­ies to racial minor­it­ies, a cohes­ive minor­ity group with 30, 40, or 49.9% of the popu­la­tion will no longer find itself shut out of the city coun­cil.

Districts, however, are not the only means to provide racial and language minor­it­ies with effect­ive elect­oral oppor­tun­it­ies.  In some circum­stances — and it is import­ant that it is only true in some circum­stances — a muni­cip­al­ity may be able to main­tain an at-large district, and instead change its voting rule, in order to ensure that minor­ity votes are not diluted.  Juris­dic­tions across the coun­try are discov­er­ing, or redis­cov­er­ing, these sorts of altern­at­ive voting systems to provide for minor­ity voters.[I] 

I would like to briefly sketch three of the prin­cipal altern­at­ive voting systems that you may wish to consider, as you look to ensure minor­ity voting oppor­tun­it­ies in the next decade.  In the right envir­on­ment, each is able to empower protec­ted minor­ity groups, and create the means for such groups to mean­ing­fully parti­cip­ate in the elec­tion process.

Cumu­lat­ive voting

The first altern­at­ive method, cumu­lat­ive voting, is a prac­tice well known in corpor­ate elec­tions.[ii]  Many Amer­ican corpor­a­tions use this mech­an­ism to ensure a voice for minor­ity share­hold­ers, and it can bring similar salut­ary effects to a muni­cipal elec­tion. 

Both cumu­lat­ive voting and tradi­tional voting in an at-large elec­tion allow voters to cast as many votes as there are seats up for elec­tion, and in both cases the candid­ates with the most votes win.  But while tradi­tional voting in an at-large elec­tion gives each elector one vote per seat, cumu­lat­ive voting allows citizens to cast multiple votes for a given candid­ate for a given seat.  So, return­ing to our example juris­dic­tion with five seats up for elec­tion, a tradi­tional voting struc­ture would allow a citizen to cast at most one vote for each of the five seats.  A cumu­lat­ive voting struc­ture would give our voter five votes to alloc­ate as she wished.  She could choose to vote once for five differ­ent candid­ates, but she might also cast all five of her votes for the same candid­ate, or alloc­ate three votes to one candid­ate and two to another.  (A vari­ant asks the voter to choose her desired number of candid­ates, and her total votes are equally alloc­ated among the candid­ates chosen; a voter choos­ing two candid­ates in the example above would effect­ively give 2.5 votes to each.)[iii]  The winning candid­ates are determ­ined by a simple plur­al­ity, with the top over­all vote-getters winning office.  So in our example elec­tion for five seats, the candid­ates with the five highest vote totals will win the elec­tion.

Cumu­lat­ive voting allows for minor­ity oppor­tun­it­ies because it enables a voting bloc to concen­trate its votes effect­ively.  Imagine a cohes­ive and polar­ized minor­ity group compris­ing 30% of our city’s voting popu­la­tion.  Their votes are completely diluted in a tradi­tional at-large voting struc­ture.  In a cumu­lat­ive voting system, however, if members of the minor­ity group coalesce around one candid­ate, and each minor­ity voter casts all five of her votes for that candid­ate, the group would be able to ensure that its candid­ate of choice receives at least 30% of the over­all vote.  And with at least 30% of the over­all vote, the minor­ity’s candid­ate of choice would be guar­an­teed a seat on the five-member coun­cil.

 Limited voting

A second altern­at­ive voting system is known as limited voting.  In a limited voting system, each voter casts fewer votes than the number of seats up for elec­tion.  For example, in our city coun­cil elec­tion with five seats, each voter might be entitled to vote for only two candid­ates.  As with cumu­lat­ive voting, the candid­ates with the most votes win. 

Limited voting makes it more diffi­cult for the juris­dic­tion’s major­it­ies to win all of the avail­able seats, and there­fore allows cohes­ive minor­ity groups to coalesce behind certain candid­ates to gain seats in the legis­lature.  For example, if our minor­ity group with 30% of a city’s voting popu­la­tion coalesces behind the same candid­ate with one of its two limited votes, that candid­ate will be guar­an­teed a seat on the five-member coun­cil.  The more that voters are limited in the number of votes they may cast, the greater the chances that minor­ity popu­la­tions will be able to win a seat.

 Ranked choice voting

 The third altern­at­ive voting system that I will mention today is ranked choice voting, also known as pref­er­ence voting, or the single trans­fer­able vote.  Vari­ants of this system are used by several of the major sports leagues to select MVPs, and by the Academy of Motion Picture Arts and Sciences — the Academy of “I’d like to thank the Academy” — to choose Oscar winners.[iv]  In a ranked choice voting system, each voter ranks the candid­ates that she supports: next to the names on the ballot, voters indic­ate their top choices, in order of pref­er­ence.  That is, a voter ranks her most preferred candid­ate as #1, her next choice as #2, and so on. 

Determ­in­ing the winning candid­ates in a ranked choice system takes a bit more math than the other systems I’ve described.  There are several possible algorithms, each aimed at ensur­ing that as few votes as possible are “wasted.”  Consider one formula, at work in our example city, with five seats for the city coun­cil, and 3000 voters total.  The winners are calcu­lated based on an oper­at­ive threshold of 501 — the number will change, based on the total number of seats and total number of votes.[v]  First, look to the first choices; any candid­ate clear­ing the 501-vote threshold wins a seat on the coun­cil.  Next, take the “surplus” votes, above the 501-vote threshold, for each of these winning candid­ates, and trans­fer those votes propor­tion­ally to the second-choice candid­ates on each voter’s ballot; again, any candid­ate clear­ing the 501-vote threshold wins a seat.  If there are still seats remain­ing, elim­in­ate the candid­ate with the fewest votes, and again trans­fer that candid­ate’s votes propor­tion­ally to the second-choice candid­ates on each voter’s ballot.  The process contin­ues until enough candid­ates clear the threshold to fill all five seats.[vi]

From the elec­tion offi­cial’s perspect­ive, the algorithm for count­ing votes in a ranked choice voting system is admit­tedly less intu­it­ive than for more tradi­tional systems.  Yet this algorithm is relat­ively straight­for­ward to imple­ment in prac­tice.  And from the voter’s perspect­ive, the math is far less import­ant: voters need only rank their preferred candid­ates in order.[vii]  In a ranked choice voting system, as with the others above, if minor­ity groups rank their preferred candid­ate or candid­ates toward the top of the pack, those candid­ates will also be awar­ded seats on the legis­lat­ive body.

 Eval­u­at­ing the propri­ety of altern­at­ive voting systems

Each of the altern­at­ive voting systems above has the poten­tial to trans­late minor­ity pref­er­ences to seats on the govern­ing legis­lat­ive body, without the need to draw district lines divid­ing a city into pock­ets.[viii]  Indeed, partic­u­larly if a minor­ity popu­la­tion is well-integ­rated geograph­ic­ally, or if two or more minor­ity popu­la­tions with distinct voting pref­er­ences live side by side, or if there are small pock­ets of minor­ity voters spread in non-compact fash­ion through­out the city, it may be quite diffi­cult to draw district lines that achieve equit­able minor­ity voting power.  District lines are also usually stable over the course of a decade, making it diffi­cult to reflect rapid demo­graphic change in the minor­ity popu­la­tion. In such situ­ations, altern­at­ive voting rules in an at-large elec­tion may promote minor­ity pref­er­ences in ways that are both more effect­ive and more legally secure than districts.[ix]

That said, altern­at­ive voting systems work well only in certain circum­stances, and it is as import­ant to consider their limit­a­tions as it is to consider their advant­ages.  First, each of these altern­at­ive voting systems will likely be less famil­iar to voters than the ballots to which they are accus­tomed.  This novelty need not repres­ent an insu­per­able obstacle: the concepts are straight­for­ward, partic­u­larly for limited voting (“vote for two”) or ranked choice voting (“rank your choices, 1 to 5”), and recent exit poll stud­ies have found no reason to believe that either minor­ity voters or less educated voters were confused even by cumu­lat­ive voting, which is slightly more complic­ated to explain.[x]  Still, in order to ensure that voters are not confused by the change, a city would be wise to under­take a concer­ted educa­tional effort when switch­ing systems.[xi]

Altern­at­ive voting systems are also differ­ently impacted by vari­ations in turnout than are districts.  Often, districts can be drawn with an eye to past turnout, compens­at­ing for low histor­ical turnout by draw­ing districts in which eligible minor­ity voters comprise a substan­tial super­ma­jor­ity.  The bene­fit of such an approach is that it provides a safe­guard for minor­ity popu­la­tions’ abil­ity to achieve repres­ent­a­tion if turnout contin­ues to be depressed; the down­side is that minor­ity voting power may actu­ally be diluted by such super­ma­jor­it­ies if turnout substan­tially improves, with minor­ity voters over-“packed” into hyper­con­cen­trated districts.  Altern­at­ive voting systems in at-large districts have neither the bene­fits nor the detri­ments above.  That is, if turnout falls below a certain level, minor­ity popu­la­tions may not be able to elect their preferred candid­ates, and so altern­at­ive voting systems may not be appro­pri­ate where there is reason to believe that minor­ity turnout will remain espe­cially low even when minor­ity popu­la­tions would other­wise be able to control a seat.  In juris­dic­tions where minor­ity turnout can be expec­ted to be moder­ate to large, however, altern­at­ive systems have the capa­city to more effi­ciently distrib­ute minor­ity votes, and thereby to increase their voting power.

It is easi­est to demon­strate these turnout effects with numer­ical examples.  Before assess­ing turnout, though, we need to estab­lish the effect­ive level of control over the outcome of an elec­tion.  In a distric­ted system, the effect­ive level of control is clear: if a cohes­ively voting popu­la­tion exceeds fifty percent within any given district, it will be able to win that district’s seat.  It is possible for a group to win seats at lower rates, of course, if the oppos­i­tion is split between two or more candid­ates.  The fifty percent threshold is simply an expres­sion of the threshold at which a win can be guar­an­teed, under the “worst-case” circum­stances for the group in ques­tion. 

It is simil­arly possible to calcu­late the threshold level of control — known as the “threshold of inclu­sion” — for a group to be guar­an­teed a seat in an altern­at­ive-vote system, using a similar “worst-case scen­ario” analysis.  These calcu­la­tions assume voters are both perfectly cohes­ive within their own group, and that they are perfectly stra­tegic: that is, they assume that the oppos­i­tion does not unne­ces­sar­ily split their voting power.  Below is a chart of the “threshold of inclu­sion” for differ­ent numbers of legis­lat­ive seats, under each of the altern­at­ive systems discussed above.  This means, for example, that if our example city with five city coun­cil seats chose to run its elec­tions with cumu­lat­ive voting, 16.7 percent of the popu­la­tion could band together and ensure that its chosen candid­ate would win a seat on the coun­cil, even if the oppos­i­tion was perfectly cohes­ive and perfectly stra­tegic.[xii]  To determ­ine the threshold for winning two seats in a cumu­lat­ive voting system, double the number: 33.3 percent of the popu­la­tion could vote stra­tegic­ally to win two seats on the five-seat coun­cil.[xiii] 

Threshold of inclu­sion[xiv]



Cumu­lat­ive Voting &

Ranked Choice Voting

Limited Voting

1 vote

2 votes

3 votes












































Now let us apply this threshold to an example involving a minor­ity popu­la­tion, to see how turnout affects the math.  Assume that, in a perfectly polar­ized envir­on­ment, 1000 voters of our 3000-voter juris­dic­tion are cohes­ive and stra­tegic minor­ity citizens, prepar­ing for the city coun­cil elec­tion with five seats.  Depend­ing on hous­ing patterns, it might be possible to draw three districts aimed at provid­ing minor­it­ies an oppor­tun­ity to elect candid­ates of choice.  With each of the five districts hold­ing 600 voters, it might be possible to draw the district lines such that three of these districts held 333 minor­ity voters apiece.  If each minor­ity voter turned out at the same rate as their white coun­ter­parts, the minor­ity voters would reli­ably control three seats of the five.  However, if minor­ity turnout were 80% of the white turnout, the minor­ity group would reli­ably lose each of the five seats. 

In order to accom­mod­ate these sorts of turnout concerns, oppor­tun­ity districts are often drawn with a super­ma­jor­ity of minor­ity voters.  Let’s assume, then, that our juris­dic­tion draws two oppor­tun­ity districts with 400 minor­ity voters each.  Even if minor­ity turnout were just over 50% of the white turnout, minor­it­ies would reli­ably control at least two of the five seats.  But these super­ma­jor­it­ies limit the poten­tial upside for the minor­ity group as well.  If minor­ity turnout were equal to the white turnout, minor­it­ies would still control only two of the five seats.  And if the white major­ity were less than perfectly stra­tegic, divid­ing their vote among multiple candid­ates in the other districts, minor­it­ies would still likely control only two of the five seats.

Now turn to the altern­at­ive voting systems discussed above, and assume that a ranked choice or pref­er­ence voting struc­ture is in place.  No district lines are neces­sary — the elec­tion takes place over the entire 3000-voter juris­dic­tion.  If minor­ity voters turn out at the same rate as their white coun­ter­parts, they would reli­ably control at least two of the five seats.  Under lower minor­ity turnout, the altern­at­ive systems fare less well: if minor­ity turnout were 50% of the white turnout, minor­it­ies would reli­ably control only one of the five seats.  But the altern­at­ive systems also allow the poten­tial to capture a signi­fic­ant upside: if the white major­ity were less than perfectly stra­tegic, and the minor­ity community turned out in force, it would be possible for minor­it­ies to control more than two seats.

The partic­u­lar example above high­lights perhaps the most signi­fic­ant aspect, not only of this present­a­tion, but of many topics discussed today: the import­ance of context.    Real­istic turnout projec­tions are crucial factors in decid­ing which elect­oral struc­tures most effect­ively promote minor­ity voting oppor­tun­it­ies in a partic­u­lar loca­tion.  So are the extent of polar­iz­a­tion and the geographic disper­sion or concen­tra­tion of a minor­ity community. 

Staggered seats are import­ant elements in the over­all context as well.  A staggered elec­tion struc­ture effect­ively limits the number of avail­able seats in any one at-large elec­tion; limit­ing the number of avail­able seats increases the threshold of inclu­sion.  For example, imagine that our five-seat city has been divided into districts, with district #1 designed to give minor­it­ies the oppor­tun­ity to elect their candid­ate of choice.  If elec­tions are conduc­ted for districts #1, 3, and 5 in one year, and districts #2 and 4 have elec­tions in a differ­ent year, the stag­ger­ing will not usually affect minor­it­ies’ oppor­tun­it­ies to elect a candid­ate in district #1.[xv]  If, however, the elec­tions are held at-large, and seats #1, 3, and 5 are up for office in one year, with seats #2 and 4 staggered to another year, the scen­ario changes.  Now, instead of a 5-seat elec­tion, with minor­it­ies guar­an­teed to win one seat if they can muster 16.7% of the vote, the juris­dic­tion has one 3-seat elec­tion in one year (minor­it­ies have the chance to win only if they collect at least 25% of the total vote), and one 2-seat elec­tion in the next elec­tion cycle (minor­it­ies have the chance to win only if they collect at least 33.3% of the total vote).  That staggered struc­ture will make it compar­at­ively more diffi­cult for minor­it­ies to win in an at-large district.

The limit­a­tions discussed above make it unwise to adopt at-large altern­at­ive voting systems in every case.  I do, however, urge you to add these altern­at­ives to your menu of options, if state law permits you to do so.  In the right circum­stances, altern­at­ive voting systems allow the bene­fits of an at-large struc­ture, while simul­tan­eously promot­ing effect­ive oppor­tun­it­ies for minor­ity voters.[xvi]

 In the real world

The altern­at­ive voting systems described above are not merely pie-in-the-sky polit­ical scient­ists’ ventures.  Several have an extens­ive Amer­ican pedi­gree; cumu­lat­ive voting, for example, was used for more than 100 years to elect members of Illinois’ state house of repres­ent­at­ives.[xvii]  In part because they help preserve juris­dic­tion-wide elect­or­ates while giving effect­ive voice to minor­ity popu­la­tions, they have become increas­ingly popu­lar options for muni­cip­al­it­ies. In the past twenty years, cumu­lat­ive voting has been adop­ted by more than fifty local govern­ments, and limited voting by more than thirty.[xviii]  Ranked choice voting was once more in vogue for multi-seat elec­tions than it is currently;[xix] at present, it is used to elect city coun­cil members in Aspen, CO; Cambridge, MA; Hende­r­son­ville, NC; and will be used in Minneapolis for the first time in 2009.[xx]  Other local juris­dic­tions have author­ized the use of ranked choice voting, but have not yet imple­men­ted the system.[xxi]

In addi­tion to those muni­cip­al­it­ies that have chosen to enact such altern­at­ives unpromp­ted, both limited voting and cumu­lat­ive voting have been imple­men­ted to address liab­il­ity under the Voting Rights Act.[xxii]  Several juris­dic­tions with substan­tial African-Amer­ican, Latino/Hispanic, or Amer­ican Indian minor­ity popu­la­tions — includ­ing Lake Park, Flor­ida, as recently as Octo­ber 2009 — have entered into consent decrees that estab­lished cumu­lat­ive or limited voting.[xxiii]  Just this past year, an Ohio federal court chose limited voting — one of two altern­at­ive voting systems proposed by the Euclid, Ohio, school board — as an appro­pri­ate remedy for a Section 2 viol­a­tion,[xxiv] and a decision is currently pending in a New York federal court regard­ing the Village of Port Chester’s efforts to adopt a cumu­lat­ive voting scheme as a remedy for its Section 2 liab­il­ity.[xxv] 

The cases above all show the use of altern­at­ive voting systems as respons­ive meas­ures, in response to litig­a­tion.  With the Census around the corner, I would like to suggest a differ­ent approach.  Some of you may repres­ent juris­dic­tions with a minor­ity popu­la­tion about to achieve signi­fic­ant numbers for the first time.  Others may repres­ent areas covered by section 5 of the Voting Rights Act like Calera, Alabama, where the minor­ity popu­la­tion has dwindled or become less geograph­ic­ally concen­trated over the past decade, making it more diffi­cult to draw districts that avoid retro­gres­sion.[xxvi]  You will no doubt be watch­ing the Census care­fully, to see whether demo­graphic change augurs change to your elect­oral struc­ture.  In that context, to the extent permit­ted by state law, I suggest adding altern­at­ive voting systems to the menu of options you should be consid­er­ing to promote minor­ity voting oppor­tun­it­ies well before the Depart­ment of Justice comes call­ing.  Under the right circum­stances, they can be very effect­ive means to repres­ent minor­ity voices, without sacri­fi­cing the bene­fits of an at-large struc­ture. 

Pris­ons and redis­trict­ing

Thus far, our discus­sion has concerned altern­at­ives to draw­ing district lines — but many of you have determ­ined or will determ­ine that districts are the most appro­pri­ate struc­ture in your juris­dic­tion.  I have a “hot voting rights topic” for you, too.   It concerns the distor­tions in the redis­trict­ing process that result from the way in which incar­cer­ated persons are tallied.

The status quo

 At present, incar­cer­ated indi­vidu­als are coun­ted by the Census at the blocks where they are incar­cer­ated, rather than at their prior addresses, in their home communit­ies.  These two addresses are usually far from each other, and coupled with the nation’s rising incar­cer­a­tion rate, lead to a system­atic distor­tion of the popu­la­tion picture.

When local redis­trict­ing decisions rely directly on the Census tallies, they build this distor­tion into the distri­bu­tion of local demo­cracy.  Districts are construc­ted on the backs of “ghost voters,” pack­ing in pris­on­ers who count toward the district size but who, with few excep­tions, are not permit­ted to vote, and who are not connec­ted to the other resid­ents of the district.   This arti­fi­cially inflates the polit­ical power of voters in prison districts, and arti­fi­cially deflates the power of voters every­where else, so that some votes are worth far more than others.[xxvii] 

For example, in Lake County, Tennessee, after the 2000 Census, 88% of the repor­ted popu­la­tion of one County Commis­sioner district was allot­ted to a local correc­tional facil­ity.  As a result, each resid­ent in this district has more than 8 times the voting power of every other resid­ent in the county. And in Anamosa, Iowa, 1300 of the 1400 indi­vidu­als allot­ted to City Coun­cil ward 2 are incar­cer­ated there, giving the 100 voters in that ward massively dispro­por­tion­ate polit­ical power.  Indeed, in districts so distor­ted by the prison popu­la­tion, candid­ate recruit­ment is diffi­cult and turnout suffers greatly; only two voters of the 100 eligible in Anamosa’s ward 2 actu­ally bothered to vote for the single candid­ate on the ballot in 2006.[xxviii]  

 Local redis­trict­ing altern­at­ives

 Rather than count­ing pris­on­ers where they are incar­cer­ated, the most appro­pri­ate policy for redis­trict­ing purposes is to count pris­on­ers where they lived before their incar­cer­a­tion, which is where virtu­ally all pris­on­ers return when they are released.  This more accur­ately accounts for prison popu­la­tions in the communit­ies to which they have an actual connec­tion.

Some of the juris­dic­tions that you repres­ent may well have the admin­is­trat­ive capa­city to determ­ine pris­on­ers’ last addresses before incar­cer­a­tion, and the legal author­ity to adjust the census count for local redis­trict­ing purposes.  I heart­ily encour­age you to seek this data and adjust the count accord­ingly.

In other muni­cip­al­it­ies, acquir­ing the last address of incar­cer­ated indi­vidu­als may entail substan­tial logist­ical complex­it­ies.  In such circum­stances, I recom­mend an altern­at­ive halfway meas­ure that corrects a sizable portion of the skew.  The way that prison popu­la­tions are currently tallied creates two distor­tions: count­ing incar­cer­ated indi­vidu­als where they are imprisoned and refus­ing to count pris­on­ers in their home communit­ies.  Even if it is not imme­di­ately possible to determ­ine the home addresses where incar­cer­ated indi­vidu­als should be coun­ted for redis­trict­ing purposes, it is possible to elim­in­ate half of the skew by ensur­ing that pris­on­ers are not coun­ted where they are incar­cer­ated.  Instead, incar­cer­ated popu­la­tions would be tallied as part of a special precinct not tied to any specific geograph­ical loca­tion for redis­trict­ing purposes.  This is akin to the way that the Census Bureau tallies members of the milit­ary and federal govern­ment person­nel who are over­seas when the Census is conduc­ted: such indi­vidu­als are assigned to the state whence they came for congres­sional appor­tion­ment purposes, but not to a specific address within that state that would affect redis­trict­ing.[xxix]

Remov­ing this half of the skew would not completely resolve the prison distor­tion by recon­nect­ing the incar­cer­ated popu­la­tion to the communit­ies to which they most tangibly belong. But it would take a signi­fic­ant step toward resolv­ing the mass inequit­ies in voting power occa­sioned by districts in which the major­ity of the popu­la­tion is imprisoned.  And this step would bene­fit all of the remain­ing voters in the juris­dic­tion, whose polit­ical power is presently diluted in favor of the relat­ively small number of voters in prison districts.

As with the altern­at­ive voting systems discussed earlier, adjust­ing the Census count in this fash­ion for local redis­trict­ing purposes may be unusual, but it is hardly unpre­ced­en­ted.  Several juris­dic­tions, includ­ing about 1/3 of the counties with pris­ons in New York State, have excluded pris­on­ers from the popu­la­tion figures used to conduct local legis­lat­ive appor­tion­ment or redis­trict­ing.[xxx]  And where state law does not preclude local adjust­ment to Census determ­in­a­tions for redis­trict­ing purposes, other juris­dic­tions have sought and received approval by their state attor­neys general for permis­sion to do the same.[xxxi]  Indeed, some state law mandates such an adjust­ment.  Color­ado, for example, requires its county commis­sion districts to exclude incar­cer­ated popu­la­tions for redis­trict­ing purposes.[xxxii]  Missis­sippi law appears to correct the skew entirely; a 2002 opin­ion by the Missis­sippi Attor­ney General, for example, appears to require local juris­dic­tions in that state to count incar­cer­ated persons in their former communit­ies when redis­trict­ing.[xxxiii]

 Advocacy for state adjust­ment

 In addi­tion to the local redis­trict­ing correc­tion above, your juris­dic­tions may have a role in advoc­at­ing for a correc­tion to the redis­trict­ing prac­tice statewide.  In many areas, the sizable prison insti­tu­tions are located in more rural portions of the state.  This means that the current prac­tice of tally­ing pris­on­ers where they are incar­cer­ated gener­ally inflates the district size and polit­ical power of rural popu­la­tions across the coun­try, at the expense of small, mid-size, and larger urban communit­ies.

For example, in the Bren­nan Center’s home state of New York, 91% of pris­on­ers from New York City are housed outside the city. And 66% of the state’s pris­on­ers as a whole are from New York City.  The fail­ure to account for these shifts in redis­trict­ing deprives New York City of the repres­ent­a­tion in the state legis­lature to which it should be entitled.  Indeed, if incar­cer­ated popu­la­tions were tallied in the communit­ies to which they are tied, New York City would gain substan­tial repres­ent­a­tion, draw­ing from seven New York state legis­lat­ive districts that would be too under­pop­u­lated to exist.  Even if pris­on­ers were not coun­ted in their home communit­ies, simply elim­in­at­ing the present skew would redis­trib­ute the repres­ent­a­tion unduly lodged in those seven districts, to communit­ies all around the state.

Few cities, of course, would be affected by a redis­trict­ing recon­fig­ur­a­tion of incar­cer­ated persons in as extreme a manner as would New York City.  Yet your muni­cip­al­it­ies are likely also losing repres­ent­a­tion based on the way pris­on­ers are tallied in the redis­trict­ing process.  Moreover, given the heav­ily minor­ity compos­i­tion of the communit­ies from which most incar­cer­ated popu­la­tions are drawn, and given the demo­graph­ics of voters in most districts where pris­ons are located, this repres­ent­a­tional skew is likely work­ing to dilute minor­ity votes.  The issue was mentioned in a 2006 Voting Rights Act case in the Second Circuit, and though it has not been fully litig­ated, it repres­ents a plaus­ible cause of action on the hori­zon.[xxxiv]

Some state legis­lat­ors and exec­ut­ives have expressed interest in correct­ing the prison-count skew for state legis­lat­ive districts, and bills have been filed this year in at least five states to accom­plish that end, in whole or in part.[xxxv]  Just as indi­vidual muni­cip­al­it­ies have the author­ity to adjust the census count for local redis­trict­ing purposes, so too may states choose to correct the distor­tion caused by prison popu­la­tions for purposes of state legis­lat­ive redis­trict­ing.  Many of you may have an interest in advoc­at­ing for such a correc­tion on behalf of your muni­cipal clients, to return to them the repres­ent­a­tion to which they should right­fully be entitled.  Indeed, to the extent that the skew unduly dilutes the votes of racial minor­it­ies, correct­ing the distor­tion may well be required by the Voting Rights Act.  I encour­age you to raise the issue with the respons­ible state author­it­ies, before the redis­trict­ing process begins for the next cycle.


Some of the cities that you repres­ent have long had sizable minor­ity popu­la­tions; else­where, the 2010 census will confirm that a signi­fic­ant minor­ity popu­la­tion has recently emerged.  In the months ahead, you will likely be asked to provide advice on the import of these demo­graphic trends for ensur­ing minor­ity enfran­chise­ment, and for avoid­ing the wrath of the Justice Depart­ment. 

In provid­ing this advice to your client juris­dic­tions, I urge you to consider the two topics I have briefly intro­duced today, to the extent state law allows you to add them to your toolkit.  First, in addi­tion to eval­u­at­ing the use of districts or redraw­ing district lines to ensure minor­ity oppor­tun­it­ies, consider the poten­tial for altern­at­ive voting systems to provide oppor­tun­it­ies for minor­ity voters within an at-large struc­ture, partic­u­larly where minor­ity popu­la­tions are geograph­ic­ally diffuse or in small pock­ets that make compact districts diffi­cult to draw.  Second, if you find that districts are more appro­pri­ate, consider design­ing those districts with prison popu­la­tions tallied either as part of their home communit­ies, or in non-geograph­ical precincts, so that the districts with pris­ons do not enjoy arti­fi­cially inflated voting power, at the expense of voters through­out the remainder of the city.  And I suggest that you further consider advocacy with state decision­makers, to reflect similar correc­tions in statewide redis­trict­ing proced­ures, to ensure that your client juris­dic­tions are allot­ted the repres­ent­a­tion to which they should be entitled.

Thank you, once again, to IMLA for allow­ing me to bring these topics to your atten­tion, and to all of you for your parti­cip­a­tion in today’s very import­ant confer­ence.  I am happy to take any ques­tions that you may have; if you wish to ask ques­tions beyond today’s gath­er­ing, I can best be reached by email at justin.levit­





* Note: This present­a­tion was prepared and submit­ted on Novem­ber 2, 2009.  Several of the cases or elec­tions described in this draft as pending have since been resolved.

[I] For more inform­a­tion on these altern­at­ive voting systems, see gener­ally, e.g., South­ern Center for Stud­ies in Public Policy & Center for Voting and Demo­cracy, Full Repres­ent­a­tion: Elec­tion Systems Manual 11–22 (1999) [here­in­after Full Repres­ent­a­tion]; Steven J. Mulroy, Altern­at­ive Ways Out: A Remedial Road Map for the Use of Altern­at­ive Elect­oral Systems as Voting Rights Act Remed­ies, 77 N.C. L. Rev. 1867 (1999); Lani Guinier, The Tyranny of the Major­ity: Funda­mental Fair­ness in Repres­ent­at­ive Demo­cracy (1994); Todd Donovan with Heather Smith, Propor­tional Repres­ent­a­tion in Local Elec­tions (1994), at­tions.pdf; Richard L. Engstrom, Modi­fied Multi-Seat Elec­tion Systems as Remed­ies for Minor­ity Vote Dilu­tion, 21 Stet­son L. Rev. 743 (1992).

[ii] See gener­ally Anupam Chander, Minor­it­ies, Share­holder and Other­wise, 113 Yale L.J. 119 (2003); Jeffrey N. Gordon, Insti­tu­tions as Rela­tional Investors: A New Look at Cumu­lat­ive Voting, 94 Colum. L. Rev. 124 (1994); Lani Guinier, No Two Seats: The Elusive Quest for Polit­ical Equal­ity, 77 Va. L. Rev. 1413 (1991).

[iii] Full Repres­ent­a­tion, supra note 1, at 11.

[iv] See Press Release, Pref­er­en­tial Voting Exten­ded to Best Picture on Final Ballot for 2009 Oscars, Aug. 31, 2009, at­releases/2009/20090831a.html.

[v] The threshold described here, known as the “Droop quota,” is calcu­lated by divid­ing the total number of votes (3000) by one more than the total number of seats (6), and then adding one.  Other aggreg­a­tion algorithms use slightly differ­ent threshold formu­las.

[vi] If voters are not required to rank all of the candid­ates, it may be that none of the last remain­ing candid­ates clear the threshold.  In such cases, the remain­ing seat or seats are usually awar­ded to the candid­ates with the most votes.

[vii] Also, from polit­ical groups’ perspect­ive, ranked choice voting renders unne­ces­sary stra­tegic choices about the optimal number of candid­ates; in other voting systems, includ­ing tradi­tional at-large voting, running “too many” aligned candid­ates can create a prob­lem by caus­ing support­ing groups to divide their vote. 

[viii] See Full Repres­ent­a­tion, supra note 1, at 29–30, 34 (tabu­lat­ing minor­ity voter percent­age and minor­ity candid­ate wins in juris­dic­tions with altern­at­ive voting systems).

[ix] Two strands of Supreme Court juris­pru­dence have raised seri­ous ques­tions about districts drawn in order to promote voting oppor­tun­it­ies for racial minor­ity popu­la­tions that are not “compact” — a concept that has not been well-defined in the courts, but that essen­tially refers to popu­la­tions concen­trated fairly closely together.  League of United Latin Amer­ican Citizens v. Perry, 548 U.S. 399 (2006), held that section 2 of the Voting Rights Act protects only compact popu­la­tions of racial minor­it­ies.  See id. at 430–35; see also Thorn­burg v. Gingles, 478 U.S. 30, 50–51 (1986).  And a series of cases begin­ning with Shaw v. Reno, 509 U.S. 630 (1993), has held that there may be consti­tu­tional diffi­culties with draw­ing districts predom­in­antly to promote the voting power of racial minor­it­ies, if such districts are not required by the Voting Rights Act.  For minor­ity popu­la­tions that are not compact, altern­at­ive voting systems in an at-large district can provide effect­ive minor­ity voting oppor­tun­it­ies without creat­ing these consti­tu­tional ques­tions.

[x] Richard L. Engstrom & Robert R. Brischetto, Is Cumu­lat­ive Voting Too Complex?  Evid­ence from Exit Polls, 27 Stet­son L. Rev. 813, 821‑26 (1998).

[xi]  In 1994, the Depart­ment of Justice objec­ted to a preclear­ance request under section 5 of the Voting Rights Act for a cumu­lat­ive voting system by the city of Morton, Texas, because it had not taken suffi­cient steps to explain the cumu­lat­ive voting struc­ture to minor­ity voters.  Hear­ing on H.R. 1173, the “States’ Choice of Voting Systems Act”, Before the Subcomm. on the Const. of the H. Comm. on the Judi­ciary, 106th Cong. (Sept. 23, 1999) (state­ment of Anita Hodgkiss, Deputy Asst. Att’y Gen., Civil Rights Div.,  Dep’t of Justice).  After plans for voter educa­tion had improved, the cumu­lat­ive voting system was precleared.  Id.



[xii] The formula to determ­ine the threshold for winning a seat in a cumu­lat­ive or ranked choice voting system is fairly straight­for­ward:


1 + (number of seats up for elec­tion)

[xiii] It is more complic­ated to determ­ine thresholds for winning multiple seats in a limited voting system.  Rob Richie has developed a formula called the “Code to Win,” to calcu­late the thresholds involved.  Email from Rob Richie, Exec­ut­ive Director of Fair­Vote, to Justin Levitt (Oct. 25, 2009).  If s is the total number of seats up for elec­tion, t is the number of seats that the cohes­ive group is target­ing, and v is the number of votes that each voter is allowed, the Code to Win is as follows:

max ( t , v )

max ( s+1, 2v, t+v, 1+s+v–t  )

[xiv] Mark E. Rush & Richard L. Engstrom, Fair and Effect­ive Repres­ent­a­tion? Debat­ing Elect­oral Reform and Minor­ity Rights 42 (2001).

[xv] The caveat is that the years of stag­ger­ing must be years with similar turnout, with simil­arly motiv­at­ing races at the top of the ticket.  Stag­ger­ing races so that districts designed for minor­ity oppor­tun­it­ies have elec­tions only in low-turnout years, partic­u­larly if minor­ity voters are less likely to turn out than white voters, increases the chance that minor­ity voting power will be diluted.

[xvi] See sources cited in Engstrom & Brischetto, supra note 5, at 816 n.25.

[xvii] See Full Repres­ent­a­tion, supra note 1, at 11.

[xviii] See Shaun Bowler, Todd Donovan, & David Brock­ing­ton, Elect­oral Reform and Minor­ity Repres­ent­a­tion: Local Exper­i­ments with Altern­at­ive Elec­tions 120–122 (2003); Full Repres­ent­a­tion, supra note 1, at 8; Engstrom & Brischetto, supra note 10, at 814 n.10.

[xix] See sources cited in Mulroy, supra note 1, at 1879 n.61.

[xx] See City of Aspen, Instant Runoff Voting Proced­ures Manual (2009), at; Aspen, Colo., Home Rule Charter § 2.7; Mass. Gen. Laws ch. 43, § 96; Cambridge, Mass., Muni­cipal Code § 2.02.030; Hende­r­son­ville County Bd. of Elec­tions, Hende­r­son­ville Pilots Instant Runoff Voting (2009), at; Chris Steller, New Instant-Runoff Voting Could Factor in 22 Minneapolis Races, Minn. Inde­pend­ent, July 22, 2009, at; Minneapolis, Minn., Charter ch. 2, § 5B.

Ranked choice voting has been used more often in distric­ted or single-seat elec­tions, where it is more commonly known as instant-runoff voting; this method has partic­u­lar advant­ages for milit­ary and over­seas elect­ors who might have more diffi­culty parti­cip­at­ing in a second runoff elec­tion.  Basalt, CO; San Fran­cisco, CA; Takoma Park, MD; Cary, NC; Burl­ing­ton, VT; and Pierce County, WA are among the juris­dic­tions that have recently used ranked choice voting for single seats.  See, e.g., Basalt, Colo., Home Rule Charter § 2.8; City and County of San Fran­cisco, Cal., Muni­cipal Charter § 13.102; City and Takoma Park, Md., Muni­cipal Charter art. VI, § 606; Jordan Schrader, To Stem Runoffs, New Ballots Have Voters Rank Top 3, USA Today, Oct. 18, 2007; Vt. Stat. Ann. tit. 24 app., ch. 3, § 5; Pierce County, Wash., Charter § 4.10.

[xxi] See, e.g., Ferndale, Mich., Charter ch. IV, § 17; Sara­sota, Fla., Charter art. IX, § 5.

[xxii] Cf. Holder v. Hall, 512 U.S. 874, 897, 910 (1994) (Justice Thomas, concur­ring) (“Perhaps the most prom­in­ent feature of the philo­sophy that has emerged in vote dilu­tion cases since Allen has been the [Supreme] Court’s pref­er­ence for single member district­ing schemes, both as a bench­mark for meas­ur­ing undi­luted minor­ity voting strength and as a remedial mech­an­ism for guar­an­tee­ing minor­it­ies undi­luted voting power . . . .  But noth­ing in our present under­stand­ing of the Voting Rights Act [limits] the author­ity of federal courts [so as to] prevent them from insti­tut­ing a system of cumu­lat­ive voting as a remedy under § 2, or even from estab­lish­ing a more elab­or­ate mech­an­ism for secur­ing propor­tional repres­ent­a­tion . . . .” ).

            When courts have rejec­ted the use of altern­at­ive voting systems to remedy viol­a­tions of the Voting Rights Act, it has gener­ally been because the altern­at­ive system was imposed despite the affected juris­dic­tion’s pref­er­ence for districts.  See, e.g., Cane v. Worcester County, 35 F.3d 921, 928–929 (4th Cir. 1994); Harper v. City of Chicago Heights, 223 F.3d 593, 601–03 (7th Cir. 2000).

[xxiii] Consent Judg­ment and Decree, United States v. Town of Lake Park, Flor­ida, No. 09–80507 (S.D. Fla. Oct. 26, 2009) (limited voting); Agreed Settle­ment and Release of Claims, Reyn­oso v. Amarillo Indep. Sch. Dist., No. 98–00186 (N.D. Tex. May 17, 1999) (cumu­lat­ive voting); Moore v. Beaufort County, 936 F.2d 159, 164 (4th Cir. 1991) (limited voting); Dillard v. Town of Cuba, 708 F. Supp. 1244 (M.D. Ala. 1988) (limited voting); Dillard v. Chilton County Bd. of Educ., 699 F.Supp. 870 (M.D. Ala. 1988) (cumu­lat­ive voting); Buckanaga v. Sisseton Indep. Sch. Dist., No. 84–01025 (D.S.D. Oct. 26, 1988) (cumu­lat­ive voting); Amended Final Consent Decree and Full Resol­u­tion of Action exh. B, Banks v. City of Peoria, No. 87–02371 (C.D. Ill. Aug. 10, 1988) (cumu­lat­ive voting); Stip­u­lated Find­ings of Fact and Conclu­sions of Law, Vega v. City of Alamogordo, No. 86–0061-C (D.N.M. Mar. 2, 1987) (cumu­lat­ive voting).

[xxiv] United States v. Euclid City School Bd., 632 F.Supp.2d 740 (N.D. Ohio 2009).  The Euclid school board consisted of five members, each elec­ted in odd-numbered years to four-year staggered terms.  The African-Amer­ican voting-age popu­la­tion had grown from 27.8% in 2000, at the start of the proceed­ings, to 40.2% at the time of the remedy trial.  Id. at 746, 756–57.  Yet until 2008, no African-Amer­ican had ever been elec­ted as a school board member.

[xxv] United States v. Village of Port Chester, 2008 WL 190502 (S.D.N.Y. 2008).  Port Chester had main­tained an at-large voting system to elect its Mayor and its six-member Board of Trust­ees.  The Village’s Hispanic community had been rapidly grow­ing, and by 2000 Hispan­ics consti­tuted a substan­tial portion of the popu­la­tion — yet in ten consec­ut­ive contests for Trustee in which the Hispanic candid­ate of choice differed from the non-Hispanic candid­ate of choice, the candid­ate preferred by non-Hispanic voters won nine of the elec­tions.

[xxvi] Because the minor­ity popu­la­tion in Calera is “spread city­wide, they were no longer able to draw the [city coun­cil district] lines in a way that kept one district with a major­ity black popu­la­tion.”  Mary Orndorff, Court to Hear Case on Need for U.S. Elec­tions Over­sight, Birm­ing­ham News, Apr. 27, 2009.  Instead, they turned to an at-large elec­tion for city coun­cil, using a limited voting system that was recently precleared by the Justice Depart­ment. Scot­tie Vick­ery, Justice Clears Calera Plan for Elec­tions, Birm­ing­ham News, Sept. 26, 2009.  Of the nine candid­ates for six seats, one is African-Amer­ican.  Id.  The elec­tion will take place on Novem­ber 10, 2009, after this paper has been submit­ted.

[xxvii] See, e.g., Peter Wagner & Elena Lavarreda, Import­ing Constitu­ents: Pris­on­ers and Polit­ical Clout in Oklahoma (2009), at; John Hejduk & Peter Wagner, Import­ing Constitu­ents: Pris­on­ers and Polit­ical Clout in Wiscon­sin (2008), at; Peter Wagner & JooHye Della­Rocco, Phantom Constitu­ents in Tenness­ee’s Boards of County Commis­sion­ers (2008), at; Peter Wagner et al., Phantom Constitu­ents in the Empire State (2007), at

[xxviii] Maur­een Turner, The Prison Town Advant­age, Valley Advoc­ate, Oct. 8, 2009.

[xxix] National Research Coun­cil, Once, Only Once, and in the Right Place: Resid­ence Rules in the Decen­nial Census 327–28, 330–31 (Daniel L. Cork & Paul R. Voss, eds. 2006); Frank­lin v. Massachu­setts, 505 U.S. 788, 805–06 (1992).  Other over­seas citizens, and members of U.S. flag merchant vessels docked abroad, are not coun­ted in the Census at all, even if their iden­tit­ies are known.  National Research Coun­cil, supra at 301.

[xxx] Letter from Daniel Jenkins et al. to U.S. Census Bureau Director (July 9, 2004), avail­able at http://www.pris­on­er­­cit­izens/.  Peter Wagner has assembled a running list of local juris­dic­tions that exclude incar­cer­ated popu­la­tions for redis­trict­ing purposes, at http://www.pris­on­er­ (last updated Oct. 6, 2009). 

[xxxi] See, e.g., Op. No. 91–601, 74 Ops. Cal. Atty. Gen. 162, 1991 WL 495473 (Cal. A.G. 1991).

[xxxii] Colo. Rev. Stat. §§ 30–10–306, 30–10–306.7.

[xxxiii] Op. No. 2002–0060, 2002 WL 321998 (Miss. A.G. 2002).

[xxxiv] The case, Hayden v. Pataki, 449 F.3d 305 (2d Cir. 2006), concerned a direct chal­lenge to the prac­tice of felony disen­fran­chise­ment under the Voting Rights Act.  The court, sitting en banc, dismissed the chal­lenge at hand, but remanded to consider whether the plaintiffs had stated a vote dilu­tion claim based on redis­trict­ing that counts incar­cer­ated popu­la­tions where they are incar­cer­ated.  Id. at 328–29.  In the trial court, plaintiffs clari­fied that they had not stated such a claim, and the case was dismissed.  Memor­andum and Order, Hayden v. Pataki, No. 00–8586, 2006 WL 2242760 (S.D.N.Y. Aug. 4, 2006).

[xxxv] See, e.g., H.B. 4650, 96th Gen. Assem. (Ill. 2009); S. 1633, 2009 Leg. (N.Y. 2009); H.B. 2930, 75th Leg. Assem. (Or. 2009); H.B. 2855, 81st Leg. (Tex. 2009); Assem. J. Res. 63, 99th Leg. (Wis. 2009). Cf. H.B. 5969, 2009 Leg., Jan. Sess. (Conn. 2009) (adjust­ing popu­la­tion data to count incar­cer­ated persons in their home communit­ies for fund­ing purposes).