Is Groton the Next Evenwel?
Paul H. Edelman*
In Evenwel v. Abbott the Supreme Court left open the question of whether states could employ population measures other than total population as a basis for drawing representative districts so as to meet the requirement of “one person, one vote” (OPOV). It was thought that there was little prospect of resolving this question soon as no appropriate instances of such behavior were known. That belief was mistaken. In this Essay I report on the Town of Groton, Connecticut, which uses registered voter data to apportion seats in its Representative Town Meeting and has done so since its incorporation in 1957. The resulting apportionment arguably meets the requirements of OPOV as applied to registered voter data but badly fails if total population is employed. Thus, it would make a good test case to resolve some of the open questions in Evenwel.
From the start of the reapportionment revolution in the 1960s, the Supreme Court has been ambiguous about what measure of population should be used to assess “one person, one vote” (OPOV). Does OPOV guarantee equal representation, or does it guarantee equal voting power? The former would require districts of equal population, while the latter would argue for districts with an equal number of voters. The rhetoric of the Court in its OPOV cases could be used to support either of these positions.
In spite of this ambiguity, the vast majority of jurisdictions have employed total population to measure the size of their districts. But in Evenwel v. Abbott some residents tried to change that. The plaintiffs argued that the state of Texas was required to use citizen voting age population as the basis for its districting of the state senate. The Court unanimously rejected this claim and found that Texas was permitted to use total population as the basis for the districting. Still left open, though, was the question of whether the state would be permitted to use an alternative measure of population, such as registered voters or citizen voting age population.
In the aftermath of Evenwel there was considerable speculation about what the Court would do if confronted by the question of whether a state is permitted to use voting population rather than total population data. All of the speculation was hypothetical, however, in that no one had an example of a jurisdiction that clearly posed the issue. Surprisingly there is such a jurisdiction that seems to have gone unnoticed for sixty years. The Town of Groton, Connecticut has been districting its Representative Town Meeting (RTM) using registered voters since its founding in 1957.
In this Essay, I will discuss the significance of the Town of Groton’s political organization. In Part I, I explain why one should care about the governmental structure of a little town in Connecticut. In Part II, I describe how Groton is organized, how the Court would likely assess its adherence to OPOV, and why Groton presents an interesting test case that might force the Court to address some of the open questions left from Evenwel. I close with a few final remarks.
I. Who Cares About Groton?
Is there a reason, beyond its potential as a test case, to consider how Groton governs itself? I believe so. It provides insight into how registered voter data can be used in practice. It advances the discussion of the relative advantages of different methods of apportionment. Finally, it further highlights the variety of local government.
While it has been asserted that the use of registered voting data is fraught with practical problems, Groton seems to have managed just fine for sixty years. Indeed, Groton does not even retain total population information for its voting districts. This is not to say that concerns about using registered voter data are unfounded, but in at least this one instance no one seems to have a problem with it.
It is certainly possible that the use of registered voter data in a larger jurisdiction would be more problematic. But its use for sixty years in Groton opens the possibility that using registered voter data is perfectly practical for local jurisdictions even if it would pose problems for state or federal legislatures. Moreover, the use of voter data at the local level may be constitutional regardless of whether it would violate OPOV if adopted for state or federal legislative districting. Courts have routinely made the distinction between what is required for states and the federal government and what is acceptable for local jurisdictions. The Court has enforced the rigors of OPOV on virtually every general-purpose legislative body, but it has also given local governments more leeway and flexibility in meeting the requirement. It routinely allows a larger deviation in the traditional measures of OPOV for local governments than for state or federal representatives. The Court has also approved completely nonstandard legislative arrangements for local governments, such as the weighted voting schemes that are common throughout upstate New York. This indulgence of local governmental arrangements might well support using different populations to comply with OPOV at different levels of government.
One advantage of Groton’s approach is that it can reapportion the RTM every four years instead of every ten. Whatever the difficulties with the registered voter data, we know that the total population data immediately begins to degrade after the decennial census. Perhaps the more often revised but messier registered voter data is better than old total population data at the end of the ten-year cycle. I am unaware of any other jurisdiction that reapportions their legislative body this often. It seems like an interesting idea. And it seems to have been remarkably uncontroversial in Groton.
Finally, while the focus of election law scholars is on state and federal elections, Groton is more evidence that there is a diverse set of political arrangements at the local level that do not comport with traditional analysis. This is not to say that the focus on state and federal government is inordinate, but local governments might be a source of insight that has been underutilized, if not outright ignored. The study of these unusual arrangements might lead to a more nuanced analysis of when, not if, alternative interpretations of OPOV are appropriate.
II. A Good Test Case?
A. Describing a Test Case
Would Groton’s districting provide a good test case for the questions left open in Evenwel? An optimal test case would have the following properties:
- An elected legislative body that has sufficiently broad governmental power so as to be governed by OPOV,
- The districting for that body is based on some measure of the voting population, not the total population,
- The districting fails to meet OPOV on the basis of total population, and
- The districting meets the requirement of OPOV based on the chosen measure of voting population.
The first two of these conditions are obviously necessary. The third condition will ensure that a Court cannot dodge the question by claiming that the chosen method of districting leads to a result similar to one gotten by using a total population measure. The last condition would force the Court to decide something about the uses of voting population in this instance.
These conditions seem like a high bar to surmount. Nevertheless, as I will show in this Part, Groton clearly satisfies the first, second, and third of these conditions. The last, that the districting meets the OPOV requirement as measured with respect to voting population, is at least arguable although not a slam dunk. But to show how the conditions are satisfied we must investigate in considerably more detail the structure of Groton’s government.
B. The Groton Town Structure
The government of the Town of Groton has the Town Council-Town Manager-Representative Town Meeting structure, one of the structures available under Connecticut state law. The Town Council (TC), consisting of nine members elected at-large, is generally responsible for policy. They appoint a Town Manager who serves as the executive arm of the Council.
For our purposes the more significant body is the Representative Town Meeting (RTM). It is a legislative body that acts as the overseer of the TC. According to Town Charter section 4.1.2 the RTM has the power to:
(a) approve or amend and approve budgets passed by the Council in accordance with Chapter IX of this Charter;
(b) approve or reject supplemental appropriations in excess of ten thousand dollars ($10,000) in accordance with Chapter IX;
(c) approve or reject bonding ordinances in accordance with Section 184.108.40.206;
(d) accept or reject gifts, grants, and bequests of real and/or personal property with a value in excess of ten thousand dollars ($10,000);
(e) accept or reject all open spaces and roads not a part of an approved subdivision. Reference is hereby made to Section 5.4.10;
(f) all powers granted under the CGS to the town meeting where the objective of such statute cannot be otherwise accomplished by the Town pursuant to CGS, the Charter, or ordinances.
The breadth of its powers, and the fact that it is an elected body, make it subject to the OPOV requirement under Avery v. Midland.
The members of the RTM are elected at-large from within voting districts, the number of which is determined by the TC. There were 8 voting districts until 2013 when the number was reduced to 7. By charter, the RTM is reapportioned every four years following a presidential election. The members are apportioned among the districts on the basis of the number of “electors” in each district, where “elector” means a registered voter at the time of the preceding presidential election.
The method of apportioning the representatives among the voting districts, as well as determining how many representatives there are, is detailed in section 4.2.2 of the Groton Town Charter. First, the total number of electors is divided by 45—the maximum size of the RTM as designated in the Charter—and the resulting number is rounded down to get an integer we will call the divisor. The divisor is then divided into the number of electors in each district and the resulting quotient is again rounded down. The resulting integer is the number of representatives allocated to that voting district.
To illustrate this process, Table 1 shows the structure as of December 2012. (Note that Groton redistricted in 2013; I will get to that later.) The total number of electors is 20,742. The quotient 20,742/45= 460.93 is rounded down to get a divisor of 460. To get the number of representatives allocated to District 1, which contains 3,176 electors, we round down the quotient 3,176/460=6.90 to get the assignment of 6 electors. The other numbers are similarly calculated.
Table 1: Groton RTM 2012
|Voting District||2010 Population||Electors 2012||RTM reps|
C. Does the Structure Meet OPOV?
Does this allocation meet the strictures of OPOV with respect to either total population or the number of electors as a measure of population? The Court established a very clear method to decide this question in Abate v. Mundt. First you compute the amount of representation each district receives by computing the number of people per representative. Then you compute the percent difference of that amount from what a perfect representation would be; this percentage is called the percent deviation. And finally you look at the gap between the most over-represented district and the most under-represented district—which is called the total deviation. That number serves as the quantity of interest and the Court sets different thresholds on how large that is permitted to be depending on the level of government being examined.
As an example, in Voting District 1, the amount of representation is 3,607/6=601.67 if we use total population as the measure of people and 3,176/6=529.33 if we use electors as the measure. Perfect representation would result in 40,115/40=1,002.88 people per representative and 20,742/40=518.55 electors per representative. The percent deviation for District 1 is (601.67−1,002.88)/1,002.88=−0.4006 or −40.06% using total population and 529.33−518.55)/518.55=0.0208 or 2.08% percent deviation with respect to the number of electors. The full table of values is given in Table 2.
Table 2: Groton 2013 Deviation
|Elector/Rep||% dev Ele/Rep||Pop/Rep||% dev Pop/Rep|
The total deviation is given by the difference between the percent deviation of the most underrepresented district and the percent deviation of the most overrepresented. In this example, using registered voter population, District 4 is the most underrepresented with a percent deviation of 21.97%. District 5 is the most overrepresented district with a percent deviation of −9.99%. Thus, the total deviation of the RTM using registered voter data is 21.97%− (−9.99%) = 31.96%. A similar calculation using the total population data as of 2012 shows a total deviation of 184.70%.
The total deviations for both total population and electors are quite large, 31.96% using electors and 184.70% using total population. No court has upheld any legislative scheme as adhering to OPOV with such large total deviations. The largest total deviation ever found to be in compliance was about 17%. The Supreme Court rejected as outrageous a total deviation of 78% in Board of Estimate of New York v. Morris. There seems to be little by way of justification for having such a large total deviation.
These computations merely show that the Town of Groton was skating on thin ice with respect to OPOV. These numbers are not helpful in providing a test case for the use of voting data to measure OPOV in 2012 since no matter what measure of population is used Groton seems to fall short.
Fortunately for us, in 2013 the town decided to change the number of voting districts to 7 and then after the 2016 presidential election they reapportioned the RTM. The relevant data is collected Table 3.
Table 3: Groton 2017 Deviation
|District||Electors 2016||RTM Reps||Elector/Rep||% dev Ele/Rep||2010 Population||Pop/Rep||% dev Pop/Rep|
We see from Table 3 that the total deviation is 19.77% using registered voter data and at least 136.52% using total population. The latter would be a clear violation of OPOV under any analysis. The question of whether a total deviation of 19.77% is acceptable is somewhat more fraught. The Court has validated districting schemes with a total deviation of 16.40% arguing that the state (Virginia in this case) had a legitimate interest in preserving political boundaries in their districting that resulted in this amount of deviation. On the other hand, the Court has invalidated a districting plan with total deviation of 20.14%. Whether Groton could mount an argument similar to Virginia’s is not clear. However, it is also true that local governments are given considerably more latitude than state governments when it comes to achieving OPOV, so it is more than plausible that the current districting of Groton would meet the strictures of OPOV using registered voter data even though it would fail if using total population. And that is exactly what we seek for a test case.
Just a few final remarks to close:
(1) The total deviation in the RTM could be reduced if only Groton had not chosen such an odd method of apportionment. For example, had they moved one representative from District 6 (leaving 6 representatives there) to District 2 (resulting in 5 in that district) the resulting total deviation drops to 18.25%.
The Town Charter allows for up to 45 representatives, and if the Town were able to use this flexibility they could lower the total deviation even more by using a different apportionment method. Employing an algorithm designed to minimize total deviation, I find that the apportionment 7, 5, 6, 8, 5, 7, 5 of 43 representatives to District 1, 2, etc., respectively, results in a total deviation of 13.01%, which falls within the range of deviations that have already been found acceptable.
(2) Why Groton established this method of districting (and their even odder method of apportionment of representatives) remains a mystery. I suspect that it is similar to the reason that Hawaii uses the same registered voter data—there is a large naval presence in Groton. The Naval Submarine Base, New London is the home of 6,500 military personnel. That is roughly one-sixth of the total population of the town. By choosing to district with respect to registered voters, the influence of the local military is likely to be minimized. What makes the situation different from Hawaii, however, is that the districts drawn using registered voters are not close to what would be required using total population.
There are other ways to minimize the influence of transient military members on local government if that is what Groton wished to accomplish. States have more flexibility than the federal government to exclude certain groups from the count of total population for districting purposes. For example, Hawaii, Kansas, and Washington all exclude nonresident military personnel from the total population count.
Would changing the accounting of total population be superior to the current system in Groton? I am not so sure.
(3) Can Groton’s example be generalized beyond the class of small New England towns? The answer to that is largely dependent on one’s view of the reliability of the registered voter data.
First, if the jurisdiction needs to be concerned about producing minority-majority districts, the registered voter data may be inadequate. Only nine states collect racial data as part of the registration process, making it insufficient to draw minority-majority districts. On the other hand, total population data is not sufficient either; ultimately, some enumeration of the voters—or potential voters—by race is required if such districts are to be drawn.
A second alleged drawback of registered voter data is that registered voter data is subject to political manipulation. The data is subject to purges that could be politically motivated. It is certainly possible that such shenanigans might be employed. But it is also true that the census itself is subject to some manipulation by political forces—witness the current plan to include a question on the census about citizenship. Is it obvious which of the two population measures has more systematic bias? If there were more at stake with the registered voter data, would that encourage more manipulation, or would it act to encourage more vigilance to keep the data clean?
Two other concerns about registered voter data are related: the registration lists can vary considerably from election to election depending on the salience of the contests, and the “registration lists are notoriously inaccurate.” Again, it is useful to compare these claims against total population data. As noted earlier, the census data begins to degrade as soon as it is published. By the end of the ten-year cycle it is hard to believe that it is any more accurate than registered voter data. Moreover, as Groton demonstrates, if the use of registered voter data is paired with more frequent reapportionments then perhaps the combination is better than the use of census data alone over a longer period of time.
Clearly these arguments are not definitive, but they do indicate that, perhaps, the experience of Groton can be duplicated elsewhere.
(4) In November 2018 there will be a proposition on the ballot to revise Groton’s town charter to eliminate the RTM, among other reforms. The principal concern about the RTM seems to be that it is ineffective in monitoring the town budget. This proposal was originally made by a charter revision committee but it was rejected by the TC. Now a petition has received a sufficient number of signatures to have it appear on the ballot. Should this proposal pass, much of what I argue in this note is likely to be moot.
(5) If a court rejected the use of registered voter data for districting it might still leave open the prospect of some other voter-related measure. The one proposed in Evenwel was citizen voting age population. So the prospect of clearly resolving the question of what population measure is appropriate could be postponed further.
While most practitioners focus on federal and state elections, the Town of Groton illustrates why an examination of local elections remains fruitful. There is more variation in governmental structure in local government than most academics appreciate. Groton would give the courts an opportunity to weigh in on the acceptability of these local innovations, as well as address broader questions left open by Evenwel. I hope this tale of the Town of Groton will also encourage others to investigate these local innovations in representation.
* Professor of Mathematics and Law, Department of Mathematics and the Law School, Vanderbilt University. Thanks to Ed Cheng, Rick Hasen, Suzanna Sherry, and Ganesh Sitaraman for helpful comments. Additional thanks to Deborah Schander of the Vanderbilt Law Library and Weixing Zhang of the Connecticut State Data Center for research help.
. For more extensive treatment, see generally Robert W. Bennett, Should Parents Be Given Extra Votes on Account of Their Children: Toward a Conversational Understanding of American Democracy, 94 Nw. U. L. Rev. 503, 516 (2000) (discussing the logic behind the decision to count total population for measuring the size of districts); Joseph Fishkin, Weightless Votes, 121 Yale L.J. 1888, 1892 (2012) (analyzing the difficult question of how to “weight” votes in districting and how it impacts elections); Ronald Keith Gaddie et al., Seats, Votes, Citizens, and the One Person, One Vote Problem, 23 Stan. L. & Pol’y Rev. 431, 432 (2012) (discussing citizen apportionment and the flexibility the Court has provided States in determining how to count population for districting purposes); Bernard Grofman & Howard A. Scarrow, The Riddle of Apportionment: Equality of What?, 70 Nat’l Civic Rev. 242, 243 (1981) (noting that “the Supreme Court has never clearly answered” what representation means in the context of reapportionment); Sanford Levison, One Person, One Vote: A Mantra in Need of Meaning, 80 N.C. L. Rev. 1269, 1270 (2002) (lamenting the “confusions” resulting from a lack of clarity about the definition of representation in the one person, one vote doctrine).
. Evenwel v. Abbott, 136 S. Ct. 1120, 1124 (2016). But there is one notable exception. In Burns v Richardson, 384 U.S. 73 (1966), the state of Hawaii was permitted to use registered voting age population because of its large military population. The Court accepted this mostly because “the apportionment achieved by use of a registered voters basis substantially approximated that which would have appeared had state citizen population been the guide.” Since Burns, which is considered at best sui generis and at worst just bad law, there have been no further Supreme Court cases addressing the issue of the nature of OPOV.
. Population data is from the US Census provided by the Connecticut State Data Center. Email from Weixing Zhang, Conn. State Data Ctr., to Paul H. Edelman, Professor of Mathematics and Law, Vanderbilt Univ. (May 17, 2018, 10:51 EST) (on file with the Michigan Law Review). The Elector and representative data is from the Town Clerk of Groton, Connecticut. See Email from Moukawsher, supra note 7; Town Council, Town of Groton, Conn., Meeting Minutes 7 (Dec. 18, 2012).
. This method of apportioning goes back to the original Town of Groton charter in 1957, although the original charter allowed for up to 60 representatives. Groton Town Charter, ch. 3, § 1 (1957) (on file with the Michigan Law Review). I have no idea of the origin of this method. It is unlike the methods used to apportion the U.S. House of Representatives in that the number of members of the RTM is not a priori fixed but will vary depending on the outcome of the calculations. I do not know of any other legislative body in the United States or elsewhere that employs a similar method.
. A comment on the data. When Groton redistricted and lost a district they consolidated the 4th and 5th districts and designated it as District 4. Then they moved some of District 8 into District 1 and renumbered Districts 6–8 to become District 5–7. Compare Town Council, Town of Groton, Conn., Meeting Minutes 7 (Mar. 5, 2013), with Town Council, Town of Groton, Conn., Meeting Minutes 6 (Jan. 3, 2017). As a result, the populations of Districts 2 and 3 remained unchanged, the new District 4 has a population which is the sum of the populations of the old 4th and 5th districts, and District 6 has the population of the old District 7. Compare Beeson et al., supra note 5, at 17 (2012), http://www.groton-ct.gov/about/TownGovGuide.pdf [https://perma.cc/ED33-M23P] (showing the original 8 districts), with Town of Groton, Conn., supra note 19 (showing the current 7 districts). I am not able to recompute the 2010 population in the new Districts 1 and 7, which is what the ??? represent. Thankfully, the data that I do have is sufficient to make the point that the total deviation based on total population exceeds any reasonable threshold.
−7.10% resulting in a total deviation of 12.67−(−7.10)=19.77%. Using 2010 total population District 4 has a percent deviation of 101.4% and District 6 has a percent deviation of −35.38%. Thus, the total deviation is at least
101.14−(−35.38)=136.52%. We can only say “at least” because we do not have all of the population information.
. Paul H. Edelman, Minimum Total Deviation Apportionments, in Mathematics and Democracy, Recent Advances in Voting Systems and Collective Choice 55 (Bruno Simeone & Friedrich Pukelsheim, eds., 2006); Paul H. Edelman, Getting the Math Right: Why California Has Too Many Seats in the House of Representatives, 59 Vand. L. Rev. 297, 329–34 (2006).
. Those states are Alabama, California, Florida, Georgia, Louisiana, North Carolina, Pennsylvania, South Carolina, and Tennessee. Email from Carl Klarner, Academic Consultant, to Paul H. Edelman, Professor of Mathematics and Law, Vanderbilt University (June 7, 2018, 1:06 PM) (on file with the Michigan Law Review).
. Nick Brown, U.S. Census Citizenship Question Panned by Scientists, Civil Rights Groups, Reuters (Aug. 9, 2018, 8:14 PM), https://www.reuters.com/article/us-usa-census-citizenship/u-s-census-citizenship-question-panned-by-scientists-civil-rights-groups-idUSKBN1KV00T [https://perma.cc/QX4G-YFQM].
. Deborah Straszheim, Groton Residents Petition Rejected Charter Revision Report to November Vote, Day (Jan. 22, 2018, 7:38 PM), https://www.theday.com/local-news/20180122/groton-residents-petition-rejected-charter-revision-report-to-november-vote [https://perma.cc/M3KP-A49R].
. The proposition referred to above was defeated on November 6, 2018. See Erica Moser, Groton Voters Reject Charter Changes, Day (Nov. 6, 2018, 9:55 PM), https://www.theday.com/local-news/20181106/groton-voters-reject-charter-changes [https://perma.cc/2VBH-9LVY]. This leaves the RTM intact so the problems identified in this essay will continue into the indefinite future.