Clough v. Guzzi

OPINION LEVIN H. CAMPBELL, Circuit Judge. In this action brought pursuant to 28 U.S.C. § 2201, plaintiff seeks a declaratory judgment that sections 34 and 45 of chapter 53 and sections 33, 41 and 42 of chapter 54 of the Massachusetts General Laws are unconstitutional as violative of rights guaranteed by the fourteenth amendment. The challenged statutes provide that the official ballots in primary and final elections shall list incumbents first, followed by other can *1059 didates in alphabetical order, and shall designate incumbents as candidates for re-election. 1 Jurisdiction is asserted under 42 U.S.C. § 1983 and 28 U.S.C. § 1343(3) & (4). As plaintiff seeks an injunction with respect to ballot arrangement in future elections, a three-judge court was requested and convened pursuant to 28 U.S.C. § 2281 et seq. Plaintiff, Charles I. Clough, Jr., was a candidate for Middlesex county commissioner in the September 1974 Democratic primary. The original defendant, John F. X. Davoren, was Secretary of the Commonwealth, with duties of supervising state and town elections, preparing ballots, and arranging for voting machines. Paul H. Guzzi, the current Secretary of the Commonwealth, has been substituted as defendant. Plaintiff contends that the Massachusetts ballot system, insofar as it designates incumbents as candidates for re-election and places them first on the ballot, gives incumbents a significant and unfair advantage in the results of an election. He asserts that lesser advantages accrue to both incumbency and first ballot position when taken separately, and that in combination an even more significant advantage results. As Massachusetts thereby confers an advantage on one class of candidates (incumbents), without conferring a like advantage on another (non-incumbents) of which he is a member, he claims that he is denied equal protection of the laws. 2 *1060 The complaint was filed on June 3, 1974. On June 19, 1974, a requested temporary restraining order was denied by the single judge. On August 7, 1974, this court of three judges denied a preliminary injunction, ruling that plaintiff had not shown a sufficient probability of success on the merits. We indicated, however, that the questions raised would not be rendered moot by the upcoming elections, and should be decided. Thus upon plaintiff’s motion, we held a hearing on the merits on May 18, 1976. New primary elections are to be held in Massachusetts this September, and while plaintiff is not a candidate, he indicates his continuing interest in the controversy. 3 Prior Judicial Proceedings This is not the first attempt by candidates and voters to raise this issue either in the federal or Massachusetts courts. Until now, however, plaintiffs have encountered difficulty in securing a clear-cut adjudication. The statutes now before us were challenged in Tsongas v. Davoren, CA No. 74-2084-F (D.Mass.) (Tsongas I), as violative of the federal constitution, and in Tsongas v. Secretary of the Commonwealth, 362 Mass. 708, 291 N.E.2d 149 (1972) (Tsongas II), as- violative of Article IX of the Massachusetts constitution. 4 A three-judge court was convened in Tsongas I, but it stayed consideration of the federal constitutional question pending disposition of the state question in Tsongas II. In Tsongas II, a single justice of the Massachusetts Supreme Judicial Court, Justice Reardon, held, in a memorandum dated July 26,1972, that since the statutory objecfive was primarily to establish a favored position for incumbents on the ballot, the constitutional mandate of equality had not been observed. But the full bench of the S.J.C. stayed Justice Reardon’s order and remanded to him for an evidentiary hearing, for findings of fact as to whether “a candidate given a position at the head of the ballot has [a] distinct advantage over other candidates”, and for such other rulings as he might deem advisable. Thereafter Justice Reardon filed certain findings which are part of the present record. He first noted, “Given the complex nature of voting behavior influenced by numerous variables, not all of which can be identified and few of which can be accurately measured, and given the inherent difficulty of conducting a direct empirical study of the effect of ballot position, proof of an advantage associated with being first on the ballot is necessarily imprecise.” He found that “[although first ballot position may not always, in all types of races, confer a systematic advantage, in a significant number of elections it appears to be the most advantageous.” He went on to find that various other factors influenced the extent of the first position advantage: the visibility of the race; the number of candidates on the ballot; and the position of the particular race on the overall ballot. He also found that, in the races in which plaintiffs were competing, county commissioner and treasurer, position effect was maximized to the extent that first position held a “distinct advantage”. Finally, he found that incumbency was a more signifi *1061 cant and consistent advantage than first position, but that it did not negate the advantage conferred by first position. The full bench accepted Justice Reardon’s findings as not “plainly wrong”, but vacated his decree and dismissed the complaint. As the primary election had by that time passed (two of the plaintiffs had won and one had lost), the Court considered plaintiffs’ prayer for injunctive relief moot. The Court also declined, despite plaintiff’s urging, to make a declaration as to the constitutionality of the challenged statutory provisions, stating, 5 “It is not clear that [the state] question is in any way different from the question whether these statutes deny equal protection of the laws in violation of the Fourteenth Amendment to the Constitution of the United States. The plaintiffs, however, have chosen to present the latter question to a Federal court. We are not asked to pass on it, and it would not be appropriate for us to do so. Thus, any answer by us would be incomplete and ‘would not terminate the uncertainty or controversy giving rise to the proceedings.’ If the Federal case were pressed to a decision, it might produce a conclusion different from ours. If we upheld the statutes, and the Federal case were not decided on the merits, it is not clear that our decision would serve any substantial public purpose.” 362 Mass, at 714, 291 N.E.2d at 153. [citation omitted]. Justice Kaplan in dissent expressed his disagreement with this course: “the muddle of the double litigation in which the parties are involved would largely be disposed of if we should hold the present procedure invalid, and something can be said for doing that particular job without looking to the Federal Building.” 362 Mass, at 722, 291 N.E.2d at 157. In the meantime, after the elections, motion was made in Tsongas I to dissolve the three-judge court since injunctive relief was no longer sought therein. Tsongas I was thereafter consolidated with other similar actions then pending before a single judge which also sought only declaratory relief. Those actions apparently have not as yet worked their way through the heavy calendar in the District of Massachusetts. 6 Historical Background Before turning to the factual and legal arguments raised, we think it instructive to relate the history of the challenged statutes, through their various amendments and re-enactments. Provision for the order of names on the ballot and for the designation of incumbents in both primary and final elections has long been part of Massachusetts law. 7 Originally names were to be listed in alphabetical order; a lottery was next instituted for certain elections; and *1062 finally the present incumbents-first system was instituted for most state-wide elections. Cities and towns presently are permitted to choose their own methods. Primary Elections The first legislative enactment on the subject provided that candidates for all elective offices be listed alphabetically by surname and that a star (*) be placed beside the names of candidates for re-election. St. 1894, c. 504, § 24. In 1895, the statute was amended to include candidates for town committees. Further change was made for certain electoral offices in 1911, providing that in elections for ward or town committees and for delegates to the state convention, ballot position be determined by lot. St.1911, c. 550, § 12. By this time, the star used to indicate incumbency was no longer in use. House no. 5312, Legislative Research Council Report Relative to Order of Names on the Ballot, at 17 (Jan. 9, 1974) (hereinafter Legislative Research Report). The lottery was abandoned in favor of the former alphabetical arrangement for all names, St.1923, c. 302, § 1, but was reinstated shortly thereafter. St.1925, c. 312, § 1. The lottery is still used in elections for ward and town committees. Mass.Gen.Laws c. 53, § 34, par. 4 (1976-77 Supp.). The present system for state primaries, giving first position to “candidates for renomination to state offices of which they are the elected incumbents”, was instituted in 1938. St.1938, c. 473, § 9. Other candidates follow in alphabetical order. The statute in basically its present form was finally enacted in 1953. St.1953, c. 406, § 3. Amendments since then have merely broadened the definition of incumbency to include certain appointed incumbents in addition to elected incumbents. 8 Final Elections The historical background of ballot position in final elections, provided for in c. 54, § 42, followed a pattern similar to that of primary elections. Names were originally listed in alphabetical order in all elections. In 1913, in the first departure from the alphabetical system, the City of Boston was authorized to determine ballot position in municipal elections by lottery, St.1913, c. 835, § 259, and similar authority was granted to other cities in 1932. St.1932, c. 135, § 5. In 1941, a new arrangement for statewide elections was enacted which placed candidates from the two major political parties first. St.1941, c. 292. The alphabetical system was, however, retained in other state elections, as well as in county, city, and town elections, except as otherwise provided by city charters. Incumbents were accorded first position on the ballot in final elections for the first time in 1948, St.1948, c. 272, or ten years after a similar provision was enacted for primary elections. That- system was later extended to municipal elections except as otherwise provided in city charters. St. 1953, c. 432. The methods which the commonwealth’s cities and towns have selected in accordance with their statutory grant are described in Legislative Research Report, supra, at 22-23. Of the state’s thirty-nine cities, fourteen use a lottery in primary elections but follow the state system in final elections; three use a combined lottery-alphabetical system; eight, including Boston, use a lottery in both primary and final elections; thirteen use the state system in both primary and final elections; and one uses a rotation system. Towns generally, by choice, follow the state system in their local elections, with the exception of six towns which use a lottery system. The Evidence Plaintiff argues that first ballot position, particularly when filled by an incumbent, carries with it a certain statistical advan *1063 tage. He attributes this phenomenon to the voting habits of a segment of the total electoral vote sometimes referred to as the “windfall vote” or “donkey vote”, i. e. the vote cast by citizens who are either uninformed about or indifferent to any or all of the candidates for a particular office on the ballot. Plaintiff contends that these voters uncritically check off whoever is at the top of the ballot, especially if that candidate is also an incumbent. Given this distinct advantage, he argues that the Commonwealth is thereby conferring a certain benefit on incumbents not shared by non-incumbents, and so is denying non-incumbents equal treatment under the law. We first analyze the evidence before us to determine if and to what extent plaintiff has established that the Massachusetts system of designating and ordering incumbents does indeed accord them an advantage. A number of written studies have been introduced in evidence purporting to demonstrate the effects of the designation of incumbency and of first position on the outcome of elections. Some of them support, and some contradict, plaintiff’s factual premise. Also introduced in evidence were the findings of Justice Reardon in Tsongas II; a transcript of testimony before Justice Reardon by Dr. Gary C. Byrne and J. Kristian Pueschel, co-authors of an extensive study of the Massachusetts ballot system; 9 a transcript of testimony by Dr. Henry Bain, an expert in the field of ballot position effect, in a similar California state court proceeding; and the Legislative Research Report earlier referred to, which has collected much of the literature in the field, which we also refer to herein. We have also heard testimony from plaintiff’s expert, Dr. Marian Just, who is expert in the field of quantitative political science, involving the application of statistical analysis to voting behavior. The most relevant study to the case before us is that of Byrne and Pueschel. This analyzes in detail the effects of both ballot position and incumbency in Massachusetts primary elections for three offices — county commissioner, state senator, and representative to the General Court — during the years 1926 to 1970. 10 Covering over 20,000 candidates, it is the largest study of its kind. It includes elections both before and after incumbents were given first position on the ballot (the changeover having occurred in 1938). Byrne and Pueschel cumulated and analyzed separately the results of their study for each of the three offices. In elections for representative to the General Court, involving over 18,000 candidates, by far the largest number, they found that incumbency was the dominant electoral influence. As for first position, their findings were mixed: under some conditions, first position offered an advantage; and under certain control conditions, for example where an incumbent was elsewhere on the ballot, first position frequently offered a disadvantage. After 1938, when incumbents began to be listed first, they found a twelve-fold increase in the advantage of first position. In elections for state senator, involving over 2,573 candidates, they found that Democratic and Republican incumbents had a substantial advantage over other candidates even before the 1938 change. After 1938, when incumbents moved into first position, their advantage increased marginally. (The increase may have been only marginal because the pre-1938 advantage for incumbency was already so substantial.) Indeed, Byrne and Pueschel concluded that while first position did increase somewhat the ad *1064 vantage already accruing to an incumbent by virtue of his office, clearly the dominant influence was incumbency. And after analyzing races with no incumbent candidate, in order to isolate the first position effect, they found that first position was generally advantageous, although in some cases disadvantageous, depending on the number of candidates in the race. The final office for which elections were studied, county commissioner, was of a somewhat different character than the other two. It was a lower level and less visible office and was placed lower on the ballot. (This was also the office for which plaintiff was running.) In elections involving 1,179 non-incumbents and 267 incumbents, or I, 446 candidates in all, Byrne and Pueschel again found incumbency advantageous — regardless of the incumbent’s position on the ballot or the number of other candidates in the race. They did find that first position offered a slight advantage, depending on several factors, when isolated from the effect of incumbency, and that the first position advantage became substantial when interacting with incumbency. They also concluded, interestingly, that the influence of incumbency was so powerful that it was somewhat lessened when combined with the varying advantage factor of first position. While Byrne and Pueschel recognized some advantage to first position, they tended to minimize it and attached less importance to it than did plaintiff’s expert, Dr. Just. Dr. Just cited as her principal authority a study by Henry M. Bain 11 and Donald S. Hecock, Ballot Position and the Voter’s Choice (1957), which analyzed the 1952 Democratic primary for state senate in Michigan. 12 Bain and Hecock found that first position in a vertical list was favored on paper ballots, and that the top row was favored on voting machines. As a related effect, they found additional positions after the first also favored as the number of candidates for a particular office increased. On long lists, they found that the last, in addition to the first position was favored (called the “J curve effect”). Their statistics also showed that voters’ familiarity with a particular candidate modified the position effect; the better known candidate was less disadvantaged by occupying a lower position, and the magnitude of the position effect was greater outside a candidate’s home town, where he would be less well known. Turning to Dr. Just’s own testimony, she stated that political scientists generally agree that the name appearing first on the ballot has a better chance of being chosen. She agreed with scholars who believe that the position effect causes at least a 5% increase in the first candidate’s total vote, see below, and further claimed that the increase is actually as high as 15%. She concluded that the longer the ballot, the more important is first position, although on extremely long lists, the “J curve effect” comes into play. The less important the office and the less well known the candidate, the more important is position. Hence in primary elections, where candidates are generally less well known, position is a more significant factor than in final elections. Dr. Just acknowledged that ethnicity also might be a significant factor influencing the so-called “windfall vote”. Dr. Just expressed the view that the fairest means to determine ballot position would be a rotation system, whereby as many forms of ballots would be used as there are candidates, and ballot positions would be rotated so that each candidate is listed in each ballot position an equal number of times. Lot was her second choice. Dr. Just criticized Byrne and Pueschel’s results as being contrary to most other literature. She felt that Byrne and Pueschel were working from a specific hypothesis, *1065 that incumbency is a stronger factor than position, and therefore did not consider position effect carefully. She interpreted their statistics, nonetheless, as demonstrating that both incumbency and first position separately confer a significant advantage, and that in combination the two factors confer an advantage which is even greater than the sum of the two effects. As for position effect with voting machines, which are also utilized in Massachusetts elections, Dr. Just testified that the data is inconsistent. (Bain and Hecock and Byrne and Pueschel did study both machines and paper ballots.) Dr. Just stated that the name at the “top” of the voting machine would be the one at the extreme left. Also, the top row would be preferred. She described those voting machines with which she was familiar as the type used in New York, which have vertical rows. She has not observed the type of machines used in Massachusetts. Other studies of elections outside Massachusetts were also introduced into evidence. The findings and conclusions in these are somewhat mixed. A 1950 study of a two-person state senatorial contest in Ohio in which ballot positions were rotated, showed that the first position candidate carried 85% of the precincts. 13 (Only voting machines were used in that election, and first position referred to the top line.) The results of a 1969 Los Angeles primary race, in which 133 candidates ran for seven positions on a junior college board of trustees, were found to be characterized by the “J curve effect”. 14 Another study of ten primary contests in California revealed a first ballot position advantage, except in the race for governor. 15 The author concluded that the advantage amounted to a minimum of 5% of the first-listed candidate’s vote. The author also emphasized that a precise determination of an average bias figure was impossible because the size of the bias also depended on factors such as the importance of the office, publicity, the educational level of the electorate, and the complexity of the voting procedure. Finally, a British study of two, three, and four-person races in the 1959 and 1964 British general elections concluded that first place was not advantaged, but that last place was disadvantaged. 16 Findings of Fact The range of expert opinion demonstrates that the advantage conferred by incumbency and by first position on the ballot is not one which is easily susceptible to measurement. Voter familiarity with the candidates, the importance of the office, ethnic affiliation, and the educational level of the electorate, to name but a few factors, make it difficult to isolate with confidence the effects of incumbency and first position. Measurement is made more difficult by the fact that studies cannot be made in controlled laboratory conditions. Still we do not altogether discount the likelihood that skilled analysis can, to a certain extent, reveal patterns resulting from the effects of incumbency and first position independent of other variables. On the basis of the present record we are persuaded, and so find, that the designation of incumbency does confer a distinct benefit on the incumbent candidate. We also find *1066 that first ballot position, in combination with the designation of incumbency, sometimes confers some further increment of advantage in favor of the incumbents so positioned. On the other hand, we find that plaintiff has not proved a substantial advantage inherent in first ballot position alone. We note that we have some difficulty with the definition of the “windfall vote”, the segment of the electorate presumably influenced by the factors just mentioned (i. e. incumbency reinforced by first position). Is an “uninformed” voter who selects an incumbent because of his status as an incumbent, without knowing anything more, part of the windfall vote? Or is that a rational choice? Incumbency is highly relevant to a candidate’s qualifications or lack thereof. We raise this question as it indicates the difficulty of clearly identifying the extent of the unequal or, at least, unfair treatment which plaintiff asserts, a difficulty which certainly weighs in our final decision. Do the Challenged Statutes Violate the Fourteenth Amendment? Plaintiff contends that, as the designation of incumbency and first position for incumbents together give a statistical advantage to incumbents without giving a like advantage to non-incumbents like himself, Massachusetts is thereby denying him equal treatment under the law, in violation of the fourteenth amendment. He likens his interest as a candidate to that of voters who support him or other non-incumbents. Therefore, he asserts, a fundamental right, the right to vote, is also being infringed by the challenged statutes. And where a fundamental right is involved, he contends that the Massachusetts statutes can only pass constitutional muster if they rest on some compelling state interest. We agree that Massachusetts accords some advantage to incumbents by means of its challenged practices, and we have so found. But it does not necessarily follow that these statutes violaté the fourteenth amendment. Whether there is a constitutional violation depends on whether or not the classification implicit in the Massachusetts legislation is reasonably justifiable. And in making this assessment, the initial and generally critical question concerns the nature of the interest asserted by plaintiffs, i. e. whether it is the type of fundamental interest to which all but compelling state interests must yield; or rather a non-fundamental interest which may be restricted or otherwise affected by legislation having a rational basis. Or, put differently, the question is what standard of scrutiny should be applied in assessing the state interest asserted by the defendant in justification of the classification. Equating his interest as a candidate with that of voters, plaintiff argues for strict scrutiny on the basis that the right to vote is a fundamental right. See Massachusetts Board of Retirement v. Murgia, - U.S. -, 96 S.Ct. 2562, 49 L.Ed.2d -, 44 U.S.L.W. 5077, 5079 & n.3 (1976). This identifying of a candidate’s interests with those of voters is not without support in considering what level of judicial review is proper. Bullock v. Carter, 405 U.S. 134, 142-43, 92 S.Ct. 849, 31 L.Ed.2d 92 (1972). The right to run for public office, and with it the status of candidacy, is not itself viewed as a fundamental right which in and of itself warrants strict scrutiny. Id; cf. Turner v. Fouche, 396 U.S. 346, 362, 90 S.Ct. 532, 24 L.Ed.2d 567 (1970). But a practice which, by impinging upon the status of candidacy also has the effect of placing a limitation or incidental burden on the exercise of voting rights, may be subject to rigorous judicial review. See Lubin v. Panish, 415 U.S. 709, 94 S.Ct. 1315, 39 L.Ed.2d 702 (1974); Bullock v. Carter, supra 405 U.S. at 143, 92 S.Ct. 849, citing McDonald v. Board of Election, 394 U.S. 802, 89 S.Ct. 1404, 22 L.Ed.2d 739 (1969). Whether it is subject to such rigorous review requires an examination “in a realistic light” of the extent and nature of the impact of the questioned practice on the right to vote. Bullock v. Carter, supra. In the instant case, we are not persuaded that the alleged burden on candidacy has a *1067 sufficient impact on the right of Massachusetts voters to choose a representative government as to warrant strict judicial scrutiny. See Sununu v. Stark, 383 F.Supp. 1287, 1292 (D.N.H.1974) (concurring opinion), aff’d, 420 U.S. 958, 95 S.Ct. 1346, 43 L.Ed.2d 434 (1975); Chimento v. Stark, 353 F.Supp. 1211, 1218 (D.N.H.) (concurring opinion), aff’d, 414 U.S. 802, 94 S.Ct. 125, 38 L.Ed.2d 39 (1973). We first note what the challenged statutes do not do: they do not impose direct restrictions on the exercise of the right to vote, e. g., Harper v. Virginia Board of Elections, 383 U.S. 663, 86 S.Ct. 1079, 16 L.Ed.2d 169 (1966) (poll tax); they do not effectively bar candidates from an election by imposing prohibitively burdensome candidacy filing fees, Bullock v. Carter, supra; and they do not restrict the ballot to candidates of only certain political parties, Williams v. Rhodes, 393 U.S. 23, 89 S.Ct. 5, 21 L.Ed.2d 24 (1968). Powerful restrictions of this sort have been held to have such a severe impact on the right to vote as to warrant greater judicial concern, and accordingly stricter scrutiny, lest a fundamental right be infringed. Here we are told that the Massachusetts statutes afford to incumbents such an unfair advantage, by according them the windfall vote, as effectively to dilute the weight of the vote of those electors who support the class of non-incumbents. See Gould v. Grubb, 14 Cal.2d 661, 122 Cal.Rptr. 377, 536 P.2d 1337, 1343 (1975). Plaintiff analogizes this “dilution” to cases where a state has been found to have quantitatively diluted the weight of votes by the malapportionment of legislative districts, e. g. Reynolds v. Sims, 377 U.S. 533, 562-63, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964). Such practices have so limited the effect of one’s vote that they have been viewed as an infringement of the fundamental right to exercise one’s vote, and have been strictly scrutinized, id. Similarly, plaintiff argues the votes of informed citizens who are familiar with the candidates and the issues are subject to dilution by the state’s near guarantee that the votes of uninformed citizens, who simply check off the first name on the list, will go to the incumbent candidate. The state practice here differs significantly from those in the apportionment and other voting cases where the courts have exercised strict scrutiny, in that it does not create the same kind of immutable or absolute advantage or preclusion. In the apportionment cases, the district lines and populations were fixed and the effects of malapportionment unavoidable for voters within the disadvantaged districts. In contrast, the dilution effect herein depends upon the existence of a pool of presumably uninformed voters. Yet non-incumbents and their supporters have access to those voters and may, in theory and possibly in practice, so educate them as to eliminate the donkey vote and thus eliminate the statistical position bias. Furthermore, the impact of the Massachusetts statutes on the total vote is not so clear that we can really define with anything approaching certainty, as would be possible in the apportionment cases, how significantly the state is directly affecting the weight of anyone’s vote. The principle of dilution as affecting the fundamental right to vote would seem better reserved for the more clear-cut and certain cases of inequality. Voters have no constitutional right to a wholly rational election, based solely on reasoned consideration of the issues and the candidates’ positions, and free from other “irrational” considerations as a candidate’s ethnic affiliation, sex, or home town. We conclude that, even assuming some positional advantage here, the voters’ right to choose their representatives is not sufficiently infringed as to warrant strict scrutiny of the Massachusetts statute and underlying legislative purpose. Our task is therefore simply to decide whether Massachusetts has a legitimate, rational reason for designating incumbents and placing them first on the ballot. We note that the states have wide latitude to determine “the nature of their own machinery for filling local public offices.” Oregon v. Mitchell, 400 U.S. 112, 125, 91 S.Ct. 260, 265, 27 L.Ed.2d 272 (1970) *1068 (Black, J.); accord, Evans v. Cornman, 398 U.S. 419, 422, 90 S.Ct. 1752, 26 L.Ed.2d 370 (1970). Massachusetts justifies its method of ballot ordering, unique among the states, as an informational device. This method indicates in a clear manner who is the candidate for re-election, both by marking his or her name and by placing that name first. The Commonwealth further contends, and plaintiff’s expert, Dr. Just, conceded, that the most important decision which the voter must make is whether to retain or to replace the incumbent. Its statutory order serves to underscore that decision and, the Commonwealth asserts, helps eliminate the possibility of voter confusion. The Commonwealth also asserts that, if there is an uninformed segment of the voters, those votes should go to the incumbent candidate who at least has some experience in the business of government. We cannot say that these are illegitimate considerations beyond the authority of the Commonwealth properly and lawfully to advance. The designation of incumbency is material, and the statutory order relevant, to the electoral decision-making process. See Bullock v. Carter, supra, 405 U.S. at 145, 92 S.Ct. 849; Morey v. Doud, 354 U.S. 457, 465, 77 S.Ct. 1344, 1 L.Ed.2d 1485 (1957); cf. Anderson v. Martin, 375 U.S. 399, 84 S.Ct. 454, 11 L.Ed.2d 430 (1964). The fact that some statistical advantage may at the same time accrue to one of the candidates by virtue of his or her incumbency does not for constitutional purposes invalidate that otherwise legitimate purpose, especially where that advantage remains problematic and variable from election to election. And whether for purposes of a more absolute fairness that advantage warrants a different statutory scheme is properly a legislative consideration. We therefore hold that the Massachusetts statutes are constitutionally permissible under the equal protection clause of the fourteenth amendment. We add, as a further consideration supporting the rationality of Massachusetts’ choice, that none of the available alternatives are themselves without disadvantages. Alphabetical order or a lottery would, in the end, give only one candidate first position, and would arguably entail an even more arbitrary system than the present one. The rotational system, which plaintiff advocates and which a number of states have adopted, would presumably allow all candidates to occupy first position on an equal number of ballots, and thus share equally in the advantage. However, the system is more burdensome to administer and more costly because of the necessity of printing more- than one ballot; some critics say that it is also more susceptible to tabulation error. Without meaning to overstate these difficulties, which may well be offset by the greater equity or appearance of equity provided by the rotational system, still we cannot say that a legislature could not rationally give some weight to them in declining to adopt such a system. Complaint dismissed. Judgment for defendant. So ordered. Footnotes [1] Primary elections are governed by the provisions of chapter 53, Mass.Gen.Laws. The order of names on the ballot is laid out in § 34: . . Names of candidates for each elective office shall be arranged alphabetically according to their surnames except as otherwise provided. “Names of candidates for nomination for all offices to be voted for at a state primary of which they are the elected incumbents or the incumbents chosen by the senate and house of representatives, or appointed by the governor, or appointed by the justices of the supreme judicial or superior court, or appointed by the county commissioners, or appointed by the county commissioners and the clerk of the courts for a county, shall be placed first in alphabetical order and names of other candidates shall follow in like order. “Names of candidates for state committee who are either elected or appointed incumbents shall be placed first in alphabetical order, and names of other candidates for said office shall follow in like order. “Names of candidates for ward and town committees shall be arranged in groups in such order as may be determined by lot Information concerning incumbency is to be supplied in a candidate’s nomination papers, § 45, par. 1: “Every nomination paper [may] state, in addition to the name of the candidate, . the public offices which he holds or has held, showing clearly that he is a former incumbent thereof if such is the case and, if he is an elected incumbent of an office for which he seeks renomination, that he is a candidate for such renomination . . . This statement is then placed beside the name of the candidate on the ballot, § 34. Final elections are governed by the provisions of chapter 54, Mass.Gen.Laws. The following provisions are made therein for the order of names on the ballot and for the designation of candidates for re-election: Section 42, par. 1: “Except as provided in section forty-one A [as to governor and lieutenant governor], under the designation of the office, the names of the candidates for re-election to any office to be filled at a state election of which they are the elected incumbents shall, except in places where voting machines are used, as provided by section thirty-three, be placed first on the ballot in alphabetical order according to their surnames; next and in like order the names of candidates of political parties, as defined in chapter fifty, and the names of all other candidates shall follow in like order.” (Similar provision is made for governor and lieutenant governor, who are listed together as a group, § 41 A.) Section 41, par. 3: “. . .To the name of a candidate for a state or city office who is an elected incumbent thereof there shall be added . the words ‘Candidate for Re-election’, except in the case of political party candidates for the offices of governor and lieutenant governor. ... To the name of each candidate for a town office upon an official ballot who is an elected incumbent thereof shall be added the words ‘Candidate for Re-election’. Section 33, par. 5: “The arrangement of names and questions on voting machines and on ballot or ballot labels used in electronic voting systems shall be, in general, the same as on the official ballot, except as hereinbefore provided.” [2] A similar argument was advanced in Gould v. Grubb, 14 Cal.3d 661, 122 Cal.Rptr. 377, 536 P.2d 1337 (1975), challenging a California bal *1060 Iot system similar to the Massachusetts. The high court of California declared that system unconstitutional. See also Williamson v. Fortson, 376 F.Supp. 1300 (N.D.Ga.1974). [3] It is not now contended that the action is moot, nor do we find it to be so. Because of the short duration of the election process compared to the protracted time span of most federal court litigation, cases challenging election practices will rarely be settled while a particular election is still live. Such controversies properly fall, we think, in the category of those “ ‘capable of repetition, yet evading review’ ”. Dunn v. Blumstein, 405 U.S. 330, 333 n.2, 92 S.Ct. 995, 998, 31 L.Ed.2d 274 (1972); Reed v. Board of Election Commissioners, 459 F.2d 121, 124 n.5 (1st Cir. 1972). As plaintiff has not abandoned his prayer for injunctive relief, there is no reason to dissolve the panel of three judges. [4] Article IX provides: “All elections ought to be free; and all the inhabitants of this commonwealth, having such qualifications as they shall establish by their frame of government, have an equal right to elect officers, and to be elected, for public employment.” [5] The Court apparently desired further proof in order to evaluate alternative systems, even though it agreed with Justice Reardon that proof in this context is “necessarily imprecise.” 362 Mass, at 720, 291 N.E.2d at 156. After discussing alternative methods of fixing ballot position, such as rotation, lottery, or alphabetical listing, Justice Braucher concluded: “The various possibilities suggested above rest largely in speculation. They were not canvassed in the testimony before the single justice nor indeed in the briefs. Since the controversy giving rise to this case is now moot, we do not think it would be appropriate to remand the case for further hearings. But in the absence of further data we cannot with confidence make a binding declaration of rights.” 362 Mass, at 720-721, 291 N.E.2d at 156-157. [6] In a related action also seeking injunctive relief, Murray v. Shea, CA No. 73-2676-T, statutes affecting ballot position in elections for Springfield City Council were challenged. A complaint had also been filed in the state court (Supreme Judicial Court, Equity No. 73-78), and that case was heard by Justice Quirico, sitting as a single justice of the Supreme Judicial Court. He found against the plaintiff candidates. Preliminary injunctive relief was then denied by the federal three-judge court pending appeal to the full Supreme Judicial Court of Justice Quirico’s decision. Plaintiffs did not, however, pursue their appeal to the full bench, and the federal action has since been dismissed. [7] This historical development is set forth in House No. 5312, Legislative Research Council Report Relative to Order of Names on the Ballot (Jan. 9, 1974). [8] Amendments have brought within section 34 incumbents who have been chosen by the Senate and House of Representatives, St. 1954, c. 225; appointed by the governor, St. 1958, c. 289; appointed by the Supreme Judicial Court or Superior Court, St.1960, c. 216; appointed by the county commissioners, St. 1962, c. 758; and appointed by a vote of a board of aldermen or city council, St. 1964, c. 55, § 1. [9] “An Analysis of Ballot Position and Incumbency as Influencing Factors” (unpublished 1972). This study was commissioned by the then Attorney General of Massachusetts. [10] Byrne and Pueschel have also written another survey, entitled “But Who Should I Vote for for County Coroner” (unpublished). Therein they analyze approximately five hundred Democratic and Republican county central committee elections in California from 1948 to 1970. Contrary to plaintiff’s contention, they found that in those elections first place conferred no advantage; second, third, and fourth places conferred a 10% disadvantage; and the lower positions experienced an increasing advantage, with last position being the most favored. [11] In testimony during the case of Gould v. Grubb, supra note 2, Dr. Bain generally reiterated his written findings of twenty years ago. Dr. Bain has, since 1957, done other statistical analyses of voting, but none which related to position effect. [12] Byrne and Pueschel criticized the Bain and Hecock study as time-bound, geographically limited, and confusing with respect to the role of voting machines. [13] H. White, “Voters Plump for First on the List”, National Municipal Review, Feb., 1950, at 110. Byrne and Pueschel criticize this study as an attempt by the author, a candidate in that election, to explain why he lost. [14] J. Mueller, “Choosing Among 133 Candidates” (unpublished). [15] W. Scott, Jr., “California Ballot Position Statutes: An Unconstitutional Advantage to Incumbents”, 45 S.Cal.L.Rev. 369 (1972). Byrne and Pueschel criticized this study as covering only a small number of elections and hence candidates, and as reporting mixed results, i. e., that certain candidates received fewer votes in first position than in other positions. [16] C. Bagley, “Does Candidates’ Position on the Ballot Paper Influence Voters’ Choice? A Study of the 1959 and 1964 British General Elections”, 19 Parliamentary Affairs 173 (1966). Bagley’s premise, as described by Byrne and Pueschel, was that name attractiveness is a more important variable than ballot position; names beginning with letters near the front of the alphabet are the most attractive and therefore the most preferred.