Opinion
C.A. No. 08-468 S.
December 12, 2008
Timothy J. Robenhymer, Esq., Warwick, RI, for Plaintiffs.
Joseph Avanzato, Esq., Adler Pollock Sheehan, P.C., Raymond A. Marcaccio, Esq., Oliverio Marcaccio, Armando E. Batastini, Nixon Peabody LLP, Providence, RI, David V. Igliozzi, Igliozzi Reis, LLP, Cranston, RI, for Defendants.
AMENDED OPINION AND ORDER
Pending before the Court is Plaintiffs' Motion for Preliminary Injunction. Also, before the Court is a Motion to Dismiss filed by the Rhode Island Board of Elections. Plaintiffs seek to enjoin the Board of Elections and the Board of Canvassers for the Town of Smithfield, Rhode Island from issuing certificates of election to the prevailing candidates in the November 4, 2008 election for Smithfield Town Council. Plaintiffs, who also seek class certification, are voters from the Town of Smithfield who claim that their civil rights were infringed in violation of 42 U.S.C. § 1983 as a result of having voted using ballots that included the name of a candidate who had withdrawn from the race.
For reasons discussed below, the Court will reserve ruling on this motion at this time.
Plaintiffs' amended complaint, filed approximately three weeks after the election, contains a total of five counts; however, only two causes of action are actually asserted. Count II claims a violation of Plaintiffs' right to substantive due process alleging it was fundamentally unfair for Plaintiffs to have received and voted using an incorrect ballot, part of which was not counted. Count III asserts a violation of the right to equal protection premised on the theory that the Plaintiffs were victims of discrimination because they were among those voters who received the incorrect ballots.
The Board of Elections argues that the equitable doctrine of laches bars this action because Plaintiffs waited three weeks after the election to file. The Court disagrees and believes two weeks to realize that a claim exists and find a lawyer, plus another week for that lawyer to prepare a case is not unreasonable; moreover, the Court perceives no prejudice. See In re Bankvest Capital Corp., 375 F.3d 51, 61 (1st Cir. 2004) (stating laches applies when a Court determines (1) plaintiffs' delay in bringing suit was unreasonable and (2) the delay has prejudiced the defendant).
The remaining counts represent Plaintiffs' requests for class certification and injunctive relief against the Board of Elections and the Smithfield Board of Canvassers. See Fed.R.Civ.P. 23(a) 65(a).
Plaintiffs originally filed this action the day before Thanksgiving. The Court granted Plaintiffs' request for a temporary restraining order preventing the Smithfield Board of Canvassers from certifying the election results to preserve the status quo and scheduled a preliminary injunction hearing for December 3, 2008. Subsequently, the Rhode Island Supreme Court issued a stay in a companion case involving a challenge filed by one of the losing candidates.
See Hawkins v. R.I. Bd. of Elections, No. 2008-318-M.P. (R.I. Dec. 4, 2008) (order granting motion to stay).
After conducting a preliminary injunction hearing, this Court finds that Plaintiffs have failed to demonstrate a substantial likelihood of success on the merits of their case. Accordingly, Plaintiffs' request for a preliminary injunction must be denied.
I. Findings of Fact
The Court finds the following facts from the stipulations of the parties and the evidentiary hearing conducted on December 3. On election day November 4, 2008, about 9,500 voters in the Town of Smithfield, Rhode Island participated in what was undoubtably our nation's most historic election. In addition to casting their vote for President of the United States, qualified voters were also electing new members to the Smithfield Town Council. The Town Council for the Town of Smithfield is comprised of five members, all of whom were up for re-election. The field of thirteen candidates consisted of four Democrats, five Republicans, and four Independents. Voters were allowed to vote for a maximum of five candidates and the top five vote-getters would then make up the new governing body. Unfortunately, the conduct of the election was far from flawless.
During the morning hours of the election, from about 7 a.m. to 10 a.m., at polling locations throughout the Town approximately 2,900 voters were given ballots that contained the name of Richard A. DiIorio, Jr. ("Mr. DiIorio"), a candidate who had withdrawn from consideration well before the election. Voters who cast a vote for Mr. DiIorio on the incorrect ballot did not have that one vote counted. All other votes contained on the incorrect ballot were considered in determining the outcome of the election.
After the votes were counted (and recounted), only a 39 vote differential separated the fifth place candidate, Ms. Maxine Cavanagh ("Ms. Cavanagh") a Republican, and the sixth place candidate, Mr. Bernard Hawkins ("Mr. Hawkins") a Democrat. The election of Ms. Cavanagh would result in a change of majority control of the Town Council from Democrat to Republican.
At the evidentiary hearing, the Director of the Elections Division for the Rhode Island Secretary of State, Ms. Janet L. Ruggiero ("Ms. Ruggiero"), testified that the mistake occurred because an employee at the company hired to print the ballots had inadvertently used the wrong digital file to print the official ballots. Ms. Ruggiero explained that the Secretary of State had initially approved an electronic ballot template containing Mr. DiIorio's name on September 22, 2008. The Secretary of State then forwarded the September 22 template to the printing company responsible for printing the ballots. Eight days later, however, the Elections Division received a certificate of withdrawal from Mr. DiIorio directing that his name be taken off the ballot. The very next day, October 1, 2008, the Elections Division contacted the printer to remove Mr. DiIorio's name from the ballot. As requested, the printer made the change, which was reflected in the fact that both the mail ballots and the sample ballots (which were printed from the correct files) did not contain Mr. DiIorio's name; however, when it came time to print the official ballots in mid-to-late October, the employee in charge of printing those ballots failed to use the updated version saved on the company's computer network system. Instead, he relied on the outdated ballot that he had previously saved to his workstation hard drive and had not updated. Ballots were then printed using the wrong file and sent to the various polling locations. Election officials did not notice the mistake until the morning of election day.
To the credit of the Secretary of State, after being notified of the mistake around 8 a.m., his office immediately procured corrected ballots from the printer and distributed them to Smithfield polling locations by 10 a.m. The damage, however, had already been done. During the three hour period before corrected ballots could be issued, approximately 2,900 people voted using the wrong ballot, 570 of whom cast a vote for Mr. DiIorio.
At the conclusion of the hearing, Plaintiffs and the Intervenor, Ms. Cavanagh, submitted a Joint Statement of Facts. For purposes of this decision, the Court will assume, as Plaintiffs contend, that 570 people actually voted for Mr. DiIorio using the incorrect ballot. And, of those 570 ballots, the Court finds (again relying on Plaintiffs' assertion) that 458 contain a vote for Mr. Hawkins. Thus, from this evidence, the Court deduces that Mr. Hawkins could have garnered additional support from only 112 of the 570 ballots.
Without objection, the Court approved Ms. Cavanagh's motion to intervene in this action at a status hearing held on December 1, 2008.
The Court notes that despite the title there is very little agreement contained in the document. Nonetheless, the Court will seek to give Plaintiffs the benefit of the doubt and will primarily rely on their figures.
The Court assumes that the approximately 2,300 other voters who received the incorrect ballot and did not cast a vote for Mr. DiIorio were not influenced at all by his presence and thus would not have changed their vote using a correct ballot.
The Joint Statement of Facts also indicates there were 11 under-votes and 11 over-votes (9 of which contained a Hawkins vote). Thus, these 11 under-votes must be subtracted from the 112 potentially outcome altering ballots because those voters exercised their right not to vote for a full slate of candidates and were not forced to choose between Mr. DiIorio and other candidates. It is well known that voters sometimes choose to "maximize" their vote by voting for fewer than the allowable number of candidates and not for the maximum allowed. No evidence was presented to suggest that voters who under-voted were doing anything other than maximizing their vote; to count these votes as potential Hawkins votes would be contrary to common sense and the evidence. Additionally, the 2 over-votes that did not contain a Hawkins vote must also be subtracted because it seems clear those voters made the conscience choice not to vote for Mr. Hawkins. Thus, the Court concludes that the number of votes that potentially could have affected the outcome of the election between the fifth and sixth place finishers is 99. It is on these facts that the ensuing analysis is based.
The statement defines under-vote as a ballot where the voter cast a vote for less than the five candidates and over-vote as a ballot where the voter cast a vote for more than the five candidates.
See e.g. Harmon v. Baldwin, 107 Ohio St.3d 232, 837 N.E.2d 1196, 1201 (2005) (stating a voter's intent in an under-vote could reasonably be determined as being an intent not to vote).
II. Discussion
The standard for obtaining preliminarily relief is familiar. In order to obtain a preliminary injunction, a Plaintiff must demonstrate: (1) a significant risk of irreparable harm absent the injunction; (2) a substantial likelihood of success on the merits of their claim; (3) the scales balancing the hardship tip in their favor; and (4) the injunction will not harm the public interest. Wine Spirits Retailers, Inc. v. Rhode Island, 418 F.3d 36, 46 (1st Cir. 2005); see also New Comm Wireless Servs., Inc. v. SprintCom, Inc., 287 F.3d 1, 9 (1st Cir. 2002) ("[t]he sine qua non of this four-part inquiry is likelihood of success on the merits"); Mass. Coalition of Citizens with Disabilities v. Civil Def. Agency Office of Emergency Preparedness of Commonwealth of Mass., 649 F.2d 71, 76 n. 7 (1st Cir. 1981) (stating that a preliminary injunction should be used sparingly). Moreover, in matters involving state and local election disputes, federal courts must tread lightly and strive to avoid becoming involved in matters that are more appropriately the ken of local election officials and state courts.
A. Due Process
The right to vote is a bedrock principle of our democracy. See Reynolds v. Sims, 377 U.S. 533, 554, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964); Yick Wo v. Hopkins, 118 U.S. 356, 370, 6 S.Ct. 1064, 30 L.Ed. 220 (1886). Interference with the right to vote can in some circumstances raise constitutional claims where voters are disenfranchised or where the vote is sufficiently diluted. Griffin v. Burns, 570 F.2d 1065, 1077 (1st Cir. 1978); see also Rossello-Gonzalez v. Calderon-Serra, 398 F.3d 1, 16 n. 29 (1st Cir. 2004) (citing Bush v. Gore, 531 U.S. 98, 121 S.Ct. 525, 148 L.Ed.2d 388 (2000)). However, "[d]espite this bedrock federal interest, a federal court may not inject itself into the midst of every local electoral dispute." Bonas v. Town of N. Smithfield, 265 F.3d 69, 74 (1st Cir. 2001). "The federal court is not equipped nor empowered to supervise the administration of a local election. If every election irregularity or contested vote involved a federal violation, the court would 'be thrust into the details of virtually every election, tinkering with the state's election machinery, reviewing petitions, registration cards, vote tallies, and certificates of election for all manner of error and insufficiency under state and federal law.'" Griffin, 570 F.2d at 1077 (quoting Powell v. Power, 436 F.2d 84, 86 (2d Cir. 1970)).
With these principles of nonintervention in mind, the Court has given careful consideration to the Board of Elections' request that the Court abstain. However, the Court declines this invitation at this stage and notes the decision to abstain is highly discretionary. See Duncan v. Poythress, 657 F.2d 691, 696 (5th Cir. Unit B 1981) (stating the extraordinary nature of abstention stems from the duty of decision imposed upon the district courts by Congress). It is insufficient to justify a decision to abstain because a case concerns a state electoral process. See Smith v. Cherry, 489 F.2d 1098, 1101 (7th Cir. 1973). Although the First Circuit has recognized abstention is plausible in state election cases, see Bonas v. Town of N. Smithfield, 265 F.3d 69, 76 n. 5 (1st Cir. 2001) this Court believes the constitutional nature of Plaintiffs' claims compel the exercise of jurisdiction, at least to resolve this request for preliminary relief. If this case proceeds beyond this stage, the Court may reconsider the abstention question.
Before an election error becomes a key that unlocks the restraints on the federal court's authority to act, the Plaintiffs must demonstrate either an intentional election fraud or an unintentional error resulting in broad-gauge unfairness. Griffin, 570 F.2d at 1077; see also Navedo v. Acevedo, 752 F.Supp. 523, 528-530 (D.P.R. 1990), aff'd, 932 F.2d 94 (1st Cir. 1991).
Here, Plaintiffs concede, as they must, there are no suggestions of intentional fraud. Instead, Plaintiffs proceed on the theory that the unintentional injection of incorrect ballots into the election process created a patent and fundamental unfairness. Plaintiffs must clear a high bar to prevail on this theory, however, because the First Circuit has held that save for a "total and complete disenfranchisement of the electorate as a whole" garden variety election errors do not typically harbinger patent and fundamental unfairness. Bonas, 265 F.3d at 75.
The First Circuit's leading cases in this area suggest several factors for the Court to consider in assessing whether an election error rises to the level of fundamental unfairness. See Rossello-Gonzalez, 398 F.3d at 14; Bonas, 265 F.3d at 73; Partido Nuevo Progresista v. Barreto Perez, 639 F.2d 825, 827 (1st Cir. 1980); Griffin, 570 F.2d at 1077. These include but are not limited to: the breadth of the unfairness, whether the unfairness played a role in determining the outcome, whether the error induced any detrimental reliance of the part of the voters, and whether there is an adequate state administrative and judicial corrective process.
The breath of the unfairness here is plain based on the number of incorrect ballots utilized (2,900) out of the total cast (9,500). These numbers, however, only tell a small part of the story as the discussion below reveals.
Although not germane to the constitutional analysis, the Court notes the similarity of these factors to the standard outlined in the leading Rhode Island Supreme Court election case. See Buonanno v. DiStefano, 430 A.2d 765 (R.I. 1981). In that case, the Rhode Supreme Court clearly expressed that there is a strong public policy favoring stability and finality of election results and that an election should not be upset absent a compelling reason. Id. at 770. A mere mathematical possibility, as opposed to solid probability, that an error affected the results is insufficient to disturb an outcome. Id.
Plaintiffs' complaint appears to sound in detrimental reliance: that voters were induced by the incorrect ballot to vote for Mr. DiIorio. To convince the Court of this claim, Plaintiffs only offered their self serving affidavits. They presented no evidence as to whether any voters actually relied on the presence of Mr. DiIorio to the exclusion of Mr. Hawkins or anyone else; nor was evidence presented as to whether election officials took steps to inform morning voters they were receiving an incorrect ballot. However, even if the Court were to assume voters relied to their detriment on the ballot, Plaintiffs have totally failed to show that this fact somehow could have made a difference in the outcome.
When reviewing evidence of election results, it is important to keep in mind that "mathematical certainty" is not required, and instead Plaintiffs may show that the irregularity could have altered the outcome. Griffin, 570 F.2d at 1080. At the close of the hearing, the Court requested, and the parties subsequently submitted, the complete Town Council election results on a precinct basis for both sets of ballots (those that contained Mr. DiIorio's name and those that did not). The Court also requested a recount report for Ms. Cavanagh and Mr. Hawkins generated by the Board of Elections that includes the final totals for ballots cast with and without Mr. DiIorio's name. At the hearing, Plaintiffs presented no expert to assist the Court in analyzing this evidence for outcome determinativeness. Rather, Plaintiffs relied on their bald assertions that because the number of ballots containing a vote for Mr. DiIorio but not a vote for Mr. Hawkins is greater than the margin separating the fifth and sixth place finisher, the outcome could have been affected. However, this argument, like most easy solutions, is "neat, plausible, and wrong."
See H.L. Mencken, The Divine Afflatus, in A Mencken Chrestomathy. chapter 25, p. 443 (1949) ("There is always an easy solution to every human problem — neat, plausible, and wrong.").
In determining whether Mr. Hawkins could have secured victory from the 99 voters the Court earlier deduced could have made difference one thing is clear: for Mr. Hawkins to have overcome the 39 vote advantage held by Ms Cavanagh, he would have to receive over 40 percent of those 99 votes. However, based Mr. Hawkins's performance on the corrected ballot, it seems nearly certain that he has virtually no chance of conquering this deficit.
In accordance with a district court's inherent authority to appoint technical advisors, see Reilly v. United States, 863 F.2d 149, 155 (1st Cir. 1988); see also Ex parte Peterson, 253 U.S. 300, 312-13, 40 S.Ct. 543, 64 L.Ed. 919 (1920), the Court contacted Jennifer L. Lawless, Ph. D., Associate Professor of Political Science at Brown University, to assist the Court with analyzing the election data that the Court requested. Professor Lawless is a nationally renowned scholar and a leading authority in the field quantitative analysis of election data. At best, the calculations Professor Lawless ran suggest Mr. Hawkins would have received 12 percent of the 99 ballots, leaving him far short of overcoming Ms. Cavanagh's lead. Her computations of the election data suggest a "compelling statistical improbability" that Mr. DiIorio's name being on the ballot cost Mr. Hawkins the election. The Court concludes, as Professor Lawless advises, that:
The Court notes that it asked Professor Lawless to advise in this case because she possesses the specialized skill of applying statistical analysis to election data. While the Court could have undertaken this type of analysis own its own, the prudent approach was to do so guided by the hand of a neutral and detached academic. The Court greatly appreciates the willingness of Professor Lawless to assist in this case on short notice and without compensation. Attached as an appendix to this decision is the report prepared by Professor Lawless that aided the Court's review of the evidence.
Considering that the aggregate and precinct-level analysis demonstrates that Mr. Hawkins did not fare considerably worse in the places we might expect when Mr. DiIorio's name appeared on the ballot, coupled with the fact that the data imputation in the most plausible scenario does not alter the outcome of the election, it seems highly unlikely that even more precise estimates would predict a 5th place finish for Mr. Hawkins.
While Professor Lawless's report does leave room for the remote possibility that a more in-depth analysis might reveal a different conclusion, the Court concludes based on the mathematical calculations she performed that it is not probable.
Professor Lawless performed a total of six statistical analyses on the election data evidence. And, in only one analysis was she able to produce a result different from election day with Mr. Hawkins coming in fifth. See Table 2 of Appendix I. However, the Court agrees with Professor Lawless that this scenario was unlikely to have occurred and is reminded you can "torture numbers, and they will confess to anything." Gregg Easterbrook, The Progress Paradox, p. 10 (2003).
Furthermore, Professor Lawless's report also assists in debunking Plaintiffs' related contention that another candidate, in addition to Mr. Hawkins, was harmed by the incorrect ballot. At the hearing, Plaintiffs suggested Richard A. Poirier ("Mr. Poirier") may have been able to prevail but for the erroneous ballot. However, it appears that Mr. Poirier would have actually faired worse not better if the correct ballot had been in use for the entire election. It is also inconceivable to believe that Mr. Poirier would have received over 50 percent of the 536 votes Plaintiffs claim altered the outcome of Mr. Poirier's race given that no candidate received more than 12 percent of the vote.
The Court need go no further. Plaintiffs have not met their burden of showing a likelihood of success on the merits of their due process claim and to probe the remaining factors would be pointless. See New Comm Wireless, 287 F.3d at 9 ("if the moving party cannot demonstrate that he is likely to succeed in his quest, the remaining factors become matters of idle curiosity").
With regard to the adequacy of the state process, the Board of Elections took the position that Plaintiffs should have filed their complaint as part of the challenge filed by Mr. Hawkins. However, when pressed the Board pointed to no authority other than general language in the Board's enabling legislation that supports the contention that affected voters have standing to pursue such matters in front of the Board of Elections. See R.I. Gen. Laws § 17-7-5(a)(1), (11). This Court need not resolve that question here.
B. Equal Protection
Plaintiffs' argument that their likelihood of success is stronger on their equal protection claim fares no better. While Plaintiffs may have been discriminated against on election day, the Supreme Court has made clear that "the Equal Protection Clause does not make every minor difference in the application of laws to different groups a violation of our Constitution." Williams v. Rhodes, 393 U.S. 23, 30, 89 S.Ct. 5, 21 L.Ed.2d 24 (1968). If Plaintiffs fell into a "discrete monitory" once election officials made the decision to switch out the erroneous ballots with the corrected ones, see Bonas, 265 F.3d at 74, this was not because of some arbitrary treatment designed to discriminate. See Bush v. Gore, 531 U.S. 98, 104, 121 S.Ct. 525, 148 L.Ed.2d 388 (2000); see also Torres-Rivera v. Calderon-Serra, 412 F.3d 205, 211 n. 7 (1st Cir. 2005) (stating that under the equal protection clause of the Fourteenth Amendment a showing of disproportionate impact alone is not enough to establish a constitutional violation; plaintiff still must show purposeful discrimination). Rather, election officials decided to advance the legitimate and valid goal of preservation of the integrity of the electoral process. Rosario v. Rockefeller, 410 U.S. 752, 761, 93 S.Ct. 1245, 36 L.Ed.2d 1 (1973). Moreover, the First Circuit has refused to find violations of equal protection in situations where the rationale for the decision to discriminate is based on avoiding voter confusion and assuring that the winner is the choice of a majority, or at least a large plurality, of those voting. See Hopfmann v. Connolly, 769 F.2d 24, 25 (1st Cir. 1985) (quoting Storer v. Brown, 415 U.S. 724, 732, 94 S.Ct. 1274, 39 L.Ed.2d 714 (1974)). Thus, the Plaintiffs' equal protection claim is not stronger, and in fact may even be weaker on the merits than their due process claim. Given the analysis above, Plaintiffs have failed again to present evidence sufficient to meet their burden of likelihood of success on the merits.
III. Conclusion
Based on the foregoing, Plaintiffs' Motion for Preliminarily Injunctive Relief is DENIED.
The Defendant Rhode Island Board of Elections' Motion to Dismiss is DENIED in part (laches); the Court reserves ruling on Defendants' other motions.
IT IS SO ORDERED.
December 8, 2008
APPENDIX
December 8, 2008
Honorable William E. Smith
District Judge
United States District Court
BROWN UNIVERSITY Department of Political Science, Box 1844 Providence, RI 02912 (401) 863-1575 fax: (401) 863-7018 jennifer_lawless@brown.edu
One Exchange Terrace
Providence, RI 02903
Re: Joseph M. Bennett, et al. v. Ralph Mollis, et al.
C.A. No. 08-468S
Dear Judge Smith,
Based on a series of conversations last week, you asked me to analyze the election returns from the November 4, 2008 Smithfield Town Council race and attempt to determine how many votes Mr. Hawkins would have drawn from ballots that included a vote for Mr. DiIorio, but not Mr. Hawkins.
Let me note from the outset that we are somewhat limited in the extent to which we can draw conclusions about individuals' voting patterns when all we have to analyze are precinct-level data. In other words, we cannot predict with a high degree of precision the percentages of the DiIorio vote that would have gone to each of the other candidates. Nevertheless, we can discern certain patterns in voting behavior across precincts, as well as generate comparisons between the total number of votes Mr. Hawkins received when Mr. Di-Iorio's name appeared on the ballot and when it did not. Ultimately, the precinctlevel data analysis suggests a compelling statistical improbability that Mr. DiIorio's name on the ballot cost Mr. Hawkins a 5th place finish.
Below, I present the data and logic through which I arrive at this conclusion.
I. Imputing the Morning Vote
Perhaps the best place to begin is with an assessment of the difference in the total percentage of the vote each candidate received in the morning (when Mr. DiIorio's name appeared on the ballot) versus the afternoon (once Mr. DiIorio's name was removed). Because we can consider each vote as the unit of analysis, we can remove the DiIorio votes from our calculations, but still count the other votes on each ballot.
The overall data indicate that Mr. Hawkins performed better when Mr. DiIorio's name did not appear on the ballot. Indeed, when Mr. DiIorio's name appeared, Mr. Hawkins placed 7th overall, compared to his 5th place finish when voters did not have the option of choosing Mr. DiIorio. The issue, therefore, is whether the votes Mr. DiIorio drew in the morning account for Mr. Hawkins' loss. In other words, would the votes Mr. DiIorio received at Mr. Hawkins' expense be enough to put Mr. Hawkins over the top?
In order to answer this question, we can consider three scenarios by which to impute the number of votes each candidate would have received in the morning had Mr. DiIorio's name not appeared on the ballot. In Scenario A (Table 1), I calculated the "imputed morning vote" working under the assumption that Mr. DiIorio's votes would be distributed across the remaining 13 candidates in proportion to the vote each drew in the morning. Without access to individual ballots (through which we could assess patterns and combinations of vote choices), this is a plausible way to proceed. Table 1: Vote Totals and Candidate Rank Based on Imputing the Morning Vote (Scenario A) Percent Vote Total Imputed AM Total Vote Total Received in the Vote Received (PM Total + Percent Overall Overall AM (excluding (redistributing Imputed of Vote Rank Rank DiIorio Votes) DiIorio Votes) AM Total) (Imputed) (Imputed) (Actual) Archambault Flynn Hawkins Cavanagh Cerroni Poirier Fanning Coutu Manni Begin Esposito LaBrie Tocco Dilorio
0.105339924 1359.833077 4598.8331 0.109436 1 1 0.110606920 1427.824731 4414.8247 0.105057 3 3 0.090430273 1167.364395 3874.3644 0.092196 6 6 0.098857467 1276.151041 3973.1510 0.094547 5 5 0.103233125 1332.636415 4461.6364 0.106171 2 2 0.097074791 1253.138481 3793.1385 0.090263 7 7 0.089295843 1152.720039 3776.7200 0.089873 8 8 0.064743538 835.7743295 2529.7743 0.060200 9 9 0.098209221 1267.782838 4028.7828 0.095871 4 4 0.038165465 492.6779840 1672.6780 0.039804 10 10 0.035086298 452.9290171 1627.9290 0.038739 12 12 0.035734543 461.2972206 1624.2972 0.038653 13 13 0.033222591 428.8704319 1646.8704 0.039190 11 11 — — — — — 14 Notes: Imputations are based on reallocating Mr. DiIorio's votes in proportion to the votes each of the candidates received when Mr. DiIorio's name appeared on the ballot. All calculations rely on the recount tallies. In this scenario, Mr. Hawkins still finishes 6th, and Ms. Cavanagh leads Mr. Hawkins for 5th place by roughly 99 votes.Scenario B errs on the side of Mr. Hawkins and assumes that the voting patterns we saw in the afternoon are exactly what the voting patterns would have looked like in the morning had Mr. DiIorio's name not appeared on the ballot (see Table 2). In this case, Mr. Hawkins actually moves into 5th place and replaces Ms. Cavanagh as the winner by a margin of 14 votes. Table 2: Vote Totals and Candidate Rank Based on Imputing the Morning Vote (Scenario B) Percent Vote Total Imputed AM Total Vote Total Received in the Vote Received (PM Total + Percent Overall Overall AM (excluding (redistributing Imputed of Vote Rank Rank DiIorio Votes) DiIorio Votes) AM Total) (Imputed) (Imputed) (Actual) Archambault Flynn Hawkins Cavanagh Cerroni Poirier Fanning Coutu Manni Begin Esposito LaBrie Tocco DiIorio
0.105339924 1436.156179 4675.156179 0.111252 1 1 0.110606920 1324.420657 4311.420657 0.102597 3 3 0.090430273 1200.270076 3907.270076 0.092979 5 6 0.098857467 1195.836127 3892.836127 0.092636 6 5 0.103233125 1387.382737 4516.382737 0.107474 2 2 0.097074791 1126.223123 3666.223123 0.087243 8 7 0.089295843 1163.468297 3787.468297 0.090128 7 8 0.064743538 751.1110119 2445.111012 0.058185 9 9 0.098209221 1224.213402 3985.213402 0.094834 4 4 0.038165465 523.2060177 1703.206018 0.040530 11 10 0.035086298 520.9890431 1695.989043 0.040359 12 12 0.035734543 515.6683039 1678.668304 0.039946 13 13 0.033222591 540.0550251 1758.055025 0.041836 10 11 — — — — 14 Notes: Imputations are based on reallocating Mr. DiIorio's votes in proportion to the votes each of the candidates received when Mr. DiIorio's name did not appear on the ballot. All calculations rely on the recount tallies. There are several theoretical reasons to believe, however, that the patterns we saw in the afternoon (when Mr. DiIorio's name did not appear on the ballot) would not be mirrored exactly in the morning (when his name did appear). The percentage of straight ticket voters, for example, was 6.5 percent higher in the morning than in the afternoon (17.8 percent in the morning, compared to 16.7 percent on the afternoon). This might seem like a minor difference, but in such a close election, it is important, especially since it suggests that Mr. Hawkins drew more votes in the morning than he otherwise might have because of the straight ticket benefit.Thus, in Scenario C, I split the difference and impute the morning vote such that it is the average total vote share each candidate received in the morning and the afternoon. Under this scenario (see Table 3), Mr. Hawkins does not retake the lead; Ms. Cavanagh maintains her 5th place finish (leading Mr. Hawkins by 42 votes) and the election results do not change.
Overall, therefore, based on the most plausible scenario by which to reallocate the votes Mr. DiIorio received, Mr. Hawkins falls short of gaining a sufficient number of votes to overtake Ms. Cavanagh and place 5th. Table 3: Vote Totals and Candidate Rank Based on Imputing the Morning Vote (Scenario C) Percent Vote Total Imputed AM Total Vote Total Received in the Vote Received (PM Total + Percent Overall Overall AM (excluding (redistributing Imputed of Vote Rank Rank DiIorio Votes) DiIorio Votes) AM Total) (Imputed) (Imputed) (Actual) Archambault Flynn Hawkins Cavanagh Cerroni Poirier Fanning Coutu Manni Begin Esposito LaBrie Tocco DiIorio
0.105339924 1397.994628 4636.994628 0.110344207 1 1 0.110606920 1376.122694 4363.122694 0.103827016 3 3 0.090430273 1183.817236 3890.817236 0.092587803 6 6 0.098857467 1235.993584 3932.993584 0.093591452 5 5 0.103233125 1360.009576 4489.009576 0.106822682 2 2 0.097074791 1189.680802 3729.680802 0.088753321 8 7 0.089295843 1158.094168 3782.094168 0.090000575 7 8 0.064743538 793.4426707 2487.442671 0.059192411 9 9 0.098209221 1245.998120 4006.998120 0.09535250 4 4 0.038165465 507.9420008 1687.942001 0.040167099 12 10 0.035086298 486.9590301 1661.959030 0.039548795 11 12 0.035734543 488.4827623 1651.482762 0.039299497 13 13 0.033222591 484.4627285 1702.462728 0.040512641 10 11 — — — — — 14 Notes: Imputations are based on reallocating Mr. DiIorio's votes in proportion to the votes each of the candidates received overall (when Mr. DiIorio's name did not appear on the ballot). All calculations rely on the recount tallies.II. Precinct-Level Analysis
The second way we can analyze the data involves looking at the patterns within each precinct and determining whether Mr. DiIorio's name on the ballot affected certain precincts more than others. If the presence of Mr. DiIorio's name weakened Mr. Hawkins disproportionately in the precincts where he fared the best in the afternoon, for instance, then that suggests that the outcome may have been different had Mr. DiIorio's name not appeared on the ballot.Table 4 presents Mr. Hawkins' vote share in each precinct when Mr. DiIorio's name appeared on the ballot, and when it did not. In general, the data do not support the notion that Mr. DiIorio's name systematically disadvantaged Mr. Hawkins.
More specifically, we see that Mr. Hawkins' afternoon vote share exceeded his morning vote share in 8 of the 10 precincts. But his margin did not disproportionately increase in the precincts where he fared the worst in the morning. For example, precincts 3103, 3105, and 3106 were Mr. Hawkins' 3 weakest precincts when Mr. DiIorio's name appeared on the ballot. Precincts 3105 and 3106 remained among his 3 weakest precincts when Mr. DiIorio's name did not appear on the ballot. In other words, as we see in the last column of Table 4, Mr. Hawkins' vote share in the afternoon was quite consistent with his morning performance in most precincts. In short, Mr. Hawkins does not appear to have suffered a sizeable systematic disadvantage. Table 4: Precinet-Level Comparison of Mr. Hawkins' Vote Share AM Vote Percentage AM Vote Percentage PM Vote Margin of Increased Precinct (with DiIorio votes) (excluding DiIorio votes) Percentage Support in the PM
3101 0.092166 0.095808 0.095499 0.003333 3102 0.095682 0.100444 0.101345 0.005663 3103 0.077277 0.081003 0.094841 0.017564 3104 0.091698 0.095855 0.098879 0.007181 3105 0.077807 0.082090 0.089990 0.012183 3106 0.078614 0.082058 0.082371 0.003757 3107 0.083703 0.088098 0.092201 0.008498 3108 0.085938 0.089249 0.082633 -0.003305 3109 0.094146 0.097867 0.108805 0.014659 3110 0.086372 0.090361 0.077972 -0.008400 Notes: All calculations rely on the recount tallies. It is also important to recognize that we cannot view Mr. Hawkins' precinct performance separate from that of Ms. Cavanagh. Indeed, Ms. Cavanagh performed better in 4 of the 10 precincts when Mr. DiIorio's name did not appear on the ballot, too (see Table 5). Table 5: Precinct-Level Comparison of Ms. Cavanagh's Vote Share AM Vote Percentage AM Vote Percentage PM Vote Margin of Increased Precinct (with DiIorio votes) (excluding DiIorio votes) Percentage Support in the PM 3101 0.095622 0.099401 0.094236 -0.001386 3102 0.087214 0.091556 0.087593 0.000379 3103 0.095676 0.100289 0.084598 -0.011078 3104 0.097893 0.102332 0.089106 -0.008787 3105 0.095491 0.100746 0.092686 -0.002805 3106 0.097935 0.102225 0.096484 -0.001451 3107 0.091463 0.096266 0.092566 0.001103 3108 0.110352 0.114604 0.104575 -0.005777 3109 0.079662 0.082811 0.083648 0.003986 3110 0.102687 0.107430 0.107371 0.004684 Notes: All calculations rely on the recount tallies.Finally, we should note that even though Mr. Hawkins fared disproportionately better than did Ms. Cavanagh in the morning than in the afternoon, Ms. Cavanagh still outperformed Mr. Hawkins in 5 of the 10 afternoon precincts. Ms. Cavanagh won precincts 3105, 3106, 3107, 3108, and 3110 in both the morning and the afternoon. This suggests, then, that at least in half of the precincts, Mr. DiIorio's name on the ballot did not produce a different result between Mr. Hawkins and Ms. Cavanagh. If Mr. Hawkins is to recoup the votes that separate him from Ms. Cavanagh, then he would likely need to net them all in only half of the precincts, thereby further reducing his likelihood of success.
III. Analyzing the Questionable Ballots
The final approach we can take to assessing whether Mr. DiIorio's name affected the outcome of the election involves focusing on the ballots that included a vote for Mr. DiIorio, but not a vote for Mr. Hawkins. Once we omit ballots with under votes and over votes, my understanding is that 99 ballots included a vote for Mr. DiIorio, but not Mr. Hawkins. I am also under the impression that the parties agree that 39 votes separate Mr. Hawkins from Ms. Cavanagh. If this is the case, then Mr. Hawkins would have to receive 40 of the 99 DiIorio votes.
Working from Mr. Hawkins' premise — that a vote for Mr. DiIorio came at the expense of Mr. Hawkins — we can assume that the other 4 votes on each of these 99 ballots would be unchanged by the removal of Mr. DiIorio's name. If such is the case, then Mr. Hawkins would need to receive 40 percent of the votes in question, which is an incredibly high bar, considering that in no precinct, in either the morning or the afternoon, did any one candidate receive more than 12 percent of the vote. Moreover, in each of these cases, we are dealing with split ticket voters, so Mr. Hawkins' party identification plays no role and there is no reason to expect that the overall correlation between the votes he and Mr. DiIorio received would hold for these 99 ballots.
Conclusion
Again, let me reiterate that unless we conduct a more sophisticated individual-level analysis of voters' ballots and the combinations of candidates they chose, we cannot arrive at more precise estimates. Considering that the aggregate and precinct-level analysis demonstrates that Mr. Hawkins did not fare considerably worse in the places we might expect when Mr. DiIorio's name appeared on the ballot, coupled with the fact that the data imputation in the most plausible scenario does not alter the outcome of the election, it seems highly unlikely that even more precise estimates would predict a 5th place finish for Mr. Hawkins.
That said, if the parties are interested in such an analysis, then I recommend that the focus be on precincts 3101, 3103, and 3104, precincts in which Mr. Hawkins placed ahead of Ms. Cavanagh when Mr. DiIorio's name did not appear on the ballot. Moreover, I suggest that we compare the 99 ballots in question to a random sample of afternoon ballots, weighted by precinct.
Please let me know if you need any additional information, or if you have any questions.
Sincerely,
Jennifer L. Lawless
Assistant Professor