The Florida Vote: What If It Had Happened Here?

Originally published in The Pennsylvania Lawyer (March-April 2001)

— Justice Hugo L. Black, Wesberry v. Sanders, 376 U.S. 1, 17 (1964).

Pennsylvania’s actual popular vote margin for the Gore-Lieberman ticket was more than 200,000 votes, as reported in the morning newspapers of Wednesday, Nov. 8, 2000. What if the margin in Pennsylvania had instead been a virtual tie — a narrow 1,210 votes — as was actually announced that morning for the Bush-Cheney ticket in Florida? What if Pennsylvania’s 25 electoral votes would have determined the next president? Would sorting out the will of the voters have been any easier in Pennsylvania — or a great deal more difficult — than the process attempted in Florida?

The efforts in Pennsylvania and Florida to count every precious vote, to leave no vote behind, can be compared or contrasted under six main points.

In the early morning of Wednesday, Nov. 8, the Florida secretary of state announced the official election results and explained that because the margin for president was less than half of 1 percent, Florida’s automatic recount statute would require a statewide recount. The Florida secretary of state did not specify how that recount would be conducted.

Pennsylvania’s Election Code does not require any official statewide election results immediately following an election. Indeed, in Pennsylvania, there are no official statewide election results until several weeks after an election. The statewide numbers that are published in the news media on the day following the election are assembled by the Associated Press, using a group of journalists stationed in each of the 67 counties who report to 15 persons in the AP Bureau in Philadelphia. In most of the counties, as the precinct election officials deliver the voting materials to the county boards of elections, the "unsealed" official return sheets are periodically totaled up by board of elections employees and given to the waiting journalists. In the seven counties using punch-card ballots, the ballots are unofficially counted at a central location in the county using the same card-reading equipment as will later be used in the official count.

One or two precincts from a few larger counties and several dozen precincts from Philadelphia are omitted from the totals created by the Associated Press before the process stops at 5:30 a.m. on Wednesday morning. The process is resumed later in the day on Wednesday and concluded that afternoon with totals that include results from more than 99 percent, but not quite 100 percent, of the 9,405 precincts in Pennsylvania.

For the numbers from Philadelphia, which contains 1,681 of the 9,405 precincts, the Associated Press relies on results gathered by a semi-official process financed in part by the news media. Under this system, unofficial forms — referred to as "flash returns" — are handwritten at each polling place, gathered up by police officers, and processed though a city-owned data entry system to generate a computer database of the votes cast in each precinct for each candidate. The Philadelphia process generally stops with more than two dozen polling places not reported.

The continued absence in Pennsylvania of any statutory procedure for statewide official vote totals on the day following the election may reflect the excellent job that the Associated Press has done at private expense in gathering results from more than 99 percent of the precincts. In recent years the secretary of the commonwealth has also posted unofficial county returns (and the official county returns) on the Department of State Web site (www.dos.state.pa.us). No one can recall when the absence of immediate official statewide results seemed to make a difference.

The omission of any statutory requirement that such totals be officially gathered may also reflect the historical fact that when the Election Code was adopted in 1937, each political party had an effective system of gathering accurate returns from an army of elected precinct committee workers. The duties of those precinct workers included bringing an accurate return sheet to a local party headquarters, where accurate tallies were kept and rapidly passed on to the state committees in Harrisburg. Although some of that party structure survives in some counties, there is no longer any dependable method by which the political parties in Pennsylvania can generate accurate statewide results on election night.

As the world learned, part of the Florida election statute seemed to require that each county file its certified vote totals with the secretary of state on the seventh day after the election, although the Florida Supreme Court held that the seven-day requirement could be extended to accommodate manual recounts.

Pennsylvania’s more leisurely official vote-counting process would have surprised journalists if Pennsylvania’s vote had decided the presidency. Pursuant to the 1937 Election Code, the process in each Pennsylvania county of publicly reading aloud and adding up the precinct results — called the "official canvass of the returns" — does not even start until the Friday following the election. In two large counties — Philadelphia and Montgomery — the board of elections convenes on Friday morning as required by the Election Code, but then adjourns until the following Monday. Because the Election Code also requires a five-day hiatus between the public announcement of the final total for each candidate and the county’s certification of its totals, a county that finishes the official count on the Friday after the election cannot certify until the next Wednesday, eight days after the election, so the secretary of the commonwealth does not start to receive county certifications from the smallest (or most efficient) counties until nine days after the election. Although the Election Code requires that all certifications be submitted to Harrisburg within 20 days of the election, no one can recall an election when all counties have finished within that statutory standard. In Pennsylvania, 30 days from the election has been the actual standard for the completion of the first official statewide vote totals, although the process can be accelerated when there is a reason for haste (as when a legislative office is vacant and the results of a special election in that district must be certified in order to swear in a successor and perhaps when the statewide vote will determine the next president).

Although both Pennsylvania and Florida, like the majority of states, delegate to county governments the selection of voting methods and the administration of elections, the Pennsylvania system is much more decentralized than in Florida. Unlike the prominent role played in Florida by the secretary of state, Pennsylvania’s secretary of the commonwealth has very limited powers concerning the count of an election. There is no power vested in the secretary of the commonwealth to promulgate regulations to control the vote count or the certification of returns or to issue binding advisory opinions. In a close presidential race, the role of the secretary of the commonwealth would likely be limited to an effort to cajole each county board of elections to finish its vote count more swiftly than usual and carry the certification to Harrisburg.

In Florida, an automatic recount follows from a less than half of 1 percent margin in any election. Pennsylvania’s Election Code lacks any procedure for an automatic statewide recount, regardless of the closeness of the vote. Moreover, the Pennsylvania Election Code omits any provision for a statewide recount at the request of a candidate or political party.

The only statutory method by which to obtain a recount of the votes cast in a Pennsylvania election is through an archaic procedure that requires the filing of a petition for each precinct signed by three voters of that precinct requesting a "recanvass." As required by a statutory formula, the petition must set forth allegations to the effect that "upon information which they consider reliable, they believe that fraud or error, although not manifest on the general return of votes made therefrom, was committed in the canvassing of the votes cast" in that precinct. If filed with the county board before the first vote totals are computed, which rarely occurs, no fee is required. If filed with the court of common pleas after the vote totals have been computed and announced, but not yet certified, the proponents of the petition are required to deposit $50 for each precinct to be recounted, with the money forfeited to the county unless "fraud or substantial error" is discovered when the recount occurs.

The statutory procedure is so harsh as to invite joint efforts by both sides in a close election to seek a recount without going through the precinct-by-precinct petitioning process. In 1974, when the votes for a state House of Representatives seat for Adams County appeared to be tied in the unofficial post-election news media results and the other seats in the state House were evenly split — 101 Republican and 101 Democratic — the president judge of Adams County approved a request from both candidates for a complete recount of all the 39,000 votes (cast on traditional paper ballots). The Delaware County Board of Elections in 1984 approved a joint request to recount the votes cast on mechanical voting machines and absentee ballots in an election for U.S. Congress. But these instances of joint requests for a complete recount in a particular district are not provided for in the Election Code, nor in living memory has there ever been in Pennsylvania a complete statewide recount.

In Philadelphia County, there is a unique tradition, not based on any explicit language in the Election Code. On the day after the official vote totals have been publicly announced, but before the vote is certified, all the voting machines are opened and made available for inspection in the voting-machine warehouse. Any candidate can examine the counters on the machines and compare those numbers with the numbers on the handwritten "official return sheets" that have been used to prepare the tentative official totals. This procedure was first ordered by Common Pleas Judge Harvey Schmidt in 1979, following a mistake in the layout of the official return sheets used in almost half the city in a hotly contested primary election. The mistake caused many return sheets to be erroneously completed on election night and required the retention of an accounting firm to prepare new official returns for each affected polling place. Schmidt ordered that all the voting machines be available for a free inspection without payment of the statutory $50 per precinct in an effort to allay fears that the vote had not been accurately recorded. In subsequent years, his precedent has been followed by the Philadelphia Board of Elections, notwithstanding that there is nothing in the Election Code that provides for the procedure.

Florida’s election statute provides for an automatic statewide recount if the margin is less than half of 1 percent but does not specify how that automatic recount is to be conducted. In Florida counties using punch-card ballots, the ballots were read a second time by machine, but in 20 other Florida counties the "recount" consisted only of reconciling precinct returns with the originally reported county totals, rather than a recount of the ballots themselves. When a manual recount is requested in a Florida county, the statutory procedure provides for an initial manual count of test precincts selected by the candidate requesting the recount and a further manual recount of the entire county at the discretion of the canvassing board of that county. In Miami-Dade, the canvassing board started but later declined to complete a manual recount, for reasons that were disputed between the lawyers (and journalists) and never definitively resolved.

Pennsylvania certainly needs a new and better recount procedure, but so does Florida.

The U.S. Supreme Court held in the second presidential election decision that Florida’s failure to set a detailed standard for the counting of disputed ballots created a violation of equal protection, which required the U.S. Supreme Court to terminate the contest process then under way in the Florida Circuit Court of Leon County. Although the Florida Supreme Court applied a statutory standard that no vote should be disregarded "if there is a clear indication of the intent of the voter," the U.S. Supreme Court held that this standard was not sufficiently precise and would permit differing interpretations, thus denying equal protection of the laws to some group of voters, presumably to the voters whose ballots were machine-readable and hence not subject to a difference of opinion on manual review.

The Florida Supreme Court had earlier seen its first decision vacated by the U.S. Supreme Court and remanded for reconsideration of whether the state court had erroneously relied on the Florida Constitution as a source of law by which to construe apparently conflicting provisions of the Florida election statutes. It had been urged by the Republican lawyers that reliance on the Florida Constitution to construe the election statutes would violate the provision of Article II, §1, cl. 2 of the U.S. Constitution by which the presidential electors of each state shall be appointed "in such Manner as the Legislature thereof may direct," thus arguably delegating the authority to the legislature of each state, not to the state itself. It had also been urged that because the Florida Supreme Court had attempted to reconcile the apparently conflicting provisions of the state law in a decision reached after election day, the Florida Court had violated the safe-harbor provision of 3 U.S.C. § 5 by "changing the rules of the game" after the game had been played. Hence the Florida Supreme Court, in its second ruling directing the manual recount of all ballots not machine-readable, appeared reluctant to establish any standard more detailed than the statutory mandate, borrowed from another part of the election statute, to count those ballots from which the canvassing board (or reviewing court) could determine the "clear intention of the voter." Republicans again accused the Florida Supreme Court of trying to change the law.

Gov. George W. Bush of Texas signed into law during 1999 a set of criteria for counting punch-card ballots. Those criteria include the same standards that his Florida lawyers argued were illegal and unconstitutional under the equal protection clause. The U.S. Supreme Court majority accepted that argument. If Florida had enacted a similar law prior to the 2000 election, would the U.S. Supreme Court have allowed the votes to be counted?

"Intent of the voter" is the standard in many states, including Pennsylvania, by which to determine whether an irregular ballot counts as a legal vote. All the voting methods that are used in Florida are used in Pennsylvania, but Pennsylvania’s Election Code does not provide specific or detailed standards for determining whether a punch-card ballot from which the chad is not detached should be counted as a legal vote. In a 1974 decision, the Pennsylvania Supreme Court held that the "intent of the voter" was the controlling standard for determining whether traditional paper ballots should be counted, despite irregularities, or rejected. The court thus discarded earlier precedents in which highly technical arguments had been used to disqualify ballots with unusual markings, such as an ill-defined "X" or a marking made with a pen other than the statutorily permissible "blue, black or blue-black ink." Those earlier precedents frequently relied on specific language in the Election Code as the basis on which to reject irregular ballots.

Because the standard of "intent of the voter" is established in Pennsylvania only by judicial precedent, it can be argued that Pennsylvania would have been even more vulnerable to the U.S. Supreme Court’s disapproval than was Florida. Indeed, in the Alice Through the Looking Glass world of the Article II, §1, cl. 2 analysis by the U.S. Supreme Court majority, the failure of the Pennsylvania Legislature to have set that standard might well have been a fatal constitutional infirmity. The temporal fact that the standard was declared by the Pennsylvania Supreme Court in 1974, long before the presidential election, would not solve the Article II problem because the Legislature of Pennsylvania had not acted to set the standard. If the choice of Pennsylvania’s presidential electors had depended on a few irregularly marked or punched ballots, the omission from the Pennsylvania Election Code of detailed and specific standards might have invited U.S. Supreme Court intervention. Nor is Pennsylvania’s judicially created standard any more precise than the Florida Supreme Court’s formula, which the U.S. Supreme Court struck down as insufficiently detailed.

If Pennsylvania had adopted, prior to the 2000 election, a detailed and precise statutory scheme for determining which punch-card ballots would be legal votes, as had been signed into law in Texas by Gov. Bush in 1999, the potential "equal protection" issue might have been avoided. And the same is true in Florida.

The "contest" phase of the Florida presidential election, which followed the termination on Nov. 26 of the "recount" phase, was assigned under Florida law to the Circuit Court of Leon County because that is the county in which the seat of the state government is located. Jurisdiction of a contested statewide election in Florida is similar to Pennsylvania procedure before 1968 — a statewide contest is tried in the circuit court of the county in which the state government is established, just as statewide election issues used to be tried in the Court of Common Pleas of Dauphin County. In Leon County, from the Democratic perspective, the initial delays in the progress of the contest phase were highly prejudicial.

Although Pennsylvania lacks any statutory provision for a statewide recount, the Election Code does have clear provisions for a statewide contest of the results, which must be filed within 20 days after the election. Pennsylvania’s creation in 1970 of Commonwealth Court as a centralized court for governmental litigation would likely have assured a better adjudication of such a contest proceeding than occurred in Leon County, Florida. Commonwealth Court not only handles statewide election issues but also decides appeals from election cases originally tried in the courts of common pleas. Commonwealth Court has developed an expertise in dealing with election cases fairly and expeditiously, exactly as the framers of the 1968 constitution intended. In Pennsylvania, Commonwealth Court would have done whatever was required to adjudicate a statewide contest within the time required.

The existence of Commonwealth Court might also have caused one side or the other to file a contest petition much earlier than the 20-day limit in the Election Code, in an effort to use the contest procedure — in a manner not contemplated by the Election Code — to obtain a centralized recount proceeding to avoid the "equal protection" issues that could have resulted from different interpretations county-by-county of irregular ballots.

This article looks at what might have happened in Pennsylvania if very few votes had separated the Gore-Lieberman and Bush-Cheney tickets, but it assumes that every other aspect of the election in Pennsylvania on Nov. 7 took place as it actually occurred that day. I conclude that Pennsylvania ultimately would have been vindicated, in the eyes of the nation and the world. Despite early embarrassments during the recount ("recanvass") phase, nothing in the conduct of the election, either by the counties or by the voters, would have given either side an issue on which to conduct a successful contest.

Pennsylvania, like Florida, allows each county to determine the method of voting that will be used in that county, subject to general standards set forth in the election statutes. Pennsylvania displays a crazy-quilt pattern of voting methods not dissimilar from Florida. Those methods include traditional paper ballots still used in a few counties, mechanical lever machines dating from the turn of the last century, punch-card ballots voted by removing a perforation with a stylus and read by machine, paper ballots marked by hand and read by an optical scanning machine, and state-of-the-art "direct recording electronic" machines. Despite Pennsylvania’s use of all the same methods of voting, several of which created major problems in Florida, there appears to have been nothing in Pennsylvania on Nov. 7 from which one might have constructed a contest petition.

The statistics for punch-card voting in Pennsylvania’s two largest counties that use that method are dramatically different from the Florida numbers. Chester County is the largest Pennsylvania county using punch-card ballots. Of the 189,750 voters on Nov. 7, there were only 701 over-votes in which the voter punched more than one perforation in the presidential part of the ballot card, or less than four-tenths of 1 percent (.00369), and 1,372 under-votes, in which no machine-readable space was punched through for president, or approximately seven-tenths of 1 percent (.00723), some of which may be intentional abstentions. Florida over-vote percentages to contrast with the percentages from Chester County are 4 percent in Palm Beach County and 8 percent in Duval County, where the presidential candidates’ names were spread over two ballot pages but the county’s official sample ballot, published in a newspaper before the election, erroneously instructed voters to "vote all pages." Florida’s under-vote percentages were 2.29 percent in Palm Beach County, 1.14 percent in Broward County and 1.64 percent in Miami-Dade County. Justice John Paul Stevens, relying on data presented in the 11^th Circuit litigation, stated that the percentage of nonvotes in the election in Florida counties using a punch-card system was 3.92 percent, (aggregating the over-votes and the under-votes).

Centre County is the next largest Pennsylvania county using punch-card balloting.

According to Joyce E. McKinley, director of voter registration and elections, of the 50,110 ballots cast on Nov. 7, there were 164 were over-votes for president, also less than four-tenths of one percent (.00327), and 477 were under-votes for president, a slightly higher percentage than in Chester County but still less than one percent (.00951).

There are demographic differences between these two Pennsylvania counties and the three Florida counties, but one factor not commented upon in reports of the Florida voting is the importance of the use of high-quality ballots. According to Linda Cummings, director of voter services in Chester County, the specifications for the quality of the card stock and the cutting of the perforations in the ballot cards are critical to the low number of voting failures. On one occasion in which the specifications were relaxed and less expensive ballots were used, the number of problems was exponentially higher.

Statistics from Philadelphia County in the last three presidential elections tend to confirm that the Chester and Centre County under-vote percentages are not the result of deficiencies in the punch-card method as administered in those two counties. All voting in Philadelphia is on lever mechanical voting machines, except for a small percentage of absentee votes cast on traditional paper ballots. The percentage of persons who signed in to vote but did not cast a vote for president was six-tenths of 1 percent in 2000, eight-tenths of 1 percent in 1996 (Clinton versus Dole), and only three-tenths of 1 percent in the exceptionally high-turnout year of 1992. The variation over the three presidential elections suggests that there are voters who intentionally abstain, perhaps more than half of those who do not vote, in addition to some voters who mistakenly fail to vote.

There is a need for reform of parts of the Pennsylvania Election Code. However, the gaps probably would not have mattered in the year 2000 election, assuming that the U.S. Supreme Court declined to intervene, even if Pennsylvania’s vote for presidential electors had been as close as in Florida. The conduct of the election in Pennsylvania would not have provided any factual basis for additional litigation after the tedious recount process had been completed. The recounted result would likely have been accepted by the losing candidate and by the entire nation as a fair and just resolution of an extremely close election.