Company News · October 11, 2004
Election Nightmare Scenarios for 2004, Part Four
By edison
by Joe Lenski, Edison Executive VP
This week I finish my discussion of Election Nightmare Scenarios. In this fourth and final installment I write about the procedural issues dealing with the vote counting process in many states that may leave the final outcome of the Presidential election undetermined on Election Night
The first installment of this series examined the possibility of an Electoral Vote tie and the ramifications of the Proportional Electoral Vote Referendum in Colorado. The second installment examined the possibility that control of the Senate and the U.S. House may not be known until days or weeks after Election Day. The third installment dealt with the procedures in place to deal with the disruption of an election due to another terrorist attack or the death or disability of a presidential or vice-presidential candidate. This article examines how absentee voting, provisional ballots, late-arriving overseas and military ballots and just plain miscounting of votes could leave the winner of the presidential race undetermined for days or weeks.
Scenario #11 – Absentee voting
Situation: One candidate leads in the Election Day voting but the other candidate makes up the difference in absentee votes.
Likelihood: Probable. It happened in at least two states in 2000.
As part of the recount process in Florida in 2000, it became clear that Al Gore had won the Election Day vote, but George W. Bush had a slightly bigger margin among those who had cast absentee votes. In fact, Gore won the Election Day vote in Florida by more than 2 points (a margin of 131,373 votes) but Bush won the vote among the 12% who cast their vote in Florida by absentee by 20 points (a margin of 131,910 votes).
The situation was exactly the reverse in Iowa. George W. Bush actually won the Election Day vote in Iowa by just under 1 point (a margin of 7,253 votes), but Al Gore won the Absentee voters by 4 points (a margin of 11,397) which allowed him to win Iowa’s electoral votes by a total margin of just over 4,000 votes.
The two parties know these facts and have learned a lot of lessons from the 2000 Election about the value of absentee voting. Each Party and their unofficially affiliated 527s are spending considerable time and money to increase the number of their partisans who are casting absentee votes. Approximately 16% of all voters in 2000 cast a ballot by-mail or in-person before Election Day. In 2004 this percentage is likely to grow to more than 20% of the total vote. All voting in Oregon is now done by mail. In the state of Washington more than two-thirds of the vote will be cast by mail. There are another six states (Arizona, Colorado, Nevada, New Mexico, Tennessee and Texas) where at least 40% of the vote will be cast before Election Day.
So why is this a problem when it comes to learning the winner of the election on Election Day?
The biggest problem is that some states do not count many of the absentee votes on Election Day. In addition some states only dictate that an absentee ballot be postmarked by Election Day and does not have to be received by election officials until later. Thus, the votes are unable to be counted because they are literally still in the mail.
There are several states in which uncounted absentee votes could delay knowing the winner of the election.
In Oregon only 73% of the mail-in votes were counted on Election Night in 2000. At 6AM the next morning George W. Bush led the vote count by over 31,000. Days later Gore pulled ahead and won the state by just under 7,000 votes.
In Washington more than half of the absentee votes were not counted on Election Night in 2000. This did not affect the presidential race but it did keep the winner of the Senate race from being determined for more than three weeks.
There are seven states with critical presidential or senate contests that do not count a significant portion of their votes on Election Night. In 2000, Alaska (13% of the total vote uncounted), Arizona (14%), California (16%), Maine (13%), Michigan (8%), New Mexico (13%) and North Carolina (7%) had a considerable number of votes left uncounted on Election Night.
The other problem with Absentee voting is the potential for voting fraud. There are already news stories about potentially fraudulent voter registrations in states like Nevada and Ohio. In an election where the stakes are so high and the election is so close it is not inconceivable that someone could organize some hanky-panky to increase the number of absentee votes via fraud.
There is a history of this in several locales. The results of the 1997 election for mayor in Miami, Florida were thrown out after evidence of 5,000 fraudulent absentee ballots was discovered. In 1993 the results of a Pennsylvania State Senate election in Philadelphia were thrown out and campaign officials thrown in jail after evidence of absentee fraud was found in that race. It is not out of the realm of possibility that someone might try to do this, and also not out of the realm of possibility that they will eventually be caught which might cause the election to be thrown out. [See all the complications that would cause under Scenario #6.]
Scenario #12 – Provisional Voting
Situation: Under provisions of the Help America Vote Act (HAVA), voters who show up at the polling place and find that they are not on the voter registration list are given the opportunity to cast a “provisional” ballot which is not counted on Election Night but is counted later if the election officials can validate that the voter is indeed registered to vote. These uncounted provisional ballots could provide the margin of victory in a closely contested race
Likelihood: Probably pretty likely. Remember that Bush officially won Florida by 537 votes, and Gore won New Mexico officially by 365 votes. Three other states (Iowa, Oregon and Wisconsin) were also decided by less than half of one percent of the vote. It is not unlikely that the number of uncounted provisional ballots will exceed the unofficial vote margin on Election Night in several states.
The 2000 election was abnormally close on many levels but it was not that out of line with the number of close states in a typical presidential election. In 1992 and 1996, two states in each election were decided by 1% or less; in 1980 three states were decided by 1% or less; in 1976 there were four states; and in 1960 six states were decided by 1% of the vote or less. Thus, past history suggests that if the race is close nationally, there will be at least several individual states close enough for provisional ballots to make the difference.
So how are these provisional ballots validated? Well, that varies from state to state. In 26 states and the District of Columbia election officials have ruled that a provisional ballot must be cast in the correct precinct or it will not be counted. Other states have ruled that as long as the voter is indeed registered to vote the provisional ballot will be counted even if that voter went to the wrong polling place to cast the ballot.
Different Standards? That sounds familiar courtesy of the Supreme Court’s ruling in Bush v. Gore. You can bet that if provisional ballots could indeed make the difference in an election the lawyers will be arguing over which ballots should be counted and which ones should not.
How long could such a legal process take? The process of validating a provisional ballot usually takes seven to ten days. The legal challenges could take a lot longer.
We can look at the 2002 House Election in Colorado’s 7th Congressional District as an example of how this would play out in a close election. After the election on November 5, 2002 Republican Bob Beauprez led his opponent Democrat Mike Feeley by 386 out of more than 160,000 votes counted. However, there were 6,200 provisional ballots cast that were not yet counted. After about one week of arguing, election officials in the three counties that make up the CD decided that about 2,000 of the ballots were valid and should be counted. Once they were counted Beauprez’s margin dropped to 122 votes. On November 21st, Feeley demanded a recount which was not completed until December 10th – 35 days after the election. The result: Beauprez won by 121 votes. The arguments over which provisional ballots should be counted in a close race for president could last for weeks if not months.
Scenario #13 – Late arriving military and oversees ballots
Situation: Late arriving military and overseas ballots determine the outcome of an election.
Likelihood: Well, it happened in Florida in 2000. Florida counts overseas and military ballots that arrive within 10 days of Election Day. In 2000 Florida counted 2,490 overseas and military ballots. They provided George W. Bush with a margin of 739 votes which was more than the 537 votes by which he officially won the state.
Believe it or not, another key state in this year’s election has a similar provision. Ohio will count overseas and military ballots which arrive as late as November 12th this year as long as the ballot has a November 2nd postmark. There are 10 other states that have similar post-Election Day deadlines for overseas and military ballots including the battleground states of Arkansas, Iowa, Washington and West Virginia.
Scenario #14 – Miscounting of Votes
Situation: Through human or mechanical error, votes are miscounted on Election Night leading to the reporting of the wrong winner of a race.
Likelihood: Again this was part of the confusion in Florida in 2000. Significant miscounts of the vote during Election Night took place in Brevard, Duval and most significantly Volusia Counties.
After all that happened in 2000, one would think that vote miscounts would be less likely to happen subsequently. However in the Election of 2002 there were still miscounts of the votes.
The most critical miscount occurred in Baldwin County, Alabama for the Governor’s Election. Baldwin County initially reported 7,000 extra votes for incumbent Governor Don Siegelman. This temporarily put Siegelman ahead statewide by about 4,000 votes late in the evening and led the Associate Press to momentarily declare him the winner. Baldwin County corrected its tally later in the evening and Bob Riley ended up being the victor by 3,120 votes. This miscount led to several weeks of confusion until Siegelman finally conceded 13 days after the election.
The mistake in Alabama was quickly corrected. However, there are plenty of vote count mistakes that are never caught.
In researching the 2002 vote returns for Arkansas we discovered a similar mistake that has been “officially certified” by the State of Arkansas. Precinct 4D in Garland County reported that 951 votes had been cast for Senate but in the same precinct 5,142 votes had been cast for Governor. Reviewing the candidate returns for each race it becomes clear what happened.
Obviously there was a typo when the results were recorded for Governor Huckabee – an extra “3” was tacked on to Huckabee’s vote. Awarding Huckabee 465 votes in this precinct as opposed to 4,653 means that 954 votes would have actually been cast in this precinct for Governor, almost exactly the same number as the votes cast in this precinct for Senate.
This means that because of a simple typo Governor Huckabee received 4,188 extra votes in the “official certified” results. Huckabee’s margin of victory statewide was 48,843 votes so these 4,188 extra votes did not change the winner. However, if typographical errors this large can occur in a single precinct, it means that any race with a small margin of victory could be change by a simple mistake in recording the votes in a single precinct.
Conclusion
This series of articles examined 14 possible “nightmare” election scenarios that could keep us guessing about who has won this year’s election until days or weeks after the final ballots have been cast. These possibilities also demonstrate how legal challenges in the courts may influence the final outcome of the election.
I welcome your comments about these scenarios or about other possibilities that I have overlooked. You can contact me at jlenski@edisonresearch.com.
Get our latest insights delivered to your inbox.
by Joe Lenski, Edison Executive VP
This week I finish my discussion of Election Nightmare Scenarios. In this fourth and final installment I write about the procedural issues dealing with the vote counting process in many states that may leave the final outcome of the Presidential election undetermined on Election Night
The first installment of this series examined the possibility of an Electoral Vote tie and the ramifications of the Proportional Electoral Vote Referendum in Colorado. The second installment examined the possibility that control of the Senate and the U.S. House may not be known until days or weeks after Election Day. The third installment dealt with the procedures in place to deal with the disruption of an election due to another terrorist attack or the death or disability of a presidential or vice-presidential candidate. This article examines how absentee voting, provisional ballots, late-arriving overseas and military ballots and just plain miscounting of votes could leave the winner of the presidential race undetermined for days or weeks.
Scenario #11 – Absentee voting
Situation: One candidate leads in the Election Day voting but the other candidate makes up the difference in absentee votes.
Likelihood: Probable. It happened in at least two states in 2000.
As part of the recount process in Florida in 2000, it became clear that Al Gore had won the Election Day vote, but George W. Bush had a slightly bigger margin among those who had cast absentee votes. In fact, Gore won the Election Day vote in Florida by more than 2 points (a margin of 131,373 votes) but Bush won the vote among the 12% who cast their vote in Florida by absentee by 20 points (a margin of 131,910 votes).
The situation was exactly the reverse in Iowa. George W. Bush actually won the Election Day vote in Iowa by just under 1 point (a margin of 7,253 votes), but Al Gore won the Absentee voters by 4 points (a margin of 11,397) which allowed him to win Iowa’s electoral votes by a total margin of just over 4,000 votes.
The two parties know these facts and have learned a lot of lessons from the 2000 Election about the value of absentee voting. Each Party and their unofficially affiliated 527s are spending considerable time and money to increase the number of their partisans who are casting absentee votes. Approximately 16% of all voters in 2000 cast a ballot by-mail or in-person before Election Day. In 2004 this percentage is likely to grow to more than 20% of the total vote. All voting in Oregon is now done by mail. In the state of Washington more than two-thirds of the vote will be cast by mail. There are another six states (Arizona, Colorado, Nevada, New Mexico, Tennessee and Texas) where at least 40% of the vote will be cast before Election Day.
So why is this a problem when it comes to learning the winner of the election on Election Day?
The biggest problem is that some states do not count many of the absentee votes on Election Day. In addition some states only dictate that an absentee ballot be postmarked by Election Day and does not have to be received by election officials until later. Thus, the votes are unable to be counted because they are literally still in the mail.
There are several states in which uncounted absentee votes could delay knowing the winner of the election.
In Oregon only 73% of the mail-in votes were counted on Election Night in 2000. At 6AM the next morning George W. Bush led the vote count by over 31,000. Days later Gore pulled ahead and won the state by just under 7,000 votes.
In Washington more than half of the absentee votes were not counted on Election Night in 2000. This did not affect the presidential race but it did keep the winner of the Senate race from being determined for more than three weeks.
There are seven states with critical presidential or senate contests that do not count a significant portion of their votes on Election Night. In 2000, Alaska (13% of the total vote uncounted), Arizona (14%), California (16%), Maine (13%), Michigan (8%), New Mexico (13%) and North Carolina (7%) had a considerable number of votes left uncounted on Election Night.
The other problem with Absentee voting is the potential for voting fraud. There are already news stories about potentially fraudulent voter registrations in states like Nevada and Ohio. In an election where the stakes are so high and the election is so close it is not inconceivable that someone could organize some hanky-panky to increase the number of absentee votes via fraud.
There is a history of this in several locales. The results of the 1997 election for mayor in Miami, Florida were thrown out after evidence of 5,000 fraudulent absentee ballots was discovered. In 1993 the results of a Pennsylvania State Senate election in Philadelphia were thrown out and campaign officials thrown in jail after evidence of absentee fraud was found in that race. It is not out of the realm of possibility that someone might try to do this, and also not out of the realm of possibility that they will eventually be caught which might cause the election to be thrown out. [See all the complications that would cause under Scenario #6.]
Scenario #12 – Provisional Voting
Situation: Under provisions of the Help America Vote Act (HAVA), voters who show up at the polling place and find that they are not on the voter registration list are given the opportunity to cast a “provisional” ballot which is not counted on Election Night but is counted later if the election officials can validate that the voter is indeed registered to vote. These uncounted provisional ballots could provide the margin of victory in a closely contested race
Likelihood: Probably pretty likely. Remember that Bush officially won Florida by 537 votes, and Gore won New Mexico officially by 365 votes. Three other states (Iowa, Oregon and Wisconsin) were also decided by less than half of one percent of the vote. It is not unlikely that the number of uncounted provisional ballots will exceed the unofficial vote margin on Election Night in several states.
The 2000 election was abnormally close on many levels but it was not that out of line with the number of close states in a typical presidential election. In 1992 and 1996, two states in each election were decided by 1% or less; in 1980 three states were decided by 1% or less; in 1976 there were four states; and in 1960 six states were decided by 1% of the vote or less. Thus, past history suggests that if the race is close nationally, there will be at least several individual states close enough for provisional ballots to make the difference.
So how are these provisional ballots validated? Well, that varies from state to state. In 26 states and the District of Columbia election officials have ruled that a provisional ballot must be cast in the correct precinct or it will not be counted. Other states have ruled that as long as the voter is indeed registered to vote the provisional ballot will be counted even if that voter went to the wrong polling place to cast the ballot.
Different Standards? That sounds familiar courtesy of the Supreme Court’s ruling in Bush v. Gore. You can bet that if provisional ballots could indeed make the difference in an election the lawyers will be arguing over which ballots should be counted and which ones should not.
How long could such a legal process take? The process of validating a provisional ballot usually takes seven to ten days. The legal challenges could take a lot longer.
We can look at the 2002 House Election in Colorado’s 7th Congressional District as an example of how this would play out in a close election. After the election on November 5, 2002 Republican Bob Beauprez led his opponent Democrat Mike Feeley by 386 out of more than 160,000 votes counted. However, there were 6,200 provisional ballots cast that were not yet counted. After about one week of arguing, election officials in the three counties that make up the CD decided that about 2,000 of the ballots were valid and should be counted. Once they were counted Beauprez’s margin dropped to 122 votes. On November 21st, Feeley demanded a recount which was not completed until December 10th – 35 days after the election. The result: Beauprez won by 121 votes. The arguments over which provisional ballots should be counted in a close race for president could last for weeks if not months.
Scenario #13 – Late arriving military and oversees ballots
Situation: Late arriving military and overseas ballots determine the outcome of an election.
Likelihood: Well, it happened in Florida in 2000. Florida counts overseas and military ballots that arrive within 10 days of Election Day. In 2000 Florida counted 2,490 overseas and military ballots. They provided George W. Bush with a margin of 739 votes which was more than the 537 votes by which he officially won the state.
Believe it or not, another key state in this year’s election has a similar provision. Ohio will count overseas and military ballots which arrive as late as November 12th this year as long as the ballot has a November 2nd postmark. There are 10 other states that have similar post-Election Day deadlines for overseas and military ballots including the battleground states of Arkansas, Iowa, Washington and West Virginia.
Scenario #14 – Miscounting of Votes
Situation: Through human or mechanical error, votes are miscounted on Election Night leading to the reporting of the wrong winner of a race.
Likelihood: Again this was part of the confusion in Florida in 2000. Significant miscounts of the vote during Election Night took place in Brevard, Duval and most significantly Volusia Counties.
After all that happened in 2000, one would think that vote miscounts would be less likely to happen subsequently. However in the Election of 2002 there were still miscounts of the votes.
The most critical miscount occurred in Baldwin County, Alabama for the Governor’s Election. Baldwin County initially reported 7,000 extra votes for incumbent Governor Don Siegelman. This temporarily put Siegelman ahead statewide by about 4,000 votes late in the evening and led the Associate Press to momentarily declare him the winner. Baldwin County corrected its tally later in the evening and Bob Riley ended up being the victor by 3,120 votes. This miscount led to several weeks of confusion until Siegelman finally conceded 13 days after the election.
The mistake in Alabama was quickly corrected. However, there are plenty of vote count mistakes that are never caught.
In researching the 2002 vote returns for Arkansas we discovered a similar mistake that has been “officially certified” by the State of Arkansas. Precinct 4D in Garland County reported that 951 votes had been cast for Senate but in the same precinct 5,142 votes had been cast for Governor. Reviewing the candidate returns for each race it becomes clear what happened.
Obviously there was a typo when the results were recorded for Governor Huckabee – an extra “3” was tacked on to Huckabee’s vote. Awarding Huckabee 465 votes in this precinct as opposed to 4,653 means that 954 votes would have actually been cast in this precinct for Governor, almost exactly the same number as the votes cast in this precinct for Senate.
This means that because of a simple typo Governor Huckabee received 4,188 extra votes in the “official certified” results. Huckabee’s margin of victory statewide was 48,843 votes so these 4,188 extra votes did not change the winner. However, if typographical errors this large can occur in a single precinct, it means that any race with a small margin of victory could be change by a simple mistake in recording the votes in a single precinct.
Conclusion
This series of articles examined 14 possible “nightmare” election scenarios that could keep us guessing about who has won this year’s election until days or weeks after the final ballots have been cast. These possibilities also demonstrate how legal challenges in the courts may influence the final outcome of the election.
I welcome your comments about these scenarios or about other possibilities that I have overlooked. You can contact me at jlenski@edisonresearch.com.