States hope to improve on missing vote tally

By THOMAS HARGROVE and MICHAEL COLLINS
Scripps Howard News Service
October 19, 2004

- America's top election officials, reacting to growing public anxiety over the voting errors made in Florida four years ago, are promising they will scrutinize next month's election results like never before.

At least 1.6 million ballots did not record a preference for president in 2000, or nearly 2 percent of the vote. Such an electoral fog could easily obscure a clear winner in the current, tight race for the White House.

Ballots are much more likely not to count for major offices like president or governor when state officials fail to check for missing votes, according to a Scripps Howard News Service study based on interviews with chief election officers or their surrogates in all 50 states and the District of Columbia.

Most - but not all - of these election officers vow they will be looking for missing votes following disclosures that inaccurate balloting machines and inadequate counting methods have marred several recent elections.

"We suddenly are responsible, so we'd better take on more responsibility," said Amy Naccarato, Utah's director of elections. "We've started to get more involved, much to the dismay of my county clerks. We know we have to do a better job. We are tracking all kinds of statistics that we've never tracked before."

They have at least two reasons to be concerned:

- States that didn't look for lost votes four years ago had a 39 percent higher rate of so-called "undervoting" in which ballots do not record votes in major races, the Scripps Howard study found.

- Election officials concede that when they aggressively look for missing votes, they improve the accuracy of their elections by uncovering tabulation errors or determining when bad ballot designs or old voting machines are confusing people into making disqualifying errors at the polls.

Asking if votes might be missing is an important step toward repairing the credibility of the electoral process and reassuring Americans that every vote counts, election experts say.

"Just asking that question is revolutionary," said DeForest Soaries, chairman of the new U.S. Election Assistance Commission created by Congress to correct voting errors. "And acquiring data (about the undervote) is very instructive. We plan to check this for the entire country this November."

Thirty-one states and the District of Columbia compared the number of ballots cast against the number of votes counted in the 2000 presidential election, according to the Scripps Howard survey of all top state election officials. However, several said they looked for missing votes for the first time four years ago to determine whether they suffered Florida-like problems.

This year, 42 states and the District of Columbia have promised to conduct a study of missing votes next month.

"Yeah, we'll be checking," said Idaho Elections Director Tim Hurst. "If there is something that looks fishy, we'll go back and ask the counties to look at the numbers again. We've spent a lot of time with our clerks, making sure they understand they are responsible for these tabulations."

A few states, including those suffering the nation's worst problems in 2000, still say election supervision is essentially a local matter. They argue that a large undervote does not necessarily prove something's amiss.

"An undervote is the choice of the voter. And all of those results are handled at the local level," said Florida Secretary of State Glenda Hood. "What we get at the state level is the number of votes cast for particular candidates or particular issues."

Florida drew international attention four years ago when 178,145 of its ballots failed to register for president. George W. Bush officially carried the state by 537 votes.

Hood said her office will not be gathering the necessary information from county governments to calculate the undervote in next month's election. "We don't receive any of that information at the state level. That's all local," she said.

Illinois had a worse undervote than Florida in 2000 because of its 190,084-vote discrepancy between the number of ballots cast and the number of votes counted for president, according to a Scripps Howard study of official election returns.

Six Illinois counties including Chicago's Cook County reported that more than 5 percent of their ballots cast did not record a vote for president. Election experts suggest that the undervote should be investigated whenever it exceeds 2 percent in major elections.

"I don't have an explanation for that," Illinois State Board of Elections Executive Director Daniel White said of the large undervote. "Unless there are allegations of fraud, the counties certify the results to us and we certified the results statewide. I don't think we have legal authority to question the results."

White said his office has never studied the undervote in Illinois, but he recently ordered county officials to provide the necessary information to allow such a study. "A lot of people are going to want this information this time," White said.

The Scripps Howard survey found that 98.2 percent of all ballots cast in 2000 recorded a vote for president in states where officials examined the undervote. In contrast, only 97.5 percent of the ballots recorded presidential votes in states where officials have never checked for missing votes.

Put another way, the rate at which ballots failed to count in the 2000 presidential election rose from 1.8 percent in states that checked for missing votes to 2.5 percent in states that don't check, an increase of 39 percent.

These findings are not coincidence, many state officials said.

"I'm surprised to hear some states aren't checking this," said Mary Kiffmeyer, secretary of state in Minnesota where 99.3 percent of ballots recorded a presidential vote. "How could you not reconcile the number of votes to the number of ballots? How can you not do that?"

In Nevada, "we've put a lot of reforms in place," Secretary of State Dean Hellar said. "We will have the lowest undervote in the country, I assure you."

The nation's lowest statewide undervote four years ago occurred in Louisiana, where presidential votes were counted in 99.4 percent of the ballots cast.

"That is not an accident," said Louisiana Secretary of State Fox McKeithen. "We routinely generate statistics to make sure there isn't something wrong with some machine somewhere."

Sometimes problems become obvious after a study of the undervote, according to Washington state's Secretary of State Sam Reed. "It's the first thing we look for," he said. "If counties have a high number, we meet with them to find out what is going on."

Washington has a lower-than-average statewide rate of undervoting, but Franklin and Yakima counties reported only 95 percent of their ballots registered a presidential vote four years ago. "They use punch cards, which we think was a factor. They also have a large Spanish speaking population," Reed said, adding that Latinos may not have been able to read balloting instruction that four years ago were only printed in English.

Even more curious is Douglas County, Wash., which reported 12,855 presidential votes were counted out of only 12,225 ballots cast four years ago. (More than two-dozen counties nationwide reported more votes were counted for president in 2000 than ballots cast, according to the Scripps Howard study.)

"We did a recount and discovered that a precinct in Douglas County was double-counted," Reed said. "There was no excuse for that. We sent a review team over there to make sure that all personnel are well trained."

Scripps Howard mailed a county-by-county analysis of the undervote to top elected officials in the 40 states that provided complete election data four years ago. (Ten states did not count the number of ballots cast then, making an undervote study in those states impossible.)

North Carolina officials last month made inquiries after asked by Scripps Howard why Union and Clay counties reported more votes than ballots four years ago.

"The turnout figure for Union County is incorrect," said Don Wright, general counsel to the North Carolina State Board of Elections. "It is listed as 47,000 on our report, but a call to Union confirmed their turnout in 2000 to be 48,638."

Wright said Clay County officials were not immediately able to report the number of ballots cast four years ago "but confirmed the turnout figure on our report is not right."

Several officials conceded they had never examined the undervote trends.

"Our falloff was 1.37 percent. I made that calculation after your letter came to me," said North Dakota Secretary of State Alvin Jaeger. "Your listing (of the undervote) by county, well, I didn't do that. But I can see that it varies all over the place."

Jean Jensen, the new secretary of the Virginia Board of Elections, said she was surprised to learn that the undervote approached 10 percent in some of her counties.

"This is fascinating," Jensen said. "We really can't come up with any explanation for why it (the undervote) was so high, other than that this was the first year that new equipment was being used in some places."

Delaware Election Commissioner Frank Calio began an investigation after learning that 3.6 percent of the ballots cast in urban New Castle County did not register a presidential vote four years ago. "I was surprised that so many people chose not to vote," Calio said.

Twelve states did not report how many ballots were cast when they certified their presidential votes four years ago, but most of these have promised to collect the information and calculate the undervote in their states next month.

"Everybody in the world is going to want to know that this year," said Tennessee Elections Coordinator Brook Thompson. "We will get a reading from our counties as to exactly how many ballots were issued."

After the voting errors came to light in Florida four years ago, South Dakota took a closer look at its undervote and even asked some counties with higher undervotes to go back and recheck their tallies, said Secretary of State Chris Nelson.

"It showed exactly what we expected," Nelson said. "The counting machines were counting correctly, and they simply had races further down the ballot that people cared more about than the presidential race."

Still, South Dakota will study its undervote in next month's election, Nelson said.

Massachusetts did not check its undervote four years ago and would do so this year only if it becomes an issue or if the election results in the state are close, said Secretary of the Commonwealth William Galvin. "We would absolutely look at it if there was a recount," he said.

Colorado did not study its undervote in the 2000 election, but plans to do so this year, said Secretary of State Donetta Davidson. However, she said, in Colorado a ballot is counted as an undervote only when it is totally blank.

Oregon did not study the undervote in 2002, although it made an informal check in 2000 because of the embarrassments in Florida. But Oregon will be checking the undervote in November.

"In fact, it is now a directive from this office that it be done in every election because of the additional scrutiny and concerns from the 2000 election," said Ann Martins, spokeswoman for Oregon Secretary of State Bill Bradbury. "We did this because, well, people are just worried."

Eight states have said they do not have plans to study the undervote this year: Arizona, Florida, Iowa, Massachusetts, Montana, New Mexico, North Dakota and Pennsylvania.

"We do not routinely make a study of undervotes, particularly because our undervotes are low," said Mike Sullivan, spokesman for Montana Secretary of State Robert Brown. (About 98.3 percent of Montana's ballots recorded a presidential vote, slightly better than average.)

"We don't have any plans to do so in 2004 unless there is something really inconsistent or egregious," Sullivan said.

Arizona State Election Director Joe Kanefield said the state isn't required by law to examine its undervote. "Of course, we look at the numbers and we notice things," he said. "The numbers have never given us any reason to believe there is anything else other than a voter deciding not to vote in a particular election. We are confident in our equipment and believe that (undervote) is merely a reflection of the voter's independent choice."

New Mexico, a battleground state in this year's presidential race, has no plans to check for missing votes. Only 97.3 percent of ballots cast recorded a presidential vote four years ago.

The state did recheck the votes in conservative Roosevelt County after the county's undervote exceeded 10 percent and Republicans suggested that Bush should have gotten more votes there. Officials determined that three absentee ballot counters had not been programmed properly. Bush picked up another 100 votes after the counters were reset.

Despite that problem, Secretary of State Rebecca Vigil-Giron said there are no plans to check the undervote this year unless one of the parties again questions the results.

"That is not part of the law," she said. "I'm not in a position to question the intent of an individual voter. My only requirement is what constitutes a vote, and the definition of that is built into our rules and the way those paper ballots are counted."