Most undervotes come from counties that Obama won

By Brian Bakst

Associated Press

ST. PAUL — An Associated Press analysis of the nearly 25,000-vote difference in presidential and Senate race tallies shows that most ballots lacking a recorded Minnesota Senate vote were cast in counties won by Democrat Barack Obama.

The finding could have implications for Republican Sen. Norm Coleman and Democrat Al Franken, who are headed for a recount separated by the thinnest of margins — a couple hundred votes, or about 0.01 percent.

Though some voters may have intentionally bypassed the race, others may have mismarked their ballot or optical scanning machines may have misread them. A recount due to begin Nov. 19 will use manual inspection to detect such ballots.

Three counties — Hennepin, Ramsey and St. Louis — account for 10,540 votes in the dropoff between the two races. Each saw Obama win with 63 percent or more.

Larry Jacobs, a University of Minnesota political science professor, said the dropoff analysis creates a "zone of uncertainty" that could become a focal point for the campaigns and election officials.

"These numbers present a roadmap for a Franken challenge," he said. "It suggests there are about 10,000 votes in the largest Democratic counties that are potentially going to tilt in Franken’s direction."

The Minnesota ballots that showed a presidential vote, but no Senate vote, are called the "undervote."

Statewide, more than 18,000 of those ballots came from counties won by Obama with more than half the vote. About 6,100 were in counties won by McCain with at least 50 percent.

In 13 counties, the two ran about even; in all, those counties combined for 707 ballots without a Senate preference.

Some areas of the state would seem to favor Coleman in a recount based on the dropoff, but most of those were smaller counties where the undervote was in the dozens. The largest of those pro-McCain counties were Anoka, where 1,189 ballots didn’t choose a Senate candidate, and Stearns, where 681 did not.

Only one county — Jackson — reported one Senate vote for every presidential vote.

There’s one more critical statistic: About 8,900 people weren’t recorded as voting for president, according to county-by-county turnout estimates kept by the Secretary of State. That nearly 9,000 people would skip the closely watched race is questionable, raising the possibility that as many as 33,700 ballots might be subject to change in a hand recount.

Minnesota ballots are fed into optical scanners, which depend on voters filling in ovals to make their choice.

Kim Brace, president of the consulting firm Election Data Services Inc., said there’s no reason a ballot without a vote for a particular race would be rejected.

"Usually they’re set to kick back to the voter if there is an overvote," said Brace, who has been an expert witness in court cases stemming from disputed elections. "But in most instances they’re not set to kick back to the voter if there is an undervote. After all, the public has a right to not vote for somebody for a particular office."

What recount teams will be looking for is whether stray or light marks on ballots signaled a voters’ preference.

Secretary of State Mark Ritchie, a Democrat, said any process involving humans and machines is subject to error.

"Humans are marking ballots and some humans may not have pressed hard enough in the oval so the machine may not have caught their ballot’s intention," Ritchie said. "The recount is designed to make sure every eligible ballot is included."

Michael Shamos, a Carnegie Mellon University computer science professor who studies election systems, said the dropoff didn’t strike him as suspiciously large.

"All that means to me is they consider the presidential race more important than the Senate race," he said. "They either didn’t make up their mind. They didn’t care. They didn’t make a mark in that race."

To be sure, similar dropoffs have happened in the past.

In 2000, the last year with both races on Minnesota’s ballot, about 19,000 presidential voters were not recorded as casting a Senate vote. Fewer people voted that year, making the dropout rate similar to this year’s.

Since Democratic candidate Mark Dayton won clearly that year, there was no review of ballots to see whether voters intentionally skipped the race.