Monday's adjustment of the poll formula also impacts the numbers in the 4th Congressional District contest between Democratic Rep. Jim Matheson and GOP challenger Mia Love, but not to the extent of the county mayor's election.
The crux of the problem was Mason-Dixon's figures included too many Republicans in both the county and 4th Congressional District and undersampled Democrats in both races.
The revised numbers show Love leading Matheson 50-43 a narrower margin than the original 52-40 advantage that was reported. The new spread is within the poll's 4 percent margin of error, but Coker said it is "not big enough to change my interpretation that Love is the favorite.
Both campaigns say it doesn't change the race.
"At the end of the day, polls have been all over the map in this race. I have always said it would be a close contest. If you are a Matheson supporter, go and cast your vote," said Matheson.
And the Love campaign said it comes down to their message.
"When we were ahead in September and we were ahead last week, it never changed what our plan was," said Ivan DuBois, senior adviser to the Love campaign.
In the Salt Lake County mayor's race, neither candidate felt the earlier numbers would influence the election's outcome.
"I don't think the polls drive decisions. Our sense and our numbers have always been that we're up three to four [percentage] points," said Crockett. But his campaign spokeswoman, Evelyn Call, said later Monday she thought it was "fishy" for a pollster to change results, especially so close to Election Day.
While McAdams said the new numbers are "much more in line with what our polling has shown for a month that this is a very close race," he added that last week's deficit in The Tribune poll was so inconsistent with other polls that he did not alter his strategy for chasing votes on the final weekend before Tuesday's general election.
But Kelly McBride, a senior faculty member specializing in ethics at the Poynter Institute, which promotes excellence and integrity in journalism, said The Tribune let its readers down by not asking probing questions about the poll numbers and the surprisingly big lead they seemed to give Crockett before they were published. And that could influence election results, she added.
"It's a real shame that happened this close to the election. The damage may not be insignificant," McBride said, noting that when one candidate is reported to be far ahead of another, it can keep people from voting.
"Polls are controversial because they do impact how people vote," she said. "We have to be super-sensitive to make sure the information is accurate and scientific."
And Quin Monson, director of the Center for the Study of Elections and Democracy at Brigham Young University who specializes in political polling, said the changes in the numbers are surprising and the lack of transparency in the poll remains troubling.
"A nine-point change [in the mayor's race] is just shocking, actually," he said. "The real problem here is a lot of people saw these results and, consciously or unconsciously, they affect how voters who haven't cast their ballots think about the election, and it will have some effect on people, whether they realize it or not. That's the sad part.
"As a news organization, having released numbers that could now affect the final outcome, that becomes a real ethical problem," he said.
Tribune editor Nancy Conway acknowledged the problem.
"We are as concerned about this as anyone," she said Monday. "As soon as we understood there was a problem we worked to correct it.
"We had no reason to doubt the poll until we saw others conducted over the same period and could see differences in the numbers. That raised questions," Conway said. "We contacted our pollster who did additional research on Salt Lake County demographics and found there was indeed a flaw.
"We knew right then that we needed to correct our mistake and that's what we are doing," Conway said.
Pollster Coker said his firm interviewed 625 likely voters statewide to get statistically valid responses to several poll questions; 240 of those people were from Salt Lake County. Then, to get valid numbers for Salt Lake County issues, another 385 likely county voters were surveyed. Similar approaches were taken for polls involving the 2nd and 4th congressional districts.
To help ensure those respondents represent a true cross section of their communities, he added, the results are "weighted" in the polling formula to reflect the area's political breakdown.
Twenty percent of those sampled said they were Democrats, 48 percent said Republican and 32 percent said independent. That was close to accepted breakdowns of party affiliation statewide, said Coker, "but intuitively we know Salt Lake County is more Democratic than the rest of the state." It would have been more accurate to weight Democratic affiliation in Salt Lake County at closer to 25 percent, with Republicans at 43 percent, he added.
But "we finished the poll on Wednesday night. We're pulling [the results] together at 3 a.m. I'm cranking to get this out to [The Tribune] Thursday morning," Coker said, and he did not notice that the party-affiliation figures for weighting were off. "It slipped between the cracks when we were pulling that information together and trying to put it out quick. … That's something I normally would have caught."
He checked the numbers again after being asked over the weekend to do so by The Tribune.
On the other hand, Coker said, it's possible the higher Republican Party preference voiced by poll participants could be accurate, reflecting a greater-than-usual eagerness to claim GOP allegiance because that party's ticket is headed this year by presidential nominee Mitt Romney, a Utah favorite due to his LDS faith and leadership role in running Salt Lake City's 2002 Winter Olympics.
"If there is a big extra Republican turnout for Mitt Romney, that works for Crockett's advantage," he said. "We could wake up Wednesday morning and Crockett will have won by 10 [percent]."
Crockett said he expects voter turnout Tuesday to be high. "People really want to vote this year," he said. "They don't want to miss their opportunity to vote for Romney."
McAdams said confidence in his campaign's internal polling and its prognosis that he has a slight edge have left him feeling that "voters will make the right decision [Tuesday]. We have not let up and will continue working to take this across the finish line."
Maryann Martindale, executive director of the liberal Alliance for a Better Utah, chided The Tribune for its "unfortunate decision to announce methodologically flawed poll results" but applauded the decision to publish the revised numbers.
State Democratic Party Chairman Jim Dabakis was more biting in his assessment, saying with apparent sarcasm that "we suspected that the poll was wrong when it showed Al Gore up by 5." Democrat Gore lost the 2000 presidential election to Republican George W. Bush.
Tribune reporters Robert Gehrke and Christopher Smart contributed to this article.