IN Colorado "there must have been a colossal breakdown" in South Dakota there was "a colossal screwup"; in Iowa the results were simply "wrong." In state after state this year, public opinion polls - the ball and bat of American politics - miscued, sometimes with disastrous consequences for candidates.
The quotes above came from interviews with pollsters, candidates and political consultants. Though their views predictably varied, an informal series of interviews suggests widespread belief that this was a bad year for the polls.
After last week's election results, said one liberal Democrat who has worked in dozens of campaigns. "I would no longer put the dependency on polls that I used to."
"Were we polling the wrong people, or asking the wrong questions?" asked Joe Rothstein, a political consultant who had unusually bad luck with both poll and election results this year.
"What we all need to do is reexamine the whole art of public opinion sampling," said an aide to Sen. Thomas MeIntyre (D-N.H.), who lost last Tuesday in a stunning upset.
McIntyre's defeat was perhaps the single most startling example of a polling miscue this year. In mid-October, a poll conducted for the 16-year Senate veteran found that he led his archconservative challenger. Gordon Humphrey, by 59.5 percent to 30 percent. The poll found deep, firm support for McIntyre, and no significant trend toward Humphrey. On election day, Humphrey won.
That poll was conducted under the direction of Cambridge Survey Research Inc., the firm run by interview yesterday. Caddell said this poll was conducted by volunteers, not "professionals," but officials of the McIntyre campaign noted that Cambridge Survey Research had drawn up the sample and trained those volunteers.
According to one MyIntyre associate, Caddell's partner, John Coleman told McIntyre he could have high confidence in the poll's results.
Peter D. Hart, another leading pollster and consultant, made no excuses for his firm's failure to perceive and coaster the late trend in Iowa that led to the surprising defeat there of Democratic Sen. Dick Clark.
"I don't feel that we served him well," Hart said of Clark, whom his firm advised on campaign strategy and advertising an apparently unprecedented post-mortem survey of the Iowa voters it questioned last summer and fail to try to find out what went wrong.
Peter D. Hart, another leading pollster and consultant, made no excuses for his firm's failure to perceive and counter the late trend in Iowa that led to the surprising defeat there of Democrate Sen. Dick Clark.
"I don't feel that we served him well," Hart said of Clark, whom his firm advised on campaign strategy and advertising. The Hart firm is now undertaking an apparently unprecedented post-mortem survey of the lows voters it questioned last summer and fail to try to find out what went wrong.
Hart's last full poll Iowa voters, conducted Oct. 4, found Clark leading challenger Roger Jepsen, a conservative Republican, 57 to 27 percent. Hart declined to provide those precise numbers - they came from another source - but acknowledged that "we did not have it tight, and we did not have Jepsen moving up" in October.
Hart noted - and many other professionals interviewed for this article agreed - that polls like his may have been right when they were taken, but wrong on election day. It is already a piece of conventional wisdom that the 1978 electorate was extraordinarly "volatile."
Douglas Bailey, a moderate Reubulican campaign consultant who had a good year for his clients, said he thought it was possible, even probable, that many polls that performed erratically during the campaign were correct at the time they were taken.
Bailey advised Sen. Charles H. Percy (H-III.) on his reelction campaign, and watched Percy fall behind challenger Alex Seith by nearly 20 points in the Chicago Sun-Times straw poll in late October, only to win last week by 5 1/2 to 4 percent.
In fact, the final Sun-Times poll and other Illinois polls, including Percy's own, showed the race much closer than the final outcome, though the last minute movement to Percy was widely perceived.
"The Sun-Times poll saved Percy," Bailey said in an interview, by alerting both Republican workers and the voting public to Seith's workers and the voting public to Seith's unexpected strength. Bailey said he thought published polls cold have a powerful impact on election results by mobilizing underogs' supporters or discouraging heavy favorites' backers from voting.
The Des Moines Register and Tribune's final Iowa poll gave Clark a nine-point lead in his race.
Most of Bailey's campaigns were associated with Robert Teeter, a Detroit pollster who worked for Gerald R. Ford in 1976, and who had more accurate results this year than some of his competitors.
Caddell's firm had one of the most erratic records. In Colorado, for instance, he called the Senate race a tossup a week before election day, but it turned out to be a landslide victory for Rep. William Armstrong (R) over incumbent Sen. Floyd Haskell.
Rothstein, the political consultant who aided the Haskell campaign, said he was convinced that voters' attitudes "were being affected by things we didn't understand" this year.
Teeter said he found unusual indecision this year, reflected both in shifts in public views toward candidates during the campaign and an unusually large undecided bloc right up to election day. There was also indecision about whether to vote, Teeter said.
He and other pollsters agreed that low turnouts make their predictions riskier. "As turnout declines it becomes tougher and tougher to measure who in your sample will really vote," Teeter said.
Teeter, Caddell and others noted the success of negative campaigns this year, particularly in Senate races. Caddell said that every Senate candidate who dramatically changed his standing in the polls did so with a negative campaign directed against his opponent.
Noting the large turnover in Senate seats in the last two elections, Caddell hypothesized that voters may increasingly went their frustrations on senators rather than House members or local officials.
Hart, whose polls last summer showed the potential for a negative campaign against Percy in Illinois, noted that polls are often too superficial to be significant. As an example, he cited his won polling in the Minnesota Senate primary three months before the voting.
In a head-to-head contest between Rep. Don Fraser and Robert Snort, Hart said. Fraser then led 67 to 16 percent. But if the poll dropped the candidates' names and described them both in terms of their positions and backgrounds, the candidate with Fraser's atributes got 49 percent, the candidates with Short's 42 percent. In the end, Short narrowly won, a result Hart said was presaged by that "hypothetical pairing" three months earlier.
Polls can also be sloppily conducted. In South Dakota's first congressional district, for example, overlapping polls by Caddell's firm and Hoggard-Smith and Associates of Washington produced startling discrepancies. Caddell found that 48 percent of the district's voters thought of themselves as "conservative" while Hoggard-Smith said that number was 34 percent.
Caddell said the district was 34 percent old people, Hoggard-Smith said it was 17 percent elderly. (The census says the figure is 20 percent.)
No one interviewed for this article suggested that polls would be abandoned because of this year's results. "This process depends on understanding public attitudes," consultant Rothstein said, "so polls will have to be improved."