Organized labor is taking a hard look at its political influence and voter turnout operations after the two union-backed candidates that were to dominate the Iowa caucuses sank instead.

Unions were to be the powerhouse in Iowa, with white-collars for Howard Dean (searchand blue-collars for Dick Gephardt (search). But in a stunning upset, they emerged battered in third- and fourth-place behind Sens. John Kerry (searchand John Edwards (search).

Leaders of industrial unions that formed a coalition supporting Gephardt, called Americans for Economic Justice, were to confer Wednesday to assess what happened in Iowa and map out a political future.

But already, union leaders say the lesson from Iowa is that organized labor remains split over which Democrat is best suited to challenge President Bush in November. Of the 64 unions in the AFL-CIO, less than half were committed to a candidate.

"I don't anticipate us arriving at another candidate," said Donald Kaniewski, political director of the Laborers' International Union of North America, a member of the coalition that backed Gephardt. "We've got plenty of work to do on jobs, health care and trade, and that work will continue."

Unions are looking ahead to contests in labor-dense states such as Missouri on Feb. 3, where more than 13 percent of the work force belongs to a union, and Michigan on Feb. 7, with more than 21 percent of its work force unionized. In Iowa, about 11 percent of workers are union members.

The Service Employees International Union, the largest in New Hampshire with 10,000 members, hopes it can give Dean a victory in Tuesday's primary after his disappointing loss in Iowa, where the union has few members. 

"We've been talking to members, and a good chunk are undecided," said spokeswoman Sara Howard. But by an overwhelming margin, she said SEIU members in New Hampshire "remain steadfast in their commitment to Dean."

As labor's candidate, backed by almost two dozen unions, Gephardt went into Iowa with high expectations — probably too high, union supporters now say. He withdrew from the race Tuesday as a result of his Iowa finish.

Conventional wisdom was that the candidate with the best organizational effort to turn out caucus participants would win.

Gephardt had the industrial unions, including the Teamsters, Machinists, Steelworkers and Laborers. Dean was backed by two of the country's largest unions — SEIU and the American Federation of State, County and Municipal Employees.

But union members and their families were outnumbered by new caucus-goers turned off by the negative campaigning between the two, union leaders say. And Gephardt failed to expand his base of support beyond labor.

Fewer than one-fourth of participants in the caucuses were from union households, according to entrance polls conducted for The Associated Press and the television networks by Edison Media Research and Mitofsky International. In 2000, one-third of Iowa voters were from union households.

"We're not going to make any moves right now," said Rob Black, spokesman for the Teamsters' union, which had endorsed Gephardt. "We're going to reassess our options and start the process over again." 

Some union members also were turned off by the negative campaigns, and fled for Kerry or Edwards.

"We found members that were highly supportive of Dick Gephardt and would have walked across the Mississippi River to help him if they could," Sloan said. "But we also had members who were turned off by the process, and there were members who supported other candidates. That's their right."

Gephardt's message that unfair trade is sending jobs overseas may not have been the right focus for the last few weeks of the campaign, Sloan said. "It was not the cutting issue in Iowa," which has an unemployment rate of just 4.2 percent, according to the latest Labor Department figures.

While Gephardt did better among union households than among most other groups, Kerry had a slight edge and Edwards and Dean did as well as Gephardt among the labor vote, according to the interviews of 1,665 people entering 50 randomly selected Democratic caucus sites. Results were subject to sampling error of plus or minus 4 percentage points.