WITH A LITTLE push from the administration, the public works bandwagon has started rolling again in Congress. What started as a well- meaning effort to create jobs for the unemployed may well turn into a hefty addition to the congressional pork barrel.

Congressmen like big public works projects--public buildings, dams, military installations and so on. These projects attract the support of local contractors and construction unions. They are also tangible evidence that a congressman is helping his district economically. To no one's surprise, despite all the cutting in social programs, public works programs not only survived the onslaught, but actually were expanded by $5 billion by the recent gas tax legislation.

But popular or not, public works construction is a very poor way to combat unemployment. Construction projects are slow to start. Because of high costs for materials and equipment and federal wage-setting rules, they create few jobs for each dollar spent. If administration policy-makers--who are now said to be considering a step-up in public works--doubt this, perhaps they will want to consult pages 39-41 of the president's new Economic Report, in which the case against public works is clearly stated.

A much better approach was recently proposed by Sen. Dan Quayle, chairman of the subcommittee on employment and productivity. The bill would restrict federal job creation to quick start-up, labor-intensive activities. Unlike CETA, which let states and localities decide what kinds of jobs to create, the new program would limit jobs to areas that Congress, through earlier legislation, has already decided are of major federal interest. Job-cost limits would keep wages low enough that workers who could find private-sector jobs would want to take them.

Other features--similar to those proposed by the administration in its 1984 budget--would build on the job-training legislation passed last year. There would be an immediate expansion of the small program for retraining workers whose jobs have been permanently lost. This would be helpful, especially if the procedure were combined with jobs that allow workers to support their families while they're in training.

The legislation would also let people convert federally extended unemployment benefits into vouchers that they could give to employers who hire them. This short-term subsidy isn't likely to create any new jobs--only a permanent increase in demand for products will induce employers to step up production--but it might give people who have been unemployed a long time an advantage over less needy applicants.

Much has been learned--both good and bad-- about job-creation programs from the experience of earlier recessions. Congress would do well to remember these lessons before wasting billions on projects that will do little for the unemployed.