Into the new national debate over merit pay for teachers roars Dallas school superintendent Linus Wright, with all the finesse of a tank commander.

"Educators are as afraid of merit pay as they are of leprosy," says Wright, who has just rammed through, over his board's qualms and teachers' opposition, the nation's first computerized merit pay plan.

The Dallas approach differs from other classroom pay-for-performance plans in one key respect--it measure teachers by output, not input.

Each school will be measured against its projected student achievement scores on the Iowa Test of Basic Skills. The projections come from a computer. They are based on--and this Wright says he believes is the key to the plan's credibility--the past three years of test results of each student in the school.

The incentive pay system works as a sort of beat-the-computer exercise. In the top 25 percent of schools that outperform their computer projections, each teacher pockets an extra $1,500 at the end of the year and each support staffer gets $750. The average salary for teachers in Dallas is $22,000.

"We looked at every merit pay system ever tried and the one thing that brings them all down is subjectivity," said Robby Collins, Wright's director of employe relations. "Once you have human beings evaluating other human beings, the systems produce jealousies, morale problems. The beauty of our system is that it's done totally by computer."

Beauty of course, is with the beholder. The Wright plan has no end of detractors, and when the Board of Education approved it by a 5-to-4 vote last month it was for a one-year trial only.

"I'm afraid we'll wind up with graduates who know how to get good test scores but still don't know how to function very well," said board member Kathryn Gilliam, who voted with the minority.

A more ominous line of criticism comes from the teachers. Awarding bucks on the basis of test results, they say, will trigger an epidemic of cheating--by teachers.

"If you enact this plan without beefing up test security , the cheating already taking place by teachers desperate to improve scores is going to be rampant," Sharon Morgan, a middle-school math teacher, warned the board.

"In every stressful situation, you find people who cannot handle the stress," says Maureen Peters, president of the Dallas Federation of Teachers. "Even before this plan, Wright has put a tremendous stress on test scores in evaluating his principals. Principals in turn put heat on teachers. And we keep hearing stories of teachers walking around the room while the test is under way and saying 'uh-huh' or 'un-uhh' as they look over the shoulders of the kids."

Wright says he finds such arguments "embarrassing," and his director of research and evaluation, William Webster, says "it gives me a stomachache."

Wright is planning either to hire 50 part-time proctors or to have the tests administered by out-of-school counselors as a way to beef up security. In addition, says Webster, "If there is wholesale cheating, we ought to get some funny results that the computer will flag."

He says his faith in the policing powers of human nature has not been bolstered by reports from the Iowa test publishers that they are getting an unusal number of requests from Dallas for advance copies of this year's tests.

There are other objections to the merit pay plan. "It is strange that you would implement a plan that punishes good teachers in 75 percent of the schools and rewards bad teachers in 25 percent of the schools," Harley Hiscox, president of the United Teachers of Dallas, told the board.

Wright acknowledges the inequity. Originally he had proposed a multi-faceted merit pay scheme that included rewards for teacher "input" by measuring teacher competence with peer review. That approach, now being implemented elsewhere around the country, was turned down by the Dallas board for budgetary reasons.

Meanwhile, the superintendent's staff sees pluses to the team-award approach, which it borrowed from industry.

"We're hoping the stronger teachers will have a stake in helping out the weaker ones," Webster says.

The pay plan is structured to encourage other virtues. No school can even get in the running for bonuses without meeting base-line criteria on student attendance and teacher absenteeism.

"This plan has the capacity to pay for itself," Wright says, noting that for every 1 percent increase in average daily attendance, the system gets another $1 million in state aid. The first-year cost of the plan is $3 million.

The 128,000-student Dallas system is half black and 22 percent Hispanic, but the debate about racial and cultural bias in standardized tests was not raised by the three minority board members who opposed the Wright plan. Rather, they worried that social skills, values, creativity and other broad goals of education, not to mention subject matters like art, music and even science, were not being tested.

Wright says he knows education must cater to "the total child." Still, he says, "Achievement tests are the only measurement the public can relate to as far as success is concerned . . . . And besides, the bottom line is that if you master the basic skills, you're likely to get good results in the other areas, too."

Wright, 56, came to the superintendency of the system from an unlikely background. He was a boxer, a paratrooper in World War II and he spent little time in the classroom. He scaled the education bureaucracy on the business and administrative side.

Each morning his rises at 5 a.m., jogs three miles and reads passages from the Bible. He says he believes that success in public education lies with strong leadership, especially from school principals. This year he has taken the extraordinary step of placing four principals on probation, letting them know that if their schools' test scores do not go up they will be fired.

In fact, the computer model he is using for the merit pay plan was developed earlier this year to evaluate principals.

"When merit pay was suddenly placed on the public agenda by the president, I decided to seize the moment," Wright says.

One reason he has been so bold is that Dallas has an unusually sophisticated test data base. It can track the test history of each student. Thus when a student moves from one school to another--high mobility is a common feature of urban school systems--his data moves with him, keeping the school's projected test scrores in line with its ever-changing population.

A student's past three years of test scores is "an amazingly good predicter of his future test scores," Webster says. "I defy you to read the literature and find any better. It accounts for 80 percent of the variance in future scores."

Webster adds that the empirically based computer formulas have created a "level playing field" for all types of schools. "We know for example that it is easier to see dramatic gains in test scores among students in the middle achievement range than among students at either the high or the low ends. But the formula corrects for that, so no school is theoretically disadvantaged."

Webster says he thinks his system is foolproof, but his nature is cautious. "We haven't flown this in combat yet," he says. " I'll be the first to admit there could be something wrong with it. I just haven't found it yet."