Two-year-old Joy Griffith had wandered into the living room to watch cartoons on television when she apparently became trapped between the seat and the leg rest of her grandfather's brown recliner.

As she struggled to free herself, the footrest squeezed her neck. It was five minutes before Joy was discovered, and by then, doctors said, the brain damage was irreversible.

The accident was more than her father could bear. Eight months later, Charles Griffith visited her hospital crib with a $40 handgun and fired two fatal shots into the chest of the comatose child.

"I was wrong, your honor, but it wasn't my fault," Griffith later told a Florida judge who sentenced him to life in prison for first-degree murder. "I just couldn't see her suffer anymore."

In 1985, when the Consumer Product Safety Commission (CPSC) learned of the deaths of Joy and eight other children in similiar accidents, it alerted furniture manufacturers and the group quickly agreed to urge modifications in the way the chairs were made and to issue warnings to customers.

"We recognized it as a responsibility of the industry," said Joe Gerard, vice president of the American Furniture Manufacturers Association. Costs "really didn't matter," he said.

But back at the CPSC, the federal agency with authority to order unsafe products off the market, agency economists also examined the chair issue, using a controversial "cost-benefit" formula that the Reagan administration has pressed all regulatory agencies to employ before undertaking action.

The results shocked some agency officials, as well as furniture manufacturers. The economists concluded the agency should not support the changes in recliners.

"It is our recommendation that nothing be done beyond mentioning {the problem} in safety alerts . . ." wrote Warren J. Prunella, an assistant to the director of the agency's economic analysis staff, in an internal memo.

Even warnings should be tempered, Prunella urged. "The psychic costs associated with the anxiety that accompanies the release of information on household hazards is to be considered against any accompanying benefits."

To consumer activists, many of whom have expressed alarm over the CPSC's conservatism under the Reagan administration, Prunella's memo is a dramatic example of how cost-benefit analysis can be used to slow regulatory action.

"To use a cost-benefit formula to make a decision is impossible," argues Joan Claybrook, head of the Ralph Nader group Public Citizen and one of the CPSC's strongest critics. Her sentiment was backed by a number of former CPSC officials at a recent congressional hearing, including several former commission chiefs who attacked the proposal as misguided and certain to hinder the agency's ability to remove unsafe products from the market.

Sen. Albert Gore Jr. (D-Tenn.) called the efforts to expand use of the formula "kind of silly," and typical of the efforts of the agency's controversial chairman, Terrence M. Scanlon. Under Scanlon, Gore charged, the agency has signaled that it is "open season on consumers."

David Schmeltzer, the commission's compliance director, has described a proposal to mandate use of the formula as "one of the most critical issues to come before the commission." The agency's "whole approach to recalling hazardous products {is} at stake," he said.

Others, including Scanlon and Paul Rubin, a conservative who heads the commission's economics analysis staff, challenge that conclusion, calling the formulas a vital tool that can help the agency decide which problems it wishes to attack with its limited resources.

Agencies, other than the independents such as CPSC, have been required since 1981 to apply cost-benefit formulas during formal rule-making. Scanlon, a strong disciple of the formulas, is pressing the case that the commission should accept the same rules as other agencies, and even that it should expand the use of cost-benefit analysis to other commission matters, such as which hazards to study. Hearings on the issue are expected next month, officials said.

In some cases, Rubin said, the formula can make a strong case for regulatory action. Rubin said he did that recently, arguing on economic grounds that the commission should require a fourth wheel on three-wheel versions of the controversial all-terrain vehicles (ATVs). The commission has declared the vehicles unsafe and is asking the Justice Department to back a 2-to-1 vote for an ATV recall.

Rubin notes that other agencies also have used the formula in major decisions. He cites the National Highway Traffic Safety Administration, which overwhelmingly proved the benefit of requiring a third tail light on American cars after studying a dramatic drop in rear-end accidents on postal vehicles and taxis with the lights; and the Environmental Protection Agency, which based its decision to phase out leaded gasoline in part on cost-benefit studies.

To understand why the formulas have inflamed so many passions at the CPSC, pitting the agency's once-dominant compliance staff against its economists, consider the recliner case.

When Prunella drafted his memo Dec. 24, 1985, commission officials knew of nine deaths and three serious injuries in the preceding 12 years that involved children who had been entrapped in recliners, or one per year.

The costs of the hazard were simple to calculate. Figuring that each death or injury had a value of $1 million, Prunella placed "the costs" of the recliner hazard at $1 million a year.

Under the formula, that figure became the limit for what the economists believed society should have to pay to improve the product, the other side of the cost-benefit equation.

The commission had estimates that 40 million recliners were in use, so Prunella divided the $1 million by 40 million chairs and came up with an estimate of how much the repair should cost to be economically beneficial.

Prunella's answer was 2 1/2 cents per chair per year. Since Prunella figured each chair would last 10 years, he calculated that the repair costs could be spread over 10 years: 10 times 2 1/2, or 25 cents.

His conclusion: ". . . a remedial strategy that is fully effective in eliminating the hazards of entrapment will be cost effective only if it adds less than a quarter to consumers' costs."

Representatives of the American Furniture Manufacturers Association, a major industry group, had agreed two months earlier to recommend that its members install a device between the leg rest and the seat in any recliners made after October 1986. An association spokesman said he didn't know what the cost to most firms was, but that it was more than 25 cents. Despite the agency's recommendation against it, they proceeded with the modification.

There are ways other than the formula to rate hazards, a task that in the past has fallen largely to Schmeltzer's compliance staff. It has rated hazards on a scale that involves the number of reported injuries and, according to Rubin, requires a more subjective analysis than the cost-benefit formula.

Some commission officials have said privately that Rubin believes in his discipline so strongly that he would like to see economics given priority in many of the agency's deliberations. And since Rubin was hired by Scanlon, many suspect he may have an edge over others in the current debate.

Schmeltzer, one of the few remaining commission executives strongly praised by consumer groups, has emerged as Rubin's principal opponent, warning of dire consequences if the formula is set in place.

If the formula becomes a standard tool of the commission, product recalls would drop "dramatically," he said earlier this year in a memo. "A valuable part of the commission's effort to prevent injuries would be irretrievably destroyed."

Even critics of the formula concede there are instances in which it can be effectively used. The key, Public Citizen's Claybrook said, is "when you have good data."

The problem, Schmeltzer said, is that "the cost-benefit analyses performed by economics over the past year have not been meaningful but often have been misleading and unrealistic."

Rubin's formula fails to consider what Schmeltzer said are "many of the benefits" of changing a product. There is no credit given a company for the good will a safer product will create, nor are there credits for the reduction in lawsuits that a safer product could allow, Schmeltzer said.

Nonetheless, the economics staff has argued that one of the costs to be considered in a recall is the impact on a company's stock. "This reduction in stock prices means a higher cost of capital to firms, and ultimately to consumers," said one commission memo.

Noting that Rubin's staff has sometimes used cost-benefit analysis to argue against issuing news releases on some hazards, Schmeltzer said his "most serious objection" to the formula is a fear that the commission's devotion to it could "result in consumers being deprived of their right to be informed."

In addition, Schmeltzer argues, $1 million is too low a figure to place on a life. "Understating the costs of death reduces the benefits of commission action and biases ecomomic analysis toward inaction, rather than safety."

Rubin conceded that the $1 million figure, which he says his staff drew from well-accepted studies, is controversial. He defends the figure, while noting that in more recent memos his staff has been citing figures that range from $1 million to $2 million per fatality.