New guidelines proposed last April by the Environmental Protection Agency would enable the agency to label virtually anything it wants as cancer-causing -- regardless of what the science says, according to agency-watchers.

• Critics say that while science has never been EPA's strong suit, past EPA cancer risk assessments were at least rooted in science by its traditional guidelines.

• Those guidelines held that epidemological results must be statistically significant -- that they must have at least a 95 percent confidence level.

• Because of this statistical threshold, EPA has not been able to demonstrate that chlorinated drinking water, electromagnetic fields, dioxin, second-hand smoke, dietary pesticide residues and Superfund sites -- to name a few examples -- are cancer-causing.

• Critics say that with its new guidelines, EPA would free itself of the "statistical significance" straightjacket and gain much more latitude to label various phenomena as cancer risks.

Observers say the EPA is making this move because of its inability to verify cancer risks in two areas in particular: electromagnetic fields and environmental tobacco smoke.

• l A vast majority of studies on electromagnetic fields were not statistically significant at the 95 percent level, which made it impossible for the EPA to throw its weight behind regulations favoring burying power lines -- which would have cost an estimated $20 billion.

• l When the agency studied the dangers of environmental tobacco smoke, it lowered the necessary confidence limit from 95 percent to 90 percent -- tainting scientific means for regulatory ends.

Scientific observers are fearful that if the EPA has its way in this regulatory sleight-of-hand, it would set a precedent for regulators world-wide to ignore statistical significance.

Source: Steven J. Milloy (publisher of "The Junk Science Home Page" on the World Wide Web), "The EPA's Houdini Act," Wall Street Journal, August 8, 1996.