If industry or the medical profession or another private entity reviews its own performance, there is always a suspicion that the results might be biased and self-serving. Thus, an independent outside review may be demanded: often by a government agency with the power to destroy private persons or enterprises.

And who shall review the reviewers?

Government regulation rests on the assumption that public servants are only working for the public interest and have no potentially corrupting private agenda. The primary check on a government agency is the public notice and comment required by the Administrative Procedure Act. However, there is no higher authority to assure that the agency responds to comments in a legitimate manner.

The Environmental Protection Agency (EPA), which possesses awesome and growing power that impacts the life of every American, is under an Executive Order to conduct cost-benefit analyses of its edicts. However, there is no mechanism to check the analyses. Both EPA personnel and the outside contractors it may hire face very clear disincentives to produce results critical of proposed policies. Once a rule has been published in the Federal Register, the agency has already invested a great deal of time, money, and reputation in drafting it. As the Heartland Institute observes, ``In such an environment, cost-benefit analyses are not surprisingly viewed by the agency as a marketing opportunity, instead of a rigorous self-examination.''

Two recent, high-profile EPA analyses have now been branded as ``junk science'': the secondhand smoke study and the Great Lakes Initiative.

In a blistering 92-page decision, Judge William Osteen of the Middle District of North Carolina threw out the EPA's 1993 risk assessment linking secondhand smoke to cancer. He stated that ``EPA publicly committed to a conclusion before research had begun; excluded industry by violating the Act's procedural requirements; adjusted established procedure and established scientific norms to validate the Agency's public conclusions; and aggressively utilized the Act's authority to disseminate findings to establish a de facto regulatory scheme intended to restrict Plaintiff's products and to influence public opinion.''

If the ruling stands on appeal, it will set a precedent that risk assessments are subject to judicial review when they have a regulatory impact. ``This prospect is nothing short of a nightmare for federal agencies which are not accustomed to seeing their scientific pronouncements challenged in court.'' Having an analysis invalidated for violating standard statistical practices is a ``humiliation'' for the EPA, raising serious questions about the science of other EPA regulatory decisions, including last year's decision to tighten standards for particulate matter and ground-level ozone (see July issue). The latter action is also being challenged in court, with a ruling expected during the next year (EPA Watch, 7/31/98).

The Great Lakes Initiative (GLI) is a case study for the question: ``Should EPA Cost-Benefit Analyses Be Conducted By Independent Experts?'' The Heartland Institute Policy Study #86 by Daniel W. Smith, Ph.D., of Conestoga-Rovers & Associates presents such a review.

The 1995 EPA initiative would increase the stringency of pollution controls at a cost of $80 million to $380 million per year (EPA estimate) or billions of dollars per year (some critics' estimate). The agency boasted that it had used cost-benefit analysis throughout the development of the rule, which was touted as a ``cost effective,'' ``common sense'' regulation that would ``restore the health and economy of the Great Lakes.'' Among the claimed benefits: averting 25 to 47 cancers over the next 70 years.

According to Dr. Smith, the EPA exaggerated benefits by 312,000% by ``making assumptions it knew to be false, using data it knew to be erroneous, and using methodologies that violate its own guidelines.'' After correcting the EPA's ``many fabrications,'' he calculated that the true cost-benefit of the GLI is about 185,000 to 1, or $1 million in costs for every $5 in benefits, and that one cancer death might be averted sometime in the next 6,000 years.

Some of the EPA errors were: assuming an impossibly high fish consumption (310%); using concentration data based primarily on lake trout, which are at the top of the food chain (250%); the selective use of toxicologic data (400%); and overestimating the percentage loading due to point sources (630%).

The broader conclusion of the report, which has been widely distributed to legislators and opinionmakers, is that ``cost-benefit analyses of EPA proposals need to be conducted by independent outside parties, rather than by the agency itself.''

A flyer provided by Lawrence Parks, a speaker at the last DDP meeting, is enclosed. The principles apply to any issue.

In an August 13 article in Nature, Wentz and Schabel published corrections to satellite temperature data that account for the ``satellite dropping effect.'' Having been leaked to the press months before, it was old news by the time it was published. The press did cover a commentary by Vice Presidential climate modeler James Hansen, who triumphantly concluded that ``the bottom line of this convergence [of satellite data, surface data, and models] could be a sea change in the global warming debate. Until now, the MSU data have been the principal refuge for those who deny the reality of global warming. We believe that the warming trends of both the surface and troposphere are now sufficiently clear that the issue should no longer be whether global warming is occurring, but what is the rate of warming, what is its practical significance, and what should be done about it'' (Science 281:930-932, 1998).

Building on the crude correction proposed by Wentz and Schabel, Spencer and Christy, who collected the original data, produced a very detailed and accurate correction. They stated that the proper way to remove the bias was to apply the correction individually to each satellite. They also corrected for a slight east-west orbital drift. The conclusion: a decadal temperature trend of -0.01 degree Centigrade.