META-ERRORS

The dotted line labeled "relative risk = 1" corresponds to no effect of radon on lung cancer at all - no increase and no decrease. The dashed line labeled "pooled miner estimate" is calculated using the no-threshold linear hypothesis to extrapolate from the lung cancer rates of miners who received very high radiation doses and suffered an ex- pected correspondingly increased cancer incidence. The error bars are at two standard deviations or 95% reliability.

Cohen's minimum lung cancer plateau at 2.3 pCi per liter corre- sponds to (37)(2.5) = 92.5 or about 93 Bq / m in the units of this graph Lubin and Boice use these plotted points to contend that this data verifies the "pooled miner estimate" and no-threshold linear hypothesis and that it disproves Cohen's experimental work

Look at the error bars! A straight line average of this data is acceptable, since the values are so scattered that any higher order assumption is probably not justified by the data. To use, however, this data to verify the existence of a straight line relationship is obviously ridiculous. Even when pooled together, these eight studies involved about 30-fold fewer homes than the Cohen research.

Moreover, about two-thirds of these values are above the radiation dose range studied by Cohen, so Lubin and Boice have about 100-fold less data than Cohen in the relevant range. This increases their error by about a factor of 10, which is the square root of 100. Assuming the primary observations to be of equivalent quality, Cohen should be able to discern the shape of the real curve with 10-fold greater reliability.

Moreover, in the region below 80 Bq / m3 these average values actually trend below 1.0, which we could cite in support of Cohen. This would be, of course, erroneous sleight-of-hand First, the scatter is too great for this conclusion. Second, if we move the threshold up to 120 Bq / m3, this "trend" disappears. The hormesis discovered by Cohen is, however, well within the error of the Lubin and Boice data.

This "meta-analysis" proves nothing except that the tax-financed de&ens of the National Cancer Institute ante becoming so nervous about Cohen's results that they are wilhng to publish virtually anything that claims to be a refutation. NCI is entirely dependent for funds upon the Clinton Administration. This Administration is filled from top to bottom with 1960's radical agitators who are "anti-nuke" to the core.