Scientific Misconduct: Falsifying the Data

Dr. Otto Raabe summarized the intentional misrepresentation of the radium dial painter data presented in BEIR IV in response to the following statement:

At 01:47 PM 4/19/99 -0500, Mike McNaughton wrote:

>Caution: the radium-dial painter data are consistent with the linear model. The data look inconsistent because they are drawn on a logarithmic graph. On this graph, the linear model transforms to an exponential, and it is possible to draw a reasonable "exponential" fit through these data.

-----------------------------------------------------------------------------------------

April 21, 1999 Davis, CA

Actually, in the November 1974 issue of the Health Physics Journal Robley Evans showed definitively that NO linear model of radiation-induced bone cancer is consistent with the U.S. data on radium in people (Robley D. Evans, "Radium in Man", HEALTH PHYSICS 27:497-510, 1974). He used linear (not logarithmic) plots and rigorous mathematical tests of several hypothetical linear models (Figures 4 and 5 in his paper). His analysis demonstrated that it is highly unlikely that these data can be explained by any linear dose-response model and that all of the linear dose-response models were "strongly rejected by the chi-square test for goodness of fit."

By grouping the Evans data into six non-uniform dose groups selected so that only one dose group included no bone cancer cases (one with average skeletal alpha doses from zero to about 500 rad or 10,000 rem)and so that the next highest dose group included a few cases of bone cancer (cases were only observed for average skeletal alpha radiation doses that exceeded 1,000 rad or 20,000 rem), Chuck Mays and Ray Lloyd created the appealing, but misleading, linear plot shown on page 198 of BEIR IV. In their plot the "threshold" region, which is below 1,000 rad, is obscured near the origin since the abscissa is extended to 16,000 rad and only one dose group was assigned to this region. Their plot proves nothing about linearity. Evans's analysis shows that no linear model fits these data.

Otto

*********************************************************
Prof. Otto G. Raabe, Ph.D., CHP
Institute of Toxicology & Environmental Health (ITEH)
(Street address: Building 3792, Old Davis Road)
University of California, Davis, CA 95616
Phone: 530-752-7754 FAX: 530-758-6140
E-mail ograabe@ucdavis.edu
*********************************************************

The RADSAFE Frequently Asked Questions list, archives and subscription information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html