>'“But, over all, would you say the temperature of the water was the key?” I asked. “The water temperature’s a factor. The water chemistry would also have contr

>'“But, over all, would you say the temperature of the water was the key?” I asked. “The water temperature’s a factor. The water chemistry would also have contributed.” “So a combination of the temperature and the salinity?” “But also the calcium content. That’s absolutely important.” I added that to my list of drivers: “Temperature, salinity, calcium . . .”' Of course... what is more important to determining the volume of a box? Is the height a key variable? Sure, but you also need to know the width. Not only that but also the height. Surely irrelevant things like the color of the box, or material it is made of, will also have non-zero correlations with the volume. This is one reason why the NHST strategy of finding isolated "effects" will never lead to an "equation for cancer", but is very good at generating large numbers of red herrings. You need a model of the process of carcinogenesis, and then to collect data allowing you to estimate (at least get some decent upper/lower bounds on) the parameters of the model (mutation rates, division rates, number of cells of each type, % aberrant cells that get cleared by the immune system, % that commit apoptosis, etc). Armitage and Doll started this way back in the 1950s and had quite an effect on the current cancer paradigm (slow accumulation of genetic errors in a single cell lineage -> cancer), but since then it has been only a tiny minority of cancer researchers working on the actual problem.