I don't quite follow your logic in the various threads here. It seems that you are attempting to make the claim that since some cases don't follow the trend (sm

I don't quite follow your logic in the various threads here. It seems that you are attempting to make the claim that since some cases don't follow the trend (smoking can lead to a reduction in cancer for some people), therefore the statistics overall must have been wrong (smoking leads to increased rates of cancer). But that does not follow logically or mathematically. In this study, they were looking at the conditional probabilities of what is unique about people who were non-cancerous. Since in this study they excluded people with cancer, we cannot make any claim about expected lung cancer rates, since they excluded that data up-front. As for the other claims, I haven't done any research, but it sounds vaguely like Simpson's paradox could be starting to dominate the graphs, so they stopped publishing them. (i.e. there are other confounding variables that now dominate, so it is not worth the research time and money to continue collecting that data specifically for smoking rather than looking for new causes)