The debate in another thread has shifted to the climate debate, so perhaps we should keep it separate.
Publication bias may be detected by what I think is called a funnel plot. Imagine a funnel lying on its side.
On the X-axis, you have the power of the study - high powered studies are nearer to the truth so they lie in the stem of the funnel.
On the Y-axis you have the finding of each study - whether the activity is beneficial or not. The middle of the neck of the funnel is the best estimate of the true value.
At the left of the plot, the wide bit of the funnel, lie low powered studies. Some will show that the activity is beneficial, some the reverse. So if you look at the risk of smoking, some low powered studies should have shown that it was beneficial. IIRC, studies showing that smoking was beneficial were not published. That may be because the authors chose not to submit, or editors chose not to accept.
I have no idea whether this sort of plot has been done for the climate debate, but it ought to have been.
I accept David Z's argument that the climate has warmed and cooled long before industry appeared (even on a Roman scale), but what bugs me is the doctrine that we cannot afford to get it wrong.
Does anybody here know how man-made energy compares with the amount which arrives from the sun?
