Scientists live in a world of absurdly, dangerously high standards. They are ruled by 95% confidence intervals. Statisticians spend their lives propping up the presumption that this is the only responsible way to do science. Yes, confidence intervals are a measure of the reliability of the method used, but that is related closely to the perception of the certainty of the conclusion, and studies generating conclusions that are only probably true (let alone possibilities that are less than probably true) tend to be dismissed by reviewers.
Yet none of us really lives our lives by such standards. Nor should we. If a substance has even a small chance of causing a life-threatening disease, we would be wise to avoid it. The civil law, recognising the impracticability of scientific standards of proof, instead requires proof on the balance of probabilities.
The precautionary principle is poorly served by scientists. So, therefore, are prudent people.
I propose, therefore, a new journal of medical research. (Other specialities should follow, but let’s do a pilot with medicine). Statistical standards are important: this mustn’t become simply a journal of vacuous speculation. The journal would publish research whose findings were justified (if you like to a 95% confidence standard) on the balance of probabilities. It would be far more useful for real decision-making of certain types than the pathologically fastidious journals we have now.
The views, opinions and positions expressed by these authors and blogs are theirs and do not necessarily represent that of the Bioethics Research Library and Kennedy Institute of Ethics or Georgetown University.