|
Abstract:
|
Most research studies are concerned with a number of different outcomes. In giving statistical advice, I soon became aware of the importance of considering the total number of inferences and the increase in the probability of drawing false conclusions if methods don't take this multiplicity into account. Most of my research has been in this area. The usual aim is to reject as many false hypotheses as possible while controlling some overall probability of rejecting true hypotheses. Some time ago, based on researcher reactions, I became interested in looking also at the pattern of rejections, so that the total conclusion would be as interpretable as possible. I've taken up this research again recently with some new approaches, and will describe the original motivation, the development of concepts in this area, and some results and possible future developments.
|
