Experts are good at betting which scientific experiments can replicate

EXCITING results from a scientific study are in effect meaningless if they cannot be replicated. All too often, at least in psychology experiments, that seems to be the case. A new report by a scientist who looks at this area, Brian Nosek of the University of Virginia, has once again showed that a high proportion of psychology studies failed to replicate. And this time, Dr Nosek and his colleagues may have found a shortcut to identify which fall into this category.In most circumstances, a study is considered to be significant if the odds are 5%, or lower, that the result would have occurred by chance. So for every 20 studies that get published, it is reasonable to expect that one will have results that are not correct. In 2015 Dr Nosek, working with a different team, found something alarming: that a whopping 64% of 97 psychology experiments that he re-ran failed to replicate.