I counted that as “repeat”, given that’s part of how this problem was first discovered.
On reflection though, that only works when you know roughly how many negative results get published: part of the problem with unreproducable results is that negatives have been difficult to publish, biasing publications so that entirely competent and honest mistakes are much more common than one would expect from just the p-value.
On reflection though, that only works when you know roughly how many negative results get published: part of the problem with unreproducable results is that negatives have been difficult to publish, biasing publications so that entirely competent and honest mistakes are much more common than one would expect from just the p-value.