The website FiveThirtyEight has a thoughtful post reflecting on the recent study in psychology in which 100 experimental studies were replicated . At least two points are worth highlighting. First, despite the concern about p-hacking, it turns out that one of the best predictors of replicability in that study was p-level. So while 5 percent may be too generous, there is valuable information in the p-value. The second point is that psychology is confronting its replication demons straight on. Economics? Hmmm. To read more, click here.