Again, the distribution of the data in the datasets being used really has an effect on the results. Minor changes in the distribution of the data can have an disproportionate effect on the median.
Medians are too blunt an instrument to really see what’s going on. I think that headline overstates it. It also gives the bulls an opportunity for a “cry wolf” comeback if/when that percentage eases up, as I’m sure it will.
.
