Hubbard U Parameter for Fe


I was wondering why in all of the publications from people from the MP group (Jain, Urban, Ceder et al.) where iron containing oxides are investigated with GGA+U, a U value of 4.0 is reported, whereas on the MP pages and in the INCAR files, you always use 5.3? Is there a paper I missed where U(Fe)=5.3 is motivated?

There is an earlier post dealing with this problem, which clarifies thing: