This is the estimate of the element's weight percent based on the ratio of the counts, uncorrected by the matrix correction routines.

K-raw(%) = (cts unk / cts std) * (beam current std / beam current unk) * 100

This is a measure of how similar the standard's counts are to that of the unknown's. If "K-raw(%)" is around 100%, then that element has about the same abundance in both the standard and the unknown.