Unfortunately, we had mixed up some results when we plotted the graphs in Figure 7 of the above paper. This affected only the two blue curves in each of the three graphs, for $2\times 2$ MIMO in Random-LOS with LP and CP incidences, and had no influence on the discussion and conclusions in the paper. The correct graphs of Figure 7 are given here, in addition to an explanation of how the best level of the curves are related to the directivity of the antenna under test.
