Comments on the Determination of Significance Levels of the Coherence Statistic