In order to establish whether the non-uniformity in the skill and total
errors was consistent with a 24 h rhythm, a cosinor analysis (Nelson et al.,
1979) was carried out on the corrected hourly frequencies. Phase and
amplitude were assessed, and the amplitude was subjected to a t test for the null hypothesis of zero amplitude. The results indicate that the variability
of skill-based errors and total errors was consistent with a 24 h
rhythm (Table 1).