Statistical analysis is a crucial step in all experimental studies, including sport sciences, because inappropriate analysis can lead to erroneous assumptions of performed experiments. Statistical analyses of the training-related data are required to make the training process more efficient. The analyses of various parameters are performed in repeated cycles, requiring appropriate statistical tests. STATISTICA software (version 10) offers a Friedman test for non-parametric analyses of more than 2 groups of repeated measures (which often takes place). Unfortunately, there is no post hoc test to verify which groups decide of the statistical significance of the results. The solution to this problem may lie in the normalization of the data with one of the most popular logarithmic transformations. It allows performing multiple comparisons for the 1-way ANOVA with repeated measures, as well as appropriate post hoc test to precisely determine which group of data is responsible for the statistical significance of the differences.
K-complexes - phenomena occurring in sleep EEG - pose severe challenges in terms of detection as well as finding their physiological origin. In this study, K-complexes (KCs) were evoked by auditory stimuli delivered during sleep. The use of evoked KCs enables testing the sleeping nervous system under good experimental control. This paradigm allowed us to adopt into the KC studies a method of signal analysis that provides time-frequency maps of statistically significant changes in signal energy density. Our results indicate that KCs and sleep spindles may be organized by a slow oscillation. Accordingly, KCs might be evoked only if the stimulus occurs in a certain phase of the slow oscillation. We also observed middle-latency evoked responses following auditory stimulation in the last sleep cycle. This effect was revealed only by the time-frequency maps and was not visible in standard averages.