Statistical characterization of the EEG: the use of the power spectrum as a measure of ergodicity.

In order to interpret statistical summaries of treatment-related changes in EEG power spectra, the internal consistency of the set of data from each condition must first be known. A technique for the assessment of such across-record homogeneity (ergodicity) is derived from the defining characteristics of the power spectrum. A simplified graphic approximation is also detailed. Computer simulations and practical validation confirm it as a useful means of evaluating the ergodicity of sets of EEG data.