The effect of smoothing functions on data obtained from a FFT

Abstract

This research investigated the effect on power spectra when data-smoothing functions were used on EEG data prior to submitting them to a FFT. We used two smoothing function options: not using any smoothing function and using the Parzen smoothing function. We developed a program to evaluate each of these functions with real and standard data. When a set of data is submitted to smoothing prior to being submitted to a FFT, there are statistically significant differences in the power spectra obtained from the FFT. This finding holds true for standard waveforms as well as for real EEG data. © 1986 Psychonomic Society, Inc.