This work presents a comparison performed between two Spectral
F-Tests (SFT) used in an incremental analysis of EEG records
containing four different classes of Steady-State Visual Evoked
Potentials (SSVEP). The features extracted from these tests are
classified by a decision tree. The statistic tests were evaluated
according to the analyzed interval of EEG signal. The results
indicate that a spectral F-test for phase-locked changes is more
appropriated to smaller signal segments. The obtained
classification rate is up to 82\% with a ITR of 62.1 bits/min and
a processing time of 50 ms. The decision interval is 1 s and the
data were analyzed to be suitable to this interval. These are very
good features for an efficient online Brain-Computer Interface
(BCI) implementation.