If you are planning only group-level analyses, using absolute power or log-transformed power should not really matter, because both absolute and log-transformed individual PSD values are supposed to be normally distributed in the population of subjects.

However, in the single-subject data the absolute PSD values follow a chi-square distribution over the sample of trials or FFT epochs, which complicates within-subject statistical tests between experimental conditions. Taking the logarithm of PSD normalizes the data, i.e. log(PSD) values are normally distributed across epochs within a subject. Therefore log-transformation of single-trial PSD allows to use t-statistics for evaluating significance of experimental effects at the single-subject level.

I am assuming from this question that you are using the newtimef function in EEGLAB to perform a time-frequency analysis (either within the study interface or outside it). This function performs baseline correction and then averages over trials before log transforming the data for conversion into decibels. In terms of reporting a baseline, because the ERSPs that you report are output in dB it seems to me that your baseline should also be reported in dB to be comparable.

You should be able to extract these baseline values via the 'powbase' output that is returned by the newtimef function. Or in case you are interested how these values are calculated, you can also calculate them by modifying this basic script, which takes the raw output of the newtimef function (in uV) and step-by-step transforms it into the dB converted baseline:

%% Perform TFA

ScalpElecs = [13]; %% Electrodes to extract in TFA

BaselineOnset = -400; %% Onset time of baseline (in ms) relative to stimulus onset

BaselineOffset = 0; %% Offset time of baseline (in ms) relative to stimulus onset

for y = ScalpElecs %% Loop over each electrode of interest

[ersp,itc,powbase,times,freqs,erspboot,itcboot,tfdata] = newtimef(EEG.data(y,:,:), EEG.pnts, [EEG.xmin*1000 EEG.xmax*1000], EEG.srate, [3 0.8] , 'topovec', 1, 'elocs', EEG.chanlocs, 'chaninfo', EEG.chaninfo, 'baseline',[-400 0], 'freqs', [.5 20], 'plotersp', 'off', 'plotitc' , 'off', 'plotphase', 'off', 'timesout', 400, 'padratio', 8, 'freqscale', 'log');

freqIdx = find(freqs>8 & freqs BaselineOnset & times< BaselineOffset); %% Get baseline times to extract

Powertrialcalc = abs(tfdata).^2; %% Convert function output to (uV^2/Hz)

Powertrialcalcavg = mean(Powertrialcalc,3); %% Average data over each trial

Basetrialcalc = mean(Powertrialcalcavg(:,ERSPBaseTimes),2); %% Get averaged baseline times

PowerBaseTotal = 10 * log10(Basetrialcalc); %% Convert averaged baselines to dB

PowerBaseFinal = mean(PowerBaseTotal(freqIdx, :)); %% Average over frequencies/times in baseline

end

Thank you both for your answers, these have been very helpful!